3069 — The Quality of Pain Screening and Management for Veterans with Advanced Cancer: A Pilot Study of the ASSIST Pain Indicators

O'Neill SM (VA Greater Los Angeles Healthcare System), Lorenz KA (VA Greater Los Angeles Healthcare System), Asch SM (VA Greater Los Angeles Healthcare System), Tisnado DM (Department of Medicine, University of California, Los Angeles), Walling AM (Department of Medicine, University of California, Los Angeles), Malin JL (VA Greater Los Angeles Healthcare System)

Objectives:
Because pain is common among veterans with cancer, but quality indicators (QIs) have been lacking, we piloted the ASSIST (Assessing Symptoms Side Effects and Indicators of Supportive Treatment) QIs in patients with advanced cancer at the VA Greater Los Angeles Healthcare System (VA GLAHS).

Methods:
We operationalized 14/15 QIs. Eligible veterans were those with: a cancer registry-documented stage IV solid tumor diagnosed in 2006, > = 1 outpatient visit, and survival > = 1 month. Trained oncology nurses abstracted from electronic medical records, using a two-hour protocol on a computer-based abstraction tool.

Results:
Eligible veterans (N = 118) were predominantly male (98%); 38% were non-white. Mean (SD) age was 64 (10) years. Median survival was 12.1 months on median follow-up of 24.9 months. The most common tumors were lung (23%), head/neck (23%), prostate (17%), and colon (9%). We collected data for 14/15 indicators, but due to our sampling strategy, only 4/14 of these QIs had data from more than five patients. We present the results of those four QIs, measured at the patient level: 93% (101/109) of veterans ever received recommended pain screening; 35% (27/78) of veterans with pain had an assessment of etiology documented; 37% (27/74) of veterans with pain were assessed for functional impairment; and 18% (4/22) of veterans’ pain regimens were modified for severe/worsening pain or medication side effects. As a proportion of 990 total outpatient visits, the QI adherence to recommended pain screening was 50%, and showed considerable variation across clinical settings (range: 29-77%). In visits where pain was present by 5th vital sign, a clinician also documented its presence two-thirds of the time.

Implications:
Using a two-hour abstraction protocol, we successfully piloted 14/15 Cancer-Quality ASSIST Pain QIs. This represents a promising advance in the use of abstracted data for the evaluation of cancer pain management. Data abstracted from charts provided for the analysis of four indicators; and time requirements suggest that pharmacologic data is best derived administratively. Despite national attention to pain management, many veterans with advanced cancer did not receive recommended care.

Impacts:
The ASSIST indicators demonstrate the feasibility of evaluating the quality of cancer pain screening and management. These results highlight promising targets for improving pain management for seriously ill veterans.