Lippincott Williams & Wilkins · Web viewDemographic simulation (age, sex and mortality) The model...
Transcript of Lippincott Williams & Wilkins · Web viewDemographic simulation (age, sex and mortality) The model...
Appendix – Model Technical details and validation
This supplementary material provides technical details of the model design including parameter values, model assumptions and results of model validation. Figure 1 shows the basic model design. Briefly, the model is an individual-based simulation of the full population of Zimbabwe, representing births and death, HIV incidence and disease progression and the development of several NCDS. The model uses data and estimates that are available for the period 1950 to 2015, and is then used to forecast future trends in the period 2015 and 2035.
Figure 1. Schematic of the multi-disease model for Zimbabwe. The model simulates demography (blue), HIV-infection (orange), and non-communicable diseases (green). The model accounts for demographic and medical risk factors for HIV-infection and for development of non-communicable diseases (red arrows to individual conditions and group of conditions). The model starts in 1950 and runs until 2035, making projections from 2015 onwards. Demographic and non-communicable disease patterns are simulated from 1950 and the HIV-epidemic from 1975. From 2015, the model is used to forecast the timing and nature of non-communicable diseases in the general population and HIV-positive population.
1. Technical details
Demographic simulation (age, sex and mortality)
The model simulated the whole population of Zimbabwe from 1950 until 2035. Demographic simulation of the population of Zimbabwe are based on data from United Nation’s (UN) Department of Economic and Social Affairs.(1) These include sex-specific age composition in 1950, age-specific fertility rates between 1950 and 2015, age-and-sex-specific mortality rates between 1950 and 2015. Rates are presented by 5-year time periods, with the model assuming that rates within these periods are constant. Cause-specific mortality was fitted by hand in the model (including mortality for HIV and each of the NCDs included) by reducing overall mortality estimates in HIV-negative and patients without NCDs by matching to overall mortality estimates provided by UN estimates. Projections beyond 2015 assume that fertility and mortality rate remain constant at the 2015 levels.
HIV epidemic
The model simulates the HIV epidemic from 1975, including new HIV infection, CD4 count progression, ART initiation and HIV-related death. Annual age-and-sex-specific HIV incidence rates (including paediatric infection) were taken from the official UNAIDS estimates.(2) These incidence estimates are derived through fitting a model to available data, largely prevalence estimates in Demographic and Health Surveys. ART roll-out and scale-up, including the expanding ART eligibility criteria over time, are based on data on the number of people starting ART in Zimbabwe from UNAIDS estimates, including number of persons on ART by CD4 count.(2)
Estimates of CD4 count at seroconversation (Table 1), CD4 count progression rates (Table 2), and mortality by CD4 count (Table 3) were taken from a study by Mangal et al, AIDS (under review) of a joint analysis of all seroconverter data in Europe, Africa, South America and Asia (3). Projections of the number of new HIV-infections and people starting ART assume that incidence and ART trends remain constant the 2015 level.
Table 1. CD4 count at HIV-infection. Age-and-sex-specific probability of being in specific CD4 count category at seroconversion. Source: Mangal et al AIDS (3)
Age Category
CD4 count category
>500
500-350
350-250
Men
<30
0.58
0.25
0.17
30-40
0.53
0.26
0.20
40-55
0.48
0.26
0.23
>55
0.4
0.26
0.28
Women
<30
0.63
0.22
0.15
30-40
0.58
0.23
0.18
40-55
0.53
0.23
0.21
>55
0.44
0.23
0.26
Table 2. Rate of CD4 count progression. Source: Mangal et al AIDS (3)
CD4 count progression
From >500 to 500-350
From 500-350 to 350-250
From 350-250 to 250-200
From 250-200 to 200-100
From 200-100 to 100-50
From 100-50 to <50
Men
0.19835
0.32913
0.33628
0.50936
0.57271
0.86702
Women
0.18456
0.30458
0.31120
0.47137
0.52999
0.80236
Table 3. Age-and-sex-specific mortality rate by CD4 count. Source: Mangal et al AIDS (3)
Age Category
CD4 count category
>500
500-350
350-250
250-200
200-100
100-50
<50
Men
<25
0.00321
0.00912
0.00733
0.01281
0.02161
0.03741
0.84544
25-35
0.00403
0.01145
0.00921
0.01608
0.02713
0.04696
1.06123
35-45
0.00497
0.01412
0.01136
0.01984
0.03347
0.05794
1.30933
>45
0.00712
0.02022
0.01626
0.02841
0.04792
0.08296
1.87488
Women
<25
0.00334
0.00948
0.00762
0.01331
0.02245
0.03887
0.87847
25-35
0.00419
0.01189
0.00957
0.01671
0.02819
0.04879
1.10269
35-45
0.00517
0.01467
0.01180
0.02061
0.03478
0.06020
1.36048
>45
0.00740
0.02101
0.01690
0.02952
0.04980
0.08620
1.94813
NCDs
The model simulates the development of a number of NCDs including: asthma, chronic kidney disease (CKD), depression, diabetes, hypertension, strokes and cancers. The individual cancers, which were chosen based on the most prevalent non-AIDS defining cancers (as determined by 2012 GloboCan (4)) in Zimbabwe, were: colorectal, liver, oesophageal, breast, cervical and prostate cancer as well as ‘other’ cancer, which comprise all other cancers.
An extensive literature review was carried out to obtain age-specific incidence or prevalence to parameterise the development of NCDs. The literature review focused on published studies recording age-specific prevalence or incidence rates for Zimbabwe or neighbouring countries (with the implicit assumption that that neighbouring countries experience similar age-specific prevalence patterns as Zimbabwe). The review found a paper by Negin et al, using survey of a representative sample of people in South Africa report data on 2008 age-specific prevalence of asthma, depression, diabetes, hypertension and stroke.(5) Questions included whether individuals had every been diagnosed with a number of NCDs, with asthma and depression validated against a set of symptom questions and related diagnostic algorithm.(5) A paper from Tanzania by Stanifer et al was identified, presenting age-specific estimates of CKD prevalence for 2014 (6). CKD was defined as estimated glomerular filtration rate ≤60ml/min/1.73m2 using the Modification of Diet in Renal Disease equation and/or persistent albuminuria (6), Age-specific cancer incidence estimates were taken from the GloboCan dataset for Zimbabwe, which used a number of data sources and methods, including regional incidence data and mortality data to provide national cancer-specific incidence estimates.(4) Age-standardized cancer rates are presented in Table 4.
Table 4. Age-standardized cancer rates (ASR) per 100,00 (both sexes). Source: Globocan (4)
cc
Breast cancer
Cervical cancer
Colorectal cancer
Liver cancer
Oesophagus cancer
Prostate cancer
Other cancers
ASR
38.8
40.1
8.6
5.4
17.6
31.6
39.7
Where studies presented prevalence estimates, age-specific incidence rates were calculated, assuming age-specific prevalence estimates are constant over time as per Equation 1, fitting I(a) to solve for P(a), where I(a) stands for incidence at age a and p(a,b) for the probability of an individual infected at age a surviving to age b. We assumed that NCD prevalence/incidence only varied significantly by HIV status if found to do so in a comprehensive study by Schouten and colleague in HIV-positive patients and age-and-sex-matched uninfected controls (7). Therefore, where prevalence estimates were presented by HIV-status, prevalence data was pooled between HIV-negative and HIV-positive individuals to provide average estimates for both groups. Estimates were pooled by weighing for the proportion of patients in each group as generated by the model. For example, if 80% of people in Zimbabwe in the year of the reported prevalence were HIV-negative and 20% were HIV-positive, prevalence for HIV-negative counted 8/10 and HIV-positive 2/10 towards the pooled average. We used age-specific population estimates (number of people in the population) generated by the model to back-calculate estimates of the number of new cases diagnosed to match prevalence levels observed. These were used to calculate incidence estimates.
Equation 1.
The model further accounts for the propensity for one NCD to be associated with an increased risk to develop another– for example for hypertension to increase the risk of having a stroke through common causal pathways. Parameters defining these increased risks were based on the output of an extensive literature review including studies from all regions.(8) These links are illustrated by the red arrow in Figure 1. The increased risk for HIV-positive individuals to develop certain NCDs was incorporated into the model through parameters based on a large study The Netherlands that compares NCD risk amongst HIV-positive individuals and uninfected controls.(7) The mechanisms driving increased NCD risk amongst HIV-positive patients are complex and their relative contribution a topic of ongoing research.(9)
2. Model calibration
A number of model checks were carried out to ensure that the model was able to reconstruct demographic and epidemiological trends robustly. This was done by comparing model output to available data.
Demography
The model-based simulation of the demographic trends in Zimbabwe were compared to data from United Nations’ (UN) Department of Economic and Social Affairs for Zimbabwe (1). The model was able to accurately reconstruct the Zimbabwean demographic transitions observed by national surveillance between 1950 and 2015. Figure 2
shows the comparison between UN surveillance data and model output. The model was able to reconstruct the population growth between 1950 and 2015 (Figure 2), as well as accurately reconstruct age distribution in between 1950 and 2015 (Figure 3 shows comparison to 2015 data as a snapshot). Fertility and mortality patterns were accurately simulated in the model, recreating both the number of births and death between 1950 and 2015 (Figure 4A and Figure 4B) and births and deaths by age group (Figure 4C and Figure 4D). Any discrepancies between the data and the model can in part be explained by the stochastic nature of the model, with results presented below based on one model run.
We have further checked the model between 2015 and 2035 against all available projections generated from UN Department of Economic and Social Affairs for Zimbabwe (Figure 5) (1). We compared both the projected number of people in Zimbabwe and the age composition of the population between 2015 and 2035. We find that the model is able to recreate the future projected trends well (Figure 5).
Figure 2.Comparison of the total population (both sexes) measures for Zimbabwe from UN Department of Economic and Social Affairs for Zimbabwe to the model output between 1950-2015. Data source: UN Department of Economic and Social Affairs for Zimbabwe.(1)
Figure 3. Comparison of the age structure in July 2015 for Zimbabwe from UN Department of Economic and Social Affairs for Zimbabwe to the model output. Data source: UN Department of Economic and Social Affairs for Zimbabwe.(1)
Figure 4. Comparison of demographic measures for Zimbabwe from UN Department of Economic and Social Affairs for Zimbabwe to the model output between 1950 and 2015. Specifically; A. The number of births between 1950 and 2015, B. Number of births by age group of the mother between 2005 and 2015, C. Number of deaths between 1950 and 2015, and D. Number of deaths by age group in July 2015. Data source: UN Department of Economic and Social Affairs for Zimbabwe.(1)
A. B.
C. D.
Figure 5. Comparison of demographic projections for Zimbabwe from UN Department of Economic and Social Affairs for Zimbabwe to the model output between 2015 and 2035 for A. Total population and B. population by age groups. Data source: UN Department of Economic and Social Affairs for Zimbabwe (1).
A. B.
HIV epidemic
The model-based simulation of the HIV epidemic in Zimbabwe was compared to data from UNAIDS for Zimbabwe.(2) The output demonstrates that the model-based reconstruction of the HIV epidemic in Zimbabwe was able to reconstruct both the incidence and prevalence transition in the country between 1975 and 2015, as well as trends in treatment between 2004 and 2015.
Figure 5 A and B shows the number of new HIV-infections from the start of the epidemic in 1975 to 2015 amongst men and women respectively, illustrating that the model was able to capture the increase in the number of new infections from 1975, the peak in 1993 and gradual decrease from 1993 to 2015. When comparing the patterns of overall HIV-infections, the output also managed to reconstruct the increase in HIV prevalence amongst men and women (Figure 5 C and D respectively) from 1975 to the peak in 2000, followed by a gradual decrease, as well as the age-specific prevalence estimates as reported by DHS (2,10).
Figure 6 shows that the model was able to recreate the increasing number of persons on antiretroviral therapy (ART) between 2004 and 2015, including trends by sex (Figure 6 B and C) and in children Figure 6D). The model was further able to match trends in ART initiation by CD4 counts in men and women (Figure 6E and F). Differences between UNAIDS estimates (Figure 6 - data) and our computed prevalence estimates (Figure 6 - model) arise due to differing assumptions relating to natural history and disease progression.
Figure 6. Comparison of measures of HIV epidemic in Zimbabwe from UNAIDS estimates to mode output. A. and B. show the number of new HIV-infections between 1975 and 2015 in men and women respectively. C. and D. show the year total number of HIV-infections men and women respectively between 1975 and 2015. E. compares age-specific prevalence estimates from the model to DHS data. Data stands for UNAIDS estimates and DHS data (for E). Data source: UNAIDS epidemic estimates for Zimbabwe.(2,10).
A. B.
C. D.
E.
Figure 7. Comparison of number of people on ART from UNAIDS estimates for Zimbabwe and model output between 2004 and 2015 for A. both sexes, B. women, C. men, D. children. Comparison by CD4 count are shown in E and F for men and women respectively. The solid colours illustrate data and the dotted colours model output. Data source: UNAIDS epidemic estimates for Zimbabwe (2).
A. B.
C. D.
E. F.
NCD epidemic
We compared model-based age-specific prevalence in 2008 of asthma, depression, hypertension, diabetes and ever having had a stroke to 2008 age-specific prevalence data from Negin et al (Figure 7 A to E). We also compared age-specific prevalence estimates of CKD generated by the model to 2014 age-specific prevalence of CKD from Stanifer et al (Figure 7 F). The results show that the model was able to accurately reconstruct age-specific prevalence patterns.
Figure 8. Comparison of age-specific prevalence from data to mode output for Zimbabwe, including A. asthma, B. depression, C. diabetes, D. stroke, E. hypertension, and F. CKD. A to E were compared to 2008 age-specific prevalence data from Negin et al (2008) (5) and F to 2014 age-specific prevalence data from Stanifer et al (6).
A. B.
C. D.
E. F.
Cancer epidemic
We compared model-based overall cancer-specific prevalence in 2012 of colorectal, liver, oesophagus, and stomach cancer as well ‘other’ cancers (any other cancer except the aforementioned) to 2012 GloboCan estimates for Zimbabwe (Figure 8)(11). The results show that the model was able to accurately reconstruct cancer-specific prevalence patterns in 2012.
Figure 9. Comparison of cancer-specific prevalence from model to 2012 GloboCan in Zimbabwe. Data source: GloboCan estimates for Zimbabwe (11).
Mortality
We compared model-based cause-specific deaths in 2013 to the Institute for Health Metrics and Evaluation’s Global Burden of Disease 2013 estimates of causes of death for Zimbabwe (Figure 9) (12). The results show that the model was able to accurately reconstruct the percentages of death contributed by individual conditions.
Figure 10. Comparison of causes of death from model to 2013 Global Burden of Disease. Data source: Institute for Health Metrics and Evaluation’s Global Burden of Disease 2013 estimates of causes of death for Zimbabwe (12).
Sensitivity analyses
Changes in HIV incidence and ART coverage
The model was re-run with changes in both HIV incidence and ART coverage. Results comparing baseline (HIV incidence remains stable at 2015 level and ART coverage gradually increases to 90:90:90 targets by 2035) to a scenario where HIV incidence increases or decreases annually by 10% and where ART coverage remains stable at 2015 level are presented in Table 5. We have also tested scenarios where ART was never introduced in 2004 (in which case mean age would be 34 by 2035), HIV incidence remained at highest level (mean age would be 31 by 2035), and HIV had no impact on NCD incidence (36% would be diagnosed with at least one NCD by 2035).
Table 5. Sensitivity analyses comparing baseline scenario for HIV incidence and ART coverage to variations in ART coverage and HIV incidence.
Projection in 2035
Baseline*
ART coverage remains at 2015 level
10% annual increase in HIV incidence
10% annual reduction in HIV incidence
Mean age
45
45
45
45
Total HIV population
1.0 million
0.95 million
1.02
0.98
Proportion with ≥
1 NCD
59%
58%
59%
59%
Number of people with at least one NCD
0.84 million
0.77 million
0.84 million
0.84 million
*Baseline defined as HIV incidence that remains stable at 2015 levels up until 2035 and where ART coverage increases steadily to 90:90:90 targets.
Addition of Myocardial Infraction
Due to the lack of robust age-specific data on myocardial infraction (MI) in the region, MI was not included in the model. We did however carry out additional analyses, where MI was included in the model and incidence rates were assumed to be similar to the incidence of stroke. We assumed interactions between pre-existing NCDs and risk of MI (Figure 1) to be the same as stroke. The results showed that the proportion with at least one NCD by HIV status would not vary greatly, reflecting the low incidence of MIs assumed. The proportion of HIV-negative people suffering from at least one NCD would increase from 14.3% to 14.4% in 2015 and 21.2% to 21.3% by 2035. Amongst HIV-positive this would increase from 33.2% to 33.3% in 2015 and from 59.4 to 59.5% by 2035.
References
1. United Nations. United Nations - Department of Economic and social Affairs [Internet]. 2015 [cited 2016 Aug 1]. Available from: http://esa.un.org/unpd/wpp/Download/Standard/Population/
2. UNAID. UNAIDS - Epidemic Projection Package [Internet]. 2015 [cited 2016 May 5]. Available from: http://www.unaids.org/en/dataanalysis/datatools/spectrumepp
3. Mangal TD, UNAIDS Working Group on CD4 Progression and Mortality Amongst HIV Seroconverters including the CASCADE Collaboration in EuroCoord. Joint estimation of CD4+ cell progression and survival in untreated individuals with HIV-1 infection. AIDS. 2017 May 15;31(8):1073–82.
4. GloboCan 2012: Estimated Cancer Incidence, Mortality and Prevalence Worldwide in 2012 [Internet]. [cited 2016 May 27]. Available from: http://globocan.iarc.fr/old/age-specific_table_r.asp?selection=227716&title=Zimbabwe&sex=0&type=0&stat=0&window=1&sort=0&submit=%C2%A0Execute
5. Negin J, Martiniuk A, Cumming RG, Naidoo N, Phaswana-Mafuya N, Madurai L, et al. Prevalence of HIV and chronic comorbidities among older adults. AIDS. 2012 Jul 31;26 Suppl 1:S55–63.
6. Stanifer JW, Maro V, Egger J, Karia F, Thielman N, Turner EL, et al. The Epidemiology of Chronic Kidney Disease in Northern Tanzania: A Population-Based Survey. PLoS ONE. 2015 Apr 17;10(4):e0124506.
7. Schouten J, Wit FW, Stolte IG, Kootstra N, van der Valk M, Geerlings SG, et al. Cross-sectional comparison of the prevalence of age-associated comorbidities and their risk factors between HIV-infected and uninfected individuals: the AGEhIV Cohort Study. Clin Infect Dis. 2014 Sep 2;
8. Smit M, Brinkman K, Geerlings S, Smit C, Thyagarajan K, van Sighem A, et al. Future challenges for clinical care of an ageing population infected with HIV: a modelling study. The Lancet Infectious Diseases [Internet]. 2015 Jun [cited 2015 Jun 15]; Available from: http://linkinghub.elsevier.com/retrieve/pii/S1473309915000560
9. Althoff KN, Smit M, Reiss P, Justice AC. HIV and ageing: improving quantity and quality of life. Curr Opin HIV AIDS. 2016 Jul 8;
10. Goverment of Zimbabwe, The DHS Program, ICF International, Rockville, Maryland USA. Zimbabwe: Demographic and Health Survey 2015 [Internet]. [cited 2017 Jul 11]. Available from: http://dhsprogram.com/pubs/pdf/FR322/FR322.pdf
11. GloboCan. GloboCan - Estimated Incidence by age [Internet]. 2012 [cited 2016 May 27]. Available from: http://globocan.iarc.fr/old/age-specific_table_r.asp?selection=227716&title=Zimbabwe&sex=0&type=0&stat=0&window=1&sort=0&submit=%C2%A0Execute
12. IHME. Institute for Health Metrics and Evaluation - MDG Viz [Internet]. [cited 2016 Aug 20]. Available from: http://www.healthdata.org/institute-health-metrics-and-evaluation
2
25 20 15 10 5 0 5 10 15 20 25
0-45-9
10-1415-1920-2425-2930-3435-3940-4445-4950-5455-5960-6465-6970-7475-7980+
Percentage
AgeGrou
p
Women- Model Women- Data
Men- Model Men- Data
2520151050510152025
0-4
5-9
10-14
15-19
20-24
25-29
30-34
35-39
40-44
45-49
50-54
55-59
60-64
65-69
70-74
75-79
80+
Percentage
A
g
e
G
r
o
u
p
Women-ModelWomen-Data
Men-ModelMen-Data
0.0
500.0
1,000.0
1,500.0
2,000.0
2,500.0
3,000.0Nu
mbe
rofb
irths(1,000)
Calendartime
Data
Model
0.0
500.0
1,000.0
1,500.0
2,000.0
2,500.0
3,000.0
N
u
m
b
e
r
o
f
b
i
r
t
h
s
(
1
,
0
0
0
)
Calendartime
Data
Model
0
100
200
300
400
500
600
700
800
900Nu
mbe
rofb
irth(1,000)
Agegroup
Data
Model
0
100
200
300
400
500
600
700
800
900
N
u
m
b
e
r
o
f
b
i
r
t
h
(
1
,
0
0
0
)
Agegroup
Data
Model
0200400600800
1,0001,2001,4001,6001,8002,000
Numbe
rofd
eaths(1,000)
Calendartime
Data
Model
0
200
400
600
800
1,000
1,200
1,400
1,600
1,800
2,000
N
u
m
b
e
r
o
f
d
e
a
t
h
s
(
1
,
0
0
0
)
Calendartime
Data
Model
0
50
100
150
200
250
300
350Nu
mbe
rofd
eaths(1,000)
AgeGroups
Data
Model
0
50
100
150
200
250
300
350
N
u
m
b
e
r
o
f
d
e
a
t
h
s
(
1
,
0
0
0
)
AgeGroups
Data
Model
0
50,000
100,000
150,000
200,000
250,000
300,000
2015 2020 2025 2030 2035
Num
bero
fpeople(x1,000)
Year
Population projection
Data
Model
0
50,000
100,000
150,000
200,000
250,000
300,000
20152020202520302035
N
u
m
b
e
r
o
f
p
e
o
p
l
e
(
x
1
,
0
0
0
)
Year
Populationprojection
Data
Model
0
10,000
20,000
30,000
40,000
Numbe
rofp
eople
Year
AgeGroupsprojections
0-24 25-69 70
0
10,000
20,000
30,000
40,000
N
u
m
b
e
r
o
f
p
e
o
p
l
e
Year
AgeGroupsprojections
0-2425-6970
020,00040,00060,00080,000
100,000120,000140,000160,000180,000200,000
Nrofn
ewHIVinfections
Year
ModelData
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
180,000
200,000
N
r
o
f
n
e
w
H
I
V
i
n
f
e
c
t
i
o
n
s
Year
Model
Data
020,00040,00060,00080,000
100,000120,000140,000160,000180,000200,000
Nrofn
ewHIVinfections
Year
ModelData
0
20,000
40,000
60,000
80,000
100,000
120,000
140,000
160,000
180,000
200,000
N
r
o
f
n
e
w
H
I
V
i
n
f
e
c
t
i
o
n
s
Year
Model
Data
0
500,000
1,000,000
1,500,000
2,000,000Nrofm
enwith
HIV
Year
Data
Model
0
500,000
1,000,000
1,500,000
2,000,000
N
r
o
f
m
e
n
w
i
t
h
H
I
V
Year
Data
Model
0
500,000
1,000,000
1,500,000
2,000,000Nrofw
omen
with
HIV
Year
Data
Model
0
500,000
1,000,000
1,500,000
2,000,000
N
r
o
f
w
o
m
e
n
w
i
t
h
H
I
V
Year
Data
Model
0%10%20%30%40%50%60%70%80%90%100%
15-19 20-24 25-29 30-34 35-39 40-44 45-49 50-54
Prevalen
ce
Age Groups
Age-specificprevalence
Data- MenModel- MenData- Women
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
15-1920-2425-2930-3435-3940-4445-4950-54
P
r
e
v
a
l
e
n
c
e
AgeGroups
Age-specificprevalence
Data-Men
Model-Men
Data-Women
0
200,000
400,000
600,000
800,000
1,000,000
Numbe
rofp
eople
Year
NumberofpeopleonART- bothsexes
DataModel
0
200,000
400,000
600,000
800,000
1,000,000
N
u
m
b
e
r
o
f
p
e
o
p
l
e
Year
NumberofpeopleonART-bothsexes
Data
Model
0
200,000
400,000
600,000
800,000
1,000,000Nu
mbe
rofp
eople
Year
NumberofpeopleonART- men
DataModel
0
200,000
400,000
600,000
800,000
1,000,000
N
u
m
b
e
r
o
f
p
e
o
p
l
e
Year
NumberofpeopleonART-men
Data
Model
0
200,000
400,000
600,000
800,000
1,000,000Nu
mbe
rofp
eople
Year
NumberofpeopleonART- men
DataModel
0
200,000
400,000
600,000
800,000
1,000,000
N
u
m
b
e
r
o
f
p
e
o
p
l
e
Year
NumberofpeopleonART-men
Data
Model
0
200,000
400,000
600,000
800,000
1,000,000Nu
mbe
rofp
eople
Year
NumberofpeopleonART- children
DataModel
0
200,000
400,000
600,000
800,000
1,000,000
N
u
m
b
e
r
o
f
p
e
o
p
l
e
Year
NumberofpeopleonART-children
Data
Model
0
100,000
200,000
300,000
400,000
500,000
600,000
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Numbe
rofp
eople
Year
ARTbyCD4count- Men
>500 350-500 250-350 200-250 100-200 50-100 <50
>500 350-500 250-350 200-250 100-200 50-100 <50
0
100,000
200,000
300,000
400,000
500,000
600,000
200420052006200720082009201020112012201320142015
N
u
m
b
e
r
o
f
p
e
o
p
l
e
Year
ARTbyCD4count-Men
>500350-500250-350200-250100-20050-100<50
>500350-500250-350200-250100-20050-100<50
0
100,000
200,000
300,000
400,000
500,000
600,000
2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015
Numbe
rofpe
ople
Year
ARTbyCD4count- Women
>500 350-500 250-350 200-250 100-200 50-100 <50
>500 350-500 250-350 200-250 100-200 50-100 <50
0
100,000
200,000
300,000
400,000
500,000
600,000
200420052006200720082009201020112012201320142015
N
u
m
b
e
r
o
f
p
e
o
p
l
e
Year
ARTbyCD4count-Women
>500350-500250-350200-250100-20050-100<50
>500350-500250-350200-250100-20050-100<50
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-49 50-59 >60
Prevalen
ce
AgeGroups
Asthma
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-4950-59 >60
P
r
e
v
a
l
e
n
c
e
AgeGroups
Asthma
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-49 50-59 >60
Prevalen
ce
AgeGroups
Depression
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-4950-59 >60
P
r
e
v
a
l
e
n
c
e
AgeGroups
Depression
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-49 50-59 >60
Prevalen
ce
AgeGroups
Diabetes
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-4950-59 >60
P
r
e
v
a
l
e
n
c
e
AgeGroups
Diabetes
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-49 50-59 >60
Prevalen
ce
AgeGroups
Hypertension
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-4950-59 >60
P
r
e
v
a
l
e
n
c
e
AgeGroups
Hypertension
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-49 50-59 >60
Prevalen
ce
AgeGroups
CKD
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-4950-59 >60
P
r
e
v
a
l
e
n
c
e
AgeGroups
CKD
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-49 50-59 >60
Prevalen
ce
AgeGroups
CKD
Data
Model
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
18-4950-59 >60
P
r
e
v
a
l
e
n
c
e
AgeGroups
CKD
Data
Model
Hypertension
Diabetes
Stroke
Depression
Asthma
HIV & ART
Age and Sex
Breast cancer
Cervical cancer
Oesophageal cancer
Liver cancer
Prostate Cancer
Kidney disease
Other Cancer
Colorectal cancer
Mortality
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
18,000
20,000
1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015
Num
bero
fpeo
ple(1,000)
Year
Data
Model
0
2,000
4,000
6,000
8,000
10,000
12,000
14,000
16,000
18,000
20,000
19501955196019651970197519801985199019952000200520102015
N
u
m
b
e
r
o
f
p
e
o
p
l
e
(
1
,
0
0
0
)
Year
Data
Model