Summary of findings for the main comparison. EGFR tyrosine kinase inhibitor (TKI) compared to observation alone for maintenance treatment of epithelial ovarian cancer after first‐line chemotherapy.

EGFR tyrosine kinase inhibitor (TKI) compared to observation alone for maintenance treatment of epithelial ovarian cancer after first‐line chemotherapy
Patient or population: maintenance treatment of epithelial ovarian cancer after first‐line chemotherapy Setting: hospital outpatient treatment of women with ovarian/fallopian tube/primary peritoneal cancer after response to first‐line chemotherapy Intervention: EGFR tyrosine kinase inhibitor (TKI) Comparison: observation alone
Outcomes	Anticipated absolute effects* (95% CI)		Relative effect (95% CI)	No. of participants (studies)	Certainty of the evidence (GRADE)	Comments
	Risk with observation alone	Risk with EGFR tyrosine kinase inhibitor (TKI)
Overall survival	Median 50.8 months in the EGFR TKI group vs 59.1 months in the observation arm See comment		HR 0.99 (0.81 to 1.20)	835 (1 RCT)	⊕⊕⊝⊝ LOWa	Outcome unlikely to be affected by blinding. due to the way HRs are calculated, the assumed and corresponding risks were not estimated.
Progression‐free survival	Median PFS of 12.7 months in the EGFR TKI group vs 12.4 months in the observation arm See comment		HR 1.05 (0.90 to 1.23)	835 (1 RCT)	⊕⊝⊝⊝ VERY LOWb	due to the way HRs are calculated, the assumed and corresponding risks were not estimated.
Quality of life assessed with EORTC QLQ‐C30 and OV28 (Ovarian Cancer Module) questionnaires	"Global health/QOL scores showed a significant overall difference between the two treatment arms during the first year (P 0.0102) in favour of the observation arm. In addition, the QLQ‐C30 found statistically significant differences at the 5% level in symptom levels for diarrhoea, loss of appetite, nausea/vomiting, and fatigue, with worse symptom scores for the erlotinib arm. None of the scales, however,reported differences of 10 points except for the diarrhoea [sic] scale in which differences of more than 20 points were observed at most assessments during the first year. Sensitivity analyses by means of imputation revealed similar results".		‐	835 (1 RCT)	⊕⊝⊝⊝ VERY LOWc	Analysable data were not provided in the published paper or after communication with the study author, and so we were unable to analyse or exclude selective reporting bias.
*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). CI: confidence interval; EGFR: epidermal growth factor receptor; HR: hazard ratio; RR: risk ratio; OR: odds ratio; PFS: progression‐free survival; QOL: quality of life; QLQ‐C30: Quality of Life Questionnaire C30; RCT: randomised controlled trial; TKI: tyrosine kinase inhibitor.
GRADE Working Group grades of evidence High‐certainty: we are very confident that the true effect lies close to that of the estimate of the effect Moderate‐certainty: we are moderately confident in the effect estimate: the true effect is likely to be close to the estimate of the effect, but there is a possibility that it is substantially different Low‐certainty: our confidence in the effect estimate is limited: the true effect may be substantially different from the estimate of the effect Very low‐certainty: we have very little confidence in the effect estimate: the true effect is likely to be substantially different from the estimate of effect

^aDowngraded by two levels for imprecision (confidence intervals that cross zero and single study) and inability to assess inconsistency, as results were based on a single study.
 ^bDowngraded by three levels due to risk of bias (unblinded study); imprecision (confidence intervals that cross zero and single study); and inability to assess inconsistency, as results were based on a single study.
 ^cDowngraded by three levels due to the possibility of selective reporting bias (incompletely reported predefined outcome, so possibility of selective outcome reporting); imprecision; and risk of bias (unblinded), as no data were available to analyse effect and outcome was highly likely to be affected by lack of blinding.