To the Editor: In the study on the association between medical care use and personal health records (PHRs), Dr Palen and colleagues1 found increased use of medical care among those with PHRs—more visits, telephone calls, hospitalizations, and emergency and after-hour visits. This contrasted with expectations and much of the previous literature.
We, however, were not surprised by the conflicting findings.2 Although Palen et al attempted to statistically balance the samples, research involving patients' voluntary actions may not be amenable to post hoc adjustments, no matter how sophisticated. This is a classic example of common and powerful biases in the medical literature (eg, volunteer selection bias) making it difficult to differentiate between patient characteristics and intervention effects.3,4 Patients without interest in PHRs may not be comparable with volunteer technophiles or with the worried patients who use more medical care.