Cost-effectiveness analysis of an active 30-day surgical site infection surveillance at a tertiary hospital in Ghana: evidence from HAI-Ghana study

Objective To assess the cost-effectiveness of an active 30-day surgical site infection (SSI) surveillance mechanism at a referral teaching hospital in Ghana using data from healthcare-associated infection Ghana (HAI-Ghana) study. Design Before and during intervention study using economic evaluation model to assess the cost-effectiveness of an active 30-day SSI surveillance at a teaching hospital. The intervention involves daily inspection of surgical wound area for 30-day postsurgery with quarterly feedback provided to surgeons. Discharged patients were followed up by phone call on postoperative days 3, 15 and 30 using a recommended surgical wound healing postdischarge questionnaire. Setting Korle-Bu Teaching Hospital (KBTH), Ghana. Participants All prospective patients who underwent surgical procedures at the general surgical unit of the KBTH. Main outcome measures The primary outcome measures were the avoidable SSI morbidity risk and the associated costs from patient and provider perspectives. We also reported three indicators of SSI severity, that is, length of hospital stay (LOS), number of outpatient visits and laboratory tests. The analysis was performed in STATA V.14 and Microsoft Excel. Results Before-intervention SSI risk was 13.9% (62/446) as opposed to during-intervention 8.4% (49/582), equivalent to a risk difference of 5.5% (95% CI 5.3 to 5.9). SSI mortality risk decreased by 33.3% during the intervention while SSI-attributable LOS decreased by 32.6%. Furthermore, the mean SSI-attributable patient direct and indirect medical cost declined by 12.1% during intervention while the hospital costs reduced by 19.1%. The intervention led to an estimated incremental cost-effectiveness ratio of US$4196 savings per SSI episode avoided. At a national scale, this could be equivalent to a US$60 162 248 cost advantage annually. Conclusion The intervention is a simple, cost-effective, sustainable and adaptable strategy that may interest policymakers and health institutions interested in reducing SSI.

Few comments below: please spell out the abbreviations in abstract (i.e.HAI) please describe the intervention in abstract please clarify which cost is measured, direct medical cost?
Introduction: this session is informative and well organized. please consider add more background information of the SSI in Ghana.
Methods: could you please justify, why a before and during intervention strategy was used? why not measure results after intervention?
Could you please explain a bit more of the patient cost? is it outof-pocket or total medical cost? what is the role of NHIS?
Results: could you please present P value for Table 3 and 5? please clarify, only direct medical cost was measured in this study. but across the text, "societal" was used multiple times. Did you measure indirect cost as well?

REVIEWER
McFarland, Agi Glasgow Caledonian University, Nursing and Community Health REVIEW RETURNED 01-Nov-2021

GENERAL COMMENTS
Many thanks for the opportunity to review your study. I have a few questions and some additions which would improve the current manuscript, below: • No information is provided on ethical approval; please add this in • What diagnostic criteria did you use for SSI? And which type of SSI were included from which surgeries?
• Why was the follow up limited to 30 days given that SSI may be diagnosed beyond that given current diagnostic criteria from CDC?
• What software did you use for the modelling? Please specify • You mention discounting at 2.5% on page 5 but then refer to a 2.5% inflation using the same reference on page 6; it is not clear which you did and needs attention • Model parameter values need further detail in relation to units of measurement • There were significant differences in wound class breakdown and surgery types between your comparator groups which may have under estimated the overall impact of your intervention. This is worth highlighting in the discussion • Please also consider the limitations of using mean LOS comparisons; a useful article in this regard: https://pubmed.ncbi.nlm.nih.gov/29902486/ • Your point about the LMIC setting is an important one given the lack of such studies identified in this recent review: https://pubmed.ncbi.nlm.nih.gov/32417433/

Reviewer 1
Comment 1: This article assessed the cost-effectiveness of 30-days SSI surveillance in a teaching hospital in Ghana. This article brought important cost and effectiveness information, and hopeful inform the policy making process. However, the causal-relationship between intervention and observed effect is limited by the study design. Response: Thank you for this comment. We acknowledged the limitations of the design under subsection "Strength and limitations of the study". (see page 2 main document). Comment 2: Please spell out the abbreviations in abstract (i.e. HAI).
Response: All the abbreviations in the abstract are spelled out in the first instance and used subsequently in order not to exceed the word limit of the abstract.  Tables 3 and 5? Response: Resolved. Comment 9: Please clarify, only direct medical cost was measured in this study. but across the text, "societal" was used multiple times. Did you measure indirect cost as well? Response: We measured both direct and indirect medical costs as patient costs and also report patient productivity loss due to absenteeism from work. (Cost breakdown is detailed in Table 5).
Reviewer 2 Comment 1: No information is provided on ethical approval; please add this in Response: Ethics approval information is provided under subsection "Ethics Approval Statement." Comment 2: a) What diagnostic criteria did you use for SSI? b) And which type of SSI were included from which surgeries? Response: a) SSI diagnosis was based on CDC criteria. b) Types of SSI included were those related to superficial, deep, and organ space wounds resulting from eleven surgical procedures (Table 2). Comment 3: Why was the follow-up limited to 30 days given that SSI may be diagnosed beyond that given current diagnostic criteria from CDC?. Response: The 30-day surveillance follows standard methods described by the CDC. We excluded SSI from implant surgery, which may require more than 30-days follow-up. Comment 4: What software did you use for the modeling? Please specify. Response: We used both STATA version 14.0 and Microsoft Excel to perform the whole analysis. Comment 5: You mention discounting at 2.5% on page 5 but then refer to a 2.5% inflation using the same reference on page 6; it is not clear which you did and needs attention. Response: Resolved. This was an oversight. We used a 2.5% discount rate to compare costs before and during the intervention.

GENERAL COMMENTS
Please specify the year of CDC diagnostic criteria and provide a reference. Study perspective: please be specific in who the "provider" is