Despite an emphasis on quality performance and a system-wide electronic health record, implementation of automated eGFR reporting among VHA laboratories was incomplete and varied over many years following its adoption mandate. This wide variation could not be explained by facility-level organizational characteristics as there were no significant differences in these characteristics between the stages of implementation. There was only one significant difference by implementation status: presence of dialysis services was associated with implementation. Importantly, delayed or absent implementation of automated eGFR reporting could have translated into missed opportunities for earlier diagnoses of CKD among US veterans. Also, these findings illustrate that implementation of laboratory IT is not associated with level of healthcare system integration or presence of facility-level QI characteristics.
Rates of implementation of automated eGFR reporting have been evaluated in other studies. A survey from non-VHA settings in 2007 revealed that 40% of U.S. clinical laboratories reported eGFR with creatinine values in 2007 [5
].That same year, the College of American Pathologists’ Annual Survey revealed eGFR reporting was used in approximately 50% of laboratories [12
]. In the current study, we found that approximately 68% of VHA laboratories were reporting eGFR by 2007. Although these approximations of automated eGFR reporting usage were obtained by different methods, they all show a similar trend of incomplete implementation many years from initial software availability and consensus recommendations. This data suggests VHA laboratories had higher prevalence of automated eGFR reporting than non-VHA laboratories in 2007, and this may be explained by the VHA’s highly integrated healthcare system and electronic health record. Although when compared to another large integrated healthcare system, Alberta Health Services in Alberta, Canada, VHA laboratories had gradual implementation. In 2004, both healthcare systems adopted eGFR reporting software, but only Alberta Health Services had complete implementation of automated eGFR reporting software in its laboratories within that year [13
]. Given that implementation of this innovation can vary regardless of system integration, more specific organizational characteristics likely explain this variation.
We sought to identify specific organizational characteristics that describe the variation in implementation of automated eGFR reporting within the VHA. Our analyses revealed that facilities that implemented the software were more likely to provide dialysis services than facilities that did not implement. In the VHA, facilities with dialysis units typically offer tertiary care services, and the demands of a more complex patient population and the better availability of financial resources in these facilities differs from facilities that provide only primary care services. As a result, the association of dialysis services to implementation of automated eGFR reporting suggests that presence of dialysis services may be highly correlated with one or all of the following: 1) presence of other resources in a facility, such as experienced laboratory information technology staff; 2) high demand to provide complex medical services (so facility is more apt to incorporate new innovations); or 3) efficient chain of command (facilities with dialysis units are larger and may incorporate more organizational hierarchy that promotes better accountability to tasks). Overall, the association of implementation of automated eGFR reporting and presence of dialysis services is consistent with previous research that indicates environmental barriers, such as lack of resources, may reduce provider adherence to clinical practice guidelines [14
Aside from this finding, none of the other organizational characteristics were associated with implementation of automated eGFR reporting. These null results may have occurred because the variables used to describe the organizational domains of the conceptual model were only surrogate measures of implementation of automated eGFR reporting (Table ). For example, the use of clinical champions is an important tool for implementation of healthcare innovations [15
], and the CPOS-COS survey item for this variable was examined as a surrogate measure to determine whether use of clinical champions is important for implementation of automated eGFR reporting. All of the facilities reported similar frequencies in use of clinical champions for general QI initiatives. Although these similarities exist for general QI initiatives, the survey didn’t solicit the opinions of laboratory personnel to directly assess the use of clinical champions in implementation of automated eGFR reporting. As a result, the importance of clinical champions cannot be ruled out because our methods don’t include direct measurement of this variable. Similarly, all other organizational variables and domains from the conceptual model should be measured more directly before one concludes that there is no association with implementation of automated eGFR reporting. A similar survey distributed to laboratory personnel could provide more direct and specific assessment of these organizational variables. Not only that, direct query of laboratory decision-making processes or laboratory and IT leadership opinions, could provide more insight beyond the domains of this conceptual model.
We did not expect to find that none of the facility level characteristics from the CPOS-COS survey were associated with time to implementation of automated eGFR reporting. Intrinsic to our study design, we used the CPOS-COS survey for facility-level organizational variables because other similar studies demonstrated variable performance and practice patterns among VHA facilities that are related to facility- and clinic-level organizational characteristics [10
]. We further justified use of the CPOS-COS survey because general QI characteristics can affect implementation of new innovations in individual facilities [22
]. Conversely, our results show that general QI characteristics were not predictive of implementation of automated eGFR reporting. Automated eGFR reporting and other laboratory IT innovations are implemented differently from clinical innovations and are less affected by the extent of clinical QI infrastructure. This is an important lesson for future studies that attempt to assess the role facility-level organizational characteristics have in implementation of laboratory IT innovations.
This study has limitations that should be acknowledged. First, we did not assess laboratory-specific organizational characteristics for association with implementation of automated eGFR reporting. Without laboratory-specific variables, an explanation for the delayed implementation of automated eGFR reporting remains unclear. Second, we could not use verified start dates (of implementation) for each facility because of data transmission errors within the administrative data. As a result, we derived a definition for initiation of automated eGFR reporting that provided approximate dates of implementation. Last, the retrospective study design limited the data available for analysis of the process of software implementation. To counter this limitation, we utilized the 2006 CPOS-COS survey responses to obtain organizational characteristics that were measured within two years of the VHA’s initial adoption of automated eGFR reporting in 2004.
The wide variation in implementation of automated eGFR reporting in the VHA and other laboratories draws attention to the ongoing need for quality improvement in CKD management. Over the past decade, there has been interest in improving CKD identification, and automated eGFR reporting has been endorsed as a tool to enhance detection of CKD [2
]. In fact, many studies, as described in a systematic review, have shown that automated eGFR reporting is associated with earlier detection of CKD [1
]. Consequently, CKD detection may have been delayed in some VHA facilities as a result of the wide variation in implementation of automated eGFR reporting. This potential delay in diagnoses could be associated with disparate health outcomes between veterans who receive care at eGFR reporting facilities and those who do not. Specifically, late diagnosis of CKD leads to late nephrology referral which has been associated with increased mortality among those who progress to kidney failure [23
]. Because of inconsistent implementation of automated eGFR reporting in other U.S. laboratories, disparate health outcomes may also exist outside the VHA. Given the growing prevalence of CKD, a concerted effort to enhance early detection and management of CKD remains important to prevent adverse outcomes and slow disease progression [6
This study also has implications for future laboratory reporting innovations in nephrology. Although laboratories currently report eGFR from the MDRD equation, the MDRD equation may eventually be replaced by newer estimating equations, such as the Chronic Kidney Disease Epidemiology Collaboration (CKD-EPI) equation or estimates based on alternative biomarkers (e.g., cystatin-C) [25
]. Alternative equations for estimation of GFR will necessitate the development and activation of additional software patches in clinical laboratories in the VHA and worldwide. Without a clear perspective on predictors of implementation, widespread implementation of additional eGFR reporting equations may be delayed.
Although we did not identify organizational characteristics clearly associated with rate of implementation of automated eGFR reporting, further investigation is warranted to inform implementation of future laboratory IT innovations which may lead to timely implementation and improved disease management. Future studies could include a qualitative analysis of facilities that did not implement automated eGFR reporting to reveal barriers to implementation. These barriers could be further evaluated prospectively with the implementation of similar laboratory IT. Additionally, an effort to develop a conceptual model specific for laboratory IT could enhance further implementation research.