Plus Sutures for preventing surgical site infection: a systematic review of clinical outcomes with economic and environmental models

The searches for the NICE submission were conducted between 01 February 2021 and 08 February 2021 in multiple databases and information resources (see Table 2). These included: databases covering biomedical healthcare and nursing journal literature; databases of controlled trials, systematic reviews and health technology assessments; databases containing conference abstracts; and databases containing information on ongoing trials. The full search strategy for MEDLINE (Supplementary Fig. 1) was translated appropriately for the other databases (Supplementary Appendix A).

Searches of economic databases were included because search results were also considered for use in the cost-effectiveness model of Plus Sutures.

Reference lists of any relevant systematic reviews published in the last five years were checked for any eligible studies that may have been missed by the database searches, and the manufacturer provided details of any ongoing or unpublished trials with which they were associated.

Results were downloaded into Endnote bibliographic software [29], deduplicated using several algorithms, and the duplicate references held in a separate EndNote database. A single researcher assessed the search results according to their relevance in providing information on the clinical efficacy and safety of the intervention and comparator and removed the obviously irrelevant records.

Two reviewers independently assessed the titles and abstracts of remaining records for relevance against the eligibility criteria, with disagreements adjudicated by a third reviewer. Assessment of full texts was then conducted by the same three reviewers; two reviewers independently assessed the records, and the third reviewer adjudicated any disagreements.

The first reviewer extracted data from the eligible studies and the second reviewer quality checked all the data points against the papers to ensure accuracy. Data were extracted as reported by study authors, with calculations performed only where the required data were not presented in the format required for the meta-analyses. Calculations were minimal and were based only on reported data.

As recommended by Cochrane guidance [30], timepoints at which data were to be extracted were specified prior to starting the review. One timepoint per study was extracted (the optimum timepoint for each surgery type was either 30 or 90 days post-surgery); if a paper reported data at more than one timepoint, CDC guidance [3] was used to select the most appropriate timepoint.

The NICE Risk of Bias tool [31] was used to quality assess each of the included studies, with one researcher completing the assessment and a second verifying this. This tool considers seven different criteria.

A high-level assessment of the similarity of studies and availability of data was performed, guided by the Australian Pharmaceutical Benefits Advisory Committee criteria [32]. Where meta-analysis was possible, i.e. sufficient homogenous studies reported suitable data, statistical methods were used to analyse and summarise the results of the included studies.

For outcomes for which meta-analysis was not possible, a narrative summary exploring the quality of the studies, the relationship between studies and any patterns discerned in the data were presented.

The base case analysis included all studies that provided sufficient data. Subgroup analysis considered four subgroups as defined in the NICE scope (adults only, children only, patients classified as having clean wounds, and patients classified as having non-clean wounds). Studies in which the authors did not explicitly state wound type were mapped to the subgroups by surgery type. Finally, a sensitivity analysis was carried out removing Stratafix Plus as an intervention to explore its potential effect on heterogeneity.

For the subgroup analyses by wound type, we recorded authors’ descriptions of the status of the wounds assessed in each study. Where the authors did not explicitly report this information, the independent opinion of three clinicians was sought as to the likely wound status following the surgery detailed in each of the studies. The categorisation of the wound status was then compared across the clinicians and any divergence of opinion discussed before a consensus was reached. The decisions reached determined which subgroup analysis each study would contribute to.

A decision-tree model was developed to estimate changes in healthcare costs (from the perspective of the NHS) associated with the use of Plus Sutures compared with conventional sutures in people undergoing a surgical procedure. The population considered in the model was all people undergoing a surgical procedure with subgroup analysis conducted for adults (≥ 18 years of age), children, clean and non-clean wounds. Following their surgical procedure people enter the model and receive either Plus Sutures or conventional sutures and then follow the pathway of SSI or no SSI, followed by survival or death. The model structure is presented in Fig. 1.

Patients who developed an incisional SSI accrued additional mortality risk and costs which incorporated hospital readmission, increased length of stay plus other resource use required during SSI treatment. Given that no system changes were anticipated to implement the technology and no additional training was required for health care professionals to use Plus Sutures, no additional resource costs were included in the model.

Adverse events related to sutures were not included in the model because no events were identified from the clinical review that were judged to have a substantial impact on health-related quality of life (HRQoL) or health care resource use. Those reported were considered unlikely to be related to the Plus technology.

The model structure was aligned with the clinical pathway and other SSI models in the published literature as well as those developed or assessed by NICE, where the risk of SSI in both arms was captured and the cost of this applied [36‐45]. The input parameters were not aligned with these studies; rather the best available evidence relevant to the NHS at the time of constructing the model (March 2021) was used instead, to ensure the model was current and relevant to the UK.

The model inputs are presented in Supplementary Table 1. In addition to the searches run to inform the systematic review, reports of adverse events associated with the technology were sought via searches of the Manufacturer and User Facility Device Experience (MAUDE) database and Medicines and Healthcare products Regulatory Agency (MHRA) resources. Results of these searches did not inform the systematic review but were used to check for any adverse events not reported in the published literature that might need to be considered in the model.

The model generated the total per patient costs in each arm over a one-year time horizon. The incremental cost per patient is presented as well as a cost per SSI averted and the cost per death averted.

One-way deterministic sensitivity analysis was conducted to explore the impact of varying individual parameters and identify key drivers of the analysis. Threshold/breakeven analysis was conducted around the baseline SSI incidence with comparator sutures, the cost of SSI, the RR reduction of SSI incidence with Plus Sutures, and the number of sutures. A tornado diagram (see Fig. 4) was used to present one-way analysis for all model inputs. Ranges reported were taken from the literature, and where unavailable, clinical opinion or assumptions were used.

Probabilistic sensitivity analysis (PSA) was conducted to explore second order uncertainty in the results using 1,000 iterations to ensure model stability.

The ranges considered in the sensitivity analyses are presented in Supplementary Table 1.

Subgroup analysis was conducted as described throughout for adults, children, clean and non-clean wounds.

All outcomes, except incidence of SSI, were unsuitable for assessment using meta-analysis. The main reason was a lack of reporting of the outcomes; when reported, the information was often incomplete. A summary of the reported outcomes, including safety outcomes, is presented in Supplementary Sect. 5.1.

The included studies reported minimal adverse events related to triclosan-coated sutures. Since adverse events are likely to emerge within the follow-up time of the RCTs, the evidence seems robust enough to exclude the possibility of significant adverse events related to the triclosan-coated sutures.

Thimour-Bergström 2013 [83] contributed two datasets to the meta-analysis; further details of these data can be found in the Supplementary Sect. 2.

The fixed effects model for SSI including all groups of patients estimated a RR of 0.71 (95% confidence interval (CI): 0.64 to 0.79) indicating those in the triclosan-coated sutures (including Stratafix Plus) group had a 29% reduction in the risk of developing an SSI compared to those in the control group (p < 0.001). The random effects model estimated a RR of 0.70 (95% CI: 0.58 to 0.84; p < 0.001). These results are presented in Fig. 3 and are based on 6,852 and 6,969 total patients, and 503 and 708 events in the triclosan-coated sutures (including Stratafix Plus) and control arm, respectively. No outliers or substantial publication bias (see Supplementary Fig. 5) were noted during the analysis.

Triclosan-coated sutures were found to be associated with a significant reduction in the risk of developing an SSI compared with those in the control group in all analyses conducted (25% to 48% reduction using fixed effects models across main analysis and subgroups). Full details of subgroups are presented in Supplementary Fig. 6a to d. Consistent findings were reported when Stratafix Plus was removed from the analyses (28% reduction; Supplementary Fig. 7).

The clinical evidence is drawn from a good number of RCTs (31) including a high number of SSIs (1,211) across a wide range of surgery types. In 27 of the included studies, incidence of SSI was the primary outcome, with most studies using a commonly accepted definition of SSI and a clearly defined patient population.

As Plus Sutures were the only triclosan-coated sutures on the market at the time of conducting the review [J&J to provide citation as detailed above], these analyses reflect the available evidence base and may not be applicable to any future antibiotic suture products that may come to market.

The 31 studies represented eight middle income and eleven high income countries. This range of countries provides good generalisability to the middle / high income global context, although low income countries are not represented in this review. The FALCON trial [96] (published after searches for this review were conducted) has recently suggested that in 54 hospitals located in low and middle income countries (Benin, Ghana, India, Mexico, Nigeria, Rwanda and South Africa) the use of triclosan-coated sutures did not reduce the risk of SSI. Some of the pragmatic choices made in the conduct of the trial limit the generalizability of the results. Among them, (i) allowing for different techniques of skin closure, (ii) use of triclosan-coated sutures only for fascia closure and not for skin closure, (iii) compliance with the individual component of the WHO checklist [10] was not mandated and (iv) no standard postoperative management of the wounds was prescribed. SSI are multifactorial and the absence of clear standardisation in pre-, intra-, and post-operative management of the wounds in the trial raises the risk that the effect of single components (like the use of Plus Sutures) is obscured by other non-controlled factors. This risk may be heightened in emergency procedures which comprised over 60% of the cases in FALCON. The results of FALCON may therefore not be applicable outside the geographical context, the surgical technique used, and the population included in the trial.

A recent systematic review and analysis of 29 trials in clean and clean-contaminated procedures, including FALCON (Otto-Lambertz 2023 [93]), found a significant 24% reduction in postoperative wound infection rate with the use of triclosan-coated sutures.

An exploratory analysis (reported in Supplementary Fig. 8) was conducted to assess the impact of the FALCON trial on the meta-analysis described in this paper, and found similar results to those of Otto-Lambertz. No publication bias was identified, and the fixed effects model indicated that patients in the triclosan-coated sutures (including Stratafix Plus) group had a 18% reduction in the risk of developing an SSI compared with those in the control group. This reduces the effect of the main analysis, which showed a 29% reduction of risk, but still suggests benefits associated with the triclosan-coated sutures compared to standard sutures.

Studies included in this review covered a wide range of surgical interventions, both emergency and planned, as laid out in Table 4. Most of the studies included patients with comorbidities including diabetes, chronic obstructive pulmonary disease (COPD), malignant diseases, chronic renal insufficiency, anaemia, and people living with obesity or malnourishment. This large heterogeneity in patient population in conjunction with the positive result and low heterogeneity of the meta-analysis suggests that the intervention can be recommended to a wide population of patients.

Not all studies were blinded in the same way; 15 of the studies were double-blind and the remaining 16 were either single-blind, open label, or not clearly reported. In addition, the studies were conducted in a wide range of countries over a fifteen-year date span (from 2005 to 2020). During this time, the clinical pathways are likely to have changed. Despite these factors, heterogeneity across studies was not substantial, and the use of random effects models accounted for any variability.

As recommended by Cochrane to increase precision of the results [97], the current review incorporated evidence from all available trials without removing the high-risk of bias studies. We acknowledge that only limited number of RCTs (three of the total 31 included in the review) were deemed to have a low risk of bias.