Authors: Eric M. Lander (1Division of Hematology/Oncology, Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee), Xuanyi Li (1Division of Hematology/Oncology, Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee), Li-Ching Huang (2Department of Biostatistics, Vanderbilt University Medical Center, Nashville, Tennessee), Amanda S. Cass (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Wade T. Iams (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Emily A. Skotte (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Jennifer G. Whisenant (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Robert A. Ramirez (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Sally J. York (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Travis J. Osterman (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Jennifer A. Lewis (1Division of Hematology/Oncology, Department of Medicine, Vanderbilt University Medical Center, Nashville, Tennessee; 3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee; 4Veterans Health Administration-Tennessee Valley Healthcare System, Geriatric Research, Education and Clinical Center (GRECC) and Medicine Services, Nashville, Tennessee; 5Center for Clinical Quality and Implementation Research, Vanderbilt University Medical Center, Nashville, Tennessee.), Christine M. Lovly (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Yu Shyr (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee), Leora Horn (3Vanderbilt-Ingram Cancer Center, Vanderbilt University, Nashville, Tennessee)
Categories: Article
Source: Journal of the National Comprehensive Cancer Network : JNCCN
Authors: Eric M. Lander, Xuanyi Li, Li-Ching Huang, Amanda S. Cass, Wade T. Iams, Emily A. Skotte, Jennifer G. Whisenant, Robert A. Ramirez, Sally J. York, Travis J. Osterman, Jennifer A. Lewis, Christine M. Lovly, Yu Shyr, Leora Horn
More than 50% of patients with lung cancer are admitted to the hospital while receiving treatment, which is a burden to patients and the healthcare system. This study characterizes the risk factors and outcomes of patients with lung cancer who were admitted to the hospital.
A multidisciplinary oncology care team conducted a retrospective medical record review of patients with lung cancer admitted in 2018. Demographics, disease and admission characteristics, and end-of-life care utilization were recorded. Following a multidisciplinary consensus review process, admissions were determined to be either “avoidable” or “unavoidable.” Generalized estimating equation logistic regression models assessed risks and outcomes associated with avoidable admissions.
In all, 319 admissions for 188 patients with a median age of 66 years (IQR, 59–74 years) were included. Cancer-related symptoms accounted for 65% of hospitalizations. Common causes of unavoidable hospitalizations were unexpected disease progression causing symptoms, chronic obstructive pulmonary disease exacerbation, and infection. Of the 47 hospitalizations identified as avoidable (15%), the median overall survival was 1.6 months compared with 9.7 months (hazard ratio, 2.07; 95% CI, 1.34–3.19; P<.001) for unavoidable hospitalizations. Significant reasons for avoidable admissions included cancer-related pain (P = .02), hypervolemia (P = .03), patient desire to initiate hospice services (P = .01), and errors in medication reconciliation or distribution (P<.001). Errors in medication management caused 26% of the avoidable hospitalizations. Of admissions in patients receiving immunotherapy (n = 102) or targeted therapy (n = 44), 9% were due to adverse effects of treatment. Patients receiving immunotherapy and targeted therapy were at similar risk of avoidable hospitalizations compared with patients not receiving treatment (P = .3 and P = .1, respectively).
We found that 15% of hospitalizations among patients with lung cancer were potentially avoidable. Uncontrolled symptoms, delayed implementation of end-of-life care, and errors in medication reconciliation were associated with avoidable inpatient admissions. Symptom management tools, palliative care integration, and medication reconciliations may mitigate hospitalization risk.
Hospitalization results in significant financial cost and diminished quality of life for patients with cancer and their caregivers. An estimated 35% of all patients with cancer experience an unplanned hospitalization within the year of diagnosis.^1^ Inpatient hospitalization is the largest driver of regional spending variation and has only recently fallen to be the second largest driver of cost for cancer care behind systemic therapy.^2–5^ Furthermore, the average cost of hospitalization in patients with cancer is higher (13,800 per stay).^6^
Despite high frequency and financial costs of hospitalization for patients with cancer, inpatient admission is not associated with better outcomes or within patients’ goals of care. Hospital stays in patients with cancer are generally 1.5 days longer and carry a 2-fold higher risk of in-hospital mortality compared with inpatient admissions for other conditions.^6^ Most patients with cancer are admitted for symptom control, which may be managed on an outpatient basis.^7,8^ Additionally, most patients would prefer to spend their remaining days at home rather than in the hospital when faced with a terminal illness.^9,10^
Although some unplanned oncologic hospitalizations, such as for neutropenic fever due to chemotherapy, are harder for clinicians to anticipate and easier for patients to overcome, others may be the easier to predict and/or difficult to overcome. Prior studies of patients with gastrointestinal cancers and acute myeloid leukemia demonstrated that 12% to 27% of inpatient admissions could potentially be avoided if identified risks included age ≥70 years, third-line or greater palliative chemotherapy, oncologist advice to consider hospice, readmission due to previous premature hospital discharge, and failure of timely outpatient follow-up.^11–13^ More than half of patients with lung cancer experience unplanned admissions following their diagnosis, and in 2017, an estimated 122,000 hospitalizations in the United States contained a primary diagnosis of lung cancer with an aggregate cost of $2.3 billion.^6^ Thus, strategies that mitigate potentially avoidable hospitalizations are germane to advancing patient-centered, high-quality care while reducing financial toxicity to patients and the healthcare system.^14–18^ One previous study investigated whether emergency department (ED) visits by patients with lung cancer who were receiving anticancer therapies could be avoided^19^; however, to our knowledge, there is minimal published data on avoidable inpatient admissions in patients with lung cancer.
We aimed to characterize avoidable hospitalizations among patients with lung cancer to inform future strategies aimed at preventing inpatient admissions. Through direct assessment of medical records and a multidisciplinary, consensus-driven review process, we identified clinically relevant risk factors for potentially avoidable hospitalizations (PAHs) and compared disposition and survival outcomes of patients between groups. We hypothesized that patients with PAHs have worse survival outcomes compared with patients with unavoidable hospitalizations.
Each medical record number of a patient admitted to Vanderbilt University Medical Center (VUMC) from January 1, 2018, until December 31, 2018, with an ICD-10-CM diagnosis code of C34 was pulled from the electronic medical record (EMR). A total of 815 hospitalizations were screened (supplemental eFigure 1, available with this article at JNCCN.org). Patients were included if they had established care in the outpatient adult thoracic oncology clinic before they were hospitalized. A hospitalization was excluded if it was for a planned surgery, the patient was treated by an oncologist outside VUMC, the patient was not yet established in an oncology clinic, an incorrect diagnosis code was used, or the patient completed treatment >3 years from the date of admission and the admission was unrelated to malignancy. One patient with mesothelioma was excluded. One reviewer (E.M. Lander) screened all hospitalizations for inclusion and exclusion. After screening, 319 hospitalizations of 188 unique patients were included in the study. This study was approved with exempt status by the VUMC Institutional Review Board.
To perform a comprehensive review and reduce bias, a multidisciplinary team of thoracic oncology providers conducted a multistage, consensus-driven review of each hospitalization. The framework for the review process was based on a previous high-quality study of avoidable hospitalizations in patients with gastrointestinal cancer.^13^ The case review team consisted of 2 medical oncologists (E.M. Lander, L. Horn), 1 hospitalist (E.M. Lander), 1 thoracic oncology nurse practitioner (E.A. Skotte), and 1 outpatient thoracic oncology pharmacist (A.S. Cass). The multidisciplinary team discussed patient and disease variables that may portend an increased risk of hospitalization and then developed a comprehensive assessment tool for each admission (supplemental eAppendix 1) with data stored in REDCap.^20^ Each reviewer then completed specific components of the assessment tool for each demographics (E.M. Lander), disease characteristics (E.A. Skotte, E.M. Lander), disease status at admission (E.A. Skotte, E.M. Lander), end-of-life services (E.M. Lander), admission information (E.M. Lander), and medication reconciliation (A.S. Cass). Reviewers assessed the contents of the EMR, including oncology, palliative care, infusion, ED, and hospital admission notes, as well as imaging and laboratory data. However, only information up to the date of admission and during the admission was considered when determining whether a hospitalization was avoidable.
The determination of PAHs occurred in 2 stages. In the first review stage, 1 physician (E.M. Lander) flagged every potentially avoidable inpatient admission through the standardized assessment tool.^20^ Each hospitalization that was deemed potentially avoidable based on case review and knowledge of best practice then progressed to the second stage, during which the record was independently reviewed by each member of the multidisciplinary team of 4 thoracic oncology care providers (E.M. Lander, A.S. Cass, E.A. Skotte, L. Horn). The multidisciplinary team then convened and voted for a final consensus decision as to whether each hospitalization was avoidable. If no consensus agreement was made, the hospitalization was deemed to be unavoidable. The multidisciplinary team mutually agreed on the primary reason for each PAH.
Patient demographics and disease and treatment characteristics at the time of admission were extracted from the EMR. Demographic and baseline health variables included age, sex, Brief Health Literacy Screen score,^21–23^ insurance type, smoking status, comorbid diseases, and Charlson comorbidity index score.^24–26^ Disease characteristics included tumor histology, metastatic (stage IV) disease present at time of admission, and time in months from date of lung cancer diagnosis until date of admission. The reviewers assessed whether patients were receiving palliative or curative-intent therapy and had received chemotherapy, immunotherapy, targeted therapy, or radiation therapy within the 30 days before hospitalization. Enrollment in a clinical trial at the time of admission was assessed.^27^
Hospitalization characteristics were gleaned from the EMR during the avoidability assessment, and outcome data were obtained after determination of hospitalization avoidability. Inpatient admission data included where the patient was admitted from (ED, transfer from outside hospital, oncology clinic, or infusion center attached to the hospital [open Monday through Friday from 00 AM until 00 PM]). The oncology clinic and infusion center at VUMC were open from Monday through Friday from 00 AM until 00 PM; thus, patients who needed to be admitted outside these hours were admitted from the ED or as transfers from an outside hospital. Other hospitalization data included whether the encounter was a readmission (defined as <30 days from a previous discharge), the categorical reason for admission (eg, symptom management not related to treatment, treatment-related adverse effect, initiation of hospice services, noncancer medical condition), and chief complaint (clinical sign or symptom) causing hospitalization. The timing of palliative care involvement, advice to consider hospice, and hospice enrollment were examined (timing before admission, during admission, within 30 days after discharge, or none of these criteria). The number of medication changes made at discharge and the next clinic visit, and the number of medication errors in the discharge summary medication reconciliation were reviewed. Outcomes included length of stay, discharge destination, and, in cases in which patients had died, the location of death and length of survival measured as the number of days from admission until death. Due to an occasional paucity of EMR data, information regarding date and location of death was found via online review of obituaries for a small number of patients.
Patient and hospitalization characteristics were summarized using medians with interquartile ranges for continuous variables and frequencies with proportions for categorical variables. To evaluate the associations of avoidable hospitalizations with patients’ demographics, clinical characteristics, reasons for hospitalization, clinical outcomes, and anticancer therapies, univariable logistic regression models were used and the odds ratios (ORs) with 95% confidence intervals were reported. The correlations among hospitalizations on the same patients were accounted for by using generalized estimating equations (GEEs) with an exchangeable correlation matrix. The associations between clinical outcomes of days after last discharge and avoidable hospitalization were examined using logistic regression analysis.
To establish the relationship between PAHs and more than one patient’s characteristics, we performed GEE multivariable logistic regression analyses. With the sample of 47 avoidable hospitalizations, we limited the model to include 4 age ≥70 years, intent of therapy (curative vs no active treatment, palliative vs no active treatment), lines of therapy (≥3 vs none to 2), and readmission within 30 days of previous hospital discharge. These variables were preselected unanimously among the multidisciplinary team (E.M. Lander, A.S. Cass, E.A. Skotte, L. Horn) based on clinical considerations.
For the outcome of time to death (defined as the number of months from date of admission to date of death or of last follow-up if alive), Kaplan-Meier estimates and Cox proportional hazards regression analysis were used, and GEE with independence correlation matrix was used to account for correlation among observations within same patients. Two-sided P values ≤.05 were considered statistically significant. All analyses were conducted using R version 4.1 (R Foundation for Statistical Computing).
A total of 319 hospitalizations among 188 unique patients with lung cancer who were admitted from January 1, 2018, through December 31, 2018, were examined (Table 1). Patients were mostly White (87%), had high self-reported health literacy (median score, 13 of 15), and possessed Medicare (65%) or private insurance (22%).
Of all hospitalizations, the median length of stay was 3 days (supplemental eTable 1). Most patients had metastatic cancer and were receiving palliative chemotherapy. In all, 163 admissions (51%) came from the ED, and 72 hospitalizations (23%) were readmissions (within 30 days of prior admission). The most common cause of hospitalization was symptom management (n = 208; 65%) followed by treatment-related adverse effects (n = 65; 20%) and noncancer medical conditions (n = 62; 19%). Pneumonia and/or other infections caused 34% of all hospitalizations. Other symptoms that were among the primary causes of hospitalization included cancer-related pain (n = 40; 13%), acute hypoxemic respiratory failure (n = 39; 12%), failure to thrive (n = 39; 12%), altered mental status (n = 23; 7%), and nausea or vomiting (n = 22; 7%).
Based on a multidisciplinary 2-stage review process, 47 (15%) of 319 hospitalizations were deemed potentially avoidable. During the initial reviewing stage, 52 hospitalizations were identified as avoidable. During the second reviewing stage, 47 hospitalizations (90%) were unanimously selected as PAHs.
None of the patient demographic or disease characteristics were significantly associated with PAHs in univariable analysis (data not shown). Patients were more likely to have an avoidable hospitalization if they were readmitted (OR, 2.26; 95% CI, 1.19–4.28; P =.01). Significant reasons for avoidable admissions included cancer-related pain (OR, 2.58; 95% CI, 1.16–5.76; P =.02), hypervolemia (OR, 3.11; 95% CI, 1.1–8.79; P =.03), and patient desire to initiate hospice services (OR, 6.27; 95% CI, 1.53–25.67; P =.01). A total of 12 hospitalizations (4%) were related to medication errors and thus avoidable, accounting for 26% of PAHs. Most medication errors resulting in hospitalizations were related to incorrectly prescribed medication doses and inadequate medication filling (supplemental eTable 2).
The outcomes of hospitalizations are shown in supplemental eTable 3, and Table 2 shows the associations among end-of-life care utilization, survival, and avoidable hospitalizations. A total of 44 (23%) patients were still surviving at the time of the final analysis. In all, 37 (26%) patients died in the hospital or inpatient palliative care unit. The median survival time of patients with a PAH was lower at 1.64 months (95% CI, 0.99–9.40 months) compared with 9.67 months (95% CI, 5.66–14.70 months) for unavoidable hospitalization (hazard ratio [HR], 2.07; 95% CI, 1.34–3.19; P =.001; Figure 1). Patients were approximately 3 times more likely to be alive at 30 and 60 days from the date of admission if their hospitalization was unavoidable (P =.008 and P =.01, respectively). Patients with PAHs were more likely to enroll in hospice during hospitalization or within 30 days after discharge compared with patients with unavoidable hospitalizations (OR, 2.74; 95% CI, 1.44–5.21; P =.002). There was no difference between groups in the rates of palliative care consultation or frequency of hospice recommendation at any point before hospitalization (P =.8 and P =.7, respectively).
The multivariable analysis investigated clinical variables that were preselected by the multidisciplinary oncology team (Table 3). Readmission remained positively associated with PAHs (OR, 1.98; 95% CI, 1.05–3.72; P = .035), whereas hospitalizations among patients receiving curative-intent therapy tended to be unavoidable compared with those for patients not actively receiving treatment (OR, 0.14; 95% CI, 0.02–1.19; P = .071).
Because a smaller study evaluated whether ED visits in patients with lung cancer were due to medical cancer therapies, this study included a similar subgroup analysis (Table 4).^19^ In 68% of hospitalizations, patients were actively receiving systemic therapy in which 47% of regimens included immunotherapy and 20% included targeted therapy. Patients receiving immunotherapy and targeted therapy were not more likely to have PAHs compared with patients not receiving the treatments (P = .3 and P = .1, respectively), and patients receiving immunotherapy and targeted therapy tended to be hospitalized less often from treatment than patients receiving cytotoxic chemotherapy (P = .09 and P = .10, respectively). Adverse effects from treatment were generally unexpected and resulted in unavoidable hospitalizations (OR, 0.23; 95% CI, 0.07–0.72; P = .01).
Avoiding unnecessary hospitalization is crucial for improving patient quality of life and lessening the burden of cancer care on our health care system. Although PAHs have been studied in patients with gastrointestinal cancers^13^ and acute myeloid leukemia,^12^ our investigation is the largest and first known study to research PAHs and associated outcomes in patients with lung cancer, which remains the leading cause of cancer death in the United States.^28^
We identified several themes in PAHs in patients with lung cancer, including uncontrolled symptoms, medication errors, and lack of end-of-life care integration. Common symptoms provoking hospitalization were pain, volume overload, and intractable vomiting. Readmission was significantly associated with PAHs in univariable and multivariable analyses.
In a review of 173 ED visits among 97 patients who were actively receiving systemic therapy for lung cancer at Stanford Medical Center, Shah and Neal^19^ found that 34% of encounters were potentially preventable or unnecessary—a larger proportion than the 15% of PAHs from our study. ED visits were therapy-related in 21% of patients receiving chemotherapy, 12% of patients receiving immunotherapy, and 2% of patients receiving tyrosine kinase inhibitor (TKI) therapy. Our data yielded similar proportions of patients receiving chemotherapy (19%) and immunotherapy (9%), but a higher proportion of patients receiving TKI therapy (9%) who were admitted due to treatment toxicity. This may be explained by differences in the TKIs prescribed. Limitations in the study by Shah and Neal^19^ include a smaller sample size, no inpatient admission data, no data on patient outcomes, and no multidisciplinary consensus review process. Notably in our cohort, patients receiving immunotherapy and targeted therapy tended to be hospitalized less often from treatment than patients receiving cytotoxic chemotherapy (P =.09 and P =.10, respectively). If validated, this finding may support providers’ decisions to select targeted therapy or immunotherapy over conventional chemotherapy in situations of clinical equipoise to lower the risk of hospitalization.
Through identifying the most common causes of PAHs, interventions to reduce admission may be designed. Patients with lung cancer tend to have numerous other comorbidities that place them at risk for admission.^29^ However, symptoms related to their disease or treatment, such as uncontrolled pain and nausea/vomiting, can be managed in the outpatient setting with closely coordinated care. One clinical trial standardized outpatient symptom management via an intervention of serial computer-based symptom monitoring that was incorporated into routine clinic visits. Patients in the intervention group had improved health-related quality of life and fewer ED visits compared with usual care, and the intervention tended to improve the rate of hospitalization (45% vs 49%; P = .08).^30^ Health systems could integrate risk scores into the EMR to identify patients at high risk of acute hospitalization or develop new loci for urgent cancer care.^31,32^
Our study is unique in finding that medication errors constituted 26% of all PAHs and that most errors were related to issues with outpatient timeliness in filling medications or improper doses being prescribed. It may benefit practices to routinely conduct a pharmacist-driven medication reconciliation at a patient’s first posthospital clinic visit to reduce errors in medication dosing and filling.^33,34^ Oncology pharmacists improve patient symptom management, reduce drug-related problems, and promote cost savings.^35–37^ Four patients experienced hospitalizations that could have been avoided if their chronic narcotic prescriptions had been refilled sooner upon request. Although prescriptions should not automatically be refilled in most cases, medical record automation for patients with cancer-related pain, which flags the narcotic prescriber when the patient is nearing time for refill, may prompt providers and staff to follow up with patients before they run out of medications.
The end-of-life outcomes in our study reinforce that earlier end-of-life care integration is needed to prevent PAHs. Patients with PAHs had a lower median overall survival than those with unavoidable hospitalizations (1.6 vs 9.7 months, respectively; P = .001), and only 28% of patients with PAHs were alive >60 days from time of admission. More than 1 of 3 patients with PAHs enrolled in hospice during the admission or within 30 days of discharge. Before inpatient admission, however, palliative care was not involved, and hospice was not discussed more often in patients with PAHs. Similar to findings of previous clinical trials,^38–42^ we found that many hospitalizations could have been avoided with earlier integration of palliative care into the patient’s care.
Strengths of our study include the interdisciplinary approach, which enabled further investigation into medication errors as a cause of PAHs. We found that improper medication filling and dosing was a major contributor to PAHs, which was not evaluated in previous studies. Furthermore, our study is the first to identify risk factors and outcomes of PAHs in patients with lung cancer. Although PAHs in gastrointestinal cancers and acute myeloid leukemia have been examined along with ED visits in patients with lung cancer, our analysis contains the largest number of events and patients across previously published studies.^12,13,19^ We were also able to study PAHs in the era of targeted therapy and immunotherapy.
Limitations of our study included subjectivity in determining whether an admission is avoidable and limited generalizability of a single-center study. We used a 2-tier review process with reviewers that included a hospitalist, medical oncologists, a thoracic oncology pharmacist, and a thoracic oncology nurse practitioner and included only hospitalizations that were unanimously determined to be avoidable to reduce bias. Because our analysis included only hospitalization at a single center, we potentially missed hospitalizations outside our system. It was not feasible to obtain external hospitalization information. The demographic of included patients tended to be White and well-insured with high health literacy according to their Brief Health Literacy Screen score; this does not reflect all practice settings, and it does not sufficiently address disparities in socioeconomic status that may exist at our center.
We studied hospitalization in patients with lung cancer and found that 15% of hospitalizations were potentially avoidable. Hospitalizations may be prevented with more aggressive outpatient symptom management, earlier hospice discussion with at-risk patients, and outpatient pharmacist review of medications after hospital discharge. Patients receiving immunotherapy or targeted therapy were only rarely admitted as a result of adverse effects of treatment, and hospitalization due to therapy-related adverse effects was generally unavoidable.