SOCIO-ENVIRONMENTAL FACTORS AND SUICIDE IN …eprints.qut.edu.au/30317/1/Xin_Qi_Thesis.pdf ·...
Transcript of 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
TABLE 3-4: SPEARMAN CORRELATIONS BETWEEN OF SOCIO-
ENVIRONMENTAL DETERMINANTS AND MALE ASM OF
SUICIDE------------------------------------------------------------------------- 52
TABLE 3-5: SPEARMAN CORRELATIONS BETWEEN OF SOCIO-
ENVIRONMENTAL DETERMINANTS AND FEMALE ASM OF
SUICIDE ------------------------------------------------------------------------ 53
TABLE 3-6: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL
DETERMINANTS AND MALE SUICIDES-------------------------------55
TABLE 3-7: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL
DETERMINANTS AND FEMALE SUICIDES ---------------------------55
TABLE 3-8: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL
DETERMINANTS AND YOUNG MALE SUICIDES ------------------ 56
TABLE 3-9: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL
XI
DETERMINANTS AND YOUNG FEMALE SUICIDES --------------- 56
TABLE 3-10: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL
DETERMINANTS AND MIDDLE AGE MALE SUICIDES ----------- 57
TABLE 3-11: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL
DETERMINANTS AND MIDDLE AGE FEMALE SUICIDES ------- 57
TABLE 3-12: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL
DETERMINANTS AND SUICIDE AMONG MALE ELDERLY----- 58
TABLE 3-13: POISSON REGRESSION OF SOCIO-ENVIRONMENTAL
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
TABLE 4-8: GEE REGRESSION OF SOCIO-ENVIRONMENTAL
DETERMINANTS OF SUICIDE (MALE) -------------------------------- 86
TABLE 4-9: GEE REGRESSION OF SOCIO-ENVIRONMENTAL
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.
Authors, year Country,
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
Authors, year Country,
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).
**. Correlation is significant at the 0.01 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).
**. Correlation is significant at the 0.01 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)
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population); UER (unemployment rate).
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population).
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population); UER (unemployment rate).
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population); UER (unemployment rate).
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
Parameter ß SE 95% CI
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population).
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population).
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
population with low individual income); PPLEL (proportion of population with low educational level)
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)
ASM RF Tmin Tmax IRSAD IRSD IER IEO PIP UER PPLII
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 0.05 level (2-tailed)
**. Significant at the 0.01 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)
84
Table 4-6: Pearson correlations (seasonal data, female)
ASM RF Tmin Tmax IRSAD IRSD IER IEO PIP UER PPLII
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 0.05 level (2-tailed)
**. Significant at the 0.01 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)
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)
Parameter ß SE 95% CI
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population); UER (unemployment rate); PPLII (proportion of
population with low individual income); PPLEL (proportion of population with low educational level)
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)
Parameter ß SE 95% CI
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population); UER (unemployment rate); PPLII (proportion of
population with low individual income); PPLEL (proportion of population with low educational level)
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)
Parameter ß SE 95% CI
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
*Note: RF (rainfall); Tmax (maximum temperature); IRSD (Index of Relative Socio-economic Advantage and
Disadvantage); PIP (proportion of Indigenous population); UER (unemployment rate); PPLII (proportion of
population with low individual income); PPLEL (proportion of population with low educational level)
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
90
employment rates, increased income and more accesses to training and education,
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
employment rates, increased income and more accesses to training and education,
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 (2001). Demography, Queensland, 2000. Cat. No. 3311.3, ABS, Canberra.
ABS (2002). Demography, Queensland, 2001. Cat. No. 3311.3, 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). Australian Standard Geographical Classification (ASGC) 2000. Cat. No.
1216.0, ABS, Canberra.
ABS (2000). Suicides, Australia, 1921–1998. Cat. No. 3309.0, ABS, Canberra.
ABS (2001). Australian Standard Geographical Classification (ASGC) 2001. Cat. No.
1216.0, ABS, Canberra.
ABS (2002). Australian Standard Geographical Classification (ASGC) 2002. Cat. No.
1216.0, ABS, Canberra.
ABS (2002). Year Book Australia, 2002. Cat. No. 1301.0, ABS, Canberra.
107
ABS (2003). Australian Standard Geographical Classification (ASGC) 2003. Cat. No.
1216.0, ABS, Canberra.
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
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.
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.
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.
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.
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.
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
Organization, Geneva, Switzerland.
WHO (2003). The world health report: shaping the future. WHO Press, World Health
Organization, Geneva, Switzerland.
119
WHO (2002). World report on violence and health. WHO Press, World Health
Organization, Geneva, Switzerland.
WHO (1999). Injury: a leading cause of the global burden of disease. World Health
Organization, Geneva, Switzerland.
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.
120