Gram-positive bacilli represent a diverse species of bacteria that range from commensal flora to pathogens implicated in severe and life-threatening infection. Following the isolation of Gram-positive bacilli from blood cultures, the time to species identification may take upward of 24 hours, leaving clinicians to conjecture whether they may represent a contaminant (inadvertent inoculation of commensal flora) or pathogenic organism. In this study, we sought to identify patient variables that could help predict the isolation of contaminant versus pathogenic Gram-positive bacilli from blood cultures.

Retrospective cohort study.

One quaternary academic medical center affiliated with the University of Toronto.

Adult inpatients were admitted to hospital over a 5-year period (May 2014 to December 2019).

A total of 260 unique Gram-positive bacilli blood culture results from adult inpatients were reviewed and analyzed in both a univariable and multivariable model.

Malignancy (aOR 2.78, 95% CI 1.33–5.91, p = 0.007), point increments in the Quick Sepsis Related Organ Failure Assessment score for sepsis (aOR 2.25, 95% CI 1.50–3.47, p < 0.001), peptic ulcer disease (aOR 5.63, 95% CI 1.43–21.0, p = 0.01), and the receipt of immunosuppression prior to a blood culture draw (aOR 3.80, 95% CI 1.86–8.01, p < 0.001) were associated with an increased likelihood of speciating pathogenic Gram-positive bacilli from blood cultures such as Clostridium species and Listeria monocytogenes.

Such predictors can help supplement a clinician’s assessment on determining when empirical therapy is indicated when faced with Gram-positive bacilli from blood cultures and may direct future stewardship interventions for responsible antimicrobial prescribing.

Our objective was to conceptualize and assess the potential value of a disease management approach to health technology assessment (HTA) in Canada.

We conducted 18 semi-structured interviews between April 2022 and October 2022 to elicit informant views on potential opportunities for re-conceptualizing the decision-problem in HTA as a disease management problem versus a technology management problem. Participants were purposefully sampled from national and provincial HTA agencies and related organizations in Canada to achieve representation across the disease pathway including prevention, screening, and treatment, and the decision-making pathway including HTA organizations, expert committee members, and decision-makers. Data were analyzed using thematic organizations (based on the interview guide) and manual line by line coding of the data. Ethics approval was received from the Health Sciences Research Ethics Board at the University of Toronto.

Three key features of a disease management approach to HTA (i.e., disease-based, multi-interventional, and dynamic) emerged from informants that differed from traditional HTA processes in Canada. The concept was generally not perceived to be a new idea – some informants indicating that it was implicit in the HTA analysis framework. There was general support for an explicit disease management approach to HTA if the impact of the approach could be demonstrated, if the assessment could be completed within an appropriate time frame, and if the assessment could include the equity, ethical and implementation domains of HTA. Informants indicated that the reconceptualization of HTA could lead to effective and efficient decision-making throughout a technology’s lifecycle, help breakdown system silos, and offer a platform for greater consideration of non-drug alternatives and upstream interventions. The impact of this approach was anticipated to contribute to a proactive health system that could improve population health, enhance the patient experience, and ensure appropriate stewardship of health care resources.

A disease management approach to HTA has international relevance as an approach that could promote integrated, proactive, sustainable, and resilient health systems.

Health technology assessment (HTA) traditionally informs decision making for single health technologies, which could lead to ill-informed decisions, suboptimal care, and system inefficiencies. We explored opportunities for conceptualizing the decision space in HTA as a disease management question versus an intervention management question.

Semistructured interviews were conducted between April 2022 and October 2022 with purposefully selected individuals from national and provincial HTA agencies and related organizations in Canada. We conducted manual line by line coding of data informed by our interview guide and sensitizing concepts from the literature. One author coded the data, and findings were independently verified by a second author who coded a subset of transcripts

Twenty-four invitations were distributed, and eighteen individuals agreed to participate. A disease management approach to HTA was differentiated from traditional approaches as being disease-based, multi-interventional, and dynamic. There was general support for an explicit care pathway approach to HTA by informing discussions around patient choice and suboptimal care, creating a space where decision makers can collaborate on shared objectives, and in setting up a platform for open dialogue about managing high-cost and high-severity diseases. There are opportunities for a care pathway approach to be implemented that build on the strengths of the existing HTA system in Canada.

A disease management approach may enhance the impact of HTA by supporting dynamic decision making that could better inform a proactive, resilient, and sustainable healthcare system in Canada.

To evaluate inter-physician variability and predictors of changes in antibiotic prescribing before (2019) and during (2020/2021) the coronavirus disease 2019 (COVID-19) pandemic.

We conducted a retrospective cohort analysis of physicians in Ontario, Canada prescribing oral antibiotics in the outpatient setting between January 1, 2019 and December 31, 2021 using the IQVIA Xponent data set. The primary outcome was the change in the number of antibiotic prescriptions between the prepandemic and pandemic period. Secondary outcomes were changes in the selection of broad-spectrum agents and long-duration (>7 d) antibiotic use. We used multivariable linear regression models to evaluate predictors of change.

There were 17,288 physicians included in the study with substantial inter-physician variability in changes in antibiotic prescribing (median change of −43.5 antibiotics per physician, interquartile range −136.5 to −5.0). In the multivariable model, later career stage (adjusted mean difference [aMD] −45.3, 95% confidence interval [CI] −52.9 to −37.8, p < .001), family medicine (aMD −46.0, 95% CI −62.5 to −29.4, p < .001), male patient sex (aMD −52.4, 95% CI −71.1 to −33.7, p < .001), low patient comorbidity (aMD −42.5, 95% CI −50.3 to −34.8, p < .001), and high prescribing to new patients (aMD −216.5, 95% CI −223.5 to −209.5, p < .001) were associated with decreases in antibiotic initiation. Family medicine and high prescribing to new patients were associated with a decrease in selection of broad-spectrum agents and prolonged antibiotic use.

Antibiotic prescribing changed throughout the COVID-19 pandemic with overall decreases in antibiotic initiation, broad-spectrum agents, and prolonged antibiotic courses with inter-physician variability. These findings present opportunities for community antibiotic stewardship interventions.

We sought to evaluate the impact of antibiotic selection and duration of therapy on treatment failure in older adults with catheter-associated urinary tract infection (CA-UTI).

We conducted a population-based cohort study comparing antibiotic treatment options and duration of therapy for non-hospitalized adults aged 66 and older with presumed CA-UTI (defined as an antibiotic prescription and an organism identified in urine culture in a patient with urinary catheterization documented within the prior 90 d). The primary outcome was treatment failure, a composite of repeat urinary antibiotic prescribing, positive blood culture with the same organism, all-cause hospitalization or mortality, within 60 days. We determined the risk of treatment failure accounting for age, sex, comorbidities, and healthcare exposure using log-binomial regression.

Of 4,436 CA-UTI patients, 2,709 (61.1%) experienced treatment failure. Compared to a reference of TMP-SMX (61.9% failure), of those treated with fluoroquinolones, 56.3% experienced failure (RR 0.91, 95% CI: 0.85–0.98) and 60.9% of patients treated with nitrofurantoin experienced failure (RR 1.02, 95% CI: 0.94–1.10). Compared to 5–7 days of therapy (treatment failure: 59.4%), 1–4 days was associated with 69.5% failure (RR 1.15, 95% CI: 1.05–1.27), and 8–14 days was associated with a 62.0% failure (RR 1.05, 95% CI: 0.99–1.11).

Although most treatment options for CA-UTI have a similar risk of treatment failure, fluoroquinolones, and treatment durations ≥ 5 days in duration appear to be associated with modestly improved clinical outcomes. From a duration of therapy perspective, this study provides reassurance that relatively short courses of 5–7 days may be reasonable for CA-UTI.

We explore how the scope of analysis is defined in health technology assessment (HTA) and economic evaluation and consider the potential implications of these decisions.

The scope of analysis, including decisions about which methods and domains of HTA to include in the assessment, which costs, and health outcomes are most meaningful, and which comparators are the most relevant are typically informed by the needs of the decision-maker. We undertook two systematic scoping reviews to assess: (i) to what extent systems thinking is considered in literature-based technology assessments; and (ii) how the scope of the analysis is defined in economic evaluation using Clostridioides difficile infection as an exemplar. We synthesized the findings from these reviews and offer three key observations for future research and exploration in the field of HTA.

Our scoping reviews found that the scope of analysis in economic evaluations typically focus on single interventions, often ignoring upstream and downstream interventions. Similarly, published technology assessments have narrowly defined and inconsistent scopes of analysis, with limited consideration of indirect health and non-health impacts. Three key observations for the field of HTA include: (i) economic evaluations focus on the value of single heath interventions. A focus on a single health intervention may simplify the analysis; however, will this siloed decision-making lead to optimal health resource allocation? (ii) published assessments have inconsistently defined scopes of analysis. A decision problem that focuses on the needs of the decision-maker is practical; however, will inconsistencies in perspectives across assessments create unfair conceptualizations of value? (iii) HTA is technology-focused, not patient-focused. A technology-focused HTA system aligns with the technology diffusion process; however, does this move us away from the patient-centered mandate of HTA?

The dynamic nature of HTA leads to many conceptualizations of value. Considering the potential implications of narrowly defined, inconsistent, and technology-focused scopes of analyses may have consequences on achieving a patient-centered high-quality health system.

To evaluate the utility of autologous bone-flap swab cultures performed at the time of cranioplasty in predicting postcranioplasty surgical site infection (SSI).

Retrospective cohort study.

Patients undergoing craniectomy (with bone-flap storage in tissue bank), followed by delayed autologous bone-flap replacement cranioplasty between January 1, 2010, and November 30, 2020.

Tertiary-care academic hospital.

We framed the bone-flap swab culture taken at the time of cranioplasty as a diagnostic test for predicting postcranioplasty SSI. We calculated, sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios.

Among 282 unique eligible cases, 16 (5.6%) developed SSI after cranioplasty. A high percentage of bone-flap swab cultures were positive at the time of craniectomy (66.7%) and cranioplasty (59.5%). Most organisms from bone-flap swab cultures were Cutibacterium acnes or coagulase-negative staphylococci (76%–85%), and most SSI pathogens were methicillin-susceptible Staphylococcus aureus (38%). Bone-flap swab culture had poor sensitivity (0.07; 95% CI, 0.01–0.31), specificity (0.4; 95% CI, 0.34–0.45), and positive likelihood ratio (0.12) for predicting postcranioplasty SSI.

Overall, autologous bone-flap swab cultures performed at the time of cranioplasty have poor utility in predicting postcranioplasty SSI. Eliminating this low-value practice would result in significant workload reductions and associated healthcare costs.

To evaluate different prospective audit-and-feedback models on antimicrobial prescribing at a rehabilitation hospital.

Retrospective interrupted time series (ITS) and qualitative methods.

A 178-bed rehabilitation hospital within an academic health sciences center.

ITS analysis was used to analyze monthly days of therapy (DOT) per 1,000 patient days (PD) and monthly urine cultures ordered per 1,000 PD. We compared 2 sequential intervention periods to the baseline: (1) a period when a dedicated antimicrobial stewardship (AMS) pharmacist performed prospective audit and feedback and provided urine culture education followed by (2) a period when ward pharmacists performing audit and feedback. We conducted an electronic survey with physicians and semistructured interviews with pharmacists, respectively.

Audit and feedback conducted by an AMS pharmacist resulted in a 24.3% relative reduction in total DOT per 1,000 PD (incidence rate ratio [IRR], 0.76; 95% confidence interval [CI], 0.58–0.99; P = .04), whereas we detected no difference between ward pharmacist audit and feedback and the baseline (IRR, 1.20; 95% CI, 0.53–2.70; P = .65). We detected no statistically significant change in monthly urine-culture orders between the AMS pharmacist period and the baseline (level coefficient, 0.81; 95% CI, 0.65–1.01; P = .07). Compared to baseline, the ward pharmacist period showed a statistically significant increase in urine-culture ordering over time (slope coefficient, 1.04; 95% CI, 1.01–1.08; P = .02). The barrier most identified by pharmacists was insufficient time.

Audit and feedback conducted by an AMS pharmacist in a rehabilitation hospital was associated with decreased antimicrobial use.

To determine whether combinations of diagnosis and procedures codes can improve the detection of prosthetic hip and knee joint infections from administrative databases.

We performed a validation study of all readmissions from January 1, 2010, until December 31, 2016, following primary arthroplasty comparing the diagnosis and procedure codes obtained from an administrative database based upon the International Classification of Disease, Tenth Revision (ICD-10) to the reference standard of chart review.

Four tertiary-care hospitals in Toronto, Canada, from 2010 to 2016.

Individuals who had a primary arthroplasty were identified using procedure codes.

Chart review of readmissions identified the presence of a prosthetic joint infection and, if present, the surgical procedure performed.

Overall, 27,802 primary arthroplasties were performed. Among 8,844 readmissions over a median follow-up of 669 days (interquartile range, 256–1,249 days), a PJI was responsible for or present in 586 of 8,844 (6.6%). Diagnosis codes alone exhibited a sensitivity of 0.88 (95% CI, 0.85–0.92) and positive predictive value (PPV) of 0.78 (95% CI, 0.74–0.82) for detecting a PJI. Combining a PJI diagnosis code with procedure codes for an arthroplasty and the insertion of a peripherally inserted central catheter improved detection: sensitivity was 0.92 (95% CI, 0.88–0.94) and PPV was 0.78 (95% CI, 0.74–0.82). However, procedure codes were unable to identify the specific surgical approach to PJI treatment.

Compared to PJI diagnosis codes, combinations of diagnosis and procedure codes improve the detection of a PJI in administrative databases.

Interfacility patient movement plays an important role in the dissemination of antimicrobial-resistant organisms throughout healthcare systems. We evaluated how 3 alternative measures of interfacility patient sharing were associated with C. difficile infection incidence in Ontario acute-care facilities.

The cohort included adult acute-care facility stays of ≥3 days between April 2003 and March 2016. We measured 3 facility-level metrics of patient sharing: general patient importation, incidence-weighted patient importation, and C. difficile case importation. Each of the 3 patient-sharing metrics were examined against the incidence of C. difficile infection in the facility per 1,000 stays, using Poisson regression models.

The analyzed cohort included 6.70 million stays at risk of C. difficile infection across 120 facilities. Over the 13-year period, we included 62,189 new cases of healthcare-associated CDI (incidence, 9.3 per 1,000 stays). After adjustment for facility characteristics, general importation was not strongly associated with C. difficile infection incidence (risk ratio [RR] per doubling, 1.10; 95% confidence interval [CI], 0.97–1.24; proportional change in variance [PCV], −2.0%). Incidence-weighted (RR per doubling, 1.18; 95% CI, 1.06–1.30; PCV, −8.4%) and C. difficile case importation (RR per doubling, 1.43; 95% CI, 1.29–1.58; PCV, −30.1%) were strongly associated with C. difficile infection incidence.

In this 13-year study of acute-care facilities in Ontario, interfacility variation in C. difficile infection incidence was associated with importation of patients from other high-incidence acute-care facilities or specifically of patients with a recent history of C. difficile infection. Regional infection control strategies should consider the potential impact of importation of patients at high risk of C. difficile shedding from outside facilities.

Clostridium difficile spores play an important role in transmission and can survive in the environment for several months. Optimal methods for measuring environmental C. difficile are unknown. We sought to determine whether increased sample surface area improved detection of C. difficile from environmental samples.

Samples were collected from 12 patient rooms in a tertiary-care hospital in Toronto, Canada.

Samples represented small surface-area and large surface-area floor and bedrail pairs from single-bed rooms of patients with low (without prior antibiotics), medium (with prior antibiotics), and high (C. difficile infected) shedding risk. Presence of C. difficile in samples was measured using quantitative polymerase chain reaction (qPCR) with targets on the 16S rRNA and toxin B genes and using enrichment culture.

Of the 48 samples, 64·6% were positive by 16S qPCR (geometric mean, 13·8 spores); 39·6% were positive by toxin B qPCR (geometric mean, 1·9 spores); and 43·8% were positive by enrichment culture. By 16S qPCR, each 10-fold increase in sample surface area yielded 6·6 times (95% CI, 3·2–13) more spores. Floor surfaces yielded 27 times (95% CI, 4·9–181) more spores than bedrails, and rooms of C. difficile–positive patients yielded 11 times (95% CI, 0·55–164) more spores than those of patients without prior antibiotics. Toxin B qPCR and enrichment culture returned analogous findings.

Clostridium difficile spores were identified in most floor and bedrail samples, and increased surface area improved detection. Future research aiming to understand the role of environmental C. difficile in transmission should prefer samples with large surface areas.

Antibiotic use varies widely between hospitals, but the influence of antimicrobial stewardship programs (ASPs) on this variability is not known. We aimed to determine the key structural and strategic aspects of ASPs associated with differences in risk-adjusted antibiotic utilization across facilities.

Observational study of acute-care hospitals in Ontario, Canada

A survey was sent to hospitals asking about both structural (8 elements) and strategic (32 elements) components of their ASP. Antibiotic use from hospital purchasing data was acquired for January 1 to December 31, 2014. Crude and adjusted defined daily doses per 1,000 patient days, accounting for hospital and aggregate patient characteristics, were calculated across facilities. Rate ratios (RR) of defined daily doses per 1,000 patient days were compared for hospitals with and without each antimicrobial stewardship element of interest.

Of 127 eligible hospitals, 73 (57%) participated in the study. There was a 7-fold range in antibiotic use across these facilities (min, 253 defined daily doses per 1,000 patient days; max, 1,872 defined daily doses per 1,000 patient days). The presence of designated funding or resources for the ASP (RRadjusted, 0·87; 95% CI, 0·75–0·99), prospective audit and feedback (RRadjusted, 0·80; 95% CI, 0·67–0·96), and intravenous-to-oral conversion policies (RRadjusted, 0·79; 95% CI, 0·64–0·99) were associated with lower risk-adjusted antibiotic use.

Wide variability in antibiotic use across hospitals may be partially explained by both structural and strategic ASP elements. The presence of funding and resources, prospective audit and feedback, and intravenous-to-oral conversion should be considered priority elements of a robust ASP.

The relationship between hospital antibiotic use and antibiotic resistance is poorly understood. We evaluated the association between antibiotic utilization and resistance in academic and community hospitals in Ontario, Canada.

We conducted a multicenter observational ecological study of 37 hospitals in 2014. Hospital antibiotic purchasing data were used as an indicator of antibiotic use, whereas antibiotic resistance data were extracted from hospital indexes of resistance. Multivariate regression was performed, with antibiotic susceptibility as the primary outcome, antibiotic consumption as the main predictor, and additional covariates of interest (ie, hospital type, laboratory standards, and patient days).

With resistance data representing more than 90,000 isolates, we found the increased antibiotic consumption in defined daily doses per 1,000 patient days (DDDs/1,000 PD) was associated with decreased antibiotic susceptibility for Pseudomonas aeruginosa (−0.162% per DDD/1,000 PD; P=.119). However, increased antibiotic consumption predicted increased antibiotic susceptibility significantly for Escherichia coli (0.173% per DDD/1,000 PD; P=.005), Klebsiella spp (0.124% per DDD/1,000 PD; P=.004), Enterobacter spp (0.194% per DDD/1,000 PD; P=.003), and Enterococcus spp (0.309% per DDD/1,000 PD; P=.001), and nonsignificantly for Staphylococcus aureus (0.012% per DDD/1,000 PD; P=.878). Hospital type (P=.797) and laboratory standard (P=.394) did not significantly predict antibiotic susceptibility, while increased hospital patient days generally predicted increased organism susceptibility (0.728% per 10,000 PD; P<.001).

We found that hospital-specific antibiotic usage was generally associated with increased, rather than decreased hospital antibiotic susceptibility. These findings may be explained by community origins for many hospital-diagnosed infections and practitioners choosing agents based on local antibiotic resistance patterns.

Infect Control Hosp Epidemiol 2017;38:1457–1463

In Canadian hospitals, clinical information is coded according to national coding standards and is routinely collected as administrative data. Administrative data may complement active surveillance programs by providing in-hospital MRSA infection data in a standardized and efficient manner, but only if infections are accurately captured.

To assess the accuracy of administrative data regarding in-hospital bloodstream infections (BSIs) and all-body-site infections due to MRSA.

A retrospective study of all (adult and pediatric) in-hospital MRSA infections was conducted by comparing administrative data against surveillance data from 217 acute Canadian hospitals (124 in Ontario, 93 in Alberta) over a 12-month period. Hospital-associated MRSA BSI cases in Ontario, and for all-body-site MRSA infections in Alberta were identified. Pearson correlation coefficients were used to compare the number of hospital-level MRSA cases within administrative versus surveillance datasets. The correlation of all-body-site MRSA infections versus MRSA BSIs was also assessed using the Ontario administrative data.

Strong correlations between hospital-level MRSA cases in administrative and surveillance datasets were identified for Ontario (r=0.79; 95% CI, 0.72–0.85) and Alberta (r=0.92; 95% CI, 0.88–0.94). A strong correlation between all-body-site and bloodstream-only MRSA infection rates was identified across Ontario hospitals (r=0.95; P<.0001; 95% CI, 0.93–0.96).

This study provides good evidence of the comparability of administrative and surveillance datasets in identifying in-hospital MRSA infections. With standard definitions, administrative data can provide estimates of in-hospital infections for monitoring and/or comparisons across hospitals.

Infect Control Hosp Epidemiol 2017;38:436–443

High-quality cost estimates for hospital-acquired Clostridium difficile infection (CDI) are vital evidence for healthcare policy and decision-making.

To evaluate the costs attributable to hospital-acquired CDI from the healthcare payer perspective.

We conducted a population-based propensity-score matched cohort study of incident hospitalized subjects diagnosed with CDI (those with the International Statistical Classification of Diseases and Related Health Problems, 10th Revision, Canada code A04.7) from January 1, 2003, through December 31, 2010, in Ontario, Canada. Infected subjects were matched to uninfected subjects (those without the code A04.7) on age, sex, comorbidities, geography, and other variables, and followed up through December 31, 2011. We stratified results by elective and nonelective admissions. The main study outcomes were up-to-3-year costs, which were evaluated in 2014 Canadian dollars.

We identified 28,308 infected subjects (mean annual incidence, 27.9 per 100,000 population, 3.3 per 1,000 admissions), with a mean age of 71.5 years (range, 0–107 years), 54.0% female, and 8.0% elective admissions. For elective admission subjects, cumulative mean attributable 1-, 2-, and 3-year costs adjusted for survival (undiscounted) were $32,151 (95% CI, $28,192–$36,005), $34,843 ($29,298–$40,027), and $37,171 ($30,364–$43,415), respectively. For nonelective admission subjects, the corresponding costs were $21,909 ($21,221–$22,609), $26,074 ($25,180–$27,014), and $29,944 ($28,873–$31,086), respectively.

Hospital-acquired CDI is associated with substantial healthcare costs. To the best of our knowledge, this study is the first CDI costing study to present longitudinal costs. New strategies may be warranted to mitigate this costly infectious disease.

Infect Control Hosp Epidemiol 2016;37:1068–1078

Hospital-acquired infections (HAIs) develop rapidly after brief and transient exposures, and ecological exposures are central to their etiology. However, many studies of HAIs risk do not correctly account for the timing of outcomes relative to exposures, and they ignore ecological factors. We aimed to describe statistical practice in the most cited HAI literature as it relates to these issues, and to demonstrate how to implement models that can be used to account for them.

We conducted a literature search to identify 8 frequently cited articles having primary outcomes that were incident HAIs, were based on individual-level data, and used multivariate statistical methods. Next, using an inpatient cohort of incident Clostridium difficile infection (CDI), we compared 3 valid strategies for assessing risk factors for incident infection: a cohort study with time-fixed exposures, a cohort study with time-varying exposures, and a case-control study with time-varying exposures.

Of the 8 studies identified in the literature scan, 3 did not adjust for time-at-risk, 6 did not assess the timing of exposures in a time-window prior to outcome ascertainment, 6 did not include ecological covariates, and 6 did not account for the clustering of outcomes in time and space. Our 3 modeling strategies yielded similar risk-factor estimates for CDI risk.

Several common statistical methods can be used to augment standard regression methods to improve the identification of HAI risk factors.

Infect. Control Hosp. Epidemiol. 2016;37(4):411–419

Surgical site infections (SSIs) are common hospital-acquired infections. Tracking SSIs is important to monitor their incidence, and this process requires primary data collection. In this study, we derived and validated a method using health administrative data to predict the probability that a person who had surgery would develop an SSI within 30 days.

All patients enrolled in the National Surgical Quality Improvement Program (NSQIP) from 2 sites were linked to population-based administrative datasets in Ontario, Canada. We derived a multivariate model, stratified by surgical specialty, to determine the independent association of SSI status with patient and hospitalization covariates as well as physician claim codes. This SSI risk model was validated in 2 cohorts.

The derivation cohort included 5,359 patients with a 30-day SSI incidence of 6.0% (n=118). The SSI risk model predicted the probability that a person had an SSI based on 7 covariates: index hospitalization diagnostic score; physician claims score; emergency visit diagnostic score; operation duration; surgical service; and potential SSI codes. More than 90% of patients had predicted SSI risks lower than 10%. In the derivation group, model discrimination and calibration was excellent (C statistic, 0.912; Hosmer-Lemeshow [H-L] statistic, P=.47). In the 2 validation groups, performance decreased slightly (C statistics, 0.853 and 0.812; H-L statistics, 26.4 [P=.0009] and 8.0 [P=.42]), but low-risk patients were accurately identified.

Health administrative data can effectively identify postoperative patients with a very low risk of surgical site infection within 30 days of their procedure. Records of higher-risk patients can be reviewed to confirm SSI status.

Infect. Control Hosp. Epidemiol. 2016;37(4):455–465