Impact of transfer from non-acute care centers on clinical outcomes in patients with congestive heart failure

Study objective To compare the clinical outcomes in patients with congestive heart failure who are transferred to an acute care hospital from non-acute care centers with patients who are admitted as regular hospital admissions. Design This was a retrospective cohort study. Setting We utilized the National Inpatient Sample database from 2016 to 2018. Participants Our cohort consisted of hospitalized patients who were at least 18 years old with a primary diagnosis of congestive heart failure. Interventions These patients were either transferred from non-acute centers or presented as regular hospital admissions. Main outcome measurements We matched patients in a greedy nearest neighbor 1:1 model with caliper set at 0.2. Multivariable logistic regression, adjusted for age, sex, race and comorbidities, was used to compare mortality in our matched cohort. Results This study included 35,010 non-acute care transfers and 951,189 regularly admitted patients. Compared to patients who were not transferred, non-acute care transfers were older, predominantly female, White and less racially diverse. After matching, there were 6689 patients in each cohort. When adjusted for age, race, sex and comorbidities, non-acute care transfers with congestive heart failure had 2.20 times higher odds of suffering in-hospital mortality compared to regular, non-transferred admissions (aOR 2.20, 95 % CI: 1.85–2.61; p < 0.001). Conclusion Our findings illustrate that non-acute care transfers are a vulnerable population that require additional medical support in the acute care setting.


Introduction
Non-acute care is an umbrella term that encompasses numerous subsets of healthcare.It includes assisted-living centers, intermediate care facilities and outpatient physician clinics [1,2].Non-acute care facilities are popular across the United States for numerous reasons.Nursing homes and assisted living centers help support our older population [3,4].Intermediate care centers, such as urgent care clinics, are becoming more prevalent due to ease of access and increasing healthcare demand [1,5,6].Non-acute care facilities help treat patients that do not require intensive medical management.In theory, this allows for a more focused distribution of hospital resources towards acutely ill patients.This decreases emergency department burden and also reduces financial costs for both our patients and healthcare system [7,8].As the number of non-acute care centers continues to expand to meet the increasing healthcare demand, it is imperative that we understand these long-term ramifications on our healthcare system [3,9].Only a few studies have compared patients who are transferred to an acute care hospital from non-acute care centers with patients who are admitted as regular hospital admissions.A prior study by Elkaryoni et al. suggests that patients with an ST-elevation myocardial infarction (STEMI) who are transferred from a non-acute care center have higher odds of inhospital mortality than patients who are non-transferred STEMI admissions [2].To our knowledge, no study yet exists that conducts a retrospective analysis on the outcomes of non-acute care transfers with congestive heart failure (CHF).Thus, in this study, we aim to explore the short-term prognostic outcomes of patients who are transferred to acute care hospitals from non-acute care centers with congestive heart failure.

Methods
This was a retrospective cohort study comparing adult patients who are transferred to the hospital from non-acute centers with patients who are non-transferred hospital admissions between 2016 and 2018.Patients were selected from the National Inpatient Sample (NIS) database, which is the largest publicly available all-payer medical sample in the United States.The NIS provides a 20 % probability sample of non-federal hospital discharges.All discharges from these hospitals are recorded and weighted to ensure that they are representative of the national demographic [10].Each identified discharge record includes one primary diagnosis and up to 29 secondary diagnoses, using the International Classification of Diseases, Tenth Edition, Clinical Modification (ICD-10-CM).We chose 2016 as the beginning of the study period as this was the first full calendar year for ICD-10-CM code usage.
Our cohort consisted of hospitalized patients 18 years or older.We identified patients who were either transferred from non-acute centers or not transferred at all using the NIS data element "TRAN_IN."Using ICD-10-CM codes, we then identified hospitalizations with a primary diagnosis of congestive heart failure.The specific ICD-10-CM codes that were included are listed in Appendix A. The primary outcome was the inhospital mortality rate among congestive heart failure patients transferred from non-acute care centers compared to patients admitted as regular hospital admissions.The secondary outcomes included mortality by race, length of stay and total cost to the hospital.
In Multivariable logistic regression, adjusted for age, sex, race and comorbidities, was used to compare mortality in our matched cohort.The comorbidities adjusted for included coronary artery disease, chronic kidney disease, chronic obstructive pulmonary disease, hypertension, obesity, prior myocardial infarction and type 2 diabetes mellitus.This was performed to eliminate confounding effects.The final effect size is reported as an odds ratio (OR) for binary variables and median with interquartile range (IQR) for continuous variables.The analysis was two-tailed with the threshold for significance set at p < 0.05.All analyses were performed using STATA Version 17.

Results
Among the 986,199 patients who met our inclusion criteria, 35,010 (3.5 %) were transferred from non-acute care centers (Table 1).In our unmatched cohort, the non-acute care transfers were older (77.2 years vs 71.2 years, p < 0.001), predominantly female (53.3 % vs 47.1 %, p < 0.001) and White (80.7 % vs 69.0 %, p < 0.001, Table 1).Non-acute care transfers had more patients in the highest APR-DRG Mortality score category (27.7 % vs 16.5 %, p < 0.001).The APR-DRG Mortality score is calculated from discharge billing codes and based on discharge diagnosis, pre-existing medical conditions and age.Non-acute care transfers were also more likely to have Medicare (83.0 % vs 72.2 %, p < 0.001) and be treated in small bed sized hospitals (25.3 % vs 21.1 %, p = 0.002, Table 1).In terms of comorbidities, non-acute care transfers had higher rates of chronic obstructive pulmonary disease (19.6 % vs 19.0 %, p = 0.002, Table 1), but decreased rates of coronary artery disease, chronic kidney disease, hypertension, obesity, prior myocardial infarction and type 2 diabetes mellitus (Table 1).
After nearest neighbor propensity-score matching, 6689 patients were in each cohort.Both cohorts had similar mean age, sex, median household income, hospital bed size, hospital region and hospital location distributions (Table 2).When adjusted for age, race, sex and comorbidities, non-acute care transfers with congestive heart failure had 2.20 times higher odds of suffering in-hospital mortality compared to regular, non-transferred admissions (aOR 2.20, 95 % CI: 1.85-2.61;p < 0.001, Fig. 1).When separated by diagnosis, non-acute care transfers still had a higher likelihood of mortality as well.Non-acute care transfers with heart failure with preserved ejection fraction (HFpEF) had 2.08 times higher odds of suffering in-hospital mortality compared to regular admissions (aOR 2.08, 95 % CI: 1.59-2.73;p < 0.001, Fig. 1).Non-acute care transfers with heart failure with reduced ejection fraction (HFrEF) had 2.48 times higher odds of suffering in-hospital mortality compared to regular admissions (aOR 2.48, 95 % CI: 1.90-3.24;p < 0.001, Fig. 1).

Discussion
In our nationally representative sample of inpatient hospitalizations across the United States, non-acute care transfers with congestive heart failure had higher odds of suffering in-hospital mortality compared to regular, non-transferred hospital admissions.Even when separated by race, patients transferred from non-acute care centers had higher odds of inpatient mortality.Non-acute care transfers were also more likely to have acute kidney injury, atrial fibrillation and cardiogenic shock.Our findings are highly generalizable given the large, heterogeneous population that was studied.There are some plausible theories as to why congestive heart failure patients transferred from non-acute care centers are more likely to suffer in-hospital mortality.One possible explanation is that non-acute care facilities, such as urgent care centers and nursing homes, are not as wellequipped with the same resources, (imaging modalities, medications, specialized staff, etc.), as hospitals are for managing acutely ill patients [1,7,8].The lack of these resources can cause delays in diagnosis, leading to adverse outcomes.
The location of a non-acute care center can also impact a patient's outcome.According to the Urgent Care Association's 2018 Benchmarking Report, only 6.7 % of urgent care centers can be found in rural areas [1,5].Furthermore, 138 rural hospitals have closed since 2010 and in February 2019, 673 additional rural hospitals were at risk of closing as well [5,11].This additional time it takes for a patient to find a healthcare provider or be transferred to an acute care center from an underserved, rural location can also negatively impact a patient's outcome [12][13][14].
Another alternative explanation is that some patients are using nonacute care centers as their primary source of healthcare.In the Urgent Care Association's 2018 Benchmarking Report, it was reported that 35 % of urgent care patients are unaffiliated with a primary care provider or medical home [15].While urgent care centers have their own utility within our medical system, it is important for patients to have regular follow-up with a primary care provider for close management of their chronic illnesses.
Additionally, the shortage of physicians and nurses is another contributing factor.A 2021 survey conducted by the American Health Care Association found that 99 % of nursing homes and 96 % of assisted living facilities endorsed facing staffing shortages [16].Unfortunately, this issue does not seem to be rectified soon.The Association of American Medical Colleges predicts a shortage of between 37,800 and 124,000 physicians in the United States by 2034 [17].This high patientto-provider ratio allows for less time to be spent with each patient, increasing the chances for medical errors to occur.Furthermore, along with this shortage of physicians, our older population continues to grow.The Population Reference Bureau predicts that by 2060, there will be 95 million Americans ages 65 and older, nearly double from 2018 [18].As our population's average life expectancy increases, chronic diseases will need closer follow-up and management.We must continue to find innovative solutions to address this shortage of healthcare providers.
Given that non-acute care centers are also a staple of medical systems in other countries, this study is highly generalizable on a global scale as well.For instance, the Market Research Report from March 2018 projected that the global urgent care center market will grow from 19.2 billion dollars in 2018 to 25.9 billion dollars in 2023 [19].While North America accounts for most of this development, Europe and Asia are also areas of growth as well.Furthermore, based on a United Nations study from 2017, Asia is expected to see a twofold increase in the number of individuals over the age of 60 from 549 million in 2017 to 1.3 billion in 2050 [20].Thus, integrated healthcare systems, which utilize non-acute care centers, are becoming increasingly popular [21].This study sheds light at both a national and international level on the importance of nonacute care centers and how vital they are to the health and well-being of elderly populations.
Our findings illustrate that non-acute care transfers are a vulnerable population that require additional medical support in the acute care  setting.This study highlights the need for further health policy discussion, specifically pertaining to the design and structure of these nonacute care centers.Furthermore, an increased emphasis on medical resource reallocation to help improve accessibility and the quality of healthcare received by this vulnerable population should also be considered.
Our study has some limitations.Firstly, this was a retrospective cohort study done by using an administrative database, which limits the uniformity of data collection and has the potential for misclassification.When using a database such as NIS, there is an inherent risk for miscoding and under coding as it was not possible for us to determine the diagnostic criteria used at each hospital.Data such as the left ventricular ejection fraction, medications administered, readmission rate and outcomes after discharge was also not available.Furthermore, out of the 35,010 non-acute care transfer patients in our dataset, only 6689 (19.1 %) were included in our final analysis.This causes a selection bias and potentially affects the generalizability of our results as these patients were chosen from the calculated propensity scores.Secondly, in the prematched data in Table 1, the non-acute care transfers tended to have higher APR-DRG Mortality scores.Thus, it is possible that the non-acute care centers worked as triage centers and selectively transferred sicker patients, while taking care of less sick patients locally.This also creates the potential for selection bias.Propensity score matching helped overcome this possible issue.Further research should be conducted to better understand why non-acute care transfers are more likely to suffer in-hospital mortality and what changes can be made to address this issue moving forward.

Conclusion
Among patients with congestive heart failure, non-acute care transfers had higher odds of suffering in-hospital death compared to regularly admitted, non-transferred patients.Non-acute care transfers with congestive heart failure are a high-risk population that requires close inpatient monitoring.Further discussion focusing on better supporting this vulnerable cohort in the acute care setting and improving the design and structure of these non-acute care centers should be a point of emphasis moving forward.

Fig. 1 .
Fig. 1.Adjusted odds ratios comparing disease mortality in non-acute care transfers and regular admissions.Adjusted odds ratios calculated by multivariable logistic regression, adjusted for age, race, sex and comorbidities.

Table 1
Baseline characteristics of non-acute care transfers and regular hospital admissions before propensity score matching.
A. Gill et al.

Table 2
Baseline characteristics of non-acute care transfers and regular hospital admissions after propensity score matching.
a APR-DRG scores are calculated from discharge billing codes and based on discharge diagnosis, pre-existing medical conditions and age.A. Gill et al.