Case-control study

from Wikipedia, the free encyclopedia

A case-control study is a form of epidemiological study in medicine . People with a certain disease are examined and compared with suitable control groups. Looking back, a cause is sought for the disease. It is a retrospective study of a sample made up of sick people (case) and a sample made up of healthy people (control). For both groups it is now determined whether there has been any past exposure to potential risk factors. A significant difference between the two groups means a correlation between risk factor and disease. Under no circumstances can one infer a cause / effect relationship. A retrospective study arrangement such as the case-control study is particularly suitable for uncovering the causes of rare diseases.


Classification of clinical studies
 
 
Intervention study
 
 
 
 
 
Observational study
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
comparative
groups
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Randomized
controlled study
 
non-randomized
controlled
study
 
Descriptive
study
 
Analytical
study
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
Cohort
study
 
Case-control
study
 
Cross-sectional
study

Selection of cases

For a case-control study, cases are selected based on disease, not exposure. There are various sources to choose from, such as hospital patients or patients from the residential area. In order to prevent the risk factors discovered from being attributable to a specific hospital, it is advisable to select cases from different hospitals or federal states. When selecting the cases, it must also be determined whether new (incidental) or existing (prevalent) cases are to be included. In the case of case-control studies, it is recommended to include incidence cases, because with prevalence cases there is a risk that the detected risk factor indicates a connection with survival and not with the cause of the disease. If, for example, the majority of the sick should die shortly after the diagnosis, these patients will be underrepresented in the examination with prevalent cases and ultimately distort the result. In the case of incidental cases, however, it should be noted that the period in which a new disease occurs may be longer than in prevalent cases.

Selection of controls

The selection and comparability of the controls are important criteria for ensuring the validity of the case-control study. So that comparable exposure assumptions can be made, the cases and controls should belong to the same baseline population. The best way to ensure comparability is to select controls using the population approach. The controls are taken from a random sample from the same population as the cases. The selection can be made, for example, by randomly selected telephone numbers from the same region. Cases and controls from a cohort are also easily comparable . In that case the study is called a nested case-control study . Overall, cases and controls should meet the same inclusion criteria, such as region or age.

Hospital controls and controls from the hospital catchment area are another way of generating controls. When making a selection, care must be taken to ensure that the controls do not have any diseases that are associated with the exposure or the disease of interest. For example, people with a certain type of cancer represent the cases, and the controls are recruited from patients who had to visit the clinic because of traffic accidents. However, this type of control is more susceptible to interference than selection from the population.

In order to prevent the cases and controls from showing differences in properties and exposure that are not actually the focus of the study, the participants are linked with regard to characteristics such as age, gender, cultural background, socio-economic status and occupation (technical term matching ). To make the study more robust, each case is often matched with several, sometimes even eight, controls.

Assessment of the risk factor with blinding

An attempt should be made to blind the examiners who assess the risk factor retrospectively . This means that when determining the risk factor, the investigators should not know whether the test person is one of the cases or one of the controls in order to avoid unconscious rounding errors in the hypothesis-compliant direction. If the risk factor has to be ascertained in an interview, it is sometimes unavoidable that the examiner learns the assignment. For better traceability, it can be useful to include whether the interviewer or data carrier believes that it is a case or a control person.

Difficulties in the case-control studies

The main difficulty of the case-control study itself is the correct selection of the controls - because the overall validity of the study depends on it. If one or more controls of the same age from the same sex and the same place of residence are selected for each case, everything that is related to sex, age or place of residence is excluded from the outset as a risk factor. However, as mentioned earlier, cases and controls must come from the same baseline population.

The main difficulty in collecting data is the limits of the participants' memory (recall). Because in order to obtain information about the exposures, the persons are asked retrospectively. The most common form is called recall bias . The persons may have forgotten a possible exposure because the persons concerned often deal more intensively with possible causes of the disease, which for the controls do not reveal any connection or have been forgotten. Thus, the different ability to remember between the cases and controls can lead to possible misinterpretations.

To avoid this, the case-control study can be expanded to include risk factors that have already been documented in writing. Examples of this are place of residence and address if the suspected risk factor is related to pollutants in the air or drinking water, or medical records requested by clinics if the suspected risk factor is related to a previous illness or treatment.

rating

The main advantage of case-control studies is that they provide results quickly and can be used for rare diseases, while cohort studies often take years to produce meaningful results. The main disadvantage is that the researchers cannot strictly define the risk factor (for example by assigning patients to placebo and active ingredient, or to two different treatment methods). A case-control study is therefore of less informative value than a cohort study.

In addition, case-control studies are relatively inexpensive and can also be carried out by small groups or individual researchers in individual research institutions, which would often not be possible for cohort studies. They have shown the way to several important discoveries and advances, but it is precisely their success that has resulted in their being too trusted and damaging their credibility. This is in large part due to erroneous assumptions about such studies.

Examples of case-control studies

This method met with great success in 1951 when a study initiated by Sir Richard Doll found a relationship between tobacco smoking and lung cancer. Skeptics, mostly with the support of the tobacco industry, had argued - and correctly - for many years that this type of study could not establish a conclusive relationship between cause and effect ( causality ), but in the end the results of the British Doctors Study carried out confirmed it , a cohort study, impressively presented the results of the case-control study.

Another important case-control study was published in 1971. Daughters suffered from a certain, otherwise very rare, vaginal cancer more often if their mothers took diethylstilbestrol during pregnancy , a remedy that was initially approved for vaginitis , breastfeeding problems and menopausal complaints , and was then also prescribed for pregnancy problems . The only eight cases were born between 1946 and 1951, and fell ill between 1966 and 1969. Research into medical records and interviews with patients and their relatives revealed that the mother used the above-mentioned drug during pregnancy.

Individual evidence

  1. ^ Gaus Wilhelm, Muche Rainer: Medical statistics: Applied biometrics for doctors and health professions . Schattauer, 2017, ISBN 978-3-7945-3241-4 , pp. 47 ( limited preview in Google Book search).
  2. Leon Gordis: Epidemiology. Fourth edition. Sauders Elsevier, Philadelphia 2009; Robert H. Fletcher, Suzanne W. Fletcher. Clinical epidemiology. Basics and application. 2nd Edition. Verlag Hans Huber, Bern 2007, pp. 180-185
  3. Robert H. Fletcher, Suzanne W. Fletcher. Clinical epidemiology. Basics and application. 2nd Edition. Verlag Hans Huber, Bern 2007, pp. 134-136
  4. Leon Gordis: Epidemiology. Fourth edition. Sauders Elsevier, Philadelphia 2009, pp. 180-185; Robert H. Fletcher, Suzanne W. Fletcher. Clinical epidemiology. Basics and application. 2nd Edition. Verlag Hans Huber, Bern 2007, pp. 134-136
  5. Leon Gordis: Epidemiology. Fourth edition. Sauders Elsevier, Philadelphia 2009, pp. 185-186; Robert H. Fletcher, Suzanne W. Fletcher. Clinical epidemiology. Basics and application. 2nd Edition. Verlag Hans Huber, Bern 2007, pp. 135-137
  6. Marcus Müllner: Successful scientific work in the clinic: Evidence Based Medicine . Springer-Verlag, 2013, ISBN 978-3-7091-3755-0 , pp. 60 ( limited preview in Google Book search).
  7. Leon Gordis: Epidemiology. Fourth edition. Sauders Elsevier, Philadelphia 2009, p. 190
  8. ^ Daff, ME, Doll, R., & Kennaway, EL (1951). Cancer of the Lung in Relation to Tobacco. British Journal of Cancer, 5 (1), 1-20.
  9. Herbst, AL, Ulfelder, H., & Poskanzer, DC (1971). Adenocarcinoma of the vagina: Association of maternal stilbestrol therapy with tumor appearance in young women. New England journal of medicine, 284 (16), 878-881.

literature

  • Ruth Bonita, Robert Beaglehole, Tord Kjellström: Introduction to Epidemiology. 2nd Edition. Hans Huber Verlag, Bern 2008, ISBN 978-3-456-84535-7
  • Robert H. Fletcher, Suzanne W. Fletcher. Clinical epidemiology. Basics and application. 2nd Edition. Publisher Hans Huber, Bern 2007
  • Leon Gordis: Epidemiology. Fourth edition. Sauders Elsevier, Philadelphia 2009
  • Oliver Razum, Jürgen Breckenkamp, ​​Patrick Brzoska: Epidemiology for Dummies. WILEY-VCH Verlag, Munich 2009