Further evidence of misclassification of the injury deaths in South Africa: When will the barriers to accurate injury death statistics be removed?

BACKGROUND: Contrary to the World Health Organization's internationally recommended medical certificate of cause of death, the South African (SA) death notification form (DNF) does not allow for the reporting of the manner of death to permit accurate coding of external causes of injury deaths. OBJECTIVES: To describe the injury cause-of-death profile from forensic pathology records collected for the National Cause-of-Death Validation (NCoDV) Project and compare it with profiles from other sources of injury mortality data. In particular, the recording of firearm use in homicides is compared between sources. METHODS: The NCoDV Project was a cross-sectional study of deaths that occurred during a fixed period in 2017 and 2018, from a nationally representative sample of 27 health subdistricts in SA. Trained fieldworkers scanned forensic records for all deaths investigated at the forensic mortuaries serving the sampled subdistricts during the study period. Forensic practitioners reviewed the records and completed a medical certificate of cause of death for each decedent. Causes of death were coded to the International Statistical Classification of Diseases, 10th revision (ICD-10), using Iris automated coding software. Cause-specific mortality fractions for injury deaths were compared with Injury Mortality Survey 2017 (IMS 2017) and Statistics South Africa 2017 (Stats SA 2017) datasets. The cause profile for all firearm-related deaths was compared between the three datasets. RESULTS: A total of 5 315 records were available for analysis. Males accounted for 77.6% of cases, and most decedents were aged between 25 and 44 years. Homicide was the leading cause of death (34.7%), followed by transport injuries (32.6%) and suicide (14.7%). This injury cause profile was similar to IMS 2017 but differed markedly from the official statistics, which showed markedly lower proportions of these three causes (15.0%, 11.6% and 0.7%, respectively), and a much higher proportion of other unintentional causes. Investigation of firearm-related deaths revealed that most were homicides in NCoDV 2017/18 (88.5%) and IMS 2017 (93.1%), while in the Stats SA 2017 data, 98.7% of firearm deaths were classified as accidental. Approximately 7% of firearm-related deaths were suicides in NCoDV 2017/18 and IMS 2017, with only 0.3% in Stats SA 2017. CONCLUSION: The official cause-of-death data for injuries in SA in 2017 differed substantially from findings from the NCoDV 2017/18 study and IMS 2017. Accurate data sources would ensure that public health interventions are designed to reduce the high injury burden. Inclusion of the manner of death on the DNF, as is recommended internationally, is critically important to enable more accurate, reliable and valid reporting of the injury profile.

Estimating the changing burden of disease attributable to alcohol use in South Africa for 2000, 2006 and 2012.

BACKGROUND: Alcohol use was one of the leading contributors to South Africa (SA)'s disease burden in 2000, accounting for 7% of deaths and disability-adjusted life years (DALYs) in the first South African Comparative Risk Assessment Study (SACRA1). Since then, patterns of alcohol use have changed, as has the epidemiological evidence pertaining to the role of alcohol as a risk factor for infectious diseases, most notably HIV/AIDS and tuberculosis (TB). OBJECTIVES: To estimate the burden of disease attributable to alcohol use by sex and age group in SA in 2000, 2006 and 2012. METHODS: The analysis follows the World Health Organization (WHO)'s comparative risk assessment methodology. Population attributable fractions (PAFs) were calculated from modelled exposure estimated from a systematic assessment and synthesis of 17 nationally representative surveys and relative risks based on the global review by the International Model of Alcohol Harms and Policies. PAFs were applied to the burden of disease estimates from the revised second South African National Burden of Disease Study (SANBD2) to calculate the alcohol-attributable burden for deaths and DALYs for 2000, 2006 and 2012. We quantified the uncertainty by observing the posterior distribution of the estimated prevalence of drinkers and mean use among adult drinkers (≥15 years old) in a Bayesian model. We assumed no uncertainty in the outcome measures. RESULTS: The alcohol-attributable disease burden decreased from 2000 to 2012 after peaking in 2006, owing to shifts in the disease burden, particularly infectious disease and injuries, and changes in drinking patterns. In 2012, alcohol-attributable harm accounted for an estimated 7.1% (95% uncertainty interval (UI) 6.6 - 7.6) of all deaths and 5.6% (95% UI 5.3 - 6.0) of all DALYs. Attributable deaths were split three ways fairly evenly across major disease categories: infectious diseases (36.4%), non-communicable diseases (32.4%) and injuries (31.2%). Top rankings for alcohol-attributable DALYs for specific causes were TB (22.6%), HIV/AIDS (16.0%), road traffic injuries (15.9%), interpersonal violence (12.8%), cardiovascular disease (11.1%), cancer and cirrhosis (both 4%). Alcohol remains an important contributor to the overall disease burden, ranking fifth in terms of deaths and DALYs. CONCLUSION: Although reducing overall alcohol use will decrease the burden of disease at a societal level, alcohol harm reduction strategies in SA should prioritise evidence-based interventions to change drinking patterns. Frequent heavy episodic (i.e. binge) drinking accounts for the unusually large share of injuries and infectious diseases in the alcohol-attributable burden of disease profile. Interventions should focus on the distal causes of heavy drinking by focusing on strategies recommended by the WHO's SAFER initiative.

Estimating the changing burden of disease attributable to high systolic blood pressure in South Africa for 2000, 2006 and 2012.

BACKGROUND: Ongoing quantification of trends in high blood pressure and the consequent disease impact are crucial for monitoring and decision-making. This is particularly relevant in South Africa (SA) where hypertension is well-established. OBJECTIVE: To quantify the burden of disease related to high systolic blood pressure (SBP) in SA for 2000, 2006 and 2012, and describe age, sex and population group differences. METHODS: Using a comparative risk assessment methodology, the disease burden attributable to raised SBP was estimated according to age, se, and population group for adults aged ≥25 years in SA in the years 2000, 2006 and 2012. We conducted a meta-regression on data from nine national surveys (N=124 350) to estimate the mean and standard deviation of SBP for the selected years (1998 - 2017). Population attributable fractions were calculated from the estimated SBP distribution and relative risk, corrected for regression dilution bias for selected health outcomes associated with a raised SBP, above a theoretical minimum of 110 - 115 mmHg. The attributable burden was calculated based on the estimated total number of deaths and disability-adjusted life years (DALYs). RESULTS: Mean SBP (mmHg) between 2000 and 2012 showed a slight increase for adults aged ≥25 years (127.3 - 128.3 for men; 124.5 - 125.2 for women), with a more noticeable increase in the prevalence of hypertension (31% - 39% in men; 34% - 40% in women). In both men and women, age-standardised rates (ASRs) for deaths and DALYs associated with raised SBP increased between 2000 and 2006 and then decreased in 2012. In 2000 and 2012, for men, the death ASR (339/100 000 v. 334/100 000) and DALYs (5 542/100 000 v. 5 423/100 000) were similar, whereas for women the death ASR decreased (318/100 000 v. 277/100 000) as did age-standardised DALYs (5 405/100 000 v. 4 778/100 000). In 2012, high SBP caused an estimated 62 314 deaths (95% uncertainty interval 62 519 - 63 608), accounting for 12.4% of all deaths. Stroke (haemorrhagic and ischaemic), hypertensive heart disease and ischaemic heart disease accounted for >80% of the disease burden attributable to raised SBP over the period. CONCLUSION: From 2000 to 2012, a stable mean SBP was found despite an increase in hypertension prevalence, ascribed to an improvement in the treatment of hypertension. Nevertheless, the high mortality burden attributable to high SBP underscores the need for improved care for hypertension and cardiovascular diseases, particularly stroke, to prevent morbidity and mortality.

Estimating the changing disease burden attributable to smoking in South Africa for 2000, 2006 and 2012.

BACKGROUND: Ongoing quantification of the disease burden attributable to smoking is important to monitor and strengthen tobacco control policies. OBJECTIVES: To estimate the attributable burden due to smoking in South Africa for 2000, 2006 and 2012. METHODS: We estimated attributable burden due to smoking for selected causes of death in South African (SA) adults aged ≥35 years for 2000, 2006 and 2012. We combined smoking prevalence results from 15 national surveys (1998 - 2017) and smoking impact ratios using national mortality rates. Relative risks between smoking and select causes of death were derived from local and international data. RESULTS: Smoking prevalence declined from 25.0% in 1998 (40.5% in males, 10.9% in females) to 19.4% in 2012 (31.9% in males, 7.9% in females), but plateaued after 2010. In 2012 tobacco smoking caused an estimated 31 078 deaths (23 444 in males and 7 634 in females), accounting for 6.9% of total deaths of all ages (17.3% of deaths in adults aged ≥35 years), a 10.5% decline overall since 2000 (7% in males; 18% in females). Age-standardised mortality rates (and disability-adjusted life years (DALYs)) similarly declined in all population groups but remained high in the coloured population. Chronic obstructive pulmonary disease accounted for most tobacco-attributed deaths (6 373), followed by lung cancer (4 923), ischaemic heart disease (4 216), tuberculosis (2 326) and lower respiratory infections (1 950). The distribution of major causes of smoking-attributable deaths shows a middle- to high-income pattern in whites and Asians, and a middle- to low-income pattern in coloureds and black Africans. The role of infectious lung disease (TB and LRIs) has been underappreciated. These diseases comprised 21.0% of deaths among black Africans compared with only 4.3% among whites. It is concerning that smoking rates have plateaued since 2010. CONCLUSION: The gains achieved in reducing smoking prevalence in SA have been eroded since 2010. An increase in excise taxes is the most effective measure for reducing smoking prevalence. The advent of serious respiratory pandemics such as COVID-19 has increased the urgency of considering the role that smoking cessation/abstinence can play in the prevention of, and post-hospital recovery from, any condition.

Estimating the changing burden of disease attributable to low levels of physical activity in South Africa for 2000, 2006 and 2012.

BACKGROUND: Physical activity is associated with a lower risk of cardiovascular outcomes, certain cancers and diabetes. The previous South African Comparative Risk Assessment (SACRA1) study assessed the attributable burden of low physical activity for 2000, but updated estimates are required, as well as an assessment of trends over time. OBJECTIVE: To estimate the national prevalence of physical activity by age, year and sex and to quantify the burden of disease attributable to low physical activity in South Africa (SA) for 2000, 2006 and 2012. METHODS: Comparative risk assessment methodology was used. Physical activity was treated as a categorical variable with four categories, i.e. inactive, active, very active and highly active. Prevalence estimates of physical activity levels, representing the three different years, were derived from two national surveys. Physical activity estimates together with the relative risks from the Global Burden of Disease, Injuries, and Risk Factors (GBD) 2016 study were used to calculate population attributable fractions due to inactive, active and very active levels of physical activity relative to highly active levels considered to be the theoretical minimum risk exposure (>8 000 metabolic equivalent of time (MET)-min/wk), in accordance with the GBD 2016 study. These were applied to relevant disease outcomes sourced from the Second National Burden of Disease Study to calculate attributable deaths, years of life lost, years lived with disability and disability adjusted life years (DALYs). Uncertainty analysis was performed using Monte Carlo simulation. RESULTS: The prevalence of physical inactivity (<600 METS) decreased by 16% and 8% between 2000 and 2012 for females and males, respectively. Attributable DALYs due to low physical activity increased between 2000 (n=194 284) and 2006 (n=238 475), but decreased thereafter in 2012 (n=219 851). The attributable death age-standardised rates (ASRs) declined between 2000 and 2012 from 60/100 000 population in 2000 to 54/100 000 population in 2012. Diabetes mellitus type 2 displaced ischaemic heart disease as the largest contributor to attributable deaths, increasing from 31% in 2000 to 42% in 2012. CONCLUSIONS: Low physical activity is responsible for a large portion of disease burden in SA. While the decreased attributable death ASR due to low physical activity is encouraging, this burden may be lowered further with an additional reduction in the overall prevalence of physical inactivity, in particular. It is concerning that the attributable burden for diabetes mellitus is growing, which suggests that existing non-communicable disease policies need better implementation, with ongoing surveillance of physical activity, and population- and community-based interventions are required in order to reach set targets.

Estimating the burden of disease attributable to ambient air pollution (ambient PM2.5 and ambient ozone) in South Africa for 2000, 2006 and 2012.

Background: Globally, a growing body of research has shown that ambient air pollution is one of the most critical environmental issues, especially in relation to human health. Exposure to ambient air pollution leads to serious health conditions such as lower respiratory infections, cancers, diabetes mellitus type 2, ischaemic heart disease, stroke and chronic obstructive pulmonary disease. Objectives: To estimate the burden of disease attributable to ambient air pollution in South Africa (SA) for the years 2000, 2006 and 2012. Methods: Comparative risk assessment method was used to determine the burden of disease due to two pollutants (particulate matter (PM2.5) and ambient ozone). Regionally optimised fully coupled climate chemistry models and surface air pollution observations were used to generate concentrations of PM2.5 and ozone for each SA Census Small Area Level, for the year 2012. For 2000 and 2006, population-weighted PM2.5and ozone were estimated, based on the 2012 results. Following the identification of disease outcomes associated with particulate matter with aerodynamic diameter <2.5 µm (PM2.5) and ozone exposure, the attributable burden of disease was estimated for 2000, 2006 and 2012. Furthermore, for the year 2012, the burden of disease attributable to ambient air pollution exposure was computed at provincial levels. Results: In 2012, approximately 97.6% of people in SA were exposed to PM2.5 at levels above the 2005 World Health Organization guideline: 10 µg/m3 annual mean. From 2000 to 2012, population-weighted annual average PM2.5 increased from 26.6 µg/m3 to 29.7 µg/m3, and ozone 6-month high 8-hour daily maximum increased from 64.4 parts per billion (ppb) to 72.1 ppb. At a national scale, in the year 2000, 15 619 (95% uncertainty interval (UI) 8 958 - 21 849) deaths were attributed to PM2.5 exposure, while 1 326 (95% UI 534 - 1 885) deaths were attributed to ozone. In 2006, an estimated 19 672 deaths (95% UI 11 526 - 27 086) were attributed to PM2.5, and a further 1 591 deaths (95% UI 651 - 2 236) to ozone exposure. In 2012, deaths attributed to PM2.5 were 19 507 (95% UI 11 318 - 27 111), and to ozone 1 734 (95% UI 727 - 2 399). Additionally, population-weighted provincial scale analysis showed that Gauteng Province had the highest number of attributable deaths due to both PM2.5 and ozone in 2012. Conclusion: The study showed that ambient air pollution exposure is an important health risk in SA, requiring both short- and long-term intervention. In the short term, the SA Ambient Air Quality Standards and industrial minimum emissions standards need to be enforced. In the longer term, to reduce air pollution and the associated disease burden, the combustion of fossil fuels as a source of energy for power generation and transportation, as well as industrial and domestic uses, needs to be replaced with clean renewable energy sources. In addition to local measures, when the southern African prevalent anticyclonic air dynamics that transport regionally emitted pollutants into SA (especially from biomass burning) are considered, it is also advisable to establish long-term regional co-operation in reducing air pollution.

Estimating the changing disease burden attributable to iron deficiency in South Africa, 2000, 2006 and 2012.

BACKGROUND: Worldwide, iron deficiency, and consequent iron-deficiency anaemia, remains the most common nutritional disorder. Iron-deficiency anaemia mostly affects young children and women of reproductive age, especially in Asia and Africa. Iron deficiency may contribute to disability directly or indirectly as a risk factor for other causes of death, and may rarely contribute to death. OBJECTIVES: To estimate the changing burden of disease attributable to iron deficiency in males and females (all ages) for the years 2000, 2006 and 2012 in South Africa (SA). METHODS: The comparative risk assessment methodology developed by the World Health Organization (WHO) and the Global Burden of Diseases, Injuries, and Risk Factors Studies was used to estimate the burden attributable to iron deficiency in SA for the years 2000, 2006 and 2012. We attributed 100% of the estimated iron-deficiency anaemia burden across all age groups by sex to iron deficiency. For maternal conditions, the attributable burden to iron deficiency was calculated using the counterfactual method and applied to all women of reproductive age. The population attributable fraction calculated for these selected health outcomes was then applied to local burden estimates from the Second SA National Burden of Disease Study (SANBD2). Age-standardised rates were calculated using WHO world standard population weights and SA mid-year population estimates. RESULTS: There was a slight decrease in the prevalence of iron-deficiency anaemia in women of reproductive age from ~11.9% in 2000 to 10.0% in 2012, although the prevalence of anaemia fluctuated over time (25.5% - 33.2%), with a peak in 2006. There has been a gradual decline in the number of deaths from maternal conditions attributable to iron deficiency in SA between 2000 (351 deaths (95% uncertainty interval (UI) 248 - 436)) and 2012 (307 deaths (95% UI 118 - 470)), with a peak in 2006 (452 deaths (95% UI 301 - 589)). Furthermore, our analysis showed a 26% decrease between 2000 and 2012 in the age-standardised burden rates from maternal conditions (truncated to 15 - 49 years) attributable to iron deficiency. Between 2000 and 2012, the age-standardised disability-adjusted life year (DALY) rate from iron-deficiency anaemia attributable to iron deficiency markedly decreased by 33% in males, and increased by 3% in females of all ages. Approximately 1.1 - 1.4% of all DALYs in SA from 2000 to 2012 were attributable to iron deficiency. CONCLUSION: Iron-deficiency anaemia prevalence can be markedly reduced if iron deficiency is eliminated. Hence it is essential to encourage, reappraise and strengthen the measures that have been put in place to address iron deficiency, especially in women of reproductive age and children.

Estimating the changing disease burden attributable to raised low-density lipoprotein cholesterol in South Africa for 2000, 2006 and 2012.

BACKGROUND: Low-density lipoprotein cholesterol (LDL-C) is the most important contributor to atherosclerosis, a causal factor for ischaemic heart disease (IHD) and ischaemic stroke. Although raised LDL-C is a key contributor to cardiovascular disease (CVD), the exact attributable disease risk in South Africa (SA) is unknown. The the first SA comparative risk assessment (SACRA1) study assessed the attributable burden of raised total cholesterol, and not specifically LDL-C. OBJECTIVES: To estimate the national mean serum LDL-C by age, year and sex and to quantify the burden of disease attributable to LDL-C in SA for 2000, 2006 and 2012. METHODS: The comparative risk assessment (CRA) method was used. Estimates of the national mean of LDL-C, representing the 3 different years, were derived from 14 small observational studies using a meta-regression model. A theoretical minimum risk exposure level (TMREL) of 0.7 - 1.3 mmol/L was used. LDL-C estimates together with the relative risks from the Global Burden of Disease Study 2017 were used to calculate a potential impact fraction (PIF). This was applied to IHD and ischaemic stroke estimates sourced from the Second National Burden of Disease Study. Attributable deaths, years of life lost, years lived with disability and disability-adjusted life years (DALYs) were calculated. Uncertainty analysis was performed using Monte Carlo simulation. RESULTS: LDL-C declined from 2.74 mmol/L in 2000 to 2.58 mmol/L in 2012 for males, while in females it declined from 3.05 mmol/L in 2000 to 2.91 mmol/L in 2012. The PIFs for LDL-C showed a slight decline over time, owing to the slight decrease in LDL-C levels. Attributable DALYs increased between 2000 (n=286 712) and 2006 (n=315 125), but decreased thereafter in 2012 (n=270 829). Attributable age-standardised death rates declined between 2000 and 2012 in both sexes: in males from 98 per 100 000 members of the population in 2000 to 78 per 100 000 in 2012, and in females from 81 per 100 000 in 2000 to 58 per 100 000 in 2012. CONCLUSIONS: Mean LDL-C levels were close to 3 mmol/L, which is the recommended level at which cholesterol-lowering treatment should be initiated for people at low and moderate risk for cardiovascular outcomes. The decreasing trend in the age-standardised attributable burden due to LDL-C is encouraging, but it can be lowered further with the introduction of additional population-based CVD prevention strategies. This study highlights the fact that high LDL-C concentration in relation to the TMREL in SA is responsible for a large proportion of the emerging CVD, and should be targeted by health planners to reduce disease burden.

Estimating the changing burden of disease attributable to high fasting plasma glucose in South Africa for 2000, 2006 and 2012.

BACKGROUND: Worldwide, higher-than-optimal fasting plasma glucose (FPG) is among the leading modifiable risk factors associated with all- cause mortality and disability-adjusted life years (DALYs) due to the direct sequelae of diabetes and the increased risk for cardiovascular and chronic kidney disease. OBJECTIVES: To report deaths and DALYs of health outcomes attributable to high FPG by age and sex for South Africa (SA) for 2000, 2006 and 2012. METHODS: Comparative risk assessment methodology was used to estimate the burden attributable to high FPG. A meta-regression analysis was performed using data from national and small-area studies to estimate the population distribution of FPG and diabetes prevalence. Attributable fractions were calculated for selected health outcomes and applied to local burden estimates from the second South African National Burden of Disease Study (SANBD2). Age-standardised rates were calculated using World Health Organization world standard population weights. RESULTS: We estimated a 5% increase in mean FPG from 5.31 (95% confidence interval (CI) 5.18 - 5.43) mmol/L to 5.57 (95% CI 5.41 - 5.72) mmol/L and a 75% increase in diabetes prevalence from 7.3% (95% CI 6.7 - 8.3) to 12.8% (95% CI 11.9 - 14.0) between 2000 and 2012. The age-standardised attributable death rate increased from 153.7 (95% CI 126.9 - 192.7) per 100 000 population in 2000 to 203.5 (95% CI 172.2 - 240.8) per 100 000 population in 2012, i.e. a 32.4% increase. During the same period, age-standardised attributable DALY rates increased by 43.8%, from 3 000 (95% CI 2 564 - 3 602) per 100 000 population in 2000 to 4 312 (95% CI 3 798 - 4 916) per 100 000 population in 2012. In each year, females had similar attributable death rates to males but higher DALY rates. A notable exception was tuberculosis, with an age-standardised attributable death rate in males double that in females in 2000 (14.3 v. 7.0 per 100 000 population) and 2.2 times higher in 2012 (18.4 v. 8.5 per 100 000 population). Similarly, attributable DALY rates were higher in males, 1.7 times higher in 2000 (323 v. 186 per 100 000 population) and 1.6 times higher in 2012 (502 v. 321 per 100 000 population). Between 2000 and 2012, the age-standardised death rate for chronic kidney disease increased by 98.3% (from 11.7 to 23.1 per 100 000 population) and the DALY rate increased by 116.9% (from 266 to 578 per 100 000 population). CONCLUSION: High FPG is emerging as a public health crisis, with an attributable burden doubling between 2000 and 2012. The consequences are costly in terms of quality of life, ability to earn an income, and the economic and emotional burden on individuals and their families. Urgent action is needed to curb the increase and reduce the burden associated with this risk factor. National data on FPG distribution are scant, and efforts are warranted to ensure adequate monitoring of the effectiveness of the interventions.

Estimating the changing burden of disease attributable to high body mass index in South Africa for 2000, 2006 and 2012.

BACKGROUND: A high body mass index (BMI) is associated with several cardiovascular diseases, diabetes and chronic kidney disease, cancers, and other selected health conditions. OBJECTIVES: To quantify the deaths and disability-adjusted life years (DALYs) attributed to high BMI in persons aged ≥20 years in South Africa (SA) for 2000, 2006 and 2012. METHODS: The comparative risk assessment (CRA) methodology was followed. Meta-regressions of the BMI mean and standard deviation from nine national surveys spanning 1998 - 2017 were conducted to provide estimates by age and sex for adults aged ≥20 years. Population attributable fractions were calculated for selected health outcomes using relative risks identified by the Global Burden of Disease Study (2017), and applied to deaths and DALY estimates from the second South African National Burden of Disease Study to estimate the burden attributed to high BMI in a customised Microsoft Excel workbook. Monte Carlo simulation-modelling techniques were used for the uncertainty analysis. BMI was assumed to follow a log-normal distribution, and the theoretical minimum value of BMI below which no risk was estimated was assumed to follow a uniform distribution from 20 kg/m2 to 25 kg/m2. RESULTS: Between 2000 and 2012, mean BMI increased by 6% from 27.7 kg/m2 (95% confidence interval (CI) 27.6 - 27.9) to 29.4 kg/m2 (95% CI 29.3 - 29.5) for females, and by 3% from 23.9 kg/m2 (95% CI 23.7 - 24.1) to 24.6 kg/m2 (95% CI 24.5 - 24.8) for males. In 2012, high BMI caused 58 757 deaths (95% uncertainty interval (UI) 46 740 - 67 590) or 11.1% (95% UI 8.8 - 12.8) of all deaths, and 1.42 million DALYs (95% UI 1.15 - 1.61) or 6.9% (95% UI 5.6 - 7.8) of all DALYs. Over the study period, the burden in females was ~1.5 - 1.8 times higher than that in males. Type 2 diabetes mellitus became the leading cause of death attributable to high BMI in 2012 (n=12 382 deaths), followed by hypertensive heart disease (n=12 146), haemorrhagic stroke (n=9 141), ischaemic heart disease (n=7 499) and ischaemic stroke (n=4 044). The age-standardised attributable DALY rate per 100 000 population for males increased by 6.6% from 3 777 (95% UI 2 639 - 4 869) in 2000 to 4 026 (95% UI 2 831 - 5 115) in 2012, while it increased by 7.8% for females from 6 042 (95% UI 5 064 - 6 702) to 6 513 (95% UI 5 597 - 7 033). CONCLUSION: Average BMI increased between 2000 and 2012 and accounted for a growing proportion of total deaths and DALYs. There is a need to develop, implement and evaluate comprehensive interventions to achieve lasting change in the determinants and impact of overweight and obesity, particularly among women.

Estimating the changing burden of disease attributable to unsafe water and lack of sanitation and hygiene in South Africa for 2000, 2006 and 2012.

BACKGROUND: The incidence of diarrhoeal disease is closely linked to socioeconomic and environmental factors, household practices and access to health services. South African (SA) district health information and national survey data report wide variation in the incidence and prevalence of diarrhoeal episodes in children under 5 years of age. These differentials indicate potential for reducing the disease burden through improvements in provision of water and sanitation services and changes in hygiene behaviour. OBJECTIVES: To estimate the burden of disease attributed to unsafe water, sanitation and hygiene (WASH) by province, sex and age group for SA in 2000, 2006 and 2012. METHODS: Comparative risk assessment methodology was used to estimate the disease burden attributable to an exposure by comparing the observed risk factor distribution with a theoretical lowest possible population distribution. The study adapts the original World Health Organization scenario-based approach for estimating diarrhoeal disease burden from unsafe WASH, by assigning different standards of household water and sanitation-specific geographical classification to capture SA living conditions in rural, urban and informal settlements. RESULTS: SA experienced an improvement in water and sanitation supply in eight of the nine provinces between 2001 and 2011, with the exception of Northern Cape Province. In 2011, 41% of South Africans lived with poor water and sanitation conditions; however, wide provincial inequalities exist. In 2012, it was estimated that 84.1% of all deaths due to diarrhoeal disease were attributable to unsafe WASH; this equates to 13 757 deaths (95% uncertainty interval (UI) 13 015 - 14 300). Of these diarrhoeal disease deaths, 48.2% occurred in children under 5 years of age, accounting for 13.9% of all deaths in this age group (95% UI 13.1 - 14.4). Between 2000 and 2012, the proportion of deaths attributable to diarrhoea reduced from 3.6% to 2.6%. Gauteng and Western Cape provinces experienced much lower WASHattributable death rates than the more rural, poorer provinces. CONCLUSION: Unsafe WASH remains an important risk factor for disease in SA, especially in children. High priority needs to be given to the provision of safe and sustainable sanitation and water facilities and promoting safe hygiene behaviours. The COVID-19 pandemic has reinforced the critical importance of clean water for preventing and containing disease.

Estimating the burden of disease attributable to household air pollution from cooking with solid fuels in South Africa for 2000, 2006 and 2012.

BACKGROUND: Household air pollution (HAP) due to the use of solid fuels for cooking is a global problem with significant impacts on human health, especially in low- and middle-income countries. HAP remains problematic in South Africa (SA). While electrification rates have improved over the past two decades, many people still use solid fuels for cooking owing to energy poverty. OBJECTIVES: To estimate the disease burden attributable to HAP for cooking in SA over three time points: 2000, 2006 and 2012. METHODS: Comparative risk assessment methodology was used. The proportion of South Africans exposed to HAP was assessed and assigned the estimated concentration of particulate matter with a diameter <2.5 µg/m3 (PM2.5) associated with HAP exposure. Health outcomes and relative risks associated with HAP exposure were identified. Population-attributable fractions and the attributable burden of disease due to HAP exposure (deaths, years of life lost, years lived with disability and disability-adjusted life years (DALYs)) for SA were calculated. Attributable burden was estimated for 2000, 2006 and 2012. For the year 2012, we estimated the attributable burden at provincial level. RESULTS: An estimated 17.6% of the SA population was exposed to HAP in 2012. In 2012, HAP exposure was estimated to have caused 8 862 deaths (95% uncertainty interval (UI) 8 413 - 9 251) and 1.7% (95% UI 1.6% - 1.8%) of all deaths in SA, respectively. Loss of healthy life years comprised 208 816 DALYs (95% UI 195 648 - 221 007) and 1.0% of all DALYs (95% UI 0.95% - 1.0%) in 2012, respectively. Lower respiratory infections and cardiovascular disease contributed to the largest proportion of deaths and DALYs. HAP exposure due to cooking varied across provinces, and was highest in Limpopo (50.0%), Mpumalanga (27.4%) and KwaZulu-Natal (26.4%) provinces in 2012. Age standardised burden measures showed that these three provinces had the highest rates of death and DALY burden attributable to HAP. CONCLUSION: The burden of disease from HAP due to cooking in SA is of significant concern. Effective interventions supported by legislation and policy, together with awareness campaigns, are needed to ensure access to clean household fuels and improved cook stoves. Continued and enhanced efforts in this regard are required to ensure the burden of disease from HAP is curbed in SA.

Estimating the burden of disease attributable to a diet low in fruit and vegetables in South Africa for 2000, 2006 and 2012.

BACKGROUND: Low intake of fruit and vegetables is associated with an increased risk of various non-communicable diseases, including major causes of death and disability such as cardiovascular disease, diabetes mellitus and cancers. Diets low in fruit and vegetables are prevalent in the South African (SA) population, and average intake is well below the internationally recommended threshold. OBJECTIVES: To estimate the burden of disease attributable to a diet low in fruit and vegetables by sex and age group in SA for the years 2000, 2006 and 2012. METHODS: We followed World Health Organization and Global Burden of Disease Study comparative risk assessment methodology. Population attributable fractions - calculated from fruit and vegetable intake estimated from national and local surveys and relative risks for health outcomes based on the current literature - were applied to the burden estimates from the second South African National Burden of Disease Study (SANBD2). Outcome measures included deaths and disability-adjusted life years (DALYs) lost from ischaemic heart disease, stroke, type 2 diabetes, and five categories of cancers. RESULTS: Between 2000 and 2012, the average intake of fruit of the SA adult population (≥25 years) declined by 7%, from 48.5 g/d (95% uncertainty interval (UI) 46.6 - 50.5) to 45.2 g/d (95% UI 42.7 - 47.6). Vegetable intake declined by 25%, from 146.9 g/d (95% UI 142.3 - 151.8) to 110.5 g/d (95% UI 105.9 - 115.0). In 2012, these consumption patterns are estimated to have caused 26 423 deaths (95% UI 24 368 - 28 006), amounting to 5.0% (95% UI 4.6 - 5.3%) of all deaths in SA, and the loss of 514 823 (95% UI 473 508 - 544 803) healthy life years or 2.5% (95% UI 2.3 - 2.6%) of all DALYs. Cardiovascular disease comprised the largest proportion of the attributable burden, with 83% of deaths and 84% of DALYs. Age-standardised death rates were higher for males (145.1 deaths per 100 000; 95% UI 127.9 - 156.2) than for females (108.0 deaths per 100 000; 95% UI 96.2 - 118.1); in both sexes, rates were lower than those observed in 2000 (-9% and -12%, respectively). CONCLUSION: Despite the overall reduction in standardised death rates observed since 2000, the absolute burden of disease attributable to inadequate intake of fruit and vegetables in SA remains of significant concern. Effective interventions supported by legislation and policy are needed to reverse the declining trends in consumption observed in most age categories and to curb the associated burden.

Overview.

BACKGROUND: South Africa (SA) faces multiple health challenges. Quantifying the contribution of modifiable risk factors can be used to identify and prioritise areas of concern for population health and opportunities for health promotion and disease prevention interventions. OBJECTIVE: To estimate the attributable burden of 18 modifiable risk factors for 2000, 2006 and 2012. METHODS: Comparative risk assessment (CRA), a standardised and systematic approach, was used to estimate the attributable burden of 18 risk factors. Risk exposure estimates were sourced from local data, and meta-regressions were used to model the parameters, depending on the availability of data. Risk-outcome pairs meeting the criteria for convincing or probable evidence were assessed using relative risks against a theoretical minimum risk exposure level to calculate either a potential impact fraction or population attributable fraction (PAF). Relative risks were sourced from the Global Burden of Disease, Injuries, and Risk Factors (GBD) study as well as published cohort and intervention studies. Attributable burden was calculated for each risk factor for 2000, 2006 and 2012 by applying the PAF to estimates of deaths and years of life lost from the Second South African National Burden of Disease Study (SANBD2). Uncertainty analyses were performed using Monte Carlo simulation, and age-standardised rates were calculated using the World Health Organization standard population. RESULTS: Unsafe sex was the leading risk factor across all years, accounting for one in four DALYs (26.6%) of the estimated 20.6 million DALYs in 2012. The top five leading risk factors for males and females remained the same between 2000 and 2012. For males, the leading risks were (in order of descending rank): unsafe sex; alcohol consumption; interpersonal violence; tobacco smoking; and high systolic blood pressure; while for females the leading risks were unsafe sex; interpersonal violence; high systolic blood pressure; high body mass index; and high fasting plasma glucose. Since 2000, the attributable age-standardised death rates decreased for most risk factors. The largest decrease was for household air pollution (-41.8%). However, there was a notable increase in the age-standardised death rate for high fasting plasma glucose (44.1%), followed by ambient air pollution (7%). CONCLUSION: This study reflects the continued dominance of unsafe sex and interpersonal violence during the study period, as well as the combined effects of poverty and underdevelopment with the emergence of cardiometabolic-related risk factors and ambient air pollution as key modifiable risk factors in SA. Despite reductions in the attributable burden of many risk factors, the study reveals significant scope for health promotion and disease prevention initiatives and provides an important tool for policy makers to influence policy and programme interventions in the country.

Estimating the changing burden of disease attributable to high sodium intake in South Africa for 2000, 2006 and 2012.

BACKGROUND: Elevated sodium consumption is associated with increased blood pressure, a major risk factor for cardiovascular and chronic kidney disease. OBJECTIVES: To quantify the deaths and disability-adjusted life years (DALYs) attributed to high sodium intake in persons aged ≥25 years in South Africa (SA) for 2000, 2006 and 2012. METHODS: Comparative risk assessment (CRA) methodology was used and population attributable fractions (PAFs) of high sodium intake, mediated through high blood pressure (BP), for cardiovascular and chronic kidney disease were estimated. This was done by taking the difference between the PAF for elevated systolic BP (SBP) based on the estimated SBP level in the population and the PAF based on the estimated SBP that would result if sodium intake levels were reduced to the theoretical minimum risk exposure level (1 g/day) according to population group and hypertension categories. A meta-regression based on data from nine national surveys conducted between 1998 and 2017 was used to estimate the prevalence of hypertension by age, sex and population group. Relative risks identified from international literature were used and the difference in PAFs was applied to local burden estimates from the second South African National Burden of Disease Study. Age-standardised rates were calculated using World Health Organization (WHO) standard population weights. The attributable burden was also estimated for 2012 using an alternative target of 2 g/day proposed in the National Strategic Plan for the Prevention and Control of Non-communicable Diseases (NSP). RESULTS: High sodium intake as mediated through high SBP was estimated to cause 8 071 (95% uncertainty interval (UI) 6 542 - 15 474) deaths in 2012, a drop from 9 574 (95% UI 8 158 - 16 526) in 2006 and 8 431 (95% UI 6 972 - 14 511) in 2000. In 2012, ischaemic heart disease caused the highest number of deaths in persons (n=1 832), followed by haemorrhagic stroke (n=1 771), ischaemic stroke (n=1 484) and then hypertensive heart disease (n=1 230). Ischaemic heart disease was the highest contributor to deaths for males (27%), whereas for females it was haemorrhagic stroke (23%). In 2012, 1.5% (95% UI 1.3 - 2.9) of total deaths and 0.7% (95% UI 0.6 - 1.2) of total DALYs were attributed to high sodium intake. If the NSP target of <2 g/day sodium intake had been achieved in 2012, ~2 943 deaths and 48 870 DALYs would have been averted. CONCLUSION: Despite a slight decreasing trend since 2006, high sodium intake mediated through raised BP accounted for a sizeable burden of disease in 2012. Realising SA's target to reduce sodium intake remains a priority, and progress requires systematic monitoring and evaluation.

Estimating the changing burden of disease attributable to interpersonal violence in South Africa for 2000, 2006 and 2012.

BACKGROUND: South Africa (SA)'s high rate of interpersonal violence persists as a leading public health problem for the country. The first South African Comparative Risk Assessment Study (SACRA1) in 2000 quantified the long-term mental and physical health burden attributable to interpersonal violence by supplementing the direct injury burden of disease attributable to interpersonal violence injuries with the substantial contribution of mental health, behavioural and reproductive health consequences accruing from exposure to intimate partner violence (IPV) and child sexual abuse. OBJECTIVES: To revise and improve these estimates by including the additional burden from other forms of child maltreatment, community violence, sexual violence by non-partners, and bullying victimisation in SA for 2000, 2006 and 2012, and trends over time. METHODS: We used comparative risk assessment methods to calculate population attributable fractions (PAFs) for interpersonal violence. This method requires inputs on the prevalence of exposure to the interpersonal violence risk factor subtypes, namely child maltreatment, bullying, IPV, sexual violence by non-partners and other community violence; the burden of related health outcomes (mortality and morbidity); and relative risks of health outcomes in individuals exposed to the risk factor v. those unexposed. We estimated the PAF for the combinations of all interpersonal violence subtypes together to estimate the burden attributable to interpersonal violence overall for 2000, 2006 and 2012. RESULTS: Between 2000 and 2012, there was a decrease in interpersonal violence age-standardised attributable death rates from 100 to 71 per 100 000. In the second South African Comparative Risk Assessment Study (SACRA2), estimates of the attributable disability-adjusted life years (DALYs) for interpersonal violence for the year 2000 were revised, from 1.7 million to 2 million DALYs, taking into account attributable mortality and disability from additional forms of violence. There was a decrease in DALYs attributable to interpersonal violence from 2 million in 2000 to 1.75 million in 2012, accounting for 8.5% of the total burden for SA, ranking second highest, after unsafe sex, among 18 risk factors evaluated in 2012. CONCLUSION: Overall, interpersonal violence-attributable DALYs decreased substantially but remain high. The reduction in age-standardised attributable death rates indicates that some policy and social intervention aspects are effective. Further strengthening of existing laws pertaining to interpersonal violence, and other prevention measures, are needed to intensify the prevention of violence, particularly gender-based violence. Additional forms of violence included in this analysis have improved our understanding of the interpersonal violence burden, but the attributable burden in males, although exceedingly high, remains an underestimate. There is a need to improve the epidemiological data on prevalence and risks for the different types of interpersonal violence, particularly for males.

Estimating the changing burden of disease attributable to childhood stunting, wasting and underweight in South Africa for 2000, 2006 and 2012.

BACKGROUND: National estimates of childhood undernutrition display uncertainty; however, it is known that stunting is the most prevalent deficiency. Child undernutrition is manifest in poor communities but is a modifiable risk factor. The intention of the study was to quantify trends in the indicators of child undernutrition to aid policymakers. OBJECTIVES: To estimate the burden of diseases attributable to stunting, wasting and underweight and their aggregate effects in South African (SA) children under the age of 5 years during 2000, 2006 and 2012. METHODS: The study applied comparative risk assessment methodology. Data sources for estimates of prevalence and population distribution of exposure in children under 5 years were the National Food Consumption surveys and the SA National Health and Nutrition Examination Survey conducted close to the target year of burden. Childhood undernutrition was estimated for stunting, wasting and underweight and their combined 'aggregate effect' using the World Health Organization (WHO) 2006 standard. Population-attributable fractions for the disease outcomes of diarrhoea, lower respiratory tract infections, measles and protein-energy malnutrition were applied to SA burden of disease estimates of deaths, years of life lost, years lived with a disability and disability-adjusted life years for 2000, 2006 and 2012. RESULTS: Among children aged under 5 years between 1999 and 2012, the distribution of anthropometric measurements <‒2 standard deviations from the WHO median showed little change for stunting (28.4% v. 26.6%), wasting (2.6% v. 2.8%) and underweight (7.6% v. 6.1%). In the same age group in 2012, attributable deaths due to wasting and aggregated burden accounted for 21.4% and 33.2% of the total deaths, respectively. Attributable death rates due to wasting and aggregate effects decreased from ~310 per 100 000 in 2006 to 185 per 100 000 in 2012. CONCLUSION: The study shows that reduction of childhood undernutrition would have a substantial impact on child mortality. We need to understand why we are not penetrating the factors related to nutrition of children that will lead to reducing levels of stunting.

The impact of the SARS-CoV-2 epidemic on mortality in South Africa in 2020.

BACKGROUND: The impacts on mortality of both the SARS-CoV-2 epidemic and the interventions to manage it differ between countries. The Rapid Mortality Surveillance System set up by the South African Medical Research Council based on data from the National Population Register (NPR) provides a means of tracking this impact on mortality in South Africa. OBJECTIVES: To report on the change in key metrics of mortality (numbers of deaths, life expectancy at birth, life expectancy at age 60, and infant, under-5, older child and adolescent, young adult, and adult mortality) over the period 2015 - 2020. The key features of the impact are contrasted with those measured in other countries. METHODS: The numbers of registered deaths by age and sex recorded on the NPR were increased to account for both registered deaths that are not captured by the NPR and an estimate of deaths not reported. The estimated numbers of deaths together with estimates of the numbers in the population in the middle of each of the years were used to produce life tables and calculate various indicators. RESULTS: Between 2019 and 2020, the number of deaths increased by nearly 53 000 (65% female), and life expectancy at birth fell by 1 year for females and by only 2.5 months for males. Life expectancy at age 60 decreased by 1.6 years for females and 1.2 years for males. Infant mortality, under-5 mortality and mortality of children aged 5 - 14 decreased by 22%, 20% and 10%, respectively, while that for older children and adolescents decreased by 11% for males and 5% for females. Premature adult mortality, the probability of a 15-year-old dying before age 60, increased by 2% for males and 9% for females. CONCLUSIONS: COVID-19 and the interventions to manage it had differential impacts on mortality by age and sex. The impact of the epidemic on life expectancy in 2020 differs from that in most other, mainly developed, countries, both in the limited decline and also in the greater impact on females. These empirical estimates of life expectancy and mortality rates are not reflected by estimates from agencies, either because agency estimates have yet to be updated for the impact of the epidemic or because they have not allowed for the impact correctly. Trends in weekly excess deaths suggest that the drop in life expectancy in 2021 will be greater than that in 2020.

Unnatural deaths, alcohol bans and curfews: Evidence from a quasi-natural experiment during COVID-19.

BACKGROUND: Coronavirus disease-19 (COVID-19) restrictions, particularly relating to the sale of alcohol and hours of curfew, have had a marked effect on the temporal pattern of unnatural deaths in South Africa. Methods. Death data were collected over 68 weeks from January 2020 to April 2021, together with information on the nature of restrictions (if any) on the sale of alcohol, and hours of curfew. Data were analysed using a simple ordinary least square (OLS) regression model to estimate the relative contribution of restrictions on the sale of alcohol and hours of curfew to the pattern of excess unnatural deaths. Results. The complete restriction on the sale of alcohol resulted in a statistically significant reduction in unnatural deaths regardless of the length of curfew. To the contrary, periods where no or limited restrictions on alcohol were in force had no significant effect, or resulted in significantly increased unnatural deaths. Conclusions. The present study highlights an association between alcohol availability and the number of unnatural deaths and demonstrates the extent to which those deaths might be averted by disrupting the alcohol supply. While this is not a long-term solution to addressing alcohol-related harm, it further raises the importance of implementing evidence-based alcohol control measures.