Matched case–control study with reporting bias

• Consider matched case–control study in the presence of selection bias.
• Exposure of interest can be both binary and continuous.
• Additional information from external source has been used.
• Simulation study indicates both unbiasedness and efficiency.
• Application to the analysis of spontaneous adverse drug reaction reports is considered.

Matching is widely used to improve the efficiency in case–control studies. To achieve this, it is required that both case and control samples are representative of respective part in the source population. Often this characteristic is violated and we fail to draw valid inference about the parameter of interest. In this work, we consider matched case–control study subject to such bias known as selection bias. This kind of bias arises, for example, in the analysis of spontaneous adverse drug reaction (ADR) reports, where the event of ADR, or its complement, is reported with some probability depending on the use of drug. Here it is known as reporting bias. For unbiased analysis of such data, we consider a reference sample from the source population. We also discuss the asymptotic properties of the estimates of the model parameters. An example is considered for the purpose of illustration and some simulation studies are carried out to investigate the properties of the estimators.