Associations of Health Care Utilization and Therapeutic Alliance in Patients with Advanced Cancer

Abstract

Introduction:

Therapeutic alliance (TA), or the extent to which patients feel a sense of caring and trust with their physician, may have an impact on health care utilization. We sought to determine if TA is associated with: (1) emergency department (ED) visits within 30 days of death and (2) hospice enrollment.

Methods and Materials:

This is a secondary analysis of data from a randomized clinical trial. We used restricted cubic splines to assess the relationship between TA scores and health care utilization.

Results:

Six hundred seventy-two patients were enrolled in the study, with 331 (49.3%) dying within 12 months. Patients with higher TA were less likely to have an ED visit in the last 30 days of life, but there was no evidence of a relationship between TA and enrollment in hospice.

Conclusions:

Higher TA was associated with decreased ED visits within 30 days of death. There was no association between TA and rates of hospice enrollment.

Clinical Registration Number: NCT02712229.

Introduction

Healthcare utilization among cancer patients increases significantly in the last six months before death and is associated with decreased quality of life.^1–3 Despite efforts to increase enrollment and access, hospice remains underutilized among patients with advanced cancer.⁴ Given the high rates of health care utilization near the end of life in advanced cancer patients, decreasing end of life emergency department (ED) visits and increasing hospice enrollment rates are commonly used quality metrics.^5–8

Therapeutic alliance (TA) is the extent to which patients feel a sense of mutual understanding, caring, and trust with their oncologist. Recent literature has demonstrated that high TA correlates with improved treatment adherence and improved quality of life among patients with advanced cancer, perhaps because strong patient-provider TA improves patients' abilities to proactively manage their health care and, therefore, leads to less urgent, potentially avoidable care.^9,10 TA also may be associated with decreased time in the intensive care unit at the end of life.¹¹ However, there is a little research on the relationship between TA and ED visits and hospice enrollment in the advanced cancer setting.¹²

In this study of patients with advanced cancer, we aim at determining if TA is associated with: (1) ED visits within 30 days of death, and (2) hospice enrollment. We hypothesized that high TA would be associated with fewer ED visits in the last 30 days of life and higher hospice utilization among those who died during the study.

Methods and Materials

Study design

This is a secondary analysis of CONNECT, a cluster randomized clinical trial of 672 patients with advanced cancer that evaluated the impact of primary palliative care in oncology clinics. The research protocol was approved by the University of Pittsburgh Institutional Review Board (PRO15120154) and was registered with clinicaltrials.gov (NCT02712229).^13,14

Participants

Inclusion criteria were: (1) age ≥21 years old, (2) metastatic solid tumor diagnosis, (3) treating oncologist answered yes to “would not be surprised if patient died in the next year,” (4) ECOG score ≤2, and (5) patient willing to be seen by a participating oncologist at least monthly for visits. Exclusion criteria were: (1) unable to read and respond to questions in English, (2) cognitively impaired or unable to consent to treatment, (3) unable to participate in a baseline interview, and (4) diagnosed with hematologic malignancy.

Measures

TA was measured using The Human Connection (THC) scale at patients' baseline assessment. THC is a validated 16-item Likert-type scale in which patients evaluate their relationship with their oncologist and a higher score indicates a higher TA.¹¹ Before our analysis, we considered if the CONNECT intervention had an impact on TA over time and whether TA scores changed over time within treatment groups. We found that the intervention did not significantly influence TA scores nor did TA scores differ over time (Supplementary Fig. S1). In addition, we did not find an association between intervention group and rates of health care utilization. Therefore, we used patient's baseline TA scores in both the intervention and control groups in our analysis.

As part of the CONNECT study, patients and caregivers were contacted on a monthly basis up to 12 months after study enrollment to assess health care utilization. Study staff also completed medical record review and if there was a discrepancy, the reported utilization was considered the gold standard. ED visits within 30 days of death and hospice enrollment were coded. These metrics are based on National Quality Forum measures for preferred practices in palliative care.⁵

Data analysis

Continuous variables are summarized as mean ± standard deviation; categorical variables are summarized with frequencies and percentages. Restricted cubic splines illustrate the estimated probability of experiencing relevant health care utilization outcomes across the entire range of TA scores. This approach was used rather than linear regression to allow for potential non-linearities in relationships (e.g., the possibility that within a lower range of TA scores there would be an increasing relationship between TA scores and a health care utilization outcome that flattened out in the upper range of TA scores).

For example, an odds ratio of 0.79 at a TA score of 20 means that a patient with a TA score of 30 has 0.79 times the odds of the outcome as a patient with a TA score of 20. The estimated probabilities for each outcome across the full range of TA scores are plotted visually with accompanying 95% confidence intervals. We present results without any covariate adjustment and with adjustment for sex, education, and time since diagnosis. We selected covariates that were expected to influence the relationship between TA and health care utilization outcomes. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC).

Results

Of the 672 patients, nearly half of patients (n = 331, 49.3%) died within 12 months of enrolling in the study. Demographic and clinical characteristics are available in Table 1. Overall, patients reported high levels of TA (mean score: 56.4 ± 7.4). This score was similar among those alive 12 months after enrollment (56.4 ± 7.7) and those who died by within 12 months (56.5 ± 7.2).

Table 1.

Baseline Characteristics (Overall Population and for Those Who Died Within 12 Months)

	All randomized patients	Survived follow-up (12 months)	Died during follow-up (12 months)
# Patients	672	341	331
Age, years (mean ± standard deviation)	69.3 ± 10.2	69.6 ± 10.7	69.0 ± 9.7
Sex
Male	312 (46.4%)	141 (41.3%)	171 (51.7%)
Female	360 (53.6%)	200 (58.7%)	160 (48.3%)
Race
Caucasian/White	632 (94.0%)	323 (94.7%)	309 (93.4%)
African American/Black	33 (4.9%)	14 (4.1%)	19 (5.7%)
Asian	5 (0.7%)	2 (0.6%)	3 (0.9%)
Other	2 (0.3%)	2 (0.6%)	0 (0.0%)
Ethnicity
Latino	9 (1.3%)	3 (0.9%)	6 (1.8%)
Education
High school degree or less	335 (49.9%)	171 (50.1%)	164 (49.5%)
Some college or 2-year degree	201 (29.9%)	97 (28.4%)	104 (31.4%)
4-year degree	88 (13.1%)	48 (14.1%)	40 (12.1%)
Master's degree, professional school, or doctoral degree	41 (6.1%)	20 (5.9%)	21 (6.3%)
Declined to answer	6 (0.9%)	5 (1.5%)	1 (0.3%)
Cancer type
Bladder	15 (2.2%)	3 (0.9%)	12 (3.6%)
Brain	3 (0.4%)	2 (0.6%)	1 (0.3%)
Breast	85 (12.6%)	54 (15.8%)	31 (9.4%)
Colon	69 (10.3%)	32 (9.4%)	37 (11.2%)
Endometrial	6 (0.9%)	3 (0.9%)	3 (0.9%)
Esophageal	34 (5.1%)	14 (4.1%)	20 (6.0%)
Gallbladder	2 (0.3%)	0 (0.0%)	2 (0.6%)
Gastric	17 (2.5%)	8 (2.3%)	9 (2.7%)
Head and neck	13 (1.9%)	3 (0.9%)	10 (3.0%)
Kidney	27 (4.0%)	17 (5.0%)	10 (3.0%)
Liver	8 (1.2%)	2 (0.6%)	6 (1.8%)
Lung	243 (36.2%)	125 (36.7%)	117 (35.3%)
Melanoma	9 (1.3%)	5 (1.5%)	4 (1.2%)
Ovarian	22 (3.3%)	13 (3.8%)	9 (2.7%)
Pancreatic	69 (10.3%)	32 (9.4%)	37 (11.2%)
Prostate	29 (4.3%)	19 (5.6%)	10 (3.0%)
Rectal	11 (1.6%)	5 (1.5%)	6 (1.8%)
Sarcoma	6 (0.9%)	2 (0.6%)	4 (1.2%)
Urethral	2 (0.3%)	1 (0.3%)	1 (0.3%)
Other	2 (0.3%)	1 (0.3%)	2 (0.6%)
Time since diagnosis
<1 month	23 (3.4%)	12 (3.5%)	11 (3.3%)
≥1 month but <6 months	174 (25.9%)	83 (24.3%)	91 (27.5%)
≥6 month but <1 year	96 (14.3%)	45 (13.2%)	51 (15.4%)
≥1 year but <2 years	109 (16.2%)	53 (15.5%)	56 (16.9%)
≥2 years but <5 years	131 (19.5%)	66 (19.4%)	65 (19.6%)
≥5 years	136 (20.2%)	81 (23.8%)	55 (16.6%)
Declined to answer	1 (0.1%)	0 (0.0%)	1 (0.3%)
Time with current oncologist
Less than one month	42 (6.3%)	19 (5.6%)	23 (6.9%)
1–6 months	208 (31.0%)	102 (29.9%)	106 (32.0%)
6 months–1 year	114 (17.0%)	60 (17.6%)	54 (16.3%)
1–2 years	118 (17.6%)	65 (19.1%)	53 (16.0%)
2–5 years	119 (17.7%)	55 (16.1%)	64 (19.3%)
>5 years	67 (10.0%)	37 (10.9%)	30 (9.1%)
Decline to answer	3 (0.4%)	2 (0.6%)	1 (0.3%)
Receiving chemotherapy at study enrollment	463 (68.9%)	226 (66.3%)	237 (71.6%)

During the study, 160 of the 331 patients who died (48.3%) had an ED visit within 30 days of death. The average TA score for patients who did not go to the ED within the last 30 days of life was 57.1 ± 6.4 whereas the score for those who did go to the ED was 55.8 ± 7.9 (p = 0.08 for comparison between groups). Patients with higher baseline TA scores in the unadjusted and adjusted models were less likely to have an ED visit than patients with lower baseline TA scores (Fig. 1A and Table 2).

FIG. 1.

Probability of (A) Emergency Department visit within 30 days of death and (B) Hospice enrollment before death based on therapeutic alliance using restricted cubic splines.

Table 2.

Likelihood of Having an Emergency Department Visit Within 30 Days of Death and Hospice Enrollment Before Death During the 12-Month Study per 10-Point Increase in Therapeutic Alliance

Odds Ratio	Unadjusted		Adjusted*
	Estimate	95% confidence limits	Estimate	95% confidence limits
Likelihood of having emergency department visit within 30 days of death per 10-point increase in TA scores at
TA Score = 20	0.57	0.14–2.35	0.57	0.13–2.44
TA Score = 30	0.58	0.15–2.25	0.57	0.14–2.33
TA Score = 40	0.70	0.42–1.18	0.70	0.41–1.20
TA Score = 50	0.81	0.53–1.24	0.82	0.54–1.26
TA Score = 60	0.82	0.53–1.26	0.83	0.53–1.29
Likelihood of hospice enrollment before death per 10-point increase in TA scores at
TA Score = 20	0.79	0.22–2.83	0.77	0.21–2.81
TA Score = 30	0.79	0.23–2.71	0.78	0.23–2.69
TA Score = 40	0.91	0.57–1.45	0.90	0.57–1.44
TA Score = 50	1.00	0.66–1.52	1.01	0.66–1.54
TA Score = 60	1.01	0.66–1.54	1.01	0.66–1.56

^* Adjusted for sex, education, and time since diagnosis.

Results for splines should be interpreted that a lower odds ratio means an increased likelihood of utilizing health care metric. For example, the odds ratio of 0.57 in the upper left cell suggest that compared with a TA score of 20, a patient with a 10-point increase in TA has an odds ratio of 0.57 for having an emergency department visit in the last 30 days. This may be interpreted as meaning that a patient with a TA score of 30 has 0.57 times the odds of an ED visit within their last 30 days as a patient with a TA score of 20, meaning that higher TA scores are associated with a lower probability of ED visit in the last 30 days of life.

During the study, 166 of the 331 patients who died (50.2%) received hospice care. After adjusting for covariates, we found there was no correlation between TA score and probability of hospice enrollment (Fig. 1B and Table 2). In addition, there was no correlation between TA score and probability of hospice enrollment within three days of death, a National Quality Forum metric.⁵

Conclusion

This secondary analysis of data from a cluster randomized clinical trial of patients with advanced cancer demonstrated that patients with higher TA scores were less likely to have ED visits within the last 30 days of life. In addition, we found that TA score was not associated with hospice enrollment rates.

Consistent with literature from advanced cancer populations, we found that nearly half of patients had an ED visit within 30 days of death.^15–17 Congruent with our hypothesis, we found that patients with higher baseline TA scores had lower rates of ED visits within 30 days of death. It is plausible that patients with higher TA scores have fewer ED visits because they are comfortable communicating with their oncologists to proactively manage complications and concerns as they arise.^18–20

These findings contribute to a growing body of literature associating increased stronger patient-oncologist relationships with improved end-of-life outcomes.^21,22 Future research should focus on better understanding why a higher TA reduces ED visits in dying patients to develop targeted interventions.

We found that TA scores are not associated with hospice enrollment. In fact, there is a slight trend toward lower TA scores being associated with higher rates of hospice enrollment. This is in contrast to our hypothesis that patients with higher TA scores would be more likely to enroll in hospice before death due to strong relationships with their providers and improved coping skills.

This finding may be due to providers wanting to maintain strong relationships with their patients and fearing that recommending hospice may hurt that relationship or be seen as “giving up.”²³ Prior research has found that oncologists miss opportunities to discuss end-of-life care with patients, perhaps due to discomfort discussing death or fear that patients may not respond well to these conversations.²⁴

It is also plausible that, even with a strong oncologist-patient TA, decision making around hospice enrollment is not associated because these decisions rarely occur with active decision making.²⁵ Future research should focus on understanding the longitudinal relationship between patients and oncologists and how it impacts hospice enrollment.^26,27

There are several limitations. Per the National Quality Forum, we opted to use hospice enrollment versus no hospice enrollment as our metric, but we recognize there is not an ideal hospice metric. In addition, our participant population lacked racial/ethnic diversity, which may limit application of our findings to marginalized patients. Future research should focus on assessing TA in a more diverse patient population and determine if concordance or discordance between patient-provider plays a role in TA and health care utilization.

Our findings suggest that TA plays an important role in health care utilization in advanced cancer patients. By placing an emphasis on this relationship, oncologists and health systems may impact end-of-life health care through decreased emergency visits. Future research should focus on determining additional predictors of ED use at end of life and understanding better the role of an oncologist in hospice enrollment.

Funding Information

The following funding sources supported this study: (Schenker) NCI R01CA197103; (Schenker) Palliative Research Center (PaRC) at the University of Pittsburgh; (Schenker) K24 AG070285/AG/NIA; (Schenker) NCI P30 through UPMC Hillman Cancer Center.

Author Disclosure Statement

K.J.S.: Grant support from Sanofi-Pasteur (influenza vaccination) and NICO Corporation (neurosurgical device); R.M.A.: Board of VitalTalk, Royalties from Cambridge University Press and UptoDate; and Y.S.: Royalties from UptoDate and EMMI Solutions.