The “Comfortable Dying” Measure: How Patient Characteristics Affect Hospice Pain Management Quality Scores

Introduction

The Centers for Medicare and Medicaid Services (CMS) requires that quality measures be collected and reported for many Medicare-funded programs. ¹ Increasingly, these quality measures are being publicly reported. In theory, such measures allow patients and families to make more informed choices among health care providers. ¹

In a preliminary step toward public reporting in hospice, hospices were required to report to CMS the “Comfortable Dying” Measure. This measure is defined as the proportion of patients whose pain was brought under control within 48 hours of hospice enrollment, and is known by its National Quality Forum number (#0209). ² Although it is unclear whether the #0209 measure will be discontinued or modified, interpretation of these data is critical to the assessment of future quality measures. With more than 1.6 million patients enrolled in hospice in 2011, ³ future measures have the potential to affect a very large population.

As quality measures are implemented for public reporting in hospice, it will be particularly important to understand how a hospice's performance is influenced by its patient case mix. It is not known which patient characteristics are associated with a better score on the #0209 measure. When such characteristics influence #0209 scores but are beyond the hospice's control, they should be considered for inclusion in methods of case-mix adjustment. Therefore, the goal of this study was to describe the implementation of the #0209 measure in a network of hospices, and to identify patient characteristics that could be used to develop appropriate case-mix adjustment methods for hospices reporting quality data.

Methods

Patient data were extracted from the electronic medical records of 10 hospices in the CHOICE network (Coalition of Hospices Organized to Investigate Comparative Effectiveness). CHOICE is a research-focused collaborative of hospices that all use Suncoast Solutions Electronic Health Record (EHR) Software, and which have agreed to share their data for research purposes. CHOICE projects are defined and approved by a steering committee comprising leaders from hospices in the network. Hospices participating in this study range in size from 400 to 1700 patients/day and are located in New Mexico, California, Florida, Pennsylvania, Wisconsin, Michigan, Ohio, Texas (two hospices), and Kansas/Missouri. All are not-for-profit.

Patients were included if they were admitted to a participating hospice between October 1, 2012 and December 31, 2012. We extracted a dataset containing basic demographic variables (age, gender, race), admitting diagnosis, and NFQ #0209 data. Calculated #0209 scores indicate the number of patients with pain at admission who reported that their pain had been controlled to a comfortable level within 48 hours. It is important to note that this proportion underestimates the number of patients whose pain was controlled, because patients who were incapable of reporting their pain on the second assessment are counted as having inadequate pain control. Coding included site of care at the time of enrollment (home, long-term care facility, hospital, hospice inpatient unit). We also extracted clinical data elements that were markers of the severity of the patient's illness and the complexity of care (e.g., presence of Foley catheter, use of oxygen, and use of opioids at the time of hospice enrollment). Additionally, Palliative Performance Scale (PPS) scores for each patient were extracted. The PPS is an 11-point scale (scored 0–100 in 10-point increments) in which higher numbers indicate better function. ⁴ We initially described the PPS score as a continuous variable. In subsequent analysis, for ease of interpretation in calculating predicted survivals, we grouped PPS scores into two categories (0–20; 30–100).

To identify patient characteristics that were independently associated with #0209 scores, we first used univariate regression models. Next, we created a multivariable model using forward stepwise selection. We considered all variables that had an association that reached at least a moderate level of significance (p<0.25) in the univariate models. ⁵ We added variables systematically in all possible combinations and selected the one with the lowest Aikake Information Criterion. ⁶ Models used robust standard errors, clustered by hospice, to account for hospice-level effects.

Stata statistical software (Stata MP2 version 11.0; StataCorp. LP, College Station, TX) was used for all analysis. The University of Pennsylvania Institutional Review Board provided approval for use of secondary data.

Results

During the study period, 4157 patients were eligible for the #0209 measure (Table 1). At the first assessment, 28% of patients (n=1152) had no pain, 12% (n=515) were unable to respond, and 12% (n=498) had no documentation in the EHR. The remaining patients (n=1992; 48%) had pain and were eligible for the second assessment.

Of these 1992, 58% (n=1152) had their pain controlled within 48 hours of hospice enrollment (range across hospices: 46.3%–64.3%). An additional 286 (14%) patients had no second assessment recorded. Of these 286 patients, 104 (5%) had died, 18 (1%) had withdraw from hospice, 112 (6%) had a reassessment that was outside of the 72-hour window, and 52 (3%) did not have a documented reassessment.

In a multivariable logistic regression model, six patient characteristics were independent predictors of pain control on the #0209 measure within 48 hours of admission (Table 2). For instance, older patients were more likely to have their pain controlled at 48 hours (adjusted OR 1.02; 95% CI 1.02–1.03, p=0.003). Patients admitted to an inpatient hospice unit were also more likely to report that their pain was controlled (OR 1.28; 95% CI 1.08–1.47, p=0.031), as were those with a diagnosis of cancer (OR 1.37; 95% CI 1.20–1.53, p=0.008). The presence of a Foley catheter (OR 1.40; 95% CI 1.15–1.59, p=0.038), use of opioid medication at the time of hospice enrollment (OR 1.34; 95% CI 1.03–1.74, p=0.027) and PPS score (OR 1.02; 95% CI 1.01–1.03, p=0.000) were also positively associated with pain control. Conversely, presence of a Stage 2 pressure ulcer was independently associated with worse pain control (OR 0.63; 95% CI 0.31–0.96, p=0.012).

Discussion

Patients near the end of life experience serious, debilitating pain in a variety of medical settings.^7–11 Poorly controlled pain affects quality of life and functional status, and causes suffering for patients and family caregivers.^10–12 Due to the high prevalence of pain during the last week of life, the timely evaluation and treatment of pain at the time of admission to hospice should be a priority. However, before national quality measures can be developed for public reporting purposes, it will be essential to understand the factors that may influence pain scores.

In this study, we identified several patient characteristics that were associated with management of pain within 48 hours. Some of these are beyond a hospice's control. For instance, a hospice with a younger patient population is likely to have lower, or worse, #0209 scores. It is not clear why this association exists, but it is possible that younger patients have different pain perceptions, or higher expectations of pain management. In addition, age-specific differences in drug metabolism have been well described.^13–15 We also found that pain scores were worse for patients with noncancer diagnoses, which may be due to different pain etiologies and perhaps to a higher prevalence of more opioid-responsive pain in patients with cancer.^13,16 Other associations are more difficult to explain, and should be the subject of future research. Together, these results suggest the need for case-mix adjustment if the #0209 and similar measures are required for hospices, and particularly if these measures will be used for public reporting.

Another important finding of this study is that pain scores were better for patients admitted to an inpatient hospice unit. This result is surprising, because patients with more severe pain would be more likely to be admitted to an inpatient unit, and so the results reported here likely underestimate the impact of inpatient unit care on pain management. The use of inpatient care may be at least partly under hospices' control. Of course, the choice of setting of care is driven largely by patient and family preferences, and many patients may prefer to receive hospice care at home. ¹⁷ Furthermore, hospices need to justify an inpatient admission. Nevertheless, increased use of inpatient care, within the bounds of regulatory requirements and patient/family preferences, would seem to offer a potentially important opportunity to improve pain management.

This study has two limitations that should be noted. Although these results come from multiple hospices reflecting a large sample size, the patient population (88% white) limits the applicability of these findings to all hospice populations. Second, there may be other variables not described that are associated with pain scores. For instance, it is possible that hospices may face additional challenges in managing the pain of patients who live farther from the hospice's offices. In addition, #0209 scores may be influenced by care that patients received prior to hospice enrollment (e.g., previous efforts to manage pain). Therefore, further research is needed to understand the impact of other patient characteristics and other factors on this and future quality measures.

As the hospice industry continues to grow, it will be essential to develop and validate quality measures that are suitable for public reporting. When these measures are developed and implemented, however, valid methods of case-mix adjustment are needed. If factors associated with these and other quality scores are not defined, there is a significant risk that hospices will be penalized for factors beyond their control.