(1) A cost-effectiveness analysis of mupirocin and chlorhexidine gluconate for Staphylococcus aureus decolonization prior to hip and knee arthroplasty in Alberta, Canada compared to standard of care

Home

Questions

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

17

18

19

20

21

22

23

24

25

26

Papers

PMCID: 6625116 (link)

Year: 2019

Reviewer Paper ID: 1

Project Paper ID: 7

Q1 - Title

Question description: Does the title clearly identify the study as an economic evaluation and specify the interventions being compared?

Explanation: The title of the manuscript explicitly indicates it is a cost-effectiveness analysis, an economic evaluation, and specifies the interventions being compared, namely mupirocin and chlorhexidine gluconate versus standard of care for Staphylococcus aureus decolonization prior to surgeries.

Quotes:

  • A cost-effectiveness analysis of mupirocin and chlorhexidine gluconate for Staphylococcus aureus decolonization prior to hip and knee arthroplasty in Alberta, Canada compared to standard of care

Q2 - Abstract

Question description: Does the abstract provide a structured summary that includes the context, key methods, results, and alternative analyses?

Explanation: The abstract lacks a structured summary that clearly separates the sections into context, methods, results, and alternative analyses. Although it provides some details on each of these elements, the format is more narrative rather than structured.

Quotes:

  • Background: While decolonization of Staphylococcus aureus reduces surgical site infection (SSI) rates following hip and knee arthroplasty, its cost-effectiveness is uncertain.
  • Methods: Decision analytic models and a probabilistic sensitivity analysis were used for a cost-effectiveness analysis.
  • Results: Use of the decolonization bundle was cost saving compared to usual care ($153/person), and resulted in 16 complex Staphylococcus aureus SSIs annually as opposed to 32.
  • Conclusions: Decolonization for Staphylococcus aureus prior to hip and knee replacements resulted in cost savings and fewer SSIs, and should be considered prior to these procedures.

Q3 - Background and objectives

Question description: Does the introduction provide the context for the study, the study question, and its practical relevance for decision-making in policy or practice?

Explanation: The introduction provides context by describing the frequency of hip and knee replacements in Canada, the risk of surgical site infections, and the suggested decolonization protocol. It outlines the study question by evaluating the cost-effectiveness of a decolonization protocol. The practical relevance is discussed in terms of reducing infections and healthcare costs, which are significant for decision-making in policy and practice.

Quotes:

  • There are over 100,000 hip and knee replacements performed yearly in Canada with approximately 10,000 in Alberta, Canada...1-2% of patients develop some level of periprosthetic infection after surgery.
  • Most commonly Staphylococcus aureus (S.aureus), including the virulent methicillin-resistant S.aureus (MRSA), cause post-replacement infections...there are no uniform guidelines around decolonization pre-operatively.
  • We sought to evaluate if an evidence based S.aureus decolonization protocol...was cost-effective.
  • A prior study...suggested that treating everyone is more practical and should be the strategy of choice.
  • The average cost to the Canadian healthcare system per surgical revision of an infected arthroplasty is over $17,000.

Q4 - Health economic analysis plan

Question description: Was a health economic analysis plan developed, and if so, where is it available?

Explanation: The manuscript does not indicate the development of a specific health economic analysis plan. It describes methods related to economic evaluation but does not mention a separate or detailed health economic analysis plan.

Quotes:

  • Methods section: 'Economic evaluation A cost effectiveness analysis (CEA) was conducted to assess the efficiency of implementing a decolonization protocol prior to hip and knee arthroplasty in Alberta.'
  • Methods section: 'The model used was a simple decision tree (Fig. 1) and was constructed using decision analysis software.'
  • Methods section: 'The model was evaluated by clinical experts in order to establish face validity.'

Q5 - Study population

Question description: Are the characteristics of the study population (e.g., age range, demographics, socioeconomic, or clinical characteristics) described?

Explanation: The manuscript describes the target study population as all adult patients in Alberta undergoing elective knee or hip arthroplasty, specifying an age range and the size of the cohort. However, it does not delve into further demographic or socioeconomic characteristics, focusing primarily on the inclusion criteria and general age of patients.

Quotes:

  • The target population for this analysis was all adult patients (> 18 years of age) in Alberta who receive elective knee or hip arthroplasty.
  • Data for this cohort was based on 24,667 patients who underwent primary elective hip or knee arthroplasty in Alberta between April 1, 2012 and March 31, 2015 (mean age 66.5 years).

Q6 - Setting and location

Question description: Is relevant contextual information (such as setting and location) provided that may influence the findings of the study?

Explanation: The manuscript provides relevant contextual information about the setting and location of the study, which is Alberta, Canada. This contextual detail is crucial as it impacts both the economic evaluation and the implementation logistics in the healthcare system discussed in the study. Setting details such as centralized hip and knee clinics in Alberta and their operational framework are provided, which may influence the study outcomes.

Quotes:

  • "A cost-effectiveness analysis of mupirocin and chlorhexidine gluconate for Staphylococcus aureus decolonization prior to hip and knee arthroplasty in Alberta, Canada compared to standard of care."
  • "We modelled the cost-effectiveness of decolonization in a hypothetical cohort of adult patients undergoing hip and knee replacement in Alberta, Canada."
  • "In Alberta, nearly all patients who receive an elective hip and knee arthroplasty are triaged and managed through centralized hip and knee clinics and are assigned a nurse case manager."

Q7 - Comparators

Question description: Are the interventions or strategies being compared described, along with the rationale for their selection?

Explanation: The manuscript provides a description and rationale for comparing the decolonization strategies using mupirocin and chlorhexidine gluconate against standard care, which is no decolonization. It highlights the risk of Staphylococcus aureus infections and the cost implications of treating them as the main reason for selecting these interventions.

Quotes:

  • "We sought to evaluate if an evidence based S.aureus decolonization protocol, including intranasal mupirocin and CHG, implemented in all adult patients prior to knee and hip replacement in Alberta, compared with standard care (no routine decolonization) was cost-effective."
  • "In some decolonization protocols chlorhexidine gluconate (CHG) is used as a once daily full body wash for up to 5 days at the same time as using intranasal mupirocin antibiotic ointment to the nares twice a day."
  • "Most commonly Staphylococcus aureus (S.aureus), including the virulent methicillin-resistant S.aureus (MRSA), cause post-replacement infections. Though colonization with S.aureus is a risk factor for subsequent S.aureus SSI, there are no uniform guidelines around decolonization pre-operatively."

Q8 - Perspective

Question description: What perspective(s) were adopted by the study, and why were they chosen?

Explanation: The study adopted the perspective of the publicly funded health care system. This perspective was chosen to include the costs incurred by the inpatient hospitalizations and outpatient ambulatory care visits associated with hip and knee arthroplasty in Alberta, reflecting the impact on the healthcare system rather than individual patients.

Quotes:

  • The analysis was done from the perspective of the publicly funded health care system, and as such, included the costs of the inpatient hospitalizations and outpatient ambulatory care visits.

Q9 - Time horizon

Question description: What is the time horizon for the study, and why is it appropriate?

Explanation: The article specifies that the baseline time horizon for the study was set at 1 year. This time frame is appropriate because most SSIs following knee and hip arthroplasty occur within the first 3 months post-operatively, with a significant number happening within the first 30 days, allowing for the capture of costs and effectiveness related to the development of SSIs shortly after surgery.

Quotes:

  • In the baseline analysis, the time horizon used was 1 year, as most SSIs following knee and hip arthroplasty occur within the first 3 months post-operatively with a large proportion occurring within the first 30 days.

Q10 - Discount rate

Question description: What discount rate(s) were used, and what was the rationale for choosing them?

Explanation: The manuscript explicitly states that no discount rate was applied as the time horizon of the study was 1 year, and it does not provide any rationale for choosing this decision beyond the timeframe consideration.

Quotes:

  • No discount rate was applied as the time horizon was 1 year.

Q11 - Selection of outcomes

Question description: What outcomes were used as measures of benefit and harm?

Explanation: The manuscript does not utilize outcomes such as life-years gained, QALYs, or ICERs as measures of benefit and harm. Instead, it focuses on infections prevented and healthcare costs as the primary outcomes.

Quotes:

  • The primary outcomes of the models were infections prevented and health care costs.
  • Utilities, life years and quality adjusted life years were not considered in this base model given the short time horizon.
  • Costs (and infections avoided) were the output considered in the model.

Q12 - Measurement of outcomes

Question description: How were the outcomes used to capture benefits and harms measured?

Explanation: The manuscript does not explicitly describe how outcomes like life years gained, quality-adjusted life-years (QALYs), or incremental cost-effectiveness ratios (ICERs) were measured. It focuses on cost savings and infections prevented as primary outcomes.

Quotes:

  • The primary outcomes of the models were infections prevented and health care costs.
  • Costs and infections avoided were the output considered in the model. Utilities, life years and quality adjusted life years were not considered in this base model given the short time horizon.

Q13 - Valuation of outcomes

Question description: What population and methods were used to measure and value the outcomes?

Explanation: The article identifies the study population as all adult patients undergoing elective hip and knee arthroplasty in Alberta, Canada. The outcomes were measured and valued using a cost-effectiveness analysis relying on decision analytic models, specifically a simple decision tree and a Markov model, which were informed by a large-scale pre- and post-intervention trial and analyzed from the perspective of the publicly funded health system.

Quotes:

  • The target population for this analysis was all adult patients (> 18 years of age) in Alberta who receive elective knee or hip arthroplasty.
  • We modelled the cost-effectiveness of decolonization in a hypothetical cohort of adult patients undergoing hip and knee replacement in Alberta, Canada.
  • A cost effectiveness analysis (CEA) was conducted to assess the efficiency of implementing a decolonization protocol prior to hip and knee arthroplasty in Alberta.
  • The effectiveness of a S.aureus decolonization protocol at reducing S.aureus complex SSI was derived from a pre- and post-intervention trial.

Q14 - Measurement and valuation of resources and costs

Question description: How were the costs valued in the study?

Explanation: The costs in the study were valued using specific cost data sources, including data for mupirocin from the Alberta Blue Cross Formulary, costs for chlorhexidine from a medical supply website, nurse educator costs from AHS job postings, and costs for hospitalizations and management of SSIs from a population-based cohort study.

Quotes:

  • The cost for the mupirocin ointment was taken from the Alberta Blue Cross Formulary. The cost per gram was 45 cents and a 5-g tube would be issued.
  • The cost of 4% CHG sponges was taken from a medical supply website and five would be needed for each patient.
  • It was assumed that in order to implement the decolonization protocol at all hip and knee clinics across Alberta, a nurse educator in Infection Prevention and Control would need to be hired... This cost was estimated based on AHS job postings for nurse educators.
  • Costs for hospitalizations and management of SSIs as well as the cost of the initial arthroplasty were obtained from a population based cohort study including all patients in Alberta undergoing hip and knee arthroplasty.

Q15 - Currency, price, date, and conversion

Question description: What are the dates of the estimated resource quantities and unit costs, and what currency and year were used for conversion?

Explanation: The manuscript provides specific details about the years for the estimated resource quantities and costs, indicating they span from April 1, 2012, to March 31, 2015. It also specifies that costs were inflated to 2016 Canadian dollars (CDN$) for the analysis.

Quotes:

  • Data for this cohort was based on 24,667 patients who underwent primary elective hip or knee arthroplasty in Alberta between April 1, 2012 and March 31, 2015.
  • No discount rate was applied as the time horizon was 1 year. All costs were inflated to 2016 CDN$.

Q16 - Rationale and description of model

Question description: If a model was used, was it described in detail, including the rationale for its use? Is the model publicly available, and where can it be accessed?

Explanation: The manuscript does describe the decision analytic model used, including the rationale for its use. However, it states that the data used in the study are not publicly available, implying that the model itself might not be accessible either.

Quotes:

  • The model used was a simple decision tree and was constructed using decision analysis software (TreeAge Pro 2018 Williamstown, MA).
  • Availability of data and materials: Data sharing is not applicable to this article as no datasets were generated during the current study.

Q17 - Analytics and assumptions

Question description: What methods were used for analyzing or statistically transforming data, extrapolation, and validating any models used?

Explanation: The manuscript describes using decision tree analysis, probabilistic, and one-way sensitivity analyses for statistical transformation of data. The model was validated by clinical experts for face validity. These methodologies were used to analyze and validate the cost-effectiveness models for S. aureus decolonization.

Quotes:

  • The model used was a simple decision tree (Fig. 1) and was constructed using decision analysis software (TreeAge Pro 2018 Williamstown, MA).
  • A wide range of sensitivity analyses were undertaken to ensure the model responded appropriately.
  • The model was evaluated by clinical experts in order to establish face validity.
  • A probabilistic sensitivity analysis where we allowed for all variables to change simultaneously (though 1000 Monte Carlo simulations) was completed using distributional assumptions of the input parameters (Table 1).

Q18 - Characterizing heterogeneity

Question description: What methods were used to estimate how the results vary for different sub-groups?

Explanation: The manuscript indicates that no subgroup analyses were conducted. It explicitly states that all adults undergoing joint replacement were considered as the target population, without further breakdown into subgroups for separate analysis.

Quotes:

  • No subgroups were analyzed in this study as all adults undergoing joint replacement may derive benefit from decolonization.

Q19 - Characterizing distributional effects

Question description: How were the impacts distributed across different individuals, and were adjustments made to reflect priority populations?

Explanation: The manuscript does not provide information on how impacts were distributed across different individuals or if adjustments were made to reflect priority populations. The analysis targeted all adult individuals undergoing hip or knee arthroplasty without subgroup distinctions or priority adjustments.

Quotes:

  • The target population for this analysis was all adult patients (> 18 years of age) in Alberta who receive elective knee or hip arthroplasty. Data for this cohort was based on 24,667 patients who underwent primary elective hip or knee arthroplasty in Alberta between April 1, 2012 and March 31, 2015 (mean age 66.5 years). No subgroups were analyzed in this study as all adults undergoing joint replacement may derive benefit from decolonization.

Q20 - Characterizing uncertainty

Question description: What methods were used to characterize sources of uncertainty in the analysis?

Explanation: The analysis characterized sources of uncertainty using both one-way and probabilistic sensitivity analyses to assess how variations in input parameters affect outcomes.

Quotes:

  • Decision analytic models and a probabilistic sensitivity analysis were used for a cost-effectiveness analysis...
  • A one-way sensitivity analysis was completed to estimate the influence that the range of input values had on the overall costs associated with using a bundle. Each input variable was varied one at a time using the 95% confidence intervals.
  • A probabilistic sensitivity analysis (PSA) where we allowed for all variables to change simultaneously (though 1000 Monte Carlo simulations) was completed using distributional assumptions of the input parameters.

Q21 - Approach to engagement with patients and others affected by the study

Question description: Were patients, service recipients, the general public, communities, or stakeholders engaged in the design of the study? If so, how?

Explanation: The manuscript does not mention any engagement of patients, service recipients, the general public, communities, or stakeholders in the design of the study. The research focuses on a cost-effectiveness analysis using decision-analytic models, with details concerning statistical methods, economic evaluation, and effectiveness derived from previous trials.

Quotes:

  • A cost effectiveness analysis (CEA) was conducted to assess the efficiency of implementing a decolonization protocol prior to hip and knee arthroplasty in Alberta.
  • Data for this cohort was based on 24,667 patients who underwent primary elective hip or knee arthroplasty in Alberta between April 1, 2012 and March 31, 2015.

Q22 - Study parameters

Question description: Were all analytic inputs or study parameters (e.g., values, ranges, references) reported, including uncertainty or distributional assumptions?

Explanation: The manuscript provides a detailed table with model inputs, ranges, and includes the source of each parameter. It also describes the distributional assumptions used in the probabilistic sensitivity analysis to capture uncertainty.

Quotes:

  • Model inputs by treatment alternatives
  • Distribution for PSA
  • Ranges for Inputs (95% CI unless otherwise stated)
  • A wide range of sensitivity analyses were undertaken to ensure the model responded appropriately.
  • A probabilistic sensitivity analysis (PSA) where we allowed for all variables to change simultaneously (though 1000 Monte Carlo simulations) was completed using distributional assumptions of the input parameters (Table 1).

Q23 - Summary of main results

Question description: Were the mean values for the main categories of costs and outcomes reported, and were they summarized in the most appropriate overall measure?

Explanation: The manuscript reports the mean costs for both the decolonization protocol and standard care for the target population. These are appropriately summarized as cost savings per person, along with the corresponding effectiveness measure, which is the number of surgical site infections (SSIs) avoided. This comprehensive measure allows for a clear understanding of the outcome of both costs and effectiveness (infections prevented).

Quotes:

  • In our base case analysis, the average cost for those who received the decolonization protocol was $20,525 and for those who received standard of care the cost was $20,678 (a cost savings of approximately $153 per person, which in Alberta translates into savings of $1.26 million annually).
  • This change in effectiveness translates into 16 complex SSIs from S.aureus annually in Alberta when a decolonization bundle is used versus 32 complex SSIs with standard of care.

Q24 - Effect of uncertainty

Question description: How did uncertainty about analytic judgments, inputs, or projections affect the findings? Was the effect of the choice of discount rate and time horizon reported, if applicable?

Explanation: The manuscript reports that no discount rate was applied because the time horizon was one year, which aligns with the period when most SSIs occur. This indicates that uncertainty related to the choice of discount rate was not a factor considered in the analysis. The study's findings are presented without the effect of various discounting scenarios, focusing just on the specified one-year timeframe.

Quotes:

  • No discount rate was applied as the time horizon was 1 year. All costs were inflated to 2016 CDN$. Costs (and infections avoided) were the output considered in the model.
  • In the baseline analysis, the time horizon used was 1 year, as most SSIs following knee and hip arthroplasty occur within the first 3 months post-operatively with a large proportion occurring within the first 30 days.

Q25 - Effect of engagement with patients and others affected by the study

Question description: Did patient, service recipient, general public, community, or stakeholder involvement make a difference to the approach or findings of the study?

Explanation: There is no mention of involvement from patients, service recipients, general public, community, or other stakeholders in influencing the approach or findings of the study. The study primarily focused on cost-effectiveness analysis using pre-existing data and models.

Quotes:

  • The analysis was done from the perspective of the publicly funded health care system.
  • The effectiveness of a S.aureus decolonization protocol at reducing S.aureus complex SSI was derived from a pre- and post-intervention trial.
  • A cost effectiveness analysis (CEA) was conducted to assess the efficiency of implementing a decolonization protocol prior to hip and knee arthroplasty in Alberta.

Q26 - Study findings, limitations, generalizability, and current knowledge

Question description: Were the key findings, limitations, ethical or equity considerations, and their potential impact on patients, policy, or practice reported?

Explanation: The manuscript focuses primarily on the methodology and findings of the cost-effectiveness analysis and does not discuss ethical or equity considerations, nor does it delve into the potential impacts on policy or practice. It mentions cost savings and reduced infections, yet lacks information on the wider implications for patients, policy, or ethical considerations.

Quotes:

  • "Conclusions: We determined that a decolonization protocol prior to hip and knee arthroplasty could result in fewer complex S. aureus SSIs and cost savings at both 1 year and over a lifetime."
  • "Ethics approval for this research was obtained from the University of Calgary Health Research Ethics Board."]}

SECTION: TITLE
A cost-effectiveness analysis of mupirocin and chlorhexidine gluconate for Staphylococcus aureus decolonization prior to hip and knee arthroplasty in Alberta, Canada compared to standard of care


SECTION: ABSTRACT
Background

While decolonization of Staphylococcus aureus reduces surgical site infection (SSI) rates following hip and knee arthroplasty, its cost-effectiveness is uncertain.
We sought to examine the cost-effectiveness of a decolonization protocol for Staphylococcus aureus prior to hip and knee replacement in Alberta compared to standard care - no decolonization.

Methods

Decision analytic models and a probabilistic sensitivity analysis were used for a cost-effectiveness analysis
, with the effectiveness of decolonization based on a large published pre- and post- intervention trial. The primary outcomes of the models were infections prevented and health care costs.. We modelled the cost-effectiveness of decolonization in a hypothetical cohort of adult patients undergoing hip and knee replacement in Alberta, Canada. Information on the incidence of complex surgical site infections (SSIs), as well as the cost of care for patients with and without SSIs was taken from a provincial infection control database, and health administrative data.

Results

Use of the decolonization bundle was cost saving compared to usual care ($153/person), and resulted in 16 complex Staphylococcus aureus SSIs annually as opposed to 32
(with approximately 8000 hip or knee arthroplasties performed). The probabilistic sensitivity analysis demonstrated that the majority (84%) of the time the decolonization bundle was cost saving. The model was robust to one-way sensitivity analyses conducted within plausible ranges. There were small upfront costs associated with using a decolonization protocol, however, this model demonstrated cost savings over one year. In a Markov model that considered the impact of a decolonization bundle over a lifetime as it pertained to the need for subsequent joint replacements and patient quality of life, the bundle still resulted in cost savings ($161/person).

Conclusions

Decolonization for Staphylococcus aureus prior to hip and knee replacements resulted in cost savings and fewer SSIs, and should be considered prior to these procedures.


Electronic supplementary material

The online version of this article (10.1186/s13756-019-0568-5) contains supplementary material, which is available to authorized users.

SECTION: INTRO
Introduction

There are over 100,000 hip and knee replacements performed yearly in Canada with approximately 10,000 in Alberta, Canada. While these procedures frequently improve mobility, pain and quality
of life, 1-2% of patients develop some level of periprosthetic infection after surgery.

This is a severe complication, often requiring multiple hospital admissions, surgeries, and prolonged courses of antibiotics. The Centers for Disease Control and Prevention (CDC) separates surgical site infections (SSI) into superficial incisional, deep incisional and organ space infections. Deep incisional/organ space infections are considered complex and roughly 1% of Albertans undergoing arthroplasty will develop one of these primary infections post-operatively.

The Infectious Diseases Society of America (IDSA) guidelines suggest varying management strategies to treat SSI including repeat surgeries and long courses of antibiotics. The exact strategy and type of surgical intervention utilized depend on a number of factors including whether the SSI was complex.

The average cost to the Canadian healthcare system per surgical revision of an infected arthroplasty is over $17,000
, only considering acute care costs. Even after appropriate surgical and medical management of a primary infection, a proportion of patients (8-60%) will have a relapsed infection and require repeat surgery and potentially chronic antibiotic management.

Most commonly Staphylococcus aureus (S.aureus), including the virulent methicillin-resistant S.aureus (MRSA), cause post-replacement infections. Though colonization with S.aureus is a risk factor for subsequent S.aureus SSI, there are no uniform guidelines around decolonization pre-operatively.


In some decolonization protocols chlorhexidine gluconate (CHG) is used as a once daily full body wash for up to 5 days at the same time as using intranasal mupirocin antibiotic ointment to the nares twice a day.
For optimal efficiency, this protocol should be initiated within 10-14 days (but no more than 30 days before an operation). Usually the decolonization protocol is used for approximately 5 days leading up to the day of surgery.

A prior study completed in the US in 2012 examined the cost utility of nasal mupirocin for S.aureus decolonization prior to hip and knee arthroplasty. This study used a simple tree model to look at a "screen for S.aureus and decolonize" versus "universal decolonization protocol" versus "standard of care - no decolonization". The results suggested that a "screen and treat" and a "treat all" method were cost-effective compared to no treatment. However, the authors suggested that treating everyone is more practical and should be the strategy of choice.

We sought to evaluate if an evidence based S.aureus decolonization protocol, includi


We sought to evaluate if an evidence based S.aureus decolonization protocol, including intranasal mupirocin and CHG, implemented in all adult patients prior to knee and hip replacement in Alberta, compared with standard care (no routine decolonization) was cost-effective.


SECTION: METHODS
Methods

Economic evaluation

A cost effectiveness analysis (CEA) was conducted to assess the efficiency of implementing a decolonization protocol prior to hip and knee arthroplasty in Alberta.
A cost effectiveness analysis (CEA) was conducted to assess the efficiency of implementing a decolonization protocol prior to hip and knee arthroplasty in Alberta. The analysis assessed the impact on costs and the number of infections avoided.

Study population

The target population for this analysis was all adult patients ( 18 years of age) in Alberta who receive elective knee or hip arthroplasty.

The target population for this analysis was all adult patients ( 18 years of age) in Alberta who receive elective knee or hip arthroplasty. Data for this cohort was based on 24,667 patients who underwent primary elective hip or knee arthroplasty in Alberta between April 1, 2012 and March 31, 2015
Data for this cohort was based on 24,667 patients who underwent primary elective hip or knee arthroplasty in Alberta between April 1, 2012 and March 31, 2015 (mean age 66.5 years). No subgroups were analyzed in this study as all adults undergoing joint replacement may derive benefit from decolonization.

Study setting

In Alberta, nearly all patients who receive an elective hip and knee arthroplasty are triaged and managed through centralized hip and knee clinics and are assigned a nurse case manager.
These clinics are the medical facilities where information and decolonization materials could be supplied to the patients. Patients receive both pre-operative and post-operative follow up care at these clinics.

Perspective

The analysis was done from the perspective of the publicly funded health care system, and as such, included the costs of the inpatient hospitalizations and outpatient ambulatory care visits.


Effectiveness

The effectiveness of a S.aureus decolonization protocol at reducing S.aureus complex SSI was derived from a pre- and post-intervention trial. The pre- and post-intervention study examined 31 701 hip and knee arthroplasties (20642 pre-intervention and 11059 post intervention) completed at 16 different hospitals in the US. In the pre-intervention phase patients were treated with routine pre-operative antibiotics. In the post intervention phase patients were screened for S.aureus nasal carriage and if positive, were treated with 5 days of CHG (2% cloths or 4% body wash) baths and twice daily 2% intranasal mupirocin. They then received appropriate pre-operative antibiotics. This intervention was administered pre-operatively in both the pre- and post-intervention groups. The outcome measured was the number of complex SSIs caused by S.aureus. The rate ratio of complex S. aureus SSIs for hip and knee arthroplasties in the post-intervention group compared to the pre-intervention group was 0.51.

A factor that influenced effectiveness of the model was compliance which was taken into account in the pre- and post-intervention study, where only 83% of patients were adherent to the decolonization protocol (39% fully adherent and 44% partially adherent). Another factor that influenced effectiveness was the number of patients with a complex S.aureus SSI. If there were very few S.aureus infections then it was anticipated that there would be a loss of effectiveness.

Time horizon

In the baseline analysis, the time horizon used was 1 year, as most SSIs following knee and hip arthroplasty occur within the first 3 months post-operatively with a large proportion occurring within the first 30 days.

Statistical analyses

Base case model and data considerations

SECTION: FIG
Simple tree and Markov model for patients undergoing total hip or knee arthroplasty who either have no decolonization treatment or receive mupirocin and CHG decolonization for S.aureus prior to surgery. Abbreviations: S.aureus = Staphylococcus aureus, SSI = surgical site infection



SECTION: METHODS
The model used was a simple decision tree (Fig. 1) and was constructed using decision analysis software (TreeAge Pro 2018 Williamstown, MA).
The model used was a simple decision tree (Fig. 1) and was constructed using decision analysis software (TreeAge Pro 2018 Williamstown, MA). The model was evaluated by clinical experts in order to establish face validityThe model was evaluated by clinical experts in order to establish face validity. A wide range of sensitivity analyses were undertaken to ensure the model responded appropriately.

SECTION: TABLE
Model inputs by treatment alternatives

Variable Usual Care (No decolonization) Decolonization Bundle Ranges for Inputs (95% CI unless otherwise stated) Source Distribution for PSA Risk of S.aureus complex SSI 0.40% 0.338-0.464 AHS IPC Database Beta Risk of other complex SSI 0.64% 0.64% 0.576-0.703 AHS IPC Database Beta RR for S.aureus SSI with decolonization bundle 0.51 0.3-0.85 Schweizer et al. Beta Mupirocin Cost/ person ($)a 0 2.30 1.15-3.45 Alberta Blue Cross Formulary Gamma Chlorhexidine Cost/person ($)a 0 4.95 2.47-7.43 Dufort Lavigne Website Gamma Cost of Nurse Educator/ person ($)a 0 13.35 6.68-20.03 AHS Job Posting for Infection Control Professional Gamma 1 year costs SSI with S. aureus($) 108,175 108,175 88,223-128,127 Rennert-May et al. Gamma 1 year costs SSI with other pathogen($) 87,317 87,317 79,830-94,804 Rennert-May et al. Gamma 1 year costs with no infection ($) 19,893 19,893 11,216-28,570 Rennert-May et al. Gamma

Abbreviations: SSI Surgical Site Infection, AHS Alberta Health Services, IPC Infection Prevention and Control, S.aureus Staphylococcus aureus, PSA Probabilistic sensitivity analysis, RR rate ratio

aRanges for cost created by adding and subtracting 50%

SECTION: METHODS
The incidence and causative pathogens of complex SSI occurring in all adults in Alberta following knee and hip replacement from April 1 2012 to March 31 2015 were collected from a previous study that we completed on this population. Data inputs are displayed in Table 1.

Base case cost inputs

The cost for the mupirocin ointment was taken from the Alberta Blue Cross Formulary. The cost per gram was 45 cents and a 5-g tube would be issued. The cost of 4% CHG sponges was taken from a medical supply website and five would be needed for each patient. It was assumed that in order to implement the decolonization protocol at all hip and knee clinics across Alberta, a nurse educator in Infection Prevention and Control would need to be hired to educate medical staff and create instructional handouts for patients. This cost was estimated based on AHS job postings for nurse educators. When taking into account the number of patients receiving hip and knee replacement annually in Alberta this cost $13.35 per person (including an additional 21% increase on the hourly salary to account for benefits). The total cost for all components of the decolonization bundle was $20.60 per person.

Costs for hospitalizations and management of SSIs as well as the cost of the initial arthroplasty were obtained from a population based cohort study including all patients in Alberta undergoing hip and knee arthroplasty. The costs were a combination of average costs and microcosting data yielding very high quality costing estimates. These costs encompassed inpatient hospitalizations and outpatient visits but did not include patient borne costs (e.g. travel, outpatient antibiotics) or additional physician remuneration.

No discount rate was applied as the time horizon was 1 year.No discount rate was applied as the time horizon was 1 year. All costs were inflated to 2016 CDN$. Costs (and infections avoided) were the output considered in the model.Costs (and infections avoided) were the output considered in the model. Utilities, life years and quality adjusted life years were not considered in this base model given the short time horizon. Costing inputs are represented in Table 1.

Scenario and sensitivity analyses

A one-way sensitivity analysis was completed to estimate the influence that the range of input values had on the overall costs associated with using a bundle. Each input variable was varied one at a time using the 95% confidence intervals. For bundle component costs, dollar values were varied in each direction by 50%. A probabilistic sensitivity analysis (PSA) where we allowed for all variables to change simultaneously (though 1000 Monte Carlo simulations) was completed using distributional assumptions of the input parameters (TabA probabilistic sensitivity analysis (PSA) where we allowed for all variables to change simultaneously (though 1000 Monte Carlo simulations) was completed using distributional assumptions of the input parameters (Table 1).

We considered a scenario where no nurse educator was required as eventually the decolonization bundle could be implemented as part of the normal hip and knee clinic protocol. Additionally, a scenario was conducted where compliance was reduced by 50%. For the purposes of this scenario, we assumed that a reduction in compliance of 50% would increase the rate ratio of S.aureus by 50%, i.e. 0.77.

Finally, to assess model uncertainty, we developed a Markov model (Fig. 1) to examine the long-term impacts of developing a SSI by modeling revision arthroplasty over a lifetime in those who had received a decolonization protocol versus those who did not. It is generally accepted that every time a revision procedure is performed there is an increased risk of requiring another surgical intervention. As almost all patients with a complex SSI will receive some form of surgical procedure in conjunction with the IDSA guidelines, these patients would be at a greater risk of requiring further surgical interventions compared to those who only received an initial arthroplasty. As such, this long-term model enables consideration of the impact of repeated surgeries and having a complex SSI on the need for joint replacements in the future, and the utility of existing in each of these long-term states. This allows for a more complete assessment of how uncertainty in effectiveness impacts results. Further details regarding this model can be found in Additional file 1.

Model validity

This model was validated using guidelines for economic evaluations. Content experts were involved throughout the model creation to ensure face validity. Statistical methods that had been used previously for determining cost inputs were validated. The coding accuracy of the model was tested by changing values to extremes and ensuring the model appropriately responded.

SECTION: RESULTS
Results

Base case

SECTION: FIG
Scatterplot of the probabilistic sensitivity analysis demonstrating that with 1000 iterations, 84% of the time the decolonization protocol is less costly and more effective. Quadrant I is more effective and more costly, quadrant II is more effective and less costly, quadrant III is less effective and less costly, and quadrant IV is less effective and more costly

SECTION: RESULTS
In our base case analysis, the average cost for those who received the decolonization protocol was $20,525 and for those who received standard of care the cost was $20,678 (a cost savings of approximately $153 per person, which in Alberta translates into savings of $1.26 million annually). The risk of developing a S.aureus complex SSI was reduced from 0.4 to 0.2%|. This change in effectiveness translates into 16 complex SSIs from S.aureus annually in Alberta when a decolonization bundle is used versus 32 complex SSIs with standard of care. When the PSA was completed with 1000 iterations, 84% of the time it was more effective and less expensive to utilize the decolonization bundle (Fig. 2).

Sensitivity and scenario analyses

SECTION: FIG
Tornado diagram of one-way sensitivity analyses on the cost difference between treating patients with a decolonization bundle versus standard of care, with a black bar representing the lower end of the range and a grey bar representing the high end of the range. Abbreviations: S.aureus = Staphylococcus aureus, SSI = surgical site infection

SECTION: RESULTS
The one-way sensitivity analysis of each estimate completed, demonstrated that the use of the decolonization bundle would result in a cost savings. The smallest impact on the difference in cost between decolonization and no decolonization occurred when the probability of S.aureus complex SSI after decolonization, compared to no decolonization, increased. At the upper limit of this 95% confidence interval (when the rate ratio was 0.85), the cost savings for decolonization was only $32.00 per person. The largest impact on the cost difference between decolonization and no decolonization occurred when the rate ratio of S.aureus complex SSI was at the lower limit of the range (i.e. 0.3) (cost savings were $227 per person). A tornado diagram that represents only variables that created a difference in incremental costs is provided in Fig. 3.

In the scenario analysis where no nurse educator was required, the savings were $166 per person. In the scenario analysis where compliance was reduced by 50% (hence the rate ratio of S.aureus was 0.77), the cost savings were $61 per person.

Markov model

The Markov model demonstrated that over a lifetime the cost savings for those who were decolonized versus those who were not were $161/person. And the difference in effectiveness (i.e. QALYs) was 0.00096 between those who were decolonized versus those who were not. Additional File 1 provides further detail.

SECTION: DISCUSS
Discussion

We found the use of the decolonization bundle to prevent S.aureus complex SSI resulted in cost savings of $153 per person and 16 SSIs avoided annually (with approximately 8000 hip or knee arthroplasties done). Both the one-way sensitivity analysis and the PSA demonstrated that use of the decolonization bundle generally resulted in fewer complex SSIs, and cost savings, which was consistent with our base model findings. The Markov model discussed further in additional file 1 demonstrated that over a lifetime there were still cost savings for those who received the decolonization bundle - even if complex S.aureus SSIs were reduced from only 0.4 to 0.2%.

Our base case findings of reduced SSIs and costs associated with a decolonization protocol are consistent with prior studies in addition to the one mentioned previously. One study created an economic model to examine the cost-effectiveness of universal S.aureus decolonization with nasal mupirocin prior to hip and knee arthroplasty and determined that there were associated cost savings and reduced infections. Another recent study examined cost-effectiveness of a decolonization protocol for orthopedic procedures. The authors used effectiveness data for decolonization from a randomized controlled trial they had previously completed, examining the eradication of S.aureus. For the cost-effectiveness analysis, they estimated SSI risk from the literature. They found a "treat all" decolonization strategy resulted in the most cost-savings compared to "screen and treat" and "no treatment". However, their costs were based on average estimates from previous studies and were not specific to the patient population they were studying.

Our model was robust to sensitivity analyses demonstrating cost-savings within plausible ranges of the input variables which is consistent with previous studies. In one of the studies mentioned above, their sensitivity analysis demonstrated that the unlikely scenario of infections being reduced by less than 10% and infection related costs being less than $70,000 had to be achieved in order for a decolonization protocol to no longer be cost-saving. In the other study mentioned above they completed both one way sensitivity analyses and a probabilistic sensitivity analysis similar to the current study. In a one-way sensitivity analysis, they noted that a "treat all" approach to decolonization was always more effective and less costly than standard of care. In probabilistic sensitivity analysis, they demonstrated that it was always more effective and less costly to "treat all" with decolonization. Another study demonstrated in their sensitivity analyses that even with modest compliance, it was still cost-saving to utilize home-based CHG bathing cloths pre-operatively to prevent SSIs in patients undergoing orthopedic procedures. This was based on prior studies demonstrating the effectiveness of just CHG pre-operative bathing in reducing SSIs. It should be noted however, in the Schweizer et al. study used for effectiveness in our model, when compliance was broken down into full adherence and partial adherence, the partially adherent group did not have a statistically significant reduction in complex SSIs (RR 0.8 95%CI 0 .49-1.31) suggesting a potential need for both mupirocin and CHG bathing. Full adherence for those with S.aureus positive screening was defined as any CHG bathing and at least 3 days of mupirocin. Partial adherence was defined as any CHG bathing and any mupirocin use.

There is a risk of S.aureus resistance to topical antibiotics that would decrease the effectiveness of a decolonization protocol. In the Schweizer et al. trial, they did test for mupirocin and CHG resistance in 36 S.aureus samples and found only one sample with mupirocin resistance. Any increase in resistance would have been reflected in the probability of S.aureus with decolonization increasing, which was taken into account in the sensitivity analysis.

Our study uniquely used high quality patient-specific microcosting data and local data regarding volume of complex SSIs and causative pathogens, on a large patient population (N = 24,667). We were able to use precise and accurate costing data in our model rather than general estimates from the literature which to our knowledge has not previously been done. While the trial we utilized for effectiveness was not a randomized controlled trial, it was a pragmatic trial with a substantial volume of patients improving the accuracy of the findings. Our model demonstrated that even with a modest reduction in number of infections, decolonization was cost saving. Our study also extended a Markov model of cost-effectiveness of decolonization, specifically in orthopedic surgeries, over a lifetime. This adds value, as while our study was not the first to demonstrate cost-savings with decolonization prior to hip and knee replacements, it considered the long-term impact of CHG and mupirocin decolonization on patient quality of life as well as cost savings. This study also contributes to a paucity of Canadian literature on the cost-effectiveness of decolonization strategies for S.aureus. The epidemiology of knee and hip arthroplasty SSIs and costs associated with them are similar across Canada rendering this model relevant and generalizable nationally. While costs in different countries are not identical to Canada, the principles of the model still apply and so it provides valuable information that can be applied globally.

There were several limitations to our model. The trial used to determine effectiveness of decolonization was not a randomized controlled trial. However, even when the rate ratio for developing a complex S.aureus SSI after a decolonization bundle was as high as 0.94, in the Markov analysis, cost savings were still noted. Additionally, in this study they did not provide results separately for knee and infections. This should not have impacted the results as risk of infection and response to decolonization is likely very similar in hips and knees. This study also implemented a screen and treat strategy and we extrapolated that the rate ratio would still be applicable to our model with a treat all strategy. This was validated with a previous systematic review and meta-analysis that demonstrated that there was no difference in the effectiveness of decolonization strategies when treat all or screen and treat approaches were used.

This trial noted that screening allowed for targeting pre-operative antibiotics more effectively (i.e. vancomycin for MRSA). Targeted prophylaxis was unlikely to make a substantial difference in our outcomes and model which assumed universal decolonization, as rates of MRSA in Canada in hospital patients are fairly low (4%), and pre-operative antibiotics are targeted to cover MRSA when known to be colonized or when there are MRSA risk factors. Finally, unlike our other model inputs that were taken from local data (i.e. costs and rates of complex SSIs) the effectiveness of decolonization was extrapolated from a study conducted in an American setting. However, we feel the study setting was still applicable to our population given that the American study used data from hospitals described as urban facilities, minor teaching centers and community sites ranging from 52 to 514 beds, each having the infrastructure for quality improvement initiatives.

SECTION: CONCL
Conclusions

We determined that a decolonization protocol prior to hip and knee arthroplasty could result in fewer complex S. aureus SSIs and cost savings at both 1 year and over a lifetime.
Implementation of this bundle would have small effects on health services, requiring only a small outlay for decolonization supplies and a nurse educator. Once the program has been implemented, a dedicated nurse educator would likely no longer be required and the cost savings would increase as demonstrated in our alternate scenario. Since lower extremity arthroplasty procedures are increasing over time, we expect higher savings over time as more infections are prevented, and that this may result in a small additional capacity in operating rooms available for other needed procedures. Given our findings, we believe health systems should implement a decolonization protocol prior to hip and knee arthroplasty and continue surveillance of complex SSIs to determine if there is a reduction as anticipated. Additionally, monitoring for the development of resistance to topical antimicrobials should be pursued, particularly once a decolonization protocol is in place.

Additional file

SECTION: ABBR
Abbreviations

CDC

Centers for Disease Control and Prevention

CEA

Cost-effectiveness analysis

CHG

Chlorhexidine gluconate

IDSA

Infectious Diseases Society of America

QALY

Quality adjusted life year

RR

Rate ratio

S. aureus

Staphylococcus aureus

SSI

Surgical Site Infections

US

United States

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

SECTION: METHODS
Availability of data and materials

Data sharing is not applicable to this article as no datasets were generated during the current study.


Ethics approval and consent to participate

Ethics approval for this research was obtained from the University of Calgary Health Research Ethics Board.


Consent for publication

Not applicable.