The effect of misclassification error on risk estimation in case-control studies

Abstract

The effect of the sensitivity and specificity in classifying exposure, the prevalence of exposure in controls and true OR differed between positive and negative biases. The use of valid exposure classification instruments with high sensitivity and high specificity is recommended to mitigate this type of bias.

Association between opioid‐related deaths and persistent opioid prescribing in primary care in England: A nested case‐control study

“…Therefore, prescribers should cautiously consider the benefit and risk when continuing opioid prescription for periods beyond 1 year 14 . In addition, the MDA could potentially cause misclassification when POU occurred during a period which was split into two patient‐years and hence the proportion of patients who had ever been prescribed opioids persistently and the effect of POU on opioid‐related death could be underestimated 52,53 . Further studies should explore the various patterns of persistent opioid prescribing in general practice in the UK.…”

“…POU was not associated with opioid‐related death in the sensitivity analysis when only considering opioid prescribing in the first patient‐year. This may be attributed to a nondifferential misclassification error 52,53 because nonpersistent opioid users in the first retrospective patient‐year could have been prescribed opioids persistently in the second or the third retrospective patient‐years. Therefore, prescribers should cautiously consider the benefit and risk when continuing opioid prescription for periods beyond 1 year 14 .…”

Concordance of self-reported hormonal contraceptive use and presence of exogenous hormones in serum among African women

“…If misreporting of hormonal contraceptives is the same among women who acquire HIV and those who do not, then the reported increases in risk among DMPA users would be underestimated. [1] On the other hand, if misreporting differs between women who acquire HIV and those who do not (known as differential misclassification), results may be overestimated or underestimated [30,31]. This supports the need for well-done clinical trials to answer the question of whether hormonal contraception increases the risk of HIV acquisition.…”

Feasibility and validity of using healthcare databases to conduct cross‐national comparative studies of opioid use, its determinants and consequences

“…There are concerns based on our survey regarding hypothesis testing in individual patient databases. Firstly, as most individual patient databases include prescriptions from only primary or secondary care and cannot be linked to each other after de‐identification, opioid utilization calculated from most of those individual‐patient level databases could be underestimated and lead to misclassification errors 32,33 . Further, as most individual patient databases did not include patients at a national population level, it is crucial to ensure that the databases cover a representative population and that the results from those databases could be generalizable to the whole country.…”