METHODS

Data

The study cohort was derived from the National Cancer Data Base (NCDB) Participant User File. The NCDB is a national database that captures approximately 70% of incident cancer cases in the United States. All cases of pancreatic adenocarcinoma that presented with clinical stage IV disease from 2003 to 2011 were abstracted from the database. Interventions that were aimed at palliating or alleviating symptoms were denoted “palliative”. Patients were stratified into those who underwent a specified type of palliative therapy (surgery, radiation, systemic chemotherapy, pain management, or a combination thereof), those who received a referral for palliative therapy, or those without intervention or referral.

Covariates

Covariates of interest included patient-specific factors such as age (stratified into age groups 18–59, 60–69, 70–79, or ≥80 years), sex, race/ethnicity (white non-Hispanic, black non-Hispanic, Hispanic, or other), insurance status (private, Medicare, Medicaid, uninsured, or unknown), quartile of median household income, and number of Charlson/Deyo comorbidities (0, 1, or ≥2). Hospital-specific covariates included facility type (community cancer program, comprehensive community cancer program, academic/research program, or other) and geographic region of the United States (Northeast, South, Midwest, and West).

Statistical analysis

Demographic and facility-related covariates were compared between the group that received a specified palliative intervention and the group that did not using standard univariate statistics. Regression analyses were used to identify significant trends in use of palliative therapy over the time period of the cohort. Multivariable logistic regression was used to model receipt of palliative therapy as a function of the aforementioned covariates. Kaplan-Meier curves were constructed for the groups undergoing various palliative interventions. In order to determine the association of the various palliative- and non-palliative interventions with patient survival, multivariable survival analyses were performed using a Weibull model because survival was not consistent with a proportional hazards assumption. All analyses were performed using STATA software (version 12.1, StataCorp, College Station, TX, USA). Statistical significance was defined as P<0.05.

RESULTS

Study cohort

Within the primary study cohort, a total of 68,075 patients diagnosed with stage IV pancreatic adenocarcinoma were identified; of these, 11,449 (16.8%) received designated palliative therapy. Among this subset, the largest proportion received systemic chemotherapy (37.2%), followed by surgery (19.0%), pain management alone (15.3%), radiation (8.1%), or a combination thereof (8.7%). Compared to the 16.8% of patients receiving designated palliative therapy, rates of utilization of non-palliative therapy were as follows: systemic chemotherapy (43.6%), surgery (0.6%), radiation (1.4%), and combination therapy (4.6%). Approximately 11.7% had a referral for palliative care, but no specific interventions were identified. Univariate comparisons between those receiving and not receiving designated palliative therapy are shown in Table 1.

TABLE 1 HERE

(To view a larger version of Table 1, click here.)