Background:
The culture of academic medicine lends itself to compulsory laboratory testing and the overutilization of tests. Physicians in training constitute the primary cohort of testers, and a certain tolerance is afforded their ordering behaviors in the name of education. More than 50,000 magnesium and phosphorus assays are ordered by our inpatient teams every year. The “normal” rate for these studies is more than 90%. We chose to address the testing practices of trainees in order to contain costs and reduce the adverse effects of false‐positive results.
Purpose:
We conducted a low‐cost, low‐tech intervention to reduce laboratory testing by providers on our inpatient medicine unit. This is a summary of the results from the first phase of our intervention. We wished to prove that physicians will order fewer tests when exposed directly and indirectly to sequential interventions. In particular, we wished to understand how the test‐ordering behavior of physicians is affected by providing them charge feedback data. Our hypothesis is that this intervention will reduce the number of laboratory tests ordered on our academic general medicine unit.
Description:
We built daily QuadraMed Affinity® charge reports. These served as a daily census for providers and included the patient's name, diagnosis, LOS, cumulative lab charges, total cumulative IP charges, and insurance type. These reports were generated for each attending physician on the INPATIENT Medicine Teams. Prior to the start of a rotation, the attending physicians received a letter explaining that ihey would be given the census report daily and that the information should be used to generate a dialogue about responsible utilization of health care resources with their trainees. The number of magnesium and phosphorus tests per patient per day in the preintervention and postintervention periods were compared using a 2‐sample t test. The data was log‐transformed before analysis. In addition, regression analysis and time‐series modeling were used to adjust for case‐mix measures and potential weekday effects. The number of tests per patient per day in the medicine wards were divided by the number of tests per patient per day on the general surgery teams (our control group); the log‐transformed ratio was compared pre‐ and postintervention.
Conclusions:
There were 93.7% as many tests done postintervention as done preintervention, which is significant at the 5% level (P = 0.0076). There were 92.8% as many tests done on weekends compared with on weekdays, also significant (P < 0.001). Overall, we found that there was a 19% reduction in testing postintervention by the study group compared to preintervention, significant at the 5% level (P < 0.001). This suggests that physician test‐ordering behavior modification on our academic medical service can be achieved by providing real‐time charge feedback data. This low‐cost, low=tech intervention will continue indefinitely as we begin phase 2 of the study.
Author Disclosure:
M. Radzienda, none; A. Szabo, none; M. Conti, none.