Associations Between Degree of Hospital-Level Diagnostic Test Use, LOS, and Return Visit Outcomes

Hospital-Level Resource Utilization (Tertile)
Unadjusted AnalysisAdjusted Analysisa
 Range of hospital-level medians, d3–72–6.53.5–6N/AN/AN/A
 % difference (95% CI)bRef0.2 (−13.8–16.4)16.4 (2.3–32.4)Ref−0.8 (−14.4–14.9)13.9 (−0.7–30.6)
30-Day revisit rate
 All-cause revisit rate, %6.2–38.517.7–33.310.3–36.9N/AN/AN/A
  OR (95% CI)Ref1.10 (0.83–1.47)0.97 (0.72–1.31)Ref1.19 (0.93–1.52)1.07 (0.82–1.39)
 Inpatient/observation readmissions only, %0–35.912.8–25.910.3–31.0
  OR (95% CI)Ref1.01 (0.72–1.42)0.88 (0.62–1.25)Ref1.08 (0.80–1.45)0.93 (0.68–1.28)
 ED revisits only, %0–12.20–15.70–13.2
  OR (95% CI)Ref1.40 (0.77–2.55)1.35 (0.76–2.39)Ref1.46 (0.76–2.78)1.50 (0.84–2.68)
  • Thirty-day revisit outcomes were analyzed by mixed-effects logistic regression, with a random intercept for hospital. LOS outcome was analyzed by mixed-effects negative binomial regression, with a random intercept for hospital. ED, emergency department; N/A, not applicable; OR, odds ratio; Ref, reference.

  • a LOS analyses adjusted for patient age, sex race, Hispanic ethnicity, type of insurance (public, private, or other), hospital region (Midwest, Northeast, South, or West), and average daily census. Thirty-day revisit analysis additionally adjusted for hospital average pediatric LOS weight.

  • b LOS interpreted as the percentage mean difference in outcome (from low resource utilization).