CAUTIs and CLABSIs: Do Physicians REALLY Know What They Are?

Patients and Methods

We performed a retrospective study from January 2012 through September 2013 to investigate the incidences of two hospital-acquired conditions (HACs): Central line-associated blood stream infections and catheter-associated urinary tract infections. We compared the number of these infections identified through the UHC administrative database to the total identified by the Department of Epidemiology (CDC). Agreement between HACs identified by the Department of Epidemiology and UHC was quantified through calculation of the kappa statistic. HACs identified by the Department of Epidemiology were considered true disease and used as reference for calculating the diagnostic accuracy of HACs identified by UHC. We performed a subset analysis to evaluate those HACs identified by UHC but not CDC to determine the reasons for the misidentifications. This study was approved by the Virginia Commonwealth University IRB.

The UHC data comes from diagnostic and procedural codes associated with HACs and patient safety indicators (PSI). These parameters are set by AHRQ and CMS. At our institution, medical coders review all medical documentation in a patient's record and assign the appropriate diagnosis and procedure codes. In addition, the institutional 3M coding software has triggers in place to alert the coders to the presence of both HACs and PSIs. Once the coders complete the coding for a patient encounter, a clinical documentation nurse, coding manager, and clinical reviewer, prior to the encounter being submitted to UHC for analysis, then review the record.

The Department of Epidemiology follows patient microbial culture data daily to identify hospital acquired infections. In addition, infection control nurses track urinary and vascular catheter use throughout the institution, and record timing of placement and removal. In the setting of positive cultures, further investigation is performed to determine if the infection was present on admission, based on positive cultures present within two 2 d of admission, or if the catheter-associated infection is secondary to an infection at another site. If these exclusionary criteria do not exist, designation of a catheter-associated infection may be made using strict infection site-specific CDC guidelines related to the clinical signs and laboratory data [2].

Results

There were a total of 459 infections identified between both UHC and CDC methods. Of these, 221 were CAUTIs and 238 were CLABSIs. Out of the 221 CAUTIs, 16 (7.2%) were detected by both UHC and CDC methods. Forty-four (18.5%) of the 238 CLABSIs were identified by both UHC and CDC approaches. Using the CDC definition, more than twice as many infections were documented than were found with the UHC method (368 vs. 151). The CDC method identified 179 CAUTIs and 189 CLABSIs. Of the 179 CDC-identified CAUTIs, 91% were not identified by UHC. Of the 189 CDC-identified CLABSIs, 77% were not identified by UHC. The kappa statistic for agreement between UHC and CDC methods was poor for CAUTIs (kappa: 0.13) and CLABSIs (kappa: 0.31) as indicated by kappa scores below 0.4. Table 1 shows the incidence of HACs based on the identification method used.

Using CDC-identified HACs as the marker of true disease, the diagnostic accuracy of UHC identification of CAUTIs and CLABSIs was investigated. UHC-identification of CAUTIs and CLABSIs was associated with poor sensitivity (8.9% and 23.7% respectively) and high specificity (99.9% and 99.9% respectively). Table 2 displays the diagnostic accuracy of UHC identification.

The UHC method identified a total of 151 infections: 58 CAUTIs and 93 CLABSIs. 72.4% (42/58) of the CAUTIs identified by UHC were not identified by CDC. 52.7% (49/93) of the CLABSIs identified by UHC were not identified by CDC. The reasons for infections identified by UHC but not CDC included lack of culture data obtained, lack of culture-positivity, and infections because of catheters that were present on admission. These findings are shown in Table 3.

Discussion

Hospital acquired conditions, in particular catheter-associated infections of the blood and urine, have a major detrimental impact on the health care system [1]. Morbidity and mortality from such conditions directly compromise patient outcomes and their effects are well documented. As a result of such infections, hospital lengths of stay are prolonged, further laboratory and radiographic investigations are increased, and overall costs increase [3,4]. In an effort to characterize these implications, the Centers for Medicaid and Medicare now track closely the incidence of these diseases.

Closer observation by regulatory and payer agencies began in 2008 when the concept of HACs were introduced. At that time, individual claims were reduced if the HAC resulted in assignment of a patient to a greater Diagnosis-Related Group (DRG). As the importance of HACs and additionally patient safety indicators (PSI) increased, the financial consequences became greater. In 2014, PSI 90, which combines HACs with other indicators, was added to the hospital value based purchasing program that places 1–2% of hospital Medicare payments at risk. As of 2015, the HAC reduction program will be fully implemented, after which Medicare will reduce hospital payments by 1% for those hospitals ranked among the lowest performing quartile with regard to hospital acquired conditions. Therefore, if physicians and other health care providers do not actually understand what the definitions of these infections are and document accurately, millions of dollars may be lost by the institution.

On the basis of our results, there is a profound inconsistency between the more subjective documentation-based UHC identification and the more objective CDC criteria-based identification. The UHC definition relies on provider documentation such that simply stating in the medical record that a patient has a CAUTI or CLABSI will result in a HAC regardless of consideration of culture data, timing of catheter placement or duration between catheterization and infection-onset. Despite the rigor involved in optimizing the coding, the process is flawed as it begins with the subjective documentation. Consequently, there is great disparity in the overall number of infections identified, with many more identified through the CDC approach. Such disparities are found throughout the literature [5–7]. Our kappa statistic was below 0.4 for both CAUTIs and CLABSIs indicating poor agreement between the two identification methods, mirroring other findings of discordance between administrative database extractions and gold-standard chart review [8]. Furthermore, Cass et al. noted a 0% and 21% sensitivity for administrative databases compared with CDC for CAUTI and CLABSI detection, respectively [9]. Our findings confirmed the low sensitivity and positive predictive value associated with UHC identification, thereby making it a poor screening test. The more concerning fact is that so few of the UHC identified patients correlated with those identified by the CDC method. This finding has serious ramifications. First, it may result in patients without true infections, as defined through culture positivity, being treated with antimicrobial agents that they do not need, and secondly, it would have a deleterious effect on the institution's profile, thus increasing the likelihood of a financial penalty.

As shown in the tables, 26% of the UHC-identified CAUTIs and 53% of the UHC-identified CLABSIs may not have required any antimicrobial therapy given that they either had no positive cultures, contaminants-only, or never had cultures drawn, thus making the diagnosis suspect. However, as a result of providers documenting these infections in the chart, a substantial amount of patients likely received unnecessary antimicrobial therapy. Strong antimicrobial stewardship programs embrace appropriate therapy and minimization of duration along with de-escalation to improve patient outcomes, and such findings would be inconsistent with this practice [10,11].

It is also important to note that almost 60% of the UHC-identified CAUTIs and 31% of the UHC-identified CLABSIs were considered true positives when the charts were reviewed. Upon further investigation, the majority of these were linked to location and timing of catheter placement, whereby many of the ‘missed infections’ were because of catheters placed in the operating room and thus not having the associated catheter-placement documentation in the medical record. Similarly for the CLABSIs, peripherally inserted central catheter (PICC) documentation at our institution varies and this may have contributed to false negatives in the CDC cohort. These findings emphasize the inherent challenges with the process of infection identification.

Given the limitations of both of these approaches, it is vital for providers to have a clear understanding of what defines a hospital-acquired infection so as to appropriately treat patients and document accurately. As the HAC reduction program progresses, changes are being made to account for some of the inadequacies. Information to assist with comparison scores will be coming from two domains: One will rely primarily on the PSI 90 that combines a number of patient safety indicators as defined by UHC, and the other domain will utilize the CDC National Healthcare Safety Network (NHSN) to incorporate more objective data [12]. This development has future accuracy implications as surgical site infections will also be measured as part of the reduction program in 2016.

This study is limited by the fact that we do not have data on why the CDC identified so many more infections than UHC. Moreover, we cannot be sure whether some of these patients were under-treated as well. Finally, improving accuracy in defining these infections would have resulted in a substantial increase in UHC HACs given the much greater number of CDC defined infections. This improvement, in turn, would lower our institutional rating and jeopardize more funds such that it may not be to our institution's financial benefit. However, our goal was not necessarily to improve our financial standing but to ensure that we, along with other institutions, are being measured fairly.

In conclusion, health care providers need to have a better understanding of what defines hospital acquired conditions, as ignorance is certainly not bliss in this situation. The accuracy of identification and documentation has implications for direct patient care and long-term institutional financial viability. Investment in strong infection prevention and epidemiology expertise is critical to the success of these efforts. Ongoing education of providers to understand these definitions as well as document appropriately should improve this situation and prepare all of us for the changes that are on the horizon.