Optimising pre-surgical health
Of the 36 interventions (32 articles) targeting optimisation of pre-surgical health, 14 interventions were in patients waiting for THR,487–489,493–497,500–504,507 17 in patients waiting for TKR,500,503,509–515,517,519–521,523–525 and five in total hip and knee patients together.526–529,534
Eighteen interventions were specifically exercise based,488,494,497,502,503,509,511–515,519,520,523–525 nine were education based,52,489,495,500,501,526,527,529 while nine were multifactorial with exercise and education components and, in some cases, an occupational therapy base.487,493,496,504,507,510,517,521,528
Effect on pre-surgical physical function
Physical function before surgery was measured after 21 interventions using WOMAC (physical function or total), SF-36 physical function, HHS, HOOS, HSS, Arthritis Impact Measure Score 2 (AIMS2), AKSS or locally devised scores.
Data for meta-analysis were available for 19 interventions with 1026 participants.488,493,494,496,497,501–503,510,513–515,519,521,525,527,534 In the random-effects meta-analysis shown in and , the average SMD was 0.28 (95% CI 0.16 to 0.40) favouring intervention. Inspection of the funnel plot did not suggest publication bias. Benefit for interventions was limited to eight studies with 343 patients waiting for THR (average SMD 0.40, 95% CI 0.18 to 0.61) and two studies with 226 hip and knee patients reported together. There was no heterogeneity across all studies or in those in patients waiting for hip replacement. In eight studies with 457 patients waiting for knee replacement, there was a non-significant trend for benefit (average SMD 0.19, 95% CI –0.04 to 0.42). In three education-based interventions with 261 patients, physical function was better in intervention patients (average SMD 0.33, 95% CI 0.09 to 0.58). Similarly, in 12 exercise-based interventions with 494 patients, interventions showed benefit (average SMD 0.26, 95% CI 0.08 to 0.44).
Pre-surgical physical function outcome in studies targeting optimisation of pre-surgical health. df, degrees of freedom; IV, inverse variance. SF-36 physical function: Bitterli and colleagues. WOMAC function: Ferrara and colleagues, Gilbey and colleagues, (more...)
No heterogeneity was apparent. In four multifactorial interventions with 271 patients there was a trend for benefit (SMD 0.32, 95% CI –0.03 to 0.66). Considering studies separately in patients waiting for hip and knee replacement there was no suggestion that particular types of interventions were more effective.
In a sensitivity analysis with 10 interventions of good or reasonable quality with 610 patients, benefit was still apparent (SMD 0.21, 95% CI 0.04 to 0.38). Numbers in groups were small but benefit was mainly in patients waiting for hip replacement.
To evaluate physiotherapy content we analysed studies by intervention intensity (frequency and duration) and specific components. Higher intensity interventions showed benefit in seven studies with 308 patients waiting for hip replacement (SMD 0.39, 95% CI 0.16 to 0.62), and a trend for benefit in nine studies with 457 knee arthroplasty patients (SMD 0.19, 95% CI –0.04 to 0.42). However, numbers of studies with medium or low intensity exercise content were small.
In patients waiting for THR, five interventions focusing on strengthening showed benefit (SMD 0.36, 95% CI 0.09 to 0.62), as did three studies with cardiovascular exercise (SMD 0.48, 95% CI 0.03 to 0.93). Numbers of studies including other components targeting stretching, gait re-education, ROM, balance, endurance, flexibility, proprioception, pool-based exercises and functional exercises were too low to draw conclusions on content. There was no evidence to suggest that any interventions had an adverse effect on physical function.
In patients waiting for knee replacement, nine physiotherapy exercise interventions had a strengthening component. As shown in , there was a non-significant trend for benefit in improved physical function for interventions with a strengthening component compared with controls (SMD 0.19, 95% CI –0.04 to 0.42). As with studies in hip replacement, there were too few studies to draw conclusions on other exercise content.
Pain
Pain measured before surgery was reported for 25 interventions, with data suitable for meta-analysis in 21 interventions with 1375 patients. Interventions were associated with reduced pain (SMD 0.21, 95% CI 0.10 to 0.33), but this was limited to patients waiting for hip replacement (SMD 0.47, 95% CI 0.28 to 0.67). There was only slight heterogeneity and the funnel plot did not suggest publication bias (data not shown).
In 16 studies judged to be of good or reasonable methodological quality comprising 1141 patients,487,488,493–495,497,503,509–511,514,515,517,521,524,527 there was benefit for interventions (SMD 0.24, 95% CI 0.09 to 0.38), but this was statistically significant only in hip patients.
In 13 studies with 597 patients,488,494,497,502,503,509,511,514,515,519,524,525 exercise-based interventions showed benefit, SMD 0.23 (95% CI 0.07 to 0.39). In three education-based interventions495,501,527 and five multifactorial interventions,487,493,510,517,521 there were trends for benefit, with SMDs of 0.20 (95% CI –0.03 to 0.44) and 0.27 (95% CI –0.04 to 0.58). Numbers in some groups were small but all types of programme content showed benefit for patients waiting for THR.
Anxiety
Anxiety was reported pre-operatively for eight interventions, using the State–Trait Anxiety Inventory (STAI) in six studies,489,495,526,528,529 and HADS525 and Arthritis Impact Measure Score (AIMS) anxiety487 in one each. In eight studies487,489,495,525,526,528,529 with 660 patients, interventions were associated with lower anxiety with SMD 0.38 (95% CI 0.11 to 0.65). However, in quality assessment, five studies were classified as having risk of bias.489,495,525,526,529
Preparation for recovery in hospital
In 27 studies (26 articles), a specific focus of the intervention was preparing patients for their hospital stay, including what to expect before, during and after surgery.489–492,495,496,498,499,501,504–508,510,516–518,522,526,528–533 We identified 14 interventions in patients waiting for hip replacement,489–492,495,496,498,499,501,504–508 six interventions in patients waiting for knee replacement510,516–518,522 and seven intervention in patients waiting for either hip or knee replacement.526,528–533 Nineteen interventions489,490,492,495,498,499,501,505,508,516,518,522,526,529–533 were primarily education based and eight were multifactorial with education and exercise content.491,496,504,506,507,510,517,528
Post-surgical anxiety
Eight interventions reported anxiety outcomes within a month of surgery.489,490,492,495,506,522,529,530 Five studies with 381 patients presented data suitable for the meta-analysis in and .489,492,495,506,529 Four studies recorded anxiety using the STAI questionnaire during the hospital stay or the change from baseline and included either hip or hip and knee patients analysed together.489,492,495,529 In a random-effects model, the MD was 2.97 (95% CI 0.64 to 5.30) in favour of reduced anxiety in the intervention group. One study reported sense of uncertainty on discharge using a VAS scale.506 Inclusion of this in the meta-analysis and excluding one study with change scores gave an average SMD of 4.15 (95% CI 1.46 to 6.84).
Post-surgical anxiety outcome in studies targeting preparation for recovery in hospital (MD). df, degrees of freedom; IV, inverse variance. STAI: Butler and colleagues, Doering and colleagues (from McDonald and colleagues), Cuñádo Barrio (more...)
There was no evidence of heterogeneity and exclusion of one study with differing losses to follow-up between randomised groups did not affect the outcome.529
Post-surgical pain
Fifteen interventions reported a pain outcome up to 1 month after surgery. For meta-analysis, we included the latest data at a fixed time point up to 1 month after surgery. If not available, we used data collected at discharge but such data may be affected by time in hospital. When possible, we used pain at movement in preference to pain at rest. Using these criteria, 12 interventions with 842 patients were included in the meta-analysis. Up to 1 month after surgery, pain was non-significantly lower in intervention groups than in controls (SMD 0.18, 95% CI –0.01 to 0.38). Similar trends were noted in hip and knee patients separately. There was no evidence of publication bias from inspection of the funnel plot but some heterogeneity was evident in trials including knee replacement patients (data not shown).
Results and heterogeneity were similar after exclusion of studies with possible bias owing to differences in patient characteristics, group follow-up rates or inclusion of data collected at discharge.
Length of hospital stay
Length of hospital stay was reported after 19 interventions with 997 patients. There was no statistically significant benefit for interventions (MD –0.16 days, 95% CI –0.78 to 0.45 days) reflecting a trend for lower length of hospital stay in intervention groups. The corresponding funnel plot did not suggest presence of publication bias. Heterogeneity (I2 = 67%) was not explained by one study with different follow-up rates in randomised groups. Although benefit for reduced length of stay was noted in knee replacement patients, this was based on two studies510,517 and was not supported by studies including both hip and knee patients.528,529,531
Mobilisation after surgery
In six studies with 471 patients, time to mobilisation was reported.495,496,506,507,528,529 Studies included only hip or hip and knee patients. Measures of mobilisation were time to walking in five studies495,496,506,528,529 and time to standing in one study.507 In meta-analysis, time to mobilisation was shorter in the intervention groups (MD –0.17 days, 95% CI –0.30 to –0.04 days) and this was little changed if only time to walking was considered. Overall, there was no heterogeneity among studies.
Improving long-term outcomes
Nine pre-surgical interventions explicitly targeted improvement in long-term outcome after hip replacement487,488,493,496,501–503,506 and nine after knee replacement.503,510,513–515,519,524,525 The focus of the intervention was exercise in 12 interventions, education in one intervention and multifactorial in five interventions.
Long-term physical function
All but three studies reported a functional outcome at 3 months or longer after surgery.502,515,524 As shown in the meta-analysis in and , including 15 studies with 577 patients, there was no overall long-term benefit in intervention compared with control groups (SMD 0.08, 95% CI –0.09 to 0.26). Only moderate heterogeneity was apparent and there was no suggestion of publication bias.
Physical function in studies targeting improvement of long-term outcomes. df, degrees of freedom; ILAS, Iowa Level of Assistance Scale; IV, inverse variance. WOMAC function: Ferrara and colleagues, McGregor and colleagues (supplied by author), Rooks and (more...)
Long-term pain
As shown in , there was no benefit for reduced long-term pain after interventions in 10 studies with 414 patients (SMD 0.03, 95% CI –0.16 to 0.23). In addition, there was no heterogeneity among the studies.
