Introduction

Fractures of the proximal humerus (PHF) or distal radius (DRF) are among the most common fractures in the elderly population [1, 2]. The majority of patients suffering a PHF or DRF are aged 60 years or older, and the most representative patient is an elderly osteoporotic woman [2, 3]. Due to longer life expectancy in the elderly population, the incidence rates of these osteoporotic fractures are predicted to increase [2, 3]. In future, this increase will impose a substantial burden on healthcare systems and increase societal costs.

Both PHF and DRF can have a substantial impact on the patient´s physical function and independent living and are associated with higher morbidity and mortality [3, 4]. After sustaining a PHF or DRF, the main focus of treatment is to regain the best possible function of the shoulder or wrist. Recent evidence questions the benefit of surgical treatment compared with non-surgical treatment after PHF [5,6,7]. The same conclusion was reached in a recent review investigating the optimal treatment after DRF in which no clear benefit of surgical treatment was found in the elderly [8•]. Thus, the next important task is to address and optimize the non-surgical treatment strategy for these fractures.

The question of when to commence supervised exercise therapy is of high clinical importance. Sparse evidence suggests that there might be a preference for a short immobilization period after sustaining a PHF or DRF [9••].

Patients are usually referred to supervised rehabilitation after sustaining these types of fractures. However, supervised rehabilitation consumes considerable healthcare resources and raises the question as to what extent patients benefit from supervised exercise therapy [8, 10]. Bruder et al. have suggested that after PHF non-supervised exercises at home might be just as effective as exercises supervised by a therapist; however, this conclusion was based on scarse evidence [9••]. In 2015, the Cochrane review concluded that there is no evidence to determine the best possible non-surgical treatment after PHF [6].

Our aim therefore was to conduct a systematic review and meta-analysis to assess the benefits and harms of (1) early mobilization compared to late mobilization and of (2) supervised exercise therapy compared to non-supervised exercises after non-surgically treated PHF and DRF.

Materials and Methods

Study Design and Protocol

This is a systematic review with meta-analysis. Search strategy, trial selection, eligibility criteria, methodology assessment, data extraction, and analysis were performed in accordance with a predefined protocol (PROSPERO: CRD42020167656). Trial screening, selection of trials, data extraction, assessment of methodology, and quality of evidence were performed by two independent reviewers (H.K.Ø. and V.T.P.). Disagreements were resolved through a consensus process. The Preferred Reporting Items for Systematic Review and Meta-Analysis (PRISMA) was used as a checklist throughout the reporting [11].

Search Strategy and Trial Selection

Electronic databases were systematically searched for primary trials (Supplementary material). Searches were conducted in MEDLINE (OvidSP), Cochrane Central Register of Controlled Trials (CENTRAL), Cochrane Database of Systematic Reviews (CDSR), CINAHL, Physiotherapy Evidence Database (PEDro), Scopus, ClinicalTrials.gov, and WHO International Clinical Trials Registry Platform (WHO ICTRP). Other sources involved the hand searching of reference lists of systematic reviews or guidelines. The search was limited to randomized controlled trials, non-randomized trials, and prospective observational studies. Searches were undertaken on September 20 and December 2, 2019 and repeated on June 11, 2020. Results were loaded into EndNote (version X9.2; Clarivate Analytics) software for deduplication. Full, database-specific search strategies are available in the online appendix. The identified trials were uploaded in systematic review management software (Covidence, Aus) and screened at title/abstract level. Eligible trials were then full-text screened for final inclusion. Reference lists from the full-text trials were also screened for supplementary relevant trials.

Eligibility Criteria

Eligibility criteria were defined according to the Population, Intervention, Comparison, and Outcome (PICO) method [12]. The population consisted of adults ≥ 18 years with a verified PHF or DRF due to recent trauma and referred to non-operative treatment. The interventions and comparisons were defined as (1) early mobilization (≤ within 2 weeks after time of fracture) compared to late mobilization or (2) supervised exercise therapy compared to non-supervised exercises. The included outcome measures were function, pain, and health-related quality of life (HRQoL). Function was defined as either function assessed by patient-reported outcome measures (PROMs), performance-based function including range of motion (ROM) and strength measures, or by questionnaires comprising both of these subjective and objective measurements.

Methodological Assessment

The included trials were evaluated using the Cochrane Risk of Bias tool (RoB 2.0) [13]. The risk of bias was rated as low, unclear, or high. The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) approach was used to rate the body of evidence for each outcome as either very low, low, moderate, or high [14].

Data extraction and statistical analysis

The following trial information was extracted: author, year, country, trial design, number of participants, population characteristics, description of intervention/control, outcome measures, and time to follow-up. Effect estimates were extracted as reported in the trials. In three cases, the corresponding authors were contacted for extended data details. If a minimum of two outcome measures were found eligible for comparison, a meta-analysis was undertaken and values were presented as either mean difference (MD) or standard mean difference (SMD) with 95% CIs, using the random-effect model. The I2 value was calculated and if higher than 50%, the heterogeneity was considered substantial and not eligible for meta-analysis. In cases where meta-analysis could not be undertaken, narrative synthesis was performed based on the conclusions reported in the trials. Statistics were performed using Stata 16 (TX, USA) and R 3.6.2 (R Foundation, Vienna).

Results

Search and Selection of Trials

In total, 1924 trials were found eligible for screening. After screening and full-text reading, 15 RCTs (943 participants) were included. Furthermore, 1 abstract (98 participants) and 5 trials (736 participants) registered in ClinicalTrials.gov and WHO ICTRP were identified. A summary of the search results and trial selection process is provided in a PRISMA flow diagram (Fig. 1).

Fig. 1
figure 1

PRISMA flow chart of the included studies

Full size image

Trial Characteristics

The mean age of the participants in the included trials ranged from 52.5 to 77.3 years. The follow-up times ranged from 2 weeks to 2 years. Specific details on the outcome measures used can be found in Tables 1, 2, and 3.

Table 1 Study characteristics of the trials included in the systematic review (n = 15)
Full size table
Table 2 Study characteristics of the unpublished trials registered in ClinicalTrials.gov or International Clinical Trials Registry Platform (ICTRP) included in the systematic review (n = 5)
Full size table
Table 3 Study characteristics of the abstract included in the systematic review (n = 1)
Full size table

Benefits and harms of Early Mobilization after Proximal Humerus Fracture

Function

Functional scores were reported in three trials. [15,16,17,18]. One of the trials was outlined in two different papers [15, 16]. In addition, two unpublished trials (ClinicalTrials.gov) provided preliminary results [19, 20]. The outcome measures used are outlined in Tables 1 and 2. Data extracted from two of the trials [17, 20] allowed meta-analysis and an overall MD of 4.55 (95% CI 0.00–9.10) on the Constant Shoulder Score (CS) at 6-month follow-up in favor of early mobilization was found (Fig. 2). This finding was supported by two other trials concluding that early mobilization within 2 weeks after sustaining a PHF resulted in better function when compared to late mobilization [16, 18].

Fig. 2
figure 2

Function, 6 months after proximal humerus fracture. Displays pooled weighted mean difference for function (Constant shoulder score). 95% CI = 95% confidence interval

Full size image

ROM was reported in two trials [17, 18]. In addition, two unpublished trials reported ROM figures on which a meta-analysis could be conducted [19, 20]. No eligible evidence of a difference between the groups was found at 6-month follow-up, either in flexion with a MD equal to 2.18° (95% CI −4.83° to 9.19°) or in abduction with a MD equal to 3.75° (95% CI −5.76° to 13.27°) (Figs. 3 and 4). The same conclusion was reached in a third trial that found no eligible evidence of a difference in ROM between groups [18]. A fourth trial reported a higher active and passive abduction and anterior elevation in favor of the early mobilization group [17] (Table 1).

Fig. 3
figure 3

Range of shoulder flexion, 6 months after proximal humerus fracture. Displays pooled weighted mean difference for shoulder flexion (°) of the fractured shoulder. 95% CI = 95% confidence interval

Full size image
Fig. 4
figure 4

Range of shoulder abduction, 6 months after proximal humerus fracture. Displays pooled weighted mean difference for shoulder abduction (°) of the fractured shoulder. 95% CI = 95% confidence interval

Full size image

Pain

Pain scores were reported in two published trials [17, 18] and the two previously mentioned unpublished trials [19, 20]. Meta-analysis based on data from the two unpublished trials [19, 20] showed no evidence suggestive of between-group difference in pain 6 months after the fracture, with an MD equal to 0.20 (95% CI −0.36 to 0.76) (Fig. 5). In contrast, two other trials concluded individually that early mobilization resulted in a decrease of pain in favor of the early mobilization group [17, 18] (Table 1).

Fig. 5
figure 5

Patient-reported pain, 6 months after proximal humerus fracture. Displays pooled weighted mean difference for shoulder pain (VAS and Likert scale/0-10). 95% CI = 95% confidence interval

Full size image

Health-Related Quality of Life

The Short Form-36 (SF-36) was used to measure HRQoL in one trial. The trial reported a positive effect in the role limitation and pain domains in favor of early mobilization. This positive effect was not, however, found between groups in any of the other domains [16].

Complications

One trial identified a single case (1/32) of frozen shoulder in the late mobilization group [17] while another trial found one case (1/39 and 1/41) of reflex dystrophy in both groups [18].

Benefits and Harms of Early Mobilization after Distal Radius Fracture

Function

Functional scores were reported in five trials. The outcomes used are described in Tables 1 and 2. Two trials reported an improvement of function in favor of early mobilization after DRF [21, 22]. This finding was not, however, in agreement with the results reported in three other trials, which concluded that early mobilization after DRF did not lead to improvement of function [23,24,25]. Recovery of domestic abilities was assessed in one trial that reported a higher number of patients in the early mobilization group regained their abilities within 5 weeks [21].

ROM was reported in two trials which stated that early mobilization improved movement of the wrist [22, 25]. Grip strength was assessed in three trials [21, 22, 25]. One trial reported grip strength being higher in the early mobilization group [22]. However, no increase in grip strength was reported in the other two trials [21, 25].

Pain

Pain scores were reported in two trials. The trials did not find eligible evidence of a difference between groups at any of the time points [21, 25].

Health-Related Quality of Life

HRQoL was not assessed in any of the trials investigating early versus late mobilization after DRF.

Complications

One trial identified three cases (3/54) of treatment failure in the early mobilization group, defined as problems leading to abandonment of given treatment or operation of malunited fracture [25]. Another trial reported four cases of reflex dystrophy, but they did not state in which groups [24].

Benefits and harms of supervised exercise therapy after proximal humerus fracture

Function

ROM was assessed in two trials. One trial measured movements comprising hand on back and hand on neck [26], and one trial measured active and passive elevation [27]. Both trials concluded that supervised exercise therapy did not result in a better ROM at any time point. Muscle strength was also assessed in the two trials. One trial measured isometric muscle strength by horizontal and vertical push [27], and the other trial measured shoulder and grip strength [26]. No eligible evidence of a difference between groups at any time point was detected.

Pain

Pain was assessed in two trials [26, 27]. One trial rated pain as insignificant, moderate, or severe and the other trial used a modified Borg scale (from 0 to 8). The trials did not show a higher degree of pain relief in the supervised exercise therapy group [26, 27].

Health-Related Quality of Life

HRQoL was not assessed in any of the trials investigating supervised exercise therapy after PHF.

Complications

One trial reported three cases of frozen shoulder. Two cases (2/20) were identified in the supervised group and one (1/22) case in the non-supervised group. In addition, one (1/22) patient in the non-supervised group had an unexplained pain over a longer period [26].

Benefits and Harms of Supervised Exercise Therapy after Distal Radius Fracture

Function

Patient-Rated Wrist Evaluation (PRWE) was used in two trials and the results were inconsistent. One trial found that supervised exercise therapy improved function [28••]. The second trial, however, reported a greater improvement in the non-supervised group [29], and a third trial did not find any eligible evidence of a between-group difference [30].

ROM was reported in four trials [28, 29, 31, 32] and in one trial abstract [33]. Only one of the trials found no eligible evidence of a difference in ROM between the two groups [29], whereas the remaining trials concluded that supervised exercise therapy after DRF leads to an increased ROM [28, 31,32,33].

Grip strength was reported in four trials and the results were inconsistent [28, 29, 31, 32]. Two trials found increased grip strength in the supervised group [28, 32], the third trial found no eligible evidence of a difference between the groups at any time point [31], and the fourth trial found grip strength to be greater in the non-supervised group [29].

Pain

VAS was used to measure pain in two trials [28, 31]. One trial reported a lower degree of pain in the supervised group [28••], but another trial found no eligible evidence of a clear difference between the groups at any time point [31].

Health-Related Quality of Life

HRQoL was assessed in one trial using SF-36. The trial concluded that supervised exercise therapy after DRF did not increase HRQoL [31].

Quality of Evidence

Risk of Bias

Randomization was applied in all 15 trials. However, only six trials provided sufficient details leading to low risk of bias (Table 4). Blinding of patients in these study settings was not possible, and therefore the risk of bias was rated high in this domain in all trials. Blinding of the outcome assessors was rated as unclear in seven of the trials. Based on adherence of the methods described in the trials, selective reporting of outcomes was not found to be an overall problem, and in most cases the risk of bias was rated as low. However, only a few trials had prior protocol registrations or published protocols, and thus the assessment of selective reporting of outcomes was not comprehensive.

Table 4 Risk of bias assessment (RoB2) for the 15 included studies
Full size table

GRADE assessment

As a result of incomparable outcome measures, difference in time to follow-up, and high I2 values, meta-analysis could only be undertaken on a limited number of outcomes. Thus, GRADE assessment was based on the meta-analysis and on the substance of the narrative synthesis. As a result of risk of bias, imprecision, and inconsistency of the trial results, the quality of evidence on all outcomes was found to be low or very low (Tables 5, 6, 7 and 8).

Table 5 Summary of findings for each outcome. This table provides information on the effects of the intervention and the overall quality of evidence (GRADE) for each outcome
Full size table
Table 6 Summary of findings for each outcome. This table provides information on the effects of the intervention and the overall quality of evidence (GRADE) for each outcome
Full size table
Table 7 Summary of findings for each outcome. This table provides information on the effects of the intervention and the overall quality of evidence (GRADE) for each outcome
Full size table
Table 8 Summary of findings for each outcome. This table provides information on the effects of the intervention and the overall quality of evidence (GRADE) for each outcome
Full size table

Discussion

This systematic review and meta-analysis suggests that early mobilization within 2 weeks of fracture may result in better function after PHF. However, these findings are based on low quality evidence. Furthermore, the overall MD (4.55) on the CS is not considered to be a clinically important difference, which has been reported to be between 5.4 and 11.6 [34]. Moreover, no evidence showed early mobilization after PHF has a clear positive effect on ROM or pain. Neither did it lead to more complications. No eligible evidence was found supporting early mobilization to be superior to late mobilization after DRF in terms of improved wrist function, grip strength, HRQoL, or reduced pain. Finally, no clear evidence showed a clear benefit of supervised exercise therapy compared with non-supervised home exercises on function of the upper limb, HRQoL, or reduced pain after PHF or DRF.

These results confirm the conclusions of previous systematic reviews that have reported a lack of clear evidence to support the decision on when to commence exercise therapy and to what extent it should be supervised [6, 35].

Limitations

The low number of included trials is a limitation of the present study, and changes to the inclusion criteria may have resulted in a larger study sample. Several trials did not include PROMs as an outcome measure, and therefore most of the measurements were clinician assessed and do not necessarily reflect the patients’ own perception of function. Several trials did not provide sufficient information on the inclusion and exclusion criteria. Furthermore, sample size calculation was only reported in 6 out of the 15 trials [16, 17, 25, 28, 29, 31] and only 4 trials referred to a minimal clinical important difference for their main outcome [16, 17, 28, 31]. This adds additional uncertainty to the reported results.

Implications

The consequences of immobilizing elderly people who have sustained a PHF or DRF for longer than necessary is an important consideration. A longer period of immobilization may lead to physical inactivity, with an increased risk of compromising general health [36]. The present study identified three unpublished RCTs that had investigated the effects of early mobilization after PHF, leading us to believe that more evidence will be available in the future [19, 20, 37].

Rehabilitation after upper limb fractures must be viewed from a broader and more complex perspective than that investigated in this systematic review. Therefore, other independent risk factors for poor function after these type of fractures, such as social deprivation, low self-efficacy, and fear of movement, must be acknowledged [38, 39].

Conclusion

There is an urgent need for high-quality randomized controlled trials to substantiate the current evidence regarding the optimal time to initiate mobilization and the need for supervision after PHF and DRF.