SOCIO-ENVIRONMENTAL FACTORS AND SUICIDE IN …eprints.qut.edu.au/30317/1/Xin_Qi_Thesis.pdf ·...

SOCIO-ENVIRONMENTAL FACTORS AND SUICIDE IN

QUEENSLAND, AUSTRALIA

XIN (CHESTER) QI

MIPH, B.Med.

Submitted in satisfaction of the requirements for the degree of Master of Applied

Science (Research) in the School of Public Health, Queensland University of

Technology.

The work was mainly carried out in the School of Public Health, Queensland

University of Technology

August 2009

I

KEYWORDS

Generalized estimating equations model, geographical information system, Poisson

regression model, spatiotemporal analysis, socio-environmental factors, suicide

II

SUMMARY

Suicide has drawn much attention from both the scientific community and the public.

Examining the impact of socio-environmental factors on suicide is essential in

developing suicide prevention strategies and interventions, because it will provide

health authorities with important information for their decision-making. However,

previous studies did not examine the impact of socio-environmental factors on suicide

using a spatial analysis approach.

The purpose of this study was to identify the patterns of suicide and to examine how

socio-environmental factors impact on suicide over time and space at the Local

Governmental Area (LGA) level in Queensland. The suicide data between 1999 and

2003 were collected from the Australian Bureau of Statistics (ABS).

Socio-environmental variables at the LGA level included climate (rainfall, maximum

and minimum temperature), Socioeconomic Indexes for Areas (SEIFA) and

demographic variables (proportion of Indigenous population, unemployment rate,

proportion of population with low income and low education level). Climate data

were obtained from Australian Bureau of Meteorology. SEIFA and demographic

variables were acquired from ABS. A series of statistical and geographical

information system (GIS) approaches were applied in the analysis. This study

included two stages. The first stage used average annual data to view the spatial

pattern of suicide and to examine the association between socio-environmental factors

and suicide over space. The second stage examined the spatiotemporal pattern of

suicide and assessed the socio-environmental determinants of suicide, using more

detailed seasonal data.

In this research, 2,445 suicide cases were included, with 1,957 males (80.0%) and 488

females (20.0%). In the first stage, we examined the spatial pattern and the

III

determinants of suicide using 5-year aggregated data. Spearman correlations were

used to assess associations between variables. Then a Poisson regression model was

applied in the multivariable analysis, as the occurrence of suicide is a small

probability event and this model fitted the data quite well. Suicide mortality varied

across LGAs and was associated with a range of socio-environmental factors. The

multivariable analysis showed that maximum temperature was significantly and

positively associated with male suicide (relative risk [RR] = 1.03, 95% CI: 1.00 to

1.07). Higher proportion of Indigenous population was accompanied with more

suicide in male population (male: RR = 1.02, 95% CI: 1.01 to 1.03). There was a

positive association between unemployment rate and suicide in both genders (male:

RR = 1.04, 95% CI: 1.02 to 1.06; female: RR = 1.07, 95% CI: 1.00 to 1.16). No

significant association was observed for rainfall, minimum temperature, SEIFA,

proportion of population with low individual income and low educational attainment.

In the second stage of this study, we undertook a preliminary spatiotemporal analysis

of suicide using seasonal data. Firstly, we assessed the interrelations between

variables. Secondly, a generalised estimating equations (GEE) model was used to

examine the socio-environmental impact on suicide over time and space, as this

model is well suited to analyze repeated longitudinal data (e.g., seasonal suicide

mortality in a certain LGA) and it fitted the data better than other models (e.g.,

Poisson model). The suicide pattern varied with season and LGA. The north of

Queensland had the highest suicide mortality rate in all the seasons, while there was

no suicide case occurred in the southwest. Northwest had consistently higher suicide

mortality in spring, autumn and winter. In other areas, suicide mortality varied

between seasons. This analysis showed that maximum temperature was positively

associated with suicide among male population (RR = 1.24, 95% CI: 1.04 to 1.47) and

total population (RR = 1.15, 95% CI: 1.00 to 1.32). Higher proportion of Indigenous

population was accompanied with more suicide among total population (RR = 1.16,

95% CI: 1.13 to 1.19) and by gender (male: RR = 1.07, 95% CI: 1.01 to 1.13; female:

RR = 1.23, 95% CI: 1.03 to 1.48). Unemployment rate was positively associated with

IV

total (RR = 1.40, 95% CI: 1.24 to 1.59) and female (RR=1.09, 95% CI: 1.01 to 1.18)

suicide. There was also a positive association between proportion of population with

low individual income and suicide in total (RR = 1.28, 95% CI: 1.10 to 1.48) and

male (RR = 1.45, 95% CI: 1.23 to 1.72) population. Rainfall was only positively

associated with suicide in total population (RR = 1.11, 95% CI: 1.04 to 1.19). There

was no significant association for rainfall, minimum temperature, SEIFA, proportion

of population with low educational attainment. The second stage is the extension of

the first stage. Different spatial scales of dataset were used between the two stages

(i.e., mean yearly data in the first stage, and seasonal data in the second stage), but the

results are generally consistent with each other.

Compared with other studies, this research explored the variety of the impact of a

wide range of socio-environmental factors on suicide in different geographical units.

Maximum temperature, proportion of Indigenous population, unemployment rate and

proportion of population with low individual income were among the major

determinants of suicide in Queensland. However, the influence from other factors (e.g.

socio-culture background, alcohol and drug use) influencing suicide cannot be

ignored. An in-depth understanding of these factors is vital in planning and

implementing suicide prevention strategies.

Five recommendations for future research are derived from this study: (1) It is vital to

acquire detailed personal information on each suicide case and relevant information

among the population in assessing the key socio-environmental determinants of

suicide; (2) Bayesian model could be applied to compare mortality rates and their

socio-environmental determinants across LGAs in future research; (3) In the LGAs

with warm weather, high proportion of Indigenous population and/or unemployment

rate, concerted efforts need to be made to control and prevent suicide and other mental

health problems; (4) The current surveillance, forecasting and early warning system

needs to be strengthened, to trace the climate and socioeconomic change over time

and space and its impact on population health; (5) It is necessary to evaluate and

V

improve the facilities of mental health care, psychological consultation, suicide

prevention and control programs; especially in the areas with low socio-economic

status, high unemployment rate, extreme weather events and natural disasters.

VI

TABLE OF CONTENTS

CHAPTER 1: LITERATURE REVIEW

SUMMARY-------------------------------------------------------------------------------- 1

1. INTRODUCTION-------------------------------------------------------------------- 1

2. LITERATURY REVIEW------------------------------------------------------------ 4

2.1 METEOROLOGICAL FACTORS AND SEASONALITY---------------- 5

2.2 OTHER NON–METEOROLOGICAL FACTORS ------------------------ 14

2.2.1 SOCIOECONOMIC STATUS (SES)-------------------------------- 14

2.2.2 URBAN-RURAL DIFFERENCE------------------------------------ 16

2.2.3 POLICY------------------------------------------------------------------ 16

2.2.4 COUNTRY OF BIRTH------------------------------------------------ 17

2.2.5 INTERVENTIONS----------------------------------------------------- 18

3. SUMMARY-------------------------------------------------------------------------- 20

3.1 WHAT HAS BEEN DONE IN THIS AREA------------------------------- 20

3.2 KNOWLEDGE GAPS IN THIS AREA ------------------------------------ 20

3.3 WHAT NEEDS TO BE DONE IN FUTURE RESEARCH-------------- 22

CHAPER 2: STUDY DESIGN AND METHODS

SUMMARY------------------------------------------------------------------------------- 24

2.1 RESEARCH AIM AND OBJECTIVES------------------------------------------ 24

2.1.1 AIMS--------------------------------------------------------------------------- 24

2.1.2 SPECIFIC OBJECTIVES -------------------------------------------------- 24

2.1.3 HYPOTHESES TO BE TESTED------------------------------------------ 24

2.2 STUDY DESIGN-------------------------------------------------------------------- 25

2.3 DATA COLLECTION-------------------------------------------------------------- 25

2.4 DATA LINKAGE AND MANAGEMENT-------------------------------------- 26

2.4.1 SELECTION OF APPROPRIATE GEOGRAPHICAL UNIT FOR

FUTUTRE STUDY---------------------------------------------------------- 28

VII

2.4.2 SOCIOECONOMIC AND DEMOGRAPHIC DATA LINKAGE AND

MANAGEMENT ----------------------------------------------------------- 28

2.4.3 METEOROLOGICAL DATA LINKAGE AND MANAGEMENT--- 29

2.4.4 SUICIDE DATA LINKAGE AND MANAGEMENT ------------------- 30

2.5 ANALYTIC STRATEGIES-------------------------------------------------------- 32

2.5.1 DESCRIPTIVE STUDY----------------------------------------------------- 33

2.5.2 BIVARIABLE ANALYSIS ------------------------------------------------- 34

2.5.3 MULTIVARIABLE ANALYSIS ------------------------------------------- 34

CHAPTER 3: SOCIO-ENVIRONMENTAL DETERMINANTS OF SUICIDE: A

SPATIAL ANALYSIS

SUMMARY------------------------------------------------------------------------------- 36

3.1 INTRODUCTION------------------------------------------------------------------- 37

3.2 METHODS--------------------------------------------------------------------------- 38

3.2.1 DATA SOURCES------------------------------------------------------------- 38

3.2.2 DATA ANALYSIS------------------------------------------------------------ 39

3.3 RESULTS----------------------------------------------------------------------------- 40

3.3.1 UNIVARIABLE ANALYSIS----------------------------------------------- 40

3.3.2 BIVARIABLE ANALYSIS-------------------------------------------------- 49

3.3.3 MULTIVARIABLE ANALYSIS------------------------------------------- 54

3.3.3.1 SUICIDE AND SOCIO-ENVIRONMENTAL

DETERMINANTS -------------------------------------------------- 54

3.3.3.2 SUICIDE AND SOCIO-ENVIRONMENTAL

DETERMINANTS BY AGE AND SEX --------------------------55

3.4 DISCUSSION ----------------------------------------------------------------------- 58

3.4.1 MAJOR FINDINGS --------------------------------------------------------- 58

3.4.2 COMPARISON WITH OTHER STUDIES------------------------------- 59

3.4.3 STRENGTHENS AND LIMITATIONS ---------------------------------- 61

3.4.4 FUTURE RESEARCH DIRECTIONS ----------------------------------- 62

VIII

CHAPTER 4 PRELIMINARY SPATIOTEMPORAL ANALYSIS OF THE

ASSOCIATION BETWEEN SOCIO-ENVIRONMENTAL FACTORS AND

SUICIDE

SUMMARY------------------------------------------------------------------------------- 64

4.1 INTRODUCTION------------------------------------------------------------------- 65

4.2 METHODS--------------------------------------------------------------------------- 66

4.2.1 DATA SOURCES ------------------------------------------------------------ 66

4.2.2 DATA ANALYSIS ----------------------------------------------------------- 67

4.3 RESULTS ---------------------------------------------------------------------------- 68

4.3.1 UNIVARIABLE ANALYSIS ----------------------------------------------- 68

4.3.2 BIVARIABLE ANALYSIS ------------------------------------------------- 80

4.3.3 MULTIVARIABLE ANALYSIS------------------------------------------- 85

4.4 DISCUSSION------------------------------------------------------------------------ 86

4.4.1 MAJOR FINDINGS---------------------------------------------------------- 86

4.4.2 POSSIBLE MECHANISMS AND CONSISTENCE WITH OTHER

STUDIES----------------------------------------------------------------------- 87

4.4.3 STRENGTHENS AND LIMITATIONS ---------------------------------- 90

4.4.4 FUTURE RESEARCH DIRECTIONS------------------------------------ 91

CHAPTER 5 DISCUSSION AND CONCLUSION

SUMMARY------------------------------------------------------------------------------- 93

5.1 AN OVERVIEW OF KEY FINDINGS IN THIS STUDY--------------------- 93

5.2 BIAS AND COMFOUNDING FACTORS ------------------------------------- 94

5.3 COMPARISON WITH OTHER STUDIES-------------------------------------- 96

5.4 STRENGTHENS AND LIMITATIONS OF THIS STUDY----------------- 100

5.5 DIRECTIONS FOR FUTURE RESEARCH----------------------------------- 102

5.6 IMPLICATION FOR PUBLIC HEALTH POLICY AND

INTERVENTION------------------------------------------------------------------ 103

5.7 CONCLUSION-------------------------------------------------------------------- 104

IX

REFERENCES------------------------------------------------------------------------------ 106

X

LIST OF TABLES

TABLE 1-1: SUICIDE DEATH RATES BY STATE OF USUALLY

RESIDENCE---------------------------------------------------------------------- 4

TABLE 1-2: STUDIES OF METEOROLOGICAL AND SEASONAL FACTORS

AND SUICIDE------------------------------------------------------------------ 12

TABLE 1-3: STUDIES OF SOCIO-DEMOGRAPHIC FACTORS AND SUICIDE 19

TABLE 2-1: DETAILS OF SUICIDE IN NEWLY-ESTABLISHED LGAS, 2002–

2003------------------------------------------------------------------------------- 32

TABLE 2-2: SUICIDE CASES IN THE FINAL ANALYSIS--------------------------- 32

TABLE 3-1: SUICIDE DISTRIBUTION BY GENDER AND AGE (1999–2003) -- 43

TABLE 3-2: FREQUENCY of mortality, SMR, SEIFA AND DEMOGRAPHIC

VARIABLES ------------------------------------------------------------------- 48

TABLE 3-3: SPEARMAN CORRELATIONS BETWEEN OF SOCIO-

ENVIRONMENTAL DETERMINANTS AND SUICIDE MORTALITY

(ALL) ----------------------------------------------------------------------------- 51


ENVIRONMENTAL DETERMINANTS AND MALE ASM OF

SUICIDE------------------------------------------------------------------------- 52


ENVIRONMENTAL DETERMINANTS AND FEMALE ASM OF

SUICIDE ------------------------------------------------------------------------ 53

TABLE 3-6: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL

DETERMINANTS AND MALE SUICIDES-------------------------------55


DETERMINANTS AND FEMALE SUICIDES ---------------------------55


DETERMINANTS AND YOUNG MALE SUICIDES ------------------ 56


XI

DETERMINANTS AND YOUNG FEMALE SUICIDES --------------- 56


DETERMINANTS AND MIDDLE AGE MALE SUICIDES ----------- 57


DETERMINANTS AND MIDDLE AGE FEMALE SUICIDES ------- 57


DETERMINANTS AND SUICIDE AMONG MALE ELDERLY----- 58


DETERMINANTS AND SUICIDE AMONG FEMALE ELDERLY - 58

TABLE 4-1: FREQUENCY OF MORTALITY, SEIFA AND DEMOGRAPHIC

VARIABLES (N= 2500, 125 LGAS*4 SEASONS* 5 YEARS) -------- 79

TABLE 4-2: SUICIDE DISTRIBUTION BY MONTH (1999–2003) ----------------- 71

TABLE 4-3: SUICIDE DISTRIBUTION BY SEASON (1999–2003) ---------------- 71

TABLE 4-4: PEARSON CORRELATIONS (SEASONAL DATA, ALL) ------------ 83

TABLE 4-5: PEARSON CORRELATIONS (SEASONAL DATA, MALE) --------- 83

TABLE 4-6: PEARSON CORRELATIONS (SEASONAL DATA, FEMALE) ------ 84

TABLE 4-7: GEE REGRESSION OF SOCIO-ENVIRONMENTAL

DETERMINANTS OF SUICIDE (ALL) ----------------------------------- 85


DETERMINANTS OF SUICIDE (MALE) -------------------------------- 86


DETERMINANTS OF SUICIDE (FEMALE) -----------------------------86

XII

LIST OF FIGURES

FIGURE 1-1: THE SUICIDE TREND IN AUSTRALIA, 1921-2003------------------- 3

FIGURE 2-1: THE RELATIONSHIP BETWEEN SOCIO-ENVIRONMENTAL

DETERMINANTS AND SUICIDE---------------------------------------- 27

FIGURE 3-1: SUICIDE MORTALITY RATE (YEARLY) IN QUEENSLAND BY

GENDER (1999-2003) ------------------------------------------------------ 41

FIGURE 3-2: SUICIDE MORTALITY RATE (MONTHLY) IN QUEENSLAND BY

GENDER (1999-2003) ------------------------------------------------------ 42

FIGURE 3-3: TEMPERATURE AND RAINFALL IN QUEENSLAND (1999-2003)

------------------------------------------------------------------------------------45

FIGURE 3-4: AVERAGE ANNUAL MALE SUICIDE ASM IN QUEENSLAND

(1999-2003) ------------------------------------------------------------------- 46

FIGURE 3-5: AVERAGE ANNUAL FEMALE SUICIDE ASM IN QUEENSLAND

(1999-2003) ------------------------------------------------------------------- 47

FIGURE 4-1A: AVERAGE MALE ASM IN SPRING (1999-2003) ------------------ 72

FIGURE 4-1B: AVERAGE MALE ASM IN SUMMER (1999-2003) ---------------- 73

FIGURE 4-1C: AVERAGE MALE ASM IN AUTUMN (1999-2003) ---------------- 74

FIGURE 4-1D: AVERAGE MALE ASM IN WINTER (1999-2003) ----------------- 75

FIGURE 4-2A: AVERAGE FEMALE ASM IN SPRING (1999-2003) --------------- 76

FIGURE 4-2B: AVERAGE FEMALE ASM IN SUMMER (1999-2003) ------------- 77

FIGURE 4-2C: AVERAGE FEMALE ASM IN AUTUMN (1999-2003) ------------- 78

FIGURE 4-2D: AVERAGE FEMALE ASM IN WINTER (1999-2003) -------------- 79

XIII

ABBREVIATIONS

ABS Australian Bureau of Statistics

ASM Age-Standardised Mortality

BOM Australian Bureau of Meteorology

CI Confidence Interval

GEE Generalized Estimating Equitation

GIS Geographic Information System

GLM Generalized Linear Model

IEO Index of Education and Occupation

IER Index of Economic Resources

IRSAD Index of Relative Socio-economic Advantage and Disadvantage

IRSD Index of Relative Socio-economic Advantage and Disadvantage

LGA Local Governmental Area

PIP Proportion of Indigenous Population

PPLEL Proportion of Population with Low Educational Level

PPLII Proportion of Population with Low Individual Income

RF Rainfall

RR Relative Risk

SLA Statistical Local Areas

Tmax Maximum Temperature

Tmin Minimum Temperature

UER Unemployment Rate

XIV

STATEMENT OF ORIGINAL AUTHORSHIP

The work contained in this thesis has not been previously submitted to meet

requirements for an award at this or any other higher education institution. To the best

of my knowledge and belief, the thesis contains no material previously published or

written by another person except where due reference is made.

Signature:

Date: 17/08/2009

XV

ACKNOWLEDGEMENTS

I profoundly reserve thanks to all the people who helped me with my thesis. Prof.

Shilu Tong, Dr. Wenbiao Hu and Prof. Gerry Fitzgerald, as the supervision team, gave

me critical, thoughtful guidance, comments and advices during my master study. I

also appreciate assistance from Dr. Andrew Page, for providing suicide data and

comments on my thesis; Dr. Peter Power for providing climate data; Dr. Aaron Walker

and Mr. Hang Jin for organizing the original climate data and Dr. Adrian Barnett for

the advice on data analysis.

I am grateful to the School of Public Health and Institute of Health and Biomedical

Innovation, QUT, for all the support in my research.

As a holder of Fee Waiver and QUT Master Scholarship, I am indebted to the

Queensland University of Technology for supporting me during my study.

XVI

PUBLICATIONS BY THE CANDIDATE

JOURNAL ARTICLES

1. Qi X, Tong S, Hu W. Preliminary spatiotemporal analysis of the association

between socio-environmental factors and suicide. BMC Environmental Health.

Under review.

2. Qi X, Tong S, Hu W. Spatial Distribution of Suicide in Queensland, Australia. To

be submitted.

3. Qi X, Tong S, Hu W, Fitzgerald G. Drought and suicide: a review of

epidemiological evidence. To be submitted.

XVII

PREFACE

The data sets in this study were obtained from different sources. Suicide data were

acquired from Australian Bureau of Statistics (ABS). Meteorological data were

supplied by the Australian Bureau of Meteorology (BOM). Socio- economic Indexes

for Areas (SEIFA) and demographical data at the local governmental area (LGA) level

were from 2001 Australian Census of Population and Housing, using CDATA 2001, a

database of ABS. The author participated in the work of retrieving, linking, managing,

analyzing and interpreting the data; but did not involve in the original data collection.

The thesis is completed in the manuscript format. Chapter 1 is a review of relevant

literature. Chapter 2 describes the study design and research methods. Chapter 3

examines the spatial pattern of suicide and the possible impact of socio-environmental

factors on suicide over space. Chapter 4 assessed the spatiotemporal pattern of suicide

and explored how socio-environmental factors influenced suicide over time and space.

Chapter 5 summarised the key findings of the research, discussed strengthens and

limitations of this study and formulated recommendations.

1

CHAPTER 1. INTRODUCTION AND LITERATURE REVIEW

SUMMARY

This chapter is composed of three sections. The first section describes the pattern of

suicide in Australia. The second section critically reviews the literature about the

relationship between socio-environmental factors and suicide. The last section

discusses knowledge gaps in this area and proposes future research directions.

1 INTRODUCTION

A suicide death can be defined as ―a fatal event caused by a deliberate act of self-

harm‖ (WHO, 2006). Suicidal behaviours can be divided into three groups: 1. suicide

ideation, which means thinking of participating the act of ending one‘s life; 2. suicide

plan, which means devising or choosing particular methods through the intention of

death; and 3. suicide attempt, which means to participate any behaviour of potential

self-injury and even death (Nock et al, 2008). It is reported that there are about

877,000 suicide deaths each year globally and it is the thirteenth major cause of death

around the world (WHO, 2002, 2004). Suicide is also the sixth leading cause of

disease and health problems (WHO, 1999).

As suicide is related to psychological and psychiatric problems, the major factors

associated with suicides are depression, mood disorders, schizophrenia, anxiety,

impulsivity and sense of hopelessness (WHO, 2002). A report from the World Health

Organization indicated that a range of social and environmental factors could

2

influence suicide behaviors (WHO 2002). These factors included economic status,

financial hardship urban-rural difference, immigration and religion (WHO, 2002). As

climate change has drawn global attention, the meteorological factors were also

included as parts of socio-environmental factors and their influence on mental health

also emerged (Hoffpauir & Woodruff, 2008).

The Australian Bureau of Statistics (ABS) has established the suicide record system

since 1921. The standard suicide death rate (SSDR) experienced fluctuations from

1921 to the 1990s (Figure 1-1). During the 1930s there was a sharp decrease of SSDR

in total population and men till the bottom of 10 per 100,000 in the 1940s when

Australian troops went overseas to serve in World War II, and casualties (including

suicide and self-injuries) of these people were not regarded as part of domestic deaths

in Australia (ABS, 1997). After the World War II, SSDR increased and reached about

20/100,000 in men, 12/100,000 in women and 15/100,000 in total in the late 1960s,

and then male SSDR decreased to 17/100,000 in 1975, and rose again to 23/100,000

in 1998. SSDR of male population went up from 1975 to 1998. Female suicide SSDR

kept relative steady about 5-6/100,000 until the 1950s, when there was a drastic

increase of SSDR and reached the peak of 12/100,000 in 1963, then the SSDR

lowered and kept steady at about 5/100,000 again until 1998 (ABS, 2000). The peak

of SSDR in the 1963 owes much to the unconstrained access to hypnotic and sedative

drugs and then the restriction of such drugs by National Health Act and its

amendment in the 1960s (Oliver and Hetzel, 1972). In 1921, the ratio of male to

female suicides was about 5 to 1, and then went down at about 2.4 to 1 in the 1960

and 70s, but after that the ratio jumped again and kept about 4 to 1 since the middle of

3

the 1990s (ABS, 2000). Then the suicide mortality decreased gradually in total and by

gender (ABS, 2003, 2004).

Figure 1-1: The suicide trend in Australia, 1921-2003 (ABS, 2000, 2003, 2004).

The distribution of SSDR (1979–1998) in different states is shown in Table 1-1. In the

whole country, the suicide mortality shows some year to year fluctuations but is

steady in general, as well as in most states. The Northern Territory has the most

significant difference of suicide mortality between years (6.0 to 21.3 per 100,000)

principally because of its small population.

From 1999 to 2003, the trend of suicide rates in Australia remained steady in the

whole population. Female SSDR kept at about 5 per 100,000 and male SSDR had s

slight decrease from 20 per 100,000 in 1999 to 18 per 100,000 in 2003 (ABS, 2004).

4

Table 1-1: Suicide death rates by state of usually residence. (ABS, 2000, 2003) (a)

Year NSW VIC QLD SA WA Tas NT ACT Aust

1979 11.1 12.6 14.4 14.3 10.0 14.4 11.0 9.5 12.2

1980 11.0 11.8 13.0 11.6 10.6 11.4 14.4 6.3 11.5

1981 11.3 12.6 15.2 13.1 11.5 16.0 9.7 13.3 12.6

1982 11.4 12.1 13.0 13.3 13.4 14.5 6.0 7.9 12.2

1983 10.3 13.1 11.8 10.5 11.1 16.7 13.0 12.5 11.6

1984 9.6 11.8 13.0 11.1 13.1 12.2 8.5 14.3 11.3

1985 11.9 10.4 13.9 10.1 12.5 16.4 11.7 11.7 11.9

1986 11.3 12.6 15.2 13.1 11.5 16.0 9.7 13.3 12.6

1987 11.7 15.7 16.2 13.5 14.0 15.6 10.8 15.6 14.0

1988 12.9 12.7 15.2 13.1 13.8 16.4 18.8 11.7 13.5

1989 11.9 11.6 14.8 14.2 12.0 13.3 15.9 13.0 12.6

1990 11.6 11.4 14.9 14.8 13.6 15.5 19.5 12.7 12.7

1991 13.0 13.7 14.5 15.8 13.2 14.7 12.1 12.0 13.7

1992 12.3 12.4 14.3 14.4 13.1 20.8 14.5 10.6 13.1

1993 11.7 11.0 11.9 11.2 13.0 17.9 13.6 9.1 11.8

1994 12.8 11.3 14.3 11.3 12.8 15.0 11.6 11.9 12.6

1995 12.4 12.4 15.2 13.4 12.6 14.1 13.6 11.1 13.0

1996 13.0 10.8 16.2 12.3 12.4 13.6 20.2 11.9 13.0

1997 14.6 14.2 15.3 13.0 13.8 10.7 19.6 12.9 14.3

1998 13.3 12.1 16.3 16.1 15.2 12.4 21.3 9.5 14.0

1999 13.7 11.9 13.9 13.5 12.9 16.5 16.8 14.6 13.3

2000 11.3 10.8 15.2 13.2 13.9 10.6 21.5 9.2 12.3

2001 11.9 11.3 13.8 13.7 14.1 13.6 21.7 14.4 12.6

2002 10.4 10.8 14.5 11.2 12.6 14.8 27.8 8.1 11.8

2003 9.6 11.0 12.3 12.6 11.6 14.5 22.2 10.8 11.1

(a) Estimated by indirect method of standardisation. Rate per 100,000 persons.

2. LITERATURE REVIEW

To examine the association between socio-environmental factors and suicide, we first

searched the literature with key words of ―meteorology and suicide‖ and ―socio-

demographic factors and suicide‖ in PubMed and found relevant papers. All these

studies were reviewed by two categories: 1. Studies of meteorological and seasonal

factors and suicide (Table 1-2); 2. Studies of socio-demographic factors and suicide

(Table 1-3). The meteorological factors include temperature, rainfall, humidity and

sunshine hours. Seasonality is related to climatic conditions and its influence on

suicide is also discussed with meteorological factors below. The socio-demographical

factors consist of socio-economic status, urban-rural difference, policy, country of

birth and interventions.

5

2.1 Meteorological factors and seasonality

Ajdacic-Gross et al (2007) studied monthly weather conditions and suicide in

Switzerland, covering 120 years (1881-2000). They found a positive association

between temperature and suicide (male: r = 0.197, p < 0.001; female: r = 0.100, p <

0.001) in the whole study period. Hours of sunshine was also found to be a positive

factor influencing suicide (male: r = 0.124, p < 0.001; female: r = 0.070, p < 0.05).

Rainfall had little influence on suicide rate in this study except for female suicide

from 1971 to 2000 (r = 0.136, p < 0.01).

Benedito-Silva et al (2007) studied seasonality and suicide in different states in Brazil.

With suicide data collected from 8 states (3 in the north, 2 in the southeast and 3 in

the south) during 12 years (1979-1990), the study discovered that peak suicide

numbers emerged in November (late spring) among both sexes in Rio Grande do Sul

(male: amplitude (A) = 4.2; standard error (SE) = 1.1, p < 0.05; female: A = 2.4, SE =

0.5, p < 0.05) and among men in two states (Parana: A = 3.7, SE = 0.6, p < 0.05;

Santa Catarina: A = 1.6, SE = 0.6, p < 0.05), and in January among women in Parana

(A = 1.7, SE = 0.4, p < 0.05). It demonstrated that seasonality of suicide is evident

across geographical regions.

Deisenhammer et al (2003) studied the association between weather and suicide in

Tyrol, Austria. Mean temperature and relative humidity were associated with suicide

mortality rate. The observation days with mean temperature of 10°C and below had

the lowest suicide mortality rate, while those days with mean temperature over 20°C

6

had the highest suicide mortality rate. Days with humidity of 60% and below had the

highest suicide mortality rate, while suicide mortality was the lowest in the days with

humidity above 85%. Logistic regression was also used and showed that suicide risk

was positively associated with mean temperature (Relative Risk (RR) = 1.12, 95%

Confidence Interval (CI) 1.02 to 1.24, per 10 ºC increase) after adjustment for

geographical and social-demographic variables. In this study, rainfall information was

only categorised as yes or no rainfall days throughout study period; the effect of

rainfall on suicide was unclear.

Dixon et al (2007) studied the association between temperature and monthly suicide

mortality rate in 5 counties of the United States of America, using a simple linear

regression and discriminate analysis. They found a weak but statistically significant

association (ρ = 0.041) between temperature and suicide mortality in the five counties.

This study was implemented in small areas with limited suicides, and this study also

indicated that the suicides were not normally distributed due to months without

suicides.

Lambert et al (2003) examined effects of sunshine on daily suicides in Australia

between 1990 and 1999, using correlation and regression analysis. They discovered a

positive association between shine hours and suicide mortality. But this study did not

discuss socioeconomic status, which may have potential influence on suicide.

Lee et al (2006) studied suicide patterns in Taiwan from 1997 to 2003, and showed

that suicide had strong associations with climate. There were significant positive

associations between temperature and suicide in different groups (total: r = 0.376, p <

7

0.001; male: r = 0.336, p < 0.001; female: r = 0.382, p < 0.001; adult: r = 0.303, p <

0.001); elderly: r = 0.436, p < 0.001). The cross-correlations also suggested that

sunshine hours had a positive relationship with suicide only in elderly population (r =

0.218, p < 0.05). Humidity also greatly influenced suicide rate in different groups

(total: r = 0.368, p < 0.001; male: r = 0.350, p < 0.001; female: r = 0.398, p < 0.001;

adult: r = 0.320, p < 0.001; elderly: r = 0.438, p < 0.001). However, after adjustment

for the unemployment rate, seasonality and trend, only temperature was associated

with suicide rates (total: r = 0.036, p < 0.001; male: r = 0.036, p < 0.001; female: r =

0.020, p < 0.001; adult: r = 0.020, p < 0.001; elderly: r = 0.092, p < 0.001).

Nicholls et al (2006) used annual suicide and rainfall data in New South Wales

(NSW), Australia from 1964 to 2001 to examine the association between rainfall and

suicide. The study showed that suicide rate was negatively associated with rainfall

over time, after controlling for socioeconomic and other non-meteorological

confounders. By studying year-to-year variation in rainfall, an inverse relationship

between suicide rate and rainfall was shown, and a rainfall decline of 300 mm was

accompanied by an 8% increase in the suicide rate.

Oravecz et al (2006) studied associations between weather and monthly suicide

mortality rate in Slovenia between 1985 and 1998, using a correlation analysis. They

found a slight positive association between rainfall and suicide rate. And sunshine

hours were positively associated with suicide (r = 0.53, P < 0.001). Sunshine can

influence melatonin in human body, which affects mood regulation (Dollins et al,

1994). A peak of suicide incidence in May (1985-1993) and reduced seasonality after

1994 were also found in this study. This study considered the influence of increased

8

antidepressant use on suicide rate after 1994, but did not identify gender difference in

suicide rates.

Partonen et al (2004) studied all suicides in a 21-year-period (1979-1999) in Finland

and demonstrated that a strong seasonal effect on suicide incidence existed among the

study population (χ² = 156, P < 0.00001). Seasonal effect on suicide was strong with

the highest suicide risk in May (RR = 1.20, 95% CI: 1.13 to 1.27) and lowest in

February (RR = 0.96, 95% CI: 0.90 to 1.12). The authors hypothesised that

seasonality reflected changes of solar radiation and geomagnetic activities, which

were vital in determining human mood.

Preti et al (2007) studied suicides in Italy during 1974-2003. The study showed that

there was a strong relationship between male suicides and increasing anomalies in

monthly average temperature between May and August, and this relation was weaker

during November and December; while in January, increasing anomalies in monthly

average temperatures associated with fewer suicides. For females, the results were

less consistent and even reversed links between temperature and suicides were

observed. The authors suggested this difference was due to less-developed social

support network for males than for females, which is vital in protecting against the

outcomes of attempted suicides.

A study by Preti (1998), covering 17 towns in Italy and a period from 1974 to 1994,

revealed that annual maximum and minimum temperature had a negative relationship

with suicide in both males and females using simple regression analysis. Maximum

temperature had negative impact on suicide rates (male: ß = –0.42, p = 0.08; female: ß

9

= –0.49, p = 0.04; total: ß = –0.45, p = 0.06). There was a similar association between

minimum temperature and suicide. In this study, annual sunlight exposure had a

strong negative correlation with suicide rate among females (ß = –0.61, p = 0.03), but

a weaker association among males (ß = –0.47, p = 0.11). The explanation for this

result was that the synchronizing effect of sunlight on circadian rhythms in mammals;

and this effect was vital in controlling mood (Rudorfer et al, 1993). When there was

less sunlight, the synchronizing effect will be reduced and has less effect on

stabilizing human mood. Mood disorders in people were triggered by some extreme

climates (e.g., high temperature, natural disasters), especially in places with dry

climate (Preti, 1998).

Preti & Miottob (1998) studied all suicide cases in Italy from 1984 to 1995 with use

of Analysis of Variance (ANOVA). In the elderly aged 65 year or over, suicide cases

were higher in spring than other seasons in both sexes. While in adults (aged 25-64),

female suicides had a peak in autumn, compared with males in December, with both

genders having another peak in April. The reason for this difference was that gender

variation in the sensitivity of biological systems due to more unstable serotonin

system in females that in males, which participated in controlling polyphasic rhythm

in biological systems. For the age-related contingencies, young suicide peak coincided

with public holidays (Easter in April and Christmas in December); while peaks of

depression are usually in winter and early spring, especially among aged population

(Faedda et al, 1993).

Rock et al (2003) studied seasonality of suicide using 30-year data (1970-1999) in

Australia. Late spring peaks (November) of total suicide rate could be seen in the

10

study period. This seasonality was also seen in violent suicides (e.g., by gun). In non-

violent suicides, there was no seasonality except in females from 1990 to 1999. Male

to female ratio in total, violent and non-violent suicides had increasing trends during

the study period, with 2.26 to 3.99, 4.22 to 5.27 and 1.29 to 2.70, respectively. The

ratio of violent to non-violent suicides in total and male suicides also increased from

1970 to 1999. An explanation was that the magnitude of seasonal component has risen

in Australia during past years, which led to more violent suicides primarily by males.

Salib (1997) studied elderly suicide from 1989 to 1993 in North Cheshire, United

Kingdom, and showed a significant positive association between sunshine hours and

suicide, with odds ratio of 2.1 (95% CI: 1.2 to 20). In seasonality, suicides in summer

accounted for 45% of total suicides (Odds Ratio (OR) = 3, 95% CI: 1.4 to 17). The

study explained that sunshine hours may function as synchronizers or entraining

agents for endogenous biological rhythms that may influence an aged person's

decision to end up or endanger his or her life.

Among above studies on meteorological factors and suicide, there were three studies

in Australia, three in Italy, one in Austria, one in Brazil, one in Finland, one in

Slovenia, one in Switzerland, one in Taiwan of China, one in the United States, and

one in the United Kingdom. Some studies had included various meteorological

factors, a relative longer observation period and large area (Ajdacic-Gross, et al,

2007; Oravecz et al, 2006; Preti and Miotto, 1998; Preti, 1998); while others only

explored one specific factor (Dixon, et al, 2007; Lambert et al, 2003; Nicholls, et al,

2006; Preti et al, 2007) or seasonality (Benedito-Silva et al, 2007; Rock et al, 2003)

and suicide. However, a few studies only covered a short period (Deisenhammer et al,

11

2003; Lee et al, 2006; Salib, 1997).

In these studies, temperature, rainfall, humidity, sunshine hours were associated with

suicide to some degrees. The seasonality of suicide was weak in general. The

differences of results between these studies may be due to the difference in choosing

the study area, scale of dataset, methodology and adjusting for confounders

(Deisenhammer et al, 2003; Preti and Miotto, 1998).

12

Table 1-2: Studies of meteorological and seasonal factors and suicide

(+) means positive

(-) means negative

Note: RF: rainfall; TE: temperature; HM: humidity; SSH: sunshine hours; VL: violent; NVL: non-violent; SN: Seasonality.

Authors, year Country,

region

Observation

period

Meteorological &

other factors Statistical methods Main findings Limitations Adjustments

Ajdacic-Gross

et al, 2007

Switzerland 1877-2000 TE, SSR, RF ARIMA analysis Association between TE & suicide: (+),

esp. in summer & winter

Geographical restriction, low

magnitude of cross-correlations

Socioeconomic status

Benedito-Silva

et al, 2007

Brazil 1979-1990 Day length, SN,

latitude

Cosinor analysis,

Spearman correlation

Peak number of suicide happened in

late spring & early summer

Unreported cases due to religious

& social stigma

Urban rural difference

Deisenhammer

et al, 2003

Tyrol,

Austria

1995-2000 TE, HM, RF,

atmosphere pressure,

sultriness,

thunderstorm

Logistic regression Significant (+) for TE & thunderstorm,

(-) for HM to suicide

High correlation between

temperature and length of day

suicide time in one day,

altitude & consequent

weather in mountainous

areas Dixon et al,

2007

5 counties,

US

1991-2001 TM, monthly data Simple linear

regression, discriminate

analysis

Weak relationship between TE &

suicide, suicide seasonality

Small study areas, only 5 counties,

non-normally distributed dataset,

due to months with no suicides

Agricultural-industrial

difference

Lambert et al,

2003

Victoria,

Australia

Jan 1990- Apr

1999

SSH, daily data Correlation, regression

analysis

Daily & monthly SSH (+) with suicide Geographical variations Socioeconomic issues

Lee, et al, 2006 Taiwan,

China

1997-2003 SN & TM, monthly

data

ARIMA, cross-

correlations

Seasonality: spring peak of suicide,

temperature (+)

Misclassification, no individual

information of suicide, unreported

suicides

Socioeconomic issues

Nicholls et al,

2006

NSW,

Australia

1964-2001 RF, annual data Time series, multiple

linear regression

Long term low RF/drought increases

suicide, but not strong

Periodic variation in suicide rates Urban-rural difference,

governmental economic

assistance

Oravecz et al,

2006

Slovenia 1985-1998 TEMP, SSR, RF,

monthly data

Correlation analysis Significant correlation between suicide

& TE (SSH),1985-93

No individual information of

suicide

Rising antidepressant

use after 1994

Partonen et al,

2004

Finland 1979-1999 Solar radiation &

geomagnetic activity

Time series analysis,

locally weighted

regression

Strong SN on suicide. Increased SSH

associated with higher suicide risk

Individual data on alcohol

consumption or mental disorders

is unavailable

Socioeconomic status

Preti et al, 2007 Italy 1974-2003 TM, monthly data Gaussian low-pass

filter, linear correlation

& rank analysis

Male: higher suicide rate (May - Aug),

lower in Nov. & Dec., in Jan. rise

again; female: weaker.

Age & daily TE data unavailable Religion, tourism

13

Table 1-2: Studies of meteorological and seasonal factors and suicide (continued)

(+) means positive

(-) means negative

Note: RF: rainfall; TE: temperature; HM: humidity; SSH: sunshine hours; VL: violent; NVL: non-violent; SN: Seasonality.


region

Observation

period

Meteorological & other

factors Statistical methods Main findings Limitations Adjustments

Preti and Miotto,

1998

Italy 1984-1995 Age, gender & lethal means

(VL/NVL suicide) & SN of

suicide. monthly data

ANOVA (seasonal

analysis), simple

correlation

RF: (-) male VL. TE: (+) VL, (-) male

NVL. HM: (-) VL, (+) male NVL.

SSH: (+) VL, (-) male NVL. SN:

Aged 65 or over, higher in spring.

Adults, female peak in autumn, males

in Dec, another peak in April.

No individual data on

exposure

Mental disorder, alcohol

consumption

Preti, 1998 17 towns in

Italy

1974-1994 TE, HM, RF, SSR, monthly

data

Simple regression,

stepwise regression &

multiple regression

SSR & TE significantly (-) to

females; Stepwise regression: HM,

RF & SSH significantly relate to

suicide

Biological mechanisms on

suicide to be explored

Socioeconomic effects

& cultural habits,

unemployment,

separation & divorce

Rock et al, 2003 Australia 1970-1999 All suicide data in Australia,

seasonality

Ordinary least-squares

regression

Late spring peak of suicide rate Geographical variation Antidepressant use &

immigration

Salib, 1997 North

Cheshire,

UK

1989-1993 Aged 65 % above, TM, SSH,

RF, HM, monthly data

Pearson chi-square,

Mann-Whitney U-test,

logistical regression

SSH & HM significantly associate

with suicide rate (+)

Small number of suicide in

aged population, differential

misclassification

Psychological &

Socioeconomic

problems

14

2.2 Socio-demographic factors

Besides meteorological factors, socio-demographic factors also influence suicide

patterns (Burvill, 1998; Hall et al, 2003; Inoue et al, 2007; Page et al, 2006). These

factors include social-economic status, urban-rural difference, government policy,

country of birth and suicide prevention.

2.2.1 Socio-economic status (SES)

SES has a great impact on suicide variation in different areas. Taylor et al (2005a)

studied the association between SES and suicide in Australia from 1996 to 1998,

classifying SES into different groups aggregated by approximate equal-population

quintiles. The suicide data in the three years were aggregated. The males in the lowest

SES group had a relative risk (RR) of 1.40 (95% CI: 1.29 to 1.52) compared to those

in the highest SES group among all ages, with RR of 1.46 (95% CI: 1.27 to 1.67)

among male youth aged 25 to 34 years. As the study period is relatively short (only 3

years), it is not possible to demonstrate the change of suicide rate in any particular

SES group.

Page et al (2006) studied SES and suicide in Australia in a 25-year period (1979–

2003). This study divided the whole study period into five sessions (1979–83; 1984–

88; 1989–93; 1994–98; 1999–2003). By using Poisson regression models and after

adjusting for confounders of sex, age, country of birth (COB) and urban-rural

residence, the study showed variation in the suicide rate between groups with low and

high SES in both males and females, especially in young male groups. Suicide rate

15

ratio in young male adults (aged 20–34) ranged from 1.36 to 1.74 in low to high SES

areas during different sessions of the whole study period, and in young female adults

1.28 to 2.01. In two of 5–year time-series studies (1994–98 and 1999–2003) in this

research, young males in low SES areas had an increase of suicide rate (per 100,000)

from 44.8 (1994–98) to 48.6 (1999–2003); while in middle and high SES areas, there

were decreases from 37.3 to 33.5 (middle) and 33.0 to 27.9 (high) in the same period,

respectively. The reasons for the divergence of suicide trend among different SES

areas after 1998 were that imbalanced distribution of overall economic prosperity in

different areas; low SES areas had not gained from suicide prevention strategy and

decreased exposure to contiguous psychological and psychiatric risk factors, which

were successful in reducing suicide rate in middle and higher SES areas.

Evidence shows that unemployment is also associated with suicide. Inoue et al (2007)

studied annual suicide rates and unemployment in Japan for 27 years (1978–2004),

and discovered that annual suicide rates in males had a strong relationship with annual

employment rates (r = 0.94, p < 0.001), while the correlation was weak in females (r =

0.39, p = 0.05). The gender divergence in the relationship between unemployment and

suicide may be because the perception of work is different between males and females.

A study in Australia used yearly data and indicated that unemployment rate was

positively correlated with suicide mortality (r = 0.9, P < 0.01) among young male

adults (20–24 year-age) in the whole country between 1966 and 1996 (Morrell et al,

1998). Iverson et al (1987)‘s study applied annual data and showed that

unemployment rate was positively associated with suicide mortality by gender (male:

RR = 2.51, 95 % CI: 2.12 to 2.97; female: RR = 2.45, 95 % CI: 1.72 to 3.49) in

Denmark between 1970 and 1980.

16

2.2.2 Urban –rural difference

Page et al (2006, 2007) studied the suicide pattern in the youths (15–24 years) and

young adults (25–34 years) in rural Australia from 1979 to 2003. It showed that in

males, youth suicide rates increased in urban, rural and remote areas from 1979 to

1998. In the years between 1999 and 2003, suicide mortality rate still rose in remote

areas (from 38.8 to 47.9 per 100,000, 23% increase), but dropped in urban (from 22.1

to 16.8 per 100,000, 24% decrease) and rural (from 27.5 to 19.8 per 100,000, 28%

decrease) areas. The hypothesized reason was an imbalance in the implementation of

the youth suicide prevention strategy in different areas. Morrell et al (1999) studied

urban-rural difference of suicide in NSW of Australia between 1985 and 1994. This

study found that non-metropolitan areas had higher suicide risk, especially in males

(RR = 1.88, 95% CI: 1.69 to 2.10) rather than females (RR = 1.06; 95% CI: 0.91 to

1.25). The study by Taylor et al (2005b) also indicated that rural areas had higher

suicide risk among male population after adjustment for age, mental symptoms and

mental health services (RR = 1.19, P < 0.01). Judd et al (2006) discussed the reasons

for the urban-rural difference of suicide mortality in Australia, including

socioeconomic decline, low availability and access of health services, cultural

backgrounds (e.g., men‘s dominate role may become barriers for them to seek for help

in crisis) and owing firearms which contributed to the higher suicide mortality in rural

areas than urban areas.

2.2.3 Policy

17

As suicide has a strong relationship with socioeconomic background, government

policies which influence socioeconomic development can also have some effects on

suicide rates and the occurrence of other mental health problems. Factors contributing

to psychological and psychiatric issues, including socioeconomic consequence of

governmental policy, can be regarded as existing or potential factors associated with

suicide (Hallas et al, 2007). A study in suicide across Australia demonstrated that

during the 20th

century, NSW experienced a variation of suicide rates due to alteration

of Federal and State governments (Page et al, 2002). There was a significant

association between higher suicide risk and Federal conservative government in both

men (RR = 1.07, p < 0.01) and women (RR = 1.22, p < 0.001) compared with Federal

and state Labour governments. In reins of State conservative government, the

relationship is also evident (male: RR = 1.09, p < 0.001; female: RR = 1.17, p <

0.001). Suicide risk was highest (men: RR = 1.17, p < 0.001; female: RR = 1.40, p <

0.001) when both of Federal and State governments were conservative. If a Labour

State government coexisted with a Federal government of conservative, the relative

risks were about 1.07 to 1.09 in males (p < 0.05) and 1.09 to 1.16 in females (p <

0.001) (Page et al, 2002). Another study in UK also showed suicide rates rising during

Conservative government rein in the 20th

century (Shaw et al, 2002). One reason is

that suicide rate is associated with economic trend, which is partly resulted by

political decisions from different political parties in rein.

2.2.4 Country of birth

Country of birth and ethnic background are also associated with suicide rate. Burvill

(1998) studied suicide in Australia from 1961 to 1990 and found that European

18

migrants had higher suicide rates than those born in Australia, except for immigrants

from Southern Europe. One of the suggested reasons was that European countries

where migrants came from had higher suicide rates than Australia. It can also be

attributed to socio-environmental change of settlement (socio-cultural shock and

stress), especially for new migrants and migrants from non-English spoken countries

(e.g., Russia and Poland). While among migrants from Asia, Middle East and

Southern Europe, the suicide rate was lower than average, partly due to strong

traditional values, religion values and family influences, which had protective effects

(Burvill, 1998). Another study in Sweden obtained similar results: men born in

Finland had the highest suicide rate while women born in Finland, Poland and Eastern

Europe had the highest suicide (Westman et al, 2006). An explanation was that

migration maybe stressful to some people and caused suicide tendencies among them

(Westman et al, 2006). Lack of social networks and support was also attributed to

increased suicide risk among immigrants.

2.2.5 Interventions

Public health interventions can greatly influence suicide rate, even within a short

period. In Slovenia, the seasonal distribution of suicide decreased after 1993 due to

regular prescription and use of antidepressants in Slovenia (Oravecz et al, 2006). A

study in Australia examined prescription of antidepressants and its association with

suicide for 10 years (1991–2000) and found that total suicide rate had no significant

fall, but there was a decreased incidence of suicide in older men (rs= −0.91; 95% CI:

−0.57 to −0.98) and women (rs= − 0.76; 95% CI: −0.12 to −0.95) and increased

suicide rate in young adults (Hall et al, 2003).

19

Table 1-3: Studies of socio-demographic factors and suicide


region

Observation

period Factors Statistics methods Main findings Limitations Adjustments

Burvill, 1998 Australia 1961-1990 COB, suicide Descriptive studies European immigrants (except Southern Europe) had

higher suicide rates. Asia, Middle East & Southern

Europe immigrants had lower suicide rate

No data on the parents of

victims

Age, gender,

marital status,

language ability

Hall et al, 2003 Australia 1991-2000 Antidepressant use Spearman rank

correlations

No change in total suicide rate yet decreased in older

adults & increased in younger adults.

Unreported suicide Unemployment,

alcohol

consumption

Inoue et al, 2007 Japan 1978-2004 Suicide and

unemployment rate

Single regression

analysis & Fisher‘s

exact test

Significant correlation of suicide rates &

unemployment rates

Without regarding

confounders

Intervention of

suicide

prevention

Morrell et al,

1999

Australia 1985-1994 Urban-rural

difference

Poisson regression

models

Rural population had higher suicide risk than urban

population (male: RR=1.88, 95% CI: 1.69 to 2.10)

Individual status not

considered

Country of birth

Page, et al, 2002 NSW,

Australia

1901-1998 Federal & state

governmental type

Poisson regression

models

Suicide risk evident across the structure of

Federal/State government, both Labour (lowest),

mixed (intermediate), both conservative (highest).

Personal factors not

included

GDP change,

wartime

Page et al, 2006 Australia 1979-2003 SEIFA in 1986,

1991, 1996 and

2001

Poisson regression

models

Suicide rates: different among low and high SES

groups in males, esp. in young males.

Ecological fallacy LGA boundary

change,

Page et al, 2007 Australia 1979-2003 Urban–rural

residence, SES

Poisson regression

models

Young male suicide rate increased in urban, rural &

remote areas (1979 -1998), then dropped in urban &

rural but grown in remote (1999-2003)

Definition of ‗rurality‘ LGA boundary

change

Taylor et al, 1998 NSW,

Australia

1985-1994 Suicide, SEIFA, Generalized linear

interactive modeling

Male suicide increased when SES decreased. Suicide

RR was different among various COB groups

Unreported suicides Migrants,

country of birth

Taylor et al, 2005

a

Australia 1996-1998 GDP, political

parties holding

power,

Logistic regression

model

Suicide risk associate with continuum of federal &

state governments: both labor (lowest), mixed

(middle), both conservative (highest)

Survey questions, Demographic

confounders

Taylor et al, 2005

b

Australia 1996-1998 Urban-rural

difference

Poisson regression

models

Rural areas had higher suicide risk than urban area

among males (RR = 1.19, P<0.01)

Unreported suicides Psychological

problems

Westman et al,

2006

Sweden 1994-1999 Socioeconomic

factors among aged

25-64

Cox regression model Suicide rates different according to COB Unreported suicides, data of

outpatients, ethnicity &

social networks unavailable

Socioeconomic

factors

20

3. SUMMARY

3.1 What has been done in this area?

The studies above have explored epidemiological evidence on the relationship

between suicide and major meteorological and socioeconomic determinants.

As demonstrated in this literature review, many meteorological factors are associated

with suicide. Previous studies examined associations between suicide and many

climate variables (including rainfall, temperature, humidity and sunshine hours) and

seasonality, using a descriptive or time-series analytical approach. However, the

results differ between different studies, which may be due to the difference in the

selection of research methods (especially statistical analysis), study area and period,

population groups, and the approaches to dealing with confounders.

The influence of socioeconomic and other factors on suicide has also been examined

by many researchers. Most of these studies focus on GDP, SEIFA, unemployment,

policy and intervention. Lower SES was usually accompanied with higher suicide

mortality (Page et al, 2002, 2006, 2007; Taylor et al, 1998, 2005). Unemployment rate

was positively associated with suicide mortality (Inoue et al, 2007). Suicide mortality

varied between populations with different country of birth (Westman et al, 2006).

Public health interventions also influenced the pattern of suicide (Hall et al, 2003).

3.2 Knowledge gaps in this area

21

Socio-environmental factors are composed of meteorological, socioeconomic and

demographic factors. Previous research on the association between meteorological

factors and suicide have only mentioned confounders like socioeconomic factors and

government policy, but none of them has studied these issues in detail, especially

using analytical approaches. Very few studies explored the interrelationship among

different categories of socio-environmental factors. For example, extreme weather

may cause natural disasters, which would lead to the loss of property, damage the

industry, especially in rural areas, and reduce their income.

All of the published literature has only focused on one kind of factors (e.g.,

meteorological or socioeconomic factors), but few of them studied the different

categories of factors (e.g., socio-environmental factors). As suicide is influenced by

different factors, the impact of these factors on suicide should be studied in a

systematic way, to help formulating effective suicide prevention strategies. All the

previous studies on climate and suicide have analysed the association between

meteorological factors and suicide over time, but with very limited consideration of

geographic distribution of suicide and how the climate factors influence suicide over

different areas. Some studies have applied spatial analysis to assess the geographical

difference of suicide, but did not analyse the socio-environmental impact on suicide

(Middleton et al, 2008; Saunderson and Langford, 1996). Spatial analysis of the

impact of socio-environmental factors on suicide is critical because the distribution of

suicide deaths and its determinants varied with time and place, especially in large

geographical areas.

22

To assess the socio-environmental impact on suicide, there is a need to examine the

change of socio-environmental factors over time (different years and months) and its

association with suicide mortality rate in the same study period.

Further, there is a clear geographical variation of suicide trends, but there is little

exploration on the causes of the geographical variation, especially in relation to the

climate and socioeconomic status (Deisenhammer et al, 2003; Ajdacic-Gross et al,

2007; Preti, 1998). Further research on spatial analysis on suicide and socio-

environmental factors could uncover important geographical differences that may be

able to generate hypothesis about the causes of suicide.

3.3 What needs to be done in future research?

It is vital to use spatial analysis and geographical information system (GIS) in

examining the impact of socio-environmental factors on suicide, because socio-

demographic factors, meteorological and environmental conditions and suicide

patterns vary with geographical area. It will need to compare suicide trends,

meteorological and socioeconomic variation in each area and identify the differences

in the nature and magnitude of the associations between socio-environmental factors

and suicide over time and space.

In a study on suicide and socio-environmental change, it is essential to take into

account confounding factors such as gender, age and public health policies. The

information on these factors needs to be collected as detailed as possible.

23

Additionally, as socio-environmental change continues, a prediction model will be

required to forecast the future trend of climate, social-environmental situation and

their impact on suicide, in order to provide scientific data for policy making and

formulation of effective suicide prevention strategies.

.

24

CHAPTER 2. STUDY DESIGN AND METHODS

SUMMARY

This chapter discusses the study design and methods to examine the relationship

between socio-environmental determinants and suicide in Queensland, Australia.

2.1 RESEARCH AIM AND OBJECTIVES

2.1.1 Aims

To identify the patterns of suicide and to examine how socio-environmental factors

impact on suicide over time and space in Queensland.

2.1.2 Specific objectives:

1) To visualize the spatial pattern of suicide in Queensland, Australia (1999-2003)

2) To describe the association between socio-environmental factors and suicide in

Queensland

3) To develop multivariable regression models to assess the impact of socio-

environmental factors on suicide after adjustment for confounders

2.1.3 Hypotheses to be tested included:

a. The pattern of suicide varied at a LGA level in Queensland.

b. There is a strong seasonality of suicide in Queensland.

25

c. The spatial-temporal pattern of suicide is associated with a range of socio-

environmental factors.

2.2 STUDY DESIGN

This study examined the association of socio-environmental determinants with suicide

using a geographical information system (GIS)-based study design. A series of GIS

techniques and statistical analyses was used in this research. The study consists of

four phases: data collection, data linkage and management, descriptive study,

bivariable and multivariable analyses. The details of the phases are as follows.

2.3 DATA COLLECTION

The suicide dataset was acquired from the ABS. The suicide data, covering a 5-year

period (1999–2003), included information on age, sex, year and month of suicide,

country of birth and Statistical Local Area (SLA) code. There were altogether 2,479

suicides (1,985 males and 494 females).

The meteorological dataset was obtained from the Australian Bureau of Meteorology

(BOM). This dataset is composed of grid data by month, and covers rainfall (mm)

from 1890 to 2006 and temperature (maximum and minimum, °C) from 1950 to 2006.

In monthly data, each value represents rainfall (or min/max temperature) in one

geographical spot on the surface of the earth. The dataset contained 139 rows and 178

columns (24,742 spots/values altogether), covering the whole of Australia and its

surrounds. The interval of every two adjacent rows (latitude) and two adjacent

columns (longitude) is 0.25 degrees (about 25 km) in geography.

26

Data on socioeconomic and demographic determinants were obtained from the

CDATA 2001 of ABS, a database which provides information of 2001 Australian

Census of Population and Housing, digital statistical boundaries, base map data and

socio-economic data. CDATA 2001 also supports data analysis with tables, maps, and

graphs of data.

2.4 DATA LINKAGE AND MANAGEMENT

All original data above are in various formats, and need to be linked and managed to

be compatible with each other for further analysis. Figure 2-1 shows the data structure

for examining the relationship between socio-environmental determinants and suicide.

27

Figure 2-1: The relationship between socio-environmental determinants and suicide

Suicide (all & by

gender)

Socio-environmental

determinants

Meteorological

variables SEIFA Demographic

variables

Unemployment

rate (all & by

gender)

Proportion of

Indigenous

population (all

& by gender)

Proportion of

population with

low income (all

& by gender)

Proportion of

population with

low education (all

& by gender)

Min temperature Max temperature Rainfall

Index of Relative

Socioeconomic

Advantage & Disadvantage

Index of Relative

Socioeconomic

Disadvantage

Index of

Economic

Resources

Index of

Education &

Occupation

28

2.4.1 Selection of appropriate geographical unit for further analysis

According to ABS, there were 452 Statistical Local Areas (SLAs) in Queensland in

2001. In Queensland, there were 496 suicides on average each year from 1999 to 2003

and each SLA had only about 1 suicide every year on average (range: 0–14), so it is

difficult to detect the spatial pattern of suicide at the SLA level. Previous research on

suicide in England and Wales discussed a similar problem (Middleton et al, 2007). To

resolve this issue, geographical units covering a larger population, and hence more

suicides, were used in this study.

We chose Local Governmental Area (LGA) as a geographical unit in this study. Each

LGA has one or more SLAs (ABS 2003–05). The LGA information is from CDATA

2001 and there are 125 LGAs in Queensland. LGA information included name, code

and area (km²). Then all the data in this study need to be compiled and linked at the

LGA level.

2.4.2 Socioeconomic and demographic data linkage and management

Socioeconomic data obtained from CDATA 2001 contained Social-Economic Index

for Area (SEIFA), including four indexes: the Index of Relative Socio-economic

Advantage and Disadvantage (i.e., IRSAD, the higher IRSAD index, the higher

socioeconomic position); the Index of Relative Socio-economic Disadvantage (i.e.,

IRSD, reflecting disadvantage such as low income, low education level, high

unemployment, unskilled occupations); the Index of Economic Resources (i.e., IER,

reflecting the general level of availability to economic resources of residents and

29

households) and the Index of Education and Occupation (i.e., IEO, reflecting the

general educational level and occupational skills of people). The higher SEIFA index

(any of the four indexes) the higher SES. Demographic data included population size,

unemployment rate, proportion of Indigenous population, population with low

education level (Year 9 and below), population with low individual income (below

$200 per week) in total and by gender. SEIFA at the LGA level was created and

derived from CDATA 2001, and it can be directly combined with other data at the

LGA level. Demographic data at the LGA level were also created and derived from

CDATA. The data on unemployment rate, proportion of Indigenous population,

proportion of population with low education and low individual income (total and by

gender) were then calculated and also combined with other data at the LGA level.

2.4.3 Meteorological data linkage and management

As the raw meteorological database is composed of grid data, it needs to be

transferred into the format at a LGA level for linkage.

We used Vertical Mapper, a geographical information system (GIS) tool, to transfer

the raw meteorological data (grid data) into the LGA data. Vertical Mapper was

incorporated into the MapInfo (GIS software), and we used MapInfo as a platform to

perform data linkage, data transfer and spatial analysis.

The order of the original monthly grid data is from the left (west most) and bottom

(south most) along rows, while Vertical Mapper reads and compiles the data from the

left (west most) and top (north along) rows, so the original monthly grid data need to

30

be rearranged. We used the SharpDevelop 2.2, a free and open source programming

software package to rearrange the data, so as to make the rearranged monthly grid

data accord with Vertical Mapper application. Then we drew the LGA boundaries of

Queensland from CDATA and input them into the Vertical Mapper. After that we

used Vertical Mapper to transfer the grid data into the monthly mean data at the LGA

level. Each LGA had monthly meteorological data records (e.g., rainfall, maximum

and minimum temperature), including minimum, maximum, mean, median values and

standard deviation.

Among 125 LGAs in Queensland, 7 of them (Charters Towers, Dalby, Goondiwindi,

Redcliffe, Rockhampton, Roma and Toowoomba) did not have meteorological data

because the geographical size of these LGAs is relatively smaller than one unit (size:

0.25*0.25 degrees², or about 625 km²). To solve this problem, we used the average

meteorological data (mean value) in adjacent LGA(s) to interpolate the mean value of

meteorological data in each of the 7 LGAs. The calculation of the mean value

(MEAN) is below:

MEAN=

n

i

i

n

i

ii

area

areamean

1

1

)*(

where meani represents the mean value of adjacent LGAs and areai the area of

adjacent LGAs.

2.4.4 Suicide data linkage and management

31

Suicide data were collected at the SLA level, so they must be transferred into the

LGA-based data. The relationship between SLA and LGA code was available from

Australian Standard Geographical Classification (ASGC) (ABS, 1999-2003).

After examining all suicides in Queensland (1999–2003), we found that 26 suicide

cases (1.05% of the total) had no specific LGA code information. So these cases

cannot be incorporated into further analysis and were excluded.

From 1999 to 2003, there were some boundary changes in SLAs and LGAs, some

SLA boundary changes affected LGA boundary changes (ABS, 1999, 2000, 2002 and

2003). So the territories of some LGAs/SLAs were also changed over time. According

to the suicide database and ASGC (1999–2003), 9 suicides (3 in 1999, 2 in 2002 and 4

in 2003) were in boundary-changed LGAs/SLA. The details of these 9 cases are in

Table 2-1.

As mentioned above, further analysis will be based on information from CDATA

2001. Among these 9 suicides, only 1 suicide in 2003 (33830 Hope Vale (AC)) was

included in the analysis because the territory of LGA 33830 Hope Vale (AC) used to

belong to LGA 32500 Cook (S) in 2001, and then this case could be regarded as one

in LGA 32500 Cook according to ASGC 2001/ CDATA 2001. It is difficult to

identify the position (SLA/LGA) of other 8 cases, and therefore, these 8 cases were

excluded. The distribution of suicide cases for the final analysis is shown in Table 2-2

32

Table 2-1: Details of suicides in newly-established LGAs, 2002-2003

Year Month Age Sex SLA code & name/ LGA

code & name

Belonging to other LGA(s)/SLA(s) in

ASGC/CDATA 2001

1 2003 2 48 M 32330 Cherbourg (AC)/

32330 Cherbourg (AC)

35500 Murgon (S)/ 35500Murgon (S) & 37450

Wondai (S)/ 37450 Wondai (S)

2 2003 9 29 M 32330 Cherbourg (AC)/

32330 Cherbourg (AC)

35500 Murgon (S)/ 35500Murgon (S) & 37450

Wondai (S)/ 37450 Wondai (S)

3 2003 4 27 M 33830 Hope Vale (AC)/

33830 Hope Vale (AC)

32500 Cook (S)/ 32500 Cook (S)

4 2003 1 25 F 34420 Kowanyama (AC)/

34420 Kowanyama (AC)

32250 Carpentaria (S)/ 32250 Carpentaria (S)

& 32500 Cook (S)/ 32500 Cook (S)

5 2002 1 39 M 36070 Pormpuraaw (AC)/

36070 Pormpuraaw (AC)

32250 Carpentaria (S)/ 32250 Carpentaria (S)

& 32500 Cook (S)/ 32500 Cook (S)

6 2002 5 37 F 32154 Cambooya (S) -Pt

B/ 32150 Cambooya (S)

32154 Cambooya (S) -Pt B/32150 Cambooya

(S) & 33250 (Gatton)/ 33250 (Gatton)

7 1999 7 79 M 33968 Ipswich (C)–North/

33960 Ipswich (C)

33966 Ipswich (C)–North/ 33960 Ipswich (C)

& 31306 Karana Downs-Lake Manchester/

31000 Brisbane (C)

8 1999 11 24 F 33968 Ipswich (C)–North/

33960 Ipswich (C)

33966 Ipswich (C)–North/ 33960 Ipswich (C)

& 31306 Karana Downs-Lake Manchester/

31000 Brisbane (C)

9 1999 1 62 F 33973 Ipswich (C)–South-

West/ 33960 Ipswich (C)

33974 Ipswich (C)–South-West/ 33960 Ipswich

(C) & 30800 Boonah (S)/ 30800 Boonah (S)

Table 2-2: Suicide cases in the final analysis

Year Male Female All

1999 378 89 467

2000 444 115 559

2001 389 101 490

2002 421 100 521

2003 325 83 408

Total 1,957 488 2,445

2.5 ANALYTIC STRATEGIES

Analytic strategies, integrated with a series of statistical methods and discussion of

the results, are developed to achieve the specific objectives and test the hypotheses.

Statistical methods, including descriptive, bivariable and multivariable analyses were

applied to examine the spatial pattern of suicide and how socio-environmental

determinants influence suicide at a LGA level in Queensland. The results of the

statistical approaches, including major findings, compared with other studies,

strengths and limitations, and future research directions and public health implications,

are discussed in the following chapters. The main statistical methods are described

below.

33

2.5.1 Descriptive statistics

Distribution graphs were drawn for each of the variables, including age, gender,

country of birth, year and month of suicide, suicide place, maximum and minimum

temperature, rainfall, population size, SEIFA and demographic data. This step is vital

because we can view the pattern of distribution of each variable, and then select

appropriate approaches for bivariable and multivariable analysis.

In order to examine the spatial patterns of mortality, suicide age-standardised

mortality rates (ASM) by gender for each LGA (monthly, seasonal and yearly) were

calculated using the direct method. The distribution of population by age and gender

at a LGA level in Queensland are from ABS. The equation for calculating ASM is as

follows:

N

pNASM

ii

Where iN is the standard population size in each LGA by age and gender; ip

represents the death rate of each LGA by age and gender; N is the total population of

Queensland. In this study, firstly, the total number of suicides in the LGA by age and

gender group (the standard of age group category was as Table 3-1) was obtained;

then the age-specific rates per 100,000 of suicide deaths (age and gender group) for

each LGA were calculated; thirdly we calculated the expected number of deaths

( ii pN ) by each age and gender group by LGA; and finally the expected number of

deaths were summed, using the age structure in the total population of Queensland to

get ASM per 100,000 for each LGA.

34

The monthly and yearly mean rainfall and temperature (including maximum and

minimum) at the LGA level were also calculated. SEIFA and demographic data were

directly applied from CDATA 2001.

2.5.2 Bivariable analyses

We used MapInfo software to display the spatial and temporal distribution of suicide

cases. The spatial analysis was implemented with ASM of suicide at a LGA level.

SEIFA and demographic data were regarded as key confounders and were also

displayed at a LGA level. Chi-square tests and correlations were undertaken in

bivariable analyses. This stage was implemented to accomplish the first and second

objectives and preliminarily test the first and third hypotheses of this study.

2.5.3 Multivariable analyses

Poisson regression models were used to examine how socio-environmental factors

impact on the suicide rate, to achieve the third object and to test the second and the

third hypotheses, because the dependent variable (i.e., suicide) is a small likelihood

count, which had a Poisson distribution. We assessed the relationship between socio-

environmental determinants and suicide at different time scales (i.e. monthly, seasonal

and yearly). Confounders such as SEIFA and demographic data were adjusted for in

the multivariable model. Population in total and by gender were included as an offset

in different models.

35

In addition, a generalized estimating equations (GEE) regression model was applied

in assessing socio-environmental determinants of the suicide rate over time and

between LGAs, because this model is well suited to analyse the repeated longitudinal

data (e.g., suicide and climate data) and it fitted the seasonal data better than other

models.

36

CHAPTER 3. SOCIO–ENVIRONMENTAL DETERMINANTS OF

SUICIDE: AN EXPLORATORY SPATIAL ANALYSIS

SUMMARY

Background: The impact of socio-environmental factors on suicide has drawn

increasing attention. However, none of previous studies examined this issue using a

spatial analysis approach. This study examined the association of climate,

socioeconomic and demographic factors with suicide, using exploratory spatial and

modelling approaches.

Method: Aggregated data on suicide, demographic variables (including population,

unemployment rate, Indigenous population, population with low income and low

education) and socioeconomic indexes for areas (SEIFA) between 1999 and 2003

were acquired from Australian Bureau of Statistics. Climate data, including rainfall,

maximum and minimum temperature, were supplied by Australian Bureau of

Meteorology. A multivariable Poisson regression model was used to examine the

impact of socio-environmental factors on suicide after adjustment for a range of

confounding factors.

Results: The data analyses show that ASM of suicide varied across LGAs. Among

the male population, north, west, some of central areas had higher ASM than other

areas; while southwest and some of central areas had no suicide record. North

Queensland and some LGAs in coastal and central areas had higher female suicide

ASM, but almost half of LGAs had no female suicide record. Maximum temperature

and the proportion of Indigenous population were positively associated with suicide in

37

Queensland. Unemployment rate had a significant and positive impact on suicide

among the middle aged population. There was no significant association for other

variables.

Conclusion: Maximum temperature, the proportion of Indigenous population and

unemployment rate seem to be associated with suicide at a LGA level in Queensland.

However, it is necessary to confirm these findings using a longer term time-series data

in future research.

3.1 INTRODUCTION

It is reported that suicide is the thirteenth major cause of death around the world with

about 877,000 suicide deaths each year globally (WHO 2003 & 2002). The impact of

socio-environmental factors on suicide has drawn increasing research attention and

public interest as suicide rates have increased in some parts of the world, such as rural

Australia (Baume and Clinton, 1997; Caldwell et al, 2004; Dudley et al, 1997; Page et

al, 2007).

The impact of meteorological factors on suicide was discussed in some previous

studies (Lin et al, 2008; Linkowski 1992). Rainfall, temperature, humidity, and

sunshine hours were found having an impact on suicide and effect varied with gender

and age groups (Ajdacic-Gross et al, 2007; Nicholls et al, 2006; Preti and Miotto,

1998; Salib, 1997). Researchers also found that the distribution of suicide varied in

different seasons (Preti et al, 2007; Rock et al, 2003).

38

Previous studies also indicated that socioeconomic status (SES) can influence suicide

at different levels (Andreasyan et al, 2007; Inoue et al, 2007). Socioeconomic status,

country of birth, urban-rural difference and even government ran by particular

political party, can influence suicide mortality (Burvill, 1998; Page et al, 2002 & 2007

and Taylor et al, 1998). Increased unemployment was found to be associated with

higher suicide mortality (Westman et al, 2006).

Most of the published literature has only focused on one kind of socio-environmental

factors (e.g., meteorological factors or unemployment). As suicide is influenced by

different factors, the impact of these factors on suicide should be studied in a

systematic way, to help formulating effective suicide prevention strategies. Few of the

previous studies had consideration of geographic distribution of suicide. Spatial

analysis of the impact of socio-environmental factors on suicide is critical because the

distribution of suicide deaths and its determinants varied with time and place,

especially in large study areas.

This study aimed to use a spatial analysis approach to examine whether socio-

environmental factors are associated with suicide in Queensland, Australia, at a Local

Government Area (LGA) level.

3.2 METHODS

3.2.1 Data sources

39

The monthly meteorological data (rainfall, i.e., RF; maximum temperature, i.e., Tmax;

and minimum temperature, i.e., Tmin) in this study were supplied by Australian Bureau

of Meteorology (BOM). Suicide, Socio-economic Indexes for Area (SEIFA) and

demographic data were obtained from Australian Bureau of Statistics (ABS). SEIFA

included of four indexes: the Index of Relative Socio-economic Advantage and

Disadvantage (i.e., IRSAD); the Index of Relative Socio-economic Disadvantage (i.e.,

IRSD); the Index of Economic Resources (i.e., IER) and the Index of Education and

Occupation (i.e., IEO). The higher SEIFA index (any of the four indexes) the higher

socioeconomic status. Demographic data included population size, unemployment

rate (i.e., UER), proportion of Indigenous population (i.e., PIP), population with low

education level (i.e., PPLEL), population with low individual income (PPLII) in total

and by gender. This study involved 2,445 suicide deaths from 1999 to 2003, with

1957 males (80.0%) and 488 females (20.0%).

3.2.2 Data analysis

A series of GIS and statistical methods were used to analyse these data. MapInfo, a

geographical information system (GIS) tool (including Vertical Mapper) was used to

perform data link, data transfer and to explore spatial patterns of socio-environmental

variables and suicide.

Statistical analyses, including univariable, bivariable, and multivariable approaches,

were performed to examine the major socio-environmental determinants of suicide.

Univariable analysis was implemented to describe characteristics of each variable.

Bivariable analysis included a series of Spearman correlations. Finally a multivariable

40

Poisson regression model was developed to assess the impact of socio-environmental

factors on suicide, after adjustment for the effects of potential confounders. The

Statistical Package for the Social Sciences (SPSS) program was used for data

management and analysis.

As the original meteorological and suicide dataset were monthly data, we aggregated

the monthly data into dataset of five years altogether. Suicide counts in total and by

gender at the LGA level applied the sum counts of five year value, while suicide

mortality and ASM by gender used average yearly data. Rainfall values at the LGA

level were counted as total values, using the sum of five year data. For Tmax and Tmin

at the LGA level, we used mean value of monthly data for further analysis. SEIFA

and demographic data were directly applied, as they were from CDATA and we used

the same value of each index at the LGA level for all five years.

3.3 RESULTS

3.3.1 Univariable analysis

Figure 3-1 shows the annul distribution of suicide mortality in Queensland (1999-

2003). The suicide mortality in Queensland experienced fluctuations in this period,

higher in 2000 and 2002 and lower in 2003 among total and male mortalities. Female

mortality trend was relatively steady and between 5 and 6 per 100,000. The

distribution of suicides by age and gender in separate years was demonstrated in

Table 3-1. Most of the suicide cases were aged between 20 and 59 years. Male

suicides accounted for 80% of total deaths.

41

Figure 3-2 shows the monthly mortality of suicide in total Queensland population

(1999–2003) and by gender. The suicide mortality experienced fluctuations in the 60–

month study period, suicide peaks emerged in January (2002), March (2001), August

(1999 and 2003), October (2000) between different years, while suicide mortality

reached bottoms in June (1999), July (2000), September (2002), November (2001)

and December (2003).

Figure 3-1: Suicide mortality rate (yearly) in Queensland by gender (1999-2003) (Source: ABS)

42

Figure 3-2: Suicide mortality rate (monthly) in Queensland by gender (1999-2003) (Source: ABS)

43

Table 3-1: Suicide deaths by gender and age in Queensland (1999-2003)

Male Female Total

Age 1999 2000 2001 2002 2003 All

males 1999 2000 2001 2002 2003

All

females 1999 2000 2001 2002 2003

All

suicides

0-4 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

5-9 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

10-14 2 3 2 0 1 8 4 0 1 2 3 10 6 3 3 2 4 18

15-19 24 28 23 21 17 113 2 11 6 6 5 30 26 39 29 27 22 143

20-24 41 42 46 40 31 200 7 6 9 8 4 34 48 48 55 48 35 234

25-29 47 62 37 50 35 231 10 14 11 11 7 53 57 76 48 61 42 284

30-34 42 51 51 48 44 236 11 14 13 12 11 61 53 65 64 60 55 297

35-39 44 50 38 53 39 224 7 21 10 8 11 57 51 71 48 61 50 281

40-44 43 52 44 56 31 226 14 13 9 18 12 66 57 65 53 74 43 292

45-49 34 32 39 24 39 168 7 5 12 14 10 48 41 37 51 38 49 216

50-54 32 25 29 35 18 139 9 6 8 4 5 32 41 31 37 39 23 171

55-59 18 25 19 19 17 98 7 5 6 5 6 29 25 30 25 24 23 127

60-64 12 14 15 27 9 77 6 2 4 3 2 17 18 16 19 30 11 94

65-69 7 20 13 13 10 63 2 2 1 2 2 9 9 22 14 15 12 72

70-74 14 14 14 11 14 67 2 6 6 2 0 16 16 20 20 13 14 83

75-79 9 13 13 12 9 56 1 5 2 2 2 12 10 18 15 14 11 68

80-84 8 7 4 6 7 32 0 5 1 2 1 9 8 12 5 8 8 41

85-89 1 4 2 4 3 14 0 0 1 1 1 3 1 4 3 5 4 17

90-94 0 2 0 2 1 5 0 0 1 0 1 2 0 2 1 2 2 7

95-99 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

100+ 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0 0

Total 378 444 389 421 325 1957 89 115 101 100 83 488 457 559 490 521 408 2445

44

During 1999 - 2003, temperature (maximum, minimum and mean) kept relatively

steady in Queensland, while rainfall varied between different years (Figure 3-3). The

volume of rainfall reached peak in 2000 at 910 mm and decreased drastically to the

bottom of 393 mm in 2002. In the monthly data, rainfall reached peaks in summer

(December-February) and bottoms in winter and early spring (June-September).

Maximum, minimum and mean temperature reached the peak in January and dropped

to the bottom in July each year. As expected, the three measures of temperature

showed a strong correlation over time. Rainfall peaked in January 1999, February and

December of 2000, February 2002 and February 2003; and fell to the bottoms in

September 1999, July and September of 2000, May and August of 2001, July 2002;

and September of 2003.

Figure 3-4 shows the map of average annual male suicide ASM at a LGA level in

Queensland. It showed that northern areas (Peninsula of Cape York and coastal areas

of Gulf of Carpentaria), part of western areas, central areas, part of southern, south-

eastern coastal areas and eastern areas had higher suicide ASM, while northern-

central, south-western, southern and south-eastern inland areas had lower suicide

ASM.

Female suicide ASM map (at a LGA level) is shown in Figure 3-5. 61 LGAs (almost

half of 125 LGAs in Queensland) had no suicide record. Northern areas, part of

central, south-central, eastern and southern costal areas had highest female suicides

ASM.

45

Figure 3-3: Temperature and rainfall in Queensland (1999-2003) (Source: BOM)

46

Figure 3-4: Average annual male suicide ASM in Queensland (1999-2003)

47

Figure 3-5: Average annual female suicide ASM in Queensland (1999-2003)

48

Summary statistics of mortality, ASM by gender, climate variables, SEIFA and

demographic variables are shown in Table 3-2.

Table 3-2: Frequency of mortality, ASM, SEIFA and demographic variables (N= 125)

Note: ASM (age-standardised mortality); RF (rainfall); Tmin (minimum temperature); Tmax (maximum temperature);

IRSAD (Index of Relative Socio-economic Advantage and Disadvantage); IRSD (Index of Relative Socio-

economic Advantage and Disadvantage); IER (Index of Economic Resources); IEO (Index of Education and

Occupation); PIP (proportion of Indigenous population); UER (unemployment rate); PPLII (proportion of

population with low individual income); PPLEL (proportion of population with low educational level)

In this dataset, mortality, ASM (both genders), RF, PIP (total and by gender) were not

normally distributed between LGAs. Tmax, Tmin, SEIFA, UER (total and by gender),

PPLII (total and by gender) and PPLEL (total and by gender) did have a normal

distribution between LGAs, by normality plots with tests in the SPSS.

There was a remarkable difference across LGAs in Queensland. For example,

Brisbane City had a 1327 km²area with a population of 888,499 (2001 census data)

Mean

Std.

Deviation Minimum

Percentiles

25th 50th 75th Maximum

Mortality (per 100,000) 17.12 27.876 0.00 8.40 13.03 19.08 296.30

Male ASM rate (per 100,000) 28.11 46.873 0.00 14.11 20.72 30.65 493.00

Female ASM rate (per 100,000) 5.50 11.023 0.00 0.00 1.75 6.70 87.40

RF (mm) 3900 2084.0 1017 2718 3212 4446 13490

Tmin (°C) 15.1 2.76 9.6 12.9 15.1 17.1 24.0

Tmax (°C) 28.0 2.48 23.4 26.1 27.8 29.4 33.9

IRSAD 935.61 41.634 831.36 909.92 930.64 962.88 1059.84

IRSD 957.71 69.570 472.08 946.04 972.48 992.52 1048.88

IER 942.25 52.682 835.52 903.64 939.36 975.96 1083.76

IEO 929.05 35.702 815.68 908.88 925.84 945.96 1064.32

Total UER (%) 6.73 3.833 0.00 4.01 6.04 8.78 23.25

Male UER (%) 7.14 4.543 0.00 4.05 6.45 9.38 26.71

Female UER (%) 6.25 3.122 0.00 4.14 6.11 7.95 18.37

Total PIP (%) 7.79 14.447 0.00 1.92 2.86 6.20 87.51

Male PIP (%) 7.43 13.991 0.00 1.85 2.98 5.75 86.65

Female PIP (%) 8.23 15.044 0.00 1.87 3.14 6.62 88.67

Total PPLEL (%) 22.75 5.489 11.91 19.11 22.94 26.13 47.41

Male PPLEL (%) 24.61 6.258 11.21 19.93 25.25 28.59 50.83

Female PPLEL (%) 20.64 5.011 9.01 17.00 20.70 23.28 43.91

Total PPLII (%) 28.14 7.501 10.65 24.36 27.91 32.01 61.71

Male PPLII (%) 23.06 8.971 4.30 17.59 22.44 27.57 62.22

Female PPLII (%) 34.10 6.636 17.13 30.46 33.63 37.92 65.18

49

and 565 suicide cases (1999-2003). While Diamantina Shire covers 94832 km² but

had only 448 persons (2001 census data) and no suicides between 1999 and 2003.

3.3.2 Bivariable analysis

Spearman correlation was chosen to examine the association between suicide

mortality and socio-environmental determinants because some variables were not

normally distributed (e.g., suicide mortality). Spearman‘s correlations were

implemented, using aggregated data of the whole study period on a range of variables,

including total suicide mortality, suicide ASM by gender, climate variables, SEIFA

and demographic variables.

Table 3-3 shows the correlations between socio-environmental factors and suicide

mortality. RF was positively correlated with suicide mortality (rs = 0.249, P < 0.01).

SEIFA had negative correlation with suicide mortality (IRSAD: rs = –0.361, P < 0.01;

IRSD: rs = –0.392, P < 0.01; IER: rs = –0.261, P < 0.01; IEO: rs = –0.267, P < 0.01).

Demographic variables positively influenced suicide mortality (PIP: rs = 0.205, P <

0.05; UER: rs = 0.312, P < 0.01; PPLII: rs = 0.360, P < 0.01; PPLEL: rs = 0.225, P <

0.05). The association between temperature and suicide mortality was not significant

(Tmax: rs = –0.029, P = 0.749); Tmin: rs = 0.100, P = 0.268).

Correlations between socio-environmental factors and male ASM of suicide are

shown in Table 3-4. RF was positively correlated with male ASM of suicide. (rs =

0.251, P < 0.01). SEIFA had negative correlation with male ASM of suicide (IRSAD:

rs = –0.307, P < 0.01; IRSD: rs = –0.345, P < 0.01; IER: rs = –0.253, P < 0.01; IEO: rs

50

= –0.231, P < 0.01). PPLII was positively associated with male ASM of suicide (rs =

0.364, P < 0.01). UER had positively correlation with male ASM of suicide (rs =

0.310, P <0.01). PIP was only marginally and positively associated with male ASM

(rs = 0.167, P = 0.063). The correlations between Tmin (rs = 0.112, P = 0.215), Tmax (rs

= –0.034, P = 0.710), PPLEL (rs = 0.129, P = 0.151) and male ASM of suicide were

not statistically significant.

In examining correlations between socio-environmental factors and female ASM of

suicide (Table 3-5), only RF (rs = 0.451, P < 0.01), Tmax (rs = –0.298, P < 0.01) and

UER (rs = 0.446, P < 0.01) were significantly associated with female ASM of suicide.

IER had statistical and negative association with female ASM (rs = –0.180, P = 0.045).

Other variables were not significantly correlated with female ASM of suicide (P > 0.1

in each).

Among all the variables, some SEIFA indexes were highly correlated between each

other (e.g., IRSAD and IRSD: rs = 0.92). Thus IRSD was used in the modelling

process to avoid multicollinearity because it had the most significant association with

suicide among the four indexes. And IRSD is a general index describing the

proportion of skilled work force and people with high incomes, the vital aspects in

socioeconomic status, in each LGA (ABS, 2001). The correlation between Tmax and

Tmin was also significant (rs = 0.620). As Queensland lies in tropical and subtropical

zones, the extreme high temperature is common in many LGAs, while Tmin is

moderate between LGAs (9.6ºC to 24.0ºC) with no extreme high or low values in the

aggregated data. We selected Tmax for further analysis.

51

Table 3-3: Spearman correlations between socio-environmental determinants and suicide mortality (all)

Mortality RF Tmin Tmax IRSAD IRSD IER IEO PIP UER PPLII

RF rs 0.249** 1.000

P-value 0.005 .

Tmin rs 0.100 0.274** 1.000

P-value 0.268 0.002 .

Tmax rs –0.029 –0.429** 0.620** 1.000

P-value 0.749 0.000 0.000 .

IRSAD rs –0.361** –0.149 0.136 0.103 1.000

P-value 0.000 0.097 0.131 0.255 .

IRSD rs –0.392** –0.189* –0.211* –0.163 0.708** 1.000

P-value 0.000 0.035 0.018 0.069 0.000 .

IER rs –0.296** –0.137 0.236** 0.206* 0.921** 0.527** 1.000

P-value 0.001 0.129 0.008 0.021 0.000 0.000 .

IEO rs –0.267** 0.077 –0.057 –0.184* 0.733** 0.675** 0.492** 1.000

P-value 0.003 0.390 0.529 0.040 0.000 0.000 0.000 .

PIP rs 0.205* –0.115 0.405** 0.566** –0.144 –0.510** –0.019 –0.215* 1.000

P-value 0.022 0.202 0.000 0.000 0.108 0.000 0.835 0.016 .

UER rs 0.312** 0.566** –0.036 –0.500** –0.284** –0.329** –0.332** 0.038 –0.125 1.000

P-value 0.000 0.000 0.693 0.000 0.001 0.000 0.000 0.671 0.164 .

PPLII rs 0.360** 0.444** –0.224* –0.407** –0.706** –0.465** –0.736** –0.354** –0.118 0.592** 1.000

P-value 0.000 0.000 0.012 0.000 0.000 0.000 0.000 0.000 0.191 0.000 .

PPLEL rs 0.225* –0.148 –0.143 0.124 –0.803** –0.586** –0.765** –0.609** 0.240** 0.033 0.481**

P-value 0.012 0.100 0.113 0.168 0.000 0.000 0.000 0.000 0.007 0.717 0.000

*. Correlation is significant at the 0.05 level (2-tailed).

**. Correlation is significant at the 0.01 level (2-tailed).

Note: RF (rainfall); Tmin (minimum temperature); Tmax (maximum temperature); IRSAD (Index of Relative Socio-economic Advantage and Disadvantage); IRSD (Index of Relative Socio-

economic Advantage and Disadvantage); IER (Index of Economic Resources); IEO (Index of Education and Occupation); PIP (proportion of Indigenous population); UER (unemployment rate);

PPLII (proportion of population with low individual income); PPLEL (proportion of population with low educational level)

52

Table 3-4: Spearman correlations between socio-environmental determinants and male ASM of suicide

ASM RF Tmin Tmax IRSAD IRSD IER IEO PIP UER PPLII

RF rs 0.251** 1.000

P-value 0.005 .

Tmin rs 0.112 0.274** 1.000

P-value 0.215 0.002 .

Tmax rs –0.034 –0.429** 0.620** 1.000

P-value 0.710 0.000 0.000 .

IRSAD rs –0.307** –0.149 0.136 0.105 1.000

P-value 0.000 0.097 0.131 0.244 .

IRSD rs –0.345** –0.189* –0.211* –0.167 0.708** 1.000

P-value 0.000 0.035 0.018 0.062 0.000 .

IER rs –0.253** –0.137 0.236** 0.209* 0.921** 0.527** 1.000

P-value 0.004 0.129 0.008 0.019 0.000 0.000 .

IEO rs –0.231** 0.077 –0.057 –0.178* 0.733** 0.675** 0.492** 1.000

P-value 0.010 0.390 0.529 0.047 0.000 0.000 0.000 .

PIP rs 0.167 -0.089 0.422** 0.559** –0.144 –0.493** –0.030 –0.186* 1.000

P-value 0.063 0.325 0.000 0.000 0.108 0.000 0.738 0.038 .

UER rs 0.310** 0.577** –0.021 –0.484** –0.306** –0.331** –0.360** 0.044 -0.097 1.000

P-value 0.000 0.000 0.813 0.000 0.001 0.000 0.000 0.624 0.280 .

PPLII rs 0.364** 0.507** –0.194* –0.421** –0.715** –0.483** –0.772** –0.252** –0.054 0.669** 1.000

P-value 0.000 0.000 0.031 0.000 0.000 0.000 0.000 0.005 0.548 0.000 .

PPLEL rs 0.129 –0.229* -0.149 0.175 –0.751** –0.520** –0.733** –0.578** 0.237** –0.019 0.422**

P-value 0.151 0.010 0.097 0.051 0.000 0.000 0.000 0.000 0.008 0.830 0.000

*.Correlation is significant at the 0.05 level (2-tailed).


Note: ASM (age-standardised mortality); RF (rainfall); Tmin (minimum temperature); Tmax (maximum temperature); IRSAD (Index of Relative Socio-economic Advantage and Disadvantage);

IRSD (Index of Relative Socio-economic Advantage and Disadvantage); IER (Index of Economic Resources); IEO (Index of Education and Occupation); PIP (proportion of Indigenous

population); UER (unemployment rate); PPLII (proportion of population with low individual income); PPLEL (proportion of population with low educational level)

53

Table 3-5: Spearman correlations between socio-environmental determinants and female ASM of suicide

ASM RF Tmin Tmax IRSAD IRSD IER IEO PIP UER PPLII PPLEL

RF rs 0.451** 1.000

P-value 0.000 .

Tmin rs 0.016 0.274** 1.000

P-value 0.863 0.002 .

Tmax rs –0.298** –0.429** 0.620** 1.000

P-value 0.001 0.000 0.000 .

IRSAD rs –0.152 –0.149 0.136 0.105 1.000

P-value 0.090 0.097 0.131 0.244 .

IRSD rs –0.157 –0.189* –0.211* –0.167 0.708** 1.000

P-value 0.081 0.035 0.018 0.062 0.000 .

IER rs –0.180* –0.137 0.236** 0.209* 0.921** 0.527** 1.000

P-value 0.045 0.129 0.008 0.019 0.000 0.000 .

IEO rs 0.062 0.077 –0.057 –0.178* 0.733** 0.675** 0.492** 1.000

P-value 0.491 0.390 0.529 0.047 0.000 0.000 0.000 .

PIP rs 0.008 –0.115 0.394** 0.555** –0.147 –0.525** –0.017 –0.235** 1.000

P-value 0.932 0.201 0.000 0.000 0.103 0.000 0.852 0.008 .

UER rs 0.446** 0.508** 0.001 –0.443** –0.219* –0.321** –0.241** –0.027 –0.115 1.000

P-value 0.000 0.000 0.989 0.000 0.014 0.000 0.007 0.764 0.200 .

PPLII rs 0.108 0.187* –0.191* –0.218* –0.487** –0.291** –0.452** –0.517** –0.164 0.286** 1.000

P-value 0.230 0.037 0.033 0.015 0.000 0.001 0.000 0.000 0.067 0.001

PPLEL rs 0.122 0.010 –0.141 0.028 –0.825** –0.651** –0.754** –0.589** 0.255** 0.063 0.369** 1.000

P-value 0.177 0.908 0.116 0.758 0.000 0.000 0.000 0.000 0.004 0.485 0.000 .

*.Correlation is significant at the 0.05 level (2-tailed).





54

3.3.3 Multivariable analysis

The multivariable analyses focused on assessing the association of climate variables,

SEIFA and demographic variables with suicide mortality rate. As suicide was a small

likelihood event, Poisson regression model was applied in assessing associations

between suicide and socio-environmental determinants after adjustment for gender,

age and population size.

The generalized linear model (GLM) with Poisson link was executed with selection of

different socio-environmental variables. As most suicide deaths were between 20-59

years old, the suicides were divided into three groups by gender: young (below 20–

years), middle (20 to 59–years) and old (60–years and over). The following analyses

focused on age group and gender differences between socio-environmental factors

and suicide. In order to avoid multicollinearity, the GLM included only IRSD to

represent SEIFA in each LGA, as it had the most significant association with suicide

mortality. In terms of the demographic variables, we used two indexes: proportion of

Indigenous population and unemployment rate for modelling. We adjusted for

population size at the LGA level as an offset in the modelling process.

3.3.3.1 Suicides and socio-environmental determinants

Table 3-6 reveals the associations between socio-environmental factors and male

suicide after adjustment for age. Increased Tmax was accompanied with more suicide

deaths (Relative Risk (RR) = 1.03, 95% CI: 1.00 to 1.07, per 1 degree increase). PIP

was significantly and positively associated with suicide mortality (RR = 1.02, 95% CI:

55

1.01 to 1.03, per 1% increase). UER had a significant and positive association with

suicide mortality (RR = 1.04, 95% CI: 1.02 to 1.06, per 1% increase). RF (RR = 1.00,

95% CI: 1.00 to 1.00) and IRSD (RR = 0.98, 95% CI: 0.83 to 1.17) were not

significantly associated with suicide.

Table 3-6: Poisson regression of socio-environmental determinants and male suicides

Parameter ß SE

95% CI P-value

Lower Upper

RF (m) 0.002 0.0119 1.00 1.00 0.875

Tmax (°C) 0.033 0.0157 1.00 1.07 0.036

IRSD –0.017 0.0894 0.83 1.17 0.848

PIP (%) 0.016 0.0053 1.01 1.03 <0.001

UER (%) 0.040 0.0108 1.02 1.06 <0.001

*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and

Disadvantage); PIP (proportion of Indigenous population); UER (unemployment rate).

The associations between socio-environmental factors and female suicide are shown

in Table 3-7, after adjustment for age. UER was marginally and positively associated

with suicide (RR = 1.07, 95% CI: 1.00 to 1.16). There was no significant association

for RF (RR = 1.03, 95% CI: 0.99 to 1.08), Tmax (RR = 0.99, 95% CI: 0.92 to 1.06),

IRSD (RR = 0.98, 95% CI: 0.64 to 1.49) and PIP (RR = 1.01; 95% CI: 0.99 to 1.04).

Table 3-7: Poisson regression of socio-environmental determinants and female suicides

Parameter ß SE

95% CI P-

value Lower Upper

RF (m) 0.030 0.0227 0.99 1.08 0.192

Tmax (°C) –0.013 0.0347 0.92 1.06 0.701

IRSD –0.019 0.2176 0.64 1.49 0.930

PIP (%) 0.013 0.0118 0.99 1.04 0.263

UER (%) 0.071 0.0383 1.00 1.16 0.065



3.3.3.2 Suicide and socio-environmental determinants by age and sex

Socio-environmental determinants and suicide among young age group

56

The associations between socio-environmental factors and young male suicide (aged

below 20-years old) are shown in Table 3-8. Tmax was significantly and positively

associated with suicide (RR = 1.20, 95% CI: 1.08 to 1.33). There was no significant

association for RF (RR = 1.01, 95% CI: 0.92 to 1.10), IRSD (RR = 0.76, 95% CI:

0.46 to 1.28) and PIP (RR = 1.00, 95% CI: 0.97 to 1.03).

Table 3-8: Poisson regression of socio-environmental determinants and young male suicides

Parameter ß SE 95% CI

P-value Lower Upper

RF (m) 0.007 0.0459 0.92 1.10 0.885

Tmax (°C) 0.181 0.0535 1.08 1.33 0.001

IRSD –0.270 0.2618 0.46 1.28 0.303

PIP (%) 0.004 0.0150 0.97 1.03 0.808


Disadvantage); PIP (proportion of Indigenous population).

Table 3-9 indicates that there was a statistically significant and positive association

between RF and young female suicide (aged below 20 years old) (RR = 1.15, 95% CI:

1.03 to 1.28). Tmax was statistically and positively associated with suicide (RR = 1.23,

95% CI: 1.02 to 1.49). No significant association was observed for IRSD (RR = 0.67,

95% CI: 0.29 to 1.55) and PIP (RR = 0.99, 95% CI: 0.95 to 1.04).

Table 3-9: Poisson regression of socio-environmental determinants and young female suicides

Parameter ß SE

95% CI P-value

Lower Upper

RF (m) 0.141 0.0557 1.03 1.28 0.012

Tmax (°C) 0.206 0.0970 1.02 1.49 0.034

IRSD –0.404 0.4287 0.29 1.55 0.346

PIP (%) –0.006 0.0230 0.95 1.04 0.790



Socio-environmental determinants and suicide among middle age group

Table 3-10 shows that suicide among middle age male group (20-59 years) was

statistically and positively associated with Tmax (RR = 1.06, 95% CI: 1.01 to 1.11),

57

PIP (RR = 1.02, 95% CI: 1.00 to 1.03) and UER (RR = 1.04, 95% CI: 1.01 to 1.07).

However, there was no significant association for RF (RR = 1.02, 95% CI: 0.99 to

1.05) and IRSD (RR = 0.90, 95% CI: 0.73 to 1.12).

Table 3-10: Poisson regression of socio-environmental determinants and middle age male suicides

Parameter ß SE

95% CI P-value

Lower Upper

RF (m) 0.023 0.0150 0.99 1.05 0.134

Tmax (°C) 0.056 0.0245 1.01 1.11 0.021

IRSD –0.101 0.1071 0.73 1.12 0.345

PIP (%) 0.017 0.0064 1.00 1.03 0.007

UER (%) 0.037 0.0135 1.01 1.07 0.007



Table 3-11 reveals that UER had a significantly positive association with suicide (RR

= 1.16, 95% CI: 1.05 to 1.28) among middle aged female groups (20–59 years). The

association between PIP and suicide was marginally positive (RR = 1.03, 95% CI:

1.00 to 1.06). There was no significant association for RF (RR = 1.02, 95% CI: 0.96

to 1.09), Tmax (RR = 1.01, 95% CI: 0.91 to 1.13) and IRSD (RR = 1.36, 95% CI: 0.77

to 2.41).

Table 3-11: Poisson regression of socio-environmental determinants and middle age female suicides

Parameter ß SE

95% CI P-value

Lower Upper

RF (m) 0.022 0.0329 0.96 1.09 0.499

Tmax (°C) 0.013 0.0563 0.91 1.13 0.812

IRSD 0.307 0.2911 0.77 2.41 0.292

PIP (%) 0.027 0.0155 1.00 1.06 0.080

UER (%) 0.150 0.0493 1.05 1.28 0.002



Socio-environmental determinants and suicide among old age group

Table 3-12 indicates that Tmax had statistically significant and positive association

with suicide (RR = 1.11, 95% CI: 1.02 to 1.20) among old age male group (aged 60

58

years and above). There was no significant association for RF (RR = 1.00, 95% CI:

0.99 to 1.01), IRSD (RR = 0.87, 95% CI: 0.62 to 1.22) and PIP (RR = 0.98, 95% CI:

0.94 to 1.02).

Table 3-12: Poisson regression of socio-environmental determinants and suicides among male elderly


P-value Lower Upper

RF (m) 0.000 0.0032 0.99 1.01 0.858

Tmax (°C) 0.100 0.0424 1.02 1.20 0.018

IRSD –0.144 0.1740 0.62 1.22 0.408

PIP (%) –0.020 0.0191 0.94 1.02 0.289



Table 3-13 shows that older female suicide (60–year old and above) was not

significantly associated with RF (RR = 1.01, 95% CI: 1.00 to 1.02), Tmax (RR = 0.78,

95% CI: 0.58 to 1.03), IRSD (RR = 0.66, 95% CI: 0.31 to 1.42) and PIP (RR = 1.01,

95% CI: 0.95 to 1.07).

Table 3-13: Poisson regression of socio-environmental determinants and suicides among female elderly

Parameter ß SE

95% CI P-value

Lower Upper

RF (m) 0.009 0.0070 1.00 1.02 0.216

Tmax (°C) –0.254 0.1443 0.58 1.03 0.079

IRSD –0.412 0.3881 0.31 1.42 0.289

PIP (%) 0.008 0.0302 0.95 1.07 0.803



3.4 DISCUSSION

This study examined the relationship between socio-environmental factors and suicide

using a spatial analysis approach. A range of climate, socioeconomic and

demographical factors were included in this quantitative analysis.

3.4.1 Major findings

59

The results of this study show that north, west, some of central areas had higher male

suicide ASM than other areas; while southwest and some of central areas had no male

suicide record. North Queensland and some LGAs in coastal and central areas had

higher female suicide ASM, but almost half of LGAs had no female suicide record.

Climate, SEIFA and demographic factors influenced suicide differently in different

gender and age groups. Maximum temperature was consistently and positively

associated with all male suicide in different age groups, and was also positively

associated with female suicide in young age group at a LGA level. Unemployment

rate and the proportion of Indigenous population were significantly and positively

associated with suicide. Rainfall was only positively associated with suicide among

young females. However, there was no consistent pattern for SEIFA.

3.4.2 Comparison with other studies

Three studies have shown that maximum temperature or mean temperature was

positively associated with suicide (Ajdacic-Gross et al, 2007; Deisenhammer et al,

2003; Lee et al, 2006), which is consistent with the findings of this study. One

explanation for this trend is that higher temperature caused low availability of

tryptophan, one of the 20 standard amino acids. Then less volume of 5-

Hydroxyindoleacetic acid (5-HIAA) are synthesized from tryptophan (Maes 1994 &

1995). As 5-HIAA is vital in reducing depression among humans (Bell et al, 2001),

the lower level of 5-HIAA indirectly caused by high temperature may result in more

depression and other mental health problems among population, including suicidal

behaviours.

60

Unemployment can lead to financial burden for individuals and families, thus result in

anxiety and stress, and thus increase risk of mental health problems and even suicide

behaviour. In this study, unemployment rates in both males and females, particularly

middle age group, were significantly associated with suicide. Middle-aged people are

the main workforce Queensland, so they have more responsibility in providing

economic and financial support for their families. Thus they are more likely to suffer

depression if they become unemployed or have reduced income. Other studies also

found that higher unemployment rate was often accompanied with more suicide

deaths and suicide attempts (Inoue et al, 2007; Iverson et al, 1987; Morrell et al, 1998).

In general, there was a positive association between the proportion of Indigenous

population and suicide, especially among middle-aged males and females. The

Indigenous population is affected by the rapid social change in Australia, with

emergence of unhealthy behaviours, such as access to alcohol, drug abuse, violence

and even damage of family structure (Hunter and Milroy, 2006). While the SES in

Aboriginal and Torres Strait islander areas is still lower than other areas in

Queensland, suicide in these areas is likely to remain high (Hunter and Milroy, 2006).

Another study in Canada also supports the results of this study (Leenaars, 2006).

A few previous studies have shown that rainfall was negatively associated with

suicide (Nicholls et al, 2006; Preti and Miotto, 1998). The persistent deficiency of rain

often resulted in drought, which added to the financial burden of farmers, drove rural

population to leave for urban areas and led to a reduction of social contact and

degeneration of social support system in rural areas (Nicholls et al, 2006). These may

61

contribute to anxiety and other mental problems, which resulted in increased suicide.

As this study only used 5-year aggregated data, the lag effect of rainfall variation on

suicide was not examined. The impact of rainfall deficiency or drought on suicide

should be assessed using long-term detailed data (e.g., more than a decade).

SEIFA is a general index to reflect socioeconomic status (SES) in an area, and it

includes information about factors of economic development, income, employment,

education and occupation. The correlations between SEIFA and demographic

variables like proportion of population with low individual income and education, and

unemployment rate were significant (P < 0.01). High SES areas can also have higher

employment rates, increased income and more accesses to training and education,

compared with low SES areas. Therefore, there are usually high suicide rates in low

SES areas. Previous studies have indicated that higher SES areas usually have lower

suicide mortality, especially in a long study period (e.g., over 30 years), where suicide

prevention strategies were implemented and their effects emerged over such a period

(Page et al, 2006; Taylor et al, 2005). However, no significant association between

SES and suicide was observed in this study, which may be due to the relatively short

time series data and the use of average scores of SEIFA index at a LGA level.

3.4.3 Strengths and limitations

This study has three strengths. Firstly, it is the first study to examine the impact of

different socio-environmental factors (climate, SEIFA and demographical variables)

on suicide in Queensland at a LGA level. Secondly, spatial analysis was used to

compare suicide (suicide mortality and ASM by gender) patterns across all LGAs in

62

relation to socio-environmental variations. Thirdly, this study applied different

quantitative analytical methods (e.g., Spearman correlation and Poisson regression) to

assess how a wide range of socio-environmental factors influence suicide mortality.

This study also has several limitations. Firstly, the information of suicide methods

(WHO, 2007) was not included in the database. In some circumstances, it is difficult

to judge whether a person died due to intentional self-harm or by accident (e.g., a fall).

Secondly, climate condition varies in different areas within each LGA (e.g., the

rainfall in Cook Shire ranged from 107 mm to 336 mm (mean value 233 mm)) but it

is difficult to determine the actual climate condition in the geographical spot of each

suicide death, particularly if a LGA has large area. Thirdly, time series data on suicide

is relatively short (i.e., 5 years), so it is difficult to associate suicide trends with socio-

environmental changes. Finally, alcohol consumption and use of antidepressants also

influence suicide, but these issues were not addressed due to lack of relevant data at

the LGA level (Kalmar et al, 2008; Landberg, 2008).

3.4.4 Future research directions

To develop effective suicide prevention strategies, the impact of socio-environmental

factors on suicide should be considered. As there is a growing concern about the

effects of socio-environmental change on population health, it is important to

strengthen the surveillance systems to monitor extreme weather events such as floods,

droughts and cyclones, and to assess their impacts on local socio-economic status and

mental health (Abaurrea and Cebrián, 2002; Diaz, 2007; Hoffpauir, 2008).

Governmental officials, epidemiologists, psychiatrists, environmental health workers,

63

economists, meteorologists and community leaders should work together to design,

develop and implement effective suicide prevention and control strategies through an

integrated and systematic approach.

64

CHAPTER 4. PRELIMINARY SPATIOTEMPORAL ANALYSIS

OF THE ASSOCIATION BETWEEN SOCIO–ENVIRONMENTAL

FACTORS AND SUICIDE

SUMMARY

Background: The seasonality of suicide has long been recognised. However, little is

known about the relative importance of socio-environmental factors in the occurrence

of suicide in different geographical areas. This study examined the association of

climate, socioeconomic and demographic factors with suicide, using a preliminary

spatiotemporal approach.

Method: Seasonal data on suicide, demographic variables (including unemployment

rate, indigenous population, population with low income and low education) and

socioeconomic indexes for areas (SEIFA) between 1999 and 2003 were acquired from

Australian Bureau of Statistics. Climate data, including rainfall, maximum and

minimum temperature, were supplied by Australian Bureau of Meteorology. A

multivariable generalized estimating equations (GEE) model was used to examine the

impact of socio-environmental factors on suicide after adjustment for a range of

confounding factors.

Results: The preliminary data analyses show that far north Queensland had the

highest suicide incidence (e.g., Cook and Mornington Shires), while the south-western

areas had the lowest incidence (e.g., Barcoo and Bauhinia Shires) in most of the

seasons, especially among males. Maximum temperature, unemployment rate, the

proportion of Indigenous population and the proportion of population with low

individual income were positively associated with suicide in Queensland. Rainfall had

65

a positive association with suicide among the total population but was not

significantly associated with suicide by gender. There were weaker associations for

other variables.

Conclusion: Maximum temperature, the proportion of Indigenous population and

unemployment rate appear to be major determinants of suicide at a LGA level in

Queensland. However, the application of these research findings in the design and

implementation of suicide prevention programs need to be further examined.

4.1 INTRODUCTION

Suicide is one of the major causes of mortality around the world with about 877,000

suicide deaths each year globally (WHO 2003 & 2002). Socio-environmental impacts

on mental health, including suicide, have drawn increasing research attention,

especially in recent years as global socio-environmental conditions change rapidly

(Kefi et al, 2008; McCoy 2007).

A number of studies have examined the impact of meteorological factors on suicide.

For example, there was a negative association between rainfall and suicide (Preti and

Miotto, 1998); temperature had a positive association with suicide (Lin et al, 2008);

lowered humidity was accompanied with more suicide (Deisenhammer, et al, 2003);

and sunshine was reported having a negative association with suicide (Linkowski et al,

1992). Additionally, some studies indicated that suicide rates varied in different

seasons (Burns et al, 2008; Rocchi et al, 2007).

66

Most of the previous studies did not examine the relationship between different sets of

socio-environmental factors and suicide. As all these factors can influence suicide in

different aspects, the impact of these factors on suicide, thus, should be studied in a

systematic way, to help formulating effective suicide prevention strategies. In addition,

few of the previous studies have applied geographical information system (GIS) and

spatial analysis approaches to assess the socio-environmental impact on suicide

between different areas. Spatiotemporal analysis of the impact of socio-environmental

factors on suicide is critical because the distribution of suicide deaths and its

determinants may vary with time and place, especially in Queensland, the largest

decentralised state in Australia.

4.2 METHODS

4.2.1 Data sources

The monthly meteorological data, including monthly rainfall (i.e., RF), maximum

temperature (i.e., Tmax) and minimum temperature (i.e., Tmin) in this study were

supplied by Australian Bureau of Meteorology (BOM). Suicide, socioeconomic and

demographic data were obtained from Australian Bureau of Statistics (ABS). This

study included 2,445 suicide deaths from 1999 to 2003, with 1957 males (80.0%) and

488 females (20.0%). Socio-economic Indexes for Area (SEIFA) at a LGA level,

including the Index of Relative Socio-economic Advantage and Disadvantage (i.e.,

IRSAD), the Index of Relative Socio-economic Disadvantage (i.e., IRSD), the Index

of Economic Resources (i.e., IER) and the Index of Education and Occupation (i.e.,

IEO), were obtained from ABS, as well as demographic variables in total and by

67

gender at a LGA level, including proportion of Indigenous population (i.e., PIP),

unemployment rate (i.e., UER), proportion of population with low individual income

(i.e., PPLII) and proportion of population with low education level (i.e., PPLEL).

SEIFA and demographic data at LGA level were based on 2001 Population Census

Data. The means of seasonal meteorological data at LGA level were calculated from

monthly data (September, October and November for spring; December, January and

February for summer; March, April and May for autumn; and June, July and August

for winter). Average suicide counts in total and by gender were calculated for each

season at the LGA level. We directly applied SEIFA and demographic data acquired

from ABS, as only one census data (i.e., CDATA 2001) was available during the

study period.

4.2.2 Data analysis

A series of GIS and statistical methods were used to analyse data. MapInfo, a GIS tool

(including Vertical Mapper incorporated) was used to perform data link, data transfer

and to explore the spatial patterns of socio-environmental variables and suicide.

Statistical analyses, including univariable, bivariable, and multivariable approaches,

were implemented to examine the relationship between major socio-environmental

determinants and suicide. Univariable analysis was applied to describe characteristics

of each variable. Pearson correlations were applied for bivariable analysis after some

non-normally-distributed data (e.g., RF and PIP) were transformed into approximately

normally-distributed values. Finally multivariable generalized estimating equations

(GEE) regression models were developed to assess the possible impact of socio-

68

environmental factors on suicide, after adjustment for the effects of potential

confounders. This model is well suited to analyse the repeated longitudinal data (e.g.,

suicide and climate data) and it fitted the data better than other models, especially in

binary data or counts (Hanley et al, 2003). This approach was also applied in some

studies (Brooker et al, 2002; Chen et al, 2006). Statistical Package for the Social

Sciences (SPSS) software was used for data management and analysis.

4.3 RESULTS

4.3.1 Univariable analysis

Table 4-1 shows the characteristics of each variable, and it suggests that some

variables (e.g., suicide mortality, RF and PIP) were highly skewed. Tables 4-2

demonstrate the distribution of suicide in Queensland between 1999 and 2003 by

month. The results show a ―constant bottom‖ in April (186), May (182) and June (177)

in all the 5-year data, with monthly peaks in August (238) and October (237). Table 4-

3 shows the seasonal distribution of suicide. Summer had the highest number of

suicide deaths (639) while autumn had the lowest (575).

Figure 4-1a to 4-1d demonstrated the spatial patterns of age-adjusted standard

mortality (ASM) of male suicide in Queensland between different seasons. In spring,

some of far north, northwest, south, some of southeast and central coast areas had

higher suicide ASM, while the inland and south western areas had lower suicide ASM

or no suicide record (Figure 4-1a). During the summer time, far north, some of north

west, southeast and coastal and some of the south areas had higher suicide ASM;

69

southwest and some of the central south areas had lower suicide ASM or no suicide

record (Figure 4-1b). Figure 4-1c indicates the suicide ASM distribution in autumn.

Far north, west, central, some of the coastal and southeast areas had higher suicide

ASM; south, southwest and some of the central areas had lower suicide ASM or no

suicide record. In winter, some of far north areas, northwest, some of the southeast

and east areas had higher suicide mortality rate, while central, southwest and other

areas had lower suicide mortality rate or even no suicide record (Figure 4-1d).

The spatial patterns of ASM of female suicide across seasons were indicated in Figure

4-2a to 4-2d. In spring, far north, north and central coast and some inland areas in the

east and south had higher suicide ASM (Figure 4-2a). There was higher suicide ASM

in far north, some of north-western areas, some coastal and inland areas in the north

and southeast than other areas (Figure 4-2b). In autumn, some areas of central inland

and coast, and southeast had higher suicide ASM compared with lower suicide ASM

or no suicide record in other areas (Figure 4-2c). There was higher suicide ASM in

some parts of far north, central inland and south-eastern areas than other areas (Figure

4-2d). 68.8% to 76.8% of LGAs had no female suicide record within each seasons.

70

Table 4-1: Frequency of mortality, SEIFA and demographic variables (N = 2500, 125 LGAs*4

Seasons* 5 years)

Mean

Std.

Deviation Minimum

Percentiles

Maximum 25 50 75

Total mortality (per 100,000) 4.28 15.038 0.00 0.00 0.00 3.17 211.64

Male ASM rate (per 100,000) 6.96 25.872 0.00 0.00 0.00 3.94 410.68

Female ASM rate (per 100,000) 1.34 10.053 0.00 0.00 0.00 0.00 225.73

RF (mm) 195.0 201.53 0.1 76.5 143.8 237.8 1865.4

Tmin (°C) 15.1 5.08 2.2 12.1 15.6 18.7 26.0

Tmax (°C) 28.0 4.36 16.7 25.0 28.2 30.9 39.1

IRSAD 935.61 41.476 831.36 910.32 930.64 962.72 1059.84

IRSD 957.71 69.305 472.08 946.32 972.48 992.40 1048.88

IER 942.25 52.481 835.52 903.68 939.36 975.44 1083.76

IEO 929.05 35.566 815.68 909.60 925.84 945.52 1064.32

Total PIP (%) 7.79 14.392 0.00 1.92 2.86 6.15 87.51

Male PIP (%) 7.43 13.938 0.00 1.88 2.98 5.73 86.65

Female PIP (%) 8.23 14.987 0.00 1.89 3.14 6.61 88.67

Total UER (%) 6.76 3.819 0.00 4.03 6.04 8.76 23.25

Male UER (%) 7.14 4.526 0.00 4.06 6.45 9.36 26.71

Female UER (%) 6.25 3.110 0.00 4.16 6.11 7.93 18.37

Total PPLII (%) 28.14 7.472 10.65 24.39 27.91 31.77 61.71

Male PPLII (%) 23.06 8.937 4.30 17.66 22.44 27.44 62.22

Female PPLII (%) 34.19 6.611 17.13 30.51 33.63 37.86 65.18

Total PPLEL (%) 22.75 5.468 11.91 19.21 22.94 26.10 47.41

Male PPLEL (%) 24.61 6.234 11.21 20.00 25.25 28.53 50.83

Female PPLEL (%) 20.64 4.992 9.01 17.25 20.70 23.27 43.91

Note: ASM (age-standardised mortality); RF (rainfall); Tmin (minimum temperature); Tmax (maximum temperature);

IRSAD (Index of Relative Socio-economic Advantage and Disadvantage); IRSD (Index of Relative Socio-

economic Advantage and Disadvantage); IER (Index of Economic Resources); IEO (Index of Education and

Occupation); PIP (proportion of Indigenous population); UER (unemployment rate); PPLII (proportion of


71

Table 4-2: Suicide distribution by month (1999-2003)

Table 4-3: Suicide distribution by season (1999-2003)

Male Female Total

Month 1999 2000 2001 2002 2003 All

males 1999 2000 2001 2002 2003

All

females 1999 2000 2001 2002 2003

All

suicides

Jan 22 37 43 43 37 182 7 12 7 11 10 47 29 49 50 54 47 229

Feb 32 43 28 31 30 164 7 7 6 9 9 38 39 50 34 40 39 202

Mar 27 42 36 41 24 170 3 8 15 8 3 37 30 50 51 49 27 207

Apr 30 28 36 33 21 148 6 9 7 6 10 38 36 37 43 39 31 186

May 25 37 33 31 11 137 9 14 9 8 5 45 34 51 42 39 16 182

Jun 24 37 31 32 17 141 3 10 8 10 5 36 27 47 39 42 22 177

Jul 38 26 31 38 26 159 12 9 4 7 11 43 50 35 35 45 37 202

Aug 47 34 33 32 44 190 10 9 13 10 6 48 57 43 46 42 50 238

Sep 31 29 29 29 35 153 5 11 7 7 8 38 36 40 36 36 43 191

Oct 40 51 27 36 38 192 10 10 11 8 6 45 50 61 38 44 44 237

Nov 30 36 24 34 31 155 7 7 2 8 7 31 37 43 26 42 38 186

Dec 32 44 38 41 11 166 10 9 12 8 3 42 42 53 50 49 14 208

Total 378 444 389 421 325 1957 89 115 101 100 83 488 457 559 490 521 408 2445

Male Female Total

Season 1999 2000 2001 2002 2003 All

males 1999 2000 2001 2002 2003

All

females 1999 2000 2001 2002 2003

All

suicides

Spring 101 116 80 99 104 500 22 28 20 23 21 114 123 144 100 122 125 614

Summer 86 124 109 115 78 512 24 28 25 28 22 127 110 152 134 143 100 639

Autumn 82 107 105 105 56 455 18 31 31 22 18 120 100 138 136 127 74 575

Winter 109 97 95 102 87 490 25 28 25 27 22 127 134 125 120 129 109 617

Total 378 444 389 421 325 1957 89 115 101 100 83 488 457 559 490 521 408 2445

72

Figure 4-1a: Average male ASM in spring (1999-2003)

73

Figure 4-1b: Average male ASM in summer (1999-2003)

74

Figure 4-1c: Average male ASM in autumn (1999-2003)

75

Figure 4-1d: Average male ASM in winter (1999-2003)

76

Figure 4-2a: Average female ASM in spring (1999-2003)

77

Figure 4-2b: Average female ASM in summer (1999-2003)

78

Figure 4-2c: Average female ASM in autumn (1999-2003)

79

Figure 4-2d: Average female ASM in winter (1999-2003)

80

4.3.2 Bivariable analysis

Table 4-4 demonstrates the correlations between socio-environmental variables and

total suicide mortality. RF was not significantly associated with mortality (r = 0.031,

P = 0.125). There was a weak and positive association between Tmin (r = 0.077, P <

0.01) and Tmax (r = 0.076, P < 0.01) and suicide. SEIFA had a significant and negative

association with suicide (IRSAD: r = –0.149, P < 0.01; IRSD: r = –0.293, P < 0.01;

IER: r = –0.095, P < 0.01; IEO: r = –0.139, P < 0.01). PIP (r = 0.278, P < 0.01), PPLII

(r = 0.231, P < 0.01) and PPLEL (r = 0.151, P < 0.01) were significantly and

positively correlated with suicide. UER (r = 0.004, P = 0.842) was not significantly

associated with suicide mortality.

Table 4-5 shows the correlations between male suicide mortality and socio-

environmental variables. Tmin (r = 0.073, P < 0.01) and Tmax (r = 0.076, P < 0.01) were

significantly and positively associated with suicide mortality. There were significantly

negative associations between the SEIFA indexes (IRSAD: r = –0.147, P < 0.01;

IRSD: r = –0.296, P < 0.01; IER: r = –0.091, P < 0.01; IEO: r = –0.141 P < 0.01) and

suicide mortality. PIP (r = 0.163, P < 0.01), PPLII (r = 0.232, P < 0.01) and PPLEL (r

= 0.153, P < 0.01) had significant and positive correlation with suicide. UER (r =

0.001, P = 0.967) and RF (r = 0.017, P = 0.401) were not significantly associated with

suicide mortality.

Table 4-6 indicates the correlations between socio-environmental variables and

female suicide mortality. RF was significantly and positively associated with suicide

(r = 0.053, P < 0.01). Most of SEIFA variables had significant and negative

81

association with suicide (IRSAD: r = –0.061, P < 0.01; IRSD: r = –0.094, P < 0.01;

IER: r = –0.055, P < 0.01). PIP (r = 0.069, P < 0.01), PPLII (r = 0.059, P <0 .01) and

PPLEL (r = 0.061, P < 0.01) were significantly and positively associated with suicide.

UER (r = 0.037, P = 0.062), Tmin (r = 0.034, P = 0.088) and IEO (r = –0.034, P =

0.087) were marginally associated with suicide. However, Tmax (r = 0.013, P = 0.502)

had no significant correlation with suicide.

In the assessment of multicollinearity between socio-environmental variables, we

found that some SEIFA indexes (e.g., IRSAD and IER) were highly correlated (r =

0.90). Thus, IRSD was used to represent SEIFA in this study because of its strongest

association with suicide across four SEIFA indexes. In addition, Tmax and Tmin were

also highly correlated (r = 0.826, p < 0.01), and therefore, we use Tmax and Tmin in

separate models.

82

Table 4-4: Pearson correlations (seasonal data, all suicides)

Mortality RF Tmin Tmax IRSAD IRSD IER IEO PIP UER PPLII

RF r 0.031 1.000

P value 0.125

Tmin r 0.077** 0.449** 1.000

P value 0.000 0.000

Tmax r 0.076** 0.163** 0.848** 1.000

P value 0.000 0.000 0.000

IRSAD r –0.149** –0.007 0.029 0.017 1.000

P value 0.000 0.734 0.145 0.399

IRSD r –0.293** –0.125** –0.233** –0.196** 0.606** 1.000

P value 0.000 0.000 0.000 0.000 0.000

IER r –0.095** –0.022 0.104** 0.108** 0.901** 0.361** 1.000

P value 0.000 0.281 0.000 0.000 0.000 0.000

IEO r –0.139** 0.090** –0.053** –0.140** 0.772** 0.628** 0.445** 1.000

P value 0.000 0.000 0.008 0.000 0.000 0.000 0.000

PIP r 0.278** 0.134** 0.294** 0.283** –0.280** –0.865** –0.093** –0.341** 1.000

P value 0.000 0.000 0.000 0.000 0.000 0.000 0.000 0.000 .

UER r 0.004 0.133** –0.019 –0.221** –0.387** –0.150** –0.434** –0.067** –0.137** 1.000

P value 0.842 0.000 0.337 0.000 0.000 0.000 0.000 0.001 0.000

PPLII r 0.231** 0.120** –0.019 –0.122** –0.677** –0.630** –0.625** –0.428** 0.373** 0.493** 1.000

P value 0.000 0.000 0.338 0.000 0.000 0.000 0.000 0.000 0.000 0.000

PPLEL r 0.151** –0.071** –0.003 0.114** –0.797** –0.615** –0.699** –0.672** 0.422** 0.068** 0.557**

P value 0.000 0.000 0.873 0.000 0.000 0.000 0.000 0.000 0.000 0.001 0.000

*. Significant at the 0.05 level (2-tailed)

**. Significant at the 0.01 level (2-tailed) Note: RF (rainfall); Tmin (minimum temperature); Tmax (maximum temperature); IRSAD (Index of Relative Socio-economic Advantage and Disadvantage); IRSD (Index of Relative Socio-

economic Advantage and Disadvantage); IER (Index of Economic Resources); IEO (Index of Education and Occupation); PIP (proportion of Indigenous population); UER (unemployment rate);

PPLII (proportion of population with low individual income); PPLEL (proportion of population with low educational level)

83

Table 4-5: Pearson correlations (seasonal data, male)


RF r 0.017 1.000

P value 0.401

Tmin r 0.073** 0.449** 1.000

P value 0.000 0.000

Tmax r 0.076** 0.163** 0.848** 1.000

P value 0.000 0.000 0.000

IRSAD r –0.147** –0.007 0.029 0.017 1.000

P value 0.000 0.734 0.145 0.399

IRSD r –0.296** –0.125** –0.233** –0.196** 0.606** 1.000

P value 0.000 0.000 0.000 0.000 0.000

IER r –0.091** –0.022 0.104** 0.108** 0.901** 0.361** 1.000

P value 0.000 0.281 0.000 0.000 0.000 0.000

IEO r –0.141** 0.090** –0.053** –0.140** 0.772** 0.628** 0.445** 1.000

P value 0.000 0.000 0.008 0.000 0.000 0.000 0.000

PIP r 0.163** 0.091** 0.257** 0.280** –0.207** –0.603** –0.047* –0.277** 1.000

P value 0.000 0.000 0.000 0.000 0.000 0.000 0.019 0.000 .

UER r 0.001 0.133** –0.019 –0.221** –0.387** –0.150** –0.434** –0.067** –0.066** 1.000

P value 0.967 0.000 0.337 0.000 0.000 0.000 0.000 0.001 0.001

PPLII r 0.232** 0.120** –0.019 –0.122** –0.677** –0.630** –0.625** –0.428** 0.132** 0.493** 1.000

P value 0.000 0.000 0.338 0.000 0.000 0.000 0.000 0.000 0.000 0.000

PPLEL r 0.153** –0.071** –0.003 0.114** –0.797** –0.615** –0.699** –0.672** 0.357** 0.068** 0.557**

P value 0.000 0.000 0.873 0.000 0.000 0.000 0.000 0.000 0.000 0.001 0.000


**. Significant at the 0.01 level (2-tailed)




84

Table 4-6: Pearson correlations (seasonal data, female)


RF r 0.053** 1.000

P value 0.008

Tmin r 0.034 0.449** 1.000

P value 0.088 0.000

Tmax r 0.013 0.163** 0.848** 1.000

P value 0.502 0.000 0.000

IRSAD r –0.061** –0.007 0.029 0.017 1.000

P value 0.002 0.734 0.145 0.399

IRSD r –0.094** –0.125** –0.233** –0.196** 0.606** 1.000

P value 0.000 0.000 0.000 0.000 0.000

IER r –0.055** –0.022 0.104** 0.108** 0.901** 0.361** 1.000

P value 0.006 0.281 0.000 0.000 0.000 0.000

IEO r –0.034 0.090** –0.053** –0.140** 0.772** 0.628** 0.445** 1.000

P value 0.087 0.000 0.008 0.000 0.000 0.000 0.000

PIP r 0.069** 0.075** 0.250** 0.282** –0.189** –0.593** –0.015 –0.288** 1.000

P value 0.001 0.000 0.000 0.000 0.000 0.000 0.443 0.000

UER r 0.037 0.115** –0.018 –0.219** –0.310** –0.155** –0.314** –0.084** –0.107** 1.000

P value 0.062 0.000 0.370 0.000 0.000 0.000 0.000 0.000 0.000 .

PPLII r 0.059** 0.015 –0.026 –0.067** –0.502** –0.516** –0.338** –0.561** 0.030 0.251** 1.000

P value 0.003 0.440 0.198 0.001 0.000 0.000 0.000 0.000 0.137 0.000

PPLEL r 0.061** –0.012 0.009 0.068** –0.786** –0.657** –0.672** –0.640** 0.376** 0.054** 0.478**

P value 0.002 0.555 0.650 0.001 0.000 0.000 0.000 0.000 0.000 0.007 0.000


**. Significant at the 0.01 level (2-tailed)




85

4.3.3 Multivariable analysis

Multivariable analyses were undertaken to examine the possible impact of climate

variables, SEIFA and demographic variables on suicide mortality rate, using the GEE

model. Table 4-7 shows the associations between socio-environmental variables and

total suicide mortality. RF was significantly and positively associated with suicide

(RR = 1.11, 95% CI: 1.04 to 1.19). Tmax was marginally and positively associated

with suicide mortality (RR = 1.15, 95% CI: 1.00 to 1.32). PIP (RR = 1.16, 95% CI:

1.13 to 1.19), UER (RR = 1.40, 95% CI: 1.24 to 1.59) and PPLII (RR = 1.28, 95% CI:

1.10 to 1.48) were positively associated with suicide. There was no significant

association for IRSD (RR = 1.01, 95% CI: 0.99 to 1.04) and PPLEL (RR = 0.92, 95%

CI: 0.76 to 1.11).

Table 4-7: GEE regression of socio-environmental determinants of suicide (all)


P-value Lower Upper

RF (m) 0.104 0.0342 1.04 1.19 0.003

Tmax (ºC) 0.137 0.0721 1.00 1.32 0.058

IRSD 0.014 0.0137 0.99 1.04 0.319

PIP (%) 0.146 0.0133 1.13 1.19 <0.001 UER (%) 0.340 0.0634 1.24 1.59 <0.001 PPLII (%) 0.244 0.0743 1.10 1.48 0.001

PPLEL (%) –0.084 0.0979 0.76 1.11 0.390


Disadvantage); PIP (proportion of Indigenous population); UER (unemployment rate); PPLII (proportion of


Table 4-8 shows that Tmax (RR = 1.24, 95% CI: 1.04 to 1.47), PIP (RR = 1.07, 95% CI:

1.01 to 1.13) and PPLII (RR = 1.45, 95% CI: 1.23 to 1.72) were significantly and

positively associated with suicide in male population. RF (RR = 1.09, 95% CI: 0.95

to 1.26), IRSD (RR = 1.02, 95% CI: 0.94 to 1.09), UER (RR = 1.08, 95% CI: 0.89 to

1.32) and PPLEL (RR = 0.87, 95% CI: 0.69 to 1.09) were not significantly associated

with suicide.

86

Table 4-8: GEE regression of socio-environmental determinants of suicide (male)


P-value Lower Upper

RF (m) 0.090 0.0731 0.95 1.26 0.220

Tmax (ºC) 0.213 0.0880 1.04 1.47 0.016

IRSD –0.017 0.0235 0.94 1.03 0.483

PIP (%) 0.065 0.0299 1.01 1.13 0.029

UER (%) 0.077 0.1007 0.89 1.32 0.446

PPLII (%) 0.373 0.0863 1.23 1.72 <0.001

PPLEL (%) –0.139 0.1170 0.69 1.09 0.234




Table 4-9 reveals that PIP (RR = 1.23, 95% CI: 1.03 to 1.48) and UER were

statistically and significantly associated with female suicide (RR = 1.09, 95% CI: 1.01

to 1.18). There was no significant association for other variables.

Table 4-9: GEE regression of socio-environmental determinants of suicide (female)


P-value Lower Upper

RF (m) 0.114 0.2906 0.63 1.98 0.696

Tmax (ºC) –0.072 0.2484 0.57 1.51 0.773

IRSD 0.052 0.0896 0.88 1.26 0.565

PIP (%) 0.209 0.0938 1.03 1.48 0.026

UER (%) 0.087 0.0415 1.01 1.18 0.036

PPLII (%) –0.176 0.1365 0.64 1.10 0.198

PPLEL (%) 0.168 0.1064 0.96 1.46 0.115




4.4 DISCUSSION

This study examined the relationship between socio-environmental factors and suicide

using GIS and spatiotemporal analysis approaches. A range of climate, socioeconomic

and demographic determinants were included in this quantitative analysis.

4.4.1 Major findings

87

The results in this study indicate some key socio-environmental predictors of suicide

at the LGA level. The preliminary data analyses show that north area had highest

suicide mortality in all the seasons. Central Queensland had highest suicide mortality

in summer, while the south-western areas had no suicide records in all the seasons.

Maximum temperature was positively associated with total and male suicide. Higher

proportion of Indigenous population was accompanied with more suicides in total and

males. Unemployment rate had a positive association with total and female suicide.

Proportion of population with low individual income was significantly associated with

total and male suicide. Rainfall had a significant and positive association with total

suicide only, but not significant with suicide by gender. SEIFA index and proportion

of population with low education level had no significant association with suicide in

total and by gender.

4.4.2 Possible mechanisms and consistence with other studies

Some of the previous studies found that rainfall was negatively associated with

suicide (Nicholls et al, 2006; Preti and Miotto, 1998), which is different from the

results in this study. Usually, persistent rainfall deficiency results in drought, which

causes reduction of crops in rural areas and adds financial burden to local residents,

especially farmers. In rural areas, farmers and other residents usually have less social

support than urban residents, and this situation can get worse due to drought. All these

add stress, anxiety and mental health problems among the rural population which will

eventually lead to suicidal behaviours and even suicide. However, this study only

covered 5-year rainfall and suicide data, so it is difficult to determine the long term

88

effect of rainfall on suicide. Another explanation for this discrepancy is that

Queensland is in tropical and subtropical areas with much rainfall in general,

especially in coastal areas. Excessive rainfall prevents people from outdoor and social

activities. People usually stay in rooms with less social contact and this may cause

anxiety and other mental health problems, including alcohol drinking and suicide

behaviour. The majority of population and suicide deaths were in these areas, so the

actual association between rainfall and suicide may not be obvious, compared with

temperate climate areas.

In this study, higher maximum temperature was accompanied with increased suicide

mortality at a LGA level. This finding corroborates previous reports (Ajdacic-Gross et

al, 2007; Deisenhammer et al, 2003; Vandentorren et al, 2006). For instance, some

studies discovered that higher temperature can lead to decreased availability of

tryptophan in human body, one of the 20 standard amino acids, then the volume of 5-

Hydroxyindoleacetic acid (5-HIAA) synthesized from tryptophan greatly reduced

(Maes 1994 & 1995). As 5-HIAA can reduce depression among humans (Bell et al,

2001), and therefore, the reduced 5-HIAA indirectly caused by high temperature leads

to more depression and other mental health problems among population, even suicidal

behaviours. We also examined the association between minimum temperature and

suicide in the GEE model, but the association was very weak.

This study demonstrates a general trend that LGAs with higher proportion of

indigenous population had higher rates of suicide. As most of the Indigenous

population are located in rural areas, these communities often have lower SES and

less opportunities of healthcare, including mental health services. The rapid social

89

change in Australia may also affect the indigenous communities, with more unhealthy

behaviours such as excessive alcohol use and family violence (Measey et al, 2006).

The above factors contribute to the higher suicide occurrence and mortality rate in

communities with a high proportion of Aboriginal and Torres Strait Islanders in

Queensland (Hunter and Milroy, 2006). Other studies in the United States also

indicate that suicide mortality were higher in the areas with higher proportion of

Indigenous population than in the other areas (Else et al, 2007; Seale et al, 2006;

Wexler et al, 2008).

Increase unemployment rate directly reduces individual and family income, thus can

cause the financial burden and result in anxiety and stress among family members,

especially for a less skilled population. These may increase the risk of mental health

problems and suicidal behaviours. This can explain why unemployment had an

adverse impact on suicide. Previous studies also discovered that higher

unemployment rate can enhance the risk of suicide behaviour and suicide (Chan et al,

2007; Fergusson et al, 2007; Yasan et al, 2008). In this study, unemployment had

more significant on female suicide than male suicide, one reason is that currently

more females participated in paid jobs, while they still need to play the traditional role

of family care, thus caused heavy burden to them and may lead to mental health

problems among females.

Previous studies have indicated that higher socioeconomic status (SES) areas usually

have lower suicide mortality, especially in a long study period (e.g., over 30 years),

where suicide prevention strategies were implemented and their effects emerged over

a period (Page et al, 2006; Taylor et al, 2005). High SES areas can also have higher

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Seale%20JP%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

90


compared with low SES areas, where residents gained less from economic

development (ABS 2005). In this study, we did not find a significant association

between SEIFA and suicide, which may be due to a short time series dataset (5 years)

and the use of a snapshot measure of SEIFA (i.e., IRSD in 2001); and the results are

inconsistent with previous studies.

Some studies indicate that population with low income had higher suicide rate

(Huisman and Oldehinkel, 2008; Kalist et al, 2007). Generally, rural areas have higher

proportion of population with low income, while healthcare (including mental health

care) facilities are less developed and less accessible than urban areas. This can lead

to increased mental health problems, even suicidal behaviours, among the local

population. The results of this study are consistent with previous studies.

Some studies are implemented in temperate areas like Brazil (Benedito-Silva, et al,

2007) and Italy (Preti et al, 2007) with obvious peak of suicide in late spring and early

summer. In this study, there were suicide peaks in August (238) and October (237)

between 1999 and 2003. And summer had the most suicides among four seasons. The

results in this study were not completely consistent with previous studies, partly

because all the LGAs of Queensland are in tropical and subtropical zones, and the

four seasons are not evident in many places, especially in the LGAs in the north,

which are close to the equator.

4.4.3 Strengths and limitations

91

This study has several strengths. Firstly, this is the first study to examine an

association between a wide range of socio-environmental factors and suicide at a

LGA level in Queensland. Secondly, this study used GIS and spatiotemporal analysis

approaches to assess the socio-environmental impact on suicide. Thirdly, this study

used GEE models to examine how socio-environmental factors influence the

likelihood of suicide after taking into account a range of confounding factors.

The limitations of this study should also be addressed. Firstly, the time series data set

for analysis is short, compared with other studies (Ajdacic-Gross et al, 2007; Nicholls

et al, 2006; Preti et al, 2007). Secondly, the SEIFA index and demographical data at

the LGA level were only based on 2001 Population Census, so it cannot reflect any

changes in socioeconomic and demographical features during the whole study period.

Thus the results of this study should be interpreted cautiously. Finally, the impact of

alcohol and drug use on suicide incidence is also obvious in some studies (Evren et al,

2008; Wang and Stórá, 2009; Wojnar et al, 2009), but the relevant data are

unavailable at a LGA level in Queensland.

4.4.4 Future research directions

The impact of socio-environmental change on mental health has drawn much

attention. As global change (e.g., climate change and socio-demographic change)

continues, it is vital to strengthen surveillance system on weather extremes (e.g.,

heatwaves) and social changes (e.g., unemployment) as well as the impacts of these

changes on mental health (Abaurrea and Cebrián, 2002; Diaz, 2007; Hoffpauir, 2008).

Governmental officials, epidemiologists, psychiatrists, environmental health workers,

92

economists, meteorologists and community leaders should work together to design,

develop and implement effective suicide prevention and control strategies through an

integrated and systematic approach.

93

CHAPTER 5 DISCUSSION AND CONCLUSIONS

SUMMARY

This chapter summarizes the key findings of this study; explains the mechanisms of

the results; discusses its strengths and limitations and the future research directions,

and also draws conclusions.

5.1 AN OVERVIEW OF KEY FINDINGS IN THIS STUDY

In general, the analytical results from Chapters 3 and 4 are consistent. We examined

the spatiotemporal pattern of suicide and the possible impact of socio-environmental

factors (climate, socioeconomic and demographic variables) on suicide using a range

of GIS and statistical modelling techniques. The distribution of suicide and standard

mortality rates varied widely at a LGA level, and these variations were associated

with a range of socio-environmental factors.

The preliminary data analyses show that north area had highest suicide incidence,

while the south-western areas had no suicide records in all the seasons in general.

Central Queensland had highest suicide mortality in summer. North, west, some of

central areas had higher male suicide ASM than other areas, southwest and some of

central areas had no male suicide record. North Queensland and some LGAs in

coastal and central areas had higher female suicide ASM, but almost half of LGAs

had no female suicide record in yearly dataset. About 68% to 75% of LGAs had no

female suicide occurred in each season.

94

Climate, SEIFA and demographic factors influenced suicide differently in different

gender and age groups at a LGA level. Rainfall had a significant and positive

association with total suicide (seasonal dataset) and young females (yearly dataset).

Maximum temperature was consistently and positively associated with total suicide

and male suicide (all and by different age groups) in both yearly and seasonal dataset,

but was only positively associated with female suicide in young age group (yearly

dataset). There was a positive but very weak association between minimum

temperature and suicide.

The proportion of Indigenous population had a positive association with suicide

across different LGAs, especially in males and middle-aged population. Higher

unemployment rate were accompanied with more suicide over space, especially in the

middle-aged population, the main workforce in Queensland. Proportion of population

with low individual income was significantly associated with total and male suicide

(seasonal dataset). SEIFA index and proportion of population with low education

level had no significant association with suicide in total and by gender in different

scales of dataset.

5.2 BIAS AND COMFOUNDING FACTORS

The potential for information bias cannot be entirely ruled out in this study. For

example, 26 suicides (24 males and 2 females) were not included in the analysis due

to the lack of information on their LGA codes. Another 8 suicides (4 males and 4

females) were also excluded from the analysis due to boundary changes in different

95

years. Population, SEIFA and demographic variables at the LGA level in this study

were only based on the 2001 Population Census, so the information on changes of

these variables over time was not available. Thus the association between socio-

demographic factors and suicide over time could not be reflected. With migration of

population between different areas, it is difficult to determine whether the place

(LGA/SLA) of a suicide death in the dataset is as same as the place he/she usually

lived. However, migration bias is unlikely to be a big problem because the number of

suicide cases in migrants is likely to be very small. Some LGAs (e.g. Brisbane City

Council and Gold Coast City Council) contain a number of SLAs, with apparent

differences in SEIFA and demographical variables between SLAs within each LGA.

The data at the LGA level cannot indicate these differences and may influence the

results of analysis.

Information on some potential confounding factors was unavailable in this study.

These potential confounders included access to local health and psychological

services, and suicide prevention interventions, which can reduce the suicide risk and

prevent suicidal behaviours. Seasonal change and personal information (e.g., drug use,

medication, health status and family mental health history) (Kalmar et al, 2008;

Landberg, 2008) had impact on suicide. Some studies showed that religions can also

reduce the rate of mental illness and suicide ideation (Rasic et al, 2009; Dervic et al,

2004). Status of nutrition (e.g., minerals, vitamins and fatty acid; or food like fish,

vegetables) may also influence symptoms of psychiatric patients (Lakhan and Vieira,

2008; Li et al, 2007), as a balanced diet and enough ingestion of nutrients can benefit

physical health and improve mental health. Mental health and emergency services can

96

reduce suicide risk and suicide deaths (Pirkola et al, 2009). People with family history

of suicide and mental illness had high risk of suicide (Sørensen et al, 2009).

5.3 COMPARISON WITH OTHER STUDIES

Some of the previous studies found that rainfall was negatively associated with

suicide (Nicholls et al, 2006; Preti and Miotto, 1998), which is different from the

results in this study. Usually, persistent rainfall deficiency results in drought, which

causes reduction of crops in rural areas and adds financial burden to local residents,

especially farmers. In rural areas, farmers and other residents usually have less social

support than urban residents, and this situation can get worse due to drought. Drought

also drove rural population to leave for urban areas and led to a reduction of social

contact and degeneration of social support system in rural areas (Nicholls et al, 2006).

All these add stress, anxiety and mental health problems among the rural population

which will eventually lead to suicidal behaviours and even suicide. However, this

study only covered 5-year rainfall and suicide data, the lag effect of rainfall variation

on suicide was not examined, so it is difficult to determine the long term effect of

rainfall on suicide. The impact of rainfall deficiency or drought on suicide should be

assessed using long-term detailed data (e.g., more than a decade). Another

explanation for this discrepancy is that Queensland is in tropical and subtropical areas

with much rainfall in general, especially in coastal areas. Excessive rainfall prevents

people from outdoor and social activities. People usually stay in rooms with less

social contact and this may cause anxiety and other mental health problems, including

alcohol drinking and suicide behaviour. The majority of population and suicide deaths

97

were in these areas, so the actual association between rainfall and suicide may not be

obvious, compared with temperate climate areas.

In this study, higher maximum temperature was accompanied with increased suicide

mortality at a LGA level. This finding corroborates previous reports (Ajdacic-Gross et

al, 2007; Deisenhammer et al, 2003; Lee et al, 2006; Vandentorren et al, 2006). One

explanation for this trend is that higher temperature can lead to low availability of

tryptophan, one of the 20 standard amino acids, then the volume of 5-

Hydroxyindoleacetic acid (5-HIAA) synthesized from tryptophan also greatly reduced

(Maes 1994 & 1995). As 5-HIAA can reduce depression among humans (Bell et al,

2001), and therefore, the reduced 5-HIAA indirectly caused by high temperature leads

to more depression and other mental health problems among population, even suicidal

behaviours. We also examined the association between minimum temperature and

suicide in the GEE model, but the association was very weak.

Some studies are implemented in temperate areas like Brazil (Benedito-Silva, et al,

2007) and Italy (Preti et al, 2007) with obvious peak of suicide in late spring and early

summer. In this study, there were suicide peaks in August (238) and October (237)

between 1999 and 2003. And summer had the most suicides among four seasons. The

results in this study were not completely consistent with previous studies, partly

because all the LGAs of Queensland are in tropical and subtropical zones, and the

four seasons are not evident in many places, especially in the LGAs in the north,

which are close to the equator.

98

In general, there was a general trend that LGAs with higher proportion of Indigenous

population had higher rates of suicide, especially among middle-aged Aboriginals. As

most of the Indigenous population are located in rural areas, these communities often

have lower SES and less opportunities of healthcare, including mental health services.

The rapid social change in Australia may also affect the Indigenous communities,

with more unhealthy behaviours such as excessive alcohol use, violence and even

damage of family structure (Hunter and Milroy, 2006; Measey et al, 2006). The above

factors contribute to the higher suicide occurrence and mortality rate in communities

with a high proportion of Aboriginal and Torres Strait Islanders in Queensland

(Hunter and Milroy, 2006). Other studies in the United States and Canada also

indicate that suicide mortality were higher in the areas with higher proportion of

Indigenous population than in the other areas (Else et al, 2007; Leenaars, 2006; Seale

et al, 2006; Wexler et al, 2008).

Increased unemployment rate directly reduces individual and family income, thus can

cause the financial burden and result in anxiety and stress among family members,

especially for a less skilled population. These may increase the risk of mental health

problems and suicidal behaviours. This can explain why unemployment had an

adverse impact on suicide. Previous studies also discovered that higher

unemployment rate can enhance the risk of suicide behaviour and suicide (Chan et al,

2007; Fergusson et al, 2007; Inoue et al, 2007; Iverson et al, 1987; Morrell et al, 1998;

Yasan et al, 2008). In the study on different age groups, unemployment rates in both

males and females, particularly middle age group, were significantly associated with

suicide. Middle-aged people are the main workforce Queensland, so they have more

responsibility in providing economic and financial support for their families. Thus

99

they are more likely to suffer depression if they become unemployed or have reduced

income. The study on genders showed that unemployment had more significant on

female suicide than male suicide, one reason is that currently more females

participated in paid jobs, while they still need to play the traditional role of family

care, thus caused heavy burden to them and may lead to mental health problems

among females.

Some studies indicate that the population with low income had higher suicide rate

(Huisman and Oldehinkel, 2008; Kalist et al, 2007). Generally, rural areas have higher

proportion of population with low income, while healthcare (including mental health

care) facilities are less developed and less accessible than urban areas. This can lead

to increased mental health problems, even suicidal behaviours, among the local

population. The results of this study are consistent with previous studies.

SEIFA is a general index to reflect socioeconomic status (SES) in an area, and it

includes information about factors of economic development, income, employment,

education and occupation. The correlations between SEIFA and demographic

variables like proportion of population with low individual income and education, and

unemployment rate were significant (P < 0.01). High SES areas can also have lower


compared with low SES areas, where residents gained less from economic

development (ABS 2005). Therefore, there are usually high suicide rates in low SES

areas. Previous studies have indicated that higher SES areas usually have lower

suicide mortality, especially in a long study period (e.g., over 30 years), where suicide

prevention strategies were implemented and their effects emerged over such a period

100

(Page et al, 2006; Taylor et al, 2005). However, we did not find a significant

association between SEIFA and suicide in this study, which may be due to the

relatively short time series dataset (5 years) and the use of average scores of SEIFA

index (i.e., IRSD in 2001); and the results are inconsistent with previous studies.

In general, previous studies have indicated that temperature, proportion of Indigenous

population, unemployment rates and proportion of population with low individual

income are positively associated with suicide. These factors interacted with each other.

The results of our study are broadly consistent with other studies. However, we found

some positive impact of rainfall but other studies reported a significant and negative

association between rainfall and suicide.

5.4 STRENGTHENS AND LIMITATIONS OF THIS STUDY

This study has several strengths. Firstly, it is the first study to examine the possible

impact of a wide range of socio-environmental factors on suicide at a LGA level in

Queensland. Meteorological, socioeconomic and demographic factors were included

in this study. Secondly, this study used a comprehensive spatial dataset and a range of

quantitative analytical methods. For example, it included all the LGAs in Queensland

and only about 1% of all recorded 2,479 suicide deaths were not included in the

analysis due to lack of their LGA information or boundary change. A series of GIS

and quantitative analysis techniques were applied to compare the difference of suicide

mortality over time and space. Thirdly, this study has adjusted for a range of

confounders including gender, age, population size and SEIFA at the LGA level.

Male suicide rates are about four times higher than female suicide crates; and most of

101

suicide deaths occurred between 20 and 59-year of age. These differences can

influence the results of analysis. Finally, the results of this study may have

implications in public health policy making and implementation of suicide prevention

intervention. For example, strengthening surveillance on extreme weather events and

socioeconomic change and assessing the association between socio-environmental

change and mental health are vital, especially in rural areas and regions with low SES.

The limitations of this study also need to be acknowledged. Firstly, some suicides

may not be reported, particular in remote areas, which may affect the accuracy of

suicide mortality. The recorded suicide mortality rates in some LGAs were lower than

the actual suicide mortality rate. Secondly, climate condition varies in different areas

within each LGA, especially those covering large areas. So it is difficult to actually

determine the climate condition in the geographical spot of each suicide death.

Thirdly, ecological fallacy may occur if the association observed in this study is

interpreted at an individual level (Greenland & Morgenstern, 1989), for example,

individual employment status, income and health status. The ecological study cannot

interpret the association between socio-environmental factors and suicide at the

individual level. Fourthly, the time-series data in this study is relatively short

compared with other studies. Thus the suicide trend in a long time (e.g., 20 years or

more) as a LGA level cannot be assessed. Additionally, the data of SEIFA index and

demographic variables in this study are only based on 2001 Population Census, and

the impact of the change in these variables on suicide over time is difficult to be

assessed. For example, it is difficult to assess the association between socioeconomic

status change and suicide in one LGA. Finally, this ecological study only included

several socio-environmental variables (i.e., rainfall, temperature, SEIFA,

102

demographic variables), other factors like personal and family history of mental

health and psychiatrical problems, local health service facilities, nutrition, religion,

alcohol and drug use may also influence mental health status and suicidal behaviours,

but the relevant data are not available at the LGA level.

5.5 DIRECTIONS FOR FUTURE RESEARCH

This study examined the association of socio-environmental factors with suicide by

using a series of GIS and statistical analysis approaches. Future research should pay

more attention in the following areas.

Suicide is a serious mental health problem. Many factors such as family history

(Brodsk, 2008; Bronisch and Lieb, 2008), alcohol and drug use (Evren et al, 2008;

Wang and Stórá, 2009; Wojnar et al, 2009), country of birth (Cohen, 2008; Kposowa

et al, 2008), education and cultural background (Aubert et al, 2004, Hjelmeland et al,

2008) can influence suicide. Thus, it is vital to acquire detailed information on each

suicide case and relevant information among the population in assessing the key

socio-environmental determinants of suicide.

As climate change continues, the frequency, intensity and duration of weather

extremes (e.g., flood, drought and cyclone) are likely to increase in the coming decade.

The potential impact of climate change on mental health has drawn much attention, so

the adverse effect of weather extremes and natural disasters on mental health needs to

be addressed in future research (Abaurrea and Cebrián, 2002; Diaz, 2007; Hoffpauir,

2008).

103

Bayesian spatiotemporal models have widely been used in public health research

(Middleton et al, 2008; Poulos et al, 2008). These models can offer convenient

platforms for incorporating both spatial and temporal information. For example,

Bayesian smoothing can stabilize estimates in spatiotemporal patterns of diseases in

small areas with extremely low population (Bell and Broemeling, 2000; Tassone et al,

2009). The Bayesian conditional autoregressive (CAR) model has been used to

describe geographical variation in a specific disease risk between spatially aggregated

units, such as the administrative divisions of a country (Escaramí et al, 2008; Eksler

and Lassarre, 2008). However, such Bayesian spatiotemporal methods have rarely

been applied to examine the impact of socio-environmental factors on suicide. Based

on our findings, Bayesian model may be suitable to further compare mortality rates

and their socio-environmental determinants across LGAs in future research.

5.6 IMPLICATION FOR PUBLIC HEALTH POLICY AND INTERVENTION

This study found that a range of socio-environmental factors were associated with

suicide over time and space. So it has potential implications in public health

intervention planning and implementation in order to control and prevent suicide.

A spatiotemporal approach at the LGA level should be included in the surveillance of

socio-environmental changes and their impacts on suicide over time and space. The

information acquired from this surveillance will be useful for research, education and

training, as well as public health policy-making.

104

This study discovered that maximum temperature, proportion of Indigenous

population and unemployment can significantly affect suicide mortality. Therefore, in

the LGAs (e.g., some LGAs in the north of Queensland) with warm weather, high

proportion of Indigenous population and/or unemployment rate, concerted efforts

need to be made to control and prevent suicide and other mental health problems. It is

vital to strengthen extreme weather (e.g., high temperature) forecasting and early

warning system. The local suicide prevention and control programs (e.g.,

psychological consultation) should focus on the unemployed population. In the

Indigenous communities, health education and health promotion programmes need to

be strengthened to help the local people to build up and develop healthy behaviours

and reduce suicide risks.

In the remote and rural areas, it is necessary to strengthen facilities of mental health

care and psychological consultation, to control and prevent mental health problems

including suicide. Health education and health promotion are vital in increasing

mental health knowledge among rural population, controlling the harmful behaviours

and reducing the risk of suicide. In the disaster areas, the government, public health

emergency services and other relevant agencies need to provide necessary support to

the local communities, as well as health care, to release the financial burden of

families and mental health problems resulted from financial crisis. Financial aid and

psychological intervention are critical in reducing suicide behaviours among

unemployed people.

5.7 CONCLUSION

105

This study examined the relationship between socio-environmental factors and suicide.

GIS and relevant mapping technologies were implemented in describing the spatial

pattern of suicide at a LGA level. An explanatory spatiotemporal analysis approach

was used to assess the impact of socio-environmental factors on suicide. The study

found that climate variables, socio-economic and demographic variables were

associated with suicide mortality over time and space. These findings may have

public health implications in the control and prevention of suicide and mental health

problems in Queensland.

106

References

Abaurrea J, Cebrián AC (2002). Drought analysis based on a cluster Poisson model:

distribution of the most severe drought. Clim Res.; 22:227–235.

ABS (2000). Demography, Queensland, 1999. Cat. No. 3311.3, ABS, Canberra.

ABS (2003). Information Paper, Census of Population and Housing, Socio-Economic

Indexes for Areas, Australia, 2001. Cat. No. 2039.0, ABS, Canberra.

ABS (2001). Australian Social Trends 2001. Cat. No. 4102.0, ABS, Canberra.



ABS (1997). Australian Demographic Trends 1997. Cat. No. 3102.0, ABS, Canberra.

ABS (1999). Australian Standard Geographical Classification (ASGC) 1999. Cat. No.

1216.0, ABS, Canberra.



ABS (2000). Suicides, Australia, 1921–1998. Cat. No. 3309.0, ABS, Canberra.





ABS (2002). Year Book Australia, 2002. Cat. No. 1301.0, ABS, Canberra.

107



ABS (2003). National Income Expenditure and Product, Australia. Cat. No. 5206.0,

ABS, Canberra.

ABS (2003). Population Estimates by Age and Sex, June 1997 to June 2002.

Cat. No. 3201.0, ABS, Canberra.

ABS (2003). Population Estimates by Age and Sex, Queensland, 2002. Cat. No.

3235.3.55.001, ABS, Canberra.

ABS (2004). Population Estimates by Age and Sex, Queensland, 2003. Cat. No.

3235.3.55.001, ABS, Canberra.

ABS (2004). Population Estimates by Age and Sex, Australian States and Territories,

June 2004. Cat. No. 3201.0, ABS, Canberra.

ABS (2004). Suicide: recent trends, Australia, 1993-2003. Cat. No. 3309.0.55.001,

ABS, Canberra.

ABS (2005). Measures of Australia’s progress: summary indicators, 2005. Cat No.

1383.0.55.001, ABS, Canberra.

ACIA (2004) Impacts of a warming Arctic: Arctic climate impact assessment.

Cambridge: Cambridge University Press.

Agerboa E, Sterneb JAC, Gunnellb DJ (2007). Combining individual and ecological

data to determine compositional and contextual socio-economic risk factors for

suicide. Soc Sci Med.; 64:451–461.

108

Ajdacic-Gross V, Lauber C, Sansossio R, Bopp M, Eich D, Gostynski M, Gutzwiller

F, Rossler W (2007). Seasonal Associations between Weather Conditions and

Suicide—Evidence against a Classic Hypothesis. Am J Epidemiol., 165:561–569.

Andreasyan K, Hoy WE, Kondalsamy-Chennakesavan S (2007). Indigenous mortality

in remote Queensland, Australia. Aust N Z J Public Health; 31:422–427.

Aubert P, Daigle MS, Daigle JG (2004). Cultural traits and immigration: hostility and

suicidally in Chinese Canadian students. Transcult Psychiatry; 41:514–532.

Baume PJ, Clinton ME (1997). Social and cultural pattern of suicide in young people

in Australia. Aust. J. Rural Health; 5:115–120.

Bell BS, Broemeling LD (2000). A bayesian analysis for spatial processes with

application to disease mapping. Statist. Med.; 19:957–974.

Bell C, Abrams J, Nutt D (2001). Tryptophan depletion and its implication for

psychiatry. Br J Psychiatry; 178:399–405.

Benedito-Silva AB; Nogueira Pires ML, Calil HM. (2007). Seasonal Variation of

Suicide in Brazil. Chronobiol Int.; 24:727–737.

BOM. (2008).Time series - Australian Climate Variability and Change, see:

http://www.bom.gov.au/cgi-bin/silo/reg/cli_chg/timeseries.cgi, Accessed 30 March

2008

BOM (2008). Australian Climate Change and Variability.

http://www.bom.gov.au/silo/products/cli_chg/ (Accessed 20 April 2008)

BOM (2003). Climate Activities in Australia, 2003. Australian Academy of Science,

Australian Bureau of Meteorology and Australian Greenhouse Office.

Brodsky BS, Mann JJ, Stanley B, Tin A, Oquendo M, Birmaher B, Greenhill L,

Kolko D, Zelazny J, Burke AK, Melhem NM, Brent D (2008). Familial transmission

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Aubert%20P%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Daigle%20MS%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Daigle%20JG%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.bom.gov.au/cgi-bin/silo/reg/cli_chg/timeseries.cgi

http://www.bom.gov.au/silo/products/cli_chg/

109

of suicidal behavior: factors mediating the relationship between childhood abuse and

offspring suicide attempts. J Clin Psychiatry; 69:584–596.

Bronisch T, Lieb R (2008). Maternal suicidality and suicide risk in offspring.

Psychiatr Clin North Am.; 31:213–221.

Brooker S, Beasley M, Ndinaromtan M, Madjiouroum EM, Baboguel M,

Djenguinabe E, Hay SI, Bundy DA (2002). Use of remote sensing and a geographical

information system in a national helminth control programme in Chad. Bull World

Health Organ.; 2002; 80:783–789.

Burns A, Goodall E, Moore T (2008). A study of suicides in Londonderry, Northern

Ireland, for the year period spanning 2000–2005. J Forensic Leg Med.; 15:148–157.

Burvill PW (1998). Migrant suicide rates in Australia and in country of birth. Psychol

Med.; 28:201–208.

Caldwell TM, Jorm AF, Dear KB (2004). Suicide and mental health in rural, remote

and metropolitan areas in Australia. MJA; 181:S10–S14.

Chan WS, Yip PS, Wong PW, Chen EY (2007). Suicide and unemployment: what are

the missing links? Arch Suicide Res.; 11:327–335.

Chen XK, Yang Q, Smith G, Krewski D, Walker M, Wen SW (2006). Environmental

lead level and pregnancy-induced hypertension. Environ Res.; 100:424–430.

Cocklin C, Keen M (2000). Urbanization in the Pacific: environmental change,

vulnerability and human security. Environmental Conservation; 27:392–403.

Cohen J (2008). Safe in our hands?: a study of suicide and self-harm in asylum

seekers. J Forensic Leg Med.; 15:235–244.

Congdon P (2000). Monitoring suicide mortality: A Bayesian approach. Eur J Popul;

16:251–284.

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Smith%20G%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Wen%20SW%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Cohen%20J%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

110

Deisenhammer EA, Kemmler G, Parson P (2003). Association of meteorological

factors with suicide. Acta Psychiatr Scand.; 108:455-459.

Dervic K, Oquendo MA, Grunebaum MF, Ellis S, Burke AK, John Mann J (2004).

Religious Affiliation and Suicide Attempt. Am J Psychiatry; 161:2303–2308.

Diaz JH (2007). The influence of global warming on natural disasters and their public

health outcomes. Am J Disaster Med; 2:33–42.

Dixon PG, McDonald AN, Scheitlin KN, Stapleton JE, Allen JS, Carter WM, Holley

MR, Inman DD, Roberts JB (2007). Effects of temperature variation on suicide in five

U.S. counties, 1991–2001. Int J Biometeorol.; 51:395–403.

Dollins AB, Zhdanova IV, Wurtman RJ, Lynch HJ, Deng MH (1994). Effect of

inducing nocturnal serum melatonin concentrations in daytime on sleep, mood, body

temperature, and performance. Proc Natl Acad Sci USA; 91:1824–1828.

Eksler V, Lassarre S (2008). Evolution of road risk disparities at small-scale level:

Example of Belgium. J Safety Res.; 39:417–427.

Else IR, Andrade NN, Nahulu LB (2007). Suicide and suicidal-related behaviors

among indigenous Pacific Islanders in the United States. Death Studies; 31: 479–501.

EPA (2008). Climate change in Queensland, what the science is telling us. Office of

Climate Change, Environmental Protection Agency Queensland Government.

Escaramís G, Carrasco JL, Ascaso C (2008). Detection of significant disease risks

using a spatial conditional autoregressive model. Biometrics; 64:1043–1053.

Evren C, Sar V, Evren B, Dalbudak E (2008). Self-mutilation among male patients

with alcohol dependency: the role of dissociation. Compr Psychiatry; 49:489–495.

Faedda GL, Tondo L, Teicher MH, Baldessarini RJ, Gelbard HA, Floris GF (1993).

Seasonal mood disorders. Pattern of seasonal recurrence in mania and depression.

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Else%20IR%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Andrade%20NN%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Nahulu%20LB%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

111

Arch Gen Psychiatry; 50:17–23.

Fergusson DM, Boden JM, Horwood LJ (2007). Recurrence of major depression in

adolescence and early adulthood, and later mental health, educational and economic

outcomes. Br J Psychiatry; 191:335–342.

Greenland S, Morgenstern H (1989). Ecological bias, confounding, and effect

modification. Int J Epidemiol.; 18:269–274.

Hall WD, Mant A, Mitchell PB, Rendle VA, Hickie IB, McManus P (2003).

Association between antidepressant prescribing and suicide in Australia, 1991-2000:

trend analysis. BMJ; 326:1008.

Hallas P, Hansen AR, Stæhr MA, Munk-Andersen E, Jorgensen HL (2007). Length of

stay in asylum centres and mental health in asylum seekers: a retrospective study from

Denmark. BMC Public Health, 147:288.

Hanley JA, Negassa A, Edwardes MD, Forrester JE (2003). Statistical analysis of

correlated data using generalized estimating equations: An orientation. Am J

Epidemiol.; 157:364–375.

Hjelmeland H, Knizek BL, Kinyanda E, Musisi S, Nordvik H, Svarva K (2008).

Crisis; 29:137–144.

Hoffpauir SA, Woodruff LA (2008). Effective mental health response to catastrophic

events: lessons learned from Hurricane Katrina. Fam Community Health; 31:17–22.

Huisman M, Oldehinkel AJ (2009). Income inequality, social capital and self-inflicted

injury and violence-related mortality. J Epidemiol Community Health; 63:31–37.

Hunter E, Milroy H (2006). Aboriginal and Torres Strait islander suicide in context.

Arch Suicide Res.; 10:141–157.

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Boden%20JM%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Horwood%20LJ%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

112

Inoue K, Tanii H, Kaiya H, Abe S, Nishimura Y, Masaki M, Okazaki Y, Nata M and

Fukunaga T (2007). The correlation between unemployment and suicide rates in

Japan between 1978 and 2004. Leg Med.; 9:139–142.

IPCC, Ed. (2007). Climate Change 2007: Impacts, Adaptation, and Vulnerability.

Contribution of Working Group II to Fourth Assessment Report of the

Intergovernmental Panel on Climate Change. Cambridge: Cambridge University

Press.

IPCC, Ed. (2001) Climate Change 2001: Impacts, Adaptation, and Vulnerability.

Contribution of Working Group II to Third Assessment Report of the

Intergovernmental Panel on Climate Change. Cambridge: Cambridge University

Press.

Iverson L, Andersen O, Andersen PK, Christoffersen K, Keiding N (1987).

Unemployment and mortality in Denmark, 1970-80. BMJ; 295:879–884.

Judd F, Cooper AM, Fraser C, Davis J (2006). Rural suicide--people or place effects?

Aust N Z J Psychiatry; 40:208–216.

Kalist DE, Molinari NA, Siahaan F (2007). Income, employment and suicidal

behavior. J Ment Health Policy Econ.; 10:177–187.

Kalmar S, Szanto K, Rihmer Z, Mazumdar S, Harrison K, Mann JJ (2008).

Antidepressant prescription and suicide rates: effect of age and gender. Suicide Life

Threat Behav.; 38:363–374.

Kefi S, van Baalen M, Rietkerk M, Loreau M (2008). Evolution of local facilitation in

arid ecosystems. Am Nat; 172:E1–17.

Kovats RS, Haines A, Stanwell-Smith R, Martens P, Menne B, Bertollini R (1999).

Climate change and human health in Europe. BMJ; 318:1682–1685.

113

Kposowa AJ, McElvain JP, Breault KD (2008). Immigration and suicide: the role of

marital status, duration of residence, and social integration. Arch Suicide Res.; 12:82–

92.

Lakhan SE, Vieira KF (2008). Nutritional therapies for mental disorders. Nutr J.; 7:2

Lambert G, Reid C, Kaye D, Jennings G, Esler M (2003). Increased suicide rate in the

middle-aged and its association with hours of sunlight. Am J Psychiatry; 160:793–795.

Landberg J (2008). Alcohol and suicide in Eastern Europe. Drug Alcohol Rev.;

27:361–373.

Lee HC, Lin HC, Tsai SY, Li CY, Chen CC, Huang CC. (2006). Suicide rates and the

association with climate: A population-based study. J Affect Disord.; 92:221–226.

Leenaars AA (2006). Suicide among indigenous peoples: introduction and call to

action. Arch Suicide Res.; 10:103–115.

Li Y, Zhang J, McKeown RE (2009). Cross-sectional assessment of diet quality in

individuals with a lifetime history of attempted suicide. Psychiatry Res.; 165: 111–

119.

Lin HC, Chen CS, Xirasagar S, Lee HC (2008). Seasonality and climatic associations

with violent and nonviolent suicide: a population-based study. Neuropsychobiology;

57:32–37.

Linkowski P, Martin F, De Maertelaer V (1992). Effect of some climatic factors on

violent and non-violent suicides in Belgium. J Affect Disord.; 25:161–166.

Maes M, Scharpe S, Verkerk R, D'Hondt P, Peeters D, Cosyns P, Thompson P, De

Meyer F, Wauters A, Neels H (1995). Seasonal variation in plasma L-tryptophan

availability in healthy volunteers. Relationships to violent suicide occurrence. Arch.

Gen. Psychiatry; 52:937–946.

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Kposowa%20AJ%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22McElvain%20JP%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Breault%20KD%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

114

Maes M, De Meyer F, Thompson P et al (1994). Synchronised annual rhythm in

violent suicide rate, ambient temperature and light-dark span. Acta Psychiatr. Scand.;

90:191–196.

MapInfo (2006). MapInfo Professional Version 8.5 User Guide (Unabridged). New

York: MapInfo Corporation.

McCoy (2007). Suicide and desert men: the power and protection of kanyirninpa

(holding). Australas Psychiatry; 15:S63–S67.

McMichael AJ, Woodruff RE, Hales S (2006). Climate change and human health:

present and future risks. Lancet; 367:859–869.

McMichael AJ, Friel S, Nyong T, Corvalan C (2008). Global environmental change

and health: impacts, inequalities, and the health sector. BMJ; 336:191–194.

McMichael AJ, Campbell-Lendrum D, Kovats S, et al. (2004) Climate Change. In:

Ezzati M, Lopez AD, Rodgers A, Mathers C (eds.) Comparative Quantification of

Health Risks: Global and Regional Burden of Disease due to Selected Major Risk

Factors. Geneva: World Health Organization, 2004, pp.1543–1650.

Measey MA, Li SQ, Parker R, Wang Z (2006). Suicide in the Northern Territory,

1981–2002. MJA; 185:315–319.

Middleton N, Sterne J, Gunnell DJ (2008). An atlas of suicide mortality: England and

Wales, 1988–1994. Health Place; 14:492–506.

Morrell S, Taylor R, Kerr C (1998). Unemployment and young people‘s health. MJA;

168:236–240.

Morrell S, Taylor R, Slaytor E, Ford P (1999). Urban and rural suicide differentials in

migrants and the Australian-born, New South Wales, Australia 1985-1994. Soc Sci

Med.; 49:81–91.

115

Nicholls N, Butler CD, Hanigan I (2006). Inter-annual rainfall variations and suicide

in New South Wales, Australia, 1964–2001. Int J Biometeorol.; 50:139–143.

NSW Health (2006). Population Health Division. The Health of the people of New

South Wales—Report of the Chief Health Officer. Sydney: NSW Department of

Health; 2006.

Nock MK, Borges G, Bromet EJ, Cha CB, Kessler RC, Lee S (2008). Suicide and

suicidal behavior. Epidemiol Rev.; 30:133–154.

Oliver RG, Hetzel BS (1972). Rise and fall of suicide rates in Australia: relation to

sedative availability, MJA; 2:919–923.

Oravecz R, Czigler BA, Moore M (2006). The transformation of suicide fluctuation in

Slovenia. Arch Suicide Res.; 10:69–76.

Page A, Morrell S, Taylor R. (2002). Suicide and political regime in New South

Wales and Australia during the 20th century. J Epidemiol Community Health;

56:766–772.

Page A, Morrell S, Taylor R, Carter G, Dudley M. (2006). Divergent trends in suicide

by socio-economic status in Australia. Soc Psychiatry Psychiatr Epidemiol.; 41; 911–

917.

Page A, Morrell S, Taylor R, Dudley M, Carter G (2007). Further increases in rural

suicide in young Australian adults: Secular trends, 1979–2003. Soc Sci Med.; 65:442–

453.

Papadopoulos FC, Frangakis CE, Skalkidou A, Petridou E, Stevens RG,

Trichopoulos D (2005). Exploring lag and duration effect of sunshine in triggering

suicide. J Affect Disord.; 88:287–297.

Partonena T, Haukkaa J, Nevanlinnab H, Lonnqvista J (2004) Analysis of the

seasonal pattern in suicide. J Affect Disord.; 81:133–139.

116

Pirkola S, Sund R, Sailas E, Wahlbeck K (2009) Community mental-health services

and suicide rate in Finland: a nationwide small-area analysis. Lancet; 373:147–153.

Poulos RG, Hayen A, Chong SS, Finch CF (2008). Geographic mapping as a tool for

identifying communities at high risk of fire and burn injuries in children. Burns;

35:417–424.

Preti A (1998). The influence of climate on suicidal behaviour in Italy. Psychiatry

Res.; 78:9–19.

Preti A, Miottob P (1998). Seasonality in suicides: the influence of suicide method,

gender and age on suicide distribution in Italy. Psychiatry Res.; 81:219–231.

Preti A (1997). The influence of seasonal change on suicidal behaviour in Italy. J

Affect Disord.; 44:123–130.

Preti A, Lentini G, Maugeri M (2007). Global warming possibly linked to an

enhanced risk of suicide: Data from Italy, 1974–2003. J Affect Disord.: 102:19–25.

Rasic DT, Belik SL, Elias B, Katz LY, Enns M, Sareen J (2009). Spirituality, religion

and suicidal behavior in a nationally representative sample. J Affect Disord.; 114:32–

40.

Rocchi MB, Sisti D, Miotto P, Preti A (2007). Seasonality of suicide: relationship

with the reason for suicide. Neuropsychobiology; 56:86–92.

Rock D, Greenberg DM, Hallmayer JF (2003). Increasing seasonality of suicide in

Australia 1970–1999. Psychiatry Res.; 120:43–51.

Rudorfer, MV, Skwerer, RG, Rosenthal, NE (1993). Biogenic amines in seasonal

affective disorder: Effects of light therapy. Psychiatry Res.; 46:19–28.

117

Salib E (1997). Elderly suicide and weather conditions: is there a link? Int J Geriatr

Psychiatry; 12:937–941.

Saunderson TR, Langford IH (1996). A study of the geographical distribution of

suicide rates in England and Wales 1989-92 using empirical Bayes estimates. Soc Sci

Med; 43:489–502.

Seale JP, Shellenberger S, Spence J (2006). Alcohol problems in Alaska Natives:

lessons from the Inuit. Am Indian Alsk Native Ment Health Res.; 13:1–31.

Shaw M, Dorling D, Smith GD (2002). Mortality and political climate: how suicide

rates have risen during periods of Conservative government, 1901–2000. J. Epidemiol.

Community Health; 56:723–725.

Sørensen HJ, Mortensen EL, Wang AG, Juel K, Silverton L, Mednick SA (2009).

Suicide and mental illness in parents and risk of suicide in offspring: A birth cohort

study. Soc Psychiatry Psychiatr Epidemiol. [Epub ahead of print]

Tassone EC, Waller LA, Michele L. Casper ML (2009). Small-area racial disparity in

stroke mortality: an application of bayesian spatial hierarchical modeling.

Epidemiology; 20: 234–241.

Taylor R, Page A, Morrell S, Harrisonb J, Carterc G (2005a). Mental health and

socio-economic variations in Australian suicide. Soc Sci Med.; 61:1551–1559.

Taylor R, Page A, Morrell S, Harrison J, Carter G (2005b). Social and psychiatric

influences on urban-rural differentials in Australian suicide. Suicide Life Threat

Behav.; 35:277–290.

Taylor R, Morrell S, Slaytor E, Ford P (1998). Suicide in urban New South Wales,

Australia 1985-1994: socio-economic and migrant interactions. Soc Sci Med.;

47:1677–1686.

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Seale%20JP%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Shellenberger%20S%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

http://www.ncbi.nlm.nih.gov/sites/entrez?Db=pubmed&Cmd=Search&Term=%22Spence%20J%22%5BAuthor%5D&itool=EntrezSystem2.PEntrez.Pubmed.Pubmed_ResultsPanel.Pubmed_DiscoveryPanel.Pubmed_RVAbstractPlus

118

Vandentorren S, Bretin P, Zeghnoun A, Mandereau-Bruno L, Croisier A, Cochet C,

Ribéron J, Siberan I, Declercq B, Ledrans M (2006). August 2003 heat wave in

France: risk factors for death of elderly people living at home. Eur J Public Health;

16:583–591.

Wang AG, Stórá T (2009). Core features of suicide, gender, age, alcohol and other

putative risk factors in a low-incidence population. Nord J Psychiatry; 63:154–159.

Westman J, Sundquist J, Johansson LM, Johansson SE, Sundquist K (2006). Country

of birth and suicide: a follow-up study of a national cohort in Sweden. Arch Suicide

Res.; 10:239–248.

Wexler L, Hill R, Bertone-Johnson E, Fenaughty A (2008). Correlates of Alaska

Native fatal and nonfatal suicidal behaviors 1990-2001. Suicide Life Threat Behav.;

38:311–320.

WHO (2008). Protecting health from climate change – a toolkit for event organisers,

see: http://www.who.int/world-health-day/toolkit/toolkit_en.pdf (Accessed 10 April,

2008)

WHO (2007). International Classification of Diseases (ICD) online. World Health

Organisation. http://www.who.int/classifications/icd/en/ (accessed 20 August 2008).

WHO (2006). Preventing suicide, a resource at work. WHO Press, World Health

Organization, Geneva, Switzerland.

WHO (2004). The world health report: changing history. WHO Press, World Health


WHO (2003). The world health report: shaping the future. WHO Press, World Health


http://www.who.int/world-health-day/toolkit/toolkit_en.pdf

http://www.who.int/classifications/icd/en/

119

WHO (2002). World report on violence and health. WHO Press, World Health


WHO (1999). Injury: a leading cause of the global burden of disease. World Health


Wojnar M, Ilgen MA, Czyz E, Strobbe S, Klimkiewicz A, Jakubczyk A, Glass J, Kirk

Brower KJ (2009). Impulsive and non-impulsive suicide attempts in patients treated

for alcohol dependence. Journal of Affective Disorders; 115:131–139.

Yasan A, Danis R, Tamam L, Ozmen S, Ozkan M (2008). Socio-cultural features and

sex profile of the individuals with serious suicide attempts in southeastern Turkey: a

one-year survey. Suicide Life Threat Behav.; 38:467–480.

SOCIO-ENVIRONMENTAL FACTORS AND SUICIDE IN …eprints.qut.edu.au/30317/1/Xin_Qi_Thesis.pdf ·...

Documents

Transcript of SOCIO-ENVIRONMENTAL FACTORS AND SUICIDE IN …eprints.qut.edu.au/30317/1/Xin_Qi_Thesis.pdf ·...