The consequences of prolonged immobility during critical illness and the feasibility and associated benefits of mobilization during ICU stay are increasingly established (1–6). Data are conflicting as to whether increases in therapy during critical illness consistently improve patient outcomes (7–9), and the optimal amount of therapy is not yet established. Studies demonstrate mixed results among the outcomes of independent functional status and length of stay (LOS), and reduced prevalence of delirium (7–12). Although forthcoming studies investigate the feasibility of earlier initiation of physical therapy (PT) during critical illness (13), most have examined transitioning from “no” or “minimal” therapy to increased levels of therapy.
Among published studies of “early mobility,” baseline levels of activity were minimal and included PT starting a week after ICU admission (14), no PT (1115), passive range of motion exercises (6), less than 10 minutes of activity (16), or they did not report the amount of therapy administered as a variable or outcome (7). In comparison, the treatment groups in these mobility studies reported a wide range of therapeutic activity, including 20 minutes (1116), up to 7 days per week, sitting for 20 minutes tid (6), or approximately 40 minutes of therapy (8). One recent noteworthy study had an intervention arm that received up to 160 minutes of daily therapy (9), although this is the exception rather than the rule. In only a few studies was the quantity of mobility defined as a variable (891617). Most studies have focused on medical or mixed patients and less on surgical patients (101518).
At our institution, baseline therapy in our ICUs was already comparable to the treatment groups in most published studies of early mobility during critical illness (8916). In 2015, we implemented a quality improvement (QI) initiative designed to double available therapy time for critically ill patients from this baseline level to a level we termed “Enhanced Early Mobility.” This observational study examined the attributable effects from the QI initiative on short-term patient and quality outcomes resulting from doubling available therapy shifts.
MATERIALS AND METHODS
Our analysis is reported according to the Standards for Quality Improvement Reporting Excellence guidelines (19) (Supplemental Table 1, Supplemental Digital Content 1, http://links.lww.com/CCX/A110).
Data were extracted from the institutional electronic data warehouse (EDW) by a data scientist blinded to the goals of the analysis. EDW data at our institution includes all electronic medical record (EMR) entries, in addition to multiple other sources. It has been previously described and validated as an accurate and sufficiently complete source for research analysis in studies (2021). These data were then appended with a manually maintained ICU research database for additional outcomes such as comorbidities and past medical history, which has been previously described (22).
This study was a retrospective pre/post analysis of a QI initiative that increased available therapy staffing in two ICUs at a tertiary academic medical center. The study included a baseline period (September 8, 2014, to March 8, 2015), a 6-month period during which additional therapists were hired and trained (March 9, 2015, to September 7, 2015), and a 6-month postintervention period (September 8, 2015, to March 8, 2016). The authors sought and received Institutional Review Board approval (IRB_00084463 and AM_00025727) for observational research examining the effect of the QI intervention. The QI initiative was an increase in the quantity of current standard of care therapy within the ICUs. Accordingly, it was felt to be of probable patient benefit. Safety outcomes were monitored through the institutional safety reporting infrastructure as part of the QI process and reported as part of this analysis. No data were analyzed for this study until after the 6-month intervention period.
Eligibility for analysis was determined by patient location and date. Subjects were included if they were greater than or equal to 18 years old admitted to the cardiovascular ICU (CVICU) or surgical ICU (SICU) at the University of Utah from September 8, 2014, to March 7, 2015. Eligible patients were identified through an EMR query (Supplemental Fig. 1, Supplemental Digital Content 2, http://links.lww.com/CCX/A111). Patients were excluded if they were admitted during the 6-month intervening period between baseline and QI periods, if they had a second admission, if they had an ICU LOS less than 24 hours, or did not undergo therapy.
The primary QI intervention was to double the number of PT shifts available to patients in the two ICUs (Supplemental Fig. 2, Supplemental Digital Content 3, http://links.lww.com/CCX/A112), termed “Enhanced Early Mobility.” At baseline, PTs worked 14 shifts per week covering 28 beds between the two ICUs, and occupational therapists (OTs) worked eight shifts per week, resulting in approximately six PT sessions and four OT sessions of 50 minutes per patient per week. Each 50-minute therapy session included 20 minutes of physical activity plus 30 minutes of patient preparation and documentation by the therapist. The “Enhanced Early Mobility” phase consisted of doubling PT shifts.
The intervention doubled the therapy staff in the treatment ICUs from 14 to 28 PT shifts per week and from 8 to 15 OT shifts per week. The goal was to perform bid PT and once daily OT per patient per day. Details of the duration and manner in which therapy was administered was left to the judgment of the therapist. Staff were permanently assigned to the ICUs. The increase in staff was accomplished through new hires trained during the 6 months between data collection periods who were then assigned to the treatment groups. Patients were seen by PT/OT according to standard clinical practices. There were no interventions modifying the ordering or timing of therapy orders.
Outcome Measures and Covariate Selection
The primary outcome was administered PT time (min) per patient per day. PT time was defined as a discrete field within the PT note. OT time was not analyzed for this analysis. Secondary outcomes included ICU LOS (d), hospital LOS (d), change in functional mobility status from preadmission to discharge as defined by differences in the numerical value of the Activity Measure for Post-Acute Care [AM-PAC] score, and discharge location (ordinal). Other covariates were selected based on consistently used covariates from previously published clinical trials and observational studies of mobility among critically ill patients (11121723). These included age, sex, illness severity scores, Medicare Severity-Diagnosis Related Group (MS-DRG) weight, and number of surgeries (2425). In the analysis of change in functional mobility status, we adjusted for preadmission physical function.
Administered PT time was entered in “total” and “active” minutes in two discrete extractable fields within the EMR. Active time was time during which the patient was participating with movement (actively or passively). Total time also included patient preparation and documentation. Accuracy of each time is regularly confirmed with manual audits by Department of Physical Therapy staff. During the study, there was a change in the clinical reporting of therapy time. For consistency, affected time was manually adjusted to the baseline method of reporting for the analysis. Functional status was measured by the AM-PAC Short Form, was assessed by therapists at discharge, and was reported by patients or family for preadmission (26). The AM-PAC Short Form (“6-Clicks”) is a validated measure of patient physical function and mobility, which has high inter-rater reliability among physical therapists, convergent validity with the Johns Hopkins Highest Level of Mobility score, and has been shown to accurately predict hospital discharge. However, it has limited data in ICU settings and may have a floor effect in this setting (27–30). Scores range from 6 (fully impaired) to 24 (fully functional). Injury severity scores included the Charlson Comorbidity Index, the Acute Physiology and Chronic Health Evaluation (APACHE) II, and the Simplified Acute Physiology Score (SAPS) II. The number of surgeries were pulled from a previously described and validated research database, before and after intervention (22). Discharge location was dichotomized into “favorable/low level of care” (e.g., inpatient rehabilitation or home) and “unfavorable/high level of care” (e.g., skilled nursing facility [SNF], long-term acute care [LTAC], or other hospital) as an additional way to describe patient functional independence at discharge. Safety outcomes were assessed for all patients through a query of the institutional adverse event reporting system.
Descriptive statistics, including mean (sd), median (interquartile range), frequency, and percent were used to summarize patient characteristics. Categorical patient characteristics were descriptively compared between preintervention and postintervention phases using chi-square or Fisher exact tests. Continuous patient characteristics were descriptively compared using independent samples t tests or Wilcoxon Mann-Whitney U tests. Generalized linear models implementing segmented regression (SR) were used to estimate the change in outcomes as a result of the intervention with and without adjusting for patient characteristics (APACHE II, Charlson, SAPS II, age, sex, MS-DRG weight, number of surgeries). We also controlled for admission AM-PAC score when modeling change in AM-PAC. Our SR approach mirrors an interrupted time-series model framework, except rather than modeling data at an aggregate level, we analyzed the data at the patient level (31). As is commonly done for SR analyses, the reported outcome is the difference in the rate of change relative to the expected rate without any intervention. 95% CIs and p values were obtained directly from the models. Linear regression was used for the outcomes of mean therapy time per day and AM-PAC change, gamma regression with a log link was used for ICU LOS and total LOS due to distribution skew, and logistic regression was used for discharge location. Coefficients from gamma regression models were exponentiated and reported as percent change. Statistical analyses were conducted in R v.3.4 (R Foundation for Statistical Computing; Vienna, Austria). Significance was assessed at the 0.05 level and all tests were two-tailed.
The study included 1,515 ICU patients (Supplemental Fig. 2, Supplemental Digital Content 3, http://links.lww.com/CCX/A112). Patients increased from the baseline to the QI period by 15% overall, from 351 to 406 (CVICU), and from 352 to 406 (SICU). Table 1 presents the patient-level outcomes of the two treatment groups, before and after the intervention, stratified by unit. Baseline characteristics were unequal between the preintervention and postintervention groups. The postintervention group in the CVICU had a significant increase in patients in the lowest SAPS II tertile (10 vs 17%; p = 0.027) and APACHE II tertile (10 vs 22%; p < 0.001).
Figure 1 shows the increase in therapy time from pre-intervention to post-intervention. The total administered therapy time during each 6-month period increased in the CVICU by 60%, from 71,994 to 115,389 minutes, and in the SICU by 116% from 42,985 to 93,015 minutes. In the CVICU, the primary outcome, mean therapy minutes per patient per day increased minimally from 50 (43.9–60) to 57.4 (43.9–73.6). In the SICU, time increased from 48.3 (42.1–60) to 59.6 (43.2–83.9) (Table 2). After adjusting for covariates in the regression model, the intervention in the CVICU was associated with a nonstatistically significant 17% (95% CI, –4.9 to 43.9; p = 0.13) increase in therapy time (Table 3).The effect was stronger in the SICU at 26% (95% CI, –1 to 59.4; p = 0.06), but still not statistically significant when modeled as a percent change (Table 4). Covariate adjusted absolute minutes of therapy increased by 14.2 (95% CI, 1.53–26.61; p = 0.03) in the SICU and by 8.4 (95% CI, –2.74 to 19.24; p = 0.13) in the CVICU.
Short-Term Clinical and Quality Outcomes
Figure 2 and Supplemental Figure 3 (Supplemental Digital Content 4, http://links.lww.com/CCX/A113) show the change in ICU and total LOS from pre-intervention to post-intervention. In the CVICU, ICU LOS nonsignificantly decreased from 4 (3–7) to 3 (2–6). In the SICU, ICU LOS nonsignificantly increased from 7 (5–12) to 8 (5–13) (Table 2). After adjusting for covariates, in the CVICU the intervention was associated with a nonsignificant 27.4% decrease (95% CI, –52.5 to 10.3; p = 0.13) in ICU LOS (Fig. 2), and a nonsignificant 12.4% decrease (95% CI, –37.9 to 23.3; p = 0.45) in total LOS (Table 3 and Fig. 2). In the SICU, the adjusted ICU LOS nonsignificantly increased 19.9% (95% CI, –31.6 to 108.6; p = 0.52) (Fig. 2), but total LOS significantly increased 52.8% (95% CI, 1.0–130.2; p = 0.04) (Table 4; and Supplemental Fig. 3, Supplemental Digital Content 4, http://links.lww.com/CCX/A113).
As part of the SR model, the rate of therapy change over time is reported. Before the intervention, in both the CVICU and in the SICU, there was no significant trend of increase or decrease in therapy over time (Tables 3 and 4). After the intervention in the SICU, the model showed a significant but minimal decrease in therapy time by 0.07 minutes (95% CI, –0.13 to 0.01; p = 0.019), which corresponds to a 0.13% per day decrease throughout the intervention period.
Functional Status (AM-PAC Score)
After adjusting for baseline functional status (AM-PAC score) and other covariates, the intervention of increasing available therapy was not associated with a significant improvement in patient functional status in either CVICU patients (–2.1 [95% CI, –5.2 to 1.1; p = 0.20]) or in SICU patients (–0.8 [95% CI, –3.5 to 1.9; p = 0.55]) (Tables 3 and 4).
Among CVICU patients, the intervention was nonsignificantly associated with “favorable” discharge (defined as “home” or “inpatient rehabilitation” vs “LTAC” or “SNF” (odds ratio [OR], 2.6; 95% CI, 0.6–12.2; p = 0.22) (Table 3). Among SICU patients, the effect was stronger but still nonsignificant (OR, 3.6; 95% CI, 0.9–15.4; p = 0.08) (Table 4).
In the SICU, staff-reported safety events included one fall during the baseline period, and two falls during the QI period (p = 1.0). There were no events in the CVICU in either period.
The purpose of this study was to assess the effect of doubling available therapy during critical illness, including whether there was a patient or associated outcome benefit. The population was a heterogeneous mix of general surgical and cardiothoracic surgery patients. Overall, we found that our total therapy time increased, as was expected with the increase in staffing. There was an observed 15% increase in patients in the QI compared with baseline period, and the observed primary outcome of per patient per day therapy time did achieve adequate statistical significance. Additionally, we believe that an outcome such as LOS was likely influenced by unmeasured confounders and may not be an effective outcome measure in an observational analysis with a heterogeneous population. We feel this is supported by the wide skew of the LOS data (Fig. 2; and Supplemental Fig. 3, Supplemental Digital Content 4, http://links.lww.com/CCX/A113).
In response to the noted increase in total LOS among SICU patients, we anecdotally observed that the post-ICU (“floor”) therapists de-escalated therapy for post-SICU patients after the project began in order to prioritize their efforts on patients who had not been in the ICU. We suggest that this de-escalation of floor therapy may have influenced LOS. Additionally, we did not measure muscle bulk or function during the initiative. It is plausible that excessive therapy could lead to enough fatigue to prolong LOS in a population with an increased prevalence of sarcopenia, muscular weakness, and respiratory failure.
The data supporting early PT and mobility during critical illness is conflicting, with recent negative trials (7–9) and is not yet universally adopted (1532). Among published studies of early mobility, the control arms typically involve minimal (616) to no physical activity (111415). One recent study reported 48 minutes of PT within the control group (10), but we found no other studies with comparable activity. Likewise, within these studies, therapy within the intervention arms was at best comparable to or less than our preintervention therapy of about 20 minutes of active therapy per day (611141633). The recently published study by Schaller et al (10) reported 60 minutes of therapy within their treatment group, although the percentage of this time involving active patient movement was not reported. In our study, each session included an additional 30 minutes of therapist time per patient per day beyond the reported “active therapy,” amounting to approximately 60 minutes of therapy per patient per day. The additional time included coordination or care with various providers, review of medical records, and documentation.
We found that unit level AM-PAC scores did not change, confirming a possible floor effect during critical illness using the AM-PAC score (2334). A second possibility is that subpopulations of critically ill patients significantly benefit from increased therapy, but that others experience a ceiling effect.
The lack of an effect on ICU LOS is likely influenced by the diversity of the SICU population, which includes trauma, general surgery, orthopedic surgery, obstetrics, otolaryngology, and abdominal transplant surgery. Furthermore, in this population, the total hospital LOS increased by 16%. Investigating this, we observed that the increase in ICU therapy had the unintended consequence of post-ICU therapy being de-prioritized to post-SICU patients. If this contributed to a post-ICU LOS increase, this would further support the observation that therapy leads to shortened LOS within critically ill patients. Furthermore, as the therapists adjusted per patient therapy daily, rather than applying a fixed amount of therapy, this likely led to more impaired patients getting more therapy.
The importance of our study is that it helps to answer the question: “Will ICU patients continue to benefit from increasing physical therapy?” Guidelines recommend mobility and activity during critical illness, although studies suggest that in practice this is still rare (1532). Recommended activities include active and passive range of motion, and coordination and balance (3536) daily for 30 minutes and then progressing to 45 minutes bid (37). Despite this, we found no studies that achieved this final recommendation level. One study reported that the treatment group had PT bid upon reaching the 4th of six ascending levels of function, but this outcome is not clearly reported, and patients walked 1 day earlier than their first therapy session, suggesting a healthier population (33). Finally, no studies have examined the transition from “early mobility” to increasing levels of mobility during critical illness.
This study has several limitations. The major limitation was the increase in number of patients pre- and post-intervention which functionally diminished the administered therapy per patient. The second major limitation was the use of LOS among a heterogeneous population of ICU patients. To address this, we adjusted for covariates identified in recent publications of ICU mobility, although there may have been additional uncontrolled confounders. We additionally used a SR analysis, which examined how change in outcomes over time were affected by the intervention when adjusting for patient characteristics. However, the lack of a significant change in LOS may simply mirror the negative outcomes of multiple recent controlled trials of increasing therapy during critical illness. We excluded patients who did not have therapy documented. Although this was a limitation that could bias our results because we only analyzed patients who received the therapy, the QI initiative targeted treatment of every patient every day. We additionally excluded LOS less than 24 hours, and the incidence of a patient being admitted for 2 days without therapy was low.
Our study shows that among diverse cardiothoracic and surgical patients, doubling PT shifts is associated with increased total administered therapy time, but when distributed among a greater number of patients during the QI period, the increase per patient is tempered. Increasing therapy from a baseline of 50 minutes per patient per day was not associated with consistent changes in ICU LOS and no changes in disposition location.
We are indebted to the University of Utah Department of Physical and Occupational Therapy staff, without whose belief in our mission, and herculean physical work, this intervention would not have been possible.
1. Bolton CF. The discovery of critical illness polyneuropathy: A memoir.Can J Neurol Sci201037431–438
2. Latronico N, Bolton CF. Critical illness polyneuropathy and myopathy: A major cause of muscle weakness and paralysis.Lancet Neurol201110931–941
3. Fan E, Zanni JM, Dennison CR, et al. Critical illness neuromyopathy and muscle weakness in patients in the intensive care unit
.AACN Adv Crit Care200920243–253
4. Khan J, Burnham EL, Moss M. Acquired weakness in the ICU: Critical illness myopathy and polyneuropathy.Minerva Anestesiol200672401–406
5. Kress JP, Hall JB. ICU-acquired weakness and recovery from critical illness.N Engl J Med20143701626–1635
6. Morris PE, Goad A, Thompson C, et al. Early intensive care unit
mobility therapy in the treatment of acute respiratory failure.Crit Care Med2008362238–2243
7. Denehy L, Skinner EH, Edbrooke L, et al. Exercise rehabilitation
for patients with critical illness: A randomized controlled trial with 12 months of follow-up.Crit Care201317R156
8. Moss M, Nordon-Craft A, Malone D, et al. A randomized trial of an intensive physical therapy
program for patients with acute respiratory failure.Am J Respir Crit Care Med20161931101–1110
9. Wright SE, Thomas K, Watson G, et al. Intensive versus standard physical rehabilitation
therapy in the critically ill (EPICC): A multicentre, parallel-group, randomised controlled trial.Thorax201873213–221
10. Schaller SJ, Anstey M, Blobner M, et al.; International Early SOMS-guided Mobilization Research InitiativeEarly, goal-directed mobilisation in the surgical intensive care unit
: A randomised controlled trial.Lancet20163881377–1388
11. Schweickert WD, Pohlman MC, Pohlman AS, et al. Early physical and occupational therapy in mechanically ventilated, critically ill patients: A randomised controlled trial.Lancet20093731874–1882
12. Needham DM, Korupolu R, Zanni JM, et al. Early physical medicine and rehabilitation
for patients with acute respiratory failure: A quality improvement project.Arch Phys Med Rehabil201091536–542
13. Snelson C, Jones C, Atkins G, et al. A comparison of earlier and enhanced rehabilitation
of mechanically ventilated patients in critical care compared to standard care (REHAB): Study protocol for a single-site randomised controlled feasibility trial.Pilot Feasibility Stud2017319
14. Thomsen GE, Snow GL, Rodriguez L, et al. Patients with respiratory failure increase ambulation
after transfer to an intensive care unit
where early activity is a priority.Crit Care Med2008361119–1124
15. Hodgson C, Bellomo R, Berney S, et al.; TEAM Study InvestigatorsEarly mobilization and recovery in mechanically ventilated patients in the ICU: A bi-national, multi-centre, prospective cohort study.Crit Care20151981
16. Hodgson CL, Bailey M, Bellomo R, et al.; Trial of Early Activity and Mobilization Study InvestigatorsA binational multicenter pilot feasibility randomized controlled trial of early goal-directed mobilization in the ICU.Crit Care Med2016441145–1152
17. Engel HJ, Tatebe S, Alonzo PB, et al. Physical therapist-established intensive care unit
early mobilization program: Quality improvement project for critical care at the University of California San Francisco Medical Center.Phys Ther201393975–985
18. Nydahl P, Ruhl AP, Bartoszek G, et al. Early mobilization of mechanically ventilated patients: A 1-day point-prevalence study in Germany.Crit Care Med2014421178–1186
19. 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 Saf201625986–992
20. Lee VS, Kawamoto K, Hess R, et al. Implementation of a value-driven outcomes program to identify high variability in clinical costs and outcomes and association with reduced cost and improved quality.JAMA20163161061–1072
21. Kawamoto K, Martin CJ, Williams K, et al. Value driven outcomes (VDO): A pragmatic, modular, and extensible software framework for understanding and improving health care costs and outcomes.J Am Med Inform Assoc201522223–235
22. Lonardo NW, Mone MC, Nirula R, et al. Propofol is associated with favorable outcomes compared with benzodiazepines in ventilated intensive care unit
patients.Am J Respir Crit Care Med20141891383–1394
23. Hodgson C, Needham D, Haines K, et al. Feasibility and inter-rater reliability of the ICU mobility scale.Heart Lung20144319–24
24. Charlson ME, Pompei P, Ales KL, et al. A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation.J Chronic Dis198740373–383
25. Hester JM, Guin PR, Danek GD, et al. The economic and clinical impact of sustained use of a progressive mobility program in a neuro-ICU.Crit Care Med2017451037–1044
26. Haley SM, Andres PL, Coster WJ, et al. Short-form activity measure for post-acute care.Arch Phys Med Rehabil200485649–660
27. Jette DU, Stilphen M, Ranganathan VK, et al. Interrater reliability of AM-PAC “6-clicks” basic mobility and daily activity short forms.Phys Ther201595758–766
28. Jette DU, Stilphen M, Ranganathan VK, et al. Validity of the AM-PAC “6-clicks” inpatient daily activity and basic mobility short forms.Phys Ther201494379–391
29. Jette DU, Stilphen M, Ranganathan VK, et al. AM-PAC “6-clicks” functional assessment scores predict acute care hospital discharge destination.Phys Ther2014941252–1261
30. Hoyer EH, Young DL, Klein LM, et al. Toward a common language for measuring patient mobility in the hospital: Reliability and construct validity of interprofessional mobility measures.Phys Ther201898133–142
31. Wagner AK, Soumerai SB, Zhang F, et al. Segmented regression analysis of interrupted time series studies in medication use research.J Clin Pharm Ther200227299–309
32. Anekwe DE, Koo KK, de Marchie M, et al. Interprofessional survey of perceived barriers and facilitators to early mobilization of critically ill patients in Montreal, Canada.J Intensive Care Med2017 Jan 1:885066617696846. [Epub ahead of print]
33. Bassett RD, Vollman KM, Brandwene L, et al. Integrating a multidisciplinary mobility programme into intensive care practice (IMMPTP): A multicentre collaborative.Intensive Crit Care Nurs20122888–97
34. Hoyer EH, Friedman M, Lavezza A, et al. Promoting mobility and reducing length of stay in hospitalized general medicine patients: A quality-improvement project.J Hosp Med201611341–347
35. Hodgin KE, Nordon-Craft A, McFann KK, et al. Physical therapy
utilization in intensive care units: Results from a national survey.Crit Care Med200937561–566quiz 566–568
36. Gosselink R, Bott J, Johnson M, et al. Physiotherapy for adult patients with critical illness: Recommendations of the European Respiratory Society and European Society of Intensive Care Medicine Task Force on Physiotherapy for Critically Ill Patients.Intensive Care Med2008341188–1199
37. Hanekom S, Gosselink R, Dean E, et al. The development of a clinical management algorithm for early physical activity and mobilization of critically ill patients: Synthesis of evidence and expert opinion and its translation into practice.Clin Rehabil201125771–787
ambulation; early mobility; intensive care unit; physical therapy; rehabilitation
Supplemental Digital Content
Copyright © 2019 The Authors. Published by Wolters Kluwer Health, Inc. on behalf of the Society of Critical Care Medicine.