Surgery for degenerative rotator cuff tears: a health technology assessment
Health technology assessment
|Published
We conducted a health technology assessment of rotator cuff repair compared with non-surgical treatment for degenerative full-thickness rotator cuff tears.
Key message
We conducted a health technology assessment of rotator cuff repair compared with non-surgical treatment for degenerative full-thickness rotator cuff tears. We included five randomized and 18 non-randomized studies. We compiled the studies in meta-analyses and assessed the certainty in the effect estimates using the GRADE approach.
- For patients with tears involving only one tendon, repair makes little or no clinically relevant difference in pain, night pain, function, patient satisfaction and health-related quality of life compared to non-surgical treatment at one year (GRADE: moderate to low).
- For patients with tears involving one or two tendons, repair gives slightly higher patient satisfaction, but little or no clinically relevant difference in pain, night pain, function, and health-related quality of life compared to non-surgical treatment at one year (GRADE: moderate to low).
- Re-tears occurred in between 5% and 35% six months to ten years after repair, and an increased tear size in 59% after ten years in those treated non-surgically (GRADE: not assessed).
- Few serious adverse events were reported, but data were limited and we are uncertain about the absolute risks.
- A cost-effectiveness analysis, in terms of a main analysis with a time horizon of five years was performed. The results show that surgery is more expensive than non-surgical treatment with a cost-difference of 36 516 NOK and a difference in effect of 0,09 QALYs, resulting in an ICER of 405 733 NOK per QALY.
The potential cost savings by choosing non-surgical treatment over surgery is estimated to be 81 Million NOK, but the results are uncertain since surgery would be the appropriate treatment option for some patients.
Summary
Introduction
The rotator cuff connects scapula (the shoulder blade) to the upper arm. Together with the surrounding passive structures, the function of the rotator cuff muscles is to stabilize the shoulder joint and contribute to elevation and rotation of the arm. Painful rotator cuff tears can be treated non-surgically or surgically. Figures from the Norwegian Patient Registry indicate that there is variation in the use of surgery across the regional health authorities in Norway. The clinical community has expressed a need to evaluate the use of surgery for rotator cuff tears, particularly for degenerative lesions.
The purpose of this health technology assessment was to summarize existing evidence about the effect and safety of rotator cuff repair for degenerative full-thickness rotator cuff tears compared to non-surgical treatment, and to conduct a health economic evaluation. Both technologies are currently a part of clinical practice in Norway, and this reassessment is intended to be used as decision basis for the Decision Forum in The National System for Managed Introduction of New Health Technologies within the Specialist Health Service in Norway.
Method
We conducted systematic literature searches in relevant databases. For effect, we searched for systematic reviews, which we used to identify primary studies, and supplemented with searches for randomized studies (RCTs). We included studies that compared surgery with non-surgical treatment in patients with tear in one or two tendons. For safety (complications / adverse events), we also searched for non-randomized trials and included studies that compared different surgical procedures. Studies of patients with massive tears were excluded. Two researchers read titles, abstracts, and relevant articles in full text. One researcher extracted and compiled data, and another checked the data. Two researchers assessed the risk of bias and the confidence in the effect estimates using the GRADE approach
Cost-effectiveness analyses of surgery compared to non-surgical treatment were performed in terms of one main analysis and two scenario analyses, taking a healthcare perspective. The main analysis had a five-year time perspective, and the two scenarios had a time perspective of one - and five years, respectively. The results were presented as cost per quality-adjusted life-years (QALYs). We also carried out a simplified budget impact analysis to highlight potential cost savings at a national level.
Results
We included five RCTs and 18 non-randomized trials. The RCTs included between 56 and 190 participants with an average age of 56 to 65 years and were conducted in Norway, Sweden, Finland, and the Netherlands. All compared surgery including postoperative rehabilitation with exercises given by and/or guided by a physiotherapist, and in one study injections and drugs were also given. The non-randomized studies included between 20 and 442 participants with an average age of 55 to 70 years. Ten were from Asia and five were from Europe. Most used arthroscopic technique.
For patients with tears involving one tendon (supraspinatus), we found that repair made little or no clinically relevant difference in pain, night pain, function, patient satisfaction, and health-related quality of life compared to non-surgical treatment at one year follow-up (table). For patients with tears involving one or two tendons, we found that repair gave slightly higher patient satisfaction compared to non-surgical treatment at one year follow-up, while there was little or no clinically relevant difference in pain, night pain, function, and health-related quality of life (table).
The included studies provided scarce information on adverse events including the development of possible sequelae such as cuff-tear arthropathy. Some outcomes were not mentioned or there were very few events. Among those who underwent surgery, re-tear / non-healing occurred in between 5% and 35% after six months to ten years, and there was no evidence of an association between occurrence of re-tear and duration of follow-up. Among those who were treated non-surgically, 29%, 37% and 59% and an increase in rupture size that exceeded 5 mm after one, five and ten years, respectively (data from only one study per timepoint). After ten years, 27 % had undergone rotator cuff repair due to lack of progress (data from only one study).
The result from the main health economic analysis, estimated a cost difference of 36 516 NOK between the two alternatives and a difference in effect of 0.09 QALYs, in favour of surgery. The expected incremental cost-effectiveness ratio (ICER) was 405 733 NOK per QALY. Choosing exclusively non-surgical treatment over surgery, can provide annual cost savings of up to 81 million NOK after five years.
Table. Surgery compared to non-surgical treatment for rotator cuff tear |
|
||||||
|
Outcomes |
Anticipated absolute effects* (95% CI) |
Relative effect |
№ of participants |
Certainty of the evidence |
|
||
|
Risk with non-operative treatment |
Risk with surgery |
|
|||||
|
Main analysis (rupture in one tendon) |
|
|
|||||
|
Pain, general (VAS; 0-10, 10 is worst) |
Mean 1,3** |
MD 0.48 lower |
- |
168 |
⨁⨁⨁◯ |
|
|
|
Pain, night (NRS; 0-10, 10 is worst) |
Mean 1,3** |
MD 0.7 lower |
- |
58 |
⨁⨁◯◯ |
|
|
|
Function (Constant-Murley score; 0-100, 100 is best) |
Mean 74,1** |
MD 3 higher |
- |
168 |
⨁⨁⨁◯ |
||
|
Patient satisfaction (“satisfied or dissatisfied with the treatment outcome”) |
873 per 1 000 |
943 per 1 000 |
RR 1.08 |
110 |
⨁⨁◯◯ |
||
|
Health-related quality of life (EQ-VAS; 0-100, 100 is best) at 1 year |
Mean 82** |
MD 2 higher (4.5 lower to 8.5 higher) |
- |
58 |
⨁⨁◯◯ |
||
|
Supplementary analyses (rupture in one or two tendons) |
|
||||||
|
Pain, general (VAS; 0-10, 10 is worst) |
Mean 1,3** |
MD 0.89 lower (1.74 lower to 0,03 lower) |
- |
315 (4 RCTs) |
⨁⨁◯◯ |
|
|
|
Pain, night (NRS; 0-10, 10 is worst) |
Mean 1,3** |
MD 0.7 lower |
- |
58 |
⨁⨁◯◯ |
|
|
|
Function (Constant-Murley score; 0-100, 100 is best) |
Mean 74.1** |
MD 5,14 higher (1.82 higher to 8.45 higher) |
- |
315 (4 RCTs) |
⨁⨁⨁◯ |
|
|
|
Patient satisfaction (VAS; 0-10, 10 is best) |
Mean 7.2** |
MD 1.75 higher (0.75 higher to 2.74 higher) |
- |
102 (1 RCT) |
⨁⨁◯◯ |
|
|
|
Health-related quality of life (SF-36 mental component; 0-100, 100 is best) |
Mean 50.3** |
MD 1.3 lower |
- |
103 (1 RCT) |
⨁⨁◯◯ |
|
|
|
Health-related quality of life (SF-36 physical component; 0-100, 100 is best) |
Mean 57.5** |
MD 0.9 higher (2.72 lower to 4.52 higher) |
|
103 (1 RCT) |
⨁⨁◯◯ |
|
|
|
*The risk in the intervention group (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI); **Mean value in the (largest) non-surgical group at 1 year CI: confidence interval; EQ-VAS; EuroQoL Visual Analog Scale; MD: mean difference; NRS: Numeric Rating Scale; RR: relative risk; VAS: Visual Analog Scale |
|
||||||
|
Explanations a. Downgraded one level due to risk of bias, due to lack of blinding b. Downgraded one level due to few participants / data from only one study and low precision |
|
||||||
Discussion
Low-to-moderate-certainty evidence suggests that surgical repair does not result in clinically relevant differences in effect compared to non-surgical treatment. Confidence in the effect estimates was downgraded primarily due to lack of blinding and low precision. The direction of bias in unblinded trials is likely to favour surgery.
Few serious adverse events were reported in the included studies. However, we only identified small studies, resulting in limited data precluding estimates of absolute risks. Thus, we do not know the risk of serious adverse events. Other parts of the scope that we were unable to answer, were the effect on sleep, sick leave, participation in leisure-time activities, postoperative stiffness, and sequelae such as cuff-tear arthropathy
Health-related quality of life and shoulder function scores that were used in the health economic analysis were obtained from two different studies, where we assume there is a correlation with the assumptions made in our model. Further, there is a great variation in clinical practice, and performing outpatient rather than inpatient surgery may result in major cost savings. Many of the assumptions made in the model are based on expert opinions and are subject to great uncertainty, and we may have both underestimated and overestimated costs.
Conclusion
Our review suggests that there are no clinically relevant differences in effect between surgical and non-surgical treatment. Since the effect of surgery seems to be modest in unblinded trials, it is likely that future trials only will contribute to more precise effect estimates (and thus higher certainty), rather than changing the conclusions. Few serious adverse events were reported, but we are uncertain about the absolute risks.
Surgery is the most expensive alternative, and it is the procedure that affect the costs the most. The budget impact analysis showed that choosing exclusively non-surgical treatment over surgery can provide up to NOK 81 million in annual cost savings at a national level. The estimate is uncertain and is expected to be somewhat smaller as there always will be some patients who will need surgery, and surgery cannot be completely excluded as a treatment option.
