First, Do No Harm: Rethinking Routine Diversion in Sphincter-Preserving Rectal Cancer Resection

Abstract

Background: Although diverting stomas have reduced anastomotic leak rates after sphincter-preserving proctectomy in some series, the effectiveness of routine diversion among a broad population of rectal cancer patients remains controversial. We hypothesized that routine temporary diversion is not associated with decreased rates of leak or reintervention in cancer patients at large undergoing sphincter-sparing procedures.

Study design: The Florida State Inpatient Database (AHRQ, Healthcare Cost and Utilization Project) was queried for patients undergoing sphincter-preserving proctectomy for cancer (2005 to 2014). Matched cohorts defined by diversion status were created using propensity scores based on patient and hospital characteristics. Incidence of anastomotic leak, nonelective reintervention, and readmission were compared, and cumulative 90-day inpatient costs were calculated.

Results: Of 8,620 eligible sphincter-sparing proctectomy patients, 1,992 matched pairs were analyzed. Leak rates did not significantly vary between groups (4.5% vs 4.3%; p = 0.76), but diversion was associated with significantly higher odds of nonelective reintervention (2.37; 95% CI 1.90 to 2.96) and readmission (1.55; 95% CI 1.33 to 1.81) compared with undiverted patients. Median costs were higher among those diverted (US$21,325 vs US$15,050; p < 0.01).

Conclusions: No association between diversion and anastomotic leak was found. However, temporary diversion was associated with increased incidence of nonelective reinterventions, readmissions, and higher costs. We therefore challenge the paradigm of routine diversion in rectal cancer operations. Additional study is needed to identify which patients would benefit most from diversion.

Copyright © 2019 American College of Surgeons. Published by Elsevier Inc. All rights reserved.

Figures

Figure 1.

Flow diagram of study population selection.

Figure 2.

Reinterventions performed based on diversion status. Distribution of the 10 most common reinterventions performed within 90 days of index proctectomy among the diverted (orange) and undiverted (blue) populations. Percentages were calculated from a total of all reinterventions performed across both groups. *Each of these categories account for

Figure 3.

Indications for readmission based on diversion status. Distribution of principal diagnoses associated with readmissions in the diverted (orange) and undiverted (blue) cohorts after proctectomy for cancer. Percents were calculated from a total of all readmissions that occurred across both cohorts. *Each of these categories account for

Figure 4.

Cost differences within matched pairs. Cost difference within matched pairs is calculated by subtracting the total cost of the undiverted patient from the total cost of their matched counterpart. Negative values denote pairs where the undiverted patient generated higher costs than the diverted patient. Pairs where no leak occurred are depicted in blue. Pairs in which the diverted member suffered a leak are denoted in yellow, and pairs in which the undiverted member leaked are pictured in red. USD, US dollars; IQR, interquartile range.

eFigure 1.

Matched pair standardized mean differences before and after matching. The standardized mean difference between diverted and undiverted cohorts of patient demographic, comorbidity, and treatment related varaibles. Differences both before (red dots) and after (blue dots) matching are displayed. Sufficient balance between cohorts is assumed wherever the standardized difference is

eFigure 2.

Propensity score distributions. Distribution of propensity scores among cohorts (A) before and (B) after matching. Propensity scores of undiverted patients represented in blue, diverted scores in red.

eFigure 3.

Time to ostomy reversal. Cumulative frequency of ostomy reversal during 2-year follow-up in the (A) electively diverted and (B) initially undiverted populations; patients represented in (B) underwent subsequent diversion after leaking. Overall reversal rates within the follow-up period were 88% and 74%, respectively.