Moreover, if FDPR is poorly implemented, this could explain the absence of an effect of FPDR on patient outcomes. Fourth, hospital sites also vary in how they abstract data, and some variables—specifically the facility-reported potential resuscitation systems errors—have a large degree of missing data which challenge the interpretation of inference for these variables. Fifth, the registry is unable to capture whether there was a temporal change in a hospital’s policy for FPDR over time. Although a hospital may have added a policy in the data collection, it is unlikely that a hospital would have eliminated a pre-existing policy for FPDR. Finally, our analysis relates to an FPDR policy rather than the actual presence of family members during resuscitation care. As a consequence, we are unable to directly measure the effect of a family at the bedside on an individual resuscitation itself. This makes our study susceptible to the ecological fallacy,38 and our data should not be interpreted to mean that a lack of a correlation at the hospital-level would hold true for individual patients