Chapter 2 Nature of the evidence
Chapter overview Introduction What is epidemiology? Measuring physical activity and fitness in population studies Laboratory-based research Error: nature, sources and implications Establishing causality Summary
Epidemiology 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’. (World Health Organization)
Types of study in physical activity and health Epidemiological Mainly observational Allow nature to take its course and analyse relationships between indices of health status and other variables Laboratory-based Mainly experimental Intervene to see what happens to some/all individuals
Types of epidemiological study CategoryType of studyUnit of study Observational Descriptive studiesCase reports or seriesIndividuals Analytical studiesCorrelationalPopulations Cross-sectional surveysIndividuals Case-control studiesIndividuals Cohort studiesIndividuals ExperimentalRandomized, controlled trials Individuals
Examples of disease outcomes Simple Disease present Disease absent Graded Normal weight Overweight Obese
Cholesterol and mortality from CHD in the Seven Countries Study
Harvard Alumni Study cardiovascular diseases; longevity; diabetes; gallbladder disease; several site-specific; cancers; Parkinson’s disease; depression; suicide. A cohort study that began in 1962, has studied a range of health outcomes, including:
Measures of occurrence of health-related outcomes Prevalence: the proportion of individuals in a population that exhibits the outcome of interest at a specified time. Incidence: the number of new occurrences of an outcome that develop during a specified time interval. Best measure is person–time incidence rate.
Calculation of person–time incidence rate
Comparisons of disease occurrence between exposed and unexposed groups These are essential tools in epidemiology. They include: risk difference; relative risk; population-attributable risk; odds ratio (similar to relative risk, used in many case-control studies).
Vigorous sports and attack rate of CHD in English civil servants Episodes of vigorous sport in previous four weeks, reported in 1976 CHD cases Man-years of observation Age-standardized rate (cases per 1,000 man-years) None (reference group)41372, –31–3 377, – 12 73,3492.1
All-cause mortality risk among Harvard Alumni 1962–78 CharacteristicPrevalence (% of man-years) Relative riskPopulation- attributable risk (%) Sedentary lifestyle Hypertension Cigarette smoking
Measuring physical activity/fitness Job classification; leisure-time activity; –questionnaire –pedometer –accelerometer –total energy expenditure by doubly labelled water fitness –direct VO 2 max (treadmill or cycle ergometer) –predict VO 2 max from sub-maximal heart rate –functional measure, e.g. time to exhaustion, watts achieved, level in shuttle walking test...
Recall of lifetime participation in physical activity
A randomized, controlled, laboratory-based intervention study
Importance of control group: effect of training on heart rate
Accuracy and precision I
Accuracy and precision II Data are accurate if they are close to the true values; and precise if the same measurement, when repeated, consistently yields similar values.
Cause and effect The role of chance, random error: –sampling –measurement. Bias – systematic error: –subject selection –measurement. Confounding: –observed association is due to a third factor related to the exposure that independently affects the risk of developing the disease – a confounding variable.
Epidemiology and causality? Appropriately sequenced; measure of activity/fitness must precede onset of disease Biologically plausible, i.e. is association consistent with other knowledge? Strength – relative risk Dose–response Reversibility Strong study design Consistency in different populations
Summary Epidemiology can identify risk factors. In epidemiology, physical activity is most commonly measured by questionnaire. Relative risk estimates the strength of an association with a risk factor. Associations may reflect the true effect of an exposure, but may also reflect chance, bias or confounding. Laboratory studies can achieve excellent control and precision and indicate potential mechanisms, but are removed from clinical endpoints. The totality of the evidence (epidemiology and laboratory-based) determines decisions as to causality.