Quality of Supportive Care for Patients with Advanced Cancer in a VA Medical Center

Abstract

Purpose:

Using the Assessing Symptoms Side Effects and Indicators of Supportive Treatment (ASSIST) quality indicators (QIs), we conducted a comprehensive evaluation of the quality of care provided in our institution to patients diagnosed with advanced cancer in 2006.

Methods:

Patients diagnosed with a Stage IV solid tumor were identified from the hospital's cancer registry. Using data abstracted from medical records, care was assessed using 41 explicit QIs. Mean percent adherence to QIs was calculated overall, as well as across five clinical domains: (1) Pain, (2) Depression and Psychosocial Distress, (3) Dyspnea, (4) Treatment Toxicity, (5) Other Symptoms, and (6) Information and Care Planning.

Results:

The study cohort (n = 118) was almost all male (2% female) and mean age was 65.9 years (standard deviation [SD] 9.9 years). The most common cancers were lung and head and neck cancer (23% each); 17% had prostate cancer; 13% had colorectal cancer; and the rest (24%) had breast, esophageal, stomach, genitourinary, liver/biliary, or pancreas cancer. Patients received 51% (95% confidence interval [CI] 48%–54%) of recommended care. Adherence to recommended care within domains ranged from 38% (95% CI 35%–42%) for Other Symptoms to 79% (95% CI 73%–86%) for Information and Care Planning.

Conclusions:

This study suggests that the quality of supportive care for patients with advanced cancer can be greatly improved. Future efforts should use the ASSIST indicators to evaluate the quality of supportive care in larger and more diverse cohorts of advanced cancer patients.

Introduction

While much of the oncology literature focuses on the prevention and treatment of malignancies, one-third of patients with advanced stage cancer die within 6 months of diagnosis¹ and palliation of symptoms is essential to relieve their suffering.² Systematic reviews identify pain, fatigue, anorexia, and breathlessness among the most troublesome symptoms.^3,4 Good supportive care encompasses four important areas: (1) treatment of cancer-related symptoms, (2) prevention and treatment of the side effects associated with cancer treatment, (3) recognition of and support for patients experiencing psychosocial distress,s and (4) information and care planning for end of life. The Cancer Quality ASSIST (Assessing Symptoms Side Effects and Indicators of Supportive Treatment) quality indicators (QIs) have been shown to provide reliable and valid assessments of the quality of supportive care for patients with cancer.^5–10

In this study, we used the ASSIST QIs to conduct a comprehensive assessment of the quality of supportive care for patients with advanced cancer at the Veterans Affairs Greater Los Angeles Health Care System (VA GLAHS).

Methods

Study sample

Patients diagnosed with Stage IV solid tumors in 2006, alive 30 days post-diagnosis, with at least one outpatient visit were identified from the VA GLAHS cancer registry. Patients who died anytime between 30 days after diagnosis and the end of the study period (June 2008) were included in the “Decedent Cohort.”

Quality indicators

The ASSIST quality indicators were developed following a systematic synthesis of the oncology literature^6–9; a panel of multidisciplinary international leaders from oncology, palliative care, geriatrics, and primary care endorsed those most relevant to patient quality of life using the RAND appropriateness method.⁵ Candidate indicators were evaluated for feasibility, reliability, and validity.¹⁰ The final quality indicator set includes 41 indicators addressing the management of pain, dyspnea, depression, nausea and vomiting, diarrhea, fatigue, and other symptoms related to cancer, common complications and treatment-related toxicities, as well as information and care planning.

Data sources

Trained nurse abstractors collected data from patients' medical record for 12 months starting with the date of treatment initiation for patients who received chemotherapy or radiation or date of diagnosis for patient who did not receive any therapy. Mean kappa statistics for determining patient indicator eligibility and adherence to recommended care were 0.87 and 0.85, respectively.

Analyses

We determined the proportion of eligible patients who received the specified process of care for each QI and then calculated the mean percent adherence overall and for each of five clinical domains: (1) Pain, (2) Depression and Psychosocial Distress, (3) Dyspnea, (4) Treatment Toxicity, (5) Other Symptoms, which included nausea/vomiting, anorexia, fatigue/anemia, and delirium, and (6) Information and Care Planning. In exploratory analyses, we assessed variation in overall and domain quality scores with age, race/ethnicity, marital status, cancer type, comorbidity, treatment received, receipt of palliative care consultation, and cohort (Full Cohort vs. Decedent Cohort) using the χ² test and Fisher's exact test, as appropriate. All statistical analyses were conducted in Stata statistical software version 10.0 (Stata Corporation, College Station, TX).

Results

Patient characteristics

The cohort (n = 118) was almost all male (2% female) and the mean age was 65.9 years (standard deviation [SD] 9.9 years) (Table 1). Sixty-two percent were Caucasian and 31% were African American. Approximately half of the patients died during the study period (January 2006–June 2008). Median survival was 6.6 months (SD 5.4).

Table 1.

Patient Characteristics

	Full cohort N = 118	Decedents only N = 71
Age, mean (SD)	65.9 (9.9)	66.6 (9.8)
Female	2 (2%)	1 (1%)
Race/Ethnicity
Caucasian	73 (62%)	47 (66%)
African American	36 (31%)	20 (28%)
Hispanic	3 (3%)	1 (1%)
Asian	3 (3%)	2 (3%)
Other or unknown	3 (3%)	1 (1%)
Cancer type
Breast	1 (1%)	1 (1%)
Colorectal	15 (13%)	8 (11%)
Esophagus/stomach	10 (8%)	8 (11%)
Genitourinary	8 (7%)	8 (11%)
Head and neck	27 (23%)	11 (15%)
Liver/biliary	4 (3%)	3 (4%)
Lung	27 (23%)	23 (32%)
Pancreas	6 (5%)	5 (7%)
Prostate	20 (17%)	4 (6%)
Brain metastases	17 (14%)	14 (20%)
Died during study period	55 (47%)	100%
Survival, months		6.6 (5.4)
Palliative care consult
Ever	42 (36%)	39 (55%)
Within 90 days of treatment initiation	27 (23%)	24 (34%)

Note: due to rounding, values may not add to 100%.

Eligibility for quality indicators

A total of 1074 eligible events occurred across 118 patients for clinicians to provide care specified by the 41 QIs. The greatest number of eligible events (n = 635) was in the domain Other Symptoms, followed by the domains of Pain (n = 175) and Information and Care Planning (n = 150). The remaining three domains had a combined 114 eligible events.

Receipt of recommended care

Overall, patients received 51% (95% confidence interval [CI] 48%–54%) of recommended care (Table 2). Adherence to recommended care within domains ranged from 38% (95% CI 35%–42%) for Other Symptoms to 79% (95% CI 73%–86%) for Information and Care Planning.

Table 2.

Variation in Adherence to Quality-of-Care Scores by Domain for Full Cohort and Decedents

		Full cohort (n = 118)			Decedents only (n = 71)
	Number of indicators	% patients eligible (range)	# eligible events	Adherence	% patients eligible (range)	# eligible events	% adherence	P-value
Overall	41	0%–100%	1074	0.51 (0.48, 0.54)	0%–100%	778	0.54 (0.50, 0.57)	0.003
Pain	10	1%–90%	175	0.63 (0.56, 0.71)	1%–86%	119	0.56 (0.47, 0.65)	0.004
Depression and psychosocial distress	2	11%–15%	31	0.61 (0.44, 0.78)	14%–18%	23	0.65 (0.43, 0.84)	0.676**
Dyspnea	6	2%–15%	51	0.71 (0.56, 0.83)	1%–25%	46	0.74 (0.59, 0.86)	0.144**
Treatment toxicity	4	2%–21%	32	0.53 (0.35, 0.71)	1%–24%	27	0.52 (0.32, 0.71)	1.000**
Other symptoms	14	2%–100%	635	0.38 (0.35, 0.42)	3%–100%	421	0.40 (0.36, 0.45)	0.155
Information and care planning	5	8%–48%	150	0.79 (0.73, 0.86)	11%–79%	142	0.82 (0.75, 0.88)	0.001**

P-value of χ² tests/Fisher's exact tests(**).

Patient characteristics associated with quality of care

In bivariate analyses (Table 3), quality of care received was associated with cancer type, with patients with prostate cancer being less likely to receive recommended care (32% vs. 52%–55% for other cancer types, p < 0.01). Patients who were married or partnered had significantly higher adherence to QIs addressing pain (78% for married or partnered vs. 58% for all others, p < 0.05). Having a palliative care consult was associated with higher adherence to QIs in the domains of Dyspnea (76% vs. 64%, p < 0.05) and Information and Care Planning (87% vs. 66%, p < 0.01), but worse adherence in the Pain domain (49% vs. 73%, p < 0.01).

Table 3.

Variation in Adherence to Quality-of-Care Scores Associated with Patient Characteristics

	Overall		Pain		Depression		Dyspnea		Treatment toxicity		Other symptoms		Information planning
	# eligible events	% adherence	# eligible events	% adherence	# eligible events	% adherence	# eligible events	% adherence	# eligible events	% adherence	# eligible events	% adherence	# eligible events	% adherence
Age												*
≥ 65	500	48%	83	59%	14	57%	25	76%	19	58%	283	34%	76	78%
< 65	574	53%	92	67%	17	65%	26	65%	13	46%	352	42%	74	81%
Race/Ethnicity				†								*
Caucasian	681	53%	109	69%	23	65%	34	71%	17	47%	397	40%	101	79%
All other race/ethnicity	393	48%	66	55%	8	50%	17	71%	15	60%	238	37%	49	80%
Marital status		*		*
Married/lives with partner	285	57%	50	78%	8	50%	6	50%	7	57%	178	46%	36	83%
All others	789	49%	125	58%	23	57%	45	74%	25	52%	457	36%	114	78%
Cancer type		**								*		*
Head and neck	238	52%	35	71%	12	58%	7	86%	5	60%	153	43%	26	69%
GI and breast	461	52%	73	67%	12	67%	15	60%	17	29%	271	40%	73	82%
Lung	275	55%	44	57%	4	50%	25	68%	9	89%	149	44%	44	80%
Prostate	100	32%	23	53%	3	67%	4	25%	1	100%	62	11%	7	86%
Treatment		***		**								***
Both chemo + radiation	397	55%	61	80%	16	63%	3	67%	12	33%	257	47%	48	73%
Chemo, no radiation	165	61%	23	70%	1	0	14	57%	7	57%	98	56%	22	82%
Radiation, no chemo	183	44%	29	45%	8	38%	17	72%	7	57%	92	24%	29	90%
No treatment	329	44%	62	53%	6	17%	16	81%	6	83%	188	25%	51	78%
Comorbidity								†
ACE category 0	182	50%	25	60%	7	43%	6	83%	6	67%	115	38%	23	83%
1	504	49%	88	65%	12	67%	19	63%	17	29%	300	38%	68	75%
2	170	48%	28	71%	4	75%	10	40%	3	67%	101	36%	24	71%
3	218	58%	34	56%	8	63%	16	94%	6	17%	119	41%	35	91%
Palliative care consult				**				†						***
Ever	483	51%	71	49%	15	53%	29	76%	19	47%	254	36%	95	87%
Within 90 days of Rx initiation	308	49%	45	47%	11	64%	21	75%	13	39%	159	31%	59	88%
None	591	50%	104	73%	16	69%	33	64%	13	62%	381	40%	55	66%

ACE, adult comorbidity evaluation; chemo, chemotherapy; GI, colorectal, esophagus, stomach, liver/biliary, pancreas.

Quality of care for the full cohort vs. decedent cohort

Compared with the full (prospective) cohort, overall adherence to the quality measures was modestly higher for the decedent cohort (54% vs. 51%, p < 0.01), though there was substantial variation across domains (Table 2). Although there were no significant differences in the quality of care in the domains of Depression and Psychosocial Distress, Dyspnea, Treatment Toxicity, and Other Symptoms, adherence to QIs addressing Information and Care Planning was better in the decedent cohort than the full cohort (82% vs. 79%, p < 0.01). Of concern, the quality of care for pain was worse in the decedent cohort than observed for the full cohort (56% vs. 63%, p < 0.01).

Discussion

Using the ASSIST QIs, we conducted a comprehensive evaluation of the quality of care provided in our institution to patients diagnosed with advanced cancer in 2006. Overall, patients received approximately half of recommended supportive care and end-of-life care, although adherence varied substantially across domains, with only 38% of recommended care received for Other Symptoms and 79% received for Information and Care Planning. Although our small sample limits detailed characterization, variation in adherence to quality measures was associated with a number of patient characteristics, notably age, marital status, race, and cancer type. Older patients appeared to have lower adherence to quality measures addressing Pain, Depression, and Other Symptoms. Patients who were unmarried and of nonwhite race experienced worse quality of care for pain. We found that patients with prostate cancer were less likely to receive recommended care. This likely reflects the longer expected survival following a new diagnosis of metastatic prostate cancer compared with other solid tumors in our study. The care specified in the ASSIST quality measures may not be applicable to all newly diagnosed, advanced prostate cancer patients and perhaps eligibility should be limited to patients who have developed hormone refractory prostate cancer. Interestingly, having a palliative care consult was associated with higher adherence to QIs in the domains of Dyspnea and Information and Care Planning (but worse adherence in the Pain domain). However, this study was not designed to study the impact of the palliative care consult, and eligibility for quality indicators may have preceded the involvement of the palliative care service. We are currently conducting a national cohort study to evaluate the effect of palliative care on the quality of supportive care for patients with advanced lung, pancreas, and colorectal cancer in the Veterans Health Administration.

We noted striking differences in quality of supportive care received by advanced cancer patients at the VA GLAHS compared with findings of a concurrent study at Johns Hopkins (JHU); although remarkably the direction and magnitude of these differences vary substantially by domain.¹¹ Patients in the JHU cohort were more likely to have their symptoms assessed. Pain was assessed at 79% of JHU visits compared with just 36% of VA GLAHS visits. Nausea and vomiting were assessed in patients in 45% and 27% of visits at JHU and VA GLAHS, respectively. In contrast, 100% of patients whose death was anticipated at the VA GLAHS had an advanced directive or surrogate decision maker documented, whereas just 19% at JHU did, and, among these same patients, 84% were referred to hospice or palliative care in the VA as compared with 45% at JHU. The JHU cohort included only decedents, and comparing only care for decedents at the two sites further accentuated these differences.

We also compared the quality of care of our “Full Cohort” with a “Decedent Cohort.” Decedent cohorts are an attractive option for evaluating the quality of palliative care because of the challenges inherent in prospectively identifying patients who are at the end of life; however, researchers have raised concerns that looking back at the care received before death by a group of cancer patients who died may produce different results than if a cohort of patients who are likely to die were identified prospectively.^12,13 Comparing the quality of care of a decedent cohort with a cohort identified prospectively from the time of diagnosis of Stage IV cancer, we found that although overall adherence to the quality measures was modestly higher for the decedent cohort, the absolute difference was just 3% (54% vs. 51%) and ranged from 1% to 7% across the various domains. Although researchers should remain cognizant of potential biases with either study design, the ASSIST quality measures appear valid in both prospectively and retrospectively identified cohorts of advanced cancer patients.

Our data must be viewed in light of several limitations. First, this is a nonexperimental cohort study from a single VA medical center. Second, this study uses data abstracted from patients' medical records. To the extent that care is not accurately documented in the medical record, our results may not reflect the care actually provided to patients. This would most likely result in underestimating the quality of medical care.

Our study points to a number of potential targets for quality improvement. Assessment and management of pain and other symptoms as well as treatment toxicity, in particular, appear to be important areas for quality improvement efforts at our institution. And whereas 84% of patients who died an expected death were enrolled in hospice, only 36% of all patients and just over half of patients who died had a palliative care consultation. When these results were presented to our hospital's multidisciplinary Cancer Committee, there was strong support for a quality improvement initiative to increase palliative care consultation in advanced care patients. We are currently pilot testing several approaches to integrate a clinical reminder in our institution's electronic health record to prompt the treating clinician to consider requesting a palliative care consult. Other approaches that warrant further study include using telehealth services to provide palliative care,¹⁴ the use of routine symptom screening tools, such as the Edmonton Symptom Assessment System (ESAS),¹⁵ and home-based symptom monitoring facilitated by telehealth devices.^16,17

In summary, this study suggests that the quality of supportive care for patients with advanced cancer can be greatly improved. Future efforts should use the ASSIST indicators to evaluate the quality of supportive care in larger and more diverse cohorts of advanced cancer patients.

Footnotes

Acknowledgments

The research reported here was supported by the Department of Veterans Affairs, Veterans Health Administration, Health Services Research and Development Service SHP 08-196. The authors are solely responsible for its content and the views do not necessarily represent the views of the Department of Veterans Affairs.

Author Disclosure Statement

No competing financial interests exist.

References

Rose

, O'Toole

, Dawson

, Lawrence

, Gurley

, Thomas

, Hamel

, Cohen

. Perspectives, preferences, care practices, outcomes among older, middle-aged patients with late-stage cancer. J Clin Oncol, 2004; 22:4907–4917.

American Society of Clinical Oncology: Cancer care during the last year of life. J Clin Oncol, 1998; 16:1986–1996.

Solano

, Gomes

, Higginson

. A comparison of symptom prevalence in far advanced cancer, AIDS, heart disease, chronic obstructive pulmonary disease, and renal disease. J Pain Symptom Manage, 2006; 31:58–69.

Teunissen

, Wesker

, Kruitwagen

, de Haes

, Voest

, de Graeff

. Symptom prevalence in patients with incurable cancer: A systematic review. J Pain Symptom Manage, 2007; 34:94–104.

Lorenz

, Dy

, Naiem

, Walling

, Sanati

, Smith

, Shanman

, Roth

, Asch

. Quality measures for supportive cancer care: The Cancer Quality-ASSIST (Assessing Symptoms, Side Effects, and Indicators of Supportive Treatment) Project. J Pain Symptom Manage, 2009; 37:943–964.

, Asch

, Naeim

, Sanati

, Walling

, Lorenz

. Evidence-based recommendations for cancer fatigue, anorexia, depression, dyspnea. J Clin Oncol, 2008; 26:3886–3895.

, Asch

, Naeim

, Sanati

, Walling

, Lorenz

. Evidence-based recommendations for cancer pain management. J Clin Oncol, 2008; 26:3879–3885.

Walling

, Lorenz

, Dy

, Naeim

, Sanati

, Asch

, Wenger

. Evidence-based recommendations for information and care planning in cancer care. J Clin Oncol, 2008; 26:3896–38902.

Naeim

, Dy

, Lorenz

, Sanati

, Walling

, Asch

. Evidence-based recommendations for cancer nausea and vomiting. J Clin Oncol, 2008; 26:3903–3910.

10.

, Lorenz

, ONeill

, Asch

, Walling

, Tisnado

, Antonio

, Malin

. Cancer Quality-ASSIST Supportive Oncology Quality Indicator Set: Feasibility, reliability, and validity testing. Cancer, 2010; 116:3267–3275.

11.

, Asch

, Lorenz

, Weeks

, Kumar

, Wolff

, Malin Jennifer

. quality of end-of-life care for patients with advanced cancer in an academic medical center. Submitted for publication.

12.

Bach

, Deborah Schrag

, Begg

. resurrecting treatment histories of dead patients: A study design that should be laid to rest. JAMA, 2004; 292:2765–2770.

13.

Earle

, Ayanian

. Looking back from death: The value of retrospective studies of end-of-life care. J Clin Oncol, 2006; 24:838–840.

14.

Kidd

, Cayless

, Johnston

, Wengstrom

. Telehealth in palliative care in the UK: A review of the evidence. J Telemed Telecare, 2010; 16:394–402Epub 2010 Sep 2.

15.

Bruera

, Kuehn

, Miller

, Selmser

, Macmillan

. The Edmonton Symptom Assessment System (ESAS): A simple method for the assessment of palliative care patients. J Palliat Care, 1991; 7:6–9.

16.

Maudlin

, Keene

, Kobb

. A road map for the last journey: Home telehealth for holistic end-of-life care. Am J Hosp Palliat Care, 2006; 23:399–403.

17.

Zimmerman

, Barnason

. Use of a telehealth device to deliver a symptom management intervention to cardiac surgical patients. J Cardiovasc Nurs, 2007; 22:32–37.