No abstract

A seven-item patient-centric PRO instrument was able to separate advanced malignancy patients into cohorts who their physicians deemed were at differing stages in their cancer journey with increasing needs for advance care planning. A study to determine if the threshold scores identified in this pilot correlate with palliative/EOL consultation frequency and patient survival is underway.

86 Background: A diagnosis of advanced cancer frequently thrusts family members into the role of caregiver. Although caregiver burdens have been well documented, less is known about the level of concern borne by the patient (pt) with cancer in placing a family member in this role, known as self-perceived burden (SPB). Methods: As part of the larger “Living with Cancer” project we prospectively surveyed 1307 pts with advanced malignancies receiving treatment with non-curative intent at 17 New Jersey cancer programs within the Regional Cancer Care Associates network between Sept 2015 and Apr 2016. Pts were asked one question about SPB on family members (5-level Likert scale). Results: Pts felt that “Living with Cancer burdens my family” all day every day 68 pts (5%), part of each day 109 pts (8%), most days 136 pts (10%), occasionally 571 pts (44%) or not at all 423 pts (32%). Twenty-four percent of responses were flagged as concerned (rated most days or greater by 313 pts). In a logistic regression model, SPB was correlated with marital status (married and divorced more concerned than single) and younger age (both p < 0.05). Patients living in lower median income neighborhoods also appeared to have a higher frequency of concern (p < 0.1) Factors not correlating with the level of SPB included gender, race, solid vs liquid tumor type, and length of cancer diagnosis. SPB was also not influenced by DNR status, having developed a Living Will, or documentation of power of attorney. Distance from the pt’s home to the cancer center was not associated with SPB on the caregiver. For comparison, on the same LWC project 38% of pts with advanced cancer were concerned about the financial toxicity of their care and 33% were concerned about pain (both p < 0.01 compared to SPB). Conclusions: Self-perceived burden on a caregiver was identified in 24% of pts with advanced cancer, less than those concerned about financial toxicity or pain, in this NJ series. Divorced/married pts and younger pts with cancer are more likely to express concern. Developing an EOLC plan (DNR/Living will/POA) does not appear to influence SPB concerns.

85 Background: End-of-life care (EOLC) discussions for pts with advanced malignancies are often delayed resulting in unwanted, wasteful treatment. COTA Inc has developed a 7-item PRO instrument (LWC) to facilitate identification of pts in distress. A pilot study of 433 pts at a single institution was able to determine a threshold distress score (> 28) correlating with physician assessment of appropriateness for EOLC consultation (2015 ASCO Palliative Care Symposium). Methods: LWC explores 4 personhood and clinical domains (performance status, pain, burden, and depression). Each question presents a 5-level Likert scale supplemented by a pt derived weighting of importance. The PRO was completed by 700 pts at Regional Cancer Care Associates throughout NJ between Sept 2015 and Apr 2016. The primary oncologists separately/independently classified 611 pts as appropriate to continue therapy (Group 1) and 89 pts as appropriate to curtail therapy (Group 2). Results: A score of 29 on LWC (range 0-112) was the threshold achieving optimal sensitivity (51%) and specificity (78%) in identifying pts in Group 2 (51% of pts whose oncologists felt appropriate to curtail therapy scored > 29; 78% of pts whose oncologists felt appropriate to continue therapy scored < 29). The threshold had positive predictive value 25% and 8% false omission rate (1 - negative predictive value) (25% pts scoring > 29 should be in Group 2; 8% scoring < 29 should be in group 2). A score 29 was 2.3 times more likely in pts belonging to Group 2 (likelihood ratio), and was a predictor of physician category (logistic regression p < 0.0001). LWC had a Cronbach’s alpha coefficient 0.85, indicating internal consistency and ability to measure a single latent construct. 3 questions (Performance Status, Wants, and Lack of Pleasure) contained the most information (estimate the latent trait with precision) and had high discrimination parameters (discriminate between pts lacking latent trait and displaying latent trait). Conclusions: The LWC instrument is able to identify cancer pts for whom EOLC discussions may be appropriate. A study to determine the correlation between the threshold score, EOLC consultation rate and overall survival is underway.

10025 Background: The Living with Cancer (LWC) patient reported outcome (PRO) instrument evaluates distress from the point of view of the advanced cancer pt. The 7-item Likert survey measures 4 personhood domains (performance status, pain, burden [financial and family], depression) with scores ranging 0-112. In a pilot study of 433 cancer pts at a single center a score of >28 was associated with an increased likelihood of physician’s (blinded) opinion of need for end-of-life care discussions ( J Palliative Med 2016). Methods: The LWC instrument is a statistically validated PRO (ASCO Palliative Care Symposium 2016). LWC was administered to 1024 cancer pts receiving non-curative therapy at 7 centers (Regional Cancer Care Associates, NJ) from Sept 2015 - Oct 2016. LWC surveys were linked to the Cota database, which extracts and enriches data from EHRs. Date of survey was used as the start point in time-to-event analysis. Results: 290 (28%) pts expired during the study (median f/u 9.9 months). 267 (26%) pts exceeded the threshold score of 28 defined in the pilot set (28 was also independently this study’s optimal cut point). Pts with an LWC score >28 had inferior 6 and 12 mo overall survival (69% and 54%) vs pts with scores <29 (88% and 73%) (log rank p<0.001). A Cox model demonstrated that LWC score and cancer type were significant (LWC: p<0.001, cancer types (compared to B): GI p<0.001, GU: p=0.013, T: p<0.001, M: p=0.334) A one point score increase in LWC resulted in a 1.8% increase in expected hazard. Among solid tumor pts with LWC >28, 20% died within the next 3 mo and 35% died within the next 6 mo, indicating appropriate timing for hospice and palliative care consults, respectively. Conclusions: Pt responses to the LWC instrument predict survival among advanced cancer pts and may be useful in guiding timing of palliative care consultations. [Table: see text]