Predictors of Placement of Inpatient Palliative Care Consult Orders Among Patients With Breast, Lung, and Colon Cancer in a Safety Net Hospital System

Abstract

Background:

The provision of palliative and end-of-life care to patients who are underrepresented and underserved provides unique challenges and opportunities.

Objectives:

To examine predictors of placement of inpatient palliative care consult orders among patients with breast, lung, and colorectal cancer hospitalized in a safety net hospital in 2010.

Methods:

Simple and multivariable logistic regression of data on selected patients with cancer was performed to identify predictors of placement of inpatient palliative care consult orders.

Results:

Of 979 patients, 56% had colorectal cancer, 23% had lung cancer, and 21% had breast cancer. Of those patients, 16% received an order for inpatient palliative care consultation during the study period. Patients who had more than 20 prescriptions for opioids ordered (adjusted odds ratio [AOR]: 9.10, 95% confidence interval [CI]: 4.62-17.95), had an order for a radiation oncology consult (AOR: 2.60, 95% CI: 1.50-4.49), or had low albumin (AOR: 2.75, 95% CI: 4.71) were more likely to have an order for an inpatient palliative care consult placed. Race and ethnicity were not statistically significant predictors.

Conclusion:

In this cohort of patients in a safety net hospital, markers of pain, advanced disease, and poor prognosis were associated with placement of inpatient palliative care consult orders.

Introduction

Difficulties with access to health care and lack of insurance coverage among the underserved have long been an issue, and because of this, patients with cancer in underserved communities often present at later stages in the disease process.^1,2 These medically underserved populations with advanced cancer are likely to seek care at urban or safety net hospitals that provide significant levels of care to low-income, uninsured, and vulnerable populations.³ Having palliative care programs in place may help these patients navigate the health-care system during treatment and at the end of life. Palliative care consultation among the poor and underserved has been shown to have an effect on hospice enrollment and completion of advance care planning—components of end-of-life care that have historically been underutilized in the populations typically served within safety net hospital systems.^4,5 In fact, some programs have noted significant increases in the numbers of seriously ill patients who elect to complete do not resuscitate (DNR) orders.⁵ Additional research has shown increases in hospice enrollment, and that racial and ethnic differences in hospice enrollment may be tempered by inpatient palliative care consultation within the safety net and urban teaching hospital setting as well.⁵

Prior research has been done to examine which factors lead to palliative care consultation or referral in patients with cancer in diverse settings. For instance, some research suggests that patients who receive inpatient palliative care consultation have higher likelihoods of being at the end of life, as noted by higher hospital mortality, higher intensive care unit mortality, and more frequent discharges to hospice.⁶ Some research has identified specific patient-centered factors that may be related to palliative care referral, including older age, specific cancer type, and marital status.⁷ Although other studies have identified the need for the management of cancer-related symptoms as a significant predictor of palliative care referral.⁸ The use of palliative care consultation in the safety net hospital setting is ever evolving and so are the factors associated with palliative care consultation and hospice utilization in this population. The goal of this work is to examine factors associated with the ordering of inpatient palliative care consultation among a cohort of patients with breast, colon and lung cancer in a safety net hospital over the course of 1 year. By doing so, we hope to evaluate the differences in sociodemographic, clinical, and other patient and care characteristics of patients for whom inpatient palliative care consult orders may have been placed and conduct the formative research needed to design and implement interventions that will improve palliative and end-of-life care in this population.

Methods

Study Population and Data Collection

We conducted a secondary analysis of data previously collected for all patients hospitalized with breast, colorectal, or lung cancer from January 1, 2010, to December 31, 2010, at Parkland Health and Hospital System. Parkland is the sole safety net provider in Dallas County and is responsible for caring for the county’s uninsured and underinsured. It is a public, integrated delivery system with a comprehensive electronic medical record (EMR) used across all inpatient and outpatient settings. Parkland Hospital, a 900-bed hospital, is a major teaching affiliate of UT Southwestern Medical Center. Poor Dallas County residents without insurance are eligible for the Parkland Financial Assistance program that covers hospital and ambulatory-based care including cancer treatment, palliative care, and pain medications. In the original data set, patients were identified through the EMR by diagnosis as defined by International Classifications of Diseases, Ninth Revision Clinical Modification (ICD-9-CM) codes.⁹ The codes are as follows: lung cancer (162, 162.0, 162.2, 162.3, 162.4, 162.5, 162.8, and 162.9), breast cancer (174, 174.0, 174,1. 174.2, 174.4, 174.5, 174.6, 174.8, and 174.9), colorectal cancer (153.0, 153.1, 153.2, 153.3, 153.4, 153.5, 153.3, 153.7, 153.8, 153.9, 154, 154.0, 154.1, and 154.8), and metastatic disease or “secondary malignant neoplasm” (197.0, 197.1, 197.2, 197.3, 197.4, 197.5, 197.6, 197.7, 197.8, 198.0, 198.1, 198.2, 198.3, 198.4, 198.5, 198.6, 198.7, 198.8, and 199.0). In this secondary analysis, data regarding the specific stage of disease were not available; consequently, we examined this previously extracted data on several domains that we believe may influence prognosis and appropriateness for palliative care consultation, including: sociodemographic characteristics (age, sex, race, ethnicity, marital status, and insurance coverage), prognostic variables (cancer type, metastasis, and albumin), orders for consult services (palliative care, radiation oncology, and wound care), and pain medications (opioids and nonsteroidal anti-inflammatory drugs [NSAIDs]) prescribed during the study period. Additional data, including number of hospitalizations and number of emergency department (ED) visits during the study period, were also analyzed. These data were chosen for inclusion in the original data set because they were contained within discreet fields in the EMR that could be readily accessed.

For the purposes of this study, we examined the documentation of a palliative care consultation orders in the inpatient EMR. The palliative care team consists of an interdisciplinary team of board-certified palliative care physicians, nurse practitioners, and other ancillary staff. Palliative care referral refers to documentation in the EMR of an order for outpatient palliative care follow-up. Unfortunately, based on the data obtained, we were not able to identify what provider placed the orders for inpatient palliative care consultation or outpatient referral. We were also not able to identify whether the consult or referral was actually received based on these data. The study was approved by the [withheld] institutional review board, and all research personnel completed human subjects protection, health insurance portability and accountability act research, and conflicts of interest training.

Statistical Analysis

Descriptive statistics including the χ² test and t test were used to characterize the baseline demographic characteristics of the sample. Patient and clinical characteristics associated with inpatient palliative care consultation on univariate analysis at the P < .20 level were entered into a stepwise multivariable logistic regression model to identify independent predictors of inpatient palliative care consultation, and significance levels of P < .05 were used for all statistical comparisons. All analyses were conducted using SAS version 9.4 (SAS Institute Inc Cary, North Carolina).

Results

Patient Characteristics

There were a total of 979 patients hospitalized with breast (20.8%), lung (23.0%), or colorectal (56.2%) cancer in 2010. By race and ethnicity, 43.3% were non-Hispanic black, 25.7% were non-Hispanic white, and 26.6% were Hispanic, whereas the overall [Withheld] population is 41% non-Hispanic black, 20% non-Hispanic white, and 32% Hispanic. The majority of patients were under the age of 65 (78.5%), with an age range of 19 to 90 years and mean age of 56.4 years. The greatest percentage of patients were single (35.1%), 29.6% were married, and 11.2% were widowed. A large number of patients were Medicare recipients (42.1%), and 17.2% of the cohort had Medicaid. Of note, 35.6% of patients received charity care or were considered self-pay. Overall, 158 (16.1%) of these hospitalized patients with cancer had documentation of an inpatient palliative care consult order. These and other baseline characteristics can be found in Table 1.

Table 1.

Characteristics of Hospitalized Patients With Breast, Lung, and Colorectal Cancer.

Patient Characteristics	All, N = 979 (%)	Inpatient Palliative Care Consult, N = 158 (%)	No Palliative Care Consult, N = 821 (%)	P Value	OR (95% CI)
Age ≥ 65 (%)	21.5	17.7	22.2	.21	0.76 (0.49, 1.17)
Gender, Male (%)	45.7	51.3	44.6	.12	1.13 (0.93, 1.84)
Race/ethnicity				.11
Non-Hispanic White	25.7	27.9	25.3		Ref
Non-Hispanic Black	43.3	48.7	42.3		1.05 (0.70, 1.58)
Hispanic	26.6	21.5	27.5		0.71 (0.44, 1.16)
Other	4.4	1.9	4.9		0.35 (0.10, 1.20)
Marital status				.07
Married	29.6	26.0	30.3		Ref
Single	35.1	41.1	34.0		1.41 (0.92, 2.17)
Widowed	11.2	5.7	12.3		0.54 (0.25, 1.15)
Divorced/separated	21.1	24.1	20.6		1.37 (0.84, 2.21)
Other	2.8	3.2	2.7		1.38 (0.49, 3.85)
Insurance status				<.001
Medicare	42.1	24.7	45.4		Ref
Medicaid	17.2	25.3	15.6		2.99 (1.84, 4.85)
Charity/self-pay	35.6	45.6	33.6		2.49 (1.64, 3.80)
Commercial/other	5.2	4.4	5.4		1.52 (0.64, 3.61)
Cancer type				<.001
Colorectal	56.2	26.0	62.0		Ref
Lung	23.0	51.3	17.5		6.98 (4.59, 10.61)
Breast	20.8	22.8	20.5		2.66 (1.65, 4.30)
Metastatic disease	56.0	95.6	48.4	<.001	23.04 (10.67, 49.76)
Albumin < 2.5 g/dL	16.24	42.1	11.2	<.001	5.83 (3.98, 8.56)
Radiation oncology consult	14.4	47.5	8.0	<.001	10.34 (6.92, 15.44)
Wound care consult	12.8	23.4	10.7	<.001	2.55 (1.66, 3.91)
Outpatient palliative care referral	10.6	58.7	41.4	<.001	11.38 (7.30, 17.73)
Number of opioids ordered during				<.001
0–10	72.5	29.8	80.8		Ref
11–20	14.9	27.2	12.6		5.89 (3.71, 9.35)
>20	12.6	43.0	6.7		17.44 (10.98, 27.70)
Number of NSAIDS ordered				<.001
None	81.3	55.1	86.4		Ref
One	8.8	14.6	7.7		2.98 (1.76, 5.04)
Two or more	9.9	30.4	6.0		7.98 (5.06, 12.60)
Hospitalizations				.43
None	93.3	91.8	93.5		Ref
One	5.2	5.7	5.1		1.11 (0.53, 2.32)
Two or more	1.5	2.5	1.3		2.12 (0.66, 6.85)
ED Visits				.01
None	59.8	49.4	61.8		Ref
One	23.1	26.0	22.5		1.44 (0.95, 2.18)
Two or more	17.2	24.7	15.7		1.97 (1.28, 3.02)
Died during study period	30.4	55.7	25.6	<.001	3.66 (2.58 (5.20)

Abbreviations: CI, confidence interval; ED, emergency department; NSAID, nonsteroidal anti-inflammatory drugs; OR, odds ratio.

Univariate Predictors of Palliative care Consults

Our univariate analyses revealed that inpatient palliative care consultation orders were associated with Medicaid (odds ratio [OR]: 2.99, 95% confidence interval [CI]: 1.84-4.85) or charity/self-pay status (OR: 2.49, 95% CI: 1.64-3.80), and having an albumin <2.5 g/dL (OR: 5.83, 95% CI: 3.98-8.56). Documentation of an order for radiation oncology consult (OR: 10.34, 95% CI: 6.92, 15.44) or wound care consult (OR: 2.55, 95% CI: 1.66, 3.91) were also statistically significant predictors of placement of inpatient palliative care consult order. Patients who had prescription orders for more than 10 opioids during the study period (11-20 opioids: OR: 2.98, 95% CI: 1.76-5.04) and more than 20 opioids (OR: 17.44, 95% CI: 10.98-27.70) were more likely to have an order for inpatient palliative care consult. Additional statistically significant predictors of orders for inpatient palliative care consultation orders are noted in Table 1, including diagnosis of lung cancer and breast cancer. Of note, there were no statistically significant associations for age, gender, race/ethnicity, marital status, or number of prior hospitalizations related to inpatient palliative care consultation orders on univariate analyses.

Multivariate Predictors of Inpatient Palliative Care Consultation

The following patient and clinical characteristics were included in our multivariable model: age, gender, race/ethnicity, insurance status, number of ED visits, number of opioids, number of NSAIDs, radiation oncology consult, wound care consult, outpatient palliative care referral, cancer type, albumin <2.5 g/dL, and documentation of metastatic disease. After adjustment, age 65 years and older became a statistically significant predictor of inpatient palliative care consultation orders (adjusted odds ratio [AOR]: 2.23, 95% CI: 1.14-4.37). Charity/self-pay status remained a statistically significant predictor (AOR: 2.21, 95% CI: 1.14-4.28), as did documentation of an order for radiation oncology consult (AOR: 2.60, 95% CI: 1.50-4.49) and documentation of an order for a wound care consult (AOR: 2.34, 95% CI: 1.24-4.41). Patients who had documentation of prescription orders for more than 10 opioids (11-20 and >20) and 2 or more NSAIDs during the study period were still more likely to have an order palliative care consultation after adjusted analysis; however, those who had 2 or more ED visits during the study period were less likely to have an order inpatient palliative care consultation order (AOR: 0.46, 95% CI: 0.24-0.88). Diagnoses of lung or breast cancer were again statistically significant findings when compared to colon cancer. Please see Table 2 for all statistically significant multivariable predictors of placement of inpatient palliative care consultation orders.

Table 2.

Multivariable Predictors of Inpatient Palliative Care Consultation.

Predictor	AOR (95% CI)a
Age ≥ 65	2.23 (1.14, 4.37)
Insurance status
Medicare	Ref
Charity/self-pay	2.21 (1.14, 4.28)
Commercial/other	1.70 (0.54, 5.35)
Medicaid	1.52 (0.73, 3.16)
Cancer type
Colorectal	Ref
Lung	2.37 (1.26, 4.44)
Breast	2.12 (1.00, 4.50)
Metastatic disease	17.12 (6.50, 45.09)
Albumin < 2.5 g/dL	2.75 (1.61, 4.71)
Radiation oncology consult	2.60 (1.50, 4.49)
Wound consult	2.34 (1.24, 4.41)
Outpatient palliative care referral	3.55 (1.99, 6.34)
Number of opioids ordered
0–10	Ref
10–20	2.76 (1.53, 5.00)
>20	9.10 (4.62, 17.95)
Number of NSAIDs ordered
None	Ref
One	1.33 (0.65, 2.73)
Two or more	2.42 (1.22, 4.77)
ED visits
None	Ref
One	1.13 (0.63, 2.00)
Two or more	0.46 (0.24, 0.88)

Abbreviation: NSAID, nonsteroidal anti-inflammatory drugs.

^aAdjusted for aged ≥65, race/ethnicity, gender, insurance status, type of cancer, metastatic disease, NSAIDs use, opioid use, albumin level, and referral to outpatient palliative care, radiation oncology consult, wound care consult, and number of ED visits during the study period. Bold values are the adjusted odds ratios that are statistically significant.

Discussion

Palliative care is an important component of the care continuum for patients with serious, chronic, and oftentimes life-limiting illness. Research has repeatedly shown that palliative care may help in improving the quality of life of patients, and in some instances, early palliative care may have an effect on survival.^10–12 Palliative care programs within urban and safety net hospitals are also continuing to emerge, and research suggests that these programs are feasible, acceptable, and have an influence on certain outcomes, including completion of DNR orders and acceptance of hospice among groups who have underutilized this service for a variety of reasons.^4,5,13,14 Our goal in conducting this research is to add to this growing body of literature that examines palliative and end-of-life care among patients who rely on safety net health systems for their care.

In our study, we found that 95.6% of patients with metastatic disease had an order placed for inpatient palliative care consultation, though only 23.4% of patients who had wound care, 42.1% of patients with documentation of albumin of <2.5 g/dL, and 47.5% of those who had documentation of radiation oncology consult had an order for inpatient palliative care consultation. Low albumin level has been associated with increased mortality in persons with a variety of disease states,¹⁵ and our previous research has shown that patients with cancer having metastatic disease and low albumin have poorer prognosis.¹⁶ Radiation oncology consultation is often obtained in the advanced stages of the cancer trajectory to assist with the management of symptoms related to disease. The need for wound care may reflect declines in functional status and development of pressure ulcers, or it may signify the progression of disease in the form of cancer-related wounds that may also indicate a need for pain management.¹⁷ These predictors should certainly spur consideration of initiation of palliative care referral for symptom management and assistance with transition to end-of-life care when appropriate.

Our study examined multivariable predictors associated with placement of inpatient palliative care consult orders among patients with breast, lung, and colorectal cancer admitted to a local safety net hospital, and it has revealed some interesting findings. Our analysis revealed that patients who had charity or self-pay status were more likely to have an order for inpatient palliative care consultation when compared to Medicare beneficiaries. Our previous work has shown that of palliative care patients who received their care in a safety net hospital system, those who did not have insurance were less likely to receive hospice care.⁴ It may be that in our study population, those patients with advanced cancer who have Medicare funding are more often transitioned to hospice care and do not require palliative care consultation. Patients who are uninsured often must rely on palliative care until charity hospice services are available, and this may not be possible until they reach the very end stages of their disease trajectories.

Patient and clinical characteristics that suggest advanced disease or increased symptom burden were also found to be positive predictors of orders for inpatient palliative care consultation. These predictors included documentation of metastatic disease, low albumin, documentation of a radiation oncology consult order, and documentation of a wound care consult order. Additionally, patients who required more opioids or NSAIDs, which may suggest increased need for pain and symptom management, were also more likely to have inpatient palliative care consult orders placed. All of the positive predictors are markers of more advanced disease, need for pain and symptom management, or in some instances of poorer prognosis; consequently, palliative care consultation for assistance with the management of symptoms and addressing goals of care would be beneficial. Although these characteristics proved to be positive predictors of palliative care consultation orders being placed, our analysis of baseline characteristics suggests that the need for palliative care involvement may still be present. These findings do correlate with other studies that have identified similar predictors of palliative care consultation, including number of admissions, recurrent disease, and need for symptom management.¹⁸ Some would argue that increased utilization of ED services suggests increased need for pain and symptom management; however, our study showed that having 2 or more ED visits was a negative predictor of inpatient palliative care consult orders. Some research suggests that persons of low socioeconomic status may utilize the ED for care that could otherwise be managed in the outpatient setting.¹⁹ This may stem from perceptions that ambulatory care is less accessible and more costly.²⁰

There are some limitations that should be taken into account in this study. This research was conducted at 1 hospital system, and only certain cancer diagnoses were examined. We performed a secondary analysis of data extracted from discrete fields within the EMR and diagnoses identified by ICD-9 codes; consequently, we could not obtain data on staging or prior treatment that can affect prognosis. We did, however, include other factors that often signify advanced disease, poor prognosis, and needs for pain management in our analyses. Though these factors may challenge the external validity of the study in certain contexts, we believe that this study of more than 900 patients with breast, lung, and colorectal cancer adds to a modest body of literature that examines the use of palliative care among those who receive their care in safety net hospital settings. Finally, though we do not have data on who obtained the palliative care consult or whether palliative care consultation was offered but refused by patients and/or family members, our previous research suggests that patients and families—particularly those who are underrepresented and underserved—may not be familiar with the concept of palliative care.²¹ Future research should compare these findings to other health-care systems, identify and examine the role of the consulting provider, and further address the lack of knowledge about palliative care in the underrepresented and underserved.

Despite these limitations, we believe that certain findings should be taken into account. As seen in similar research, and as one would expect, inpatient palliative care consultation within our hospital system is largely for those with advanced disease, who require symptom management, and oftentimes have poor prognosis. Challenges within a safety net hospital system may make initiation of early palliative care more difficult; however, efforts are underway to identify patients who are in need and to provide them with education about their options for advance care planning and palliative care.¹⁶ Finally, future research efforts should target patients with other clinical conditions for which palliative care is underutilized.