Case-Control Study

872 Words2 Pages

The study’s methods may be deemed internally valid. Case-control studies are especially vulnerable to selection bias during the selection of cases and controls. This study selected all participants from a population-based surveillance system. Since the presence or absence of NVP, the exposure, had no influence on the registration of NTD cases and matched controls, selection bias due to faulty sampling procedures was minimized. Likewise, the high participation rates (>80%) and low levels of missing information (<3%) reduced selection bias due to attrition or other participation-related factors. However, two potential concerns should be noted. First, to increase the sample size, the authors used an unmatched analysis. The cases and controls were …show more content…

NVP was measured using face-to-face interviews, a method prone to recall bias. Since mothers from the case group would be more likely to recall adverse events that could be linked to NTDs in their children, the method would be more sensitive in cases than the population controls. This would lead to an overestimation of the risk. The study tries to minimize this bias by informing the participants about what NVP means (including that it necessitated bed rest or hospitalization) and giving a certain time frame to recall from (in the first trimester). Also, a limb defect control group, who too would be more likely to remember adverse exposures, was added to detect the presence of recall bias. The value of malformed controls in similar studies is debatable [3]. Nonetheless, the association between NVP and NTD when compared to this group achieved statistical significance, which strengthened the finding that an association existed, even if recall bias was present. Case-control studies are less prone to misclassification of the outcome (e.g. due to diagnostic suspicion bias) because the exposure status is known during the initial classification. Also, studies validate the two-stage process used by the surveillance system to classify NTD cases, as it has a sensitivity of 95.9% and assumed specificity near 100% …show more content…

Since most findings were statistically significant, the chance of a type 1 error was possible. In a few cases, such as during the analysis of the association between NVP and Encephalocele relative to limb defects control, the null hypothesis of no association was accepted. Here, a type 2 error may have been possible. This could be prevented by using larger samples sizes for the limb control group (123 limb defect controls) and NTD subcategory cases. However, for most purposes, the sample sizes (832 cases, 1271 population controls) used in the study were quite large, increasing the statistical power and reducing the effects of random error. This is reflected in the moderately sized confidence intervals, which further indicate that the results were moderately

More about Case-Control Study

Open Document