A recent study in Infection Control and Hospital Epidemiology found that the ICD-9-CM codes for urinary tract infections (UTIs) are unable to successfully identify hospital-acquired catheter-associated UTIs. This is bad news for the Centers for Medicare and Medicaid Services (CMS), which uses ICD-9-CM codes to identify such infections and consequently to determine which kinds of infections, under the Hospital-Acquired Conditions Initiative (commonly known as the no-pay rule ), they are not financially obligated to cover.
This discrepancy comes as no surprise to the researchers who have published multiple studies on the inability of ICD-9-CM codes to identify healthcare-associated infections, including specific infections like MRSA and C. difficile, and conditions such as catheter-associated UTIs and ventilator-associated pneumonia. Why do coding problems exist? As the authors of this study so eloquently stated: Simply put, hospital coders are not diagnosticians or clinicians; they choose diagnoses to list to justify the hospital payment requests, on the basis of a limited review of provider documentation and summaries of major test results.
When Ramanan Laxminarayan and I first assessed the ability of ICD-9-CM codes to track MRSA infections (results presented at the 19th Annual Scientific meeting of the Society for Healthcare Epidemiology of America), I spoke to a hospital coder to learn about how MRSA was coded. In short, she said they look for the word MRSA, and if they see it they code it. She speculated that they were actually under-coding MRSA since the process required specific prompting by the coding computer program. However, when we looked at patients who had the MRSA ICD-9-CM code, we found that 20% had prior history of MRSA but did not have a current MRSA infection. Similarly, Dubberke et al. found that among patients with an ICD-9-CM code for C. difficile but without a C. difficile infection, the majority either had prior history of C. difficile or were tested for C. difficile but the test was negative. Thus, the coder may have looked for the words MRSA or C. difficile and coded a condition as such, even if the patient was not infected.
In the case of hospital-acquired catheter-associated UTIs, coders would need to follow a complicated pattern of ICD-9-CM codes to trigger the no pay rule. In the case of this study, patients with these infections were coded as having a UTI and potentially as having a hospital-acquired UTI but none of these patients were coded as having a UTI that was hospital-acquired AND catheter-associated–the three criteria necessary to trigger the no pay rule. The authors state that this is because catheter use is only documented in nursing notes but coders are instructed to only use notes from physicians, physician-assistants, or nurse practitioners. CMS s strategy to not track no pay rule conditions seems to have backfired since they are still in fact paying for those conditions. However, although CMS may not be benefitting directly from the no pay rule, this initiative has brought the issue of hospital-acquired infections to the forefront. Infections such as catheter-associated, hospital-acquired UTIs are no longer thought of as an unfortunate adverse event. Instead they are now considered preventable problems.
In conclusion, before any ICD-9-CM codes are used for anything other than their intended billing purposes, they need to be validated extensively for the new purpose. It is tempting to use preexisting data for research or new initiatives, but we need to make sure we are actually measuring what we think we are measuring.
Marin L. Schweizer is an Associate in the Department of Internal Medicine at the University of Iowa Carver College of Medicine
Image credit: Flickr: jazzmasterson