In comparison, a more systematic approach is comparison group studies, where a comparable group of individuals is included for evaluating the linkage between potential risk factors and the health outcomes. Two common observational epidemiological study designs are case control studies, and cohort studies. In case control studies, individuals with health outcomes (cases) are compared with those with no evidence of health outcome (controls). Both groups are selected by the investigator. These two groups are comparable to each other in every way other than the fact that one of these groups has the illness or health outcome. These groups are evaluated with respect to their relative exposures to the different risk factors of interest. For a case control study, effect sizes are calculated as ratios of likelihood (Odds Ratios) of exposure to the risk factors (4). In a cohort study, the investigator begins with two groups both the groups are initially free of the health outcome. One of the groups is exposed to the risk factor of interest, while the other group is not. The incidences of health outcomes are followed prospectively. The incidence rates of the health outcomes are then compared among the exposed and non-exposed individuals. The effect size is expressed as the ratio (Relative Risk or Rate Ratio) of the incidence of disease among exposed versus incidence of disease among non-exposed (4).
A case control study is shorter and less expensive. It provides possibility of studying different exposures for rare diseases. However, from the perspective of deriving causal inferences, it is less reliable compared to a cohort study design in that it cannot account for temporal sequence, and case control studies are open to different types of biases. On the other hand, a cohort study, although a powerful design, is more expensive, and is open to problems of loss of study participants to follow up. A cohort study is not suitable for studying rare diseases or diseases that take a long time to develop. However, given a set of exposures, multiple outcomes can be studied using a cohort study design (3, 4). In nested case control study designs, a case control study is embedded (“nested”) within a longitudinal prospective cohort study. Typically, at the initiation of the cohort study, blood samples are stored for each individual in the cohorts. Then, down time, as a few cases start appearing, these cases and corresponding controls are used in a case control study to assess the relationship between different serum level metabolites or biochemical parameters and select health outcomes (4).
Beyond hypothesis generating studies and grouped comparison studies, meta analyses and systematic reviews generate important information by pooling together results of different epidemiological studies. Systematic reviews are compilations of results of different epidemiological studies that are selected according to fixed criteria, and then the results are pooled together to arrive at summary measures of the relationship between a risk factor and a health outcome. Meta analyses are conceptually similar and have been applied to results of randomized trials (3).
Results from epidemiological studies and effect sizes are used in different ways in formulating strategies for prevention and public health approaches. Prevalence and incidence measures are essential to quantify the health outcomes or diseases. Relative risk estimates from cohort studies can be used to identify the impact of the exposure on the outcome by calculating absolute risk reduction scores, relative risk reduction scores, and numbers needed to treat scores to translate the results into public health actions.
While epidemiological study designs can be widely used for investigating disease processes at workplaces and work sites, the primary limitations of epidemiological studies in an occupational setting using questionnaires is the concept of “healthy” worker effect. In an occupational setting, when an epidemiological investigation is conducted, the measurements are readily available on workers who are healthier and therefore present for duties. Measurements of health outcomes are missed or inadequately represented for workers who are too sick to attend their duties, or workers who degree of illness is low enough to enable them attend their job responsibilities.