In view of the findings in the previous chapter, and to try to maximise the potential conclusions using the data set, a further sensitivity analysis was planned and approved. This analysis included a new sample size power calculation.
Objective of Chapter 5
To study the effect of surgical treatment within 12 months of diagnosis of a first-time episode of TASD on recurrence rates in the 3 years that follow the diagnosis, among young adults in England.
Methods
Study design
A population-based, propensity-score-matched cohort study to control for confounding at baseline has been conducted. Young adults (aged 16–35 years) who presented with a first-time occurrence of TASD were selected from two NHS computerised databases: the CPRD and HES. Figure 12 shows a detailed illustration of the study plan that is described in more detail in the following sections.
FIGURE 12 The UK TASH-D phase 2 plan design for patients who received surgery or no surgical treatment within 12 months of diagnosis of a first-time TASD during 1 April 1997–31 March 2015, in England. (Figure not reproduced here; see the original article.)
This study used methodology identical to that described in the internal validation study in Chapter 2 and the 6-month analysis in Chapter 4; all participants had to have 2 years of clinical data within the CPRD before their first-time TASD to ensure that only ‘primary’ dislocations had been captured. The same rules were followed and the first Read code entry in the CPRD for shoulder dislocation was defined as the ‘first’ dislocation. All events were collected using the pre-agreed, validated list of CPRD Read codes (see Appendix 1). Patients experienced their first-time TASD between 1 April 1997 and 31 March 2015, allowing up to 3 years of follow-up for each patient to the end of the study on 26 April 2016. As in the previous analysis, a 6-week washout period for re-dislocation codes within the CPRD was applied to all patients following their TASD to avoid duplicate records.
The key changes from the previous analysis described in Chapter 4 were to include:
- patients having surgery up to 12 months from the date of their first-time TASD
- patients with any length of follow-up, rather than at least 2 full years
- follow-up up to 3 years, rather than 2 years.
Surgical group
The surgical group comprised patients in the CPRD with a first-time TASD who underwent shoulder stabilisation surgery after their first dislocation. This meant linking HES data to CPRD data in such a way that a HES surgical OPCS 4.7 code was seen to occur after a single first dislocation code in the CPRD before that surgical date. The timelines between first dislocation codes and OPCS 4.7 codes were recorded. If a re-dislocation occurred prior to their surgery date, the patient was allocated to the non-surgical arm.
A 6-week washout period for re-dislocation codes within the CPRD was applied to all patients in the surgical group following their surgery date to avoid duplicate records. Surgery patients were followed up for up to 3 years from the date of their surgery. Patients having surgery > 6 months following their TASD were censored on their date of surgery. The surgery dates for these patients ranged from 25 December 1997 to 30 September 2015.
Non-surgical group
As in the 6-month analysis, conservative care has been defined as ‘non-surgical intervention’ with no linked OPCS 4.7 surgical shoulder codes, producing a control cohort of patients whose first-time shoulder dislocation has been treated non-operatively. Non-surgical patients were followed up for up to 3 years following the date of their first-time TASD, as illustrated in Figure 12.
Outcome
The outcome was time to a shoulder re-dislocation as defined by the CPRD Read codes given in Appendix 1. For the surgical group, the shoulder re-dislocation can occur between 6 weeks and 2 years following the date of surgery. For the non-surgical group, the shoulder re-dislocation could occur between 6 weeks and 2 years from the date of the first-time TASD.
Any patients who died during the study were censored on their date of death.
Data sources
The analysis utilised two computerised NHS databases: the CPRD and HES, as described in Chapter 4. The linked databases were managed by a senior data manager using Python and SQL, and statistical analyses were conducted using Stata and R.
Sample size
The sample size calculation in the study plan, described fully in the previous chapter, required a total sample size of 3065 patients with 656 re-dislocations, with 90% power at the 5% significance level and allowing for a 10% loss to follow-up, to detect an absolute difference of 5% between the surgical and non-surgical groups. It was assumed that 1073 (35% of the total) patients would receive surgery. The 6-month analysis was underpowered on this basis.
A senior statistician reran the sample size calculation, reducing the power to 80%, having unequal groups (1 : 10 ratio), extending follow-up to 3 years from 2 years and looking at surgical intervention up to 12 months (rather than 6 months). To detect a HR of 0.73 (26% surgical re-dislocations vs. 19.5% non-surgical re-dislocations) would require a total of 3456 patients, of whom 314 were surgical patients and 3142 were non-surgical patients, and at least 695 re-dislocations.
Participants
Inclusion criteria
The CPRD records of 6046 patients aged 16–35 years with 2 years of data in the CPRD before a first-time TASD were identified in the internal validation study (see Table 1) and were linked to HES records.
Exclusion criteria
The following patients were excluded.
- those aged 16–35 years with a first-time TASD who could not be linked by CPRD-HES
- those with prior shoulder surgery for a shoulder dislocation
- those whose instability was treated with rotator cuff repair surgery or fracture surgery prior to or following a TASD.
Exclusions were made in accordance with the criteria above and these are described in Table 18. Following linkage to HES data, there was a linkage loss of 1234 patients and six duplicates were identified and removed. A very small proportion of patients (n = 193) was excluded for having shoulder surgery for prior dislocations, rotator cuff tears or fractures.
| Exclusions | Excluded (n) | Total (n) | % |
|---|---|---|---|
| Eligible patients in the CPRD cohort for the internal validation study following exclusions made in Table 1 | 6046 | 100 | |
| HES data linkage loss | 1234 | 20 | |
| Duplicate removal | 6 | < 1 | |
| Prior shoulder surgery for a shoulder dislocation | 155 | 3 | |
| Surgery for rotator cuff tear prior to TASD | 1 | < 1 | |
| Surgery for rotator cuff tear 6 months following TASD | 4 | < 1 | |
| Surgery for a shoulder fracture prior to TASD | 14 | < 1 | |
| Surgery for shoulder fracture in the 6 months following TASD | 19 | < 1 | |
| Total exclusions | 1433 | 24 | |
| Patients included in subsequent analyses | 4613 | 76 |
In total, 4613 patients remained available for analysis, which was greater than the minimum number of patients (n = 3456) required for a sufficient sample size at 80% power (Figure 13). Of these, 342 were surgical patients (slightly more than the minimum required, n = 314) and 4271 were non-surgical patients (much greater than the minimum required, n = 3142). Re-dislocations were observed in 912 patients (much greater than the minimum required, n = 695). Among the surgical group, 18% of patients suffered a re-dislocation, and 20% of patients suffered a re-dislocation in the non-surgical group. The difference in re-dislocation proportions between the surgical and non-surgical groups was only 2%, rather than the 5% difference used in the power calculation, which means that the power is reduced. However, overall, this study of surgical intervention up to 12 months on shoulder re-dislocations should have had sufficient power.
FIGURE 13 Shoulder dislocation exclusions flow chart for patients aged 16–35 years during 1 April 1997–31 March 2016 within CPRD in England, with linkage to HES records. (Figure not reproduced here; see the original article.)
Statistical analysis
The statistical analysis used for this sensitivity analysis was identical to that described in Chapter 4, with respect to covariates, missing data, confounding by indication and immortal time bias.
Results
Descriptive characteristics
A cohort of 4613 patients diagnosed with a first-time TASD during 1 April 1997–31 March 2015 was identified in the CPRD. The CPRD Read codes used to identify these patients are listed in Appendix 9. Of these patients, 342 underwent shoulder surgery within 6 months. The OPCS 4.7 surgical codes identified within HES are given in Appendix 10.
The descriptive characteristics of patients categorised by receiving surgery within 12 months or no surgery are described in Table 19. In the first 4 years, < 10 patients underwent surgery per year, but this increased from 2001 to 2010 to a maximum of 34 patients out of a total of 342 surgical patients. More men (82%) were diagnosed with a primary shoulder dislocation than women, and more men underwent surgery within 12 months (n = 302 men vs. n = 40 women). The majority of patients were aged 17–21 years at the time of their first-time TASD, but similar numbers were operated on among those aged 18–25 years.
| Characteristic | Patient group, n (%) | ||
|---|---|---|---|
| Whole data set | No surgery | Surgery within 12 months of TASD | |
| Total | 4613 (100) | 4271 (93) | 342 (7) |
| Year of shoulder dislocation | |||
| 1997 | 65 (1) | 64 (1) | 1 (< 1) |
| 1998 | 115 (2) | 109 (3) | 6 (2) |
| 1999 | 128 (3) | 125 (3) | 3 (1) |
| 2000 | 161 (3) | 155 (4) | 6 (2) |
| 2001 | 209 (5) | 199 (5) | 10 (3) |
| 2002 | 250 (5) | 239 (6) | 11 (3) |
| 2003 | 309 (7) | 288 (7) | 21 (6) |
| 2004 | 301 (7) | 290 (7) | 11 (3) |
| 2005 | 288 (6) | 271 (6) | 17 (5) |
| 2006 | 315 (7) | 292 (7) | 23 (7) |
| 2007 | 334 (7) | 308 (7) | 26 (8) |
| 2008 | 339 (7) | 306 (7) | 33 (10) |
| 2009 | 331 (7) | 303 (7) | 28 (8) |
| 2010 | 348 (8) | 314 (7) | 34 (10) |
| 2011 | 329 (7) | 301 (7) | 28 (8) |
| 2012 | 293 (6) | 261 (6) | 32 (9) |
| 2013 | 237 (5) | 209 (5) | 28 (8) |
| 2014 | 217 (5) | 195 (5) | 22 (6) |
| 2015 | 44 (1) | 42 (1) | 2 (1) |
| Sex | |||
| Male | 3794 (82) | 3492 (82) | 302 (88) |
| Female | 819 (18) | 779 (18) | 40 (12) |
| Age at shoulder dislocation (years) | |||
| 16 | 245 (5) | 225 (5) | 20 (6) |
| 17 | 309 (7) | 294 (7) | 15 (4) |
| 18 | 308 (7) | 281 (7) | 27 (8) |
| 19 | 324 (7) | 293 (7) | 31 (9) |
| 20 | 295 (6) | 271 (6) | 24 (7) |
| 21 | 315 (7) | 298 (7) | 17 (5) |
| 22 | 290 (6) | 260 (6) | 30 (9) |
| 23 | 258 (6) | 242 (6) | 16 (5) |
| 24 | 277 (6) | 262 (6) | 15 (4) |
| 25 | 246 (5) | 216 (5) | 30 (9) |
| 26 | 230 (5) | 212 (5) | 18 (5) |
| 27 | 207 (4) | 190 (4) | 17 (5) |
| 28 | 165 (4) | 156 (4) | 9 (3) |
| 29 | 177 (4) | 159 (4) | 18 (5) |
| 30 | 178 (4) | 169 (4) | 9 (3) |
| 31 | 148 (3) | 137 (3) | 11 (3) |
| 32 | 164 (4) | 152 (4) | 12 (4) |
| 33 | 149 (3) | 143 (3) | 6 (2) |
| 34 | 161 (3) | 152 (4) | 9 (3) |
| 35 | 167 (4) | 159 (4) | 8 (2) |
| BMI (kg/m2) | |||
| < 25 | 1542 (33) | 1439 (34) | 103 (30) |
| 25.0–29.9 | 763 (17) | 708 (17) | 55 (16) |
| ≥ 30 | 364 (8) | 338 (8) | 26 (8) |
| Missing | 1944 (42) | 1786 (42) | 158 (46) |
| IMD 2004 (quintile of deprivation) | |||
| 1 (affluent) | 1241 (27) | 1173 (27) | 68 (20) |
| 2 | 1028 (22) | 949 (22) | 79 (23) |
| 3 | 915 (20) | 846 (20) | 69 (20) |
| 4 | 834 (18) | 752 (18) | 82 (24) |
| 5 (deprived) | 592 (13) | 548 (13) | 44 (13) |
| Missing | 3 (< 1) | 3 (< 1) | 0 (0) |
| Smoking status | |||
| No | 2461 (53) | 2287 (54) | 174 (51) |
| Yes | 1335 (29) | 1233 (29) | 102 (30) |
| Ex-smoker | 392 (8) | 362 (8) | 30 (9) |
| Missing | 425 (9) | 389 (9) | 36 (11) |
| Drinking status | |||
| Yes | 2172 (47) | 2033 (48) | 139 (41) |
| No | 432 (9) | 403 (9) | 29 (8) |
| Missing | 2009 (44) | 1835 (43) | 174 (51) |
| CCI score | |||
| 0 | 4275 (93) | 3945 (92) | 330 (96) |
| 1 | 205 (4) | 199 (5) | 6 (2) |
| 2 | 84 (2) | 80 (2) | 4 (1) |
| ≥ 3 | 49 (1) | 47 (1) | 2 (1) |
| Region | |||
| East Midlands | 152 (3) | 142 (3) | 10 (3) |
| East of England | 549 (12) | 522 (12) | 27 (8) |
| London | 560 (12) | 520 (12) | 40 (12) |
| North East | 113 (2) | 104 (2) | 9 (3) |
| North West | 743 (16) | 669 (16) | 74 (22) |
| South Central | 669 (15) | 628 (15) | 41 (12) |
| South East Coast | 537 (12) | 502 (12) | 35 (10) |
| South West | 580 (13) | 524 (12) | 56 (16) |
| West Midlands | 525 (11) | 488 (11) | 37 (11) |
| Yorkshire and the Humber | 185 (4) | 172 (4) | 13 (4) |
| Epilepsy | 156 (3) | 134 (3) | 22 (6) |
| Painkiller prescriptions | |||
| Prescribed 3 months prior to TASD | 257 (6) | 235 (6) | 22 (6) |
| Prescribed 1 month following TASD | 542 (12) | 499 (12) | 43 (13) |
| Mortality | 22 (< 1) | 20 (< 1) | 2 (< 1) |
Only 8% of patients were overweight or obese, with missing data on BMI for 42% of all patients. There was a pattern of decreasing numbers of patients by deprivation quintile, but this pattern was not evident for surgical patients. Only three patients did not have a recorded IMD score. Most patients were non-smokers (53%) and there were missing data on smoking for 9% of all patients. Most patients drank alcohol (47%) and there were missing data on alcohol consumption for 44% of all patients. Most patients had no comorbid conditions (93%). Many patients were from the North West (n = 743) and South Central regions (n = 669), and fewer were from the North East region (n = 113). A small proportion of patients had been diagnosed with epilepsy (3%), but a higher proportion of all surgical patients had been diagnosed with epilepsy (6%). A small proportion (6%) of patients were prescribed painkillers in the 3 months prior to their first-time TASD, which increased to 12% in the 1 month following a TASD. Of the patients prescribed painkillers, both 3 months prior to and 1 month after their TASD, only 1% were prescribed a different painkiller. Only 22 patients died during the study period.
Multiply imputed data analyses
Multiple imputation by chained equations was conducted for patients with missing data on BMI, smoking, drinking and IMD. Appendix 11 presents the HRs, 95% CIs and p-values for each of these variables using all available data, a complete-case analysis and the multiply imputed data. The HRs are similar for each group, indicating that the multiple imputation process was successful.
Cox survival estimates for complete cases and the multiply imputed data with the factors that had an impact on re-dislocations are presented in Appendix 12. Only 2100 cases had complete data on all factors. For these patients, a protective but non-significant adjusted effect of surgery was observed (HR 0.63, 95% CI 0.40 to 1.00; p = 0.052). A younger age at first-time TASD and an epilepsy diagnosis were significant risk factors for a re-dislocation.
Using the multiply imputed data resulted in similar HRs, 95% CIs and the same risk factors. The protective effect of surgery was less marked but was significant (HR 0.73, 95% CI 0.55 to 0.95; p = 0.022).
Propensity score analysis
Logistic regression was used to calculate the propensity scores. The distributions of propensity scores in the patients receiving surgery within 12 months and in those patients who had non-operative treatment are presented in Figure 14. Prior to matching, the propensity scores tended to be higher in the surgery group, but there is a substantial amount of overlap, indicating that it is possible to proceed with propensity score matching to estimate the treatment effect. Following matching, the propensity score density plots are very similar between the surgery and non-surgery patient groups. The propensity score matching will not include the non-surgical patients in the left-hand peak of the distribution.
FIGURE 14 Propensity scores for first-time TASD patients diagnosed during 1 April 1997–31 March 2015 who had surgery within 12 months (black) or no surgery (green) within CPRD-HES, in England. (a) Prior to PS matching; and (b) following PS matching. PS, propensity score; SD, standard deviation. (Figure not reproduced here; see the original article.)
Table 20 presents the baseline characteristics of first-time TASD patients who had surgery within 12 months or non-operative treatment. The SMDs, prior to propensity score matching, show that the two groups of patients do differ for most characteristics (SMD > 0.1). During propensity score matching, 295 surgical patients were each matched to 10 non-surgical patients (n = 2950). In addition to this, the time between the date of the first-time TASD and the date of surgery was allocated to the non-surgical patients for 233 surgical patients, making a total of 3183 non-surgical patients. The remaining 62 surgical patients had no time to be re-allocated because the date of their first-time TASD was the same date as their surgery. Following propensity score matching, the SMDs were much smaller than before (all < 0.1), which suggested that the balancing was successful.
| Characteristic | All eligible primary shoulder dislocation patients (N = 4613)a | Matched analysis (10 : 1): patients matched on propensity score (N = 3478)a,b | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Surgery (n = 342) | Non-surgery (n = 4271) | SMD | Surgery (n = 295) | Non-surgery (n = 3183) | SMD | |||||
| n (%) | Re-dislocation (%) | n (%) | Re-dislocation (%) | n (%) | Re-dislocation (%) | n (%) | Re-dislocation (%) | |||
| Re-dislocation | 342 (100) | 61 (18) | 4271 (100) | 851 (20) | 295 (100) | 58 (20) | 3183 (100) | 590 (19) | ||
| Calendar year of shoulder dislocation | 342 (100) | 61 (18) | 4271 (100) | 851 (20) | 0.347 | 295 (100) | 58 (20) | 3183 (100) | 590 (19) | 0.011 |
| Sex | ||||||||||
| Male | 302 (88) | 52 (17) | 3492 (82) | 720 (21) | –0.184 | 258 (87) | 50 (19) | 2770 (87) | 528 (19) | –0.013 |
| Female | 40 (12) | 9 (23) | 779 (18) | 131 (17) | 37 (13) | 8 (22) | 413 (13) | 62 (15) | ||
| Age at shoulder dislocation (16- to 35-year-olds) | 342 (100) | 61 (18) | 4271 (100) | 851 (20) | –0.083 | 295 (100) | 58 (20) | 3183 (100) | 590 (19) | –0.019 |
| BMI (kg/m2) | ||||||||||
| < 25 | 184 (54) | 35 (19) | 2506 (59) | 523 (21) | 0.035 | 166 (56) | 34 (20) | 1835 (58) | 367 (20) | 0.013 |
| 25.0–29.9 | 106 (31) | 21 (20) | 1191 (28) | 231 (19) | 92 (31) | 19 (21) | 934 (29) | 163 (17) | ||
| ≥ 30 | 52 (15) | 5 (10) | 574 (13) | 97 (17) | 37 (13) | 5 (14) | 414 (13) | 60 (14) | ||
| IMD (quintile of deprivation) | ||||||||||
| 1 (affluent) | 68 (20) | 6 (9) | 1173 (27) | 225 (19) | 0.153 | 61 (21) | 6 (10) | 814 (26) | 144 (18) | 0.058 |
| 2 | 79 (23) | 14 (18) | 950 (22) | 201 (21) | 72 (24) | 14 (20) | 690 (22) | 146 (21) | ||
| 3 | 69 (20) | 17 (25) | 846 (20) | 158 (19) | 55 (19) | 15 (27) | 626 (20) | 104 (17) | ||
| 4 | 82 (24) | 13 (16) | 754 (18) | 150 (20) | 72 (24) | 13 (18) | 599 (19) | 102 (17) | ||
| 5 (deprived) | 44 (13) | 11 (25) | 548 (13) | 117 (21) | 35 (12) | 10 (29) | 454 (14) | 94 (26) | ||
| Smoking status | ||||||||||
| No | 197 (58) | 39 (20) | 2521 (59) | 528 (21) | –0.010 | 177 (60) | 37 (21) | 1888 (59) | 365 (19) | –0.002 |
| Yes | 112 (33) | 17 (15) | 1356 (32) | 252 (19) | 90 (31) | 16 (18) | 1011 (32) | 176 (17) | ||
| Ex-smoker | 33 (10) | 5 (15) | 394 (9) | 71 (18) | 28 (10) | 5 (18) | 284 (9) | 49 (17) | ||
| Drinking status | ||||||||||
| Yes | 280 (82) | 44 (16) | 3506 (82) | 677 (19) | 0.091 | 237 (80) | 42 (18) | 2541 (80) | 473 (19) | –0.013 |
| No | 62 (18) | 17 (27) | 765 (18) | 174 (23) | 58 (20) | 16 (28) | 642 (20) | 117 (18) | ||
| CCI score | ||||||||||
| 0 | 330 (96) | 60 (18) | 3945 (92) | 792 (20) | –0.146 | 283 (96) | 57 (20) | 3066 (96) | 567 (18) | 0.014 |
| 1 | 6 (2) | 0 (0) | 199 (5) | 37 (19) | 6 (2) | 0 (0) | 55 (2) | 15 (27) | ||
| 2 | 4 (1) | 0 (0) | 80 (2) | 15 (19) | 4 (1) | 0 (0) | 41 (1) | 5 (12) | ||
| ≥ 3 | 2 (1) | 1 (50) | 47 (1) | 7 (15) | 2 (< 1) | 1 (50) | 21 (< 1) | 3 (14) | ||
| Region | ||||||||||
| East Midlands | 10 (3) | 2 (20) | 142 (3) | 30 (21) | 0.075 | 10 (3) | 2 (20) | 103 (3) | 20 (19) | –0.012 |
| East of England | 27 (8) | 5 (19) | 522 (12) | 98 (19) | 26 (9) | 5 (19) | 268 (8) | 46 (17) | ||
| London | 40 (12) | 6 (15) | 520 (12) | 102 (20) | 37 (13) | 6 (16) | 396 (12) | 70 (18) | ||
| North East | 9 (3) | 3 (33) | 104 (2) | 26 (25) | 9 (3) | 3 (33) | 81 (3) | 18 (22) | ||
| North West | 74 (22) | 14 (19) | 669 (16) | 129 (19) | 53 (18) | 12 (23) | 592 (19) | 108 (18) | ||
| South Central | 41 (12) | 2 (5) | 628 (15) | 132 (21) | 36 (12) | 2 (6) | 433 (14) | 82 (19) | ||
| South East Coast | 35 (10) | 9 (26) | 502 (12) | 84 (17) | 34 (12) | 9 (26) | 359 (11) | 63 (18) | ||
| South West | 56 (16) | 10 (18) | 524 (12) | 110 (21) | 47 (16) | 9 (19) | 459 (14) | 88 (19) | ||
| West Midlands | 37 (11) | 7 (19) | 488 (11) | 110 (23) | 33 (11) | 7 (21) | 367 (12) | 76 (21) | ||
| Yorkshire and the Humber | 13 (4) | 3 (23) | 172 (4) | 30 (17) | 10 (3) | 3 (30) | 125 (4) | 19 (15) | ||
| Epilepsy | 22 (6) | 8 (36) | 134 (3) | 42 (31) | 0.155 | 10 (3) | 5 (50) | 128 (4) | 38 (30) | –0.033 |
| Painkiller prescriptions | ||||||||||
| 3 months prior to TASD | 22 (6) | 5 (23) | 235 (6) | 56 (24) | 0.039 | 15 (5) | 5 (33) | 181 (6) | 42 (23) | –0.027 |
| 1 month following TASD | 43 (13) | 10 (23) | 499 (12) | 89 (18) | 0.027 | 31 (11) | 9 (29) | 353 (11) | 57 (16) | –0.019 |
Figure 15 presents Kaplan–Meier estimates of probability of survival in the surgical and non-surgical groups. The hazards were not proportional as shown by the two lines crossing at approximately 750 days and by the Schoenfeld’s Residuals Test (p = 0.003). We estimated time-varying hazards, which showed that there was no effect in the first year of follow-up (p = 0.458). This means that, initially, surgical patients had a similar rate of re-dislocations to non-surgical patients. Between 1 and 3 years of follow-up, the non-surgical group was more likely to re-dislocate (p = 0.022). Overall, there appears to be a small survival advantage for surgical patients, but this is not statistically significant.
FIGURE 15 Kaplan–Meier survival estimates for first-time TASD patients diagnosed during 1 April 1997–31 March 2015 who had surgery within 12 months (green) or no surgery (black) within CPRD-HES, in England. (Figure not reproduced here; see the original article.)
Table 21 presents the final results of the effect of surgery within 12 months compared with no surgery among first-time TASD patients. The median follow-up was less in the non-surgical group. The rate of shoulder re-dislocations was similar between the surgical and non-surgical groups (0.26 per 1000 person-years, 95% CI 0.20 to 0.33 per 100,000 person-years, compared with 0.26 per 1000 person-years, 95% CI 0.24 to 0.28 per 100,000 person-years, respectively). Overall, there was no difference between the effect of surgery and non-surgery within 12 months on shoulder re-dislocations (HR 1.17, 95% CI 0.88 to 1.55; p = 0.274).
| N | Patient group | HR (IQR) | HRa (IQR) | |||||
|---|---|---|---|---|---|---|---|---|
| Surgery | Non-surgery | |||||||
| Events (n) | Median follow-up (days) (IQR) | Rate (per 1000 person-years) (IQR) | Events (n) | Median follow-up (days) (IQR) | Rate (per 1000 person-years) (IQR) | |||
| Shoulder re-dislocations | 58 | 1096 (558–1214) | 0.26 (0.20–0.33) | 590 | 1028 (430–1096) | 0.26 (0.24–0.28) | 1.18 (0.89–1.56) | 1.17 (0.88–1.55) |
To test for residual confounding, the patients were split into quintiles, based on the value of the propensity score, and HRs were produced for each quintile (see Appendix 13). The numbers of surgeries were equal in each quintile. HRs were found to differ between quintiles, suggesting that there was unmeasured confounding in the study consistent with the a priori risk factors not available in the CPRD or HES. Quintile 5 was found to have a twofold increase in the risk of dislocation after surgery (HR 2.07, 95% CI 1.23 to 3.46). Some key differences between the characteristics of patients within each stratum are shown in Appendix 14. Quintile 5, for instance, included more men, people who had a more recent first-time TASD, fewer alcohol drinkers and more people with epilepsy than the other quintiles.
Discussion
A population-based cohort of 4613 patients diagnosed with a first-time TASD during 1 April 1997–31 March 2015 with CPRD-HES linked records and up to 3 years of follow-up in England were identified. Only 342 patients in this data set received surgical treatment within 12 months of a first-time TASD. This again confirms that, in the NHS, even when extending surgery to 12 months, it is still an uncommon treatment after only one shoulder dislocation. However, extending to 12 months for this sensitivity analysis did provide some more power to the analysis, even though 47 surgical patients were not matched to controls in the propensity score analysis. On this occasion, non-proportionality of the HRs was observed, which suggests that further, more complex, statistical techniques could be considered; however, the issues of small patient numbers in the surgical group, similar proportions of re-dislocations and residual confounding mean that further analysis is unlikely to change any conclusions.
The main finding from this further propensity-score-matched analysis was that surgery within 12 months of a first-time TASD had no obvious beneficial effect compared with non-surgical interventions (HR 1.17, 95% CI 0.88 to 1.55; p = 0.274). The lack of association meant that adjusting for clustering at the general practice level and conducting a Rosenbaum bounds sensitivity analysis was inappropriate. However, when the propensity scores were split into quintiles, one quintile in particular (5) was found to be at a significantly increased risk of shoulder re-dislocation. The fact that one quintile had a different risk of the outcome is highly suggestive of the existence of residual confounding. This quintile included more men, people who had their first-time TASD more recently and fewer alcohol drinkers and had most of the people with epilepsy. So although the study now has more power, residual confounding continues. The confounders are likely to be contained within the a priori list of important risk factors, used by surgeons to make clinical decisions on the best care, that could not be identified from the data, and so these confounders could not be taken into account during the propensity scoring.