analysis of case control study

Statistical analyses of case-control studies

How Evidence-based practice (EBP) can be translated as health communication or patient education materials

How to evaluate bias in meta-analysis within meta-epidemiological studies?

Introduction.

A case-control study is used to see if exposure is linked to a certain result (i.e., disease or condition of interest). Case-control research is always retrospective by definition since it starts with a result and then goes back to look at exposures. The investigator already knows the result of each participant when they are enrolled in their separate groups. Case-control studies are retrospective because of this, not because the investigator frequently uses previously gathered data. This article discusses statistical analysis in case-control studies.

Advantages and Disadvantages of Case-Control Studies

Study Design

Participants in a case-control study are chosen for the study depending on their outcome status. As a result, some individuals have the desired outcome (referred to as cases), while others do not have the desired outcome (referred to as controls). After that, the investigator evaluates the exposure in both groups. As a result, in case-control research , the outcome must occur in at least some individuals. Thus, as shown in Figure 1, some research participants have the outcome, and others do not enrol.

Figure 1. Example of a case-control study [1]

Selection of case

The cases should be defined as precisely as feasible by the investigator. A disease’s definition may be based on many criteria at times; hence, all aspects should be fully specified in the case definition.

Selection of a control

Controls that are comparable to the cases in a variety of ways should be chosen. The matching criteria are the parameters (e.g., age, sex, and hospitalization time) used to establish how controls and cases should be similar. For instance, it would be unfair to compare patients with elective intraocular surgery to a group of controls with traumatic corneal lacerations. Another key feature of a case-control study is that the exposure in both cases and controls should be measured equally.

Though some controls have to be similar to cases in many respects, it is possible to over-match. Over-matching might make it harder to identify enough controls. Furthermore, once a matching variable is chosen, it cannot be analyzed as a risk factor. Enrolling more than one control for each case is an effective method for increasing the power of research. However, incorporating more than two controls per instance adds little statistical value.

Data collection

Decide on the data to be gathered after precisely identifying the cases and controls; both groups must have the same data obtained in the same method. If the search for primary risk variables is not conducted objectively, the study may suffer from researcher bias, especially because the conclusion is already known. It’s crucial to try to hide the outcome from the person collecting risk factor data or interviewing patients, even if it’s not always practicable. Patients may be asked questions concerning historical issues (such as smoking history, food, usage of conventional eye medications, and so on). For some people, precisely recalling all of this information may be challenging.

Furthermore, patients who get the result (cases) are more likely to recall specifics of unfavourable experiences than controls. Recall bias is a term for this phenomenon. Any effort made by the researcher to reduce this form of bias would benefit the research.

The frequency of each of the measured variables in each of the two groups is computed in the analysis. Case-control studies produce the odds ratio to measure the strength of the link between exposure and the outcome. An odds ratio is the ratio of exposure probabilities in the case group to the odds of response in the control group. Calculating a confidence interval for each odds ratio is critical. A confidence interval of 1.0 indicates that the link between the exposure and the result might have been discovered by chance alone and that the link is not statistically significant. Without a confidence interval, an odds ratio isn’t particularly useful. Computer programmes are typically used to do these computations. Because no measures are taken in a population-based sample, case-control studies cannot give any information regarding the incidence or prevalence of a disease.

Risk Factors and Sampling

Case-control studies can also be used to investigate risk factors for a rare disease. Cases might be obtained from hospital records. Patients who present to the hospital, on the other hand, may not be typical of the general community. The selection of an appropriate control group may provide challenges. Patients from the same hospital who do not have the result are a common source of controls. However, hospitalized patients may not always reflect the broader population; they are more likely to have health issues and access the healthcare system.

Recent research on case-control studies using statistical analyses

i) R isk factors related to multiple sclerosis in Kuwait

This matched case-control research in Kuwait looked at the relationship between several variables: family history, stressful life events, tobacco smoke exposure, vaccination history, comorbidity, and multiple sclerosis (MS) risk. To accomplish the study’s goal, a matched case-control strategy was used. Cases were recruited from Ibn Sina Hospital’s neurology clinics and the Dasman Diabetes Institute’s MS clinic. Controls were chosen from among Kuwait University’s faculty and students. A generalized questionnaire was used to collect data on socio-demographic, possibly genetic, and environmental aspects from each patient and his/her pair-matched control. Descriptive statistics were produced, including means and standard deviations for quantitative variables and frequencies for qualitative variables. Variables that were substantially (p ≤ 0.15) associated with MS status in the univariable conditional logistic regression analysis were evaluated for inclusion in the final multivariable conditional logistic regression model. In this case-control study, 112 MS patients were invited to participate, and 110 (98.2 %) agreed to participate. Therefore, 110 MS patients and 110 control participants were enlisted, and they were individually matched with cases (1:1) on age (5 years), gender, and nationality (Fig. 1). The findings revealed that having a family history of MS was significantly associated with an increased risk of developing MS. In contrast, vaccination against influenza A and B viruses provided significant protection against MS.

Figure 1. Flow chart on the enrollment of the MS cases and controls [1]

ii) Relation between periodontitis and COVID-19 infection

COVID-19 is linked to a higher inflammatory response, which can be deadly. Periodontitis is characterized by systemic inflammation. In Qatar, patients with COVID-19 were chosen from Hamad Medical Corporation’s (HMC) national electronic health data. Patients with COVID-19 problems (death, ICU hospitalizations, or assisted ventilation) were categorized as cases, while COVID-19 patients released without severe difficulties were categorized as controls. There was no control matching because all controls were included in the analysis. Periodontal problems were evaluated using dental radiographs from the same database. The relationships between periodontitis and COVID 19 problems were investigated using logistic regression models adjusted for demographic, medical, and behavioural variables. 258 of the 568 participants had periodontitis. Only 33 of the 310 patients with periodontitis had COVID-19 issues, whereas only 7 of the 310 patients without periodontitis had COVID-19 issues. Table 2 shows the unadjusted and adjusted odds ratios and 95 % confidence intervals for the relationship between periodontitis and COVID-19 problems. Periodontitis was shown to be substantially related to a greater risk of COVID-19 complications, such as ICU admission, the requirement for assisted breathing, and mortality, as well as higher blood levels of indicators connected to a poor COVID-19 outcome, such as D-dimer, WBC, and CRP.

Table 2. Associations between periodontal condition and COVID-19 complications [3]

iii) Menstrual, reproductive and hormonal factors and thyroid cancer

The relationships between menstrual, reproductive, and hormonal variables and thyroid cancer incidence in a population of Chinese women were investigated in this study. A 1:1 corresponding hospital-based Case-control study was conducted in 7 counties of Zhejiang Province to investigate the correlations of diabetes mellitus and other variables with thyroid cancer. Case participants were eligible if they were diagnosed with primary thyroid cancer for the first time in a hospital between July 2015 and December 2017. The patients and controls in this research were chosen at random. At enrollment, the interviewer gathered all essential information face-to-face using a customized questionnaire. Descriptive statistics were utilized to characterize the baseline characteristics of female individuals using frequency and percentage. To investigate the connections between the variables and thyroid cancer, univariate conditional logistic regression models were used. We used four multivariable conditional logistic regression models adjusted for variables to investigate the relationships between menstrual, reproductive, and hormonal variables and thyroid cancer. In all, 2937 pairs of participants took part in the case-control research. The findings revealed that a later age at first pregnancy and a longer duration of breastfeeding were substantially linked with a lower occurrence of thyroid cancer, which might shed light on the aetiology, monitoring, and prevention of thyroid cancer in Chinese women [4].

It’s important to note that the term “case-control study” is commonly misunderstood. A case-control study starts with a group of people exposed to something and a comparison group (control group) who have not been exposed to anything and then follows them over time to see what occurs. However, this is not a case-control study. Case-control studies are frequently seen as less valuable since they are retrospective. They can, however, be a highly effective technique of detecting a link between an exposure and a result. In addition, they are sometimes the only ethical approach to research a connection. Case-control studies can provide useful information if definitions, controls, and the possibility for bias are carefully considered.

[1] Setia, Maninder Singh. “Methodology Series Module 2: Case-control Studies.” Indian journal of dermatology vol. 61,2 (2016): 146-51. doi:10.4103/0019-5154.177773

[2] El-Muzaini, H., Akhtar, S. & Alroughani, R. A matched case-control study of risk factors associated with multiple sclerosis in Kuwait. BMC Neurol 20, 64 (2020). https://doi.org/10.1186/s12883-020-01635-1 .

[3] Marouf, Nadya, Wenji Cai, Khalid N. Said, Hanin Daas, Hanan Diab, Venkateswara Rao Chinta, Ali Ait Hssain, Belinda Nicolau, Mariano Sanz, and Faleh Tamimi. “Association between periodontitis and severity of COVID‐19 infection: A case–control study.” Journal of clinical periodontology 48, no. 4 (2021): 483-491.

[4] Wang, Meng, Wei-Wei Gong, Qing-Fang He, Ru-Ying Hu, and Min Yu. “Menstrual, reproductive and hormonal factors and thyroid cancer: a hospital-based case-control study in China.” BMC Women’s Health 21, no. 1 (2021): 1-8.

pubrica-academy

Related posts.

PUB - Selecting material (e.g. excipient, active pharmaceutical ingredient) for drug development

Selecting material (e.g. excipient, active pharmaceutical ingredient, packaging material) for drug development

PUB - Health Economics of Data Modeling

Health economics in clinical trials

PUB - Epidemiology designs for clinical trials

Epidemiology designs for clinical trials

Comments are closed.

An official website of the United States government

The .gov means it’s official. Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

The site is secure. The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

• Publications
• Account settings
• Advanced Search
• Journal List
• Indian J Dermatol
• v.61(2); Mar-Apr 2016

Methodology Series Module 2: Case-control Studies

Maninder singh setia.

Epidemiologist, MGM Institute of Health Sciences, Navi Mumbai, Maharashtra, India

Case-Control study design is a type of observational study. In this design, participants are selected for the study based on their outcome status. Thus, some participants have the outcome of interest (referred to as cases), whereas others do not have the outcome of interest (referred to as controls). The investigator then assesses the exposure in both these groups. The investigator should define the cases as specifically as possible. Sometimes, definition of a disease may be based on multiple criteria; thus, all these points should be explicitly stated in case definition. An important aspect of selecting a control is that they should be from the same ‘study base’ as that of the cases. We can select controls from a variety of groups. Some of them are: General population; relatives or friends; and hospital patients. Matching is often used in case-control control studies to ensure that the cases and controls are similar in certain characteristics, and it is a useful technique to increase the efficiency of the study. Case-Control studies can usually be conducted relatively faster and are inexpensive – particularly when compared with cohort studies (prospective). It is useful to study rare outcomes and outcomes with long latent periods. This design is not very useful to study rare exposures. Furthermore, they may also be prone to certain biases – selection bias and recall bias.

Introduction

Case-Control study design is a type of observational study design. In an observational study, the investigator does not alter the exposure status. The investigator measures the exposure and outcome in study participants, and studies their association.

In a case-control study, participants are selected for the study based on their outcome status. Thus, some participants have the outcome of interest (referred to as cases), whereas others do not have the outcome of interest (referred to as controls). The investigator then assesses the exposure in both these groups. Thus, by design, in a case-control study the outcome has to occur in some of the participants that have been included in the study.

As seen in Figure 1 , at the time of entry into the study (sampling of participants), some of the study participants have the outcome (cases) and others do not have the outcome (controls). During the study procedures, we will examine the exposure of interest in cases as well as controls. We will then study the association between the exposure and outcome in these study participants.

Example of a case-control study

Examples of Case-Control Studies

Smoking and lung cancer study.

In their landmark study, Doll and Hill (1950) evaluated the association between smoking and lung cancer. They included 709 patients of lung carcinoma (defined as cases). They also included 709 controls from general medical and surgical patients. The selected controls were similar to the cases with respect to age and sex. Thus, they included 649 males and 60 females in cases as well as controls.

They found that only 0.3% of males were non-smokers among cases. However, the proportion of non-smokers among controls was 4.2%; the different was statistically significant ( P = 0.00000064). Similarly they found that about 31.7% of the female were non-smokers in cases compared with 53.3% in controls; this difference was also statistically significant (0.01< p <0.02).

Melanoma and tanning (Lazovic et al ., 2010)

The authors conducted a case-control study to study the association between melanoma and tanning. The 1167 cases - individuals with invasive cutaneous melanoma – were selected from Minnesota Cancer Surveillance System. The 1101 controls were selected randomly from Minnesota State Driver's License list; they were matched for age (+/- 5 years) and sex.

The data were collected by self administered questionnaires and telephone interviews. The investigators assessed the use of tanning devices (using photographs), number of years, and frequency of use of these devices. They also collected information on other variables (such as sun exposure; presence of freckles and moles; and colour of skin, hair, among other exposures.

They found that melanoma was higher in individuals who used UVB enhances and primarily UVA-emitting devices. The risk of melanoma also increased with increase in years of use, hours of use, and sessions.

Risk factors for erysipelas (Pitché et al, 2015)

Pitché et al (2015) conducted a case-control study to assess the factors associated with leg erysipelas in sub-Saharan Africa. This was a multi-centre study; the cases and controls were recruited from eight countries in sub-Saharan Africa.

They recruited cases of acute leg cellulitis in these eight countries. They recruited two controls for each case; these were matched for age (+/- 5 years) and sex. Thus, the final study has 364 cases and 728 controls. They found that leg erysipelas was associated with obesity, lympoedema, neglected traumatic wound, toe-web intertrigo, and voluntary cosmetic depigmentation.

We have provided details of all the three studies in the bibliography. We strongly encourage the readers to read the papers to understand some practical aspects of case-control studies.

Selection of Cases and Controls

Selection of cases and controls is an important part of this design. Wacholder and colleagues (1992 a, b, and c) have published wonderful manuscripts on design and conduct of case-control of studies in the American Journal of Epidemiology. The discussion in the next few sections is based on these manuscripts.

Selection of case

The investigator should define the cases as specifically as possible. Sometimes, definition of a disease may be based on multiple criteria; thus, all these points should be explicitly stated in case definition.

For example, in the above mentioned Melanoma and Tanning study, the researchers defined their population as any histologic variety of invasive cutaneous melanoma. However, they added another important criterion – these individuals should have a driver's license or State identity card. This probably is not directly related to the clinic condition, so why did they add this criterion? We will discuss this in detail in the next few paragraphs.

Selection of a control

The next important point in designing a case-control study is the selection of control patients.

In fact, Wacholder and colleagues have extensively discussed aspects of design of case control studies and selection of controls in their article.

According to them, an important aspect of selecting a control is that they should be from the same ‘study base’ as that of the cases. Thus, the pool of population from which the cases and controls will be enrolled should be same. For instance, in the Tanning and Melanoma study, the researchers recruited cases from Minnesota Cancer Surveillance System; however, it was also required that these cases should either have a State identity card or Driver's license. This was important since controls were randomly selected from Minnesota State Driver's license list (this also included the list of individuals who have the State identity card).

Another important aspect of a case-control study is that we should measure the exposure similarly in cases and controls. For instance, if we design a research protocol to study the association between metabolic syndrome (exposure) and psoriasis (outcome), we should ensure that we use the same criteria (clinically and biochemically) for evaluating metabolic syndrome in cases and controls. If we use different criteria to measure the metabolic syndrome, then it may cause information bias.

Types of Controls

We can select controls from a variety of groups. Some of them are: General population; relatives or friends; or hospital patients.

Hospital controls

An important source of controls is patients attending the hospital for diseases other than the outcome of interest. These controls are easy to recruit and are more likely to have similar quality of medical records.

However, we have to be careful while recruiting these controls. In the above example of metabolic syndrome and psoriasis, we recruit psoriasis patients from the Dermatology department of the hospital as controls. We recruit patients who do not have psoriasis and present to the Dermatology as controls. Some of these individuals have presented to the Dermatology department with tinea pedis. Do we recruit these individuals as controls for the study? What is the problem if we recruit these patients? Some studies have suggested that diabetes mellitus and obesity are predisposing factors for tinea pedis. As we know, fasting plasma glucose of >100 mg/dl and raised trigylcerides (>=150 mg/dl) are criteria for diagnosis of metabolic syndrome. Thus, it is quite likely that if we recruit many of these tinea pedis patients, the exposure of interest may turn out to be similar in cases and controls; this exposure may not reflect the truth in the population.

Relative and friend controls

Relative controls are relatively easy to recruit. They can be particularly useful when we are interested in trying to ensure that some of the measurable and non-measurable confounders are relatively equally distributed in cases and controls (such as home environment, socio-economic status, or genetic factors).

Another source of controls is a list of friends referred by the cases. These controls are easy to recruit and they are also more likely to be similar to the cases in socio-economic status and other demographic factors. However, they are also more likely to have similar behaviours (alcohol use, smoking etc.); thus, it may not be prudent to use these as controls if we want to study the effect of these exposures on the outcome.

Population controls

These controls can be easily conducted the list of all individuals is available. For example, list from state identity cards, voter's registration list, etc., In the Tanning and melanoma study, the researchers used population controls. They were identified from Minnesota state driver's list.

We may have to use sampling methods (such as random digit dialing or multistage sampling methods) to recruit controls from the population. A main advantage is that these controls are likely to satisfy the ‘study-base’ principle (described above) as suggested by Wacholder and colleagues. However, they can be expensive and time consuming. Furthermore, many of these controls will not be inclined to participate in the study; thus, the response rate may be very low.

Matching in a Case-Control Study

Matching is often used in case-control control studies to ensure that the cases and controls are similar in certain characteristics. For example, in the smoking and lung cancer study, the authors selected controls that were similar in age and sex to carcinoma cases. Matching is a useful technique to increase the efficiency of study.

’Individual matching’ is one common technique used in case-control study. For example, in the above mentioned metabolic syndrome and psoriasis, we can decide that for each case enrolled in the study, we will enroll a control that is matched for sex and age (+/- 2 years). Thus, if 40 year male patient with psoriasis is enrolled for the study as a case, we will enroll a 38-42 year male patient without psoriasis (and who will not be excluded for other reason) as controls.

If the study has used ‘individual matching’ procedures, then the data should also reflect the same. For instance, if you have 45 males among cases, you should also have 45 males among controls. If you show 60 males among controls, you should explain the discrepancy.

Even though matching is used to increase the efficiency in case-control studies, it may have its own problems. It may be difficult to fine the exact matching control for the study; we may have to screen many potential enrollees before we are able to recruit one control for each case recruited. Thus, it may increase the time and cost of the study.

Nonetheless, matching may be useful to control for certain types of confounders. For instance, environment variables may be accounted for by matching controls for neighbourhood or area of residence. Household environment and genetic factors may be accounted for by enrolling siblings as controls.

If we use controls from the past (time period when cases did not occur), then the controls are sometimes referred to historic controls. Such controls may be recruited from past hospital records.

Strengths of a Case-Control Study

• Case-Control studies can usually be conducted relatively faster and are inexpensive – particularly when compared with cohort studies (prospective)
• It is useful to study rare outcomes and outcomes with long latent periods. For example, if we wish to study the factors associated with melanoma in India, it will be useful to conduct a case-control study. We will recruit cases of melanoma as cases in one study site or multiple study sites. If we were to conduct a cohort study for this research question, we may to have follow individuals (with the exposure under study) for many years before the occurrence of the outcome
• It is also useful to study multiple exposures in the same outcome. For example, in the metabolic syndrome and psoriasis study, we can study other factors such as Vitamin D levels or genetic markers
• Case-control studies are useful to study the association of risk factors and outcomes in outbreak investigations. For instance, Freeman and colleagues (2015) in a study published in 2015 conducted a case-control study to evaluate the role of proton pump inhibitors in an outbreak of non-typhoidal salmonellosis.

Limitations of a Case-control Study

• The design, in general, is not useful to study rare exposures. It may be prudent to conduct a cohort study for rare exposures

Since the investigator chooses the number of cases and controls, the proportion of cases may not be representative of the proportion in the population. For instance if we choose 50 cases of psoriasis and 50 controls, the prevalence of proportion of psoriasis cases in our study will be 50%. This is not true prevalence. If we had chosen 50 cases of psoriasis and 100 controls, then the proportion of the cases will be 33%.

• The design is not useful to study multiple outcomes. Since the cases are selected based on the outcome, we can only study the association between exposures and that particular outcome
• Sometimes the temporality of the exposure and outcome may not be clearly established in case-control studies
• The case-control studies are also prone to certain biases

If the cases and controls are not selected similarly from the study base, then it will lead to selection bias.

• Odds Ratio: We are able to calculate the odds ratios (OR) from a case-control study. Since we are not able to measure incidence data in case-control study, an odds ratio is a reasonable measure of the relative risk (under some assumptions). Additional details about OR will be discussed in the biostatistics section.

The OR in the above study is 3.5. Since the OR is greater than 1, the outcome is more likely in those exposed (those who are diagnosed with metabolic syndrome) compared with those who are not exposed (those who do are not diagnosed with metabolic syndrome). However, we will require confidence intervals to comment on further interpretation of the OR (This will be discussed in detail in the biostatistics section).

• Other analysis : We can use logistic regression models for multivariate analysis in case-control studies. It is important to note that conditional logistic regressions may be useful for matched case-control studies.

Calculating an Odds Ratio (OR)

Hypothetical study of metabolic syndrome and psoriasis

Additional Points in A Case-Control Study

How many controls can i have for each case.

The most optimum case-to-control ratio is 1:1. Jewell (2004) has suggested that for a fixed sample size, the chi square test for independence is most powerful if the number of cases is same as the number of controls. However, in many situations we may not be able recruit a large number of cases and it may be easier to recruit more controls for the study. It has been suggested that we can increase the number of controls to increase statistical power (if we have limited number of cases) of the study. If data are available at no extra cost, then we may recruit multiple controls for each case. However, if it is expensive to collect exposure and outcome information from cases and controls, then the optimal ratio is 4 controls: 1 case. It has been argued that the increase in statistical power may be limited with additional controls (greater than four) compared with the cost involved in recruiting them beyond this ratio.

I have conducted a randomised controlled trial. I have included a group which received the intervention and another group which did not receive the intervention. Can I call this a case-control study?

A randomised controlled trial is an experimental study. In contrast, case-control studies are observational studies. These are two different groups of studies. One should not use the word case-control study for a randomised controlled trial (even though you have a control group in the study). Every study with a control group is not a case-control study. For a study to be classified as a case-control study, the study should be an observational study and the participants should be recruited based on their outcome status (some have the disease and some do not).

Should I call case-control studies prospective or retrospective studies?

In ‘The Dictionary of Epidemiology’ by Porta (2014), the authors have suggested that even though the term ‘retrospective’ was used for case-control studies, the study participants are often recruited prospectively. In fact, the study on risk factors for erysipelas (Pitché et al ., 2015) was a prospective case case-control study. Thus, it is important to remember that the nature of the study (case-control or cohort) depends on the sampling method. If we sample the study participants based on exposure and move towards the outcome, it is a cohort study. However, if we sample the participants based on the outcome (some with outcome and some do not) and study the exposures in both these groups, it is a case-control study.

In case-control studies, participants are recruited on the basis of disease status. Thus, some of participants have the outcome of interest (referred to as cases), whereas others do not have the outcome of interest (referred to as controls). The investigator then assesses the exposure in both these groups. Case-control studies are less expensive and quicker to conduct (compared with prospective cohort studies at least). The measure of association in this type of study is an odds ratio. This type of design is useful for rare outcomes and those with long latent periods. However, they may also be prone to certain biases – selection bias and recall bias.

Financial support and sponsorship

Conflicts of interest.

There are no conflicts of interest.

Bibliography

• - Google Chrome

Intended for healthcare professionals

• Access provided by Google Indexer
• My email alerts
• BMA member login
• Username * Password * Forgot your log in details? Need to activate BMA Member Log In Log in via OpenAthens Log in via your institution

Search form

• Advanced search
• Search responses
• Search blogs
• Analysis of matched...

Analysis of matched case-control studies

• Related content
• Peer review
• Neil Pearce , professor 1 2
• 1 Department of Medical Statistics and Centre for Global NCDs, Faculty of Epidemiology and Population Health, London School of Hygiene and Tropical Medicine, London WC1E 7HT, UK
• 2 Centre for Public Health Research, Massey University, Wellington, New Zealand
• neil.pearce{at}lshtm.ac.uk
• Accepted 30 December 2015

There are two common misconceptions about case-control studies: that matching in itself eliminates (controls) confounding by the matching factors, and that if matching has been performed, then a “matched analysis” is required. However, matching in a case-control study does not control for confounding by the matching factors; in fact it can introduce confounding by the matching factors even when it did not exist in the source population. Thus, a matched design may require controlling for the matching factors in the analysis. However, it is not the case that a matched design requires a matched analysis. Provided that there are no problems of sparse data, control for the matching factors can be obtained, with no loss of validity and a possible increase in precision, using a “standard” (unconditional) analysis, and a “matched” (conditional) analysis may not be required or appropriate.

Summary points

Matching in a case-control study does not control for confounding by the matching factors

A matched design may require controlling for the matching factors in the analysis

However, it is not the case that a matched design requires a matched analysis

A “standard” (unconditional) analysis may be most valid and appropriate, and a “matched” (conditional) analysis may not be required or appropriate

Matching on factors such as age and sex is commonly used in case-control studies. 1 This can be done for convenience (eg, choosing a control admitted to hospital on the same day as the case), to improve study efficiency by improving precision (under certain conditions) when controlling for the matching factors (eg, age, sex) in the analysis, or to enable control in the analysis of unquantifiable factors such as neighbourhood characteristics (eg, by choosing neighbours as controls and then controlling for neighbourhood in the analysis). The increase in efficiency occurs because it ensures similar numbers of cases and controls in confounder strata. For example, in a study of lung cancer, if controls are sampled at random from the source population, their age distribution will be much younger than that of the lung cancer cases. Thus, when age is controlled in the analysis, the young age stratum may contain mostly controls and few cases, whereas the old age stratum may contain mostly cases and fewer controls. Thus, statistical precision may be improved if controls are age matched to ensure roughly equal numbers of cases and controls in each age stratum.

There are two common misconceptions about case-control studies: that matching in itself eliminates confounding by the matching factors; and that if matching has been performed, then a “matched analysis” is required.

Matching in the design does not control for confounding by the matching factors. In fact, it can introduce confounding by the matching factors even when it did not exist in the source population. 1 The reasons for this are complex and will only be discussed briefly here. In essence, the matching process makes the controls more similar to the cases not only for the matching factor but also for the exposure itself. This introduces a bias that needs to be controlled in the analysis. For example, suppose we were conducting a case-control study of poverty and death (from any cause), and we chose siblings as controls (that is, for each person who died, we matched on family or residence by choosing a sibling who was still alive as a control). In this situation, since poverty runs in families we would tend to select a disadvantaged control for each disadvantaged person who had died and a wealthy control for each wealthy person who had died. We would find roughly equal percentages of disadvantaged people among the cases and controls, and we would find little association between poverty and mortality. The matching has introduced a bias, which fortunately (as we will illustrate) can be controlled by controlling for the matching factor in the analysis.

Thus, a matched design will (almost always) require controlling for the matching factors in the analysis. However, this does not necessarily mean that a matched analysis is required or appropriate, and it will often be sufficient to control for the matching factors using simpler methods. Although this is well recognised in both recent 2 3 and historical 4 5 texts, other texts 6 7 8 9 do not discuss this issue and present the matched analysis as the only option for analysing matched case-control studies. In fact, the more standard analysis may not only be valid but may be much easier in practice, and yield better statistical precision.

In this paper I explore and illustrate these problems using a hypothetical pair matched case-control study.

Options for analysing case-control studies

Unmatched case-control studies are typically analysed using the Mantel-Haenszel method 10 or unconditional logistic regression. 4 The former involves the familiar method of producing a 2×2 (exposure-disease) stratum for each level of the confounder (eg, if there are five age groups and two sex groups, then there will be 10 2×2 tables, each showing the association between exposure and disease within a particular stratum), and then producing a summary (average) effect across the strata. The Mantel-Haenszel estimates are robust and not affected by small numbers in specific strata (provided that the overall numbers of exposed or non-exposed cases or controls are adequate), although it can be difficult or impossible to control for factors other than the matching factors if some strata involve small numbers (eg, just one case and one control). Furthermore, the Mantel-Haenszel approach works well when there are only a few confounder strata, but will experience problems of small numbers (eg, strata with only cases and no controls) if there are too many confounders to adjust for. In this situation, logistic regression may be preferred, since this uses maximum likelihood methods, which enable the adjustment (given certain assumptions) of more confounders.

Suppose that for each case we have chosen a control who is in the same five year age group (eg, if the case is aged 47 years, then a control is chosen who is aged 45-49 years). We can then perform a standard analysis, which adjusts for the matching factor (age group) by grouping all cases and controls into five year age groups and using unconditional logistic regression 4 (or the Mantel-Haenszel method 10 ); if there are eight age groups then this analysis will just have eight strata (represented by seven age group dummy variables), each with multiple cases and controls. Alternatively we can perform a matched analysis (that is, retaining the pair matching of one control for each case) using conditional logistic regression (or the matched data methods, which are equivalent to the Mantel-Haenszel method); if there are 100 case-control pairs, this analysis will then have 100 strata.

The main reason for using conditional (rather than unconditional) logistic regression is that when the analysis strata are very small (eg, with just one case and one control for each stratum), problems of sparse data will occur with unconditional methods. 11 For example, if there are 100 strata, this requires 99 dummy variables to represent them, even though there are only 200 study participants. In this extreme situation, unconditional logistic regression is biased and produces an odds ratio estimate that is the square of the conditional (true) estimate of the odds ratio. 5 12

Example of age matching

Table 1 ⇓ gives an example of age matching in a population based case-control study, and shows the “true’ findings for the total population, the findings for the corresponding unmatched case-control study, and the findings for an age matched case-control study using the standard analysis. Table 2 ⇓ presents the findings for the same age matched case-control study using the matched analysis. All analyses were performed using the Mantel-Haenszel method, but this yields similar results to the corresponding (unconditional or conditional) logistic regression analyses.

Hypothetical study population and case-control study with unmatched and matched standard analyses

• View inline

Hypothetical matched case-control study with matched analysis

Table 1 ⇑ shows that the crude odds ratio in the total population is 0.86 (0.70 to 1.05), but this changes to 2.00 (1.59 to 2.51) when the analysis is adjusted for age (using the Mantel-Haenszel method). This occurs because there is strong confounding by age—the cases are mostly old, and old people have a lower exposure than young people. Overall, there are 390 cases, and when 390 controls are selected at random from the non-cases in the total population (which is half exposed and half not exposed), this yields the same crude (0.86) and adjusted (2.00) odds ratios, but with wider confidence intervals, reflecting the smaller numbers of non-cases (controls) in the case-control study.

Why matching factors need to be controlled in the analysis

Now suppose that we reconduct the case-control study, matching for age, using two very broad age groups: old and young (table 1 ⇑ ). The number of cases and controls in each age group are now equal. However, the crude odds ratio (1.68, 1.25 to 2.24) is different from both the crude (0.86) and the adjusted (2.00) odds ratios in the total population. In contrast, the adjusted odds ratio (2.00) is the same as that in the total population and in the unmatched case-control study (both of these adjusted odds ratios were estimated using the standard approach). Thus, matching has not removed age confounding and it is still necessary to control for age (this occurs because the matching process in a case-control study changes the association between the matching factor and the outcome and can create an association even if there were none before the matching was conducted). However, there is a small increase in precision in the matched case-control study compared with the unmatched case-control studies (95% confidence intervals of 1.42 to 2.81 compared with 1.38 to 2.89) because there are now equal numbers of cases and controls in each age group (table 1 ⇑ ).

A pair matched study does not necessarily require a pair matched analysis

However, control for simple matching factors such as age does not require a pair matched analysis. Table 2 ⇑ gives the findings that would have been obtained from a pair matched analysis (this is created by assuming that in each age group, and for each case, the control was selected at random from all non-cases in the same age group). The standard adjusted (Mantel-Haenszel) analysis (table 1 ⇑ ) yields an odds ratio of 2.00 (95% confidence interval 1.42 to 2.81); the matched analysis (table 2 ⇑ ) yields the same odds ratio (2.00) but with a slightly wider confidence interval (1.40 to 2.89).

Advantages of the standard analysis

So for many matched case-control studies, we have a choice of doing a standard analysis or a matched analysis. In this situation, there are several possible advantages of using the standard approach.

The standard analysis can actually yield slightly better statistical precision. 13 This may apply, for example, if two or more cases and their matched controls all have identical values for their matching factors; then combining them into a single stratum produces an estimator with lower variance and no less validity 14 (as indicated by the slightly narrower confidence interval for the standard adjusted analysis (table 1 ⇑ ) compared with the pair matched analysis (table 2 ⇑ ). This particularly occurs because combining strata with identical values for the matching factors (eg, if two case-control pairs all concern women aged 55-59 years) may mean that fewer data are discarded (that is, do not contribute to the analysis) because of strata where the case and control have the same exposure status. Further gains in precision may be obtained if combining strata means that cases with no corresponding control (or controls without a corresponding case) can be included in the analysis. When such strata are combined, a conditional analysis may still be required if the resulting strata are still “small,” 13 but an unconditional analysis will be valid and yield similar findings if the resulting strata are sufficiently large. This may often be the case when matching has only been performed on standard factors such as sex and age group.

The standard analysis may also enhance the clarity of the presentation, particularly when analysing subgroups of cases and controls selected for variables on which they were not matched, since it involves standard 2×2 tables for each subgroup. 15

A further advantage of the standard analysis is that it makes it easier to combine different datasets that have involved matching on different factors (eg, if some have matched for age, some for age and sex, and some for nothing, then all can be combined in an analysis adjusting for age, sex, and study centre). In contrast, one multicentre study 16 (of which I happened to be a coauthor) attempted to (unnecessarily) perform a matched analysis across centres. Because not all centres had used pair matching, this involved retrospective pair matching in those centres that had not matched as part of the study design. This resulted in the unnecessary discarding of the unmatched controls, thus resulting in a likely loss of precision.

Conclusions

If matching is carried out on a particular factor such as age in a case-control study, then controlling for it in the analysis must be considered. This control should involve just as much precision as was used in the original matching 14 (eg, if exact age in years was used in the matching, then exact age in years should be controlled for in the analysis), although in practice such rigorous precision may not always be required (eg, five year age groups may suffice to control confounding by age, even if age matching was done more precisely than this). In some circumstances, this control may make no difference to the main exposure effect estimate—eg, if the matching factor is unrelated to exposure. However, if there is an association between the matching factor and the exposure, then matching will introduce confounding that needs to be controlled for in the analysis.

So when is a pair matched analysis required? The answer is, when the matching was genuinely at (or close to) the individual level. For example, if siblings have been chosen as controls, then each stratum would have just one case and the sibling control; in this situation, an unconditional logistic regression analysis would suffer from problems of sparse data, and conditional logistic regression would be required. Similar situations might arise if controls were neighbours or from the same general practice (if each general practice only had one or a few cases), or if matching was performed on many factors simultaneously so that most strata (in the standard analysis) had just one case and one control.

Provided, however, that there are no problems of sparse data, such control for the matching factors can be obtained using an unconditional analysis, with no loss of validity and a possible increase in precision.

Thus, a matched design will (nearly always) require controlling for the matching factors in the analysis. It is not the case, however, that a matched design requires a matched analysis.

I thank Simon Cousens, Deborah Lawlor, Lorenzo Richiardi, and Jan Vandenbroucke for their comments on the draft manuscript. The Centre for Global NCDs is supported by the Wellcome Trust Institutional Strategic Support Fund, 097834/Z/11/B.

Competing interests: I have read and understood the BMJ policy on declaration of interests and declare the following: none.

Provenance and peer review: Not commissioned; externally peer reviewed.

This is an Open Access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 3.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/3.0/ .

• ↵ Rothman KJ, Greenland S, Lash TL, eds Design strategies to improve study accuracy. Modern epidemiology. 3rd ed . Lippincott Williams & Wilkins,  2008 .
• ↵ Rothman KJ. Epidemiology: an introduction. Oxford University Press,  2012 .
• ↵ Rothman KJ, Greenland S, Lash TL, eds. Modern epidemiology. 3rd ed . Lippincott Williams & Wilkins,  2008 .
• ↵ Breslow NE, Day NE. Statistical methods in cancer research. Vol I: the analysis of case-control studies. IARC,  1980 .
• ↵ Kleinbaum DG, Kupper LL, Morgenstern H. Epidemiologic research: principles and quantitative methods. Lifetime Learning Publications,  1982 .
• ↵ Dos Santos Silva I. Cancer epidemiology: principles and methods. IARC,  1999 .
• ↵ Keogh RH, Cox DR. Case-control studies. Cambridge University Press,  2014 doi:10.1017/CBO9781139094757 . .
• ↵ Lilienfeld DE, Stolley PD. Foundations of epidemiology. 3rd ed . Oxford University Press,  1994 .
• ↵ MacMahon B, Trichopolous D. Epidemiology: principles and methods. 2nd ed . Little Brown,  1996 .
• ↵ Mantel N, Haenszel W. Statistical aspects of the analysis of data from retrospective studies of disease. J Natl Cancer Inst  1959 ; 22 :719- 48 . 13655060 .
• ↵ Robins J, Greenland S, Breslow NE. A general estimator for the variance of the Mantel-Haenszel odds ratio. Am J Epidemiol  1986 ; 124 :719- 23 . 3766505 .
• ↵ Pike MC, Hill AP, Smith PG. Bias and efficiency in logistic analyses of stratified case-control studies. Int J Epidemiol  1980 ; 9 :89- 95 . doi:10.1093/ije/9.1.89 .  7419334 .
• ↵ Brookmeyer R, Liang KY, Linet M. Matched case-control designs and overmatched analyses. Am J Epidemiol  1986 ; 124 :693- 701 . 3752063 .
• ↵ Greenland S. Applications of stratified analysis methods. In: Rothman KJ, Greenland S, Lash TL, eds. Modern epidemiology. 3rd ed . Lippincott Williams & Wilkins,  2008 .
• ↵ Vandenbroucke JP, Koster T, Briët E, Reitsma PH, Bertina RM, Rosendaal FR. Increased risk of venous thrombosis in oral-contraceptive users who are carriers of factor V Leiden mutation. Lancet  1994 ; 344 :1453- 7 . doi:10.1016/S0140-6736(94)90286-0 .  7968118 .
• ↵ Cardis E, Richardson L, Deltour I, et al. The INTERPHONE study: design, epidemiological methods, and description of the study population. Eur J Epidemiol  2007 ; 22 :647- 64 . doi:10.1007/s10654-007-9152-z .  17636416 .
• Mansournia MA, Hernán MA, Greenland S. Matched designs and causal diagrams. Int J Epidemiol  2013 ; 42 :860- 9 . doi:10.1093/ije/dyt083 .  23918854 .
• Hernán MA, Hernández-Díaz S, Robins JM. A structural approach to selection bias. Epidemiology  2004 ; 15 :615- 25 . doi:10.1097/01.ede.0000135174.63482.43 .  15308962 .

EP717 Module 5 - Epidemiologic Study Designs – Part 2:

Case-control studies.

•   Page:
•   1  
• |   2  
• |   3  
• |   4  
• |   5  
• |   6  
• |   7  

Analysis of Case-Control Studies

Which study design is best.

As with cohort studies and clinical trials one of the first steps in the analysis of a case-control study is to generate simple descriptive statistics on each of the groups being compared, i.e., the case and the controls. This helps characterize the study population, and it also alerts you and your readers to any differences between the groups with respect to other exposures that might cause confounding.

After generating the descriptive statistics for a case-control study, the next step is to organize the data using contingency tables and to calculate estimates for the odds ratio. There may be confounding factors that distort the odds ratio, but one still begins by generating crude measures of association, i.e., estimates that have not yet been adjusted for confounding factors. In a later module you will learn how to use R to adjust for confounding in a case-control study.

Selection of a study design depends on the scientific questions being addressed and should take into account ethics and feasibility. For example, randomized clinical trials provide the best opportunity to identify small but potentially important clinical associations, but it would not be ethical to address all questions with a randomized clinical trial (e.g., whether maternal smoking during pregnancy is associated with a greater risk of having a premature or low birth weight infant). Observational studies (cohort studies and case-control studies) avoid many ethical problems, because potentially harmful exposures are not being allocated by the investigators, but they frequently present potential problems with regard to confounding and bias. Clinical trials and prospective cohort studies often require large numbers of subjects and long periods of follow-up that make them too costly to perform. Retrospective cohort studies and case-control studies are best to study outcomes with long latency periods, but getting accurate exposure data may be difficult. Case-control studies are particularly useful when studying rare outcomes, dynamic populations, and in situations in which exposure information is costly or difficult to obtain.

return to top | previous page

Content ©2021. All Rights Reserved. Date last modified: April 21, 2021. Wayne W. LaMorte, MD, PhD, MPH

methods@manchester

Case-control Study

Tarani Chandola, CCSR

A case-control study is a type of observational study design that is often used in epidemiology. Two groups of people are compared; one with the condition/disease (‘cases’) and a similar group of people who do not have the condition or disease (‘controls’).

The proportion of each group having a history of a particular exposure or characteristic of interest is then compared. If there is a greater proportion of cases who are exposed to a particular factor compared to controls, there is an association between the exposure and the disease.

Case control studies are usually cheaper and easier to do than longitudinal and experimental study designs but they suffer from a number of biases including recall bias in a person’s recollection of their history of exposure to the factor of interest. 

Figure: Schematic diagram of case-control study design. Kenneth F. Schulz and David A. Grimes (2002)  Case-control studies: research in reverse . The Lancet Volume 359, Issue 9304, 431 - 434

A nested case control study utilises data from a longitudinal cohort study to select a subset of matched controls to compare with the incident cases. In a case-cohort study, all incident cases in the cohort are compared to a random subset of participants who do not develop the disease of interest. In contrast, in a nested-case-control study, some number of controls are selected for each case from that case's matched risk set.

By matching on factors such as age and selecting controls from relevant risk sets, the nested case control model is generally more efficient than a case-cohort design with the same number of selected controls. This is similar to propensity score matching techniques.

Experts/users at Manchester

• Dr Roger Webb
• Professor Jane Worthington

Key references

• Doll R, Hill AB (1952) A study of the aetiology of carcinoma of the lung. British Medical Journal 2:1271–1286
• Rothman K (2002) Epidemiology. An Introduction. Oxford University Press, Oxford, England
• Schulz KF, Grimes DA (2002) Case-control studies: research in reverse;. Lancet: 359: 431–34

Download PDF slides of the presentation ' What is a Case-Control Study? '

Study Design 101

• Helpful formulas
• Finding specific study types
• Case Control Study
• Meta- Analysis
• Systematic Review
• Practice Guideline
• Randomized Controlled Trial
• Cohort Study
• Case Reports

A study that compares patients who have a disease or outcome of interest (cases) with patients who do not have the disease or outcome (controls), and looks back retrospectively to compare how frequently the exposure to a risk factor is present in each group to determine the relationship between the risk factor and the disease.

Case control studies are observational because no intervention is attempted and no attempt is made to alter the course of the disease. The goal is to retrospectively determine the exposure to the risk factor of interest from each of the two groups of individuals: cases and controls. These studies are designed to estimate odds.

Case control studies are also known as "retrospective studies" and "case-referent studies."

• Good for studying rare conditions or diseases
• Less time needed to conduct the study because the condition or disease has already occurred
• Lets you simultaneously look at multiple risk factors
• Useful as initial studies to establish an association
• Can answer questions that could not be answered through other study designs

Disadvantages

• Retrospective studies have more problems with data quality because they rely on memory and people with a condition will be more motivated to recall risk factors (also called recall bias).
• Not good for evaluating diagnostic tests because it’s already clear that the cases have the condition and the controls do not
• It can be difficult to find a suitable control group

Design pitfalls to look out for

Care should be taken to avoid confounding, which arises when an exposure and an outcome are both strongly associated with a third variable. Controls should be subjects who might have been cases in the study but are selected independent of the exposure. Cases and controls should also not be "over-matched."

Is the control group appropriate for the population? Does the study use matching or pairing appropriately to avoid the effects of a confounding variable? Does it use appropriate inclusion and exclusion criteria?

Fictitious Example

There is a suspicion that zinc oxide, the white non-absorbent sunscreen traditionally worn by lifeguards is more effective at preventing sunburns that lead to skin cancer than absorbent sunscreen lotions. A case-control study was conducted to investigate if exposure to zinc oxide is a more effective skin cancer prevention measure. The study involved comparing a group of former lifeguards that had developed cancer on their cheeks and noses (cases) to a group of lifeguards without this type of cancer (controls) and assess their prior exposure to zinc oxide or absorbent sunscreen lotions.

This study would be retrospective in that the former lifeguards would be asked to recall which type of sunscreen they used on their face and approximately how often. This could be either a matched or unmatched study, but efforts would need to be made to ensure that the former lifeguards are of the same average age, and lifeguarded for a similar number of seasons and amount of time per season.

Real-life Examples

Boubekri, M., Cheung, I., Reid, K., Wang, C., & Zee, P. (2014). Impact of windows and daylight exposure on overall health and sleep quality of office workers: a case-control pilot study . Journal of Clinical Sleep Medicine : JCSM : Official Publication of the American Academy of Sleep Medicine, 10 (6), 603-611. https://doi.org/10.5664/jcsm.3780

This pilot study explored the impact of exposure to daylight on the health of office workers (measuring well-being and sleep quality subjectively, and light exposure, activity level and sleep-wake patterns via actigraphy). Individuals with windows in their workplaces had more light exposure, longer sleep duration, and more physical activity. They also reported a better scores in the areas of vitality and role limitations due to physical problems, better sleep quality and less sleep disturbances.

Togha, M., Razeghi Jahromi, S., Ghorbani, Z., Martami, F., & Seifishahpar, M. (2018). Serum Vitamin D Status in a Group of Migraine Patients Compared With Healthy Controls: A Case-Control Study . Headache, 58 (10), 1530-1540. https://doi.org/10.1111/head.13423

This case-control study compared serum vitamin D levels in individuals who experience migraine headaches with their matched controls. Studied over a period of thirty days, individuals with higher levels of serum Vitamin D was associated with lower odds of migraine headache.

Related Formulas

• Odds ratio in an unmatched study
• Odds ratio in a matched study

Related Terms

A patient with the disease or outcome of interest.

Confounding

When an exposure and an outcome are both strongly associated with a third variable.

A patient who does not have the disease or outcome.

Matched Design

Each case is matched individually with a control according to certain characteristics such as age and gender. It is important to remember that the concordant pairs (pairs in which the case and control are either both exposed or both not exposed) tell us nothing about the risk of exposure separately for cases or controls.

Observed Assignment

The method of assignment of individuals to study and control groups in observational studies when the investigator does not intervene to perform the assignment.

Unmatched Design

The controls are a sample from a suitable non-affected population.

Now test yourself!

1. Case Control Studies are prospective in that they follow the cases and controls over time and observe what occurs.

a) True b) False

2. Which of the following is an advantage of Case Control Studies?

a) They can simultaneously look at multiple risk factors. b) They are useful to initially establish an association between a risk factor and a disease or outcome. c) They take less time to complete because the condition or disease has already occurred. d) b and c only e) a, b, and c

← Previous Next →

© 2011-2019, The Himmelfarb Health Sciences Library Questions? Ask us .

• Himmelfarb Intranet
• Privacy Notice
• Terms of Use
• GW is committed to digital accessibility. If you experience a barrier that affects your ability to access content on this page, let us know via the Accessibility Feedback Form .