Presentation on theme: "Data collection for epidemiological statistics"— Presentation transcript:
1 Data collection for epidemiological statistics (Session 02)
2 Learning Objectives At the end of this session, you will be able to Explain the broad scope of epidemiology and why it is important to societyRecognise the primary importance of good quality routine data collection and its effective useDiscuss the place of cross-sectional surveys, longitudinal studies, research investigations in epidemiological practice
3 What is epidemiology?: 1Epidemiology is the study of the distribution and determinants of health-related states or events in specified populations, and the application of this study to control of health problems.“Study” includes surveillance, observation, hypothesis testing, analytic research and experiments.“Distribution” refers to analysis by time, place, and classes of persons affected.
4 What is epidemiology?: 2“Determinants” are all the physical, biological, social, cultural, and behavioural factors that influence health.“Health related states and events” include diseases, causes of death, behaviours such as use of tobacco, reactions to preventive regimens, and provision and use of health services.Above from Last, J.M. (ed.) A Dictionary of Epidemiology (2001, 4th ed.) Oxford U.P.
5 Routine epidemiological data Routine data, often based on returns from primary health care centres, hospitals and other health-related institutions.The title “surveillance” is given where there is systematic ongoing (“routine”) collection, collation, analysis and timely dissemination of information.“Monitoring” is where periodic or intermittent inspection of the data is undertaken.Methods need to be practical, uniform, quick.
6 Examples of routine data Mortality and morbidity reports based on death certificates, hospital records, G.P. information, or statutory notificationsLaboratory diagnosis records (pathology)Outbreak reports esp. infectious diseasesVaccine uptake and side-effect reportsEmployers’ sickness absence recordsFind and discuss two or three distinct examples of such data for your country.
7 Critique of routine data Uniform methods & recording completeness are never 100%. Need to:-make the best of what is available;where possible try to train and motivate health employees recording the data;check on data-collection circumstances, e.g. staffing or training gaps and develop ways to spot incomplete and dubious data.Sometimes need to select only sources that pass quality checks.
8 Epidemiological surveys Even where complete and ongoing data collection exists, the scope of record-keeping must be limited – cost and time.Surveys usually restricted to brief time periods & sub-areas. Areas may be a “representative” sample, or a “case study” of a few selected areas.With small no. of well-trained interviewers, a one-off field survey can explore much more complex themes than routine data.
9 Survey structure Varied objectives, usually mean compromise:- [Few questions, but large numbers] OR [Many questions, but small sample size]Example ~ WHO Expanded Programme of Immunisation. Sole objective to estimate % children having all required immunisations. Settled on norm of “30 x 7” ~ 30 clusters, & 7 children per cluster. Objective basically implies one YES/NO answer per subject.
10 Critique of survey data Sampling design may be complex; see module H6.Survey data chain often involves weak links and so poor quality data; build a checking system before accepting results.Often in reality coverage is limited by costs e.g. of travel, access difficulties etc. and generalisation from results compromised.Complex themes may address issues where people will not give complete and honest answers, or where questions incompletely address deeper issues.Users often over-interpret very specific results.
11 Survey activitiesTry to find a report of a multi-national health survey or a large national study.Look for key health results, and discuss what policy relevance they may have.Look for a description of the sample coverage: discuss its limitations.Look for clear statements about fieldwork methodology, justifications for wording of questionnaires, full use in report of all data collected : discuss reporting limitations.
13 Longitudinal data collection Diseases, for example, develop, spread, in a population, so we need measures over time. Separate surveys at two times may differ just because different individuals or sites are chosen. Comparability across time best if same units revisited:- panel study ~ ideally of same respondents, but some compromise needed. People move, household compositions change, respondents get tired/un-cooperative.
14 Sentinel site surveillance With observation at many time points, main emphasis is on time trajectories of results.Time to time comparisons important, & little interest in place to place comparisons, or “national” summation across sites. Random sampling of sites not very important and total number of sites generally small [merit of random sampling depends on large sample size !], so sites chosen non-randomly to pick up interesting site-specific results. Hence “sentinels” like look-outs at army camp.
15 Research data collection:1 Above all relate to situations where the item of interest e.g. disease is wide-spread in population. A more specialised example: some people in an area fall victim to an outbreak – cause or source unknown.In “case-control” study known cases are sampled, and a comparable sample of non-victim controls established. Then 2 groups are compared by retrospective questioning to identify likely “causes”.
16 Research data collection: 2 What the cases mostly have in common, but controls mostly do not have suggests how the outbreak came about. Method used by pioneer epidemiologist Snow identifying a polluted water supply as source of cholera in 19th century London.Case-control study needs 2 samples well matched for demographic factors e.g. age, but not for risk factors e.g. all same water source. Also reliant on memory quality of respondents.
17 Practical work follows to ensure learning objectives are achieved…