Table 3

GRADE Summary of findings for the comparison of subacromial decompression surgery with placebo surgery

Outcome, time frameMeasurement instruments and relative effectsAbsolute effect estimatesCertainty in effect estimates (quality of evidence)Plain text summary
Placebo surgerySurgery
Pain, 6 monthsMeasured by VAS and NRS scaled to 0–10.
Scale: 0–10, lower better.
MID: 1.5 units.
Based on data from 299 patients in 2 studies.
3.9 4.0 High.Surgery has little or no effect on pain at 6 months.
Difference: MD 0.07 higher
(95% CI 0.51 lower to 0.64 higher).
Pain, 1 yearMeasured by VAS and NRS scaled to 0–10.
Scale: 0–10, lower better.
MID: 1.5 units.
Based on data from 284 patients in 2 studies.
2.9 2.6 High.Surgery has little or no effect on pain at 1 year.
Difference: MD 0.26 lower
(95% CI 0.84 lower to 0.33 higher).
Function, 6 monthsMeasured by Constant Score.
Scale: 0–100, higher better.
MID: 8.3 points.
Based on data from 286 patients in 2 studies.
61 57 High.Surgery has little or no effect on function at 6 months.
Difference: MD 3.72 lower
(95% CI 8.72 lower to 1.28 higher).
Function, 1–2 yearsMeasured by Constant Score.
Scale: 0–100, higher better.
MID: 8.3 points.
Based on data from 274 patients in 2 studies.
69 72 High.Surgery has little or no effect on function at 1–2 years.
Difference: MD 2.76 higher
(95% CI 1.36 lower to 6.87 higher).
Global perceived effect, 6 monthsRelative risk: 1.04 (95% CI 0.81 to 1.34).
Based on data from 293 patients in 2 studies.
447 per 1000 465 per 1000 Moderate, due to serious imprecision*.Surgery probably has little or no global perceived effect at 6 months.
Difference: 18 more per 1000
(95% CI 85 fewer to 152 more).
Global perceived effect, 1 yearRelative risk: 1.1 (95% CI 0.94 to 1.3).
Based on data from 290 patients in 2 studies.
635 per 1000 699 per 1000 Moderate, due to serious imprecision*.Surgery probably has little or no global perceived effect at 1 year.
Difference: 64 more per 1000
(95% CI 38 fewer to 190 more).
Quality of life, 6 monthsMeasured by EQ-5D.
Scale: −0.59–1, higher better.
MID: 0.07 units.
Based on data from 292 patients in 2 studies.
0.67 0.66 High.Surgery has little or no effect on quality of life at 6 months.
Difference: MD 0.01 lower
(95% CI 0.08 lower to 0.05 higher).
Quality of life, 1 yearMeasured by EQ-5D.
Scale: −0.59–1, higher better.
MID: 0.07 units.
Based on data from 285 patients in 2 studies.23
0.73 0.70 High.Surgery has little or no effect on quality of life at 1 year.
Difference: MD 0.03 lower
(95% CI 0.11 lower to 0.06 higher).
Serious harms†Based on data from 25 240 patients in 2 observational studies.
Follow-up 30 days.
Incidence: 5 or 6 (4 or 5–7)/1000 Moderate, due to serious indirectness‡.Surgery probably causes rare serious harms†.
Frozen shoulder, RCTsRelative risk: 1.69 (95% CI 0.41 to 6.91).
Based on data from 331 patients in 2 studies.
18 per 1000
Difference, 12 more per 1000 (95% CI 11 fewer to 106 more).
30 per 1000 Low, due to very serious imprecision§.Surgery may increase the incidence of frozen shoulder.
  • *Imprecision: serious, wide CIs.

  • †Serious harms were death, bleeding (uncontrolled or requiring transfusion), cardiac arrest requiring cardiopulmonary resuscitation, myocardial infarction, cerebrovascular accident, acute renal failure, unplanned intubation, ventilator >48 hours, deep infection (surgical site or organ/space), sepsis, septic shock, wound dehiscence, pulmonary embolism, deep vein thrombosis and peripheral nerve injury.

  • ‡Indirectness: serious. Differences between the intervention of interest and those studied; observational studies included other arthroscopic shoulder surgeries in addition to subacromial decompressions only.

  • §Imprecision: very serious. Low number of patients/events, wide CIs.

  • MID, minimally important difference; NRS, numeric rating scale; RCT, randomised controlled trial; VAS, visual analogue scale.