Factors Associated with Family Reports of Pain, Dyspnea, and Depression in the Last Year of Life

Patterns of Unmet Needs With Assistance for Self-Care Activities at the End of Life by Place of Death Among Older Adults

Older Americans experience adverse consequences linked to inadequate assistance with self-care activities. These unmet care needs increase risk of hospitalization, institutionalization, and morbidity. While a growing body of research has focused on disparities in unmet care needs and place of death outcomes among older adults separately, less is known about the intersection between the two: place of death and unmet care needs at the end of life. Therefore, this study utilized data from the National Health & Aging Trends Study (NHATS) to examine patterns of place of death and unmet care needs outcomes among older adults to identify disparities in care and inform policies and programs. We conducted weighted bivariate analyses and multivariate logistic regression models using appropriate survey weights. Findings suggest more than half of older adults experienced an unmet need at the end of their lives and increasing rates of unmet needs are experienced by older adults who died in hospitals. Older adults receiving hospice care were also found to have less unmet needs at the end of their lives. Given the well-documented consequences of unmet needs with self-care activities, ongoing research is needed to better inform policies and programs that provide assistance and support to older adults with their self-care activities during their end of lives.

Patterns of Symptom Management Medication Receipt at End-of-Life Among Medicare Beneficiaries With Lung Cancer

CONTEXT

Older adults with advanced lung cancer experience high symptom burden at end of life (EOL), yet hospice enrollment often happens late or not at all. Receipt of medications to manage symptoms in the outpatient setting, outside the Medicare hospice benefit, has not been described.

OBJECTIVES

We examined patterns of symptom management medication receipt at EOL for older adults who died of lung cancer.

METHODS

This retrospective cohort used the Surveillance, Epidemiology, and End Results-Medicare database to identify decedents diagnosed with lung cancer at age 67 years and older between January 2008 and December 2013 who survived six months and greater after diagnosis. Using Medicare Part B and D claims, we identified monthly receipt of outpatient medications for symptomatic management of pain, emotional distress, fatigue, dyspnea, anorexia, and nausea/vomiting. Multivariable logistic regression estimated associations between medication receipt and patient demographic characteristics, comorbidity, and concurrent therapy.

RESULTS

Of the 16,246 included patients, large proportions received medications for dyspnea (70.7%), pain (62.5%), and emotional distress (49.4%), with lower prevalence for other symptoms. Medication receipt increased from six months to one month before death. Women and dual Medicaid enrolled were more likely to receive medications for pain, emotional distress, dyspnea, and nausea/vomiting. Receipt of symptom management medications decreased with increasing age and racial/ethnical minorities.

CONCLUSION

Symptom management medication receipt was common and increasing toward EOL. Lower use by males, older adults, and nonwhites may reflect poor access or poor patient-provider communication. Further research is needed to understand these patterns and assess adequacy of symptom management in the outpatient setting.

Polypharmacy Increases Risk of Dyspnea Among Adults With Serious, Life-Limiting Diseases

BACKGROUND

Polypharmacy is associated with dyspnea in cross-sectional studies, but associations have not been determined in longitudinal analyses. Statins are commonly prescribed but their contribution to dyspnea is unknown. We determined whether polypharmacy was associated with dyspnea trajectory over time in adults with advanced illness enrolled in a statin discontinuation trial, overall, and in models stratified by statin discontinuation.

METHODS

Using data from a parallel-group unblinded pragmatic clinical trial (patients on statins ≥3 months with life expectancy of 1 month to 1 year, enrolled in the parent study between June 3, 2011, and May 2, 2013, n = 308/381 [81%]), we restricted analyses to patients with available baseline medication count and ≥1 dyspnea score. Polypharmacy was assessed by self-reported chronic medication count. Dyspnea trajectory group, our primary outcome, was determined over 24 weeks using the Edmonton Symptom Assessment System.

RESULTS

The mean age of the patients was 73.8 years (standard deviation [SD]: ±11.0) and the mean medication count was 11.6 (SD: ±5.0). We identified 3 dyspnea trajectory groups: none (n = 108), mild (n = 130), and moderate-severe (n = 70). Statins were discontinued in 51.8%, 48.5%, and 42.9% of patients, respectively. In multivariable models adjusting for age, sex, diagnosis, and statin discontinuation, each additional medication was associated with 8% (odds ratio [OR] = 1.08 [1.01-1.14]) and 16% (OR = 1.16 [1.08-1.25]) increased risk for mild and moderate-severe dyspnea, respectively. In stratified models, polypharmacy was associated with dyspnea in the statin continuation group only (mild OR = 1.12 [1.01-1.24], moderate-severe OR = 1.24 [1.11-1.39]) versus statin discontinuation (mild OR = 1.03 [0.95-1.12], and moderate-severe OR = 1.09 [0.98-1.22]).

CONCLUSION

Polypharmacy was strongly associated with dyspnea. Prospective interventions to decrease polypharmacy may impact dyspnea symptoms, especially for statins.

The Role of Palliative Care in the Cardiac Intensive Care Unit

In the last few years, important changes have occurred in the clinical and epidemiological characteristics of patients that were admitted to cardiac intensive care units (CICU). Care has shifted from acute coronary syndrome patients towards elderly patients, with a high prevalence of non-ischemic cardiovascular diseases and a high burden of non-cardiovascular comorbid conditions: both increase the susceptibility of patients to developing life-threatening critical conditions. These conditions are associated with a significant symptom burden and mortality rate and an increased length of stay. In this context, palliative care programs, including withholding/withdrawing life support treatments or the deactivation of implanted cardiac devices, are frequently needed, according to the specific guidelines of scientific societies. However, the implementation of these recommendations in clinical practice is still inconsistent. In this review, we analyze the reasons for this gap and the main cultural changes that are required to improve the care of patients with advanced illness.

Depression and Health Care Utilization at End of Life Among Older Adults With Advanced Non-Small-Cell Lung Cancer

CONTEXT

Limited data exist regarding how depression diagnosed at different times relative to a cancer diagnosis may affect healthcare utilization at end of life (EOL).

OBJECTIVES

To assess the relationship between depression and health care utilization at EOL among older adults (ages >=67) diagnosed with advanced non-small cell lung cancer (NSCLC) from 2009 to 2011.

METHODS

Using the SEER-Medicare database, we fit multivariable logistic regression models to explore the association of depression with duration of hospice stay plus high-intensity care, for example inpatient admissions, in-hospital death, emergency department visits, and chemotherapy at EOL. We used a regression model to evaluate hospice enrollment, accounting for the competing risk of death.

RESULTS

Among 13,827 subjects, pre-cancer depression was associated with hospice enrollment (sub-hazard ratio 1.19, 95% confidence interval [CI] 1.11-1.28), 90 + hospice days (adjusted odds ratio [aOR] 1.29, 95% CI 1.06-1.58), and lower odds of most utilization; we found no association with EOL chemotherapy. Diagnosis-time depression was associated with hospice enrollment (SHR 1.16, 95% CI 1.05-1.29) but not high-intensity utilization. Post-diagnosis depression was associated with lower hospice enrollment (SHR 0.80, 95% CI 0.74-0.85) and higher odds of ICU admission (aOR 1.18, 95% CI 1.01-1.37).

CONCLUSION

EOL healthcare utilization varied by timing of depression diagnosis. Those with pre-cancer depression had lower odds of high-intensity healthcare, were more likely to utilize hospice, and have longer hospice stays. Regular depression screening and treatment may help patients optimize decision-making for EOL care. Additionally, hospice providers may need additional resources to attend to mental health needs in this population.

Machine Learning Methods to Extract Documentation of Breast Cancer Symptoms From Electronic Health Records

CONTEXT

Clinicians document cancer patients' symptoms in free-text format within electronic health record visit notes. Although symptoms are critically important to quality of life and often herald clinical status changes, computational methods to assess the trajectory of symptoms over time are woefully underdeveloped.

OBJECTIVES

To create machine learning algorithms capable of extracting patient-reported symptoms from free-text electronic health record notes.

METHODS

The data set included 103,564 sentences obtained from the electronic clinical notes of 2695 breast cancer patients receiving paclitaxel-containing chemotherapy at two academic cancer centers between May 1996 and May 2015. We manually annotated 10,000 sentences and trained a conditional random field model to predict words indicating an active symptom (positive label), absence of a symptom (negative label), or no symptom at all (neutral label). Sentences labeled by human coder were divided into training, validation, and test data sets. Final model performance was determined on 20% test data unused in model development or tuning.

RESULTS

The final model achieved precision of 0.82, 0.86, and 0.99 and recall of 0.56, 0.69, and 1.00 for positive, negative, and neutral symptom labels, respectively. The most common positive symptoms were pain, fatigue, and nausea. Machine-based labeling of 103,564 sentences took two minutes.

CONCLUSION

We demonstrate the potential of machine learning to gather, track, and analyze symptoms experienced by cancer patients during chemotherapy. Although our initial model requires further optimization to improve the performance, further model building may yield machine learning methods suitable to be deployed in routine clinical care, quality improvement, and research applications.