Impact of surgical-site infection on health utility values: a meta-analysis

Abstract

Background: Surgical-site infections (SSIs) are recognized as negatively affecting patient quality of life. No meta-analysis of SSI utility values is available in the literature to inform estimates of this burden and investment decisions in prevention.

Methods: A systematic search of PubMed, MEDLINE, CINAHL, and the National Health Service Economic Evaluation Database was performed in April 2022 in accordance with PROSPERO registration CRD 42021262633. Studies were included where quality-of-life data were gathered from adults undergoing surgery, and such data were presented for those with and without an SSI at similar time points. Two researchers undertook data extraction and quality appraisal independently, with a third as arbiter. Utility values were converted to EuroQol 5D (EQ-5D™) estimates. Meta-analyses were conducted using a random-effects model across all relevant studies, with subgroup analyses on type and timing of the SSI.

Results: In total, 15 studies with 2817 patients met the inclusion criteria. Six studies across seven time points were used in the meta-analysis. The pooled mean difference in EQ-5D™ utility in all studies combined was -0.08 (95 per cent c.i. -0.11 to -0.05; prediction interval -0.16 to -0.01; I2 = 40 per cent). The mean difference in EQ-5D™ utility associated with deep SSI was -0.10 (95 per cent c.i. -0.14 to -0.06; I2 = 0 per cent) and the mean difference in EQ-5D™ utility persisted over time.

Conclusion: The present study provides the first synthesized estimate of SSI burden over the short and long term. EQ-5D™ utility estimates for a range of SSIs are essential for infection prevention planning and future economic modelling.

Fig. 1

PRISMA flow diagram showing selection of articles for reviewSSI, surgical-site infection; QoL, quality of life.

Fig. 2

Main meta-analysis outcome measure showing mean difference in EQ-5D™ utility estimates across six included studies Mean differences are shown with 95% confidence intervals. A random-effects model was used for meta-analysis. ^a-dsignify different time points within the same study or for Hyldig the two dressing comparator groups (i.e both arms of the RCT).

Fig. 3

Forest plot for EQ-5D™ mean difference in deep surgical-site infection only Mean differences are shown with 95% confidence intervals. *Values are mean(s.d.). EQ-5D™ utility estimates. A random-effects REML model was used for meta-analysis. ^a-dsignify different time points within the same study or for Hyldig the two dressing comparator groups (i.e both arms of the RCT).

Fig. 4

Forest plot for EQ-5D™ mean difference across all surgical-site infection types at different time points Mean differences are shown with 95% confidence intervals. A random-effects REML model was used for meta-analysis. ^a-dsignify different time points within the same study or for Hyldig the two dressing comparator groups (i.e both arms of the RCT).

Fig. 5

Funnel plot of all included studies