Matching In Case-control Studies Research Articles

A flexible matching strategy for matched nested case-control studies

PurposeIndividual matching in case-control studies improves statistical efficiency over random selection of controls but can lead to selection bias if cases are excluded due to the lack of appropriate controls or residual confounding with less strict matching criteria. We introduce flex matching, an algorithm using multiple rounds of control selection with successively relaxed matching criteria to select controls for cases. MethodsWe simulated exposure-disease relationships in multiple cohort data sets with a range of confounding scenarios and conducted 16,800,000 nested case-control studies, comparing random selection of controls, strict matching, and flex matching. We computed average bias and statistical efficiency in estimates of exposure-disease relationships under each matching strategy. ResultsOn average, flex matching produced the least biased estimates of exposure-disease associations with the smallest standard errors. Strict matching algorithms that excluded cases for whom matched controls could not be identified produced biased estimates with larger standard errors. Estimates from studies with random assignment of controls were relatively unbiased, but the standard errors were larger than from studies using flex matching. ConclusionsFlex matching should be considered for case-control designs, especially for biomarker studies where matching on technical artifacts is necessary and maximizing efficiency is a priority.

Age and sex matching in case-control studies.

Age and sex matching in case-control studies.

Every Peddler Praises His Own Needle: Have Clinical Rules in the Diagnosis of Subarachnoid Hemorrhage Supplanted Lumbar Punctures Yet?

1The authors state, “Research conducted among ED [emergency department] patients with possible acute coronary syndrome suggests that patients often have much higher risk thresholds for themselves than do the treating physicians.”1 AAssuming both patients and physicians are making rational decisions, list some reasons why they may have different risk thresholds. Do you think that emergency physicians have higher or lower risk thresholds than other physicians (eg, internists, pediatricians, neurosurgeons)? Why might this be so? Should our health care system try to align patient and physician acceptable risk thresholds? If so, how might this be achieved?BAssume 1) the 97% sensitivity reported for the Mark et al decision rule is externally validated and 2) the incidence of subarachnoid hemorrhage in patients presenting to the ED with acute headache is 2% and an emergency physician treats 3,500 patients per year. On average, how many years would an emergency physician have to apply this clinical rule to miss 1 subarachnoid hemorrhage? Do these sound like reasonable numbers to you? Why or why not?CNoncontrast computed tomography (CT) of the brain and lumbar punctures have associated complications. Summarize the reported frequency of complications associated with CT and lumbar puncture. Using the same data listed in question 1B, calculate how often these complications might occur. Does it matter that lumbar puncture–associated complications typically occur acutely, whereas noncontrast CT radiation exposure is associated with late-occurring morbidity and mortality? How might patient age factor into the likelihood of complications associated with CT and lumbar puncture?DIf this clinical rule were widely adopted, in 5 years would you expect that more or fewer patients would be investigated for subarachnoid hemorrhage (the use of similar decision aids in pulmonary embolism may serve as a useful example)?2The authors performed a matched control study. AUnder what circumstances are case-control studies ideal? Consider the time course of exposure and disease, the nature of the exposure and outcomes, and the frequency of the disease.BDiscuss the importance of matching in case-control studies. Why does one match? What should one match on? What should one not match on?CWhy did the authors choose 3 controls for each case? Why not 1? Why not 6?DWhat are the biases of case-control studies in general? In this study specifically?3The authors state, “Approximately 80% of subarachnoid hemorrhage cases are due to ruptured cerebral aneurysms.”1 AWhat are other common causes of nontraumatic subarachnoid hemorrhage?BIn general terms, describe the sensitivity of a noncontrast head CT in detecting these other causes. When choosing a screening test, is it preferable to be highly sensitive or highly specific? What test characteristic (ie, sensitivity, specificity) is desired for a confirmatory study?CAfter a patient complaint about a post–lumbar puncture headache, a hospital administrator proposes that the ED replace lumbar puncture with CT angiography when attempting to exclude subarachnoid hemorrhage in patients with a negative initial head CT result. What do you think about this, assuming the combination of CT and CT angiography is 99% sensitive for diagnosing a subarachnoid hemorrhage, as recently reported?24The authors performed a sensitivity analysis to assess the rule's stability, using a 1,000-iteration bootstrap analysis. AWhat is a sensitivity analysis? Why are sensitivity analyses often performed in observational studies?BWhat is a bootstrap analysis? Why are bootstrap analyses especially important when evaluating clinical decision rule performance? What assumptions are invoked by bootstrap analyses?

Role of Matching in Case-Control Studies of Antimicrobial Resistance

Of 57 case-control studies of antimicrobial resistance, matching was used in 23 (40%). Matched variables differed substantially across studies. Of these 23 matched case-control studies, 12 (52%) justified the use of matching, and 9 (39%) noted the strengths or limitations of this approach. Analysis that accounted for matching was performed in only 52% of the case-control studies.

Why match? Investigating matched case-control study designs with causal effect estimation.

Matched case-control study designs are commonly implemented in the field of public health. While matching is intended to eliminate confounding, the main potential benefit of matching in case-control studies is a gain in efficiency. Methods for analyzing matched case-control studies have focused on utilizing conditional logistic regression models that provide conditional and not causal estimates of the odds ratio. This article investigates the use of case-control weighted targeted maximum likelihood estimation to obtain marginal causal effects in matched case-control study designs. We compare the use of case-control weighted targeted maximum likelihood estimation in matched and unmatched designs in an effort to explore which design yields the most information about the marginal causal effect. The procedures require knowledge of certain prevalence probabilities and were previously described by van der Laan (2008). In many practical situations where a causal effect is the parameter of interest, researchers may be better served using an unmatched design.

Matching criteria in case–control studies on postoperative infections

Matching is commonly used in case-control studies to control for the effect of major confounding factors. We evaluated the available evidence from case-control studies regarding postoperative infections to identify how frequently matching was performed and with what specific variables. We searched for relevant case-control studies in PubMed until August 2006 and further evaluated those that used individual matching between cases and controls. We identified and evaluated 42 relevant studies. Age was used as a matching criterion in 27 of these 42 (64.3%) case-control studies. The specific type of surgical procedure was the second most commonly used criterion in 17 of 42 studies (40.5%). Gender was used in 14/42 studies (33.3%) as a matching criterion between case and control patients. The period at risk for development of surgical site and/or other postoperative infections, i.e. time from surgery to the diagnosis of infection, was used in nine of 42 studies (21.4%), as was date of operation, and the primary diagnosis that led the case and control patients to surgery. The same surgeon or surgical team was used in seven studies (16.7%); matching according to the National Nosocomial Infection Surveillance system risk score was performed in five studies (11.9%). The findings of our analysis suggest that various characteristics are used for matching in case-control studies of postoperative infections. A more consistent use of matching with the specific type of surgical procedure may help in increasing the internal validity of a case-control study in this field of clinical research.

The use and misuse of matching in case-control studies: the example of polycystic ovary syndrome

Matching control selection strategies are often used in polycystic ovary syndrome case-control studies; however, they are infrequently used in an appropriate fashion. When properly applied, matching may offer improved study precision, but this is highly contingent on the causal pathway under consideration, strength of the associations between the matching variable and both the risk factor of interest and polycystic ovary syndrome, and the use of an appropriate stratified data analysis.

Flexible matching in case-control studies of gene-environment interactions.

Because of the lack of power of case-control study designs to detect gene-environment interactions, flexible matching has recently been proposed as a method of improving efficiency. In this paper, the authors consider a large-sample approximation method that allows estimation of the most efficient matching strategy when genotype and exposure are either independent or associated. The authors provide tables of the sample sizes required to detect gene-environment interactions if this flexible matching strategy is followed, and they make brief comparisons with other study designs.

Matching in case-control studies

EDITOR,—J Martin Bland and Douglas G Altman's review of the advantages and disadvantages of matching in case-control studies omits two important problems.1 Firstly, matching in case-control studies ensures that the matching factors, such as age or sex, are equally distributed between cases and controls. …

Examining neighbourhood confounding in a survey: an example using the National Health and Nutrition Examination Survey II.

This paper demonstrates a methodology for examining whether neighbourhood stratification lessens the confounding bias of some specific risk-factor/disease associations in analysis of the National Health and Nutrition Examination Survey II (NHANES II). The fortuitous clustering of the sample design of NHANES allows us to estimate risk-factor/disease associations with and without controlling for neighbourhood effects on the same population. We briefly discuss the implications of this methodology for neighbourhood matching in case-control studies.

The efficiency of matching in case-control studies of risk-factor interactions

We have examined the relative efficiency of matching vs not matching on a binary exposure variable in a case-control study of its interaction with another binary exposure variable. Interaction is treated as a parameter measuring departure from multiplicativity. Efficiency is considered both in the presence and absence of other matching variables, the former case requiring use of matched-pair estimators. In most circumstances examined, matching on one variable led to a less efficient estimator of the interaction parameter. If both variables were related to disease, the only circumstance in which matching always improved efficiency was when the high-risk level of the matching variable was uncommon and not positively associated with the high-risk level of the other variable. Matching always appeared to harm efficiency when the high risk level of the matching variable was common. Other disadvantages of matching are that it precludes estimation of the main effect of the matching variable and fitting of non-multiplicative models, and increases the difficulty of control selection.

Estimating the utility of matching in case-control studies

The estimation of the utility of matching in completed case-control studies is examined. The utility is defined in terms of providing an estimate of a more valid odds ratio than the crude odds ratio, and in providing a more efficient study than an unmatched study with a stratified analysis. Methods are given to estimate the utility of frequency-matching in a completed study where the cases and controls are matched into broad strata such as age or decade of birth. An example from the literature demonstrates the technique. Since the proposed analysis requires knowledge of the proportions of individuals in the study population associated with each stratum, it is recommended that estimates of these proportions be given in the results of frequency-matched case-control studies. For pair-matching where each pair is regarded as defining its own stratum (such as with neighborhood or friend controls) it is shown to be impossible to estimate the utility of the matching from a completed study. In particular, it is incorrect for either frequency- or pair-matching to try to estimate the utility of matching by performing an unmatched analysis on a completed case-control study.

Considerations in determining matching criteria and stratum sizes for case-control studies.

Despite considerable research into the statistical properties of matching in case-control studies, much disagreement remains regarding practical guidelines for the employment of the procedure. We critically examine some recent statistical recommendations, and review a number of theoretical and practical considerations that should be employed in designing matched studies. We conclude that determination of matching criteria and stratum sizes should always involve practical as well as theoretical considerations, and that global recommendations regarding matching cannot be justified on the basis of statistical considerations alone.