A research report submitted to the Faculty of Health Sciences, University of the Witwatersrand, Johannesburg, in partial fulfilment of the requirements of the award of the Masters in Science in Epidemiology in the field of Population-based Field Epidemiology April 2012 / Objective: To determine the causes of death and associated risk factors in women of reproductive age in rural South Africa. . Methods: The study population comprised all female members aged 15-49 years of 11 000 households of a rural South African Health and Demographic Surveillance Site from 2000-2009. Deaths and person-years of observation (pyo) were determined for individuals between 01 January 2000 and 31 December 2009. Cause of death was ascertained by verbal autopsy interviews, based on ICD-10 coding; cause of death were broadly categorized as AIDS/TB causes, Non-communicable causes, Communicable/maternal/perinatal/nutrition causes, Injuries and another category of undetermined (unknown) causes of death. Overall and cause specific mortality rates (MR) with 95% confidence intervals (CI) were calculated. Cox proportional hazard regression (HR, 95% CI) was used to determine risk factors associated with overall and cause-specific mortality. Results: 42703 eligible women were included; 3098 deaths were reported for 212607 person-years (pyo) of observation. Overall MR was 14.57 deaths/1000 pyo (CI;14.07-15.09), increasing from 2000-2003 (2003: MR;18.15, CI;16.41-20.08) and subsequently decreasing (2009: MR; 9.59, CI;8.43-10.91) after introduction of antiretroviral treatment (ART) for HIV in public health system facilities in South Africa in 2004. Mortality was highest for AIDS/TB (MR;10.66, CI;10.23-11.11) and the cause of death for 73.1% of all recorded deaths. Maternal mortality was 0.07 (CI; 0.04-0.11). Women aged 30-34 years had the highest MR due to AIDS/TB (MR; 20.34/1000 pyo), women aged 45-49 years due to other causes (MR; 4.29/ 1000 pyo). v In multivariable analyses, external migration status was associated with increased hazards of all cause mortality (HR; 1.87, CI; 1.56-2.26) and other causes of mortality (HR; 1.782, CI; 1.24-2.57). Self reported poor health was significantly associated with increased hazards of all cause mortality (HR; 11.052, CI; 4.24-28.82) but not with mortality due to other causes. Positive HIV status was associated with increased hazards of all cause mortality (HR; 8.53, CI; 6.81-10.67) and other causes of mortality (HR; 2.84, CI; 1.97- 4.09). Conclusion. AIDS was the main cause of death in the current study, with mortality rates declining since introduction of ART for HIV in public health facilities in the surveillance area in 2004. Further ART roll-out, increased community awareness and sensitisation messages are still needed to reduce the spread of HIV and other sexually transmitted diseases.

The effect of anti-retroviral therapy (ART) rollout can be measured in a number of ways including treatment coverage, behaviour change and the emergence of resistance. However, changes in population mortality are undoubtedly the most important measurable effect. Objectives: To describe trends in child and adult all-cause mortality versus HIV/AIDS related mortality before and after ART rollout; and to identify significant clusters of child and adult all-cause mortality versus HIV/AIDS related mortality in space-time, during pre and post ART rollout eras at Agincourt and Dikgale health and demographic surveillance systems (HDSSs) in South Africa. Design: Mortality data were extracted from both the Agincourt and Dikgale HDSSs for the period 1996–2010. Mortality rates by age group, year and village were calculated assuming a Poisson distribution and using precise person-years as the denominator. The Kulldorff spatial scan statistic was used to test for clusters of age group all-cause and HIV-related mortality both in space and time. Clusters were mapped using Quantum geographic information systems (GIS) software. Results: Both HIV-related and all-cause mortality decreased gradually over the years after the introduction of ART in 2007 for the two HDSS sites. Several statistically significant clusters of higher all-cause and HIV-related mortality were identified both in space and time. In the Agincourt HDSS, specific areas were consistently identified as high risk areas; namely, the east/south-east corner and upper central to west regions, pre ART. In the Dikgale HDSS, no significant clusters were identified using the spatial only analysis but one significant cluster, located towards the north of the Dikgale HDSS site, was identified using the space-time scanning, post ART. In Agincourt, no significant clusters of mortality were detected after the introduction of ART whereas in Dikgale, a significant cluster for all-cause mortality in the under-five age group was detected for the years after the introduction of ART. Conclusion: This work revealed the existence of spatio-temporal clusters of both child and adult mortality at the Agincourt and Dikgale HDSSs and that the introduction of ART had a substantial influence in reducing both HIV-related and all-cause mortality in rural South Africa. There is need though to take into account socio-demographic characteristics so as to determine fundamental risk factors influencing these spatio-temporal HIV-related and all-cause mortality patterns.

[Truncated abstract. Please see pdf format for complete text.] Background : The excess burden of mortality born by young Indigenous Australians and the disparity in infant and childhood mortality between Indigenous and non-Indigenous Australians have been well documented. The accuracy and completeness of national data describing the health of Indigenous Australians is inconsistent. The Western Australia (WA) Maternal and Child Health Research Database (MCHRDB), is a linked total population database that includes perinatal maternal and infant data, and infant and childhood morbidity and mortality data. Overall, these data are more than 99% complete, with a similar high level of completeness and validity for Indigenous Western Australians. Aim : The aim of this thesis is to measure Indigenous infant (0 to <1 year) and childhood (>=1 to <19 years) mortality and the disparity between Indigenous and non-Indigenous infants and children in WA for birth cohorts from 1980 to 1997 inclusive. To achieve this aim a number of secondary aims were identified, including the measurement of certain maternal and infant variables, and the age-specific, all-cause and cause-specific mortality for WA infants and children. Method : The study comprises a longitudinal birth cohort study, the primary data source being the MCHRDB. Data included on the MCHRDB are complete for all births in WA from 1980 onwards, with new birth cohorts linked on an annual basis. Maternal and infant variables and the geographical location of the residence and the time of birth and death were included in the descriptive and multivariate analyses. Each infant and childhood death was coded using a three-digit code developed primarily for research purposes. The descriptive analyses of mortality referred to the probability of dying in infancy and in childhood as the cumulative mortality risk (CMR), for various diseases and various population subgroups. Age-specific childhood rates were also calculated. The results of multivariate analyses included the fitting of Cox and Poisson regression models, and estimates of effect were represented as hazard ratios (Cox regression) and relative rates (Poisson regression). Results : Between 1980 and 1997, births to Indigenous mothers accounted for 6% of total WA births. Approximately 46% of Indigenous births were to mothers living in a remote location compared to 9% of non-Indigenous births. Indigenous mothers gave birth at an earlier age (30% of births were to teenage mothers compared to 6% of non-Indigenous births), and were more likely to be single than non-Indigenous mothers (40% Indigenous, 9% non-Indigenous). Indigenous infants had more siblings, were born at an earlier gestation and with a lower birth weight and percentage of expected birth weight. The CMR for Indigenous infants was 22 per 1000 live births compared with 6.7 for non- Indigenous infants, a relative risk (RR) of 3.3 (95%CI 3.0, 3.6). While there was a decrease in the CMR over the birth year groups for both populations, the disparity between the rate of Indigenous and non-Indigenous infant mortality increased. The Indigenous postneonatal (>28 to 365 days) mortality rate (11.7 per 1,000 neonatal survivors) was higher than the neonatal (0 to 28 days) mortality rate (10.3 per 1,000 live births). This profile differed from that for non-Indigenous infants, where the neonatal mortality rate (4.3 per 1,000 live births) was nearly twice that of the postneonatal mortality rate (2.4 per 1,000 neonatal survivors). The main causes of infant mortality among Indigenous infants were potentially preventable. These causes were infection followed by Sudden Infant Death Syndrome (SIDS), which differed from the main causes for non-Indigenous infants, sequelae of prematurity and birth defects. The CMR attributable to SIDS increased over the years amongst Indigenous infants and decreased significantly over the years in the non-Indigenous population. Furthermore, the disparity in mortality between the two populations increased and, in 1995 to 1997, was over seven times higher amongst Indigenous infants. The CMR was highest amongst infants living in remote locations for all causes of death except for Indigenous deaths attributable to SIDS, where the risk of death was highest amongst infants living in metropolitan locations. With the exception of infection, there was no difference in cause-specific mortality amongst Indigenous infants according to geographical location. Indigenous infants living in a remote location were at a significantly increased risk of death due to infection compared with their peers living in a rural or metropolitan location. The risk of death for Indigenous children was more than three times higher than for non-Indigenous children. This risk was significantly increased when most of the perinatal maternal and infant variables were considered.

This epidemiological study is a longitudinal descriptive review of the mortuary register at Benedictine Hospital, with an analysis of the trends which emerge. The descriptive component describes mortality at Benedictine Hospital during the years 1995- 2001. It describes both the actual numbers of deaths which occurred according to each sex and age group, and the causes of death as recorded in the mortuary register. The purpose of this study was twofold. First it was desired to raise AIDS awareness in the district by examining the effects of the AIDS epidemic on mortality. Second as the new district health system was being established, it was desired to develop a baseline of mortality information to be utilized for management in the Nongoma Local Municipality. In the trend analysis component of the study, first, it is assumed that most of the deaths occurred at Benedictine Hospital as it is the only health facility which handles severe illness in the Nongoma Local Municipality; therefore the number of deaths within the hospital and the population of Nongoma were used to calculate Age Specific (ASMRs) and Cause Specific Mortality Rates (CSMRs). Secondly an analysis of the age and sex distribution of deaths, ASMRs, the distribution of causes of death, and CSMRs was done. Two research questions were posed. The first research question was, has there been any change in the age distribution of death? It was demonstrated that while there was an 80% increase in the number of deaths, and although deaths increased in every age group except for the neonatal group, 80% of the increase was in the young adult ages particularly in the 20 through 39 years old age groups. By 2001 these groups were recording the largest number of deaths, 179 male deaths and 133 female deaths in the 30 through 39 years old group. Also the ASMRs of young adults had increased three to four times. The second research question was, has there been any change in the distribution of causes of death? It was demonstrated that the infectious diseases which caused the largest numbers of deaths, pulmonary tuberculosis caused 353 deaths, pneumonia 250, gastroenteritis acute and chronic 203, retro-viral disease 66, and meningitis 59, were six of the top seven causes of death in 2001. Chronic gastroenteritis, retro-viral disease, and meningitis had strengthened their position moving from the second ten into the top seven. Only trauma which was in the top five was not an infectious disease. Infectious diseases increased their share of the burden of disease from 36% in 1995 to 57% in 2001. While CSMRs for trauma and the type II non-communicable diseases were basically stable or falling, those of the infectious diseases increased three to four times. It is estimated that because the mortality pattern is similar to that of AIDS deaths in South Africa and Zimbabwe, that because it is young adult mortality that has increased and that it is infectious diseases which have increased that about 50% of mortality in Nongoma is due to AIDS. Recommendations are put forward as to how to disseminate this information and also how to institute a system to carry on monitoring mortality in Nongoma. / Thesis (M.Med.)-University of Natal, Durban, 2003.

Mortality statistics are much used although their accuracy is often questioned. Producers of mortality statistics check for errors in death certification but current methods only capture obvious mistakes. This thesis investigates whether non-obvious errors can be found by linking death certificates to hospital discharge data. Data: 69,818 deaths in Sweden 1995. Paper I: Analysing differences between the underlying cause of death from the death certificate (UC) and the main discharge condition from the patient’s last hospitalization (MDC). Paper II: Testing whether differences can be explained by ICD definitions of UC and MDC. Paper III: Surveying methods in 44 current studies on the accuracy of death certificates. Paper IV: Checking death certificates against case summaries for: i) 573 deaths where UC and MDC were the same or the difference could be explained; ii) 562 deaths where the difference could not be explained. Results: In 54% of deaths the MDC differed from the UC. Almost two-thirds of the differences were medically compatible since the MDC might have developed as a complication of the UC. Of 44 recent evaluation studies, only 8 describe the methods in such detail that the study could be replicated. Incompatibility between MDC and UC indicates a four-fold risk that the death certificate is inaccurate. For some diagnostic groups, however, death certificates are often inaccurate even when the UC and MDC are compatible. Conclusion: Producers of official mortality statistics could reduce the number of non-obvious errors in the statistics by collecting additional information on incompatible deaths and on deaths in high-risk diagnostic groups. ICD conventions contribute to the quality problem since they presuppose that all deaths are due to a single underlying cause. However, in an ageing population an increasing number of deaths are due to an accumulation of etiologically unrelated conditions.

<p>Mortality statistics are much used although their accuracy is often questioned. Producers of mortality statistics check for errors in death certification but current methods only capture obvious mistakes. This thesis investigates whether non-obvious errors can be found by linking death certificates to hospital discharge data.</p><p>Data: 69,818 deaths in Sweden 1995. Paper I: Analysing differences between the underlying cause of death from the death certificate (UC) and the main discharge condition from the patient’s last hospitalization (MDC). Paper II: Testing whether differences can be explained by ICD definitions of UC and MDC. Paper III: Surveying methods in 44 current studies on the accuracy of death certificates. Paper IV: Checking death certificates against case summaries for: i) 573 deaths where UC and MDC were the same or the difference could be explained; ii) 562 deaths where the difference could not be explained.</p><p>Results: In 54% of deaths the MDC differed from the UC. Almost two-thirds of the differences were medically compatible since the MDC might have developed as a complication of the UC. Of 44 recent evaluation studies, only 8 describe the methods in such detail that the study could be replicated. Incompatibility between MDC and UC indicates a four-fold risk that the death certificate is inaccurate. For some diagnostic groups, however, death certificates are often inaccurate even when the UC and MDC are compatible.</p><p>Conclusion: Producers of official mortality statistics could reduce the number of non-obvious errors in the statistics by collecting additional information on incompatible deaths and on deaths in high-risk diagnostic groups. ICD conventions contribute to the quality problem since they presuppose that all deaths are due to a single underlying cause. However, in an ageing population an increasing number of deaths are due to an accumulation of etiologically unrelated conditions.</p>

No abstract available. / Thesis (M.P.H.)-University of KwaZulu-Natal, Durban, 2006.

Background Recent gains in child survival are being threatened by the RN epidemic. Monitoring child mortality rates is essential to understanding the impact of the epidemic, but is constrained by a lack of data. A community-based survey was used to determine child mortality rates in a rural area with high RN prevalence, located in the Rlabisa subdistrict of the KwaZulu-Natal Province, South Africa. ii. Methods The study was conducted between 1 January 2000 and 31 December 2002 on deaths in children under the age of 15 years. Children were followed up through 4-monthly home visits. Cause of death was ascertained by verbal autopsy. Rates were calculated using Poisson methods. iii. Results Infant and under-5 mortality ratios were respectively, 59.6 and 97.1 deaths per 1000 live births. Infant and under-5 mortality rates were, respectively, 67.5 and 21.1 deaths per 1000 child-years. RN/AIDS was attributed to 41% of deaths in the under-5 age group, with a mortality rate of 8.6 per 1000 person years. Lower respiratory infections caused an estimated 24.9 deaths per 1000 person years in children under 1 year of age. iv. Discussion In rural South Africa, infant and child mortality levels are high, with RN/AIDS estimated as the single largest cause ofdeath. Improving the coverage of interventions known to impact on child mortality is required urgently. / Thesis (M.Med.)-University of KwaZulu-Natal, Durban, 2007.

Infant mortality is generally regarded as an indicator of the standard of health and it is probably one of the best measures of the general socio-economic conditions in a society. There are indications world wide (including South Africa), of slowing down in the decrease of infant mortality rate, and a continued existence of wide variation in infant mortality levels. Conversely, it has also been noted that some relatively poor countries have managed to achieve low levels of infant mortality as often experienced by wealthier nations. The case for more attention to be paid to understanding the changing dynamics of infant mortality is compelling. Using the Demographic and Health Survey, which was conducted in Transkei in 1987, this study has looked into levels and patterns of infant mortality in Transkei and factors influencing them. The accuracy of the data was determined by using the Myers' Blended Index and Life Table Model. Response errors usually arise in developing countries from omission of vital events and misstatement of dates and age. A direct method of estimation was used to determine the level of infant mortality, in Transkei, during the period 1984-1986. Socio-economic and medical factors influencing infant mortality levels were identified using univariate analysis and multiple logistic function. The results of this study indicate that these factors significantly influence infant mortality in Transkei. In particular, income, womens' employment, duration of breastfeeding, age of the mother when giving birth, spacing of births and vaccination are important factors influencing infant mortality in the region. / Thesis (M.A.)-University of Natal, Durban, 1992.

Les statistiques de mortalité permettent à l'OMS de mesurer l'ampleur des problèmes de santé dans les pays. Leur fiabilité est fonction de la qualité du certificat de décès. Il ressort que plus de 25% des données des pays Africains ne sont pas utilisables. La principale raison de ce déficit d'information serait la non-performance des systèmes et outils d'enregistrement des données. Notre travail avait pour objectif d'étudier un modèle d'enregistrement des décès qui tienne compte à la fois des normes de l'OMS et des réalités de l'Afrique. Nous avons cherché à savoir comment la certification des causes de décès pouvait améliorer la production des statistiques de mortalité en Afrique. Comme méthodologie, nous avons utilisé une recherche interventionnelle combinée à une démarche projet et à une approche épidémiologique. L'étude a eu comme résultat la conception et l'implémentation d'un modèle à trois scénarii de déploiement en fonction du niveau d'équipement des établissements de soins en TIC et en personnel qualifié. Le système a été validé et implémenté au CHU de Bobo Dioulasso. Les acteurs ont été formés et le dispositif fonctionne depuis le 1er janvier 2014. Une première évaluation du modèle a été réalisée après trois mois de fonctionnement. La mise en place d'un comité de suivi et d'un plan annuel de formation, l'implication du Ministère de la santé et de l'Organisation Ouest Africaine de la Santé, l'assistance technique de la Direction Générale de la Modernisation de l'Etat Civil et du CepiDc (France) sont des éléments d'appropriation, de pérennisation et d'espoir. Cependant, l'enregistrement des décès survenus hors des hôpitaux reste un autre défi à relever. / Mortality statistics are basic data the WHO employs to measure health problems in different countries. However, their reliability depends on the quality of death data collected by different doctors. It appears, however, that over 25% African data are of no use because they are not available on time or lack quality. The main reason for this lack of information could be the nonperformance of data logging systems and tools. Our work aimed at investigating a death registration model taking into account both WHO's standards and the realities of Africa. We tried to understand if certification of death causes could improve mortality statistics production in Africa. Our methodology was based on a combination of interventional research, project process, and an epidemiological approach. The study resulted in the design and implementation of a three scenarios model, depending on ICT equipment and qualified staff level of health care facilities. The system has been validated and implemented in the University Hospital of Bobo Dioulasso. All the actors were trained and the device operates since January 1st, 2014, after the training of the actors. A first evaluation of the model was performed after three months of operation. The establishment of a monitoring committee and of an annual training plan, the involvement of the Ministry of Health and of the West African Health Organization, the technical assistance of CepiDc (France) and of the General Directorate for Modernization of Civil Status, are elements of appropriation, sustainability and hope. However, deaths registration outside hospitals remains an active challenge.