Volar Locked Plating Versus Closed Reduction and Casting for Acute, Displaced Distal Radial Fractures in the Elderly

Abstract

Background:

It remains unclear whether volar locked plating (VLP) yields a better functional outcome than closed reduction and casting (CRC) for elderly patients with an acute, displaced distal radial fracture. Our purpose was to conduct a systematic review and meta-analysis of randomized controlled trials comparing outcomes of VLP and CRC for elderly patients (age, ≥60 years).

Methods:

Multiple databases, including MEDLINE, were searched for randomized controlled trials evaluating outcomes following distal radial fracture treatment. Raw data were obtained for studies that included patients of all ages, and the elderly subgroup was included for analysis. The primary outcome was the Disabilities of the Arm, Shoulder and Hand (DASH) score at ≥1 year of follow-up. Secondary outcomes included the 3-month DASH score, range of motion, final radiographic alignment, and complications. Effect sizes for the comparison of each outcome between groups were pooled across studies using random-effects models with the inverse variance weighting method. Changes in DASH score were compared with a minimal clinically important difference (MCID) estimate of 10 to assess clinical relevance.

Results:

Of 2,152 screened articles, 6 were included. Demographics were similar for the 274 VLP and 287 CRC patients. DASH scores were significantly better following VLP than CRC at the time of final follow-up (12 to 24 months postoperatively; score difference, −5.9; 95% confidence interval [CI], −8.7 to −3.1) and at 3 months (−8.9; 95% CI, −13.0 to −4.8). VLP yielded significantly better palmar tilt, radial inclination, and supination, with no differences in ulnar variance, flexion-extension, pronation, or total complication rates.

Conclusions:

Functional outcome was significantly better following VLP than CRC 3 months into the treatment of acute, displaced distal radial fractures in an elderly population and up to 2 years after injury. However, the observed differences in the final DASH score did not exceed published estimates of the MCID, suggesting that clinical outcomes are similar for both treatment options.

Level of Evidence:

Therapeutic Level I. See Instructions for Authors for a complete description of levels of evidence.

Distal radial fractures (DRFs) are one of the most common orthopaedic injuries¹. DRF incidence has increased over the past 4 decades¹, and follows a bimodal distribution with an initial peak for patients aged 18 to 25 years and a second peak for elderly patients in their sixties and older^1,2. The frequency of operative treatment is also increasing^3,4 despite literature suggesting that open reduction and internal fixation (ORIF) is more costly than other treatment alternatives, including closed reduction and casting (CRC) or pinning^5,6. Regardless, ORIF with volar locked plating (VLP) has become the mainstay surgical technique for these fractures^7-9. When considering these trends, and the $170 million spent for DRF care in 2007 alone for Medicare patients in the United States, it is clear that efforts to better understand the advantages and disadvantages of DRF treatment options will assist providers with counseling patients regarding the appropriate treatment plan.

Specifically, the literature offers conflicting evidence on the treatment of elderly patients with a displaced DRF^10,11. The American Academy of Orthopaedic Surgeons (AAOS) 2013 appropriate use criteria on the treatment of DRFs state “We are unable to recommend for or against operative treatment for patients over age 55 with distal radius fractures,” with an “inconclusive” strength of recommendation¹¹. Although multiple outcome studies specific to the elderly population have been published since the release of these criteria, the optimal treatment remains controversial. Some studies suggest that VLP yields better outcomes than CRC^12-15, while others have demonstrated similar outcomes^16-22.

Our aim was to perform a systematic review and meta-analysis of randomized controlled trials (RCTs) comparing ORIF using VLP to conservative management using CRC for elderly patients. Our primary null hypothesis was that no difference in the Disabilities of the Arm, Shoulder and Hand (DASH) score exists between VLP and CRC groups at the time of final follow-up. Our secondary null hypotheses were that the DASH score at the 3-month follow-up, final range of motion and radiographic alignment measurements, and total complication rate would not differ between VLP and CRC groups.

Materials and Methods

Search Method and Strategy

This study was registered with PROSPERO, an international database of prospectively registered systematic reviews (CRD42020151527, 2/27/2020), and was exempt from institutional review board evaluation at our institution. PRISMA (Preferred Reporting Items for Systematic reviews and Meta-Analyses) guidelines were followed throughout the study process^23,24. MEDLINE, Embase, CINAHL (Cumulative Index of Nursing and Allied Health Literature), CENTRAL (Cochrane Central Register of Controlled Trials), Scopus, Web of Science Core Collection, LILACS (Latin American and Caribbean Health Sciences Literature), Google Scholar, clinicaltrials.org, ISRCTN (International Clinical Trials Registry Platform), and OpenGrey (System for Information on Grey Literature in Europe) databases were searched through August 13, 2019. A comprehensive and sensitive search strategy was formulated through consultation with a medical library strategist (M.M.M.) following PRESS (Peer Review of Electronic Search Strategies) guidelines (see Appendix Table I)^25,26.

Study Selection and Data Extraction Methods

Study inclusion required an RCT study design in which DASH scores were compared between VLP and CRC groups at ≥1 year of follow-up. The initial search was not restricted by patient age. Authors of pertinent RCTs lacking age restrictions were contacted in an attempt to obtain their raw data in order to include the elderly subgroup in our analysis.^14,27 “Elderly” was defined a priori as ≥60 years of age, consistent with the lower-bound age cutoff used in the 4 pertinent elderly-specific RCTs identified prior to initiating the systematic review^12,15,21,28.

Using Covidence systematic review management software²⁹, duplicate studies were removed, and the remaining studies were screened by 2 independent reviewers (N.H.K. and A.R.S.). For studies passing the initial title and abstract screening, the corresponding full-text article was then evaluated and any reasons for exclusion were tabulated (see Appendix Table II). Disagreements in screening were resolved through joint consensus with a third reviewer (A.R.T.).

Study Quality Assessment

The Cochrane Risk of Bias (RoB) tool was used to grade included studies. Data extraction was performed by the lead author (A.R.S.) and audited by the senior author (N.H.K.). Data extracted included patient demographics, fracture classification, dominant-hand involvement, DASH score, range of motion, radiographic measurements, and complication rates. Current literature recommends inclusion of data from a minimum of 5 RCTs in a meta-analysis in order to achieve appropriate power³⁰. Accordingly, variables reported in <5 studies at a given postoperative time point were excluded from the meta-analysis.

Study Design

The primary outcome was the DASH score at ≥1 year postoperatively. Secondary outcomes included the 3-month DASH score, final range of motion and radiographic alignment measurements, and complication rates. Variation among the included studies was noted with regard to the scenario in which the development of an unacceptable loss of reduction after CRC leads to surgery using VLP (i.e., crossover to the other treatment group): some included this as a complication, whereas others did not. For the current study, this scenario was considered to align with routine practice patterns and was not considered a complication. The Patient-Rated Wrist Evaluation (PRWE), grip strength, intra-articular step-off, and radioulnar deviation were extracted but were not included in the meta-analysis because <5 studies reported these outcomes at any given time point.

Statistical Analysis

Random-effects models using the inverse variance weighting method were used to pool patient characteristics, outcomes, and effect sizes for the comparison of VLP and CRC groups³¹. Standard deviations were approximated as range/4 when not reported³². Heterogeneity among studies was examined using I², tau², and a chi-square Q test (with a p value of <0.05 indicating significant heterogeneity). Mean differences in final and 3-month DASH scores, and risk ratios with 95% confidence intervals (CIs) for complication rates between VLP and CRC groups, are presented as forest plots. DASH scores were compared with the previously reported minimal clinically important difference (MCID) of 10 to determine clinical relevance^33,34. Potential publication bias was assessed by a funnel plot (see Appendix Fig. I). All analyses were conducted using R (version 3.5; R Foundation for Statistical Computing)³⁵. Meta-analyses were performed using the “meta” package in R³⁶. P < 0.05 was considered significant, and all tests were 2-sided.

Results

Figure 1 illustrates the article screening process. Of the 6 included RCTs, 4 specifically involved elderly patients (age, ≥60 years)^15,17,21,28. The other 2 RCTs included both elderly and non-elderly patients^14,27, and the authors of those authors provided their raw data to enable inclusion of their elderly patients in the meta-analysis. Risk-of-bias assessment of the included studies is provided in Table I, and no study was excluded because of a concern for bias.

Fig. 1.

PRISMA flow diagram of study selection.

TABLE I.

Risk of Bias in Included Studies

Study	Sequence Generation	Allocation Concealment	Blinding of Participants and Personnel	Blinding of Outcome Assessors	Incomplete Outcome Data	Selective Outcome Reporting	Other
Arora21 (2011)	Low	Low	High	Low	Low	Low	Low
Bartl17 (2014)	Low	Low	High	Low	Low	Low	Low
Martinez-Mendez28 (2018)	Low	Low	High	Low	Low	Low	Low
Sirniö27 (2019)	Low	Low	High	Low	Low	Low	Low
Mulders14 (2019)	Low	Low	High	Low	Low	Low	Low
Saving15 (2019)	Low	Low	High	Low	Low	Low	Low

Study design characteristics are summarized in Table II. There were no significant differences in patient demographics or fracture type between the 274 VLP and 287 CRC patients (Table III). Final follow-up was reported for 253 VLP and 272 CRC patients. Additional demographic information is provided in Appendix Table III.

TABLE II.

Summary of Characteristics of Included Studies^*

Study	Study Design	LOE	Age for Inclusion (yr)	Inclusion Criteria	Definition of Acceptable Reduction	No. of Patients		No. of CRC Patients Who Required VLP	ITT Analysis	No. of Fractures: A2, A3		No. of Fractures: C1, C2, C3		Time to Last Radiographic Measurement (mo)	Last Follow-up† (mo)
						VLP	CRC			VLP	CRC	VLP	CRC
Arora21 (2011)	RCT‡	I	65	Independent living; available for follow-up; isolated injury; inadequate reduction or loss of reduction at 1-wk follow-up	<10° dorsal angulation; <3 mm radial shortening; <2 mm articular step-off	36	37	0	NA	3, 7	3, 9	4, 12, 10	11, 8, 6	12	12
Bartl17 (2014)	RCT	I	65	Unstable intra-articular fracture	Unclear	86	88	37	Y	0, 0	0, 0	36, 35, 15	40, 35, 13	3	12
Martinez-Mendez28 (2018)	RCT	I	60	Displaced complex intra-articular fracture, not open; history of stroke	>15° radial inclination, <15° dorsal angulation, <2 mm ulnar variance, <2 mm step-off	50	47	0	NA	0, 0	0, 0	23, 23, 4	22, 20, 5	24	24
Sirniö27 (2019)	RCT	I	60‡	Isolated displaced fracture, not open; initial visit within 1 wk of injury; acceptable initial reduction; no previous wrist/hand disorder limiting function	<10° dorsal angulation; >15° radial inclination; <2 mm ulnar variance	23	29	12	Y	16§	18§	7§	11§	24	24
Mulders14 (2019)	RCT	I	60‡	Isolated displaced extra-articular fracture, not open; no impairment of wrist prior to injury; sufficient mental capacity	>15° radial inclination; <15° dorsal angulation; <20° palmar angulation; <5 mm radial height loss	21	22	0	NA	5, 16	3, 19	0, 0, 0	0, 0, 0	12	12
Saving15 (2019)	RCT	I	70#	>20° dorsal tilt; extra-articular fracture or intra-articular fracture with <1 mm step-off; isolated non-high-energy traumatic injury; no prior impairment of either wrist; no psychiatric disorders or substance abuse; independent in daily living; ASA class 1-3; initial visit within 6 days of injury	<20° dorsal angulation; <4 mm axial shortening	58	64	0	NA	6, 33	10, 28	11, 7, 1	20, 6, 0	12	12

^*LOE = Level of Evidence, VLP = volar locking plate, CRC = closed reduction and casting, ITT = intent-to-treat, RCT = randomized controlled trial, NA = not applicable, and ASA = American Society of Anesthesiologists.

^†Last follow-up for all variables excluding radiographic measurements. Last follow-up was the same for VLP and CRC groups for each study.

^‡Raw data set provided by authors; age cutoff of 60 was defined a priori.

^§Fracture type was provided as the total for type A and for type C.

^#Initially 75, but lowered at 1 institution halfway through the study due to limited enrollment.

TABLE III.

Pooled Estimates of Baseline Demographic Factors

Variable	No. of Studies	Volar Locked Plating*	Closed Reduction and Casting*	P Value
Mean age (yr)	6	72.2 (67.5, 76.9)	72.3 (68.3, 76.4)	0.98
Female (%)	6	87.3 (80.0, 93.1)	88.1 (78.6, 95.1)	0.74
Dominant hand injured (%)	4	53.1 (41.4, 64.5)	53.9 (41.2, 66.1)	0.97
Type A (%)	6	39.0 (4.0, 83.2)	37.3 (4.1, 80.3)	0.66
Type C (%)	6	61.0 (16.8, 96.0)	62.7 (19.7, 95.9)	0.66

^*The values are given as the estimate, with the 95% CI in parentheses. The total number of patients was 274 in the VLP group and 287 in the CRC group.

Primary Outcome

DASH scores were available for all studies at the time of final follow-up. The pooled DASH score at the time of final follow-up was 11.6 (95% CI, 8.1 to 15.1) for the VLP group compared with 18.1 (95% CI, 11.6 to 24.6) for th@e CRC group, representing a difference (in favor of VLP) of −6.5 points (p < 0.001). There was no significant heterogeneity in reported effect sizes across studies (Q test, p = 0.33), with 13% of the observed variance attributed to actual differences among studies (I² = 13%). Random-effects modeling demonstrated a significantly better DASH score at the time of final follow-up in the VLP group, by a margin of −5.9 points (95% CI, −8.7 to −3.1), compared with the CRC group (Fig. 2).

Fig. 2.

Random-effects model of DASH scores at the time of final follow-up.

Secondary Outcomes

DASH scores were available for all but 1 study²⁸ at 3 months of follow-up. The pooled DASH estimate was 18.2 (95% CI, 14.4 to 22.0) for VLP and 27.5 (95% CI, 23.9 to 31.1) for CRC (p < 0.001). Random-effects modeling demonstrated that VLP yielded a significantly better DASH score at 3 months, by a margin of −8.9 points (95% CI, −13.0 to −4.8), compared with CRC (Fig. 3). There was no significant heterogeneity in reported effect sizes across studies (Q test, p = 0.18), with 36% of the observed variance attributed to actual differences among studies (I² = 36%).

Fig. 3.

Random-effects model of DASH scores at 3 months of follow-up.

Final radiographic and range-of-motion outcomes are provided in Table IV. Palmar tilt and radial inclination were both significantly better for the VLP group, whereas there was no significant difference in ulnar variance between groups. Supination was significantly greater for the VLP group, but no differences in flexion, extension, or pronation were observed.

TABLE IV.

Pooled Estimates of Radiographic and Range-of-Motion Outcomes

Variable	No. of Studies	Volar Locked Plating*	Closed Reduction and Casting*	P Value
Radiographic outcomes
Palmar tilt (deg)	6	3.3 (1.1, 5.4)	−4.2 (−9.7, 1.4)	<0.001
Radial inclination (deg)	6	20.3 (19.5, 21.1)	15.7 (13.9, 17.6)	<0.001
Ulnar variance (mm)	5	0.8 (−0.4, 2.0)	1.2 (−0.2, 2.6)	0.70
Range of motion
Flexion (deg)	5	62.8 (55.6, 69.9)	59.2 (54.1, 64.3)	0.32
Extension (deg)	5	63.5 (56.2, 70.7)	59.1 (54.1, 64.1)	0.93
Supination (deg)	5	86.7 (83.2, 90.3)	81.8 (75.7, 87.9)	0.027
Pronation (deg)	5	85.2 (83.5, 86.8)	82.9 (79.0, 86.8)	0.26

^*The values are given as the estimate, with the 95% CI in parentheses. The total number of patients was 253 in the VLP group and 272 in the CRC group.

Complications

The overall complication rate was similar for VLP and CRC based on analysis of pooled estimates (Table V) and the random-effects model (Fig. 4). However, there was substantial heterogeneity in risk ratios across studies (0.6 to 3.7), and 59% of the total variance was attributed to inherent differences between studies (I² = 0.59; Q test, p = 0.03). The rates of carpal tunnel syndrome requiring carpal tunnel release and of infection were similar between groups (Table V). See Appendix Table IV for specific complications reported in each study.

Fig. 4.

Random-effects model of total complications.

TABLE V.

Pooled Estimates of Complications

Complication	No. of Studies	Volar Locked Plating*	Closed Reduction and Casting*	P Value
Carpal tunnel syndrome requiring surgical intervention (%)	6	1.9 (0.6, 3.9)†	2.5 (0.3, 7.0)‡	0.78
Infection (%)	6	0.5 (0.0, 2.8)	0.0 (0.0, 0.6)	0.25
Total complications (%)	6	18.8 (8.9, 35.5)	14.2 (5.7, 31.3)	0.72

^*The values are given as the estimate, with the 95% CI in parentheses. The total number of patients was 253 in the VLP group and 272 in the CRC group.

^†Carpal tunnel release and hardware removal.

^‡Carpal tunnel release.

Discussion

The primary study finding was that open reduction followed by application of a volar plate yielded significantly better DASH scores, by a margin of −5.9 points, compared with closed reduction followed by application of a cast, in elderly patients (≥60 years of age) with an acute, displaced fracture of the distal aspect of the radius at a minimum of 12 months postoperatively. This significant difference in the DASH score should be carefully interpreted, however, as small differences in outcome scores may not be clinically relevant.

The MCID is a measure that defines the smallest change in an outcome score that may be perceived as clinically relevant from the patient perspective³⁷. For the DASH score, MCID estimates range from 7.9 for patients recovering from simple cubital tunnel decompression³⁸ up to 17.1 for patients with diagnoses involving the distal aspect of an upper extremity³⁹. Additional estimates include 13 in an elective shoulder surgery cohort⁴⁰ and 10³³ and 10.8³⁴ for a general hand cohort and an upper-extremity physical therapy cohort, respectively.

The point estimate for the final DASH score difference between VLP and CRC groups was 5.9, which is lower than all of the published DASH MCID estimates. Furthermore, the 95% CI for this estimate does not contain the commonly utilized anchor-based MCID estimate of 10; consequently, it can be stated with 95% confidence that the score difference between VLP and CRC groups does not exceed this MCID estimate³³. Therefore, we conclude that upper-extremity function does not differ meaningfully between VLP and CRC cohorts of elderly patients with an acute, initially displaced DRF at a minimum 1-year follow-up.

To our knowledge, no DRF-specific MCID for DASH scores has yet been calculated. Thus, it is possible that a DRF-specific anchor-based MCID value would prompt a reversal of our study conclusion. However, that is unlikely given that even the low-end estimate for the smallest detectable change reported in the literature (7.9) is greater than the DASH score difference between VLP and CRC groups at the time of final follow-up⁴¹.

Secondary findings in our study include a significantly better DASH score for VLP compared with CRC at 3 months of follow-up, with a difference of −8.9 points. Although lower than the MCID estimate of 10, the 95% CI for this estimate did include 10 points of improvement. Therefore, we cannot rule out a 10-point actual difference, and it is possible that VLP yields a clinically relevant benefit in functional outcomes at this early time point. Additionally, palmar tilt, radial inclination, and supination were significantly better for the VLP group at the final assessment. Finally, no differences in the overall complication rate or carpal tunnel syndrome requiring release were observed between groups.

Our results must be interpreted in context. For the CRC group, the final dorsal tilt was 4° with ulnar variance of +1.2 mm and radial inclination of 16°, suggesting that these fractures were aligned well^10,11. Therefore, we conclude that the final functional outcomes of VLP and CRC are similar when adequate alignment has been maintained following CRC. It deserves specific mention that our results do not allow for conclusions to be drawn regarding the relative outcomes of VLP versus nonoperative management of fractures with poor alignment. As such, we advise caution against extrapolation of our results, particularly by health-care policy makers and payers, which could misleadingly be interpreted to suggest that surgery is never indicated for this patient population. Such a conclusion would require a different study design that includes elderly patients with displaced fractures that remain unreduced or lose adequate reduction. Shared decision-making should be utilized when determining the best treatment option for patients with unacceptable loss of reduction.

Prior systematic reviews and meta-analyses, all of which included at most 2 of the RCTs in the current study, demonstrate various levels of agreement with our findings. The 2016 systematic review and meta-analysis by Chen et al.⁴² found no significant difference in pain scores, functional assessment using the DASH and PRWE, and range of motion despite better radiographic measurements and grip strength for surgically treated patients. Contrary to the current study, major complications were more frequent in the operative group. However, these observations were based on 6 retrospective studies and only 2 RCTs^17,21. The 2019 meta-analysis by Mellstrand Navarro et al. also found no difference in 12-month DASH scores between VLP and CRC, which is not surprising since the authors only included the same 2 RCTs as the Chen review in their analysis⁴³. The meta-analyses of Ju et al. and Song et al. both derived a similar conclusion through analysis of those 2 RCTs in addition to several comparative studies of lesser quality, despite better radiographic alignment following surgical treatment^44,45. The 2011 systematic review by Diaz-Garcia et al. pooled outcomes following 5 types of treatment for DRF in the elderly from multiple retrospective studies and RCTs. None of the included RCTs compared VLP with CRC. Significantly better radiographic parameters and 12-month DASH scores were observed following VLP compared with CRC. However, the difference in DASH scores was deemed to be clinically insignificant (weighted average scores of 8.2 for VLP versus 11.6 for CRC)⁴⁶. In contrast to our results, Vannabouathong et al. observed better functional outcomes at 12 months of follow-up for operative management compared with nonsurgical treatment⁴⁷. However, those authors included patients of all ages and used a subjective outcome, rather than a contemporary patient-reported outcome measure, as the primary outcome.

Limitations of the present study should be noted. We included only RCTs in an attempt to minimize bias associated with indicating of patients to operative versus nonoperative treatment in a nonrandomized fashion, and to minimize potential differences in baseline characteristics between treatment groups in light of evidence that differences in social support and depression levels affect DRF treatment outcomes^48-50. However, our study was limited by the number of RCTs available in the literature, and we were unable to include data from ongoing, unpublished RCTs^51,52. Because of the limited number of pertinent studies, we were unable to analyze several variables, including radioulnar deviation, intra-articular step-off, PRWE scores, and grip strength. Nonetheless, the literature suggests that functional outcomes do not correlate with improved range of motion, grip strength, or radiographic measurements, and therefore we believe that use of the DASH score as a primary outcome measure has merit^16,19,20,53. Although multiple DASH MCID estimates have been published, we were unable to identify DRF-specific estimates, which may limit the interpretation of our findings. The inclusion criteria within the 6 included RCTs were heterogeneous with respect to age cutoffs, definitions of acceptable alignment, fracture type, and inclusion versus exclusion of type-C fractures; therefore, our results apply to a heterogeneous population and it remains unclear whether certain elderly patient subsets may benefit from routine surgical treatment. Future studies with greater power may be able to determine whether certain patient groups benefit from VLP for a displaced DRF based on handedness, fracture type, degree of residual displacement, patient preferences, or patient comorbidities. Two included RCTs with treatment group crossover utilized an intent-to-treat analysis; although that is more rigorous than an as-treated analysis, it is possible that the inclusion of surgically treated patients in the CRC group biased the conclusions toward a negative result^17,27. Lastly, as our study was not a cost-effectiveness analysis, we cannot comment on the theoretical advantages of early return to work following VLP, or on the potential cost differences from a patient or societal perspective, for working elderly patients.

In conclusion, for elderly patients (≥60 years old) with an acute, displaced DRF, final DASH scores were significantly better following VLP than CRC, but this difference did not appear to reach the threshold of clinical relevance. Although VLP might yield a significant and clinically relevant benefit over CRC at 3 months as assessed with the DASH score, we conclude that both treatment strategies yield similar functional outcomes at a minimum of 1 year of follow-up when an adequate reduction can be maintained with CRC.

Appendix

Supporting material provided by the authors is posted with the online version of this article as a data supplement at jbjs.org (http://links.lww.com/JBJS/F899).