Quality of Care for Long-term Care Residents Living with Heart Failure in Ontario and Predictors of Hospitalization

Abstract

Background

Heart failure (HF) is a disease that is on the rise, particularly in the aging population. It is common amongst residents of long-term care homes (LTCHs). Complicating the diagnosis and treatment of HF is the interaction of geriatric symptoms and comorbidities. Literature also suggests that in addition to being under-detected, HF management is suboptimal in the long-term care setting. The combination of the complex nature of the disease in older adults, as well as poor management practices can lead to adverse outcomes such as hospitalization, depression, cognitive decline, loss in activities of daily living (ADL) and mortality. This study addressed the following research questions: 1. Upon admission, what are the clinical and demographic characteristics of residents living with HF, compared to those living without HF? 2. In residents with HF, what admission clinical and demographic characteristics are associated with hospitalization? 3. What is the quality of care for residents with HF in Ontario LTCHs? 4. Are there regional variations in quality of care for residents with HF in Ontario?

Methods The data in this study were based on the InterRAI Minimum Data Set Instrument (MDS) 2.0 assessments of residents aged 65 years and older, who were admitted to LTCHs in Ontario between January 1st, 2011 and December 31st, 2013. Residents with HF that had an end-stage disease, an expected survival of less than six months, receiving hospice care or in palliative units at admission, were excluded. Demographic and clinical information of residents with HF, and no HF at admission were summarized using means and standard deviations (SD) for continuous measures, and frequencies and percentages for categorical measures. Chi-square test was be used to evaluate whether the differences were significant in categorical measures, while continuous measures were analyzed using t-tests. To examine predictors of hospitalization, bivariate associations of demographic and clinical characteristics with spending at least one day in a hospital, were analyzed at the significance level of alpha= 0.05. In addition to p-values and odds ratios, 95% confidence intervals (CI) were used to determine whether the clinical variables were significantly associated with hospitalization. For the multivariable analysis, variables found to be significant at a bivariate level were included. Logistic regression modeling using generalized estimating equations (GEE) was used. Variables identified from the bivariate analysis were individually added to the model using step-wise selection. The C-statistic estimated the model sensitivity to predicting hospitalization

The MDS Third Generation QI scores across all local health integration network (LHIN) were used to demonstrate variability between them by quality domain. Two steps were carried out to understand the overall variability in QI scores among LHINs over time: 1) the adjusted QI scores for each LHIN were calculated within each quarter; 2) the aggregated median, interquartile range, and range in QI scores for each LHIN were calculated and plotted in a Box and Whisker Plot. The median scores were calculated in each QI per LHIN to compare quality performance amongst LTCHs located in the same region.

Results A total of 48,601 residents were included in the study with 12.3% diagnosed with HF. Compared to other residents, those with HF were slightly older, more frequently admitted from a hospital setting (43.0% vs. 34.4%), had a significantly higher number of comorbidities (6.5±2.4 vs. 4.7±2.1) and were prescribed an average of two additional medications (11.9± 4.6 vs. 9.6± 4.9) at admission. The rate of hospitalization in the sample residents with HF was 36.2%. In residents with HF, the final regression model found admission to a LTCH from a hospital setting was the strongest predictor of hospitalization (OR: 8.09, CI: 7.05-9.29), followed by a CHESS score of greater than 3, which indicates high levels of health instability (O.R 4.24, CI: 3.07-5.85). Other variables that increased the likelihood of hospitalization included monitoring for acute medical illness (O.R: 1.45, CI: 1.26-1.67). Physician visits of over three days increased odds of hospitalization by 1.6 times (CI: 1.21-2.19, P= 0.0013) and prescription with an anti-depressant (O.R: 1.16, CI: 1.0-1.33, p=0.03). Quality of care was not consistently high or low among residents in each LHIN, differing in performance across domains of quality. Of the quality indicators, decline in ADL self-performance was highest (Median: 39.6%). Approximately a third of residents had decline in mood from symptoms of depression (26.7%) and were on prescriptions of anti-psychotics without symptoms of psychosis (29.3%), while a quarter had respiratory infections (24.7%). Some QIs scores showed very little variation over time within regions (as shown by interquartile range). On the other hand, some regions demonstrated greater variations over quarters, such as ADL decline in the Central West region, which ranged from 23.6% to 35.7% (25th and 75th percentile, respectively). When comparing QI scores among LHINs, in certain aspects of quality, some regions had lower median rates, while others had higher scores. For example, mood decline in Toronto Central was at 17.1% in contrast to 30.3% in the Waterloo-Wellington region. Discussion This study described the clinical characteristics of residents living with HF in Ontario LTCHs. Findings from this study are consistent with those of previous studies describing the complex clinical profile of residents with HF in LTCH. However, some divergent findings also exist. The prevalence of HF in was 12.3%, which is lower than what has been found by other studies (Hancock et al., 2013; Foebel et al., 2013; Daamen et al., 2010). The difference in prevalence may be related to poor implementation of HF screening guidelines, lack of knowledge of HF symptoms in nursing staff in nursing homes and the complexity of HF in older adults (Marcella et al., 2012). Another important finding was that residents with HF were significantly more likely to be admitted from hospitals to LTCHs than those without HF. Admission from a hospital into LTC was found to be the strongest predictor of subsequent hospitalization in our study. Older adults that are hospitalized for HF and that are more likely to be discharged into nursing homes, have poorer health in comparison to those discharged to the community (Allen et al., 2011). Evidence from our results of QI performance among LHINs suggests that there continues to be room for improvement in providing care for residents with HF, particularly in terms of functional decline, symptoms of depression and prescription of anti-psychotics. What this suggests is that some nursing homes within regions face particular challenges in addressing these aspects of quality uniformly across conditions. However, special considerations need to be given to the complex care needs for residents living with HF. Another important finding was the variability of quality of care among LHINs, with some regions demonstrating low QI scores on certain aspects of quality in comparison to others. It should be noted that this pattern was not consistently found across other QIs, suggesting that performance is not uniform across quality domains or regions. However, these disparities in care quality can be attributed to the care setting, rather than the physical location of the nursing home (Phillips et al., 2004). The differences in regional LTC performance highlight the importance of understanding the complex context of nursing homes and its influence on care. care system

Conclusion This work shows that residents with HF living in Ontario comprise a subset of the LTC population with complex clinical characteristics. Study findings on admission characteristics that predict hospitalization can inform future research developing a risk adjusted QI measuring hospitalization in this population. The implications of this include early identification of residents facing higher likelihood of hospitalization, as well, detection of LTC practices that result in avoidable admissions. Outcomes and processes of care in nursing homes for residents with HF show that there is a need for improvement in domains of functional ability, anti-psychotic use, anti-depressants and depressive symptoms, highlighting the need to explore the aspects of LTC settings that contribute to these findings.