Research Article: Readmissions attributable to skilled nursing facility use after a colectomy: Evidence using propensity scores matching

Date Published: April 16, 2019

Publisher: Public Library of Science

Author(s): Yubraj Acharya, Amber L. Schilling, Christopher S. Hollenbeak, Stanley W. Ashley.


Postacute care (PAC) is a major driver of the rising health care costs in the United States (US). There is limited evidence on the causal effect of skilled nursing facility (SNF) use on readmission after an inpatient colectomy.

We performed a retrospective analysis of data from the Pennsylvania Health Care Cost Containment Council (PHC4) on 38,635 patients who underwent an inpatient colectomy between 2011 and 2014 in a Pennsylvania hospital. Using propensity scores, we matched patients who were discharged to a SNF to those who were discharged elsewhere. We compared the probability of readmissions within 30 days for the two groups of matched patients in a regression framework. For the subset of patients who were readmitted within 30 days, we assessed whether patients discharged to SNF were readmitted earlier than those discharged to other entities.

The use of a SNF after a colectomy significantly raises the patients’ chance of readmissions within 30 days, even after controlling for their demographic characteristics and illness severity. Based on our estimates, being discharged to a SNF raises the chance of a readmission by 7.7 percentage points. For patients who were admitted within 30 days, we find no association between discharge to a SNF and the timing of readmission.

Sending less severe patients to facilities other than a SNF following inpatient colectomy may help hospitals reduce 30-day readmission rates.

Partial Text

Postacute care (PAC) is a major driver of the rising health care costs in the United States (US) [1,2]. An emerging body of literature has examined the variation in health care spending across the postacute care options—namely, readmissions, and the use of a skilled nursing home facility (SNF), inpatient rehabilitation facility (IRF), outpatient rehabilitation facility (ORF), or a home health agency (HHA) [3,4]. These studies attempt to parse out the contribution of these PAC options to the overall health care spending.

In our sample of 38,635 patients who underwent a colectomy in Pennsylvania during 2011–2014, the average age was 63 years (Table 1, Column 1). Eighty-eight percent (88.4%) of these individuals were White and approximately 46% were male. Most colectomies were performed electively and using a non-laparoscopic approach (61.1% and 63.1%, respectively). The most common indications for having the colectomy performed were cancer (29.1%) and diverticular disease (25.1%); the “other” indication category in Table 1 includes unspecified septicemia, intestinal obstruction, non-infectious enteritis, intestinal perforation, intestinovesical fistula. Most patients (97.7%) did not have a diagnosis indicating ostomy use (colostomy or ileostomy). Of the total number of patients, 18,931 (49%) were on Medicare, 15,840 (41%) had private insurance, and 3,090 (8%) were on Medicaid.

In an attempt to generate a causal estimate of the effect of SNF utilization on readmissions, we used the PSM technique that is acceptable but, in our view, underused in health services research. We also used all-payer data, thus addressing the external validity concerns common to prior studies that use Medicare-only data. In our analysis, we were able to control for a wider range of potential confounders, especially those related to the patient’s illness severity (e.g., the urgency of the surgery and the approach used), than has been done previously. We found that being discharged to a SNF was significantly associated with a patient’s chance of readmission within 30 days following colectomy, even after controlling for demographic characteristics and surrogate markers for illness severity. Based on our estimates, being discharged to a SNF raises the risk of readmission by 7.7 percentage points. This is a large effect, given that the overall readmissions rate in our sample is 13.5%.