Pre-operative functional mobility as an independent determinant of inpatient functional recovery after total knee arthroplasty during three periods that coincided with changes in clinical pathways

Disability and poor recovery of activities during hospital stay are common complications after major surgery, especially in frail elderly patients.^1,2 This may be relevant in patients undergoing total knee arthroplasty (TKA), which is an effective treatment in patients suffering from end-stage osteoarthritis.³ It is therefore often performed in elderly patients.⁴ Slow functional recovery during hospitalisation occurs generally in 30% of patients over 70 years undergoing major surgery, regardless of the success of the surgical procedure.^5-7 These slow-recovering patients have an increased risk of being discharged to a skilled nursing facility or rehabilitation centre. Keswani et al⁸ demonstrated that prolonged hospitalisation in these post-acute facilities increases the risk of adverse events after TKA. Therefore, swift recovery of relevant activities (i.e. transferring from bed to a chair, ambulation and climbing stairs), in order to function independently at discharge, is important to prevent a prolonged length of hospital stay and/or discharge to rehabilitation facilities.⁹ Ideally, preventive measures such as pre-operative physiotherapy screening should be initiated when the patient is recommended to undergo surgery. We therefore need insight into pre-operative factors that a) are associated with delayed inpatient recovery of activities and b) can be improved prior to surgery.^2,10

From other fields of surgery we know that pre-operative functional mobility is an important modifiable factor for post-operative recovery.^2,10 Some authors advocate a routine pre-operative functional mobility assessment other than the standard pre-operative medical screening of patients scheduled for surgery. They provide objective information which could help during patient and family counselling regarding risks of surgery, as well as anticipatory peri-operative care planning.¹ However, evidence for pre-operative risk factors predicting inpatient recovery of activities after TKA is scarce.^11,12 Hoogeboom et al¹¹ studied the predictive value of patient characteristics and surgery related variables for delayed inpatient functional recovery, finding that higher age, female gender and higher BMI were independent risk factors for delayed inpatient functional recovery. However, they did not assess functional mobility as a determinant of delayed inpatient recovery of activities.¹¹ Holm et al¹² studied the role of pre-operative functional mobility as a determinant of hospital discharge based on recovery of activities. However, they did not find any pre-operative factors (except age) that predicted discharge readiness, as most of their patients undergoing fast-track TKA were discharged within two post-operative days.

It remains to be established whether functional mobility is a determinant of inpatient recovery of activities in a more heterogeneous population, and whether such prediction depends on the post-operative circumstances and provided care. The latter is relevant, as clinical pathways for TKA have undergone major changes over the last years. For example, the start of rehabilitation shifted from two days to two hours after surgery and/or general anaesthesia shifted to use of spinal anaesthesia.^13,14 Clinical care pathways are now increasingly focusing on what a patient can actually do.¹⁵

Therefore, in this study, we aimed to investigate whether pre-operative functional mobility, as measured through the Timed Up and Go test (TUG)¹⁶ or the DeMorton Mobility Index (DEMMI) (a more extensive measure of functional mobility),¹⁷ is an independent determinant of delayed inpatient recovery of activities in patients undergoing TKA during three consecutive time periods that are characterised by a change in care pathway. We also investigate whether the association between pre-operative functional mobility and inpatient recovery of activities is modified by the period during which the TKA took place.

Patients and Methods

Setting

This prospective cohort study is performed in the routine orthopaedic setting in Nij Smellinghe Hospital, Drachten, The Netherlands. Between May 2009 and March 2015, all patients undergoing primary TKA were routinely screened before, and monitored after, surgery on their functional mobility. During this six-year period, the clinical care pathway was updated twice. First, from a Joint Care pathway¹⁸ (May 2009 to Dec 2010) to a Function-Tailored pathway¹³ (Jan 2011 to Oct 2013). Second, from the function-tailored pathway to a fast-track pathway¹⁹ (Nov 2013 to Mar 2015). Table I shows the main elements of the three specific care pathways, as applied in the Nij Smellinghe Hospital.

According to Dutch law, research with anonymised care data does not require approval from a medical ethical committee and is subject to a general opt-out procedure by the hospital. This standpoint was reconfirmed by the ethical committee of the Nij Smellinghe Hospital with respect to this particular study.

Data collection

All patients placed on the waiting list for a primary TKA were routinely screened by an anaesthetist for surgical risk, by an orthopaedic physiotherapist for functional status and risk for delayed inpatient recovery of activities, and by a nurse practitioner for social and environmental issues.

After surgery, each day an orthopaedic physiotherapist assessed a patient’s recovery of activities. The healthcare team in charge of the treatment of patients scheduled for elective TKA consists of four orthopaedic surgeons, four nurse practitioners, four orthopaedic physiotherapists, and 35 nurses. To ensure that all patients were assessed in the same manner, we standardised the pre- and post-operative functional assessment by using a measurement protocol and implementing a training programme for the physiotherapists.

Inpatient recovery of activities

We measured inpatient recovery of activities using the modified Iowa Levels of Assistance Scale (mILAS).²⁰ The mILAS scores patients’ ability to perform five activities of daily life safely (i.e., supine to sit, sit to supine, sit to stand, walking (with walking aids), and stair climbing) and the degree of assistance needed to perform the tasks safely. Scores range from 30 to 0; (the lower the score, the better the activity level of the patient) with a total mILAS score of zero reflecting sufficient inpatient recovery of physical function to be safely discharged home. Stair climbing was only assessed if the patient needed to climb stairs at their discharge destination. Therefore, we excluded stair climbing from our outcome measure in those patients who did not require stair climbing. The time (in days) for a patient to achieve a mILAS score of 0/6 was used to compute the outcome variable in this study, i.e. timely versus delayed recovery.

Each update of the care pathway aimed to improve the quality and speed of inpatient recovery of activities without increasing the risk of re-admission and long-term inpatient stay in rehabilitation. Therefore, we had to define timely recovery for each specific time period. After exploring cumulative frequency distribution tables and plots, from our data, we decided to set the cut off point for delay in inpatient recovery of activities on the day that 70% of the patients were recovered, according to the mILAS, for each time period. Patients who did not achieve a mILAS score of zero before discharge from the hospital were coded as delayed recovered as they all went to inpatient rehabilitation facilities or nursing homes.

Pre-operative patient characteristics

The following potential predictors of recovery were assessed directly after the patient was indicated for TKA: age (years), gender, BMI (kg/m²), comorbidities assessed by (1) the American Society of Anesthesiologists (ASA) score (1 to 2 or ≥ 3; a higher score indicates less fit for surgery) and (2) the Charnley score (A/B/C; indicates the function of the knee with regard to the ability to walk; A is more favourable than C),²¹ and self-reported information about frailty as assessed by the Identification of Seniors at Risk score (ISAR).²² In addition, we included the element ‘presence of stairs at home’, because stair climbing is an additional activity according the mILAS that could possibly delay the recovery time of the patient.

Pre-operative functional mobility

This included the TUG test in seconds (assesses functional mobility; a lower score indicates a better functional mobility)¹⁶ and the DEMMI in points (assesses functional mobility by monitoring 15 functional activities; a higher score indicates a better functional mobility).^17,23

Statistical analysis

Pre-operative patient characteristics were described using counts and percentages for categorical variables, and means and standard deviations (sd) for continuous variables.

We used multivariate logistic regression to determine whether performance-based functional mobility indicators are determinants of inpatient recovery of activities after TKA, independent of other patient characteristics. We first made a model including other potential determinants of recovery as independent variables, i.e. age, gender, ASA score, Charnley class, ISAR, BMI, and the question ‘stairs at home?’. Subsequently, variables with a non-significant or small contribution were excluded from this model, resulting in the final reference model. The TUG or the DEMMI were each separately added as independent variables to the reference model. To investigate whether time period modified the relationship between functional mobility and inpatient recovery of activities, we added the periods (two indicator variables) and their interaction with pre-operative TUG or DEMMI data, respectively, as terms to the model. In order to avoid multicollinearity between the interaction variables and period variables, we first subtracted the mean from each of the functional mobility measurements (TUG and DEMMI). Finally, we added the DEMMI to the model including the TUG to investigate whether the DEMMI contributes more to the inpatient recovery of activities than the TUG alone. To compare the predictive value of the DEMMI and the TUG we calculated Nagelkerke’s pseudo R². All analyses were performed using Stata version 12 software, (Stata Corp., College Station, Texas). Statistical tests were performed two-sided with significance set at a p-value < 0.05.

Results

Between May 2009 and March 2015, a total of 682 people undergoing primary unilateral TKA (499 women, 73%) and a mean age of 70 years (41 to 89, sd 9.0) were screened pre-operatively (Table II). Characteristics of patients who were included during the joint care (n = 171), function-tailored (n = 360) or fast-track (n = 151) period are presented in Table II. Despite inclusion of all patients undergoing TKA in our hospital, we missed 95 (2%) of the values of pre-operative characteristics in total. The item with the largest number of missing data was the DEMMI (n = 41, 6%).

Cut-off point for timely or delayed inpatient recovery of activities

Using cumulative frequency tables, the cut-off points for timely recovery were set on six (70%), four (74%) and three (70%) days for the joint care, function-tailored care, and fast-track pathways, respectively (data not shown).

Reference model

In the logistic regression model including all selected patient characteristics as independent variables, only age (Odds Ratio (OR) 1.06, 95% confidence interval (CI) 1.04 to 1.09), ASA (OR 2.45, 95% CI 1.49 to 4.02) and ISAR (OR 1.61, 95% CI 1.36 to 1.97) were significant determinants of delayed inpatient recovery of activities (Table III). Despite the non-significant, but not negligible contribution of the BMI (p = 0.06, OR 1.03, 95% CI 0.99 to 1.08) in this model, we decided to include BMI in the final reference model to avoid possible confounding by BMI of the association between TUG (or DEMMI) and inpatient recovery of activities. The resulting reference model with age, ASA, ISAR and BMI predicted 15% of the variance in delayed inpatient recovery of activities (Pseudo R²).

TUG as independent determinant

To investigate whether the TUG added to the prediction of delayed inpatient recovery of activities, we added the TUG to the reference model. The TUG score (OR 1.10, 95% CI 1.06 to 1.15) was a significant determinant of delayed inpatient recovery of activities (Table III). The resulting model predicted 19% of the variance in delayed inpatient recovery of activities (Pseudo R²). We further investigated whether the different time period that coincided with clinical pathway changes modified the relationship between TUG and inpatient recovery of activities. The overall interaction between TUG and the time period in the model was not statistically significant (p = 0.41).

DEMMI as independent determinant

To investigate whether the DEMMI added to the prediction of timely inpatient recovery of activities, it was added to the reference model. The DEMMI score (OR 0.96, 95% CI 0.95 to 0.98) was a significant determinant of delayed inpatient recovery of activities (Table III). The resulting model predicted 20% of the variance in delayed inpatient recovery of activities (Pseudo R²). The overall interaction between DEMMI and the time period had a non-statistically significant contribution to the model (p = 0.74).

To investigate whether the more extensive DEMMI added to the prediction of inpatient recovery of activities over the TUG, we added the DEMMI variable to the model including TUG (Table III). Both functional mobility measures were statistically significant in this model, while each of the ORs were somewhat attenuated towards 1. The OR for the TUG decreased from 1.10 (95% CI 1.06 to 1.15) to 1.06 (95% CI 1.01 to 1.11), while the OR for the DEMMI increased from 0.96 (95% CI 0.95 to 0.98) to 0.97 (95% CI 0.96 to 0.99). The resulting model predicted 22% of the variance in delayed inpatient recovery of activities (Pseudo R²).

Discussion

The aim of this study was to investigate the role of functional mobility as determinant of inpatient recovery of activities after TKA. We added two measurements of functional mobility, the TUG and the DEMMI, separately to a reference model including the well-established determinants age, ASA, ISAR, and BMI. Both the TUG in seconds (OR 1.10, 95% CI 1.06 to 1.15) and the DEMMI, in points (0 to 100) (OR 0.96, 95% CI 0.95 to 0.98) proved to be independent determinants of inpatient recovery of activities.

For example, a patient who is one second slower on the TUG is 10% more likely to experience a delayed inpatient recovery of activities, compared with a patient with the same pre-operative characteristics with a TUG time one second faster. The period representing TKA care pathways, which changed twice during the period of observation, did not modify the associations between TUG or DEMMI or inpatient recovery of activities.

Our results seem in contrast to the work of Holm et al,¹² who did not find any added value of functional mobility as an independent determinant of inpatient recovery of activities in patients undergoing fast-track total joint arthroplasty surgery. These conflicting results might be explained by the relative low mean age of their population (mean 65 years, sd 9.6), as compared with our patients (mean 70 years, 41 to 89, sd 9.0), and the low score of the pre-operative TUG (mean 9.3 seconds), indicating that Holm et al’s¹² patients were functionally more mobile/fit than our patients (mean 11.4 seconds). Also the spread in the TUG was much lower in Holm et al’s¹² patients than in our patients (sd 2.3 versus 6.3).We conclude from this comparison that our patient population was more heterogeneous with respect to functional mobility, and included a substantial percentage of frail patients (25.5% in our population according the ISAR).

Our study outcomes are in line with the outcomes of Elings et al²⁴ in a risk stratification model for delayed inpatient recovery of activities in patients undergoing THA. In a multivariate logistic regression model (n = 154) with gender, age, ASA, Charnley score, and BMI, the TUG (≥ 12.5 seconds versus < 12.5 seconds) was a significant determinant of delayed recovery (OR 3.1, 95% CI 1.2 to 9.0) as measured with the mILAS. Application of the same cut-off for the TUG in our data, showed an OR 3.2 (95% CI 2.1 to 5.0) comparing patients with a TUG ≥ 12.5 seconds (mean TUG 19.3 seconds) with those with a TUG < 12.5 seconds (mean TUG 8.3 seconds). If patients experience a delayed inpatient recovery of activities, there is a serious risk of important (medical) complications or iatrogenic disability.^5,7 Besides this, inpatient recovery of activities and the length of hospital stay predict functional recovery on the long term.^11,25 As pre-operative functional mobility is a factor that can be positively influenced, our research provides a starting point in looking at how to accelerate inpatient recovery of activities in patients with a high risk for delayed recovery. To our knowledge, there is inconclusive evidence for the merit of pre-operative physical rehabilitation in people undergoing TKA.^10,26-29 However, patients at a high risk for delayed recovery of activities may benefit from pre-operative interventions,^10,30 whereas low risk patients may not.²⁶

We have shown that poor pre-operative functional mobility hampers inpatient recovery of activities in TKA patients. Any pre-operative interventions should therefore include functional screening to establish a baseline functional mobility level and to identify patients at risk for prolonged inpatient recovery of activities after the surgery. A pre-operative exercise programme tailored to optimise the functional mobility of the high-risk patient is one of the promising interventions to accomplish their return to the highest level of functioning possible as quickly as possible after surgery.^2,10

Our study has strengths and weaknesses. An important strength of this study is that the data used in the analyses were collected as part of routine care and relate to all patients undergoing TKA in our hospital. Also, pre-operative functional mobility appeared to be a strong and stable predictor of inpatient recovery of activities, independent of time period and type of care. Therefore, the findings are applicable not only to the patient population undergoing TKA in our hospital, but likely also to patient populations in other hospitals, provided that the patients have similar pre-operative characteristics. In The Netherlands, the majority of TKA patients are aged between 65 and 85 years.⁴ We have also shown that the concept of functional mobility was captured consistently by using two measurements, the TUG and the DEMMI. We had expected the DEMMI to be a more comprehensive, and therefore, better test of functional mobility than the TUG. Together, they appeared to predict slightly better than either test alone, whereas the DEMMI alone indeed performed slightly better than the TUG alone. The availability of baseline information, prospectively collected with validated instruments according to scientific standards at the pre-operative functional screening, and availability of well-monitored outcome information made it possible to apply a valid study design. This shows that well-kept, up-to-date patient records that include relevant baseline data are a source of information to monitor the quality of routine medical care in TKA. There are also limitations to our study. First, there were missing data on several study parameters. Missing data were mainly attributed to incomplete patient records. The predictor with the largest number of missing data was the DEMMI (6%), which can still be considered fairly small. Second, even though we think that delayed inpatient recovery of activities is an important determinant of long-term recovery of functional mobility, we were not able to investigate this among our patient population.

Our study has a number of important implications for daily practice and future research. First, we showed that functional mobility, measured by the TUG and the DEMMI, is an important and independent determinant of inpatient recovery of activities after TKA. Therefore, the next step is to develop a risk model in order to select high-risk patients and to investigate the effectiveness of “prehabilitation” in selected TKA patients at risk for delayed inpatient recovery of activities. Oosting et al³¹ showed in a pilot study that supervised home-based pre-operative interventions in patients at risk, are feasible and effective.

Functional mobility performance of patients undergoing TKA, as assessed by the TUG and DEMMI before surgery, is an independent and stable determinant of delayed inpatient recovery of activities after surgery. Through the addition of functional mobility to pre-operative risk assessment, clinicians have the possibility to offer pre-operative interventions to patients at high risk for delayed recovery. The effectiveness of such interventions on recovery of activities after TKA in high risk patients should be subject of future research.

Take home message:

- Poor pre-operative functional mobility hampers inpatient recovery of activities in patients undergoing total knee arthroplasty.

Author contributions:

G. van der Sluis: Contributed to conception and design of this study, Drafted the manuscript, Performed the statistical analysis, Critically reviewed the manuscript for important intellectual content, Final approval of this version to be published.

R. A. Goldbohm: Contributed to conception and design of this study, Drafted the manuscript, Contributed to the statistical analysis, Critically reviewed the manuscript for important intellectual content, Final approval of this version to be published.

J. Elings: Contributed to conception and design of this study, Critically reviewed the manuscript for important intellectual content, Final approval of this version to be published.

M. W. Nijhuis- van der Sanden: Contributed to conception and design of this study, Critically reviewed the manuscript for important intellectual content, Final approval of this version to be published.

R. P. Akkermans: Contributed to conception and design of this study, Checked the statistical analysis, Critically reviewed the manuscript for important intellectual content, Final approval of this version to be published.

R. Bimmel: Contributed to conception and design of this study, Critically reviewed the manuscript for important intellectual content, Final approval of this version to be published.

T. J. Hoogeboom: Contributed to conception and design of this study, Critically reviewed the manuscript for important intellectual content, Final approval of this version to be published.

N. L. U. van Meeteren: Contributed to conception and design of this study, Critically reviewed the manuscript for important intellectual content, Final approval of this version to be published.

No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article.

This article was primary edited by G. Scott.