Abstract
This study found that early assessment of activities of daily living (ADLs) in routine care of hospitalized patients may aid in treatment and care plan decisions, particularly for inpatients who may be at higher risk for adverse outcomes.
In 2012, there were approximately 36.5 million hospital stays in the United States, each with an average cost of $10,400 (Moore et al., 2006; Weiss & Elixhauser, 2014). In the United States, most hospital care is paid by insurance in a bundled model that is based on patient diagnosis, a fixed amount regardless of length of stay (LOS) or the specific services that are provided (Draper et al., 2006). Early identification of modifiable risk factors for adverse hospital-acquired conditions such as falls and pressure injuries, as well as other undesirable outcomes such as increased LOS or postacute care facility placement, can help providers prioritize their care, aid in discharge planning, and potentially reduce patient harms and their associated costs.
Many hospitalized patients experience difficulties performing activities of daily living (ADLs) either because of an underlying disease process (e.g., stroke) or because of the negative consequences of bed rest and hospital-acquired debility (Covinsky et al., 2003; Kim et al., 2021; Loyd et al., 2020; Martínez-Velilla et al., 2021; Reichardt et al., 2016). As such, increased difficulty performing ADLs has been shown to be associated with worse hospital-acquired conditions such as falls and pressure injuries, mortality, and patients’ ability to return home after hospitalization (Dreyfus et al., 2018; Pickham et al., 2018; Veedfald et al., 2021).
Because ADL functional deficits are potentially modifiable, early and systematic assessment of ADL function for all hospitalized patients must be prioritized. Systematic assessment of ADL function has the potential to facilitate early identification of at-risk patients and inform the plan of care for providers, such as earlier involvement of occupational therapists, but this is a current gap in practice (Hoyer et al., 2018; Jette, 2006).
At our institution, we recently incorporated the use of the Activity Measure for Post-Acute Care Inpatient Activity Short Form (AM-PAC IASF) to assess limitations in performing ADLs in the routine care of hospitalized patients. However, only one study has previously shown the potential predictive ability of the AM-PAC IASF for discharge to postacute care among a cohort of people with COVID-19 (Tevald et al., 2021). No studies have shown how the AM-PAC IASF can potentially identify patients who are at risk for major hospital-acquired conditions, inpatient falls, hospital-acquired pressure injuries, or other undesirable outcomes such as longer LOS. Hence, our goal in this study was to examine the association between ADL difficulty and these hospital-associated outcomes. This is an important first step to help establish the value of a more systematic assessment of ADLs to facilitate informed care planning in the hospital.
Method
Study Population and Setting
This was a retrospective cohort study of 1,899 consecutively admitted patients from two 32-bed acute care units using electronic medical record (EMR) data from patients who were discharged from the Johns Hopkins Hospital between January 1, 2017, and December 31, 2017. The units were selected because of their high compliance in AM-PAC IASF documentation on those units during the dates of data collection. All patients who were discharged from those units were eligible for analyses. These patients had various diagnoses, including, but not limited to, neurosurgical, stroke, and general neurology. We chose the independent and outcome variables a priori on the basis of discussions with hospital and research leaders. The Johns Hopkins Institutional Review Board approved this study with a waiver of consent.
Activities of Daily Living Measure
The AM-PAC IASF has six questions, each scored on a 4-point ordinal scale designed for the hospital population (Jette et al., 2014). Raw scores from the short form can be converted to a standardized scale score on a scale ranging roughly from 0 to 100 (Haley et al., 2004; Jette, 2006; Jette & Haley, 2005). Questions are scored on the basis of how much assistance the patient requires from another person to perform upper and lower body dressing, bathing, toileting, grooming, and eating meals (Jette et al., 2014, 2015). For our analyses, we used the standardized scale scores. We modeled the AM-PAC IASF score in decreasing 10-point increments, which is 1 SD of the population score. In a second model for disposition, we included AM-PAC IASF group (declined, improved, or stayed the same) by subtracting the first AM-PAC IASF score from the final AM-PAC IASF score. The magnitude of functional change represented by steps between raw scores (e.g., 12–13 compared with 23–24) was not the same; however, steps between standardized scores were the same (e.g., 28–29 = 44–45). Therefore, whenever comparisons are made between AM-PAC scores as an indicator of change in function, the standardized score must always be used for those comparisons to be valid.
The AM-PAC IASF has high interrater reliability (intraclass correlation coefficient, 0.91) and evidence for its validity among patients with a variety of medical conditions in the acute care setting (Jette et al., 2014, 2015). At the time of this study in our institution, the AM-PAC IASF was scored by nursing staff on admission and at least three times per week throughout a patient’s hospitalization (Hoyer et al., 2018; Klein et al., 2018). For our analysis, the primary exposure (also called initial score, and the one used in analysis) was the lowest AM-PAC IASF score recorded by nursing staff within the first 48 hours of hospitalization. We also used the last AM-PAC IASF score recorded by nursing staff in part of our analysis. Specifically, we created AM-PAC IASF groups (declined, improved, and stayed the same) by subtracting the first AM-PAC IASF score from the last AM-PAC IASF score.
Data Source and Variables
We acquired the patient’s age, sex, race, Agency for Healthcare Research and Quality (AHRQ) comorbidity index (Elixhauser et al., 1998), insurance type, diagnosis, LOS, and discharge disposition from the EMR. Where indicated, we used LOS quartile (calculated on the basis of the sample) to group patients into two groups for some analyses. Patients with an LOS in the longest quartile (more days in hospital; ≥8 days) were in one group, and those from the other three quarters were in the second group (<8 days). We grouped falls and pressure injuries by occurrence or absence, and we categorized discharge disposition by placement versus home. The AM-PAC IASF score was the independent variable of interest. To create groups of patients with similar diagnoses that were large enough for the statistical models, we grouped primary admitting diagnoses into one of the following categories for the statistical models: craniofacial surgery; spinal surgery; central nervous system, peripheral nervous system and/or other organ system infections; multiple sclerosis, myelitis, and/or amyotrophic lateral sclerosis; ischemic stroke, hemorrhagic stroke, and/or traumatic brain injury; and other. For convenience in presenting descriptive statistics (and only for those statistics), we further grouped diagnoses into one of the following categories: neurology, neurosurgery, and other. Occurrence of injurious falls and pressure injuries were recorded in the patients’ medical records in accordance with the National Database of Nursing Quality Indicators criteria. Falls were included in this study if they occurred during hospitalization, and these data were specifically obtained through the patient safety event reporting system. Pressure injuries were included in this study if they were acquired during hospitalization and were Stage II or greater.
Statistical Analysis
We used independent two-tailed t tests to examine differences in AM-PAC IASF scores between the dichotomous groups that we created for each outcome variable (LOS, ≥8 days vs. <8 days; fall, yes vs. no; pressure injury, yes vs. no; discharge disposition, placement vs. home). We then used logistic regression to estimate the relationship between the AM-PAC IASF score, and outcome variables (LOS, falls, pressure injuries, and discharge disposition). In four separate models for each outcome, we modeled the AM-PAC IASF in decreasing 10-point increments, which is 1 SD of the population score. In a fifth model, discharge disposition, we grouped AM-PAC IASF into three categories (declined, improved, the stayed the same) that were based on subtracting the first AM-PAC IASF score from the final AM-PAC IASF score. We examined the bivariate relationship between AM-PAC IASF score and outcomes of interest (Table A.1 in the Supplemental Material, available online with this article at https://research.aota.org/ajot). We adjusted all models by age, sex, race, insurance type, diagnosis, and AHRQ comorbidity index. We also adjusted the LOS model by expected LOS. Expected LOS is calculated using logistic regression on the basis of patients with a similar profile of clinical characteristics such as demographics, functional status, prognosis, clinical conditions, clinical treatments, and clinical diagnoses (University HealthSystem Consortium, 1998). We used the penalized likelihood method for logistic regression (Firth, 1993) when modeling falls and pressure injuries to improve the estimates for these rare events. For each model, we also calculated estimates for area under the curve (AUC). The AUC in this context is a measure of goodness-of-fit and was selected to compare models to determine which independent variable was best predicted by the AM-PAC activity score. We performed all statistical analyses using SAS® (Version 9.4).
Results
Study Population
There were 1,899 patients with AM-PAC IASF scores in the study period and who were included for analysis. The sample was 45% male and 64% Caucasian; 32% were Medicare recipients, and 66% had undergone a surgical procedure. The mean admission AM-PAC IASF score for patients in our sample was 39.1 (SD = 10.8). Table 1 summarizes the patient demographics by outcome (LOS, falls, pressure injuries, and disposition).
Patient Demographics by Outcome
Note. Values are n (%) unless otherwise indicated. Percentages are calculated within each outcome column unless otherwise noted. AM-PAC = Activity Measure for Post-Acute Care; IASF = Inpatient Activity Short Form
Length of stay was separated into two groups on the basis of the longest quartile and all others.
AM-PAC IASF (activities of daily living function) t-scale scores.
Length of Stay
The mean LOS was 7.0 days (SD = 7.5). We observed a statistically significant difference in mean admission AM-PAC IASF scores between those in the longest LOS quartile (M = 34.1, SD = 11.7) and all others (M = 41.0, SD = 9.8; difference, p < .0001), as reported in Table 1. In an unadjusted analysis, patients with lower AM-PAC IASF scores (every 10-point difference) had 1.9 times higher odds of being in the longest LOS quartile (≥8 days; 95% confidence interval [CI] [1.7–2.1], p < .001). The adjusted odds was 1.6 (95% CI [1.4–1.8], p < .001) for patients with the longest LOS quartile (Table 2).
Adjusted Logistic Regression Models Showing the Relationship Between AM-PAC Inpatient Activity Short Form Scores Near Hospital Admission and Outcomes of Interest
Note. AM-PAC = Activity Measure for Post-Acute Care; AUC = area under the curve; CI = confidence interval.
Separate models were adjusted by age, sex, race, insurance, diagnosis group, and Agency for Healthcare Research and Quality comorbidity index. Length of stay model was also adjusted by expected length of stay.
Values are for every negative 10-point difference in the first AM-PAC Inpatient Activity Short Form score. p < .001.
Odds of being in the longest quartile of length of stay (≥8 days).
Discharge to postacute care facility.
Hospital-Acquired Conditions
We observed a statistically significant difference in mean admission AM-PAC IASF scores between patients who had falls (mean = 33.1, SD = 8.7) and those who did not (M = 39.2, SD = 10.8; p < .0001) and between patients who acquired a pressure injury (M = 27.1, SD = 9.2) and those who did not (M = 39.1, SD = 10.8; p = .0005; see Table 1).
In an unadjusted analysis, patients with lower (every 10-point difference) AM-PAC IASF scores had 1.7 times higher odds (95% CI [1.3–2.2], p < .001) of suffering an injurious fall. For injurious falls, the adjusted odds were 1.7 (95% CI [1.3–2.2], p < .001). The unadjusted odds of acquiring a pressure injury during hospitalization were 2.9 times higher (95% CI [1.6–5.4], p < .001) for people with lower (every 10-point difference) AM-PAC IASF scores. The adjusted odds were 1.7 higher (95% CI [1.3–2.2], p < .001) for pressure injury, as seen in Table 2.
Discharge Disposition
There was a statistically significant difference in the mean AM-PAC IASF scores between patients who were discharged home (M = 41.7, SD = 10.1) and those who went to a postacute care inpatient facility (M = 32.6, SD = 9.8, p < .0001; see Table 1). In an unadjusted analysis, the odds of discharge to a postacute care facility, instead of home, were 2.6 times higher (95% CI [2.3–2.9], p < .001) for patients with lower (every 10-point difference) AM-PAC IASF scores. The adjusted odds were 2.4 higher (95% CI [2.1–2.7], p < .001) for discharge to a postacute care facility, as seen in Table 2. In an unadjusted analysis, for patients with an AM-PAC IASF score that decreased during hospitalization, the odds of being discharged to a facility were 3.3 times higher (95% CI [2.9–3.8], p < .001) than for patients with scores that improved or stayed the same. The adjusted odds were 3.0 times higher (95% CI [2.2–4.2], p < .001) for being discharged to a facility compared with those for going home, and the AUC in this additional model was 0.87 (95% CI [0.86–0.89]).
Discussion
In this retrospective study, we found that patients who have lower ADL function (toileting, bathing, grooming, etc.) on admission exhibited a longer LOS, had an increased risk for adverse hospital-acquired conditions, and were more likely to discharge to a postacute care facility instead of going home. Indeed, because the AM-PAC IASF was associated with all these important patient outcomes, use of this tool could be helpful to drive interventions for risk mitigation. Examples of mitigating interventions using the AM-PAC IASF could include early discharge planning, implementation of strategies for fall risk reduction, and occupational therapy consultations to address specific ADL impairments.
Limited research has been published on the relationship between function in ADLs and LOS, falls, and pressure injury. It has been reported that, for adults living in nonhandicapped communities, there is a relationship between the performance of ADLs and risk of future falls (Carroll et al., 2022; Hendrich, 2021; Mamikonian-Zarpas & Laganá, 2015; Newkirk et al., 2022; Strini et al., 2021). Our results from a hospitalized population were similar, showing that there are greater odds for falls among people with lower AM-PAC IASF scores (indicating lower ADL function). There continues to be debate about how to best assess someone’s risk of falling (Harrington et al., 2010; Morgan et al., 1985), and ADL ability may be an important factor to consider. Limitations in ADL function are likely caused by the same impairments (e.g., weakness, poor coordination, sequencing difficulties) that increase the risk for falls and pressure injury. Falls that occur in the hospital increase the patient’s overall length of stay (Dunne et al., 2014), but our findings may be the first that links lower ADL ability at hospital admission to longer LOS, even when accounting for other potential causes.
Assessment of ADL performance early and throughout hospitalization is important for care planning and implementation of interventions (Loyd et al., 2020). We found that patients with lower AM-PAC IASF scores had greater odds of needing postacute care. This was similar to the finding of Tevald et al. (2021), who found the same thing among people who were hospitalized for COVID-19. Additionally, we found that patients with an AM-PAC IASF score that declined during hospitalization had higher odds of being discharged to a postacute care facility rather than home. The extent to which the trajectory in ADL function influences outcomes (or can be altered through intervention) needs further investigation, but it was not surprising to find that postacute care needs were greater for people with more limitations in ADL function. The important implication here is that early and ongoing scoring of the AM-PAC IASF may aid in earlier treatment, decision making, and recommendations by occupational therapists with regard to discharge disposition. Additionally, others have shown that early discharge planning with occupational therapist involvement may even reduce patient overall length of stay (Fox et al., 2013).
One limitation to this study is that the sample was collected from two units within an urban academic medical center. Although our patients did have diverse medical conditions, future research should consider other hospitals and patient populations. Furthermore, although the relationship between AM-PAC and pressure injuries and falls was significant, there were few patients in our sample with these rare events. Therefore, the relationship between ADL function and these two outcomes needs additional investigation. Finally, this retrospective, observational study indicates important associations between the AM-PAC IASF and the outcome variables but does not allow for prediction or establish cause and effect. Despite these limitations, our robust associations are likely to be present in other populations and larger samples.
Implications for Occupational Therapy Practice
Occupational therapy is associated with improved functional performance for patients with demonstrated ADL impairment in the acute care setting (Burgess & Jensen, 2019). However, ADL deficits are often underrecognized in this setting, because they are not systematically assessed in routine care. Additionally, the acute care setting is not currently structured or staffed to provide occupational therapy intervention to all patients. Therefore, the use of a standardized assessment of ADL capacity could help identify ADL deficits early in hospitalization so that occupational therapists can prioritize therapy interventions for patients with the greatest need and potentially affect outcomes (Probasco et al., 2018).
This study has the following implications for occupational therapy practice: ▪ Early assessment of patients’ ADL functional capacity is useful in identifying patients sooner who have the potential to benefit from occupational therapy services. ▪ Systematic assessment of impairments in performing ADLs for all hospital patients may facilitate timelier, and appropriately targeted, occupational therapy interventions, which could minimize negative outcomes.
Conclusion
The decreased ADL function of hospital patients increases their risk for longer length of stay, institutional discharge disposition, inpatient falls, and hospital-acquired pressure injuries. Incorporating a systematic ADL measure into routine care could facilitate early risk mitigation strategies with the potential to improve outcomes and reduce costs.
Supplemental Material
Supplementary material for Activities of Daily Living Assessment Early in Hospitalization Is Associated With Key Outcomes
Supplementary material, sj-pdf-1-aot-10.5014_ajot.2023.050167.pdf for Activities of Daily Living Assessment Early in Hospitalization Is Associated With Key Outcomes by Annette Lavezza, Erik Hoyer, Lisa Aronson Friedman, Kelly Daley, Amber Steele, Stephanie Rosen and Daniel Young in The American Journal of Occupational Therapy
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
