Diabetic retinopathy (DR) is a leading cause of vision loss in the United States that is preventable with periodic screenings. The VHA Tele-retinal Imaging Program is designed to facilitate access to quality diabetes eye care that complies with VHA guidelines for retinal screening examinations. The program is active in all 21 VISNs with more than 169,000 unique patients having imaging encounters in FY 2011. Despite its size and scope (>700,000 patient encounters since its inception in 2006), there is significant variability across VISNs in the uptake of the program, and the overall effectiveness of the program has not yet been assessed.
The objective of this RRP was to examine the variability in uptake of VHA's tele-retinal imaging program for diabetic retinopathy (DR) and to identify potential correlates of successful implementation. The RRP was exploratory in nature and investigated the following research questions:
(1)What is the variation across VISNs in the proportion of diabetic patients who undergo tele-retinal screening as well as the proportion of screened patients who receive follow-up care?
(2)Do VISNs with a higher proportion of tele-retinal screening encounters also have a higher proportion of eye-care follow-up for diabetic patients?
(3)What are the important program, contextual, individual, and process differences between VISNs with high screening and follow-up rates (high program uptake) and those with low rates (low program uptake)?
The design for this study was an observational analysis of retrospective quantitative data and of cross-sectional qualitative data. The quantitative analysis was used to calculate the following for each VISN:
-Proportion of diabetic patients without an eye care visit in the two years preceding implementation of tele-retinal imaging who were screened by the program in the two years following implementation.
-Proportion of patients screened who had a visit to a VA medical center or clinic for eye care in the year following their screening visit.
The calculated proportions were then used for dividing all VISNs into tertiles (low, medium, and high) according to their screening and follow-up rates. VISNs that fell into the following categories were selected for inclusion in the qualitative phase of the study: low screening, low follow-up; low screening, high follow-up; high screening, low follow-up; and high screening, high follow-up. A VA medical center (VAMC) within each VISN with screening and follow-up rates similar to the VISN's overall rates was selected for recruiting interview participants. All employees who agreed to participate were interviewed using a semi-structured interview guide based on constructs from the Consolidated Framework for Implementation Research (CFIR), which consists of factors that have been shown in previous research to affect implementation success. Interviews were recorded and transcribed verbatim.
The text in the interview transcripts was then reviewed by qualitative analysts, who were blinded to degree of program uptake at each VISN and facility. The analysts coded segments of text according to the CFIR constructs, and then assigned a rating to each of the implementation constructs at the VAMC level. These ratings (-2, -1, 0, 1, or 2) represent the respondents' perception of the construct as either a facilitator (positive) or barrier (negative) in implementation of the program. The ratings were then examined for their correlation with program uptake. Constructs with ratings that correlated with uptake formed the basis for recommendations to improve program dissemination.
The percentage of diabetic patients without an eye clinic visit in the two years prior to implementation of the tele-retinal imaging program who received a retinal screen in the two years following program implementation ranged from .9 % to 18.7% (a 21-fold difference), with an average screening rate of 9.4%. Rates of follow-up to an eye clinic for these patients in the year following screening ranged from 32.7% to 60.8% (a difference of nearly 2-fold), with an average follow-up rate of 46.3%. VISNs with high rates of screening did not necessarily have high rates of follow-up; and VISNs with low rates of screening did not necessarily have low rates of follow-up.
Nine VISNs fell into the four combinations of low/high screening/follow-up rates for inclusion in the qualitative phase of the study. Interview participants included: VISN program coordinators, readers, imagers, nurse managers, and various categories of facility level coordinators (e.g. Eye Care Coordinator, Primary Care Coordinator, Tele-Retinal Coordinator, Tele-Health Coordinator). Forty-two interviews across the 9 VISNs were conducted over the phone, lasting from 18 to 86 minutes.
Seven of the 39 CFIR constructs were rated as negative or neutral in the VISNs with low screening rates and low follow-up rates (low uptake), and rated more positively in the VISNs with high screening rates and high follow-up rates (high uptake). The interview responses for each of the constructs in the low and high uptake sites were further reviewed to identify particular program features that have implications for improving dissemination of the tele-imaging program, as summarized below:
External policies and incentives. High uptake sites viewed performance measures and provision of funds and cameras as important incentives. Low uptake sites thought they were already meeting performance measures, or didn't feel the pressure to meet the performance measures because of the large number of competing measures.
Networks and communications. Participants in high uptake sites gave specific examples of extensive and open networks and communications in the facility and throughout the VISN, such as regular conference calls between the readers, imagers, and VISN coordinator. Communication in low uptake sites occurs more infrequently on an ad hoc basis.
Organizational incentives and rewards. High uptake sites specifically recognize and reward their employees who are participating in the program and helping to achieve its objectives. Low uptake sites do not provide any recognition to the participants.
Learning climate. Participants in the high uptake sites provide examples of how their input is valued and acted upon. In contrast, participants in the low uptake sites say they have little input into the design and running of the program, and some provide specific examples of trying to make suggestions for program improvements, but then perceive their suggestions as being ignored.
Access to knowledge and information. High uptake sites cited several important sources of information, including annual conferences, monthly conference calls, a local master preceptor, and access to key people at the national level. Low uptake sites have found people at the national level to be less helpful, and are not aware of tele-health expertise at the VISN or facility levels. In addition, they are not participating in monthly conference calls, either because they are not aware these are occurring, no calls are scheduled, or because the respondents are unable attend.
Personal attributes. Key participants, especially imagers, at the high uptake sites are very competent and highly motivated to promote the program and to make it work. In contrast, participants at the low uptake sites are much less motivated and essentially participate in the program because they've been told this is part of their job.
Primary care engagement. Imagers at the high uptake sites have taken a very proactive role in engaging primary care providers in the screening program, including attending team meetings, seeking out individual providers who are not referring patients, and setting up a workable process for identifying and recruiting patients. In contrast at the low uptake sites, primary care providers have been reluctant to participate in the program, and no one has taken on the task of actively working with them to engage them.
Of note is that the other 32 CFIR constructs did not distinguish between low and high uptake sites, primarily because most of these other constructs received positive ratings (i.e., were viewed as facilitators) across all or most sites, regardless of program uptake. Thus, these other constructs may be necessary for successful implementation, but are not sufficient. The seven constructs described above deserve special attention for their role in contributing to program success.
The CFIR constructs associated with variability in implementation success of VHA's tele-retinal imaging program, along with qualitative data on the context in which these constructs have been operationalized, provide useful recommendations to VHA leadership for improving uptake of the tele-retinal imaging program, as well as for successfully expanding the program to include screening for other eye diseases. The findings also contribute to implementation science, by providing specific evidence for the role of CFIR constructs in implementation.
None at this time.
Diabetes, Screening, Telemedicine