PRACTICE MAKES PERFECT: THE RELATIONSHIP BETWEEN PHYSICIAN VOLUME AND OUTCOME FOR GASTRECTOMY

Purpose: The aim of this study is to investigate the relationship between physician volume and health outcome for gastrectomy surgery.

Method: This was an observational study with retrospective data analysis, and the dataset was retrieved from 2004-2005 Taiwan National Health Insurance Research Database (TNHIRD). The study subjects were consecutive patients who underwent their gastrectomy during the period of 2004 to 2005. The dependent variables were health outcomes of the gastrectomy surgery, including infection (ICD9-CM: 998.51, 998.59), length of stay, and hospitalization expenses during that hospitalization. The surgeon-volume on gastrectomy of the physicians who implemented the surgery for the index patient in 2004 was treated as the independent variable, which was categorized into 3 levels: low(<=8 cases), medium(9-22 cases), and high(>=23 cases). Statistical analysis applied included descriptive analysis, Chi-square test, analysis of variance (ANOVA), and hiearchical linear model to examine the relationship between physician-volume and health outcome for gastrectomy surgery. By applying hierarchical linear model, we accounted for the clustered nature of the data.

Result: A total of 2,773 gastrectomy cases were analyzed in this study. Of whom 35.16% were treated by physicians with low volume, 33.18% with medium volume, and 31.66% with high volume. The mean age and standard deviation was 64.73±14.45. Over 66% of the subjects were male. The mean Elixhauser comorbidity index and standard deviation was 1.85±1.79. After controlling for the characteristics of patients and hospitals, the results of hierarchical linear models showed that comparing with low physician-volume group, patients with high physician-volume had lower risk of infection (OR=0.565, p<0.05) and shorter length of stay (β= -3.020, p<0.001). With regard to hospitalization expenses, both patients with medium and high physician-volumes expensed less than the low volume group (β= -45,809, p<0.001; β= -78,145, p<0.001).

Conclusion: By using a nationwide population-based dataset, we found that higher physician volume is associated with lower risk of infection, length of stay, and hospital expenditure in gastrectomy patients.