Analyses of 3,551 Admissions


  • Surya Shah, PhD, OTR, FAOTA,
  • Christine Tartaro, PhD,
  • Felicia Chew, MS, OTR,
  • Mike Morris, MBA, MSE


Objective: To integrate function in the electronic medical records (EMR) and to evaluate the functional outcome for the skilled nursing facilities (SNF) rehabilitation and establish a benchmark of expected outcomes. Data Sources: Data were prospectively collected on admission and discharge functional status, duration of rehabilitation and discharge destinations for 3,551 patients with arthritis undergoing SNF rehabilitation in 831 centers across 28 US states.
Study Design: A prospective longitudinal study to collect functional abilities on admission and on discharge from SNF rehabilitation.
Data Collection: Participating occupational therapists from each of the centers submitted their pre-post intervention measurements to the central IT source.
Findings: The mean age of the patients was 72 ±11 years; the mean length of stay was 15 ±12 days. The mean Modified Barthel Index (MBI) admission and discharge scores were 62.46 ±14.46 (moderate dependency) and 92.16 ±12.56 (minimal dependency). The effectiveness of returning the rehabilitation potential was 81.81 ±23.79% and the effi rate was 2.75 ±1.77 Modifi Barthel Index points/day. A total of 3,296 (93%) patients were discharged to home, 157 (4%) went to another SNF, and 98 (3%) were discharged to an acute facility.
Conclusions: For the future, 37,000 patients across ten diagnostic categories across 36 states and 1,100 centers will be analyzed to establish similar SNF rehabilitation performance benchmarks.


The therapists, stakeholders, managers, and all administrative divisions now require the rehabilitation intervention outcomes for evaluating change status and for fiscal financial accountability. Standardized patient-report measures are few and far between in the rehabilitation of persons with trauma and disease in skilled nursing facilities (SNF) rehabilitation. The emphasis needs to be on patient- centered outcome measurement as an essential and integral part of research practice and integral part of clinical outcomes.1 It is considered essential to report the performance standards of the rehabilitation teams and important to meet the needs of quality of outcomes such as efficiency, effectiveness, and presence of a benchmark of the expected standards. SNF patients are considered medically more complex, are older, have lower functional ability and have limited social support than the comprehensive rehabilitation facilities and home healthcare.
Developing and reporting of rehabilitation outcomes of impairment have been one of the highest priorities of occupational therapists. 3-4 The principal investigator (PI) consulted and collaborated with a major provider of the SNF rehabilitation services across the US to help develop a feedback system incorporating a communication process across centers and a reporting system of the patient-oriented function and the degree of dependency change in one’s management. The Modified Barthel Index (MBI) developed as the functional outcome measurement5-7 was introduced in the SNF to evaluate the change in dependency needs, the rate of change and the resulting quality of life of persons with arthritis.
In this study, we report SNF functional outcomes for ICD-9- CM Code for osteoarthritis and other arthritis as a group. Osteoarthritis (OA) is the most frequent cause of disability among adults in the US. It is estimated that that 50 million adults were reported as having some form of arthritis making it the fourth most common cause of hospitalization.8 An estimated 27 million adults have confirmed osteoarthritis which is the leading indication for joint-replacement surgery with 905,000 knee and hip replacements in 2009 at a cost of $42.3 billion. There are a large number of patient-report measures to report change in dependency status and improved quality of life following medical rehabilitation. Of the array of standardized and non- standardized measures, we chose the MBI as a change-status measure to report our findings. The MBI has been extensively used for measuring the dependency needs at admission, at discharge, and at follow up and has been used for outcome pattern analyses globally.
The authors, 13 worked with a major skilled nursing provider of the rehabilitation services that has 1,100 rehabilitation centers across 36 states in the US and employs 7,500 highly-trained therapists, to help develop a feedback system incorporating a communication process across centers, a reporting system of patient function, the change in level of their dependency, and the living arrangement on discharge. The MBI was introduced to evaluate change and establish a benchmark of expected daily outcomes in patients with arthritis treated through SNF inpatient rehabilitation. The MBI was utilized as the best indicator of productivity that is not tied to a private organization.

Aims and Objectives of the Study

The aims of the study were to incorporate functional outcome measures into the electronic medical records (EMR) and from the generated database, extract the functional status on admission and on discharge scores, and the total score on all 10 items of the MBI for all 3,551 patients with arthritis following SNF inpatient rehabilitation. The study sorted, filtered, and analyzed the data, and following activities and participation, and reports their progress towards independence upon the completion of SNF rehabilitation. The aim also was to determine the rate at which the improvement was obtained and the degree of functional recovery that was possible to achieve in patients with arthritis for the Medicare “A” and Managed Care “A” providers. The outcome of 831 centers across 28 states would serve as a benchmark of expected SNF outcome.


The study was a prospective multicenter ex post facto pre-test post-test design to determine what the functional ability changes were over what length of rehabilitation stay and the degree and the rate at which the improvement in functional changes occurred.


The MBI 10 ADL functions include (i) ambulation and wheelchair, (ii) bathing, (iii) bladder control, (iv) bowel control, (v) dressing, (vi) feeding, (vii) hygiene, (viii) stair climbing, (ix) transfers, and (x) toileting. It is an empirically derived scale with proven internal consistency (Chronbach’s Alpha = 0 .927).5 The MBI evidence also reports its reliability, repeatability and validity. The values assigned to each item in the MBI are based on the amount of physical assistance required to perform the task. A score of zero (0) indicates the patient is totally dependent on others and a score of 100 indicates that the patient is independent of assistance from others.
The MBI Likert modifications to sensitize the measure did not violate any of the original conceptualizations, weighting assigned and the minimum and maximum values (Table 1). The new guidelines were published in the Journal of Clinical Epidemiology and the additional hints were added and published in the Compendium of quality-of-life measures.

Table 1. Modified Scoring for the Barthel Index (MBI)


A nationally recognized SNF rehabilitation provider employs 7,500 highly trained professionals and who partners with over 1,100 sites, providing inpatient rehabilitation primarily to older adults in 36 states. The PI established a partnership to establish a multicenter measurement system that could be used to determine costs based on the length of stays and to evaluate outcomes as measured by change in performance scores in ADL based not on the process but on length of inpatient rehabilitation stays.
In order to incorporate the functional status in the EMR, in the first instance an accurate response system was created. A system for clinical data integrity was then established with the help of the IT to maximize accuracy of data input Optima system, to allow electronic entry of admission and discharge evaluations in the EMR. A Webinar was constructed for all therapists to learn documenting evaluations. The data on patients with complex medical/surgical conditions were extracted and a pilot study of a sample of 234 patients was conducted to ensure the system was in place for larger data sets development.


In consultation with the SNF service directors, the data analyst and the central IT team assigned to collect data extracted all inpatients with arthritis classified under ICD-9-CM Code clinical modification (Table 2). The study only utilized the major provider of services under Medicare “A” and Managed Care “A.” Of 3,551 patients, 3,247 patients were diagnosed as having osteoarthritis and allied disorders, 231 patients with other arthritis, and 73 patients with rheumatoid arthritis.

Table2. ICD Classification of 3,551 patients (Clinical Modification – ICD – 9 – CM Code)


The data on functional scores on admission and discharge on all 10 ADL functions on 3,551 patients with arthritis ICD-9–CM–Code were extracted from 831 SNF centers. All analyses were completed using IBM SPSS 19 version. Descriptive statistics provided the demography, the admission and discharge scores on all 10 individual items of the MBI, the total scores, and the length of stay from admission to discharge from rehabilitation. Since the potential improvement for patients with high initial scores is lower than those with low initial scores, achievement of potential improvement was calculated. This is expressed as actual improvement divided by potential improvement and percentaged.
Achievement of Rehabilitation Potential = (Discharge MBI – Initial MBI) ÷ (100 – Initial MBI Score) X 100%.
Rehabilitation efficiency is the amount of improvement achieved divided by the duration of rehabilitation inpatient stay and expressed as the average increase in MBI Score per day:
Rehabilitation Efficiency = (Discharge MBI Score – Initial MBI Score) ÷ Duration of Inpatient Rehabilitation Stay.
Pearson’s bivariate correlation coefficients measured the correlation between each individual’s percentage improvements on the MBI scores. Paired-sample t-tests were completed for differences between two measures. A one-way ANOVA determined whether the differences in the MBI means of three general types of discharge locations were different.

Research Findings Participants and gender
Participants and gender

There were 3,551 patients with arthritis who were treated under Medicare “A” and Managed Care “A” by SNF inpatient rehabilitation across the USA. Since gender was not made mandatory, based on recorded gender status, 60 percent of the patients were women (2,136), 23 percent (818) were identified as men, and the patients’ gender was unknown for 17 percent (597). The average age of participants was 72 ± 11.25.

Length of Days in SNF therapy

The average length of stay in therapy was 15 ± 12.1 days with a range of 196 days including all outliers. All outliers were included in the outcome study as the aim was to establish a benchmark of expected outcome of the SNF rehabilitation.

The MBI Outcome Scores

Scores for all measures, with the exception of wheelchair transfers, improved during the SNF rehabilitation stays (Table 3). Paired samples t-tests examined whether patient MBI scores increased from admission to discharge. For wheelchair transfers, discharge MBI scores were actually lower than admission scores, and the differences were statistically significant (p <.001) (Table 4). This finding indicates that a large number of wheelchair-bound patients became ambulatory and lower discharge score indicates only a few who remained dependent on wheelchairs. Patients improved on all other tasks, and the mean MBI scores improved from 62.46 ± 14.44 to 92.16 ± 12.56 (Table 4). This improvement difference was significant at the p <.001 level. This improvement change in functional status indicates that patients came to SNF rehabilitation with moderate to severe dependency status requiring 20 hours of help to remain independently functioning and were discharged to their homes with minimal to no dependency.

Table 3. Mean Admission and Discharge Scores on all MBI Items (N=3,551)

Efficiency of Occupational Therapy and Rehabilitation

Achievement of rehabilitation potential was calculated using the following equation.
Achievement of Rehabilitation Potential Equation:

Patients scored a mean of 81.81 ± 23.79 for achievement of rehabilitation potential. This result indicates that if the total functioning for each task was a score of 100, the SNF rehabilitation was able to return nearly 82% functional restoration leaving 18% as unexplained. The unexplained percentage could reflect the makeup of an individual patient that could not be explained requiring other confounding variables to be considered.
Rehabilitation effi was calculated by subtracting the admission MBI scores from the discharge MBI scores and dividing the sum by the length of SNF inpatient rehabilitation stay. The mean rehabilitation efficiency scores for the participants were 2.75 ± 1.77. Thus daily improvement in function was 2.75 MBI points per day (Table 4).

Relationship between MBI admission scores and length of stay in therapy

Pearson’s r bivariate correlation coefficients were developed to measure the correlation between each individual’s percent improvements of MBI scores. There were significant relationships between all measures of the MBI and length of stay in therapy. For all measures of the MBI, shorter stays in rehabilitation were associated with greater improvements. All of the correlations were significant at the p <.01 level (Table 5).

Table 5. SNF Rehabilitation Efficiency and Effectiveness

Discharge location

Ninety-three percent (3,296) of patients were discharged to their home or to an independent living arrangement. Four percent (157) were released to another assisted living of their choice, long termcare, or another SNF and three percent (98) were transferred to an acute care, hospital or rehabilitation setting. These acute care transfers were related to the onset of another episode or event requiring discharge from SNF rehabilitation (Figure. 1).

Figure 1. Discharge Destination for 3,551 patients following SNF Rehabilitation

Table 4. Paired Samples t-tests for ADL items on the MBI (n=3,551)

*All of the tests

significant at the .001 level
**Since some standard deviations on admission scores were higher than the means non- parametric tests were conducted.

Table 6. Pearson’s r coefficients for length of stay and % improvement for MBI scores

**correlations are significant at the .01 level

Relationship between mean MBI Admission scores and discharge location

A One-Way Analysis of Variance was conducted to determine whether the differences in means for the three general types of discharge locations were different, and significant differences were found on every type of MBI admission score. Scheffe post hoc tests were conducted to determine how the scores for each discharge location differed. Patients discharged to their homes or an independent living arrangement had higher MBI admission scores for every measure with the exception of wheelchair transfers and performance on stairs. There was no difference in the MBI scores on stairs by discharge location. For wheelchair transfers, those discharged to a home setting had the lowest mean score (0.61) compared to patients discharged to a hospital or rehabilitation unit (0.98) and those going to a long-term care or assisted-living facility (1.07). These differences were significant at the p < .05 level (Tables 7, 8, and 9). MBI admission scores were generally higher for patients discharged to rehabilitation or hospital setting when compared to those released to a long-term care or assisted- living facility. The exceptions to this were for stair climbing where the LTC/assisted-living discharge group scored slightly higher than the rehabilitation/hospital group, but this difference was not statistically significant. As noted earlier, the LTC/assisted-living group had higher scores on the wheelchair assessments as well. The differences between these two groups were not statistically significant for bathing and dressing.


The purpose of this study was to bring the science to the community. The documentation for function in rehabilitation has been subjective, sporadic, unreliable, and manual in most part. It was impressed upon the stakeholders that objective measureable and quantifiable evaluations are not an inconvenience but a must in today’s reimbursement-driven society. As a result, quality- control mechanisms, mandatory fields for documentation, and greater comprehensiveness of demographic details would be incorporated to further ensure accuracy and more completeness in outcome data. Getting the function to become part of electronic magnetic record was a major achievement for 7,500 clinicians. In addition, the study was to provide a broad brush of expected improvement per day, of potential recovery that could be achieved for the administrators, for program leaders, for stakeholders and for cost-benefit analyses. We wanted to utilize the minimum data set available without the availability of many confounding variables such as those required in predicting a specific replacement surgery or for a sub category of hip, knee, and upper limb arthritis. The study was successful in establishing an accurate response system for functional outcome data, ensuring a system for clinical data integrity by checks and cross checks, constructing a Rehab Optima data-entry system to allow electronic entry of evaluations, and developing a Webinar as a quick reference point for confirming the accuracy of the scores.
The mean age of the patients was 72 ± 11 years; the mean length of stay was 15 ± 12 days. The mean MBI admission and discharge scores were 62.46 ± 14.46 (moderate dependency) and 92.16 ± 12.56 (minimal dependency). The effectiveness of returning the rehabilitation potential was 76.5% and the efficiency rate was 2.75 ± 1.77 MBI points/day. A total of 3,296 (93%) patients were discharged to their homes, 157 (4%) went to another SNF, and 98 (3%) were discharged to an acute facility. In a specific arthritis condition17 the immediate post- operative rehabilitation outcome reported change score was 26.9 and effi was 2.6 despite less comorbidity and less severity and early rehabilitation where the improvement is better from subsidence of edema and pain.18 Similar lower effectiveness and lower effi were reported for the SNF settings. Valach12 reported efficiency of as 0.43 points per day. In another study19 of patients with orthopedic problems, the change scores were 24.5, rate of improvement was 1.75 per day with 82.5% returning to their homes or to an independent living setting. Despite older age, greater comorbidities, social isolation and late onset to commencement of rehabilitation, the findings of the current study show superior outcomes for the patients in these SNF centers. In addition, the present study included all outliers such as patients with a 100-day length of stay since the goal was a realistic benchmark for the SNF rehabilitation.

Table 7. Mean MBI Admission Score Comparisons by Discharge Location

** significant at the .01 level; *** significant at the .001 level


The gender was unfortunately not stipulated as mandatory and hence 17% of the patients had their gender missing. This omission has been rectified for data entry by therapists from 2012 onwards. While the study achieved its goal of developing a broad brush of outcome, if the clinicians at large could be enticed to record premorbid status and variables that confound outcomes, the predictive formulae would be easy to develop using many regressions techniques.

Table 8. Mean MBI Discharge Score Comparisons by Discharge Location

**ANOVA is significant at the .01 level; ***ANOVA is significant at the .001 level

Future Directions

As the functional data system is in place and integrated within the EMR system, all 37,000 patients undergoing occupational therapy and rehabilitation in 2012-2013 would be evaluated using this standardized system.

Table 9. Mean Admission Score Comparisons by Discharge Locations on all MBI Items (N=3,551)


  1. French HP. Use of questionnaire-based outcomes for the measurement of activities, and participation in the physiotherapy management of hip osteoarthritis: A review. Phy Ther Rev 2006, 11:273-288
  2. Mallinson TR, Bateman J, Tseng HY, et al. A comparison of discharge functional status after rehabilitation in skilled nursing, home health, and medical rehabilitation settings for patients after lower-extremity joint replacement surgery. Archives of Physical Medicine & Rehabilitation 2011;92:712-720
  3. American Occupational Therapy Association. Occupational therapy practice framework: Domain and process. Am J Occup Ther, 2008;62:625-683
  4. Porter ME, Teisberg EO. Redefining health care. Boston: Harvard Business School, 2006 5..Shah S, Vanclay F, Cooper B. Improving the sensitivity of the Barthel Index for stroke rehabilitation. Journal of Clinical Epidemiology 1989;42:703-709
  5. Shah S. Modifi Barthel Index – assessment. In S. Salek (Ed). Compendium of quality of life instruments, (1-6), 1998. Boston: Wiley and Sons
  6. Shah S, Tartaro C, Chew F, et al.Rehabilitation effi and effectivenessin medical-surgical conditions: A pilot study of 234 patients. Physical and Occupational Therapy in Geriatrics 2012;30:288-300
  7. Center for Disease Control. Prevalence of specifi type of arthritis. Morbidity, Mortality Weekly Report, 2010;59:1261-1265
  8. Shah S, Tartaro C, Chew F, et al. Skilled nursing facilities functional rehabilitation outcome: Analyses of 2041 stroke admissions. International Journal of Therapy and Rehabilitation 2013;11:352-360
  9. Ohura T, Ishizaki T, Higashi T, et al. Reliability and validity of a performance evaluation toll based on the Modifi Barthel Index. International Journal of Therapy and Rehabilitation 2011;21:422-428
  10. Leung SOC, Chan CCH, Shah S. Development of a Chinese version modified Barthel Index – validity and reliability. Clin Rehabil 2007;21:912-922
  11. Valach L, Selz B, Signer S. Length of stay in the rehabilitation centre, the admission functional independence measure and the functional independence measure gain. International Journal of Rehabilitation Research 2004;27:135-143
  12. Shah S, Chew F, Durham D. Why Barthel to the Modified Barthel Index and implications for therapy: A Webinar. Philadelphia, Pennsylvania: Genesis Rehabilitation Hospital (ROX-MBI Webinar) 2009
  13. Keilhofner G, Hammel J, Finlayson M, et al. Documenting outcomes of occupationaltherapy: The center for outcomes research and education. American Journal of Occupational Therapy 2004;58:15–23
  14. Medicare Learning Network. Inpatient rehabilitation therapy services: Compliance with documentation requirements. Medicare Internal Control network 2012. Available at Medicare-Learning-Network-MLN/MLNProducts/ downloads/Inpatient_Rehab_Fact_Sheet_ ICN905643.pdf
  15. Kroll C, Richman N. Skilled nursing facility documentation, reimbursement and ethicsin practice. American occupational Therapy Continuing Education 2010. Maryland: AOTA. Available at aota/prodview.aspx?TYPE=D&SKU=4843
  16. Vincent KR, Vincent HK, Lee LW, et al. Outcomes in total knee arthroplasty patients after inpatient rehabilitation. American Journal of Physical Medicine and Rehabilitation 2006;85:482-489
  17. Crosbie J, Naylor J, Harmer A, et al. Predictors of functional ambulation and patient perception following total knee replacement and short-term rehabilitation. Disability and Rehabilitation 2010;32:1088-1098
  18. Kirk-Sanchez NJ, Roach KE. Relationship between duration of therapy services in comprehensive rehabilitation and mobility at discharge in patients with orthopedic problems. Physical Therapy 2001;81:888-895