Length of Stay at Inpatient Rehabilitation Facility and Stroke Patient Outcomes

Transcription

1 FEATURE Length of Stay at Inpatient Rehabilitation Facility and Stroke Patient Outcomes Michelle Camicia 1, MSN, CRRN, CCM, Hua Wang 1, PhD, Margaret DiVita 2, PhD, MS, Jacqueline Mix 3, MPH & Paulette Niewczyk 3, PhD, MPH 1 Kaiser Foundation Rehabilitation Center, Kaiser Permanente Medical Center, Vallejo, CA, USA 2 Health Department, State University of New York at Cortland, Cortland, NY, USA 3 Uniform Data System for Medical Rehabilitation, Amherst, NY, USA Keywords Stroke; rehabilitation; length of stay; outcomes. Correspondence Michelle Camicia, Director, Kaiser Foundation Rehabilitation Center, 975 Sereno Dr. Vallejo, CA michelle.camicia@kp.org Accepted March 27, doi: /rnj.218 Abstract Purpose: To examine the association of inpatient rehabilitation facility (IRF) length of stay (LOS) with stroke patient outcomes. Design: A secondary data analysis of the Uniform Data System for Medical Rehabilitation database. Methods: Stroke patients discharged from IRFs in the United States between 2009 and 2011 were identified and divided into mild (n = 639), moderate (n = 2,065), and severely (n = 2,077) impaired groups. Study outcomes included cognition and motor functional gains measured by the Functional Independence Measure (FIM) instrument and discharge to the community. Findings: The average LOS was 8.9, 13.9, and 22.2 days for mild, moderate, and severely impaired stroke patients, respectively. After controlling for FIM admission and other important covariates, a longer LOS was associated with a modest increase in cognition gain (b = 0.038, p =.0045) for the moderately impaired patients, and a modest increase in cognition (b = 0.13, p <.0001) and motor gains (b = 0.25, p <.0001) as well as a tendency for discharge to the community (OR = 1.01, 95% CI = ) among the severely impaired patients. However, a longer LOS showed a negative association with functional gains among the mildly impaired patients as well as discharge to community for both mild and moderately impaired patients. Conclusion: The association of IRF LOS and patient outcomes varied by stroke impairment severity, positively for more severely impaired patients and negatively for mildly impaired patients. Clinical Relevance: The study provides evidence for the care of stroke patients at the IRF setting. Introduction It is a critical time for the inpatient rehabilitation hospital industry, as the allocation of resources and quality of care provided are under increasing scrutiny. Payers of health care expect that the services provided will be of value, resulting in positive outcomes relative to the fiscal expenditure. The majority of costs associated with inpatient 78

2 M. Camicia et al. Rehabilitation Length of Stay and Stroke Patient Outcomes rehabilitation are related to the number of days of the rehabilitation hospital stay. To encourage the financial incentives of inpatient rehabilitation facilities (IRF) and to adjust reimbursement for patient severity, the Centers for Medicare & Medicaid (CMS) implemented the prospective payment system in 2002 (Medicare Payment Advisory Commission [MedPAC], 2008). As part of this regulation, patients are grouped according to clinical and demographic characteristics into case mix groups (CMGs), which represent the expected resource needs of the patient. It has been reported that the average IRF LOS for stroke patients decreased from 19.6 to 16.6 days between 2000 and 2007 (Granger, Markello, Graham, Deutsch, & Ottenbacher, 2009). However, studies investigating the impact of IRF LOS on rehabilitation outcomes are limited. In the report by Granger et al. (2009), the IRF LOS decreased during the time period of , but the rate of discharge to the community decreased from 75.8% to 69.3%, and the rate of acute care hospital transfer increased from 5.6% to 10.3%. In a recent study by O Brien, Xue, Ingersoll, and Kelly (2013), Medicare beneficiaries with stroke experienced 1.8 days IRF LOS reduction between 2002 and 2007, as well as 5.4% decline in discharge to the community. They observed that after controlling for functional status at IRF admission and other covariates, a longer IRF LOS was associated with a higher Functional Independence Measure (FIM) 1 total at IRF discharge (p <.0001) but less likelihood to be discharged to the community (OR = 0.997, 95% CI = ). In addition to medical conditions (Appelros, 2007; Elwood et al., 2009; Galynker et al., 1997; Harvey et al., 1998; Saxena, Koh, Ng, Fong, & Yong, 2007) and functional status at IRF admission (Brock et al., 1997; Franchignoni, Tesio, Martino, Benevolo, & Castagna, 1998; Harvey et al., 1998; Tan, Heng, Chua, & Chan, 2009; Stillman et al., 2009), other patient characteristics and nonmedical factors influence the duration of IRF stay, including patients socioeconomic status and family structure (Tan, Heng et al., 2009), caregiver characteristics and IRF discharge planning program (Tan, Chong, Chua, Heng & Chan, 2009), and incentives for cost containment and support inpatients home environments (Stineman et al., 2001). In this study, we examined the associations of LOS and rehabilitation outcomes for stroke patients who were discharged from an IRF in the United States between 2009 and We hypothesized that a longer IRF LOS is associated with increased functional gain and discharge to the community for stoke patients. We also examined other factors that may be influential on IRF LOS in the study sample. Methods We conducted a retrospective cohort study for stroke patients who were discharged from an IRF between 2009 and The study sample was extracted from the Uniform Data System for Medical Rehabilitation (UDSMR), the largest national database on inpatient rehabilitation outcomes. UDSMR maintains patient demographic, medical, and rehabilitation data for over 70% of IRF in the United States using the Inpatient Rehabilitation Facility Patient Assessment Instrument (IRF-PAI). Included in the IRF-PAI is the FIM instrument, which is a measure of overall functional status and is comprised of 13 motor and five cognitive items. Each item is rated on a 7-level ordinal scale, with 1 indicating total assistance and 7 indicating complete independence. The sum of the item scores describes the severity of an individual s disability and reflects the burden of care and amount of assistance that is required for an individual to complete daily activities (Granger, Cotter, Hamilton, & Fiedler, 1993). All patients who were discharged between 2009 and 2011 with an impairment code of stroke (n = 286,909) were identified. To be included in the sample, patients had to have an age at IRF admission 18 years, a valid onset date, a length of stay (LOS) between 3 and 365 days; were not admitted for evaluation; and were discharged alive from the IRF. After exclusions, a total of 261,212 patients remained in the study sample. A 2% random sample was extracted (n = 5,424) from the patients who fulfilled the study criteria given the consideration that an overly large sample may lead to a loss in accuracy by rejecting a null hypothesis, even if the actual effect is very small and does not have practical importance. Since patients with an acute care hospital transfer during IRF stay usually had a short IRF LOS, we excluded this group of patients from further analysis. The final study sample consisted of 4,781 patients. The selection of about 5,000 patients was originally proposed according to our health services research experience using UDSMR database. A power analysis was also conducted based on findings from the literature to provide evidence for a sample size decision. The suggested minimum clinically important difference in FIM gain was 3 and 17, respectively, for cognition and motor measures in patients with stroke (Beninato et al., 2006). With a two-sided test, significance level of.05, and study power of 80%, our study 79

3 Rehabilitation Length of Stay and Stroke Patient Outcomes M. Camicia et al. sample allowed for the detection of a minimum cognition FIM gain between 0.8 and 2.0 as well as a minimum motor gain between 1.94 and 3.86, providing adequate power for the study. Bivariate analysis showed that there was no significant difference between the study sample and the base population in any of the exposure and outcome measures under study. Study outcomes included cognition FIM gain, motor FIM gain, and percentage of patients discharged to the community. The cognition FIM gain and motor FIM gain variables represent the gain in function in the cognitive and motor domains and are calculated by taking the difference of the summed values in each domain from admission to discharge. The percentage of patients discharged to the community represents the overarching goal of IRFs, which is to restore the patient to a functional level that enables them to return to the community. The main independent variable of interest was IRF LOS. Sociodemographic measures of interest included age, gender, marital status (married vs. single), race (White, Black, Hispanic, other), prehospital living situation (living alone or living with others), primary payer source (Medicare, Blue Cross, commercial, Medicaid, workers compensation, other), hemisphere of brain injury (left, right, bilateral, no paresis, other stroke), CMS comorbidity tier (no comorbid, mild, moderate, and severe), onset days to rehabilitation (time from stroke onset to IRF admission), and healthcare setting admitted from (acute care, community, long-term care, other, rehabilitation). In addition, facility measures such as facility type (freestanding hospital or unit in acute care hospital), number of operating beds and region of facility were evaluated. The study sample was stratified by stroke impairment using CMG, which is a proxy for patient severity on the basis of impairment, age, and functional status at IRF admission, as determined by the CMS (MedPAC, 2008). Patients in the same CMG group are expected to require similar resource utilization. Stroke severity was categorized as mildly impaired (CMG ), moderately impaired ( ), and severely impaired ( ). Descriptive statistics were calculated for all explanatory and outcome measures, including means, standard deviations, medians, and interquartile ranges for continuous variables, and frequency distributions for categorical variables. Bivariate analysis was used to calculate the unadjusted associations between exposure and outcomes, as well as among covariates with the use of corresponding parametric and nonparametric methods such as correlation, Chi-square test, one-way ANOVA. General linear models were developed to examine the association between LOS and functional outcomes, while controlling for covariates. Multiple regression analysis was utilized to examine the factors that were significantly associated with IRF LOS. The estimated coefficients, standard errors, and p-values were used to measure the strength of the adjusted associations. The statistical software SAS (version 9.13, SAS Institute Inc., Cary, NC) was used and a significance level of.05 was set for all the analyses. This study was approved by the Kaiser Permanente Northern California Institutional Review Board. Findings Table 1 presents descriptive statistics of the study sample, total, and according to stroke severity. There were 649 mildly impaired, 2,185 moderately impaired, and 2,390 severely impaired stroke patients in the study sample. Socio-demographic and medical variables between the stroke impairment groups differed significantly in most of the measures under study (p <.0001). The moderate and severely impaired patients had a higher percentage of African American and Hispanics, but a lower percentage of Whites than the mildly impaired group (p =.0163). The moderate and severely impaired groups also consisted of a higher percentage of patients in the Eastern region, but a lower percentage of patients in the Central region than the mildly impaired group (p =.0013). Table 2 presents functional measures and discharge destination, total, and by stroke severity, with significant differences among the subgroups on all measures (p <.0001). In comparison to the mildly impaired group, patients in the moderate and severely impaired groups had a lower admission and discharge FIM score, but achieved a relatively large FIM gain during the IRF stay. The mildly impaired group had the highest rate of returning to the community after IRF stay (95.6%), followed by moderate (88.3%) and severely impaired (60.0%) groups. IRF LOS and Functional Gain Figure 1 shows the correlations between IRF LOS and functional gains, for the total stroke group and separately by impairment severity groups. A positive correlation was observed between IRF LOS and FIM gain (cognition and motor) for total, moderate, and severely impaired stroke. Table 3 presents adjusted associations of IRF LOS and 80

4 M. Camicia et al. Rehabilitation Length of Stay and Stroke Patient Outcomes Table 1 Study population characteristics, UDSMR Stroke Severity Parameters Total Mild Moderate Severe p-value N (%) 4, (13.4) 2,065 (43.2) 2,077 (43.4) Age (years) Mean SD <.0001 Median (IQR) 71 (59 80) 66 (55 76) 70 (59 80) 72 (62 81) Age (years, %) < Gender (%) Male Female Marital status (%) Married Single Race (%) Black Hispanic Other White Prehospital live alone (%) Alone Other Primary pay source (%) Blue Cross <.0001 Commercial Medicaid Medicare Other Workers comp Admit from (%) Acute <.0001 Community Long-term care Other Rehabilitation Brain injury side (%) Left <.0001 Right Bilateral No paresis Other stroke CMS comorbidity tier (%) None <.0001 Severe Moderate Mild (continued) 81

5 Rehabilitation Length of Stay and Stroke Patient Outcomes M. Camicia et al. Table 1 (continued) Parameters Total Stroke Severity Mild Moderate Severe p-value Facility type Freestanding hospital <.0001 Unit in acute care hospital Operation beds (number) Mean SD <.0001 Median (IQR) 32 (20 60) 26 (18 48) 30 (20 56) 40 (21 60) Region (%) Central East West Onset to IRF admission (day) Mean SD <.0001 Median (IQR) 5.0 (4 9) 4 (3 7) 5 (3 8) 6 (4 11) Onset to IRF admission (day, %) < IRF LOS Mean SD <.0001 Median (IQR) 15 (10 22) 8 (6 11) 13 (9 17) 22 (16 26) UDSMR, Uniform Data System for Medical Rehabilitation; SD, standard deviation; IRF, inpatient rehabilitation facility; IQR, interquartile range; LOS, length of stay. functional gains by impairment groups, controlling for covariates under study. IRF LOS was negatively associated with the cognition (b =.098, p <.0001) and motor (b =.192, p =.0133) functional gain in the mildly impaired group. The association between IRF LOS and cognitive functional gain remained positive in the moderately impaired group (b =.038, p =.0045), but not with motor functional gain (b =.004, p =.9027). The IRF LOS also remained positively associated with both cognition (b =.127, p <.0001) and motor (b =.25. p <.0001) functional gains in the severely impaired group. In addition to the association of IRF LOS and functional gains, a number of covariates contributed to the functional gains in the study sample (Table 3). IRF LOS and Discharge to the Community Table 4 presents the unadjusted association between IRF LOS and discharge to the community. In the mild and moderately impaired groups, patients who were discharged to the community had a shorter LOS than those transferred to other care settings (p <.001). There was no statistically significant difference of IRF LOS between discharge destinations in the severely impaired group (p =.8321). Table 5 presents findings of the logistic model for discharge to the community by impairment severity group. Patients with a longer IRF LOS were less likely to be discharged to the community in the mild (OR = 0.905, 95% CI = ) and moderately (OR = 0.943, 95% CI = ) impaired groups. There was a tendency of being discharged to the community (OR = 1.010, 95% CI = ) for patients with a longer IRF LOS in the severely impaired group. Similarly, several covariates were also associated with discharge to the community (Table 5). Influential Factors for IRF LOS Table 6 presents factors that are significantly associated with the IRF LOS. Single marital status (p =.0114), primary payer source of Blue Cross, Commercial, Medicaid, and other vs. Medicare (p <.01), more severely impaired patients (p <.0001), a lower motor FIM rating at IRF admission (p <.0001), and a longer duration of stroke onset to IRF admission (p <.01) were associated with a longer IRF LOS. 82

6 M. Camicia et al. Rehabilitation Length of Stay and Stroke Patient Outcomes Table 2 Functional measures and discharge destination among stroke patients, UDSMR Stroke Severity Parameters Total Mild Moderate Severe p-value N (%) 4, (13.4) 2,065 (43.2) 2,077 (43.4) Cognitive FIM total Admission Mean SD <.0001 Median (IQR) 20 (14 25) 25 (20 29) 22 (17 27) 16 (11 22) Discharge Mean SD <.0001 Median (IQR) 26 (21 30) 29 (25 33) 28 (23 31) 23 (16 28) Gain Mean SD <.0001 Median (IQR) 5 (2 8) 3 (1 6) 4 (2 7) 5 (2 9) Motor FIM total Admission Mean SD <.0001 Median (IQR) 35 (24 47) 57 (53 60) 42 (37 47) 22 (16 27) Discharge Mean SD <.0001 Median (IQR) 62 (49 72) 77 (70 81) 67 (60 74) 48 (33 59) Gain Mean SD <.0001 Median (IQR) 24 (16 32) 19 (14 24) 25 (18 31) 23 (14 34) Discharge destination (%) Community <.0001 Other Long-term care Subacute setting UDSMR, Uniform Data System for Medical Rehabilitation; SD, standard deviation; IQR, interquartile range. Discussion While the average IRF LOS for stroke patients has decreased in recent years, studies on associations of IRF LOS and patient outcomes are limited. Granger et al. (2009) reported the trends of LOS in stroke patients discharged from U.S. IRFs between 2000 and 2007 and found a 3-day reduction in IRF LOS over the time period, decrease in total FIM ratings at IRF admission (from 62.6 to 55.1) and discharge (from 86.4 to 79.8), increase in total FIM gain (from 23.9 to 24.5) and LOS efficiency (from 23.9 to 24.5), a decrease in discharge to community from 75.8% to 69.3%, but an increase in the acute care hospital transfer rate from 5.6% to 10.3%. In a recent study of 371,211 Medicare beneficiaries over 65 years old and with stroke between 2002 and 2007 (O Brien et al., 2013), there was a 1.8 day reduction in IRF LOS, a decreased FIM total rating at both IRF admission and discharge, and a 5.4% decline of community discharge over the study period. Multiple regression analyses were used to examine the associations of IRF LOS and FIM total at IRF discharge as well as discharge to the community. After controlling for functional status at IRF admission and other covariates, a 1 day longer stay at IRF was associated with 0.50 total FIM increase at IRF discharge (p <.0001) as well as less likelihood to be discharged to the community (OR = 0.997, 95% CI = ). The study findings may be more generalizable to patients 65 years or older. In this study, the study sample was randomly extracted from the adult stroke population in the UDSMR database between 2009 and 2011 and was stratified into mild, moderate, and severely impaired groups based on CMG measure. The effect of IRF LOS differed between the three impairment groups. A longer IRF LOS was positively associated with cognition and motor gain as well as a higher likelihood of discharge to the community among 83

7 Rehabilitation Length of Stay and Stroke Patient Outcomes M. Camicia et al. Figure 1 IRF LOS and cognition and motor FIM gain, total and by impairment severity groups. IRF, inpatient rehabilitation facility; LOS, length of stay, r = Spearman correlation. severely impaired patients. These data show that severely impaired stroke patients require additional resources to provide necessary therapy, nursing, and physician services to address their increased medical complexity. Resources must be dedicated to family and caregiver training as, although an increase in function is realized, these patients are usually discharged requiring the assistance of others with activities of daily living and mobility. A longer IRF LOS was also positively associated with cognition gain in the moderately impaired group. However, there was no significant association between IRF LOS and motor functional gain among moderately impaired patients. Functional recovery has a steep slope of improvement in the earlier stages of rehabilitation. Discharge from the facility is indicated when functional recovery slows, even with intensive inpatient therapy. Furthermore, a longer IRF LOS had negative impact on cognition and motor gain in the mildly impaired group as well as discharge to community for both mild and moderately impaired patients. This finding may be a result of the negative effects of hospitalization, which has been noted to result in functional decline, particularly in hospitalized elderly patients (Zisberg et al., 2011). Although rehabilitation hospitals focus on increasing function, it is an institutional environment, where needs are readily met and the requirement for cognitive demand and mobility are less than in the home environment. The differences in the effects of IRF LOS on patient outcomes may be attributable to the characteristics of the impairment groups. In comparison with the severely impaired group, patients in the mild and moderately impaired groups were younger and less severe in terms of comorbidity and admission functional 84

8 M. Camicia et al. Rehabilitation Length of Stay and Stroke Patient Outcomes Table 3 General linear model: Cognition and motor FIM gain, UDSMR Cognition FIM gain Motor FIM gain Parameter Estimate Standard Error p-value Estimate Standard Error p-value Mild LOS (day) < Primary pay source Blue Cross <.0001 Commercial <.0001 Medicaid Other Medicare Admit from Community Long-term care Subacute care Acute care Adm FIM cognition < Adm FIM motor <.0001 Facility type Freestanding hospital Unit in acute care R Moderate LOS Age (year) < < < Prehospital living Alone Other Primary pay source Blue Cross Commercial Medicaid Other Workers comp Medicare Admit from Community Long-term care Other Subacute care Acute care Brain injury side Bilateral No paresis Other stroke Right brain Left brain Adm FIM cognition < <.0001 (continued) 85

9 Rehabilitation Length of Stay and Stroke Patient Outcomes M. Camicia et al. Table 3 (continued) Cognition FIM gain Motor FIM gain Parameter Estimate Standard Error p-value Estimate Standard Error p-value Adm FIM motor <.0001 Onset days (day) Facility type Freestanding hospital < <.0001 Unit in acute care R Severe LOS < <.0001 Age (year) < < < Marital status Married Single Race Black Hispanic Other White Admit from Community Long-term care Other Subacute care Acute care Brain injury side Bilateral No paresis Other stroke Right brain <.0001 Left brain Adm FIM cognition < <.0001 Adm FIM motor < <.0001 Onset days (day) < < Facility type Freestanding hospital < <.0001 Unit in acute care R UDSMR, Uniform Data System for Medical Rehabilitation; Adm, admission; LOS, length of stay, R 2, R-square statistics. 86

10 M. Camicia et al. Rehabilitation Length of Stay and Stroke Patient Outcomes Table 4 Length of stay at inpatient rehabilitation facilities by discharge destination, UDSMR Grouping Discharge Destination Community Other p-value Total N 3,681 1,100 <.0001 Mean SD Median (IQR) 14 (9 20) 20 (14 25) Mild N Mean SD Median (IQR) 8 (6 10) 10.5 (8 13) Moderate N 1, <.0001 Mean SD Median (IQR) 13 (9 16) 15 (13 19) Severe N 1, Mean SD Median (IQR) 21 (16 26) 22 (16 26) UDSMR, Uniform Data System for Medical Rehabilitation; SD, standard deviation, IQR, interquartile range. status. Age and functional measures (FIM motor and cognitive) at IRF admission are the major factors that determine the CMG and thus grouping of patients in the mild, moderate, and severely impaired groups. Due to higher degree of impairment, the moderate and severely impaired groups had the most opportunity to experience greater functional gain during the IRF stay and the mildly impaired group had the highest chance of returning to the community after IRF stay. A longer IRF LOS provides opportunity for additional treatment for patients with greater potential for improvement, but may not have a significant impact on patients with less potential for improvement or for those who have already approached their maximum recovery. This illustrates the need to closely manage the patient s progress and facilitate the transition to the next level of care when indicated. IRF LOS contributes directly to the poststroke care cost. Studies have reported that patient severity (Appelros, 2007; Elwood et al., 2009; Harvey et al., 1998), negative symptoms (Galynker et al., 1997), medical complications (Saxena et al., 2007), and functional status at IRF admission (Brock et al., 1997; Franchignoni et al., 1998; Harvey et al., 1998; Stillman, Granger, & Niewczyk, 2009; Tan, Heng et al., 2009) are the major determinants of IRF LOS. Some other nonmedical patient characteristics and organization factors are associated with duration Table 5 Logistic model for discharge to community, UDSMR Parameters Odds Ratio 95% Confidence Limits Mild LOS (day) Adm FIM cognition C Moderate LOS Age (year) Marital status Married Single Adm FIM cognition Adm FIM motor Region East West Center C Severe LOS Age (year) Race Black Hispanic Other White Prehospital living Alone Other Brain injury side Bilateral No paresis Other stroke Right brain Left brain Adm FIM cognition Adm FIM motor Facility Type Freestanding Hospital Unit in Acute Care (continued) 87

11 Rehabilitation Length of Stay and Stroke Patient Outcomes M. Camicia et al. Table 5 (continued) Parameters Odds Ratio 95% Confidence Limits Region East West Central C UDSMR, Uniform Data System for Medical Rehabilitation; Adm, admission; LOS, length of stay; C, C statistics. Table 6 Influential factors for length of stay at inpatient rehabilitation facilities, UDSMR Parameter Log-Transformed Length of Stay Estimate Standard Error p-value Marital status Married Single Primary pay source Blue Cross Commercial Medicaid Other Workers comp Medicare Case-mix group Minor <.0001 (CMG ) Moderate <.0001 (CMG ) Severe (CMG ) Adm FIM motor <.0001 Onset days (day) < R UDSMR, Uniform Data System for Medical Rehabilitation; Adm, admission. of IRF stay. For Medicare patients, length of stay is associated with payment for the IRF stay, and each year CMS publishes an expected LOS for all Medicare patients based on CMG and comorbidities (CMS, 2013). These strict rules associated with payment and LOS for Medicare patients may have some impact upon patient outcomes. Primary payer source was shown to be associated with LOS in all impairment groups. Payment for IRF in the United States under Medicare is heavily weighted on patient placement into CMGs. The system was constructed with financial incentives/disincentives for clinical practice patterns. Each CMG has an associated expected length of stay. The facility is paid equally for a patient regardless of the length of stay when a patient is discharged to the community (board and care, transitional living, or home). Payment is prorated for the length of stay up to the case rate when a patient is discharged to a skilled nursing facility (SNF) or acute care facility before their expected LOS. If a patient is discharged to a SNF at or beyond the CMG length of stay the facility receives full payment. However, commercial payment sources are usually paid on a per diem basis providing an incentive to keep patients longer in IRH than with Medicare payer. This may contribute to a longer LOS in this population. Since Medicare payment policy strictly defines average length of stay for all patients (CMS, 2013), a stratified analysis examining the relationship of LOS and gains among the three stroke groups comparing Medicare patients only to other payer sources was completed (data not shown). For the minor and severe stroke groups, no differences in the relationship between LOS and functional gains were found when stratified by payer source; there did seem to be an association among the moderate stroke patients, but the small sample sized limited the power to determine the nature of the relationship. Future analysis examining differences among outcomes comparing patients with and without Medicare will be imperative to further elucidate this association. In addition, Stineman et al. (2001) compared IRF LOS of stroke patients between U.S. Veterans Affairs (VA) and non-va healthcare system. The difference in patient population may contributed to the longer IRF LOS in the VA system, i.e., a higher proportion of single, separated, or divorced marital status; unemployed or retired as a result of disability; and non-white race/ethnicity. Fewer incentives for cost containment and less support in patients home environments may also be the important determinants of a longer IRF stay. Tan, Heng et al., 2009 and Tan, Chong et al., 2009 reported their retrospective study at Singapore and observed that higher socioeconomic status and living with nonimmediate family before stroke were associated with a longer IRF LOS; lack of caregiver training; and postdischarge support contributed to the delayed IRF discharge. In this study, more severely impaired patients and a lower motor function at IRF admission were associated with a longer IRF LOS; married patients 88

12 M. Camicia et al. Rehabilitation Length of Stay and Stroke Patient Outcomes were associated with a short IRF LOS. These agree with the study findings in the literature. In addition, patients with primary pay of Medicare and those admitted early after stroke onset had a shorter IRF LOS. Limitations This is a secondary data analysis using data obtained from the UDSMR database. There were no direct measures of stroke severity, caregiver, and type, and hours of treatment received at an IRF. However, a number of measures in the database served as proxy measures of severity and medical conditions, including CMG measure coded based on diagnosis, age, initial functional measures at IRF admission, side of brain injury, and comorbidity tier. Marital status and prehospital living alone or with others provided certain caregiver information. CMS requires adherence to the 3-hour rule for IRFs in the United States, which means patients are to receive at least 3 hours of therapy at least 5 days or a minimum of 15 hours over 7 days per week (CMS, 2013). The authors assume that the stroke populations under study received standard care during their IRF stay. In addition, while the data itself were collected prospectively, the authors examined it retrospectively. This may lead to the inability to tease out important factors that occurred during the stay that may have affected the length of stay within the IRF for certain patients, such as medical complications or personal/family issues finding the correct discharge disposition. Future prospective research should be considered for these reasons. Conclusions After controlling for FIM at IRF admission and other covariates under study, the IRF LOS associated moderately with the functional gain for the study sample. The associations varied by stroke impairment severity, positive for more severely affected patients, and negative for mildly affected patients. In addition, the severely impaired patients with a longer IRF LOS were more likely to be discharged to the community. Matching patient needs with the appropriate resources to ensure optimal patient outcomes is necessary to ensure that health care is effective and efficient. This study provides evidence for the care of stroke patients at IRF setting and the need for effective management of resources to optimize functional outcomes. Nurses with rehabilitation training, knowledge, and experience are Key Practice Points Matching patient needs with the appropriate resources ensures more optimal patient outcomes. IRS LOS appears associated with functional outcomes. Severely impaired patients with a longer IRF LOS were more likely to be discharged to the community. Nurses with rehabilitation training, knowledge, and experience are most skilled in facilitating care transitions for individuals with disabling conditions such as stroke by ensuring more efficient and effective use of healthcare resources. most skilled in facilitating care transitions for individuals with disabling conditions such as stroke to promote the most efficient and effective use of healthcare resources (Camicia et al., 2014). Further studies on factors such as payer source, care management programs, social determinants, and the structure and process of care impacting clinical outcomes of patients who have experienced a stroke are needed to ensure efficient, effective, and affordable postacute care. NOTE 1 FIM and UDSMR are trademarks of Uniform Data System for Medical Rehabilitation, a division of UB Foundation Activities, Inc. References Appelros, P. (2007). Prediction of length of stay for stroke patients. Acta Neurologica Scandinavica, 116(1), Beninato, M., Gill-Body, K.M., Salles, S., Stark, P.C., Black- Schaffer, R.M., & Stein, J. (2006). Determination of the minimal clinically important difference in the FIM instrument in patients with stroke. Archives of Physical Medicine and Rehabilitation, 87(1), Brock, K., Robinson, P., Simondson, J., Goldie, P., Nosworthy, J., & Greenwood, K. (1997). Prediction of length of hospital stay following stroke. Journal of Quality in Clinical Practice, 17(1), Camicia, M., Black, T., Farrell, J., Waites, K., Wirt, S., & Lutz, B., with the Association of Rehabilitation Nurses Task Force (2014). The essential role of the rehabilitation nurse in facilitating care transitions: A white paper by the Association of Rehabilitation Nurses. Rehabilitation Nursing, 39, Center for Medicare and Medicaid Services (2013). Inpatient rehabilitation facility prospective payment system for federal 89

13 Rehabilitation Length of Stay and Stroke Patient Outcomes M. Camicia et al. fiscal year Final rule. Federal Register, 78(151), Elwood, D., Rashbaum, I., Bonder, J., Pantel, A., Berliner, J., Yoon, S.,... Bansal, A. (2009). Length of stay in rehabilitation is associated with admission neurologic deficit and discharge destination. PM & R: The Journal of Injury, Function and Rehabilitation, 1(2), Franchignoni, F., Tesio, L., Martino, M.T., Benevolo, E., & Castagna, M. (1998). Length of stay of stroke rehabilitation inpatients: Prediction through the functional independence measure. Annali dell Istituto Superiore di Sanita, 34(4), Galynker, I., Prikhojan, A., Phillips, E., Focseneanu, M., Ieronimo, C., & Rosenthal, R. (1997). Negative symptoms in stroke patients and length of hospital stay. The Journal of Nervous and Mental Disease, 185(10), Granger, C.V., Cotter, A.C., Hamilton, B.B., & Fiedler, R.C. (1993). Functional assessment scales: A study of persons after stroke. Archives of physical medicine and rehabilitation, 74(2), Granger, C.V., Markello, S.J., Graham, J.E., Deutsch, A., & Ottenbacher, K.J. (2009). The Uniform Data System for Medical Rehabilitation: Report of patients with stroke discharged from comprehensive medical programs in American Journal of Physical Medicine & Rehabilitation, 88(12), Harvey, R.L., Roth, E.J., Heinemann, A.W., Lovell, L.L., McGuire, J.R., & Diaz, S. (1998). Stroke rehabilitation: clinical predictors of resource utilization. Archives of physical medicine and rehabilitation, 79(11), Medicare Payment Advisory Commission (MedPAC) (2008). Rehabilitation facilities (inpatient) payment system. Washington, DC: MedPAC. O Brien, S.R., Xue, Y., Ingersoll, G., & Kelly, A. (2013). Shorter length of stay is associated with worse functional outcomes for Medicare beneficiaries with stroke. Physical Therapy, 93(12), Saxena, S.K., Koh, G.C., Ng, T.P., Fong, N.P., & Yong, D. (2007). Determinants of length of stay during post-stroke rehabilitation in community hospitals. Singapore Medical Journal, 48(5), Stillman, G., Granger, C., & Niewczyk, P. (2009). Projecting function of stroke patients in rehabilitation using the AlphaFIM instrument in acute care. PM &R: The Journal of Injury, Function and Rehabilitation, 1(3), Stineman, M.G., Ross, R.N., Hamilton, B.B., Maislin, G., Bates, B., Granger, C.V., & Asch, D.A. (2001). Inpatient rehabilitation after stroke: A comparison of lengths of stay and outcomes in the Veterans Affairs and non- Veterans Affairs health care system. Medical Care, 39(2), Tan, W.S., Chong, W.F., Chua, K.S., Heng, B.H., & Chan, K.F. (2009). Factors associated with delayed discharges after inpatient stroke rehabilitation in Singapore. Annals of the Academy of Medicine Singapore, 39(6), Tan, W.S., Heng, B.H., Chua, K.S., & Chan, K.F. (2009). Factors predicting inpatient rehabilitation length of stay of acute stroke patients in Singapore. Archives of physical medicine and rehabilitation, 90(7), Zisberg, A., Shadmi, E., Sinoff, G., Gur-Yaish, N., Srulovici, E., & Admi, H. (2011). Low mobility during hospitalization and functional decline in older adults. Journal of the American Geriatrics Society, 59, Earn nursing contact hours Rehabilitation Nursing is pleased to offer readers the opportunity to earn nursing contact hours for its continuing education articles by taking a posttest through the ARN website. The posttest consists of questions based on this article, plus several assessment questions (e.g., how long did it take you to read the articles and complete the posttest?). A passing score on the posttest and completing of the assessment questions yield one nursing contact hour for each article. To earn contact hours, go to and select the Education page. There you can read the article again, or go directly to the posttest assessment by selecting RNJ online CE. The cost for credit is $10 per article. You will be asked for a credit card or online payment service number. Contact hours for this activity are available at no cost to ARN members for 60 days following the date the CE posttest is first available, after which time regular pricing will apply. The contact hours for this activity will not be available after April 30, The Association of Rehabilitation Nurses is accredited as a provider of continuing nursing education by the American Nurses Credentialing Center s Commission on Accreditation (ANCC-COA). 90