- Record: found
- Abstract: found
- Article: found
Demographic patterns and trends in Central Ghana: baseline indicators from the Kintampo Health and Demographic Surveillance System
research-article
Author(s):
Seth Owusu-Agyei
1
,
* ,
Obed Ernest A. Nettey
1 ,
Charles Zandoh
1 ,
Abubakari Sulemana
1 ,
Robert Adda
1 ,
Seeba Amenga-Etego
1 ,
Cheikh Mbacke
2
Publication date (Electronic):
20 December 2012
Journal:
Global Health Action
Publisher:
Co-Action Publishing
Keywords:
Kintampo,
demographic surveillance,
population,
Ghana,
demographic rates,
fertility,
mortality,
migration
Read this article at
There is no author summary for this article yet. Authors can add summaries to their articles on ScienceOpen to make them more accessible to a non-specialist audience.
Abstract
Background
The dearth of health and demographic data in sub-Saharan Africa from vital registration systems and its impact on effective planning for health and socio-economic development is widely documented. Health and Demographic Surveillance Systems have the capacity to address the dearth of quality data for policy making in resource-poor settings.
Objective
This article demonstrates the utility of the Kintampo Health and Demographic Surveillance System (KHDSS) by showing the patterns and trends of population change from 2005 to 2009 in the Kintampo North Municipality and Kintampo South districts of Ghana through data obtained from the KHDSS biannual update rounds.
Design
Basic demographic rates for fertility, mortality, and migration were computed by year. School enrolment was computed as a percentage in school by age and sex for 6–18 year-olds. Socio-economic status was derived by use of Principal Components Analysis on household assets.
Results
Over the period, an earlier fertility decline was reversed in 2009; mortality declined slightly for all age-groups, and a significant share of working-age population was lost through out-migration. Large minorities of children of school-going age are not in school. Socio-economic factors are shown to be important determinants of fertility and mortality.
Related collections
Most cited references22
- Record: found
- Abstract: found
- Article: found
The INDEPTH Network: filling vital gaps in global epidemiology
Osman Sankoh, Peter Byass (2012)
What is the INDEPTH Network? The International Network for the Demographic Evaluation of Populations and their Health (INDEPTH) Network is an umbrella organization for a group of independent health research centres operating health and demographic surveillance system (HDSS) sites in low- and middle-income countries (LMICs). Founded in 1998, it brought together a number of existing HDSS sites, and since then has encouraged newer HDSS sites to join. 1 The purpose of this Editorial is to set the scene for a series of profiles from INDEPTH HDSS member sites, the first examples of which are published in this edition of IJE. 2–5 All these profiles will follow a set pattern, to facilitate a systematic understanding of the multiplicity of HDSS sites involved in the Network and the various ways in which they are operated by their parent institutions. This Editorial therefore, follows the same general pattern as the individual profiles, but seeks to explore the epidemiological basis on which the HDSSs operate in general, and the role of the Network, rather than dealing with site-specific issues. At the central level, the INDEPTH Network operates from its base in Accra, Ghana, as an international NGO and is also registered as a not-for-profit entity in the USA. The emphasis on the Network’s position as a Southern-led and -based organization was an important founding tenet, and this is very welcome in a world where vestiges of colonialism still occasionally surface in relation to health data and policy. Day-to-day operations are led by the Executive Director (O.S.), and governance and oversight are provided by an international Board of Trustees and a Scientific Advisory Committee (chaired by P.B.). Why was the INDEPTH Network set up and what does it cover now? The raison d’être behind the emergence of the Network was the apparently intractable lack of reliable population-based data on health across many LMICs in Africa, Asia and Oceania. Recognizing that there are no quick fixes in terms of achieving universal individual registration of populations in LMICs, 6 the Network represents a medium-term attempt to break the link between material and data poverty. 7 Epidemiology in many LMICs suffers from a dual lack of reliable population data and human capacity to make use of them. The immediate consequence is that health policy making often lacks its essential evidence base, with the possible effect of failing to use scarce resources effectively in some of the world’s poorest countries. There are considerable global disparities in terms of epidemiological research output per population. Figure 1 shows the countries of the world shaded by a crude measure of this, namely the number of PubMed hits for a search on (‘epidemiology’ and ) per 1000 population. Much of Africa and Asia falls under the level of 0.05 per 1000, corresponding to rates which represent less than one-twentieth of some of the world’s leading countries in terms of epidemiological output. Superimposed on the map in Figure 1 are the current 43 HDSS sites run by 36 member centres of the INDEPTH Network. Although the locations of these sites are somewhat serendipitous, rather than being strategically planned, it is evident that there is considerable coverage across the areas of the world that lack substantial epidemiological output. Thus, it is clear that the INDEPTH Network, through these 43 sites in 20 countries, collectively following a population of 3.2 million people, does indeed offer possibilities for filling some of the global gaps in epidemiology. Figure 1 Countries of the world classified by PubMed citations for (‘epidemiology’ and ) per 1000 population, also showing the location of 43 HDSS site members of the INDEPTH Network (white dots) Where are the INDEPTH HDSSs? From the outset, the INDEPTH Network has operated by accepting as members already functioning independent health research centres that run HDSSs. Therefore, the Network has little influence over the locations or geographical distribution of member HDSS sites. However, since the concept of an HDSS would be somewhat irrelevant in countries with universal population registration, in practice there is self-selection of site locations in places where the lack of other reliable population-based data justifies the considerable effort involved in launching an HDSS. As is evident from Figure 1, this means that HDSS sites are located across Africa, Asia and Oceania, but by no means randomly. Several countries contain multiple HDSS sites, whereas many epidemiologically poor countries contain none. What populations are covered by the HDSSs and how are they followed up? HDSSs set out to collect epidemiological data (risks, exposures and outcomes) within a defined population on a longitudinal basis. In terms of Pearce’s classification scheme for epidemiological study designs, 8 this places HDSSs as representing ‘the most comprehensive approach since they use all of the available information on the source population over the risk period’. Unlike many epidemiological study designs, in which study participants are somehow selected to represent particular population subgroups, HDSSs generally set out to cover a real-life population and see what happens epidemiologically over a period of years and even decades. Issues of representativity and sampling are nevertheless critical considerations for all HDSSs, and need to be considered at the outset, when often little is known about potential target populations. Many HDSSs have started from intentions of covering an area that is at least subjectively thought to be typical of wider areas, maybe up to national levels. A chicken-and-egg situation arises, however, in that the motivation for having an HDSS is driven by a recognized lack of population-based health data, so that at the outset, very little may be known about candidate areas and maybe even less about the wider situation. There are no simple solutions to this conundrum. Even after identifying a target area for an HDSS, there are a number of possible design considerations. A range of different sampling strategies can be used within the target area, that have both epidemiological and practical implications. 9 In practical terms, one important consideration is whether the final population is defined as being within a contiguous area or in a collection of small areas (e.g. discrete villages) within a wider area. This has important logistic implications in terms of organizing and maintaining on-going surveillance, as well as affecting the definition of migration events (see below). The independent INDEPTH HDSSs naturally include a mixture of approaches to initially identifying target areas, within-area sampling and population contiguity. The overall size of the population within an HDSS is a further important factor, as is the case in any epidemiological study. However, an HDSS is not a classic sample survey, and so determining the size of the target population is not straightforward. Size is of course driven by considerations of the rarest event(s) of interest, which for most HDSSs are mortality-related outcomes. If specific causes of mortality are of particular concern, then the overall population size needs to be based on numbers relating to the nth ranked cause of interest. 10 Current INDEPTH member HDSS sites range in population size from tens of thousands up to around a quarter of a million. In most HDSSs the overall numbers are driven by mortality outcomes, with the result that surveillance of particular more common outcomes (such as morbidity and social measures) may in some situations be more effectively undertaken using a sample drawn from within the overall HDSS population. During the life of the INDEPTH Network, the technological and methodological possibilities for obtaining and using geographical data have advanced considerably, to the point where recording the latitude and longitude of every residential unit, and other salient features, in an HDSS using global positioning system (GPS) technology have become commonplace. Once an HDSS population is defined, an initial detailed census is usually undertaken to capture details of all residents and the social units in which they live. This usually involves assigning unique identifiers to all the residents and social units encountered in the census, using a numbering system that has sufficient capacity for expansion to reflect the addition of future residents and social units. It is not simple to arrive at generic definitions of social units across cultures and traditions, and individual HDSSs have to handle these issues in ways that make sense for their own context, both for physical structures (housing) and groups of inhabitants (families). INDEPTH has tried to standardize definitions as far as possible by publishing a resource kit for HDSS design on its website. This initial census then forms the basis of a database system that is updated on a regular basis to reflect the dynamic cohort of people living within the HDSS, as conceptualized in Figure 2. An important consideration is to determine the modality of the regular update rounds. Since HDSSs operate by definition in populations that are not otherwise enumerated, and generally have weak infrastructures, the norm is that local staff have to be recruited to undertake regular update visits to all the social units in the defined area. This forms a major component of the ongoing effort of running an HDSS, and consequently issues such as the frequency of update rounds need to be considered very carefully. Different INDEPTH HDSSs use various update frequencies, from one to four annual rounds. Certain types of events, e.g. neonatal mortality, are likely to be particularly sensitive to recall bias, which in turn is related to update frequency. Thus, it tends to be the case that more frequent updates are needed in high mortality or high migration settings, whereas in societies that are more stable, or at later stages of demographic transition, less frequent updates may prove adequate. Figure 2 Conceptual structure of the dynamic cohort model used by INDEPTH Health and Demographic Surveillance System (HDSS) sites What is being measured and how are the HDSS databases constructed? Having set up an HDSS, the next challenge is to track the progress of the dynamic cohort shown in Figure 2 by regularly updating a series of core parameters, detailed below. Naturally, the operation of an HDSS is not confined only to these core activities, and most HDSSs will have specific agendas defining what other parameters they may need to handle, e.g. in relation to the epidemiology of specific diseases, the execution of clinical trials, monitoring the effectiveness of health systems and other important issues that can be built onto the basic HDSS platform. Social units Keeping track of social units is a challenging issue, since it involves both physical structures (that can be newly built, in existence or be demolished) and the family groups associated with physical structures (that can migrate in or out as complete groups, or particular individuals can migrate to join or leave a group). In some cultures the physical structures may be large and complex compounds, perhaps housing up to 100 people and possibly containing subunits based on a polygamous social structure. At the other end of the spectrum, nuclear families may occupy small, discrete dwellings. Many HDSSs also aim to gather data on socio-economic status, often reflected by a basket of parameters including details of the physical structure, as well as owning traditional and modern assets. Births Capturing details of new births is a critical function of any HDSS, since births form a major part of new entrants to the cohort and are critical to any analyses of fertility. In some settings, traditional behaviours around childbirth (e.g. going to stay at the maternal grandmother’s residence for the birth and neonatal period) may make births more difficult to record accurately. There is a particular difficulty around detecting early neonatal deaths, and separating these reliably from intra-partum stillbirths, and this becomes more difficult with less frequent update rounds. Migrations Tracking details of migration patterns is one of the most complex areas in HDSSs, fundamentally comprising people moving into the surveillance area, within the area and out of the area. Many of these complexities are reflected in INDEPTH’s monograph on migration. 11 Every type of migration needs to be defined by rules (involving duration, intent, destination, etc.) which are appropriate to the population concerned. Some communities experience regular patterns of seasonal migration, related to employment or agricultural production. The possibility of multiple moves per individual over a period of time must be incorporated, and a further challenge can be the reliable re-identification of an individual on in-migration as being the same person who previously moved out. The design of an HDSS site in terms of the contiguity of the surveyed population is also important, since local moves in a non-contiguous population may be classified as in- and out-migrations, whereas similar moves in a contiguous area would amount to within-site migrations. Deaths Deaths, documented by age and sex, are a critical outcome measure for every HDSS and, in addition to reporting basic mortality rates, are an essential component in formulating life tables and other demographic measures for HDSS populations. As noted above, one of the most difficult issues involves reliably identifying early neonatal deaths. Causes of death Identifying the causes of death is a much more difficult issue in populations where most deaths do not occur in health facilities. The only realistic approach to attributing the cause of death is by carrying out verbal autopsy (VA) interviews with relatives or caretakers of deceased individuals, and then using those data to arrive at a likely cause of death. The INDEPTH Network was closely associated with developing a WHO standard instrument for VA interviews. 12 In many HDSSs, interpretation of the VA data was done by giving the VA data to local physicians, often more than one per case, in order to arrive at a consensus cause. However, this is an expensive and time consuming process that is gradually being superseded for most purposes by the application of computer-based probabilistic models. 13 INDEPTH is currently part of a new round of VA tool development in conjunction with WHO, which aims to simplify and shorten the VA process, as well as moving the scope of VA beyond research settings into non-enumerated populations. Databases Maintaining a database that reflects all the details of the population in a dynamic cohort is one of the most demanding tasks for most HDSSs, and a range of different approaches are used. The longitudinal nature of the HDSS data demands the use of relational database management systems (RDBMS) to handle the considerable volume of data involved over long periods of time. The basic principles of implementing an RDBMS for an HDSS have not changed fundamentally since the 1980s, when one of the longest-standing INDEPTH member HDSS sites made the transition to an RDBMS system. 14 However, appropriate hardware and software resources have progressed through several generations of development in the meantime, and that is reflected in the current range of implementations across the INDEPTH Network. These include implementations built on proprietary RDBMS systems such as Microsoft FoxPro™, Microsoft Access™ and Structured Query Language (SQL), as well as generic systems made available for the use of HDSS sites, such as the Household Registration System from the Population Council, 15 subsequently re-engineered as the paperless SQL-based ‘Open-HDS’. As commercial hardware and software specifications move on (e.g. Microsoft’s decision to cease supporting FoxPro™), long-term HDSS operations are sometimes forced to migrate their database operations onto new platforms, which is not a trivial matter for long-term databases linked to live surveillance. Ethical issues Running an HDSS over a long period raises a range of ethical issues that are different in some respects from those pertaining to many epidemiological studies. In the first place, the core HDSS data on vital events that are routinely collected in an HDSS population tend to be considered as research data, and subject to research ethics approval and informed consent, even though in countries that implement universal vital registration, it is regarded as a civic duty or even a legal obligation to provide such data. But, however population data are viewed, there are essential standards of confidentiality and anonymity that must be safeguarded. In HDSS data, there are three particularly critical types of data in this respect. Individual identities (whether by name or some other identifier) have to be protected at all stages of the process—from field interviewers observing adequate standards of confidentiality through database systems (and their backups) being held securely, to not revealing identifiers in any data sharing or outputs. Closely coupled with this, since HDSSs now commonly collect the GPS locations of households, it is important to also regard these data as confidential, since in principle they can be used to identify and locate households, and thereby their residents. Anonymizing GPS data is a much more difficult issue than simply removing names from a database. 16 Third, HDSS databases typically accumulate a large volume of personal, often medical, data (such as HIV status) that are sensitive and must be kept confidential. Key findings and publications Outputs from the INDEPTH Network mentioned here comprise those that are based on data from more than one HDSS site, or which make external comparisons. The individual HDSS site profile papers will provide further details of site-specific outputs. The INDEPTH Network website (www.indepth-network.org) provides information about the Network, its organization and current activities. One of the clear strengths of a network such as INDEPTH is its potential to collate data from member HDSS sites into outputs that enable systematic comparisons to be made. The first major INDEPTH output was a monograph published in 2002 that outlined basic HDSS concepts and gave details of 22 HDSS site members at that time. 17 Two further monographs relating to health equity in small areas 18 and migration 11 followed in 2005 and 2009, respectively. In a different format, using a supplement in an open-access journal, three sets of multi-site papers were published in 2009–10. The first related to cross-site findings on non-communicable disease risk factors from a group of INDEPTH member HDSS sites in Asia. 19– 27 The second related to mortality clustering across a range of INDEPTH member HDSS sites 28 – 36 and the third to results from eight INDEPTH member HDSS sites, which participated in the WHO–SAGE programme on ageing. 37– 46 The latter Supplement represented an innovation for the INDEPTH Network with the combined dataset used for the analyses also being published online together with the papers. Publications based on these public-domain data are now emerging. 47 A number of other papers have considered particular issues at the Network level. 48– 53 In addition, there have been some outputs that have involved inter-site collaborations but not included wide representation across the Network. 54– 59 In some cases, multiple INDEPTH members are also members of other research networks such as the RTS,S Clinical Trials Partnership 60 and the Alpha Network. 61 Several other studies have made comparisons between HDSS data from single INDEPTH HDSS sites and other sources. 62– 65 Future analysis plans As well as the substantial and continuing volume of outputs from individual HDSS sites, the INDEPTH Network will continue to produce multi-site outputs in particular topic areas. Current priorities include comparative assessments of fertility and cause-specific mortality patterns, as well as retrospective analyses of HDSS data against correspondingly timed weather data, which offer insights into the possible future population effects of changes in climatic conditions. Strengths and weaknesses HDSS sites represent an inherently strong epidemiological design, giving considerably greater analytical scope than can be achieved from e.g. cross-sectional approaches. However, the resources required to run an HDSS effectively are very considerable, particularly since the greatest gaps in health data are generally found in more logistically challenged environments. Not least this makes it very difficult for many HDSS sites to recruit and retain highly competent personnel, particularly those with experience in database management and epidemiological analysis, with the result that HDSS sites sometimes find it difficult to maximize their outputs. A recurrent issue that arises in considering HDSS data is how the site populations are, or are not, representative of the wider surrounding populations. Although this does not pose any technical issues in terms of analysing data within an HDSS site, it is of concern when it comes to interpreting HDSS data into wider epidemiological and policy arenas. There are no simple solutions to this issue, since HDSSs are always located in places where little is known about the surrounding population. It is possible to make comparisons with other data sources, such as national censuses and cluster sample surveys, 62– 65 but these sources come with their own disadvantages such as greater recall bias, and hence it is very difficult to attribute causes to observed differences. An empirical investigation into this issue used Swedish national data from 1925, a time when Sweden shared many characteristics with contemporary LMICs. 66 This showed that the majority of individual counties could have been taken as adequately representative of the national population, and the less representative counties were self-evidently so (including the capital city and the most remote regions). Although this does not offer any absolute evidence about the representativity of INDEPTH member HDSS sites, it suggests that it is not reasonable to assume by default that HDSS populations are unrepresentative. The diversity observed across the INDEPTH member HDSS sites is a further source of both strength and weakness. As discussed earlier, there has never been any master plan for establishing HDSS sites in particular locations, and there are also significant (but often locally appropriate) detailed methodological differences between HDSS sites. This brings strength in terms of having highly functional and locally supported HDSS sites in many locations, something that might not have happened so effectively in trying to locate HDSS sites more systematically. However, it also brings some weaknesses when it comes to making comparisons across HDSS sites and between the countries that they represent. In contrast, the much stricter uniformity enforced across the Demographic and Household Survey (DHS) series of cross-sectional surveys makes comparisons simpler, 67 but that stems from a completely different organizational paradigm. Nevertheless, the common core activities of all INDEPTH member DHSS sites, in following vital events longitudinally in a defined population, mean that the pooled INDEPTH data represent a major unified source of data on otherwise undocumented populations. An interesting development in some situations, e.g. in China, 68 is the concept of a distributed national network of HDSS-type surveillance, which perhaps represents a further intermediate step for the future. This has the advantage of being more widely representative, but at the same time bringing the advantages of a longitudinal approach. This may become a more common model as countries move towards universal individual registration. Data sharing and collaboration Data sharing issues have become increasingly important for all health researchers in recent years, and also continue to generate much debate. 69 There is also a continuing dialogue between researchers and funders on these issues. 70 The INDEPTH Network is firmly committed to the principles and practice of sharing data, as expressed in the INDEPTH Data Access and Sharing Policy document, available as Supplementary data at IJE online. The issues involved in sharing HDSS data are complex. By the nature of the dynamic cohort, there is never any point in time when data collection is ‘complete’, and talking about sharing data at pre-determined intervals after completion is therefore not entirely helpful. Ways to work around these conceptual difficulties therefore have to be found, involving declaring particular periods of data from an HDSS as being ready for sharing at appropriate times. INDEPTH has already launched the iSHARE portal for making data from HDSS member sites publicly available (www.idepth-ishare.org) to bona fide users, not unlike the arrangements for access to DHS data sets. In the existing version of iSHARE, data files from the participating HDSS sites are arranged in separate event files (births, deaths, migrations), but plans are underway to standardize iSHARE data into a common event-based data format. The common event attributes involved are shown in Table 1, and the range of different possible events are listed in Table 2. This structure will allow all participating sites to present HDSS core data in a straightforward and standardized format, which will facilitate a wide range of possible analytical approaches. Table 1 Common event attributes for the INDEPTH data specification Attribute Variable name Description Record number RecNr A sequential number uniquely identifying each record in the data file Centre identifier CentreId An identifier issued by INDEPTH to each member centre of the format CCCSS, where CCC is a sequential centre identifier and SS is a sequential identifier of the site within the centre in the case of multiple site centres Individual identifier IndividualId A number uniquely identifying all the records belonging to a specific individual in the data file. For data anonymization purposes, this number should not be the same as the identifier used by a contributing centre to identify the individual, but the contributing centre should retain a mapping from this identifier to their identifier Country identifier CountryId ISO 3166-1 numeric code of the country in which the surveillance site is situated Location identifier LocationId Unique identifier associated with a residential unit within the site and is the location where the individual was or became resident when the event occurred. For data anonymization purposes, this identifier should not be the same as the identifier used internally by the contributing centre, but the contributing centre should retain a mapping of this identifier to their internal location identifier Date of birth DoB The date of birth of the individual Event EventCode A code identifying the type of event that has occurred (Table 2) Event date EventDate The date on which the event occurred Observation date ObservationDate Date on which the event was observed (recorded), also known as surveillance visit date Event count EventCount The total number of events associated with this individual in this data set Event number EventNr A number increasing from 1 to EventCount for each event record in order of event occurrence Table 2 Event types for the INDEPTH data specification Event Code Definition Attributes Attribute description Birth BTH The birth of an individual to a resident female MotherId DeliveryEventId The IndividualId of the mother The RecNr of the delivery event associated with this birth Enumeration ENU Starting event for all individuals present at the baseline census of the surveillance area. It is the date on which the individual was first observed to be present in the surveillance area during the baseline census In-migration IMG The event of migrating into the surveillance area Origin Classification scheme to be developed Out-migration OMG The event of migrating out of the surveillance area Destination Classification scheme to be developed Location exit EXT The event of leaving a residential location within the surveillance area to take up residence in another residential location within the surveillance area Destination The LocationId of the location within the surveillance area to which the individual relocated Location entry ENT The event of taking up residence in a residential location within the surveillance area following a location exit event. Note that location exit and entry are actually two parts of the same action of changing residential location and as such happen on the same event date Origin The LocationId of the residential location from which the individual moved Death DTH The death of the individual under surveillance. The date of death is the event date Cause1 Cause2 Cause3 Likelihood1 Likelihood2 Likelihood3 Up to three causes of death coded using the WHO list of verbal autopsy death causes. Likelihood values associated with each possible cause of death Delivery DLV The event of a pregnancy end after 28 weeks of gestation, which may or may not result in the birth of one or more individuals (represented in this dataset by a BTH event linked to this delivery event) LBCnt SBCnt Parity Live birth count Stillbirth count The number of live births to these women prior to this delivery Observation end OBE An event inserted when a data set is right censored at an arbitrary date and this individual remained under surveillance beyond this date. The right censor date is the date of this event Last observation OBL An event indicating the last point in time on which this individual was observed to be present and under surveillance. Event date equals observation date in this instance. Normally there should be no individuals with this event as their last event if the right censoring date is prior to the start of the last complete census round Observation OBS Used to record characteristics of individuals under surveillance valid at the time of the observation. Could be used to record aspects such as educational attainment, employment status or anthropometry measures. Specific examples of this event are not part of the minimum core individual dataset, but are specified to allow for site or working group needs Conclusion Our aim here is to describe the essential nature of the INDEPTH Network as a background to detailed profiles of constituent member HDSS sites. Although all those sites have important differences, the huge volume of detailed individual data generated across Africa, Asia and Oceania by the Network constitutes a unique resource of great value to demographers, epidemiologists and health planners. Supplementary Data Supplementary Data are available at IJE online. Funding Osman Sankoh is funded by core support grants to INDEPTH from the Hewlett Foundation, Gates Foundation, Sida/GLOBFORSK and Wellcome Trust. Supplementary Material Supplementary Data
- Record: found
- Abstract: found
- Article: not found
Mortality trends and setbacks: global convergence or divergence?
Health trends over much of the past century have been generally, and notably, positive throughout the world. In several regions, however, life expectancy has declined over the past 1-2 decades. This trend suggests that the expectation that emerged in the 1960s and 1970s of worldwide gains and convergence in population health status is not guaranteed by a general deterministic process. National populations can now be clearly grouped into those that have achieved rapid gains in life expectancy; those whose gains are slower or are perhaps plateauing; and those in which the trends have reversed. Over the past two centuries, outside times of war and famine, such reversals have been rare. Exploration of these varied population health trends elucidates better the close relation between population health and the processes of economic, social, and technological change. Such analysis has shown that the health status of human populations should be a guiding criterion in the debate on sustainable development.
- Record: found
- Abstract: found
- Article: not found
Moving from Data on Deaths to Public Health Policy in Agincourt, South Africa: Approaches to Analysing and Understanding Verbal Autopsy Findings
Peter Byass, Kathleen Kahn, Edward Fottrell … (2010)
Introduction Throughout the history of public health, the concept of recording causes of individual deaths in a population and presenting them in aggregate form has been a central component of understanding health and disease at the community level. This continues to be the case, even though the extent and quality of cause of death data varies widely around the world [1]. For a large proportion of the world's communities in which individual deaths are not routinely recorded and classified by cause as part of routine civil and health service procedures, verbal autopsy (VA) has become an important technique [2]. VA involves interviewing family, friends, or carers after a death has occurred, to find out about the circumstances of death. These data are normally collected by lay interviewers, and their findings are later interpreted into possible cause(s) of death. Approaches to undertaking the interviews and interpreting the findings vary and are still developing, despite various efforts towards standardisation [3]. Much VA work has relied on physicians reviewing interview material and coming to a conclusion on cause of death, following a process closely analogous to clinical practice in which history, signs, and symptoms are used to construct a differential diagnosis. Recently, computer-based probabilistic models have become an important way of interpreting VA data, as an alternative to case-by-case physician interpretation [4]. These have the advantage of being faster, cheaper, and more internally consistent than physician review, but may lack some subtlety and nuance. Some comparisons between physician review and modelled findings have previously been made [4]–[6]. As well as characterising all-age, all-cause mortality, applications of verbal autopsy have included cause of death determination among particular groups such as women of reproductive age [7], and for assessing community interventions [8]. However, the outputs from probabilistic models have some technical differences from those typically generated by physicians, since the likelihood of a particular cause of death is also estimated quantitatively as part of the modelling. Several likely causes can be reported for a single case, and a case may remain partially or wholly indeterminate, particularly where the VA interview material is scanty. These characteristics might seem problematic from a clinical perspective that instinctively seeks a conclusive single main cause of death for each case (even though this is sometimes fudged by labelling two commonly coexisting causes as a single entity, for example the cause “HIV/AIDS and tuberculosis”). However, since VA is normally applied as a step towards community-based analyses of cause-specific mortality and public health implications, rather than as an endpoint whose primary concern is the individual case, the outcomes are essentially epidemiologically rather than clinically oriented. The proportions of deaths within a population attributable to a particular cause (cause-specific mortality fractions, CSMFs) are particularly important. Thus some uncertainty at the individual level, and possibly multiple causes per case, are not in themselves problematic, but need analytical approaches that make good sense of the data. The public health imperative to understand causes of death in terms of age and sex is also important, in order to understand burdens of premature mortality, to target potential interventions, and to inform health systems development. In this paper we aimed to assess appropriate methods for analysing and interpreting VA interview data at the population level, using both probabilistically modelled and physician-interpreted results. An example dataset from the Agincourt health and sociodemographic surveillance site (HDSS) in South Africa, a member of the INDEPTH Network (http://www.indepth-network.org), is taken to illustrate the approaches used. Existing physician-interpreted findings from the same dataset are compared with modelled results in the sense of how they are derived and analysed, leading to some comparisons between the two approaches. However, the intention here is not to validate either approach; rather the emphasis is on interpretation and analysis processes which can lead effectively from data on deaths to public health imperatives. Methods The Agincourt HDSS covers rural communities located in northeast South Africa, near the Mozambican border, and has monitored a contiguous population of around 70,000 since 1992. The background to this work is described more fully elsewhere [9], in a paper which analyses cause-specific mortality from 6,153 deaths that occurred between 1992 and 2005, on the basis of cause of death as determined by physician review. These physician reviews of VA interview material were each initially undertaken by two physicians independently. If they did not agree as to cause, a third physician arbitrated in order to reach a consensual cause of death. If consensus could not be reached, then no cause of death was recorded. The same VA interview data were compiled into an input file for the InterVA v.3 probabilistic VA interpretation model (http://www.interva.net) and processed into cause of death data. The InterVA model is based on Bayesian calculations of probabilities that a particular death was due to particular causes, given a set of symptoms and circumstances associated with the death. This is achieved using a probability matrix which generically estimates probabilities of particular symptoms and circumstances of death, given particular causes. The model was developed using an expert panel and was deliberately designed to be generic and not context-dependent, and to produce relatively broad cause-of-death categories [10]. As previously described [5], the model expects an input of “high” or “low” to reflect the local prevalence of two specific causes which often vary by more than an order of magnitude between settings: HIV and malaria; here these were set to “high” and “low” respectively. These settings do not override the handling of individual cases, but are conceptually similar to a physician knowing that a particular disease is common or rare in the local population, irrespective of a particular patient presenting in a consultation. Compiling the data input file for the InterVA model (which consists of yes/no answers for each case on around a hundred questions relating to the VA interview material) may take some days for a data manager, but processing the file into causes of death using the model then only takes a matter of minutes. This contrasts with thousands of hours of physician time, and a cost in the region of US$20,000, for reviewing a dataset of this size. The model is also totally internally consistent, meaning that rerunning data produces exactly the same output, and there is also therefore complete consistency at the individual case level over time, when considering a series of deaths that actually occurred over many years. With physician interpretation, it is unlikely that the same physicians can be available to undertake this work over an extended period, and in any case it is probable that their thinking and understanding would change over time. The model provides up to three likely causes of death for each case, or concludes that the cause is indeterminate. Each cause assigned is associated with a likelihood, and the sum of likelihoods of assigned causes has a maximum value of 1.00. If the sum of likelihoods of assigned causes is less than 1.00, then the difference reflects a lack of certainty about the overall case. It therefore seems logical to regard this uncertain proportion of each case as an indeterminate component. For analysis, a dataset was constructed from the model's output (using Microsoft FoxPro) in which each case had one or more records, each record having one cause (including the possible cause “indeterminate”) and a weight corresponding to the likelihood of that cause for the particular case. Thus over the whole dataset, the sum of all the weights was equal to the number of cases, 6,153. This dataset included a total of 11,834 records, an average of 1.92 per case. This data structure also facilitates the import of other background factors of interest (since every record contains the individual identifier variable), which can then be analysed against particular causes of death in a weighted multivariate model. Physician-interpreted material where consensus on a single main cause is required can be analysed in very similar ways, with the conceptual weighting for each case being 1. The analyses presented here were carried out using Stata 10. Surveillance-based studies in the Agincourt subdistrict were reviewed and approved by the Committee for Research on Human Subjects (Medical) of the University of the Witwatersrand, Johannesburg, South Africa (protocol M960720). Informed consent was obtained at the individual and household levels at every follow-up visit, whereas community consent from civic and traditional leadership was secured at the start of surveillance and reaffirmed from time to time. Results The same 6,153 deaths as presented previously using physician-interpreted causes [9] are shown in Table 1, with cause of death as determined from the same VA material by the InterVA model and shown by cause and age–sex group. The physician-determined CSMFs for the overall population are also shown for comparison. The ten highest ranking causes constituted 83.3% of the total according to physician interpretation and 88.2% according to probabilistic interpretation, and 8/10 of these causes were the same according to both approaches (HIV, tuberculosis, chronic cardiac, diarrhoea, pneumonia/sepsis, transport-related accidents, homicides, and indeterminate). The fractional causes of death from the model reflect the aggregation of likelihoods of particular causes over age–sex subgroups within the Agincourt population. These subgroups are the same as those used for the input file to the InterVA model. 10.1371/journal.pmed.1000325.t001 Table 1 Verbal autopsy findings for 6,153 deaths in Agincourt HDSS occurring between 1992 and 2005, by likely cause and age–sex group, using cause of death as interpreted probabilistically by the InterVA 3 model. Cause of Death Physician CSMF* (%) InterVA CSMF* (%) Number of Deaths by Age–Sex Group from InterVA Results Up to 28 Days 28 Days to 1 Year 1–4 Years 5–14 Years Men, 15–49 Years Women, 15–49 Years 50–64 Years 65+ Years Accidental drowning 0.2 0.2 — — 2.3 10.3 — — 0.5 — Accidental poisoning 0.3 0.5 — — 8.7 2.8 16.1 — 3.6 0.7 Acute cardiac 0.3 0.3 — — — — 3.2 1.0 6.2 6.7 Acute respiratory 1.1 0.6 1.8 5.0 1.8 0.4 4.9 2.2 6.9 15.5 Bloody diarrhoea 0.1 0.6 — 2.3 13.4 1.9 2.7 5.8 1.4 12.2 Chronic cardiac 4.2 4.4 0.6 — — 2.3 28.1 21.4 86.8 134.4 Chronic respiratory 0.7 1.9 — 1.1 0.5 — 7.2 6.0 33.1 70.7 Congenital malformation 0.5 0.2 9.6 0.8 — — — — — — Diabetes 1.2 2.9 — — 4.2 2.0 17.8 22.9 38.9 93.5 Disease of nervous system 0.9 0.0 — — — — 0.8 — 0.3 — HIV/AIDS related 18.6 15.3 1.6 129.4 201.7 12.8 174.2 349.6 67.4 6.1 Haemoglobinopathy 0.1 0.4 — — 7.4 10.3 2.8 6.9 — — Homicide 2.7 2.9 — — — 0.5 121.2 21.8 25.2 11.4 Indeterminate 34.8 31.0 71.1 117.8 166.7 64.0 341.0 323.0 296.8 524.7 Kidney/urinary disease 0.6 1.5 — — 1.8 1.3 8.8 3.2 19.6 56.4 Kwashiorkor 1.3 0.0 — — 2.5 — — — — — Liver disease 1.6 4.0 — 2.7 5.0 2.6 32.1 21.2 54.5 129.2 Malaria 1.6 0.1 — — 0.1 1.4 2.2 4.2 0.7 — Malignancy 5.0 0.9 — — — — 4.8 11.5 10.0 29.7 Malnutrition 0.3 0.2 — — 1.7 1.0 — — 0.2 12.0 Maternity related 0.5 0.4 — — — — — 26.8 — — Measles 0.1 0.0 — — 0.8 — — — — — Meningitis 0.7 1.4 7.1 8.4 8.1 19.6 13.6 8.8 11.1 8.3 Nonbloody diarrhoea 3.4 2.2 15.1 82.3 26.4 4.4 1.3 7.0 4.9 1.5 Other digestive disease 0.9 0.1 — — — — — — 4.9 1.5 Other fatal accident 1.8 0.2 — 0.9 0.5 0.9 1.5 0.8 0.4 7.2 Pneumonia and/or sepsis 2.3 3.7 30.7 47.6 23.1 7.9 12.2 9.0 28.7 69.7 Preterm or small baby 0.9 0.2 9.9 2.2 — — — — — — Stroke 4.4 0.7 — — — — 2.2 4.3 11.7 25.9 Suicide 1.2 0.8 — — — 3.9 27.6 2.5 5.4 8.0 Tetanus 0.1 0.2 — 1.0 — 0.7 0.6 0.6 3.4 7.1 Transport-related accident 2.6 3.3 — 0.5 13.2 25.7 96.8 30.1 18.9 17.0 Tuberculosis (pulmonary) 5.3 18.5 0.7 5.0 17.3 12.5 369.6 250.4 165.2 318.4 All causes 100 100 148 407 507 189 1,292 1,134 903 1,573 Cause-specific mortality fractions from physician coding of the same dataset are shown for comparison in the second column. *Cause specific mortality fraction, across all ages. The overall proportion of indeterminate cases was 31.0%, compared with 34.8% in the physician review process. This indeterminate category included 359 deaths for which verbal autopsies were not successfully completed. In the InterVA model, a further 375 cases were rated as completely indeterminate, and the summed weights of the uncertain proportion of the remaining cases totalled 1,170.9, an average uncertainty per case of 24.3%. In the physician coding, 1,609 individual cases were considered to be indeterminate, either because of insufficient information or failure to reach a consensus between assessing physicians. The physicians considered a further 173 cases as indeterminate for particular reasons, for example sudden deaths of unknown cause. It was also interesting to note that the physician coding process led to using a total of 250 different ICD-10 codes, but the ten most frequently used ICD-10 codes accounted for 70.7% of the deaths. Table 2 shows the five principal causes of death for each age group and period. It has been constructed to be as similar as possible to the corresponding table in the previous paper using physician interpretation of the same dataset (Table 2 in [9]), a process which involved regrouping InterVA causes of death accordingly. For each period and age group, the physician-interpreted ranks from the previous paper are also shown for comparison. 10.1371/journal.pmed.1000325.t002 Table 2 Five main causes of death by age group and time period, based on 6,153 deaths in Agincourt HDSS occurring between 1992 and 2005, using cause of death as interpreted probabilistically by the InterVA 3 model. Age Group Rank 1992–1994 1995–1997 1998–2001 2002–2005 Cause % PR* Cause % PR* Cause % PR* Cause % PR* 0–4 Years Indeterminate 43.8 38.6 35.7 25.6 1 Diarrhoea 24.7 1 HIV/TB 20.8 2 HIV/TB 31.0 1 HIV/TB 49.7 1 2 HIV/TB 9.7 — Diarrhoea 20.5 1 Acute resp infection 11.5 3 Acute resp infection 9.9 3 3 Acute respiratory infection 9.2 — Acute respiratory infection 10.0 3 Diarrhoea 11.2 2 Diarrhoea 7.5 2 4 Other infection 3.4 5 Other NCD 3.3 — Other infection 2.3 — Other infection 2.3 — 5 Accidental injury 3.3 4 Other infection 2.8 — Perinatal causes 1.7 4 Congenital 1.2 — 5–14 Years Indeterminate 33.4 23.6 34.7 38.2 1 Other infection 17.5 — Accidental injury 12.3 3 Other infection 15.8 — HIV/TB 20.1 2 2 Road traffic accident 12.8 2 Other infection 11.5 — Road traffic accident 14.3 5 Road traffic accident 15.8 3 3 HIV/TB 11.4 — HIV/TB 10.6 4 Accidental injury 8.4 1 Accidental injury 7.5 1 4 Other NCD 8.0 1 Other NCD 8.6 2 HIV/TB 6.9 3 Other infection 5.7 5 5 Diarrhoea 5.9 — Road traffic accident 8.5 5 Other NCD 5.5 — Other NCD 5.7 4 15–49 Years Indeterminate 30.3 31.1 27.3 26.3 1 HIV/TB 23.8 3 HIV/TB 34.3 1 HIV/TB 44.2 1 HIV/TB 54.4 1 2 Assault 14.3 1 Assault 9.6 2 Road traffic accident 6.7 3 Assault 4.4 2 3 Road traffic accident 12.6 2 Road traffic accident 7.7 3 Assault 5.1 2 Road traffic accident 2.9 3 4 Maternity 3.3 — Chronic liver 2.8 — Other NCD 3.3 4 Other NCD 2.2 5 5 Other NCD 3.0 — Other NCD 2.8 4 Chronic liver 2.6 — Chronic liver 2.0 — 50–64 Years Indeterminate 36.6 38.0 32.2 30.6 1 HIV/TB 14.0 — HIV/TB 21.5 1 HIV/TB 22.0 1 HIV/TB 31.9 1 2 Other NCD 11.3 4 Other cardiac 12.6 3 Other cardiac 9.4 — Other cardiac 9.4 3 3 Other infection 7.8 — Other NCD 10.0 4 Chronic liver 9.3 — Other NCD 7.9 4 4 Other cardiac 6.7 3 Chronic liver 4.2 — Other NCD 7.6 3 Chronic liver 5.0 — 5 Chronic liver 6.2 2 Assault 4.1 5 Other infection 4.3 — Acute respiratory infection 3.9 — 65+ Years Indeterminate 37.4 34.1 31.6 32.9 1 Tuberculosis 22.4 3 Tuberculosis 21.1 3 Tuberculosis 20.7 4 Tuberculosis 18.4 4 2 Other cardiac 7.4 1 Other cardiac 10.5 2 Other NCD 13.5 5 Other NCD 11.3 5 3 Acute respiratory infection 7.1 — Other NCD 7.4 — Chronic liver 9.7 — Other cardiac 9.0 2 4 Other NCD 6.8 — Acute respiratory infection 6.9 — Other cardiac 7.3 3 Chronic liver 8.9 — 5 Chronic liver 6.4 5 Chronic liver 6.0 — Acute respiratory infection 5.0 — Other infection 5.8 — *PR are physician-interpreted ranks within each period and age category, as previously published [9]. NCD, noncommunicable disease. One instance in which there was a clear difference in the estimates between the physician-coded and modelled findings was in tuberculosis as a cause of death among the elderly (over 65 years). According to the physicians, 96/1,492 (6.4%) of deaths in this age group were due to tuberculosis, compared with 318.4/1,492 (21.3%) according to the model. Of the 96 cases reported as tuberculosis by the physicians, 78 (81.3%) were also concluded to be tuberculosis by the model. However, among the 241 cases rated as tuberculosis by the model but not by the physicians, 103 (42.7%) were rated as indeterminate by the physicians. To elucidate this difference, Table 3 shows the breakdown of key VA interview parameters which might contribute to a conclusion of tuberculosis as cause of death, both for the InterVA model and for the physicians. It includes the positive predictive value (PPV) for tuberculosis for each parameter, both in the physician and model interpretation. 10.1371/journal.pmed.1000325.t003 Table 3 Pulmonary tuberculosis as a possible cause of death among 1,492 elders (65+ years) as interpreted by physician consensus (6.4%) and probabilistic modelling (21.3%), in relation to selected verbal autopsy parameters. VA Parameter Physician Interpretation Probabilistic Modelling Not TB TB PPV* Not TB TB PPV* Chest pain No 1,041 8 91% 949 100 69% Yes 359 84 225 218 Chronic cough No 1,155 6 93% 1,067 94 70% Yes 245 86 107 224 Productive cough No 1,179 8 91% 1,059 128 60% Yes 221 84 115 190 Difficulty breathing No 794 29 68% 687 136 58% Yes 606 63 486 183 Night sweats No 1,274 41 55% 1,143 172 46% Yes 126 51 31 146 Chronic fever No 908 47 49% 846 109 66% Yes 492 45 328 209 Weight loss No 426 5 95% 414 26 92% Yes 974 87 769 292 History of tuberculosis No 1,352 40 57% 1,168 225 26% Yes 48 52 18 82 *PPV: positive predictive value of the VA parameter for pulmonary tuberculosis. Discussion Having considered the causes of more than 6,000 deaths over a 14-year period, the ten highest-ranking causes accounted for 83% and 88% of all deaths by physician interpretation and probabilistic modelling respectively, and eight of the highest ten causes were common to both approaches. Probabilistic modelling was cheaper and more internally consistent than physician interpretation. Uncertainty around the cause(s) of individual deaths was recognised as an important concept that should be reflected in any overall analysis of cause-specific mortality. The advantages and disadvantages of physician-interpreted and probabilistically modelled cause of death data as evidenced by these analyses were largely as anticipated. Physician-interpreted findings included a number of quite specific, but rare, causes which were not designed to be addressed by the current model. While it is possible to build similar models with more detailed inputs and outputs, as has been done for deaths among women of reproductive age [7], this model was designed to capture major cause-of-death groupings. In principle a model designed to include greater differentiation—for example, between different cancers at particular sites—could be constructed; but the extent to which that would lead to greater understanding of population health is less clear. The very large number of specific causes used by the physicians, even though the occurrence of many was very low, could be regarded as an advantage in terms of subtlety or as a disadvantage in terms of clear overall understanding of mortality patterns (without applying further judgement calls on appropriate grouping). Probabilistically-modelled interpretation has major advantages in terms of cost (not needing to pay physicians), time (less delay in getting results after interviews) and complete consistency. A recent review of the Indepth Network accordingly concluded that the InterVA model represented the most effective way forward for standardised interpretation of VA data across the network. [11] However, there is also the possibility of there being consistent errors encapsulated in the model. The methods described here for analysing the probabilistically modelled cause of death data are relatively straightforward, taking into account that particular causes of death have been modelled with a specific likelihood and the quantifiable margin of uncertainty associated with many individual cases. These methods allow the margins of uncertainty associated with individual cause of death assignments to be carried through into the aggregated analysis process. When physicians are used to assess VA material, and particularly if, as was the case for these data, physician consensus on individual cases is taken to be an important part of the process [12], then a simpler analytical approach can be used, as evidenced in the earlier paper using these data [9]. Once each case is assigned a cause of death or is considered to be indeterminate, categorising and tabulating cases as needed is straightforward, since each death counts as a single case. However, it has to be realised in this approach that any sense of the uncertainty that may have been evident in the original physicians' consideration of individual cases, or in consensus conferences, has already been eliminated before aggregated analysis begins. Since both approaches yielded only about two-thirds certainty, incorporating uncertainty in aggregated measures of cause-specific mortality seems important. Uncertainty might be better handled in physician-interpreted data if individual physicians' opinions were used, rather than insisting on consensus. Given the very different approaches to cause-of-death interpretation and analysis as presented here and in the earlier paper (probabilistic modelling and analysis incorporating uncertainty, versus physician assessment and tabulation of definitive consensus findings), it is perhaps remarkable that many of the salient features of Table 2 here and Table 2 in the previous paper are closely similar [9]. Both give a picture of a population increasingly dominated by the burden of HIV-related mortality as time passes, together with appreciable numbers of deaths due to external causes, and relatively low infectious disease mortality (apart from the HIV/TB combination). It is also interesting to note that the overall proportion of cases to which specific causes could not be attributed is similar, despite being derived from completely different methods. There are also some potentially important differences emerging from the two approaches, even though they are not huge in the context of the entire dataset. In considering any such differences, it has to be recognised that there is no gold standard available here. Kahn et al. have previously undertaken a validation exercise between physician assessments and a limited number of well-justified hospital-based diagnoses [12], and we plan to extend this to a detailed three-way comparison including the InterVA findings for this limited subset of deaths. However, in a community such as this where many people die without contacting health services, and where hospital records are often of poor quality, the quest for a wide-ranging gold standard for VA findings which fairly represents all causes and circumstances of death has to be regarded as futile. Notable differences that do emerge include lower estimates of malignant disease in the InterVA findings and lower estimates of tuberculosis among the elderly in the physician data. The InterVA model also gave higher estimates of HIV-related mortality in the first period (1992–94), which is particularly interesting to note. This early difference may reflect a degree of false-positive HIV-related findings by the model during a period of lower HIV-prevalence, and this needs to be further investigated in terms of characterising the overall HIV prevalence for the model as “high”. On the other hand, it might reflect a difficulty among the physicians in achieving consensus on HIV as a cause of death in those relatively early days of the epidemic, and this is also something to look into further. It seems likely that during the onset of the epidemic, individual physicians' perceptions of new disease patterns might have developed quite rapidly, but not necessarily in the same ways and at the same rates, depending on their personal experiences. This process, at least for a while, might have increased the difficulties in achieving consensus on HIV-related causes. The reported rate of ill-defined or unknown causes was highest in the physician-coded material for the period 1992–94 (approximately one-third), falling to approximately one-fifth by 2002–04 [9]. The examples of factors leading to tuberculosis as a cause of death among the elderly, as detailed in Table 3, provide interesting insights into differences between the two interpretations. It is clear that the physicians mainly determined tuberculosis as a cause when chest pain, chronic cough, productive cough, and weight loss were all reported for a particular case, whereas the model took a less specific approach. This is reflected in the generally higher positive predictive values for physician interpretation. On the other hand, the high proportion of indeterminate conclusions reached by the physicians among the model's probable tuberculosis cases suggests a degree of uncertainty in their deliberations, rather than clear alternative conclusions. There may also be a question of physicians' expectations of the likelihood of tuberculosis among the elderly, given that many elders in this community will now be living in households with younger adults coinfected with HIV and tuberculosis. Recent studies from Spain [13] and China [14] reported raised tuberculosis case-fatality rates among the elderly. In any case, although this example represented one of the larger discrepancies between the two approaches, it still accounted for only 218/6,153 (3.5%) of overall deaths. The importance of conceptual categorisations of cause of death can also be seen in these comparisons. At first sight, it appears that the approaches (Table 2) gave different pictures regarding deaths due to malnutrition among the under-5s, with 0.4% from the model and 9.0% from the physicians [9]. However, if one considers that tuberculosis is probably a relatively rare cause of death in young children, even as an HIV coinfection (as evidenced in nearby Mozambique [15]), and that HIV-infected children are more likely to follow a pattern of chronic diarrhoea and malnutrition [16], then the picture changes somewhat. So, taking the physicians' “HIV/tuberculosis” grouping as mainly not being tuberculosis in this age group, and adding that to their “diarrhoea” and “malnutrition” codings, for the under-5s the proportions of deaths due to “HIV/diarrhoea/malnutrition” were 38%, 41%, 42%, and 52% for the four periods, respectively. This result is strikingly similar, in magnitude and progression, to the same grouping from the InterVA findings (34%, 41%, 42%, and 56% respectively), and would represent the largest single cause of under-5 mortality in both approaches. Thus conceptual groupings that reflect real public health issues, rather than (in this instance) rather sterile debates as to what HIV-infected children with chronic diarrhoea and wasting actually die from, are crucial. International Classification of Diseases (ICD-10) coding for causes of death may not therefore be as relevant at this conceptual level, even if they can be a useful framework at earlier stages, for example in assigning physician-coded causes. The main aim of this paper is not to provide a validation of any particular VA method, but to consider alternative approaches for handling interview data on individual deaths to give meaningful pictures of population health. These data are the basic resource for public health planning: the questions in our minds throughout these considerations have started from “If I were the local Director of Public Health…”. From these data, and irrespective of the methods used for analysis and interpretation, it is clear that the Agincourt population has undergone rapid changes, which imply new intervention target groups, expanded demands on health professionals' skills, changing demands on health services and increasing resource requirements. The pictures of the major public health themes within the Agincourt population that emerge from both of the interpretative approaches considered are encouragingly similar, both in terms of overall cause-specific mortality patterns and in the ways that they have tracked changes over time, and the adoption of one or other method of interpretation would not lead to fundamentally different public health actions. The clear development of the HIV epidemic revealed in this example, and seeing which population subgroups are vulnerable to particular diseases, both highlight some of the advantages of using VA as a public health tool. At least where VA is used within routine health services, probabilistic modelling with its consistent approach over time and place, the elimination of inter- and intra-assessor variation, faster results, and much lower cost, should be the interpretative method of choice.
Author and article information
Journal
Journal ID (nlm-ta): Glob Health Action
Journal ID (iso-abbrev): Glob Health Action
Journal ID (publisher-id): GHA
Title:
Global Health Action
Publisher:
Co-Action Publishing
ISSN
(Print):
1654-9716
ISSN
(Electronic):
1654-9880
Publication date
(Electronic):
20
December
2012
Publication date Collection: 2012
Volume: 5
Electronic Location Identifier: 10.3402/gha.v5i0.19033
Affiliations
Author notes
[*
]
Seth Owusu-Agyei, Kintampo Health Research Centre, P O Box 200 Kintampo, Ghana. Tel (Office): +233
3520 92038, Tel (Cell): +233 244 560 213. Email:
seth.owusu-agyei@
123456kintampo-hrc.org
Article
Publisher ID: 19033DOI: 10.3402/gha.v5i0.19033
PMC ID: 3529298
PubMed ID: 23273249
SO-VID: bd0c6150-5e60-470c-a027-40298cd1cefa
Copyright © © 2012 Seth Owusu-Agyei et al.
License:
This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
History
Date
received
: 26
June
2012
Date
revision received
: 13
October
2012
Date
accepted
: 21
November
2012
Categories
Subject:
Original Article
ScienceOpen disciplines: Health & Social care
Data availability:
ScienceOpen disciplines: Health & Social care
Keywords: kintampo, demographic surveillance, population, ghana, demographic rates, fertility, mortality, migration
Comments
Comment on this article
scite_
113
1
129
0
Smart Citations113
1
129
0
Citing PublicationsSupportingMentioningContrasting
See how this article has been cited at scite.ai
scite shows how a scientific paper has been cited by providing the context of the citation, a classification describing whether it supports, mentions, or contrasts the cited claim, and a label indicating in which section the citation was made.