Figure 2.

Randomization scheme. If we randomized by attending physician alone, then a primary care physician (PCP) of a patient in the usual care arm may also care for a patient in the intervention arm (ie, cared for by a different inpatient attending physician). Upon receiving notification of a test pending at discharge (TPAD) result of the intervention patient, the PCP could look up TPAD results of the usual care patient. Conversely, if we randomized by PCP alone, then an attending physician of a usual care patient may also care for a patient in the intervention arm (ie, cared for by a different PCP). Upon receiving notification of a TPAD result of the intervention patient, the attending physician could look up TPAD results of the usual care patient. Although the proposed schema reduced our sample size by 50%, it provided adequate statistical power to evaluate inpatient provider awareness while allowing for an unbiased analysis of PCP awareness.