Discussion

In the general hospital population of adults, the proportion of non-CAUTI HOUTI was over 7-fold greater compared to CAUTI. The attributable risk of increased LOS and total hospital cost of both CAUTI and non-CAUTI HOUTI were significantly higher compared to controls. Though the risk of likely secondary HOB was higher in patients with CAUTI vs non-CAUTI HOUTI, the volume of HOB was three times greater in non-CAUTI HOUTI. This finding has implications for infection control programs as non-CLABSI HOB has been shown to sustain even higher incremental risk for longer LOS, higher hospital costs, and depending on ICU status, higher mortality and 30-day readmission risk. ^{Reference Yu, Jung and Ai12}

These findings highlight the importance of recognizing both CAUTI and non-CAUTI HOUTI and are complemented by a recent practice recommendation to establish a system for defining, analyzing, and reporting data on non-catheter-associated UTIs. ^{Reference Patel, Advani and Kofman21} While CAUTI remains the more costly of the two infections, this analysis suggests that reducing not only CAUTI, but also non-CAUTI HOUTI, may result in improved patient outcomes and lower overall HOB rates. E. coli bloodstream infection rate is a UK National Health Service oversight metric ²² and research in the UK has found that E. coli bacteremia is most frequently caused by CAUTI. ^{Reference Melzer, Wickramasinghe and Welch23} This is worthy of attention given that the Centers for Medicare & Medicaid Services is currently reviewing an all-cause HOB metric. ^{Reference Stack, Dbeibo, Fadel, Kelley, Sadowski and Beeler24}

As the definition for CAUTI has become more stringent, the prevalence of CAUTI has declined. ^{Reference Sopirala, Syed, Jandarov and Lewis7,Reference Press and Metlay8,Reference Fakih, Groves, Bufalino, Sturm and Hendrich10} This study found that only 12.0% of all HOUTIs were identified as CAUTIs and thus the vast majority of HOUTIs did not meet currently reportable NHSN criteria. An analysis of cases of hospital-onset BSI, secondary to HOUTI, following introduction of the 2015 NHSN CAUTI definition changes, found that 41.9% of those HOB infections could not be classified as secondary to CAUTI because they were either fungal in nature or because the bacterial urine culture was <100,000 CFU/mL. ^{Reference Greene, Ratz, Meddings, Fakih and Saint25} Our analysis builds upon these data, highlights the importance of both types of HOUTI and associated patient outcomes, and suggests a clinical opportunity for infection prevention efforts to focus on the more serious clinical outcomes of HOB via all-cause HOUTIs irrespective of indwelling urinary catheter use or limitations of pathogen levels and species.

Prior analyses of the cost of CAUTI have yielded low values. A 2010 Agency for Healthcare Research and Quality (AHRQ) estimate for the cost of CAUTI was $1,090 increasing more than 10-fold in seven years in a subsequent analysis. ^{Reference Bysshe, Gao and Heaney-Huls26} The researchers noted that efforts to reduce the utilization of indwelling urinary catheters, and CAUTI definition changes, likely concentrated the remaining CAUTI cases to those patients with severe, more costly infections. Systematic reviews have previously estimated the cost of CAUTI to range from $1,768 to $10,197 (Medicare beneficiaries without and with ICU stays, 2016 USD) ^{Reference Hollenbeak and Schilling27} to $13,793 (2015 USD). ^{Reference Bysshe, Gao and Heaney-Huls26} This large range may be partially explained by the acuity associated with an extended LOS. Stratification by LOS and HOB during the hospitalization showed that the attributable cost of CAUTI was as low as $5,439 to $7,337 and was responsible for 2.58 to 3.07 additional LOS days in patients with LOS of ≤10 days and no matched HOB, depending on ICU status during the hospitalization.

This study is one of the first to quantify the attributable cost of non-CAUTI HOUTI. The attributable cost of non-CAUTI HOUTI was approximately $6,000 contributing 2.46 to 2.98 additional LOS days, depending on ICU status, in the least acute group of patients analyzed. This value is plausible and consistent with prior studies. ^{Reference Tabak, Sung, Ye, Vankeepuram, Gupta and McCann28} It is within the lower end of the 95% CI range for CAUTI ($5,019 to $22,568, 2015 USD) set forth in the AHRQ report ^{Reference Bysshe, Gao and Heaney-Huls26} and similar to the $6,424 (2011 USD) cost of a UTI admission determined by an analysis of the Nationwide Inpatient Sample, ^{Reference Simmering, Tang, Cavanaugh, Polgreen and Polgreen29} both of which were published in 2017. We therefore augment these analyses with more recent outcomes and cost data.

This study had several limitations. The analyzed hospitals were limited to those where total cost of care per admission and NHSN HAI reporting were both readily available. The urine culture stewardship practices of the individual hospitals were unknown, for example, it is possible that cultures may have been ordered less frequently for patients with an indwelling catheter. The association between potential secondary HOB and HOUTI may have been underestimated due to the following: (1) application of a stringent definition of HOB which required it to occur on day 4 or later and within 48 hours of the first positive urine culture collection date, and (2) patients with admissions for infection were excluded potentially eliminating the highest acuity patients with HOUTI. The location of the urine specimen collection may not have been the true location of infection onset depending on patient movement between infection onset and when the specimen was obtained. In addition, with no access to notes in the medical record, the mechanism for collecting the urine specimen and/or the mechanism of urine management, including the use of indwelling urinary catheters, could not be ascertained (note that uropathogen growth ≥10,000 CFU/mL is considered significant in specimens obtained with an in-and-out catheter per ASM recommendations in the Clinical Microbiology Procedures Handbook ^{Reference Yarbrough and Potter16} ), additional LOS could not be ascribed to the HOUTI with absolute certainty. Accounting for potential confounding was done at the design and analysis phase of the study. However, residual confounding may still be possible as with all observational studies. Finally, as the CDC definition of HOB is still evolving, the definition employed in this analysis is a signal of HOB burden and not the exact definition that may ultimately be used by the CDC and NHSN.

Despite these limitations, the study had several strengths. A stringent definition for non-CAUTI HOUTI required cases to have both: (1) an algorithmically derived indication of a likely hospital-onset UTI based on the positive urine culture, and (2) an order for complementary antimicrobials within 48 hours of that first positive culture (resulting in exclusion of 32.8% of HOUTIs to arrive at our cohort of 3,177 cases of non-CAUTI HOUTI). It is thus highly unlikely that asymptomatic bacteriuria was captured in any of the analyses. In addition, a large sample of US hospitals was utilized including sites with and without academic affiliations. Lastly, the database included both laboratory results and pharmacy orders providing a novel opportunity to evaluate non-CAUTI HOUTI and its association with HOB.

In conclusion, non-CAUTI HOUTI is a common and potentially preventable source for HOB that is associated with longer LOS and higher hospital costs compared to those in similar patients without an infection. While a higher percentage of CAUTI cases evolve into HOB, non-CAUTI HOUTI-related HOB occur more often. These data support the ongoing surveillance of not only CAUTI but also non-CAUTI HOUTI as the latter may also represent a vulnerable population that could benefit from surveillance and prevention efforts, particularly in the non-ICU setting.