Background: It is often assumed resident physicians at academic medical centers order more tests for inpatients due to different aspects of the clinical learning environment.  Despite this prevailing notion, there is very little evidence to support this claim. We sought to quantify differences in ordering practices between teaching hospitals and non-teaching hospitals for two common medical conditions – bacterial pneumonia and cellulitis.

Methods: We used the Texas Inpatient Public Use Data File, a billing level state-wide dataset, to analyze all inpatient encounters between January 2014-June 2015 with a principal diagnosis of bacterial pneumonia (N=31,948) or cellulitis (N=24,303), and calculated number of lab tests per hospital day by summing individual units billed for laboratory revenue codes and dividing by length of stay (LOS). Labs per day were adjusted individually by the illness severity of the encounter using DRG weight. We categorized hospitals with > 100 cases of each principal diagnosis into 3 types: major teaching (member of Council of Teaching Hospitals, N=14), minor teaching (N=13), and non-teaching (N=89) and compared labs per day across hospital types. We excluded patients who were <18 years of age, had an ICU stay during the hospitalization, were transferred from another hospital, or had a LOS >=2SD of the condition’s mean LOS.

We used one-way ANOVA tests to determine if severity-adjusted labs per day varied by hospital type, and multilevel quasi Poisson regression models with hospital as a random effect to determine if hospital type was associated with the raw number of labs in an encounter.

Results: Mean adjusted labs per day varied significantly by hospital type and was highest for major teaching hospitals for both pneumonia and cellulitis after adjusting for severity of illness (Figure 1). Variance in adjusted labs per day was highest for academic teaching hospitals.

Correlation analysis indicated strong association of mean severity-adjusted labs per day at the hospital level with an R2=0.80, indicating that hospitals that ordered more labs for one condition also ordered more labs for the other condition.

Quasi Poisson regression analysis indicated an association between raw number of labs per encounter and major teaching hospital (relative risk (RR) cellulitis (95% CI) = 1.44 (1.27-1.62), p<0.001; pneumonia RR= 1.48 (1.33-1.64), p<0.001) when compared to non-teaching hospitals.

Conclusions: Patients admitted to hospitals with pneumonia or cellulitis in the state of Texas may receive 20-60% more laboratory tests if they go to a major teaching hospital versus a non-teaching hospital. This relationship holds for two common inpatient conditions and after multiple adjustments for illness severity, length of stay and patient demographics.