Back to Annual Meeting Posters

OBJECTIVE: Patients who undergo pancreatic surgery are among those at the highest risk for readmission. Evidence suggests that same hospital readmission is less costly than readmission to another hospital. The objective of this study was to identify modifiable and non-modifiable predictors of same versus other hospital readmission among patients who undergo surgery for pancreatic cancer.
METHODS: Medicare beneficiaries who underwent major pancreatic resection from 2000 to 2008 were identified from the Surveillance, Epidemiology, and End Results-Medicare database. Demographics, co-morbidities, cancer-related and treatment-related variables as well as hospital characteristics were examined. Using multivariable logistic regression, predictors of readmission were identified and then compared to predictors of same versus other hospital readmission.
RESULTS: Of 2,486 patients, 512 (21%) were readmitted within 30days of discharge. One thousand six hundred eighty five (68%) had their surgery at an academic medical center and 607 (24%) at a National Cancer Institutes (NCI)-designated cancer center. One thousand six hundred forty two (66%) had an initial length of stay greater than 10 days and 471 (19%) were discharged to a skilled nursing facility (SNF). Predictors of readmission included severity of co-morbidities (Charlson Comorbiditiy Score ≥3: OR 1.54 [95%CI 1.08 - 2.2] p = 0.017), initial length of stay greater than 10 days (OR 1.66 [95%CI 1.32 - 2.17] p <0.001), and discharge to a SNF (OR 1.50 [1.15 - 1.95] p=0.003). Of the readmitted patients, 387 (76%) returned to the same hospital while 125 (24%) were readmitted to another hospital. After controlling for co-morbidities, neoadjuvant radiation and chemotherapy, and other treatment-related variables, readmission to another hospital was more likely for patients if time from diagnosis to surgery was greater than 60 days (OR 3.80 [95%CI 1.67 -8.68] p=0.005). Readmission to another hospital was also more likely if the hospital where the pancreatic surgery occurred was an NCI-designated cancer center (OR 1.91 [95% CI 1.11- 3.28], p = 0.019) or if it was affiliated with a medical school (OR 2.31 [95% CI 1.10 - 4.83] p =0.027).
CONCLUSIONS: Readmission is common in patients who undergo surgery for pancreatic cancer, and almost 1 in 4 readmitted patients go to another hospital. Risk factors for 30-day readmission include comorbidities and a complicated post-operative course resulting in prolonged length of stay and discharge to a skilled nursing facility. Predictors of readmission to another hospital relate to prolonged time from diagnosis to surgery and pancreatic surgery at a tertiary care hospital. This study identifies the group of patients who are likely referred for resection from their local setting to a tertiary care center as a target for interventions to prevent or redirect readmissions.
Multivariable Analysis Identifying Predictors Of Readmission To Another Hospital After Pancreatic Surgery

EXPLANATORY VARIBALES	Adjusted OR	95% CI	P Value
Age			0.314
66-69	Reference
70-74	1.26	(0.64 -2.49)
75-79	1.30	(0.66 - 2.59)
80+	2.02	(0.95 - 4.29)
Time From Diagnosis To Surgery			0.005
<30days	Reference
30 to 60days	1.58	(0.85 - 2.91)
>60days	3.80	(1.67 - 8.68)
Hospitalizations In Previous Year			0.767
0	Reference
1	0.85	(0.48 - 1.50)
2	0.71	(0.35 - 1.44)
3+	1.01	(0.50 - 2.03)
Initial Length Of Stay			0.113
<10days	Reference
10+ days	1.54	(0.88 - 2.71)
Index Hospital Med School Affiliation	2.31	(1.10 - 4.83)	0.027
Index Hospital Number Of Beds			0.687
<300	0.75	(0.32 - 1.75)
300-600	0.82	(0.49 - 1.36)
>600	Reference
Index Hospital NCI Designation	1.91	(1.11 - 3.28)	0.019
Index Hospital Available Hospice	0.70	(0.43-1.14)	0.148
Neoadjuvant Chemo or Radiation	0.72	(0.25 - 2.07)	0.734
DischargeTo Skilled Nursing Facility	0.60	(0.33 - 1.06)	0.077

*Controlled for stage at diagnosis and diabetes mellitus in addition to above listed variables