The Impact on Resource Utilization of Supportive Care Consults on Patients at the University of North Carolina Hospital,2010–2012

Purpose

The purpose of this retrospective study was to evaluate the effect of a supportive care intervention on the combined outcome of emergency department (ED) visits and inpatient admissions (IPs) in patients diagnosed with stage III or IV lung or head and neck cancer, two groups of patients with high symptom burdens.

Background

Palliative care in the United States began as inpatient consultative services. ¹ Increasingly, palliative care has been offered in the outpatient setting with comparable improvement in quality of life, mood, and symptoms for patients and caregivers.^1–8

The outpatient supportive (palliative) care service at the University of North Carolina (UNC; begun in 2008) consists of an oncologic pharmacist with prescriptive authority, an advanced practice nurse, and a physician who is trained in medical oncology and palliative care.^9,10 Patients are referred by the primary oncologist for complex symptom management or goals of care discussions. Patients are seen in either a free-standing clinic, an embedded clinic in oncology, or as consults in one of the oncology clinics. For the years 2010–2011, 188 new patients were seen; 384 follow-up visits were carried out.

Study Characteristics

Patients were of age 18 or older; seen by the supportive care service from January 1, 2010 through December 31, 2012 (one year of follow-up for all patients including those enrolled from December 2011); and had a pathological diagnosis of lung or head and neck cancer that was stage III or IV by the AJCC staging system. ¹¹ Patients had a new diagnosis of cancer or a new diagnosis of relapse or progression of previously known cancer. Patients had their initial contact with a UNC-affiliated oncologic provider who was primarily responsible for ongoing oncologic care that had to occur at the UNC.

Methods

This retrospective observational study was reviewed and approved by the Biomedical IRB at the UNC. Records were reviewed of patients with lung cancer or head and neck cancer enrolled in 2010–2011 who met the mentioned inclusion criteria (Table 1). Lung cancer patients were identified as those with either nonsmall cell lung cancer, small cell lung cancer, or large cell neuroendocrine cancer. Head and neck cancer patients were identified as those with squamous cell carcinoma. Two patients, one with adenoid cystic carcinoma and the other with adenocarcinoma, were excluded. We also looked at a contemporaneous group of patients with the same diagnoses that were referred to the same oncologists but were not seen by the supportive care service. Visits to the ED and IPs were the primary outcome measures of interest.

Data collected included the time of initial contact with the UNC-affiliated oncologist until the earlier of date of relapse (kept in the group only if continued to be followed at UNC) and lost to follow-up after initial contact, date of last known contact, or date of death.

Results

When looking at visits that lead to an IP, those patients with head and neck cancer were much more likely to have an IP stay (p = 0.002) (Table 2). Head and neck cancer patients who did not participate in UNC's supportive care service were much more likely to have an IP stay (p = 0.0004) and lung cancer patients who did participate in UNC's supportive care service were more likely to have an IP stay (p = 0.01). This pattern was also seen when looking at the number of IP stays, with head and neck cancer patients who did not participate in UNC's supportive care service being much more likely to have more IP stays (p = 0.0005) and lung cancer patients who did participate in UNC's supportive care service more likely to have IP stays (p = 0.02). The distance to the hospital did not appear to be a significant factor associated with whether a patient had an IP stay (p = 0.11), but did seem to be associated with whether a patient had multiple IP stays (p = 0.04).

Distance to the ED (as determined by the google maps app and zip codes) was the only variable showing a significant association (p = 0.02); patients who lived closer were more likely to go to the ED. The type of cancer a patient had was borderline significant (p = 0.08), with patients with head and neck cancer more likely to visit the ED. Whether a patient participated in UNC's supportive care service did not appear to be associated with visits to the ED (p = 0.21). Patients with head and neck cancer and those who lived closer were much more likely to go to the ED multiple times (p = 0.045 and p = 0.02, respectively).

Discussion

Reported models of care delivery in the outpatient delivery of palliative care include palliative care providers embedded in a Thoracic Oncology Clinic ⁶ ; a palliative care specialist as well as telephonic follow-up with a nurse or telephonic follow-up only^2,3; and a combination of regular visits to a free standing palliative care clinic and telephonic follow-up by a nurse. ⁷

In our previously reported model, patients are seen in a consultative manner in three venues—consultations in all adult oncology clinics, a free standing clinic, and an embedded palliative care clinic within an oncology clinic.^9,10

Despite these multiple options, we were not able to modify the frequency of ED visits. Visits were correlated with distance to the ED, a finding that has not been described before. Patients with lung cancer and especially those with head and neck cancer are populations of patients with a high symptom burden and often with poor social support, who might be expected to frequently to go to the ED. ¹² Although we did not demonstrate that the intervention overall decreased use of emergency room visits, patients with head and neck cancer showed an increase in emergency room visits and a greater likelihood of IP admissions. For patients with head and neck cancer, the enrollment in a supportive care service was associated with a reduction in IP admissions. For patients with lung cancer, the opposite effect was seen with an increase in IP visits for patients followed by supportive care. The effect of a palliative care intervention varied by tumor type in the study by Temel et al. ¹³ but not in the study by Zimmerman et al. ⁷ Outcome measures included quality of life but not resource utilization.

Neither study looked at a population of patients with head and neck cancer.

The results of enrollment of patients with head and neck cancer in a supportive care service have not been reported to our knowledge.

Previous studies showing an impact on ED visits or IP hospitalizations include a community oncology practice wherein a nurse was embedded in the practice site—IP admissions were reduced by 31% compared with historical data from the practice; 66 ED visits (that were deemed preventable) were avoided. ⁴ In a randomized trial of early palliative care (compared with as requested) embedded in an oncology clinic, Temel et al. demonstrated a reduction in ED visits from 57% to 53%, and admissions went from 77% of patients to 74% for patients with lung cancer. ⁶

Patient preferences in patients on palliative care may determine visits to ED rather than need alone. ¹⁴ A previous meta-analysis of ED visits by patients with cancer in North Carolina showed that those with lung cancer were more likely to be admitted to the hospital. ¹⁵

Conclusions

We did not show a uniform reduction in ED visits and admissions in these two populations. There are several limitations to this study. As noted earlier, this is a retrospective study with a small effective sample size that prohibits the appropriate use of multivariable modeling to adjust for multiple covariates. Larger prospective studies would be needed to address these issues. Patients with head and neck cancer are particularly challenging because of the complexity of their symptoms and frequent lack of support. ¹²

Outpatient palliative care may be able to reduce ED visits and IP admissions but the practice model that has the highest likelihood of doing so has not been fully characterized. We used personal contact with several health professionals. Telephonic follow-up alone has shown variable results.^4,6,8 Studies using personal contact have shown improvements in quality-of-life measures and potentially in resource utilization.^4,6–8 The use of personal contact with health professionals is both costly and requires clinic time; whether such a model is usable in much of the United States is not clear.