Introduction
Discharges to skilled nursing facilities (SNFs) and hospital readmissions are a frequent occurrence for Medicare beneficiaries; however, little attention has been paid to the high mortality rate for these patients (Jencks et al. Reference Jencks, Williams and Coleman2009). For Medicare beneficiaries discharged to SNF, the 1-year mortality rate has been reported as 26% and this almost doubled to 48% in 1 study for patients who are readmitted to the hospital from SNF (Hakkarainen et al. Reference Hakkarainen, Arbabi and Willis2016). In contrast, heart failure (HF) and oncology patients have been noted to have substantially higher mortality rates if discharged to SNF even without readmission. In a large study of outcomes related to discharge location in HF patients, unadjusted 1-year mortality rate was 53% if discharge to SNF versus 29% if discharge to home (Allen et al. Reference Allen, Hernandez and Peterson2011). Among older patients with new or established cancer diagnoses, the mortality rates are even higher with 56% of SNF discharges dying within 6 months of initial hospitalization compared to 36% of patients who discharged home. It is assumed that those cancer patients discharged to SNF had greater frailty and poorer functional status which led to reduced oncologic treatment and greater mortality, but the authors were unable to assess frailty factors directly (Singh et al. Reference Singh, Eguchi and Sung-Joon2021).
In this study, we wanted to determine if the high mortality rates previously published are applicable to a general population of Medicare Fee-for-Service patients or if applicable only to certain readmission and disease states. The second objective was to differentiate specific patient characteristics associated with 1-year mortality among the patients who were readmitted from SNF. Our research aims to assist clinicians in prioritizing discussions on goals of care and prognosis by highlighting the observed high mortality rates within our study. Patient subgroups who are most at risk can benefit most from timely and focused conversations regarding their care preference and expected outcomes.
Methods
This study was approved by the Institutional Review Board for Human Subjects at our institution. First, a retrospective study of administrative billing data from an accountable care organization (ACO) cohort of 847 patients with Medicare fee-for-service insurance aged 65 and above, discharged from acute care hospital to SNF in 2019 was completed to analyze readmission data, category of illness, and mortality rates. Data included age, sex, length of stay in hospital and SNF, admission events, Major Diagnostic Category (MDC) of the initial hospital stay, and date of death. Unique patient data were sorted into 4 categories based on readmission and mortality status within 400 (30 + 365) days from hospital discharge (Fig. 1).
Figure 1. Framework of charts reviewed.
Second, an in-depth chart review was completed on the 2 subsets of patients who were readmitted to our system hospitals to understand patient characteristics leading to mortality. This focus was driven by the objective to comprehensively understand the factors that may have contributed to the observed increase in mortality within this specific subgroup. Patients admitted from SNF to outside hospitals were excluded from our analysis due to absence of comprehensive access to full electronic medical records for these patients. We were able to access a total of 24/93 (26%) charts for the group who remained alive and 35/92 (38%) charts for the group who expired within 1 year. A Clinical Frailty Scale (CFS) score was determined from review of physical therapy evaluations on both hospital admission and readmission. The CFS is a tool used to screen for frailty based on fitness and overall functional status and has been validated to assess for frailty using retrospective data review for clinical practice and research (Stille et al. Reference Stille, Temmel and Hepp2020). Other data analyzed included medical comorbidities, history of falls, presence of pressure ulcers, and evidence of familial social support. Data regarding history of falls and risk of falls are determined from the Hester–Davis Fall Risk Scale completed by nurses as part of the admission process (Hester and Davis Reference Hester A and Davis2013). The presence of pressure ulcers is verified based on the comprehensive skin assessment performed on a regular basis. The presence of familial social support is ascertained from progress reports of physicians, social workers, and case managers. Other parameters analyzed included weight loss of more than 10 lb for the past year, body mass index of ≤21 kg/m2, and serum albumin of ≤3.5 g/dL. These nutritional status parameters are similarly used in calculating the frailty index as part of a comprehensive geriatric assessment (Shi et al. Reference Shi S, Olivieri-Mui and McCarthy2021). Standard descriptive statistics were used to compare groups.
Results
The sample included 847 individuals aged 65 and above admitted to SNF after an acute hospital stay. The 1-year overall mortality rate was 28.3% (240/847). The sample was divided into those who had no 30-day hospital readmission (662 patients, 78%) and those who were readmitted within 30-days (185 patients). Of those discharged to SNF with no 30-day hospital readmission, 514 (78%) were alive after a year (400 days) while 22% (148 patients) without readmission expired within a year of hospital discharge. Of those readmitted within 30 days of discharge to SNF, 50% (93 patients) remained alive at 1 year and 50% (92 patients) expired. Readmission to the hospital within 30 days from SNF was associated with a greater than twofold increase in 1-year mortality.
The 1-year mortality rates for readmitted patients varied widely among MDCs ranging from 38% to 61%. Table 1 depicts the mortality rates for the population discharged to SNF by readmission status and MDC. The most common MDCs were diseases of the nervous system (most commonly stroke), circulatory system (including HF), musculoskeletal system (including hip and other fractures), and infectious diseases (including sepsis). Patients with readmission to hospital within 30 days of discharge to SNF had a roughly threefold higher 1-year mortality rate for the more common MDCs. Interestingly, the circulatory group readmission had little impact on mortality.
Table 1. Mortality rates for the population discharged to SNF by readmission status and MDC
CNS = central nervous system; MDC = Major Diagnostic Category; SNF = skilled nursing facility.
* p ≤ 0.05.
The group that was readmitted within our hospital system was then analyzed to understand risk factors associated with 1-year mortality. Table 2 displays patient characteristics of those readmitted to our hospital system who survived compared to those who expired. The readmitted patients within a month of discharge to SNF who expired were more frequently found to have cancer as a comorbidity, weight loss, low body mass index (BMI), low albumin, higher falls risk by Hester–Davis Fall Risk Scale, and were more likely to be male. Upon review of this group’s readmission and emergency department visits, it was observed that the reasons for readmission were predominantly aimed at curative interventions rather than terminal care. This distinction is significant, indicating that despite the high-risk profile and previous admissions, the primary treatment objectives for these patients remained focused on achieving curative outcomes. There was no difference between the groups with and without mortality for history of dementia, HF, chronic kidney disease, diabetes mellitus, stroke, and pressure ulcers. Worsening of frailty status during the SNF stay bode poorly as 83% of the patients readmitted within a month of discharge to SNF presented with worse CFS at readmission and was 5 times more likely to die during the following year.
Table 2. Characteristics of the 2 groups who were readmitted to our hospital system based on mortality status after a year
CFS = Clinical Frailty Scale; CKD = chronic kidney disease; COPD = chronic obstructive pulmonary disease; SNF = skilled nursing facility.
Discussion
Discharge from the hospital to an SNF provides patients with a critical opportunity for functional recovery and improved quality of life. However, this may be a pivotal time for the older adult as the mortality rates in this subpopulation are quite high.
Consistent with the findings reported by Hakkarainen, our analysis of the Medicare fee-for-service patient population revealed an overall mortality rate of 28%. The mortality rate for patients readmitted to the hospital from an SNF was twice as high compared to those not readmitted under similar circumstances. However, we found wide variation in the mortality rates by diagnosis category. One-year mortality rate was 2–3 times higher for patients readmitted from the SNF after strokes, hip fractures, and pneumonias, all very common circumstances.
In the circulatory MDC, which was mainly patients with HF, readmission to the hospital was not associated with markedly higher mortality rates. This group had the highest 1-year mortality rate even without a 30-day readmission, but we and others have shown that patients with HF are very vulnerable to challenges with medication management at home (more so than in SNF where medications are provided and adjusted if necessary) (Agarwal et al. Reference Agarwal, Kazim and Xu2016; Orr et al. Reference Orr, Forman and De Matteis2015). For the “other” diagnoses, reflecting the remaining 18 MDC categories, readmission almost doubled 1-year mortality.
The Centers for Medicare & Medicaid Services included ACO as a quality measure for patients readmitted in the hospital. Incentives can be given to institutions that can help manage care coordination with a goal of decreasing hospital readmission rates. This is further supported by a study done by Chukmaitov et al. (Reference Chukmaitov, Harless and Bazzoli2019) which concluded that there is a decreased rate of preventable hospitalizations for ACO participating hospitals. However, this was only statistically significant for COPD, asthma, and diabetes-related medical conditions.
Patients with markers of frailty, malnutrition, and cancer were associated with a higher 1-year mortality. However, it was worsening of frailty status over the short time in the SNF that defined a group with high mortality. Fully 83% of the patients readmitted within a month of discharge to SNF that presented with worse CFS at readmission expired within 1 year. The worsening CFS while in the SNF was associated with 5 times greater likelihood of dying during the following year. However, our analysis does not enable us to ascertain whether this elevated mortality risk is a direct result of increased vulnerability due to declining health status, or if it reflects the presence of ongoing acute medical processes that both exacerbate physical deterioration and independently raise mortality risk.
Frailty has been clearly associated with increased mortality in the inpatient and community settings. A prospective cohort study of older hospitalized patients with community acquired pneumonia found that patients who met Fried’s frailty criteria had a threefold higher 1-year mortality than those who were not frail (Luo et al. Reference Luo, Tang and Sun2020). But our small sample suggests that the change in CFS while still in a therapeutic environment may provide important information not accessible with a single assessment of status. A CFS score is a simple tool that can be inferred based on the overall functional status of a patient. Physical therapy assessment notes are also useful in determining a CFS score.
Malnutrition and weight loss have also been closely linked to mortality in older adults and those sent to SNF are typical in that regard. A large study in Department of Veterans Affairs Hospital showed that patients with weight loss had significantly higher 1-year mortality after hospitalization (Liu et al. Reference Liu, Bopp and Roberson2002). More recently in patients undergoing percutaneous interventions for coronary artery disease, malnutrition was linked to all-cause mortality (Chen et al. Reference Chen, Huang and Lu2021). In our cohort of patients discharged to SNF, 89% of the patients of those who expired had a low albumin level on hospital readmission. Weight loss of at least 10 lb in the past year was associated with 80% 1-year mortality among patients who had hospital re-admission. For patients with 30-day hospital readmission from SNF, we found an alarming 90% 1-year mortality in those that presented with both a decline in CFS and a low albumin.
The main limitation of our study was that it is based on retrospective data of a smaller population from a single hospital system of Medicare fee-for-service patients. Our study only focused on the 2 subgroups of patients requiring hospital readmission from an SNF. By isolating and examining the variables and outcomes pertinent to these subsets, we aimed to identify risk factors and potential mechanisms that could explain the heightened mortality rates. Only 59 readmitted patient charts were available for in-depth chart review as we did not have access to medical records of patients who were not admitted to our hospital system. The low rate of patients staying within our system was a surprise as the patients studied all belong to an ACO and their primary care providers are associated with the hospital system studied. There is also potential for selection bias regarding which patients were readmitted to the authors’ hospital system (and included in the in-depth analysis) versus to other systems outside our system. It is hard to estimate the potential direction of these biases; however, it may be considered that patients admitted or readmitted to the hospitals that do not share the electronic medical record with their primary care and specialty doctors may be at risk for poor communication, greater harm and, perhaps, likelihood to have undesirable outcomes. Importantly, this study was limited to presenting associations between multiple confounders and the results do not infer direct causality. Lastly, we did not extend our data collection to include information on advance care directives, nor did we assess whether the reasons for hospital readmissions were consistent with those of the initial hospitalization.
Nevertheless, our findings provide a strong reminder to hospital and SNF medical teams that the population of patients who require skilled nursing stays are extremely vulnerable and have a high 1-year mortality rate whether they are readmitted or not. While we do not think the mortality reflects care in the SNF, more in-depth study is necessary to implicate care in the SNF, quality of handoffs between hospital and SNF or simply severe frailty. Patients discharging to SNF and especially those who have readmitted from SNF should be considered appropriate to have goals of care and prognostic discussions with their family and caregivers as these events are markers of high mortality.
Our small study emphasizes the findings of past studies and provides preliminary evidence of factors that should alert a physician and other providers of patients at increased risk for 1-year mortality (Burke et al. Reference Burke, Burke and Canamucio2020). Lists of triggers have been developed for specific illnesses as well as for the general population. We would add more triggers to promote discussions of goals of care, specifically discharge to SNF and more acutely readmission from SNF within 30 days. In addition, we suggest that serial frailty assessments and evaluation of nutritional status should be studied with a larger SNF population as potentially more specific triggers for serious discussions about care preferences. Table 3 is adapted from Kawashima and Evans (Reference Kawashima and Evans2023) reflecting non-disease specific triggers for initiating goals-of-care discussion. It includes findings from our current study to provide a more comprehensive understanding of the factors prompting goals-of-care discussions across various medical contexts.
Table 3. Triggers for goals-of-care discussions as adapted from Kawashima and Evans (Reference Kawashima and Evans2023)
APACHE II = Acute Physiology and Chronic Health Evaluation II; ED = emergency department; ICU = intensive care unit; SNF = skilled nursing facility; SOFA = Sequential Organ Failure Assessment.
Conclusions and implications
The 1-year mortality rates noted in those discharged and readmitted from SNF were similar in our population to previously published studies from 5 years ago but do vary according to the MDC (Hakkarainen et al. Reference Hakkarainen, Arbabi and Willis2016). Patients readmitted to hospital within 30 days of discharge to an SNF have an extremely high 1-year mortality rate for most prevalent MDCs except for circulatory disease. Evidence of malnutrition (low BMI, low albumin level and significant weight loss) and worsening CFS on readmission were also associated with high 1-year mortality in a subsample of these patients. Furthermore, while goals-of-care discussions are important for all older adults, they are critical during these transition points when patients are discharged to SNF and readmitted to the hospital within a month. From our small study, it appears that worsening score on CFS, significant weight loss and hypoalbuminemia can add further prognostic information. We suggest that further research, involving larger sample sizes, is essential to validate and expand our observations.
Acknowledgments
The authors acknowledge Nimisha Momin for her efforts supporting the initial concept of this paper, and Dr. Julia Andrieni and Dr. Ursula K. Braun for supporting our efforts to better understand outcomes for this population.
Competing interests
All the authors declare that there is no conflict of interest.
