(See the article by Vaz et al, on pages 649–655.)

Since 2008, the Centers for Medicare and Medicaid Services (CMS) has adopted a nonreimbursement policy on selected healthcare-associated infections (HAI), hoping to improve patient outcomes by decreasing HAI rates.Reference Stone ¹ This important change, which could affect direct revenues of hospitals, has received much attention. However, it is difficult for both hospital epidemiologists and infection preventionists to link direct nonreimbursement results from the CMS-selected HAI, based on billing data, with HAI identified by clinical surveillance. The national survey on the perceived impact of CMS policies among respondent infection preventionists from 317 hospitals in 2010 revealed a trend toward prioritizing resources on CMS-targeted HAI rather than nontargeted HAI.Reference Lee, Hartmann and Graham ² Because identifying HAI using administrative claims data has been regarded as problematic, with the potential for misclassification bias,Reference Edmond ³ many infection control professionals have raised concerns about the nonreimbursement policy’s negative impact. The study by Vaz et al in this issue of ICHE provides a good example of these concerns: resource-constrained, safety-net hospitals may be adversely affected by the CMS policy compared with non–safety-net hospitals.

Despite these concerns, this study reported no positive or negative impact of the CMS policy on either central-line–associated bloodstream infection (CLABSI) or ventilator-associated pneumonia (VAP), regardless of safety-net hospital status. This finding is consistent with those of a similar study of CMS impact on selected HAI rates (CLABSI, VAP, and catheter-associated urinary tract infection), in which a coauthor of the present study (Lee) and his colleagues reported no significant differences among these rates.Reference Lee, Kleinman and Soumerai ⁴ These two studies improved the sensitivity of identifying HAI by using the National Healthcare Safety Network (NHSN) definitions of HAI instead of the International Classification of Diseases, Ninth Revision (ICD-9) codes from administrative billing data. In contrast, another pre–post CMS nonreimbursement studyReference Peasah, McKay, Harman, Al-Amin and Cook ⁵ using ICD-9 diagnosis codes and present on admission (POA) coding reported a significant decrease in hospital-acquired vascular catheter-associated infections for the 3-year postimplementation period. Although Peasah et alReference Peasah, McKay, Harman, Al-Amin and Cook ⁵ claim limited code manipulation on their results, with 97% POA coding accuracy, discrepancies among study results with different HAI identification methods may reflect unknown changes in coding practices.Reference Stone ¹ Further research that compares CMS policy impact using both methods (NHSN and ICD-9 codes) in identical hospital settings may provide a clearer understanding of these discrepancies.

As noted in these prior studies, ongoing CLABSI decreases were observed before and after the implementation of the CMS nonreimbursement policy. CLABSI is a highly targeted HAI due to its substantial burden and its potential preventability, which is enough to trigger the “zero tolerance” concept. Over the last decade, more advanced, evidence-based HAI prevention efforts have been made as HAI have received greater attention as a major patient safety issue.Reference Burke ⁶ For the same period, more robust surveillance data have been available for prioritizing HAI prevention efforts; many hospitals have adopted electronic surveillance support software combined with electronic medical records. In particular, implementation of CLABSI bundles (ie, packages of various preventive measures including full-barrier precautions) has reduced CLABSI incidences substantially over the last decade, and bundles have proven a sustainable intervention.Reference Pronovost, Watson, Goeschel, Hyzy and Berenholtz ⁷ In addition, the overall HAI prevention efforts by hospitals (eg, adopting systemwide safety cultures) have been enhanced by several seminal prevention calls, such as the federal Action Plan to Prevent Healthcare-Associated Infections and public HAI reporting, concurrent with the CMS nonreimbursement policy for selected HAI, starting October 2008. Due to these overlapping efforts to reduce CLABSI, it not surprising that a uniquely significant impact of the CMS policy on reducing actual HAI was not seen. Despite the unproven impact, it remains undeniable that the CMS policy likely plays a significant role in justifying the value of HAI prevention efforts with direct financial implications for hospitals and in persuading healthcare personnel to comply with infection control guidelines.

In addition to being the first study to examine the CMS policy’s impact on HAI between safety-net hospitals and non–safety-net hospitals, the study by Vaz et al has several distinctive strengths. First, to be compatible with the CMS nonreimbursement policy, the CMS Disproportionate Share Hospital (DSH) Adjustment index was used in this study to identify eligible hospitals for additional payments to care for impoverished patient populations. Second, almost a quarter of US hospitals (N=367) were recruited for this study. Third, additional evidence was presented in this study regarding decreasing HAI over the study period (from July 2007 to December 2013 for CLABSI and to December 2012 for VAP, including 2.75 years more information after the study by Lee et alReference Lee, Hartmann and Graham ² ).

Due to study scope limitations, Vaz et al would probably be unable to conduct additional analyses to delve into the impact of other potential patient- and hospital-level factors on the observed trends in CLABSI and VAP. Although the inherent complexity of VAP makes it a poor indicator of performance, Figure 2 presents the interesting trend of decreasing VAP in safety-net hospitals, narrowing the gap in VAP rates between safety-net and non–safety-net hospitals until it disappears entirely, similar to CLABSI trends in Figure 1. In addition to the aforementioned external forces, one possible explanatory factor may be potential shifts in nurse staffing that were mandated in some states around 2010 due to a strong relationship between nurse–patient ratios and patient safety outcomes. ⁸ One study on the impact of a nurse staffing mandate on safety-net hospitals in California from 1998 to 2007 reported an unchanged skill mix (ie, the proportion of registered nurses among all nursing staff) following the mandate and found a high level of complexity in safety-net hospitals.Reference McHugh, Brooks Carthon, Sloane, Wu, Kelly and Aiken ⁹ However, updated evidence is unavailable for both potential shifts in nurse staffing and changes in VAP prevention practices in safety-net hospitals regarding the gap changes around 2010.

Although the Vaz et al study results suggest no significant disadvantages of the CMS nonreimbursement policy in safety-net hospitals, concerns remain regarding potential CMS financial penalties for large teaching, tertiary hospitals that provide complicated patient care and likely have more HAI among sicker hospitalized patients. Because CMS uses unadjusted national benchmark scores in their hospital-acquired condition (HAC) reduction program, an estimated 56.4% of large teaching hospitals may be subject to payment reductions in fiscal year 2015. ¹⁰ Therefore, adjustments for sicker patients in large teaching hospitals and safety-net hospitals still need to be considered in the CMS HAC reduction program and their non-reimbursement policy. Ideally, a symbolic level of financial incentives for ongoing HAI prevention efforts in the complex care of sicker patients may spur prevention efforts. At the least, eliminating concerns about potential financial penalties despite all HAI prevention efforts may yield more positive synergies in reducing HAI.