Clinical trial results

Study participants

Patients with an Achilles tendon rupture typically attend the emergency department at their local hospital and, following their diagnosis, are referred to the next available fracture/trauma clinic to discuss the management of their injury.

The flow of participants through the study is summarised in Figure 1. This includes details on the total number of patients referred to the trauma clinic with an Achilles tendon rupture and those randomised. The availability of the primary outcome for analysis is also reported by treatment group, as is the total number of patients excluded from the primary outcome analysis.

FIGURE 1 The UKSTAR CONSORT flow diagram. a, Two additional patients were randomised in error without giving consent to be in the study. These participants were excluded from all data analyses. Reproduced from Costa et al.49 © 2020 The Author(s). Published by Elsevier Ltd. This is an Open Access article distributed in accordance with the terms of the Creative Commons Attribution (CC BY 4.0) license, which permits others to distribute, remix, adapt and build upon this work, for commercial use, provided the original work is properly cited. See: http://creativecommons.org/licenses/by/4.0/. (Figure not reproduced here; see the original article.)

Recruitment

A total of 1076 eligible participants were screened from July 2016 to May 2018 from 39 NHS hospitals across England and Scotland (see Figure 1). Of these, 540 participants consented to take part in the trial. Reasons why patients were not included in the trial are presented. Participants attended clinic visits at the time of randomisation (baseline) and at the 8-week follow-up. Participants were also contacted by the trial team by post, e-mail or telephone to complete follow-up questionnaires at 3, 6 and 9 months post injury. Two participants were randomised in error before consenting and are therefore not included in the numbers allocated to each treatment group.

Baseline characteristics

The randomisation was stratified by centre, and the allocation of participants to the treatment groups in each centre and the overall numbers is given in Table 5. The descriptive characteristics of the participants included in the ITT population are summarised by treatment group and overall in Table 6. These values are presented as numbers and percentages for categorical factors and as means and SD or medians and IQR, as appropriate, for continuous variables. These variables all appear well balanced between the two treatment groups. The distribution of participant ages by sex at enrolment is shown in Figure 2. This distribution has a peak in male patients aged 30–40 years and in female patients aged 40–60 years. Baseline values of patient-reported outcome measures (PROMs), ATRS and EQ-5D-5L are summarised by group in Table 7. ATRS values range from 0 to 100, with lower scores indicating more functional limitations; EQ-5D utility scores range from –0.511 to 1, with higher scores indicating better QoL and 0 being equivalent to death; and EQ-5D VAS scores range from 0 to 100, with higher scores indicating better QoL. The values reported are similar in the two treatment groups.

FIGURE 2 Participant age (years) at randomisation by sex. (a) Male; and (b) female. (Figure not reproduced here; see the original article.)

Compliance

Participant compliance with treatment is presented in Table 8. The population compliant with treatment for ≥ 6 weeks was made up of 477 (88.2%) participants overall, 212 (79.7%) of whom were in the plaster cast group and 265 (96.7%) of whom were in the functional brace group. The numbers of patients compliant with treatment for ≥ 4 weeks and ≥ 2 weeks are also listed in Table 8.

Details of the treatment received following randomisation are listed in Table 9. There were 247 (92.9%) participants in the plaster cast group and 269 (98.2%) who received their allocated treatment immediately at baseline. Those who did not receive the allocated treatment at baseline received the other treatment instead or withdrew. A further 35 participants (13.2%) in the plaster cast group and 4 (1.5%) in the functional brace group changed from the treatment to which they were randomised within the first 6 weeks and received the other treatment or surgery instead. The reasons why participants changed from their allocated treatment are listed in Table 10 and include patient decision, clinician decision, medical resources unavailable, incorrect Achilles tendon rupture diagnosis, surgery and withdrawal.

Details of the treatment received in each group, including time point when the patient was allowed to fully bear weight, time point when cast/brace was removed, number of cast changes, type of brace used, number of heel wedges and VTE prophylaxis are listed in Table 11. The data collected show that the number of participants allowed to bear weight early was larger in the functional brace group than in the plaster cast group. Proportions of these patients are listed for each treatment group and are restricted to patients who did not change from their allocated treatment in the first 6 weeks following randomisation.

There were no differences between the two treatment groups in the time point when the plaster cast or functional brace was removed. On average, patients in the plaster cast group changed their plaster cast three times during the treatment period. The most frequently used functional brace make was Aircast (44.5%) and the median number of heel wedges was three, although the number varied from zero to five, depending on time point. VTE prophylaxis was offered to more patients in the plaster cast group (n = 187, 70.3%) than in the functional brace group (n = 158, 57.7%). The VTE treatments used were low-molecular-weight heparin and oral anticoagulant in similar proportions in the two groups.

Numbers analysed

The ITT population comprised all patients who were randomised and gave consent to participate in the trial. These patients were analysed in the group to which they were allocated, including in the analysis of PROMs. Data for the two randomised participants who did not give their consent were excluded from the analysis. The ITT population was made up of 540 participants overall, 266 of whom were in the plaster cast group and 274 of whom were in the functional brace group. Analyses of all primary and secondary outcomes were performed for this population.

The CACE population35 comprised all randomised participants who were compliant with treatment for ≥ 6 weeks.

Withdrawals

Table 12 provides the available data at each follow-up time point, including the number of CRFs and PROs and the number of participants who withdrew or died, according to treatment group. There was a good completion rate of CRFs and PROs across both treatment groups throughout the study period. In total, 10 patients withdrew (seven from the plaster cast group and three from the functional brace group). There were two deaths, both in the functional brace group. Reasons for withdrawal are listed in Table 13; these include clinician decision, patient decision and private treatment.

Analyses to address primary outcome

The primary outcome in this study is the ATRS measured at 9 months post injury, as described in the statistical analysis plan.

Achilles Tendon Rupture Score was assessed at baseline (pre injury), 8 weeks and 3, 6 and 9 months after the tendon rupture. The mean ATRS score and SD for each treatment group at each time point are provided in Table 14. The mean ATRS differences between the treatment groups was estimated based on a linear mixed-effects regression model both unadjusted and adjusted for the stratification factors age, sex and baseline ATRS as fixed effects, and recruitment centre and observations within participants as random effects. The adjusted analysis was prespecified as the principal analysis of the trial results.

The adjusted analysis showed no statistically significant difference in ATRS between the treatment groups at 9 months (–1.38, 95% CI –4.9 to 2.1). The 8-week follow-up results show a statistically significant difference in the ATRS in favour of the functional brace group (5.53, 95% CI 2.0 to 9.1); however, this effect fades during the 9-month follow-up. The marginal mean ATRS values from the mixed-effects model are presented from the 8-week follow-up to the 9-month follow-up in Figure 3.

FIGURE 3 Marginal mean ATRS values from the mixed-effects model and associated 95% CIs for the two treatment groups pre injury to 9 months post injury. ATRS values range from 0 to 100, with higher scores indicating better outcome. (Figure not reproduced here; see the original article.)

Supplementary fully adjusted analyses were carried out, accounting for further predefined prognostic variables, and included age, sex, baseline ATRS, diabetes mellitus and smoking status as fixed effects, with random effects for centre and observations within participant (Table 15). These results showed a similar between-group difference to those results of the primary analysis at 9 months post injury (–1.15, 95% CI –4.7 to 2.4).

Sensitivity analyses

Complier-average causal effect analysis

The number of participants compliant with the allocated treatment from ≥ 6 weeks, ≥ 4 weeks and ≥ 2 weeks is shown in Table 8. A CACE analysis was conducted to estimate the mean effect of treatment in compliers with treatment for ≥ 6 weeks (n = 265, 96.7% in the functional brace group). The numbers of participants compliant with treatment for ≥ 4 weeks and for ≥ 2 weeks were identical in the functional brace group (n = 268, 97.8%) and hence a CACE analysis was conducted including only patients compliant for ≥ 4 weeks.

We estimated the CACE using the xtivreg and xtset commands in Stata. The unadjusted and the adjusted analyses estimating the ITT effect and the CACE analysis effect are shown in Table 14. The unadjusted CACE estimate was marginally greater in modulus than the ITT (–1.70, 95% CI –5.3 to 1.9), but this difference was small given the ATRS scoring scale. The adjusted CACE analysis showed similar results to the ITT population analysis (–1.17, 95% CI –4.5 to 2.1).

Area under the curve

Parameter estimates from the mixed-effects models were used to calculate the AUC from the 8-week follow-up to the 9-month follow-up for a male participant of a mean age of 48.65 years. Results showing an overall estimate of recovery over time and a t-test comparison of the two treatment groups are presented in Table 15. Higher AUCs indicate better overall functionality. The functional brace group showed a better overall functionality than the plaster cast group; however, the difference (–5.26, 95% CI –24.66 to 14.14) was not statistically significant.

Prespecified subgroup analysis

There were no prespecified subgroups and therefore no subgroup analyses were conducted.

Analyses to address secondary outcomes

The secondary outcomes collected and analysed in the UKSTAR were EQ-5D-5L and complications evaluated at 8 weeks and 3, 6 and 9 months after the injury.

EuroQol-5 Dimensions, five-level version

The EQ-5D-5L was analysed as a continuous outcome using the utility score values from the five-level questions and based on the reported EQ-5D VAS. Summary results for the EQ-5D utility score and EQ-5D VAS are reported for each treatment group at each time point in Table 16, together with the unadjusted and adjusted mixed-effects model estimates. The analyses for both EQ-5D utility score and EQ-5D VAS are presented graphically in Figures 4 and 5. Both EQ-5D scores show a trend of improvement over time. The EQ-5D utility score analysis showed a statistically significant difference at the 8-week follow-up in favour of the functional brace group (0.069, 95% CI 0.03 to 0.1), but this difference was no longer present by the 9-month post injury follow-up.

FIGURE 4 Marginal mean EQ-5D utility values from the mixed-effects model and associated 95% CIs for the two treatment groups from baseline (post injury) to 9 months. (Figure not reproduced here; see the original article.)

FIGURE 5 Marginal mean EQ-5D VAS values from the mixed-effects model and associated 95% CIs for the two treatment groups from baseline (post injury) to 9 months. (Figure not reproduced here; see the original article.)

Sensitivity analyses

Sensitivity analyses were performed for the EQ-5D utility and VAS outcomes using CACE analysis and the AUC (Table 17). The CACE analysis was conducted using a similar approach as for the primary outcome ATRS and showed similar results to the EQ-5D ITT population analysis. The AUC summary statistics were estimated for a male participant of a mean age of 48.65 years, calculated from baseline post injury to 9 months post injury, with higher AUCs indicating better QoL.

Complications

The number of participants with one or more complications in each treatment group is presented overall from baseline to 9 months in Table 18 and at every time point for the ITT population in Table 19. Fisher’s exact and chi-squared tests showed no statistically significant results when testing for associations between the treatment groups and each type of complication across time.

Ancillary analyses

Following a presentation of the preliminary results, the TSC wanted to explore where the apparent differences at 8 weeks in EQ-5D utility score came from. The individual domains of the EQ-5D were explored using box plots in Figures 18–21 (see Appendix 3). The median score is marked with a triangle, whiskers are IQRs and the individual dots and crosses mark the outliers. The differences at 8 weeks appear to lie in the ability to self-care and usual activities only. Furthermore, in keeping with the health economics analysis plan, the distribution of the responses to the EQ-5D-5L questionnaires by treatment group and assessment point is presented in Appendix 4. That analysis showed no significant differences in the proportions of individuals reporting suboptimal health [i.e. any functional (below level 1) impairment] within dimensions between the two treatment groups at each time point.

Adverse events

Foreseeable AEs were reported as complications in Analyses to address secondary outcomes.

Two deaths were reported in this study, one of which was a SAE. This was due to a known lung cancer condition and was unrelated to the Achilles injury. The second death was due to a cardiac arrest following a bilateral PE and, as judged by the investigators, was potentially related but not unexpected. Both deaths were in the functional brace group.