The effect of hospital volume on patient outcomes in severe acute pancreatitis

Abstract

Background: We investigated the relation between hospital volume and outcome in patients with severe acute pancreatitis (SAP). The determination is important because patient outcome may be improved through volume-based selective referral.

Methods: In this cohort study, we analyzed 22,551 SAP patients in 2,208 hospital-years (between 2000 and 2009) from Taiwan's National Health Insurance Research Database. Primary outcome was hospital mortality. Secondary outcomes were hospital length of stay and charges. Hospital SAP volume was measured both as categorical and as continuous variables (per one case increase each hospital-year). The effect was assessed using multivariable logistic regression models with generalized estimating equations accounting for hospital clustering effect. Adjusted covariates included patient and hospital characteristics (model 1), and additional treatment variables (model 2).

Results: Irrespective of the measurements, increasing hospital volume was associated with reduced risk of hospital mortality after adjusting the patient and hospital characteristics (adjusted odds ratio [OR] 0.995, 95% confidence interval [CI] 0.993-0.998 for per one case increase). The patients treated in the highest volume quartile (≥14 cases per hospital-year) had 42% lower risk of hospital mortality than those in the lowest volume quartile (1 case per hospital-year) after adjusting the patient and hospital characteristics (adjusted OR 0.58, 95% CI 0.40-0.83). However, an inverse relation between volume and hospital stay or hospital charges was observed only when the volume was analyzed as a categorical variable. After adjusting the treatment covariates, the volume effect on hospital mortality disappeared regardless of the volume measures.

Conclusions: These findings support the use of volume-based selective referral for patients with SAP and suggest that differences in levels or processes of care among hospitals may have contributed to the volume effect.

Figure 1

Study flow diagram. (Note: Patients hospitalized for AP between 1996 and 1999 were excluded to ensure the inclusion of first-attack cases because most relapses occur within the first 4 years after the first "attack)."

Figure 2

Distribution of hospital volume versus hospital mortality per hospital-year in severe acute pancreatitis (Note: There were 467 hospitals contributing to a total of 2,208 hospital-years. Median hospital volume was 5 cases per hospital-year [interquartile range 2 − 13]).

Figure 3

Effect of hospital volume on hospital mortality in severe acute pancreatitis adjusting for patient and hospital characteristics (Note: Hospital volume was divided into 9 about-equal subsets).

Figure 4

Variances of outcomes (A: hospital mortality, B: hospital length of stay, C: hospital charges) explained by various variables (MV: mechanical ventilation; TPN: total parenteral nutrition; AP: acute pancreatitis).