Predictor | Quality assessment | Relative effect (95% CI) | Anticipated absolute effect | ||||||
---|---|---|---|---|---|---|---|---|---|
Risk of bias | Inconsistency | Indirectness | Imprecision | Publication bias | Overall | Baseline risk* | Risk difference (95% CI) | ||
Age: every 10-yr decrement | |||||||||
11 030 patients (22 studies), median follow-up 12 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | No serious imprecision | Undetected; symmetric funnel plot; Begg test p = 0.8; Egger test p = 0.8 | High | OR 1.36 (1.24–1.48) | 30% for age 70 yr‡ | 7% more (5% to 9% more) patients with per 10-yr decrement of age having persistent pain |
Radiotherapy: yes v. no | |||||||||
9468 patients (16 studies), median follow-up 23.5 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | No serious imprecision | Undetected; symmetric funnel plot; Begg test p = 0.6; Egger test p = 0.2 | High | OR 1.35 (1.16–1.57) | 30% | 7% more (3% to 10% more) patients with radiotherapy having persistent pain |
Axillary lymph node dissection (ALND): yes v. no or ALND v. sentinel lymph node biopsy | |||||||||
7699 patients (13 studies), median follow-up 12 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | No serious imprecision | Undetected; symmetric funnel plot; Begg test p > 0.9; Egger test p = 0.5 | High | OR 2.41 (1.73–3.35) | 30% | 21% more (13% to 29% more) patients with ALND having persistent pain |
Acute postoperative pain, measured with 10-cm pain scale: better indicated by lower values | |||||||||
1387 patients (5 studies), median follow-up 17.5 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | No serious imprecision | Uncertain: only 5 studies | High | OR 1.16 (1.03–1.30) | 30% for 1 cm on a 10-cm scale‡ | 3% more (1% to 6% more) patients with per 1-cm increment of acute pain on 10-cm pain scale having persistent pain |
Preoperative pain: yes v. no | |||||||||
2504 patients (8 studies) median follow-up 7.5 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | Serious imprecision§ | Uncertain: only 8 studies | Moderate | OR 1.29 (1.01–1.64) | 30% | 6% more (0% to 11% more) patients with preoperative pain having persistent pain |
BMI: every 5-point increment | |||||||||
3178 patients (8 studies) median follow-up 12 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | No serious imprecision¶ | Uncertain: only 8 studies | High | OR 1.11 (0.99–1.24) | 30% for BMI 25 kg/m2‡ | 2% more (0% to 5% more) patients with per 5-point increment of BMI having persistent pain |
Breast surgery: BCS v. mastectomy/modified radical mastectomy | |||||||||
8566 patients (17 studies), median follow-up 17.5 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | No serious imprecision¶ | Undetected; symmetric funnel plot; Begg test p = 0.2; Egger test p = 0.8 | High | OR 1.08 (0.90–1.30) | 30% | 2% more (2% less to 6% more) patients with BCS having persistent pain |
Chemotherapy: yes v. no | |||||||||
8481 patients (17 studies), median follow-up 12 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | No serious imprecision¶ | Undetected; symmetric funnel plot; Begg test p = 0.6; Egger test p > 0.9 | High | OR 1.12 (0.98–1.29) | 30% | 2% more (0% less to 6% more) patients with chemotherapy having persistent pain |
Endocrine therapy: yes v. no | |||||||||
8312 (11 studies), median follow-up 27 mo | No serious risk of bias† | No serious inconsistency | No serious indirectness | No serious imprecision¶ | Undetected; symmetric funnel plot; Begg test p = 0.3; Egger test p = 0.2 | High | OR 1.07 (0.94–1.22) | 30% | 1% more (1% less to 4% more) patients with endocrine therapy having persistent pain |
Note: BCS = breast-conserving surgery; BMI = body mass index; CI = confidence interval; GRADE = Grading of Recommendations Assessment, Development and Evaluation; OR = odds ratio.
↵* Baseline risk based on the subpopulation of patients undergoing sentinel lymph node biopsy with lowest absolute risk of persistent pain in the study with the largest sample size among the studies at low risk of bias. (26)
↵† Quality was not rated down on the basis of risk of bias, because the subgroup analyses and meta-regression did not show any significant difference between each risk-of-bias component and the estimates of association.
↵‡ The reference groups for age, acute postoperative pain and BMI were obtained from the largest study among those with low risk of bias that explored each of these predictors (i.e., age 70 as reference, (26) BMI of 25 (41) and acute postoperative pain of 1 cm on a 10-cm visual analogue pain scale (37)).
↵§ Quality was rated down on the basis of imprecision because the 95% CI associated with the risk difference included the predefined threshold of 10% for modifiable factors, which means that clinical actions based on the estimate of the lower or upper boundary may be different.
↵¶ Quality was not rated down on the basis of imprecision, even though the 95% CI for the pooled effect overlapped a risk difference of 0 (no effect), because clinical actions based on the estimate of the lower or upper boundary would not change, according to the predefined threshold of ≥ 20% for nonmodifiable factors.