top of page

Reducing Physical Therapy Consults for Patients with High Functional Mobility in the Acute Medical Inpatient Setting: A Difference-in-Difference Analysis


Maylyn Martinez, MD [1], Matthew Cerasale, MD, MPH [1], Mahnoor Baig, BS [2], Joshua K. Johnson, PT, DPT, PhD [3], Claire Dugan [4], MD, Ameerah Brown [5], Marla Robinson, MSc, OTR/L [6], Andrew Schram, MD, MBA [1], S. Ryan Greysen, MD, MHS, MA [7], David Meltzer, MD, PhD [1] Vineet M Arora, MD, MAPP [3]


1 Section of Hospital Medicine, Department of Medicine, University of Chicago, Chicago, IL

2 University of Illinois, Chicago IL

3 Department of Physical Medicine and Rehabilitation, Cleveland Clinic, Cleveland, OH

4 Section of General Internal Medicine, Department of Medicine, University of Chicago, IL

5 North Park University, Chicago IL

6 Inpatient Therapy Services Department, University of Chicago, Chicago, IL

7 Section of Hospital Medicine, Department of Medicine, Hospital of the University of Pennsylvania, Philadelphia, PA


Physical therapy (PT) interventions are important for preventing and treating the physical deconditioning that occurs in the inpatient setting and hospital-associated disability (HAD), which one third of hospitalized patients experience.(1). During hospitalization, patients spend 87 - 100% of their time in bed.(2,3) This extremely low mobility and resultant disability are associated with increased risk of readmission, institutionalization, permanent disability, and even death. (4,5)  Many patients who spend so much time in bed are able still able to ambulate safely alone or with appropriate unit-based mobility initiatives and support from nursing. Patients with more physical deconditioning, however, may be unable to get out of bed without physical therapists and require their specific skillset for physical rehabilitation. Given that PT is a constrained resource in most hospital settings, it is important to focus their services on those with physical deconditioning requiring skilled PT. In many cases, therapists are consulted not just for their rehabilitative skills but because of their unique role in helping to arrange safe discharges and make discharge level-of-care recommendations. (6-8) Frequently, this can lead to inappropriate requests for PT to evaluate patients with high mobility and functional independence9. These patients may not require treatment with skilled therapy during their hospitalization nor the expertise of a physical therapist to determine a safe discharge level of care. This misallocation of therapy time can contribute to the strain on PT resources. Unneeded evaluations can waste hundreds of PT hours and decrease the amount of time that deconditioned, at-risk patients spend rehabilitating with therapists.

Having standardized language to discuss basic functional mobility in the inpatient setting can impart to care teams a more nuanced understanding of a patients’ independence with mobility. This, in turn, may improve awareness of which patients may benefit from skilled PT during hospitalization, aiding efforts to properly allocate therapy services in the hospital setting. The Activity Measure Post-Acute Care “6-Clicks” Mobility Short Form (AM-PAC mobility) score is a validated tool for measuring functional mobility. It has been used to predict discharge destination within 48 hours of admission, allocate PT on neurosurgical and neurology services, and to define potential overutilization of PT on hospital medicine services.(9-12) However, no studies have evaluated the effectiveness of an AM-PAC-based clinical decision support tool as a sole intervention to optimize utilization of inpatient physical therapy. We aimed to improve inpatient skilled PT utilization using validated AM-PAC mobility score cutoffs for creating a clinical decision support tool embedded into the daily hospital medicine workflow.



Study Design and Setting

We report on our study design, data analysis and outcomes following the Standards for Quality Improvement Reporting Excellence 2.0.(13) We conducted a prospective study of all patients admitted to the hospital medicine services and general internal medicine (GIM) resident teaching services at the University of Chicago Medical Center from October 2018 – November 2021. The included hospital medicine services are covered by attending hospitalists with one service involving a GIM resident. The included GIM services are covered by GIM attendings and residents only. The populations cared for on these services consisted of general medicine, transplant (renal and lung), and cardiology patients at the medical-surgical, telemetry, and step-down level of care. Hospital medicine hepatology and liver transplant services were excluded to align with the control group who does not provide care for these populations. Hospital medicine oncology services were excluded due to an auto-order for skilled PT in their specific admission order set. All patients were hospitalized for > 48 hours.  Patients who left AMA, died, discharged to hospice, another hospital, or an inpatient psychiatric facility were excluded. For the remaining patients, we obtained age, sex, race, ethnicity, admission and discharge date and admission and discharge AM-PAC score.


Clinical Decision Support Tool

In our institution, AM-PAC scores are assessed using the 6-Clicks basic mobility short form (version 1.0) and documented by our nursing staff at the time of admission then once per shift (every 12 hours) for every hospitalized patient. The AM-PAC 6-Clicks mobility short form assesses “difficulty with” and “help required” for a patient to complete rolling in bed, sitting up at the edge of bed, moving from the bed to a chair, standing up from a chair, walking within a room, and climbing 3-5 stairs. Our nursing staff has great familiarity with the AM-PAC score, which is part of their daily patient assessments. However, the ordering providers (Hospital Medicine physicians, GIM physicians, and advanced practice providers) in our institution did not have any familiarity with the score. We hypothesized that this lack of awareness was related to an observed high proportion (38%) of all PT referrals for our Hospital Medcine patients being for those with high functional mobility who were ultimately discharge to home. We deemed these “potentially inappropriate”.(9)

To educate these clinicians, a clinical decision support tool was designed based on an AM-PAC cutoff score (> 18) that has been shown to be a strong predictor of discharge to home.(9,11) The tool was then embedded in hospital medicine “History & Physical” and “Progress Note” templates. The tool is designed so that the most recent AM-PAC score auto-populates in the assessment and plan in a list of other “things to know for all patients” (diet, DVT prophylaxis, consults, discharge plan). As with the other items in the list, a pop-up box appears showing providers a potentially appropriate mobility management plan based on current AM-PAC scores [PT ordered if AM-PAC ≤ 18, PT not ordered if AM-PAC > 18 (in which case patient will ambulate with nursing), no PT ordered if patient at functional baseline, AM-PAC > 18 but PT ordered due to new significant functional deficit, and a free text option]. To isolate the educational effect of clinical decision support, no further educational sessions about mobility scores or PT consultation practices was provided to clinicians outside of that provided within the tool. No announcements were made about the rollout of the tool. Since hospital medicine and general internal medicine log in to different departments in the Epic electronic medical record, the clinical decision support tool was only visible to hospital medicine providers. Therefore, the general internal medicine service served as a control.


Outcomes and Predictors

The primary outcome was “PT misallocation” defined as a PT consult for a patient with an admission AM-PAC score >18. Our secondary outcome was change in functional mobility during hospitalization measured as discharge AM-PAC score – admission AM-PAC score, termed “delta AM-PAC”, and calculated using AM-PAC t-scale scores.(14)  Specifically, we assessed delta AM-PAC for patients with high mobility (i.e. > 18) at the time of admission who did not receive referral for skilled PT at any time during hospitalization to assess for any ill effects of discouraging skilled PT in this group. Group allocation for comparison of outcomes was based on time (pre- vs. post-implementation of the tool) and group (treatment vs. control).


Statistical analysis

Baseline data was collected for fourteen months prior to implementation of the tool (10/1/2018 – 2/28/2020). Tool rollout occurred on February 12th, 2020. The post-intervention period was planned for 12 months but was extended to twenty months to maximize the amount of time the tool would be evaluated outside of the peaks of the COVID-19 pandemic (3/1/2020 – 10/21/2021) given possible changes to PT ordering practices and PT sessions that may have occurred throughout the hospital during pandemic surges. Pearson’s chi squared test was used to evaluate the association between implementation of the clinical decision support and PT consult rate for patients with high basic mobility. Two-sample T-test was used to assess mean change in mobility (delta AM-PAC) during hospitalization for patients admitted with high AM-PAC scores as well as for differences in mean age and length of stay (LOS) within and between groups pre- and post-intervention.  To examine the effect of the clinical decision support tool on PT misallocation we conducted a difference in difference analysis. A multivariate logit regression model was used to estimate the change in PT consults for patients who had high basic mobility at the time of admission using indicator variables for an interaction between group (control vs. treatment) and time (preintervention vs. postintervention). The model adjusted for age, sex, race, ethnicity, LOS, and change in mobility. We used the Stata logit command and did not cluster standard errors given that measurements were not repeated over time in the same individuals. P- values of < 0.05 were considered statistically significant. All analyses were performed using Stata statistical software, release 17 (StataCorp LLC). The study was determined to be a quality improvement study and, therefore, did not require IRB approval.



During the study period 20,810 admissions were eligible for the study. Table 1 reports characteristics of the two groups. Compared to the pre-intervention period, the treatment group was slightly older with a slightly higher proportion of Black patients, and lower basic mobility at the time of admission during the postintervention. The control group was slightly older with longer LOS and lower basic mobility at admission postintervention as compared to pre-intervention. Prior to the intervention, 18.5% of patients with high mobility in the treatment group were referred for skilled physical therapy, which decreased to 16.6% in the postintervention period [X2(1) = 7.01; p < 0.01]. Skilled PT referrals for the control group significantly increased in the postintervention period [18.1 % vs. 20.5%; X2(1) = 8.21; p < 0.01].

There was a statistically significant interaction between the group and time variables. In the multivariate regression model, the treatment group had lower odds of PT referrals for patients with high basic mobility [OR 0.74; 95% CI 0.64 – 0.86] controlling for age, LOS, change in mobility, race, and ethnicity (Table 2). The difference in difference analysis based on this model showed a statistically significant decrease in the marginal probability of PT referrals for patients with high basic mobility in the treatment group during the post intervention period [dy/dx -0.07; 95% CI -0.11, -0.04] (Figure 1). Compared to pre-intervention, there was no statistically significant difference in mean change in mobility during hospitalization for patients with high basic mobility not receiving PT referral.



Our study suggests that mobility score-based clinical decision support alone can decrease PT consults for patients with high baseline functional mobility. We implemented a tool within the hospital medicine workflow without any further education or announcements and demonstrated that this was independently associated with a 7% decrease in likelihood of PT referrals for patients without skilled therapy needs. This effect was sustained for at least one year. Even with the decrease in PT consults in this high mobility group, there was no negative effect on their mobility during hospitalization demonstrating the lack of necessity for skilled inpatient PT for them.

Considering PT scarcity and the financial and medical costs of HAD and low mobility, maximizing the amount of time that therapists can spend rehabilitating patients with functional impairment should be a top priority for healthcare systems. Increasing rehabilitation time for hospitalized patients has been shown to improve functional status and increase the likelihood that patients can discharge to home instead of to a post-acute care nursing facility.(15-21) In 2013, 22.3% of all hospital discharges involved use of post-acute care services22. Decreasing the proportion of these services that are used solely for physical rehabilitation after acute medical illness could have significant effects from the patient to the payor level. Accurate allocation of PT to hospitalized patients at risk of HAD and physical deconditioning could be one key step in this effort so ensuring that providers understand when to consult inpatient physical therapy is an important goal.

While ours is the first study to specifically examine the effect of an AM-PAC-based decision support tool on referral practices, Chou et al. described a similar use of the AM-PAC mobility score in another health system and its effect on patient outcomes. They demonstrated that an electronically generated referral to PT based on an AM-PAC mobility score cutoff (≥ 20) was associated with decreased readmission and death among patients with stroke.(23) That they selected an AM-PAC mobility cut-off score of 20, compared to our use of 18, suggests that there is not yet a gold-standard AM-PAC mobility cut-off score to differentiate patients who require skilled PT from those who do not. That said, an optimal cut-off score may be within this relatively tight range given the use of similar scores in two distinct health systems. In either case, the use of a cut-off score for referral to PT is not without risk. A score of 18 suggests that patients do still need some physical assistance for basic mobility tasks. Depending on the patient’s functional prognosis relative to his or her prior level of function, home set-up, and the assistance from others that he or she will have if discharged home, that level of function may not be adequate. Thus, while the clinical decision support tool we examined in this study helps providers to critically evaluate the need for skilled PT interventions, other factors remain relevant to this decision.

Some limitations to our study should be considered when interpreting our results. First, we implemented our tool at the start of the COVID-19 pandemic which, on its own, may have led to behavior changes that could have affected PT consultation practices. However, by employing a difference in difference study design we have an appropriate control group for comparison during the same time period. During that time, this group was exposed to the same hospital environment as the treatment group limiting the effect that this would have on our results. Additionally, we measured the effect of our tool for an extended period of time beyond the primary pandemic surges that created protocol and behavior changes in our hospital.  Second, the groups had statistically significant differences in important covariates before and after the intervention. However, the direction of difference minimized the clinical relevance of these discrepancies. For example, the control group had a significantly lower mean LOS in both time periods, which would have been expected to be associated with fewer unneeded PT consults, not more, which is what occurred. Last, our clinical decision support tool was not designed to prevent a PT consult at the point of ordering. It was designed to provide daily and sustainable education to ordering providers, which likely limited its impact. We designed the tool this way because with the increasing complexity of medical decision-making, it can be difficult to optimize delivery of new knowledge to clinicians already experiencing information overload. With high hospitalist staff turnover rates, leadership is tasked with maintenance of knowledge and efficiency in their medical group.(24,25) Some may consider onboarding as an opportunity to educate on standardized mobility language and protocols, but this risks knowledge dissipation over time.  Separate educational sessions may be an option but are time and resource intensive, will only be useful to those able to attend, and also have the risk of fading knowledge with time. These issues emphasize the value of developing innovative ways to educate clinicians in effective yet sustainable ways.  Our method of daily reminders through clinical decision support is one way to accomplish this.

Despite these limitations, our study underscores the importance of performing standardized mobility assessment at the time of admission for all hospitalized patients and the utility of mobility score-based clinical decision support to educate clinicians on essential considerations when prescribing inpatient physical therapy services.






1) Loyd C, Markland AD, Zhang Y, Fowler M, Harper S, Wright NC, Carter CS, Buford TW, Smith CH, Kennedy R, Brown CJ. Prevalence of Hospital-Associated Disability in Older Adults: A Meta-analysis. J Am Med Dir Assoc. 2020 Apr;21(4):455-461.e5. doi: 10.1016/j.jamda.2019.09.015. Epub 2019 Nov 14. PMID: 31734122; PMCID: PMC7469431.


2) Brown C.J, Friedkin RJ, Inouye SK. Prevalence and outcomes of low mobility in hospitalized older patients. J Am Geriatr Soc. 2004;52:1263-1270. https://


3) Fazio S, Stocking J, Kuhn B, et al. How much do hospitalized adults move? A systematic review and meta-analysis. Appl Nurs Res. 2020;51:151189. https://         


4) Zisberg A, Shadmi E, Gur-Yaish N, Tonkikh O, Sinoff G. Hospital-associated functional decline: the role of hospitalization processes beyond individual risk factors. J Am Geriatr Soc. 2015;63:55-62. jgs.13193


5) Brown CJ, Redden DT, Flood KL, Allman RM. The underrecognized epidemic of low mobility during hospitalization of older adults. J Am Geriatr Soc. 2009;57(9):1660-1665.


6) Gustavson AM, Toonstra A, Johnson JK, Ensrud KE. Reframing Hospital to Home Discharge from “Should We?” to “How Can We?: COVID‐19 and Beyond. J Amer Ger Soc. 2021;69(3):608-609.


7) Kadivar Z, English A, Marx BD. Understanding the Relationship Between Physical Therapist Participation in Interdisciplinary Rounds and Hospital Readmission Rates: Preliminary Study. Phys Ther. 2016;96(11):1705-1713.


8) Smith BA, Fields CJ, Fernandez N. Physical Therapists Make Accurate and Appropriate Discharge Recommendations for Patients Who Are Acutely Ill. Phys Ther. 2010;90(5):693-703.


9) Martinez, M., Cerasale, M., Baig, M., Dugan, C., Robinson, M., Sweis, M., Prochaska, M., Schram, A., Meltzer, D. and Arora, V.M. (2021), Defining Potential Overutilization of Physical Therapy Consults on Hospital Medicine Services. Journal of Hospital Medicine, 16: 553-555.


10) Jette DU, Stilphen M, Ranganathan VK, Passek SD, Frost FS, Jette AM. Validity of the AM-PAC “6-Clicks” inpatient daily activity and basic mobility short forms. Phys Ther. 2014;94(3):379-391.


11) Jette DU, Stilphen M, Ranganathan VK, Passek SD, Frost FS, Jette AM. AM-PAC “6-Clicks” functional assessment scores predict acute care hospital discharge destination. Phys Ther. 2014;94(9):1252-1261.

11)Young DL, Colantuoni E, Friedman LA, et al. Prediction of disposition within 48 hours of hospital admission using patient mobility scores. J Hosp Med. 2020;15(9);540-543.                                   

12) Probasco JC, Lavezza A, Cassell A, et al. Choosing wisely together: physical and occupational therapy consultation for acute neurology inpatients. Neurohospitalist. 2018;8(2):53-59.

13) Ogrinc G , Davies L , Goodman D , et al . Squire 2.0 (standards for quality improvement reporting excellence): revised publication guidelines from a detailed consensus process. BMJ Qual Saf 2016;25:986–92.doi:10.1136/bmjqs-2015-004411 pmid:


14) Jette AM, Tao W, Norweg A, Haley S. Interpreting rehabilitation outcome- measurements. J Rehabil Med. 2007;39(8):585-590.

15) Johnson JK, Rothberg MB, Adams K, Lapin B, Keeney T, Stilphen M, Bethoux F, Freburger JK. Association of Physical Therapy Treatment Frequency in the Acute Care Hospital With Improving Functional Status and Discharging Home. Med Care. 2022;60:444-452 

16) Chippala P, Sharma R. Effect of very early mobilisation on functional status in patients with acute stroke: a single-blind, randomized controlled trail. Clin Rehabil. 2016 Jul;30(7):669-75. doi: 10.1177/0269215515596054. Epub 2015 Jul 21. PMID: 26198890.

17) Pashikanti L, Von Ah D. Impact of early mobilization protocol on the medical-surgical inpatient population: an integrated review of literature. Clin Nurse Spec. 2012 Mar-Apr;26(2):87-94. doi: 10.1097/NUR.0b013e31824590e6. PMID: 22336934.

18) Schweickert WD, Pohlman MC, Pohlman AS, Nigos C, Pawlik AJ, Esbrook CL, Spears L, Miller M, Franczyk M, Deprizio D, Schmidt GA, Bowman A, Barr R, McCallister KE, Hall JB, Kress JP. Early physical and occupational therapy in mechanically ventilated, critically ill patients: a randomised controlled trial. Lancet. 2009 May 30;373(9678):1874-82. doi: 10.1016/S0140-6736(09)60658-9. Epub 2009 May 14. PMID: 19446324.

19) Landefeld CS, Palmer RM, Kresevic DM, Fortinsky RH, Kowal J. A randomized trial of care in a hospital medical unit especially designed to improve the functional outcomes of acutely ill older patients. N Engl J Med. 1995 May 18;332(20):1338-44. doi: 10.1056/NEJM199505183322006. PMID: 7715644.

20) Rajendran V, Jeevanantham D, Falk D. Effectiveness of Weekend Physiotherapy on Geriatric In-Patients' Physical Function. Gerontol Geriatr Med. 2022 May 4;8:23337214221100072. doi: 10.1177/23337214221100072. PMID: 35529693; PMCID: PMC9073106.

21) Kato M, Mori Y, Watanabe D, Onoda H, Fujiyama K, Toda M, Kito K. Relationship between average daily rehabilitation time and decline in instrumental activity of daily living among older patients with heart failure: A preliminary analysis of a multicenter cohort study, SURUGA-CARE. PLoS One. 2021 Jul 2;16(7):e0254128. doi: 10.1371/journal.pone.0254128. PMID: 34214129; PMCID: PMC8253396.

22) Agency for Healthcare Research and Quality; An all-payer view of hospital discharge to post-acute care, 2013;, Accessed March 16th, 2022

23) Chou A, Johnson JK, Jones DB, Euloth T, Matcho BA, Bilderback A, Freburger JK. Effects of an electronic health record-based mobility assessment and automated referral for inpatient physical therapy on patient outcomes: A quasi-experimental study. Health Services Research. October 2022. ePub ahead of print.

24) Pappas, MA, Stoller, JK, Shaker, V, Houser, J, Misra-Hebert, AD, Rothberg, MB. Estimating the costs of physician turnover in hospital medicine. J Hosp Med. 2022; 17: 803- 808. doi:10.1002/jhm.12942

25) McGrath, Bridget PA-C; Konold, Victoria MD; Forbes, Meggan APN-C; Murphy, Elizabeth MD; Cerasale, Matthew MD, MPH; Schram, Andrew MD, MBA. The 90-day orientation: An onboarding strategy for hospitalist PAs and NPs. Journal of the American Academy of Physician Assistants 34(9):p 52-55, September 2021. | DOI: 10.1097/01.JAA.0000758228.45700.9c

bottom of page