It is possible to obtain a valid estimate of disease prevalence from a typical case-control study.

Principal Features

Case–control and cohort studies complement each other as types of epidemiological study. In a case–control study, the groups are defined on the basis of the presence or absence of a given disease and, hence, only one disease can be studied at a time. The case–control study compensates for this by providing information on a wide range of exposures that may play a role in the development of the disease. In contrast, a cohort study generally focuses on a single exposure but can be analyzed for multiple disease outcomes. A case–control study is a better way of studying rare diseases because a very large cohort would be required to demonstrate an excess of a rare disease. In contrast, a case–control study is an inefficient way of assessing the effect of an uncommon exposure, when it might be possible to conduct a cohort study of all those exposed.

The complementary strengths and weaknesses of case–control and cohort studies can be used to advantage. Increasingly, mortality studies are being reported that utilize ‘nested’ case–control studies to investigate the association between the exposures of interest and a cause of death for which an excess has been discovered. However, case–control studies have traditionally been held in low regard, largely because they are often poorly conducted and interpreted. There is also a tendency to overinterpret the data and misuse statistical procedures. In addition, there is still considerable debate among leading epidemiologists themselves as to how controls should be selected.

Case–control studies

Case–control studies are also termed case–referent studies or, loosely and confusingly, sometimes retrospective studies. In case–control studies, the cases and noncases of the outcome in question are compared for their antecedent exposures. If exposure comparison is the purpose, the study by definition is usually a case–control study. The investigator generally has no influence over these antecedent exposures. The appropriate measure of risk in these studies is the exposure OR (see also Section 10.1 and below). The use of case–control designs is discussed further in the next section.

4.1.2.5 Case–Control Studies

Case–control studies usually choose subjects based on their disease status. A group of individuals who are disease positive (the “case” group) is compared with a group of individuals that are disease negative (the “control” group). The control group is supposed to come from the same population, ideally, that yielded the cases. Further, the case–control study looks back through time at potential exposures that both the groups (cases and controls) might have encountered. It may be pointed out here that case–control studies are usually faster and more cost-effective than cohort studies, but are sensitive to bias (both recall bias and selection bias). The main task is to identify the appropriate control group. The distribution of exposure among the members of the control group should preferably be representative of the distribution in the population that gave rise to the cases. However, there are certain drawbacks in case–control studies.

2.2.2.7 Case–Control Studies

Case–control studies are used in clinical teratology research to compare the frequency of a maternal exposure, such as treatment with a particular drug, during pregnancy among children with or without birth defects. Case–control studies are often population-based, an important factor in avoiding many kinds of ascertainment bias. Because case–control studies focus on women who have given birth to a baby with birth defects, case–control studies are usually far more statistically powerful than population-based cohort studies of an equivalent size. A major limitation of case-control studies is that they only provide information regarding the outcome or outcomes selected for study, so that an association with birth defects of an unanticipated kind, such as a previously-unrecognized pattern of minor anomalies, cannot be identified. In this regard, case definition and the system used to classify birth defects into the group(s) chosen for inclusion in the study may be of particular importance, and selection of a group of anomalies that are thought to have similar pathogenesis – for example, vascular disruption – may be more relevant than conventional anatomic classifications.

Case–control studies can be used to estimate the odds ratio and statistical significance of an association observed between birth defects in children and maternal teratogenic exposures. Like cohort studies, case–control studies depend on the quality of both the outcome data (e.g., case ascertainment) and the exposure data (e.g., exposure characterization and timing) and may be subject to serious biases and confounding. Another frequent concern with large population-based case–control studies is that many case groups involving different kinds of birth defects may be analyzed for associations with several different maternal exposures simultaneously, creating a “multiple comparisons” problem that may not be resolvable without additional investigations.

Sample size and consequent statistical power are, of course, important considerations in any epidemiological study, but the very high power of the case–control design when used with birth defect groups that are rare in the general population may raise a different issue: identification of a statistically significant risk that is real but clinically irrelevant. An exposure that doubles the risk of, say, sirenomelia to 2:100,000 from 1:100,000 in unexposed pregnancies but does not affect the risk of any other congenital anomaly might be of great interest in terms of pathogenesis but would be of little clinical consequence to an individual pregnant woman who seeks counseling.

CASE CONTROL STUDIES

Case control studies choose groups with (cases) and without (controls) the outcome of interest and look back at what different exposures they may have had to identify possible risk factors. Case control studies have been widely used in genetic studies to identify susceptibility genes and are the best design to study rare conditions, as they are efficient in use of time and money, collecting a lot of relevant information on targeted individuals. Case control studies may be “nested” within cohort studies.

Bias can be introduced when cases and controls differ in ways other than just the outcome of interest (selection bias) or when cases are not “typical” (representativeness bias). Given the increasing heterogeneity characteristic of aging, bias can be a significant problem and care needs to be taken to well match cases and controls. Recall bias may arise because subjects are able to remember events better because of their significance, or unintentionally they may be prompted to remember by investigators, who should therefore be blinded to whether the person is a case or control when assessing exposures. People who have died do not make it into case control studies and their representatives are likely to be less reliable than the person themselves at remembering exposures, introducing a potential survival bias. Although case control studies can play a pivotal role in suggesting important associations, as in the original studies linking cigarette smoking and lung cancer,14 confounding can also lead to highly misleading conclusions, as in the observational studies of combined hormone replacement therapy and cardiovascular disease in postmenopausal women.15

5.4.1.3 Case-control studies

Case-control studies utilize a group of animals diagnosed with a disease and a group that does not have the disease. The next step is to compare these groups for the presence or absence of a risk factor. An example of a case-control study was conducted by Worranut et al. (2018) in Thailand with an outbreak of white spot disease. This study concluded that postlarvae providers and farm visitors were the significant risk factors associated with the spread of the disease. Further, the researchers recommended purchasing postlarvae from reliable, tested sources and that farm visitors be limited to prevent further spread of the disease.

Case-Control Studies

Case-control studies compare animals with a specific outcome of interest (cases) with animals from the same source population but without that outcome (controls). After an outcome of interest is identified, typically a disease, a retrospective search is performed to identify exposures that could have been responsible for the outcome (e.g., exposure to oncogenic substance, or to an intervention).11 This design is particularly useful to investigate the etiology of rare diseases. Control animals should be similar to cases in all important factors except for not having the outcome in question (e.g., the disease).10 To reduce source of bias, selecting controls that are similar to cases for certain criteria (i.e., matching) may be implemented. The selection of an inappropriate control group may severely limit the validity of a case-control study. Another serious source problem of case-control study is recall bias, which occurs when there is a differential recall (i.e., different memory) of an exposure between the group of cases and the group of controls.

Case-Control Studies

Case-control studies compare prior consumption of a dietary factor in subjects with CRC and matched control subjects without CRC. Limitation in retrospective studies is the accuracy with which intake of dietary factors or supplementation can be established: individuals may misreport their habitual past diets. Furthermore, control individuals often have another disease that may be related to diet. Another problem associated with case-control studies is selection bias because of the absence of patients who do not survive long enough to be enrolled in the study. Three analyses have critically evaluated the bulk of case-control studies that address the role of dietary fiber in CRC. In a meta-analysis by Trock et al., 15 (65%) of 23 studies demonstrated either a strong or moderate protective effect of dietary fiber and vegetables. Only 2 studies (9%) lacked support for a protective effect of fiber. With fiber-rich diets, a 43% reduction in CRC risk was observed [odds ratio (OR), 0.57; 95% confidence interval (CI), 0.50–0.64] when the highest and lowest quartiles of intake were compared. Howe et al. performed a combined analysis of data from 13 case-control studies. The individual data records for 5287 case subjects and 10,470 control subjects were pooled for a common analysis. The risk of CRC was shown to decrease incrementally as dietary fiber intake increased. Consumption of more than 31 g of fiber per day was associated with a 47% reduction in the risk of CRC compared with diets incorporating less than 10 g of fiber per day (95% CI, 0.47–0.61). When all of the studies were combined and adjusted for total energy intake, age, and sex, individuals who consumed 27 g fiber per day had a 50% reduction in the risk of developing CRC compared with those who consumed less than 11 g fiber per day [relative risk (RR), 0.51; 95% CI, 0.44–0.59]. Friedenreich et al. examined the study design features and data colection methods from the 13 case-control studies to determine whether they influenced the results obtained from a pooled analysis. Their results show that subjects consuming >27 g fiber per day had a 50% reduction in the risk of developing CRC compared with those taking <11 g fiber per day (OR, 0.49; 95% CI, 0.37–0.65).

Several case-control studies have investigated the relationship between dietary fiber or fiber-rich foods and the risk of colonic adenomas, well-established precursors of adenocarcinoma. Data suggest that there exists an inverse relationship between fiber intake and the development of colonic adenomas. The magnitude of the reduction in the risk ranged from 10 to 60%. Some studies show a dose-dependent inverse association between colorectal adenoma risk and dietary intake of fiber but other studies suggest that the protective effect associated with dietary fiber is evident in women or for large (>1 cm) adenomas.

In summary, most of the published case-control studies show either a strong or a moderate protective effect of dietary fiber, consistent with the fiber hypothesis. Studies conducted in meta-analysis format provide strong support for the protective effect of dietary fiber on colorectal carcinogenesis. The strongest argument for the fiber hypothesis that can be made from case-control studies is the protective effect of dietary fiber among studies conducted in populations with different patterns of diet and CRC. The combined analysis and meta-analyses of case-control studies suggest, on average, a 50% reduction in the risk of developing CRC in individuals with the highest dietary fiber intake compared with those with the lowest fiber intake. Most case-control studies show a significant inverse dose-dependent relationship between dietary fiber intake and the risk of CRC and colorectal adenomas. Several shortcomings associated with case-control studies limit the interpretation of the results. Limitations associated with analytical methods of determining fiber content in diet, the effect of other potential anti-carcinogens present in fiber-rich foods, and questionnaires that have not been validated are all factors affecting study results.

Case-control design

Case-control studies are powerful designs that use highly selected study samples to compare a group of people with a phenotype, the “cases,” to a group of people without the phenotype, the “controls.” Thus, participants are selected for inclusion after the onset or diagnosis of the phenotype of interest, then biosamples for epigenomic assays may be obtained and information about prior exposures is typically collected retrospectively. In epigenetic case-control studies, this could mean that epigenetic features are measured within biosamples that were collected after the onset or diagnosis of the phenotype of interest and is a major limitation of this design. This introduces the possibility of reverse causation, in which the observed epigenetic differences between cases and controls could be a consequence of the phenotype, rather than a risk factor. Interestingly, case-control studies have been used for many genome-wide association studies (GWAS), partly because this phenotype-guided selection is useful for studying rare diseases, but also because genotype is established at conception and largely unperturbed throughout life, and thus is assumed to be antecedent to the emergence of the phenotype. Unfortunately, this same assumption will often be inappropriate for epigenomic studies, since many epigenetic features undergo changes with the development and aging and are responsive to exposures and pathological conditions. Thus, results from epigenetic case-control studies should be interpreted cautiously.

An additional consideration in case-control studies is how to define cases and controls. Overly restrictive case definitions can lead to small sample sizes and limited variability in the phenotype, while nonspecific case definitions can lead to misclassification in which the case group includes some people that do not actually have the phenotype of interest. The selection of controls is equally important, as this group should be composed of individuals that were at risk of developing the phenotype, but that are free of it, and their selection should be independent of any exposures that the researchers want to study.

Source attribution using systematic review of case-control studies

Case-control studies of sporadic infections are the most commonly applied approach for identifying possible exposures for sporadic foodborne disease including salmonellosis. Culture-confirmed case patients and a representative group of asymptomatic individuals (controls) are interviewed, and the frequency of exposures among cases and controls is compared and the association between disease and a certain exposure is quantified by calculating the odds ratio (OR). The proportion of cases attributed to the exposure can then be calculated and is defined as the population attributable fraction (PAF) [77]. The population attributable fractions can be used to attribute the human disease burden to specific sources.

Numerous case-control studies of sporadic infections of diseases commonly transmitted through food have been published, as reviewed by [78,79], and case-control studies are considered a valuable tool to identify potential risk factors for human infections, including sources and predisposing, behavioral, or seasonal factors [80]. Limitations of case-control studies include misclassification due to immunity, which may reduce attributable risk or even suggest protection. Likewise, misclassification of exposures due to lack of accuracy of recall may lead to an underestimation of the burden of illness attributed to specific exposures. Most studies only explain a small fraction of all cases, and cases may reflect a mixture of possible sources of exposure, which can make it difficult to distinguish between these exposures. Lastly, statistical power to determine the importance of common exposures often requires enrollment of many participants.

A systematic review (SR) of published case-control studies of sporadic infections can, in addition, provide an overview of the relevant exposures and risk factors and may identify geographical, temporal, or age-related differences regarding the most important exposures [60]. An SR follows a rigorous search strategy to identify all relevant peer-review case-control studies for a hazard, studies being conducted in a variety of countries and time periods, designed with different settings, and potentially focused on specific age groups within the population. Data from the different studies are then combined in a meta-analysis, where risk factors can be stratified according to predefined source-categorization schemes, location of exposures, and, if appropriate, frequency of exposure. The weighted summary statistics of several case-control studies may then be combined with estimates of the burden of disease caused by that hazard to estimate the burden of disease attributed to each exposure.

Recently conducted an SR of case-control studies and a meta-analysis to identify risk factors for salmonellosis [78]. Data from 35 case-control studies from 11 countries were included. Investigated risk factors represented consumption of foods, direct contact with live animals, environment transmission, predisposition, and behavioral factors. Results showed that international travel (OR 6.5), intake of antacids (OR 2.9), pre-existent medical condition (OR 2.8), previous intake of antimicrobials (OR 2.23), eating raw eggs (OR 2.78), and eating in a restaurant (OR 2.74) were the most important risk factors for human salmonellosis in the overall study population. Consumption of undercooked or raw eggs and chicken in a restaurant were the only food items identified as relevant for human disease in the analysis, and environmental routes (both drinking and recreational waters), direct contact with pets and farm animals, and various predisposition factors proved to play a role in human salmonellosis (Figure 5.5). The results of the analyses focusing on serotypes suggested that traveling abroad and consumption of eggs are particularly important risk factors for S. Enteritidis infection, while previous intake of antimicrobials was the only risk factor identified for S. Typhimurium. Available studies did not allow for an analysis by region or age group.

An SR of epidemiological studies can only be undertaken if a sufficient number of studies focusing on the same hazard have been conducted and made publically available. Also, once a SR based on all relevant studies has been conducted, a new review or update of a previous review will only add to the existent knowledge if results from a substantial number of new case-control studies are available for inclusion.