Abstract
This study reports 3-year outcomes of a randomized trial that compared close contact casting vs the usual surgery for unstable ankle fractures in older adults and that found equivalent ankle function outcomes at 6 months.
A randomized clinical trial of close contact casting vs the usual practice of surgery for treating unstable ankle fractures in older adults found equivalent ankle function outcomes at 6 months. Higher rates of radiological ankle malunion in the casting vs surgical groups (15% vs 3%, respectively) and nonunion (medial malleolus: 7% vs 1%, respectively) suggested that equivalence between the 2 groups may be lost if symptoms or functional limitations from posttraumatic arthritis manifest later. A follow-up at least 3 years after randomization was conducted to determine if equivalence persisted over time.
Methods
This study was a prespecified extended follow-up of a pragmatic, multicenter, equivalence randomized clinical trial. The National Research Ethics Service, Oxfordshire, gave approval; written informed consent was obtained. Participants were adults older than 60 years with acute unstable malleolar fracture(s) from 24 UK centers. Participants had received surgery (usual local practice internal fixation) or close contact casting, in which a minimally padded cast was applied after closed fracture reduction by an orthopedic surgeon in an operating room under anesthesia. Data were collected for at least 3 years after randomization using patient-reported postal questionnaires.
The primary outcome measure for the original trial was the Olerud and Molander Ankle Score (OMAS; range, 0-100, higher scores = better ankle function) at 6 months, with a prespecified equivalence margin of ±6 points. Extended follow-up used the same primary outcome and equivalence margin and assessed quality of life and pain as secondary outcomes (Table 1). A post hoc analysis of additional operations after 6 months was also conducted. Per-protocol primary analysis was used, consistent with the main trial. Random-effects models estimated mean differences and 95% CIs between treatments adjusted for age, sex, fracture pattern, baseline score, and time to follow-up, including the center variable as a random effect. The random-effects model was also used post hoc to assess differences in OMAS for participants with vs without radiological malunion and nonunion at 6 months. Change from baseline score was analyzed for outcomes without normal distribution. The primary outcome at extended follow-up assessed equivalence with the null hypothesis that the 2 groups were not equivalent. For all other outcomes, tests were 2-sided with a P value of .05 or less for significance. Analyses were conducted with Stata (StataCorp), version 15.0.
Table 1. Baseline Characteristics and 6-Month Outcomes of Responders and Nonresponders to 3-Year Follow-up for Casting vs Surgery for Ankle Fractures in Older Adults.
| Characteristics | Responders, Mean (SD) (n = 450) |
Nonresponders, Mean (SD) (n = 170) |
||
|---|---|---|---|---|
| Surgery Group (n = 226) | Casting Group (n = 224) | Surgery Group (n = 83) | Casting Group (n = 87) | |
| Age, y | 68.5 (6.2) | 70.7 (7.1) | 73.6 (7.6) | 73.5 (8.5) |
| Sex, No. (%) | ||||
| Men | 63 (27.9) | 56 (25.0) | 19 (22.9) | 22 (25.3) |
| Women | 163 (72.1) | 168 (75.0) | 64 (77.1) | 65 (74.7) |
| Ankle fracture classification, No. (%) | ||||
| Infrasyndesmotic or trans-syndesmotic | 197 (87.2) | 192 (85.7) | 75 (90.4) | 78 (89.7) |
| Suprasyndesmotic | 29 (12.8) | 32 (14.3) | 8 (9.6) | 9 (10.3) |
| Olerud-Molander Ankle Score, preinjurya,b | 91.2 (16.3) | 89.7 (16.8) | 86.1 (18.4) | 82.4 (18.9) |
| SF-12 mental score preinjurya,b | 54.0 (7.8) | 55.1 (7.1) | 53.1 (8.8) | 52.9 (8.2) |
| Missing data, No. | 1 | 0 | 1 | 0 |
| SF-12 physical score preinjurya,b | 51.8 (8.4) | 50.6 (9.9) | 49.3 (9.6) | 47.0 (11.1) |
| Missing data, No. | 1 | 0 | 1 | 0 |
| EQ-5D score preinjurya,c,d | 0.91 (0.17) | 0.89 (0.17) | 0.89 (0.15) | 0.83 (0.23) |
| Missing data, No. | 20 | 17 | 11 | 13 |
| EQ-5D score day of randomizationc,d | 0.06 (0.25) | 0.08 (0.25) | -0.01 (0.27) | 0.03 (0.26) |
| Missing data, No. | 32 | 29 | 17 | 18 |
| Mini-Mental State Examination scorec | 28.4 (1.9) | 28.2 (2.1) | 27.4 (2.5) | 27.1 (2.8) |
| Missing data, No. | 20 | 17 | 12 | 13 |
| Medical history, No. (%) | ||||
| Heart disease | 29 (12.8) | 30 (13.4) | 9 (10.8) | 14 (16.5) |
| Non–insulin-dependent diabetes | 20 (8.9) | 17 (7.6) | 11 (13.3) | 9 (10.6) |
| Cerebrovascular accident/transient ischemic attack | 10 (4.4) | 12 (5.4) | 4 (4.8) | 9 (10.6) |
| Osteoarthritis | 55 (24.3) | 70 (31.3) | 29 (34.9) | 30 (35.3) |
| Depression | 24 (10.6) | 26 (11.6) | 11 (13.3) | 12 (14.1) |
| Clinical Outcomes at 6 moe | Surgery (n = 225) | Casting (n = 224) | Surgery (n = 73) | Casting (n = 70) |
| Olerud-Molander Ankle Scoreb | 67.6 (20.3) | 65.7 (22.2) | 61.2 (23.6) | 61.4 (22.7) |
| SF-12 mental scoreb | 52.9 (9.7) | 53.0 (9.4) | 49.1 (12.2) | 49.7 (10.2) |
| SF-12 physical scoreb | 46.4 (9.7) | 45.2 (10.2) | 43.1 (10.7) | 41.2 (11.3) |
Abbreviations: EQ-5D, EuroQol 5 dimensions questionnaire; SF-12, 12-item Short-Form Health Survey.
Participants recalled preinjury status.
Range, 0 to 100; higher scores indicate better function.
The majority of missing scores relate to early study participants, before the measure was introduced.
Range, typically from 0 (death) to 1 (perfect health); negative scores can be obtained, reflective of a patient’s quality of life being worse than death.
Missing data for patients: responders, 1 in the surgery group; nonresponder, 1 in the casting group.
Results
From September 2013 through November 2016, 450 of the 620 randomized participants (73%) responded to follow-up at a median of 3 years (range, 2.9-9.5). Responders and nonresponders had similar characteristics (Table 1). Most responders lived in their own home (209 of 222 participants [94%] in the surgery group and 196 of 206 participants [95%] in the casting group).
Surgery and casting participants had equivalent ankle function (mean OMAS: 79.4 in the surgery group vs 76.3 in the casting group; difference, −1.3 [95% CI, −5.6 to 3.0]) and no significant differences in quality of life or pain (Table 2). Twenty-two of 222 surgery participants (10%) and 17 of 206 casting participants (8%) had operations after 6 months, including surgical implant removals (15 in the surgery group [7%] vs 8 in the casting group [4%]), arthrodesis (1 in the surgery group [0.5%] vs 3 in the casting group [1.5%]), arthroplasty (1 in the surgery group [0.5%] and 1 in the casting group [0.5%]), and infection-related procedures (2 in the surgery group [1%] and 0 in the casting group). Five casting participants (2%) had internal fixations for nonunion after 6 months. There was 1 internal fixation revision, 1 arthroscopy, and 1 hindfoot osteotomy among surgery participants. In post hoc analysis, from randomization to extended follow-up, mean total operating room procedures per participant (per protocol) were 1.2 (SD, 0.5) in the surgery group and 1.3 (SD, 0.6) in the casting group, and mean total surgical procedures per participant were 1.2 (SD, 0.5) in the surgery group and 0.3 (SD, 0.6) in the casting group.
Table 2. Primary and Secondary Outcomes at 3-Year Follow-up Among Responders to Follow-up for Casting vs Surgery for Ankle Fractures in Older Adults (Per-Protocol Analysis).
| Measure | Surgery Group | Casting Group | Adjusted Difference (95% CI)a | ||
|---|---|---|---|---|---|
| No. of Participants | Mean (95% CI) | No. of Participants | Mean (95% CI) | ||
| Olerud-Molander Ankle Scoreb | 220 | 79.4 (76.3 to 82.5) | 203 | 76.3 (72.7 to 79.9) | −1.3 (−5.6 to 3.0) |
| SF-12 mental scoreb | 204 | 51.9 (50.7 to 53.2) | 190 | 52.7 (51.3 to 54.0) | 0.5 (−1.2 to 2.2) |
| SF-12 physical scoreb | 204 | 47.2 (45.8 to 48.7) | 190 | 44.7 (42.9 to 46.4) | −0.8 (−2.6 to 1.1) |
| EQ-5D score (change from baseline)c | 187 | −0.75 (−0.79 to −0.71) | 177 | −0.68 (−0.74 to −0.63) | 0.04 (−0.01 to 0.09) |
| Olerud-Molander Ankle Score pain ratingd | 221 | 1.8 (1.7 to 1.9) | 205 | 1.9 (1.8 to 2.1) | 0.09 (−0.1 to 0.3) |
| EQ-5D score pain ratinge | 219 | 1.5 (1.4 to 1.6) | 205 | 1.5 (1.5 to 1.6) | 0.05 (−0.1 to 0.2) |
Abbreviations: EQ-5D, EuroQol-5 dimension questionnaire; SF-12, 12-Item Short-Form Health Survey.
Differences were adjusted for baseline outcome values, age, sex, recruitment hospital, fracture pattern (trans-syndesmotic and infrasyndesmotic vs suprasyndesmotic) and time to follow-up.
Range, 0 to 100; higher scores indicate better function.
Range, typically from 0 (death) to 1 (perfect health); negative scores can be obtained, reflective of a patient’s quality of life being worse than death.
Scores were from 1 to 5; 1 indicates “none” and 5 indicates “constant and severe.”
Score were from 1 to 3; 1 indicates “no pain or discomfort” and 3 indicates “extreme pain or discomfort.”
Of 67 participants with radiological abnormalities at 6 months, 43 (64%) provided extended follow-up data. Those with malleolar malunion at 6 months had significantly lower OMAS scores (n = 30; mean, 58.7 [SD, 33.1]) than those without (n = 372; mean, 79.8 [SD, 23.5]; mean difference, −16.4 [95% CI, −25.0 to −7.8]; P < .001). Participants with medial malleolar nonunion at 6 months also had significantly lower OMAS scores (n = 13; mean, 55.4 [SD, 38.5]) than those without (n = 388; mean, 79.1 [SD, 23.9]; mean difference, −13.9 [95% CI, −26.6 to −1.2]; P = .03). Treatment was not a significant covariate in these analyses.
Discussion
Equivalence in function between casting and immediate surgery strategies was maintained at 3 years. In post hoc analyses, participants with radiological malunion and medial malleolar nonunion at 6 months had lower OMAS scores at 3-year follow-up. These longer-term outcomes will support surgeon and patient decision making. The findings indicate that treatment of ankle fractures in older adults should focus on obtaining and maintaining a reduction until union, by the most conservative means possible.
The study was limited by its reliance on self-reported events requiring participant recall and by loss to follow-up. However, there were sufficient data to estimate and conclude equivalence in the primary outcome.
Section Editor: Jody W. Zylke, MD, Deputy Editor.
References
- 1.Willett K, Keene DJ, Mistry D, et al. . Close contact casting vs surgery for initial treatment of unstable ankle fractures in older adults. JAMA. 2016;316(14):1455-1463. [DOI] [PubMed] [Google Scholar]
- 2.Sanders DW. Close contact casting vs surgery for unstable ankle fractures. JAMA. 2016;316(14):1451-1452. [DOI] [PubMed] [Google Scholar]
- 3.Olerud C, Molander H. A scoring scale for symptom evaluation after ankle fracture. Arch Orthop Trauma Surg. 1984;103(3):190-194. [DOI] [PubMed] [Google Scholar]
- 4.Rabin R, de Charro F. EQ-5D. Ann Med. 2001;33(5):337-343. [DOI] [PubMed] [Google Scholar]
- 5.Gandek B, Ware JE, Aaronson NK, et al. . Cross-validation of item selection and scoring for the SF-12 Health Survey in 9 countries. J Clin Epidemiol. 1998;51(11):1171-1178. [DOI] [PubMed] [Google Scholar]