Step 1: Priority Condition/Issue: Is undiagnosed HIV infection a high-risk, high-volume clinical issue within the VA?
The observation by the CDC that 25% of HIV-infected patients in the United States do not know their status is of great relevance to the VA [1
] because the VA treats more than 20,000 veterans with HIV per year [1
]. If, as in the rest of the country, 25% of HIV-infected VA patients do not know their status, approximately 7,000 veterans are at risk of being diagnosed with HIV and treated only when they become symptomatic and severely immunosuppressed. Thus, since HIV patients benefit greatly from early diagnosis and treatment [2
], increased HIV testing in the VA can substantially reduce the number of newly diagnosed veterans who present with concurrent complications of profound immunodeficiency [3
] and extend survival for the average HIV-infected patient by 1.5 years [2
The decision by the HIV/Hepatitis QUERI Center to focus on HIV testing as a priority within HIV quality improvement was based on the above observations. This decision was approved by the nationally constituted executive committee of the HIV/Hepatitis QUERI Center. QUERI processes require that this prioritization needs to be re-justified on an annual basis through the development of a strategic plan that is reviewed and approved by VA/HSR&D Central Office appointees, representing both researchers and clinical management.
Step 2: Evidence-based Practices: Are there evidence-based recommendations for HIV testing?
The U.S. Preventive Services Task Force (USPSTF) gives a Grade A recommendation to HIV screening for all adolescents and adults who are at increased risk for HIV infection or who receive health care in a high-prevalence setting, such as where ≥ 1% of the patient population is known to be HIV-infected (Table ) [15
]. The VA has endorsed the USPSTF guidelines for HIV infection and has identified testing for HIV as being a high priority [16
US Preventive Services Task Force Grade A Recommendations for HIV Screening*
The cost-effectiveness of HIV testing in the VA is well established. As shown by a HIV/Hepatitis QUERI-affiliated investigator and others, for a population with a prevalence of HIV infection of ≥ 1%, the cost of one-time screening for HIV infection is $15,078 per quality-adjusted life-year gained [2
]. Analyses that consider the relationship of diagnosis and treatment on HIV transmission show that the cost of routine HIV screening is < $50,000 per quality-adjusted life-year gained, unless the prevalence of HIV infection is <0.05% [2
]. This level of cost effectiveness, which matches that of many well-accepted procedures such as performing colonoscopy for colorectal screening [17
], provides a strong argument in favor of implementing HIV testing programs, especially as VA patients have rates of HIV much higher than the 0.05% lower bound of cost-effectiveness [18
Step 3: Quality/Performance Gaps: Are there gaps in the performance of HIV testing in the VA?
Previous studies done by HIV/Hepatitis QUERI, the VA Public Health Strategic Healthcare Group (PHSHG, a dedicated national VA program office for guiding HIV care services), and other VA groups have shown that only 30 – 50% of VA patients with known, documented risk factors have undergone HIV testing [11
]. Furthermore, at the time of HIV diagnosis half of the veterans have advanced levels of immune suppression. These veterans have, on average, 3.7 years of VA care before their HIV is diagnosed [19
To confirm and extend these data, we evaluated the rates of HIV testing in veterans seen in the five southern California and Nevada VA facilities (otherwise referred to as Veterans Integrated Service Network 22 or VISN 22). We found that between January 1999 and December 2004 only 30% of the 45,776 at-risk veterans (i.e., VA patients with positive laboratory tests or diagnostic codes for hepatitis, sexually transmitted diseases, and/or substance abuse) had been tested for HIV infection. The rate of testing for at-risk patients ranged from a low of 8% in primary care clinics to a high of 47% in substance abuse clinics. The low testing rates and large number of at-risk veterans in VA primary care clinics (nearly ten times as great as in substance use clinics due to the vast differences in the size of these clinic populations) pointed to the need for an intervention to focus on improvements in HIV testing performance in VA primary care clinics. To better understand the source of gaps in care and to discern facilitators that would improve current practice, we reviewed VA policies regarding HIV testing and surveyed providers' practices and attitudes regarding HIV testing at two VA facilities. Following, we present these findings.
Public Law 100–322 requires that VA patients provide voluntary informed consent for HIV testing and that providers document pre- and post-test counseling [16
]. In addition, many VA providers regard HIV testing and pre- and post-counseling to be the sole provenance of specially trained HIV counselors [20
]. Furthermore, standard policy has been to require all patients with positive or negative test results to come back to clinic for face-to-face, post-test counseling. The post-test counseling appointment is problematic, as many VA providers do not have sufficient appointment slots to allow for timely in-person patient notification of test results (i.e., within 1–2 weeks of the test).
Provider willingness and ability to perform HIV testing
Surveys of a convenience sample of 30 VA primary care providers indicated that lack of knowledge of individuals' risk factors for HIV infection, the time requirements to fulfill necessary counseling processes [20
], and anxiety about post-test counseling patients who have positive test results [14
] were substantive barriers to ordering HIV tests.
Patient acceptance of HIV testing
A systematic review of 62 studies found that acceptance rates of voluntary HIV testing in the United Sates varies from 11% to 91% [23
]. Importantly, this review found that higher acceptance rates were associated with confidentiality protections (strongly upheld by VA policy and procedures), as well as the provider's belief that testing would be beneficial.
Step 4: Identify and implement interventions to promote best practices
Here we describe our QUERI step 4 activities in further detail. We are now nearing the end of phase 1 of step 4, wherein we are conducting a two-station (two-site) pilot project.
Collaborating with clinical services to design an appropriate intervention
We explicitly sought broad institutional support for this project. This included the support of the VA PHSHG, and the VISN 22 Director, Quality Improvement Council, Clinical Practices Council, and the Clinical Performance Committee. VISN 22 leadership agreed to make HIV testing a performance monitor, to support installation of the HIV Testing Clinical Reminder, and to participate as full partners in enhancing station accountability. After obtaining national and regional support, we also made presentations to, and received support from the Medical Executive Committees, Chiefs of Staff, Ambulatory Care Leadership, primary care teams, and the HIV Coordinators at our two intervention stations. Receipt of all this support was greatly facilitated by the products generated from QUERI steps 1–3.
Development and initiation of a program to improve rates of HIV testing
Conceptual basis of the implementation interventions
As required by the QUERI process, we paid careful attention to the selection of a quality improvement model upon which to base our intervention program. We elected to base our program upon the Chronic Care Model
(CCM). This decision was based on the previous, wide success of interventions based on the CCM precepts to improve clinical preventive care services [24
]. Further guidance for this implementation strategy was provided by Rogers, whose germinal work on diffusion promotes the use of opinion leaders or champions to facilitate change or innovation adoption, and highlights the importance of individuals' social networks, organizational leadership and structure [26
Key components of the CCM include a clear definition of optimal care and enumeration of targeted patients, i.e., offering HIV-testing to at-risk patients; a road map for changing the system; and an effective improvement strategy [27
]. Effective CCM implementation strategies also contain the following elements: decision support, clinical information systems, delivery system design, and patient self management [24
]. Therefore, we developed an implementation strategy that uses: clinical reminders to provide decision support
, audit/feedback as a clinical information system
, and organizational change to achieve an appropriate delivery system design
. These activities are complemented by academic-detailing and social marketing interventions to achieve provider activation
to ensure that providers have the skills and motivation to improve their performance [24
]. Both the CCM and the Institute for Healthcare Improvements Breakthrough Series have identified the need for provider activities to transform and sustain changes in group norms [24
]. Thus, we chose to implement a multi-faceted provider activation program that includes academic detailing
and social marketing
]. Finally, we promoted HIV counseling to increase patient self-management
by wide scale publicity of the HIV testing program in clinic waiting rooms and check-in areas.
Table summarizes the relationships between the barriers to HIV testing, the content of the planned implementation program, and the relationship of each intervention to the elements of our CCM-based implementation strategy, which was augmented by academic-detailing and social-marketing interventions (provider activation). In the following section, we discuss the components of these interventions and the methods of delivery in more detail.
Relationship between Identified Barriers, Elements of the Chronic Care Model and Implementation Strategy
Components of the implementation program
To leverage institutional resources, we implemented a real-time, electronic clinical reminder that had been developed by the VA PHSHG to identify veterans at higher than average risk for HIV infection – and to encourage providers to offer HIV testing to such individuals (Figure ). Widely used to implement quality improvement, clinical reminders are well-suited for use in the VA because of the system-wide computerized patient record. The HIV Testing Clinical Reminder is triggered by prior evidence of infection by Hepatitis B or C infection, illicit drug use, sexually transmitted diseases (STDs), homelessness, and/or documented risk factors for Hepatitis C infection. All these data elements can be automatically extracted from the VA electronic medical record. The reminder is resolved (i.e., appropriately addressed) by: ordering an HIV test, recording the result of an HIV test performed elsewhere, indicating that the patient is not competent to consent to testing, or specifying that the patient refuses HIV testing. Once resolved, the reminder is no longer triggered.
The HIV Testing Clinical Reminder is activated whenever a patient with HIV risk factors, who has not been previously tested for HIV, has an encounter with a healthcare provider. The format of the Reminder is as pictured.
The use of clinical reminders in individual patients, when combined with audit/feedback and organizational changes, has been shown to improve performance of vaccination, cardiovascular risk reduction, and breast and colorectal cancer screenings [9
]. Electronic clinical reminders are a standard well-developed technology with which VA providers have great familiarity and have been shown to be well suited to improve performance of tasks similar in nature to HIV test ordering and counseling. Furthermore, our previous work has shown that use of clinical reminders contributes to 10–30% increases in the rates of appropriate clinical interventions in VA HIV-infected patients [9
]. Thus, this implementation intervention satisfies the FITT (fit between individuals, task and technology) framework for assessing the suitability of using this intervention in our strategy [34
Clinical information system
We designed an audit-feedback system, wherein healthcare providers are informed of group performance in regard to HIV screening rates in at-risk patients. A meta-analysis of 85 trials demonstrates that the use of audit-feedback is effective in improving practice, especially when baseline adherence is low [35
]. We have distributed audit-feedback reports (Figure ) to senior medical center-level clinical managers, outpatient clinic managers, and primary care team leaders at the Los Angeles and the San Diego VA stations. The contents of the reports have been discussed during academic detailing visits to primary care team meetings, and in the social marketing campaign. Informal provider feedback regarding the content of these reports has been positive.
Figure 2 Sample Audit/Feedback letter. The original report identifies the station and sub-stations by name rather than as "Station A" and sub-stations 1–11. The data provided in this letter indicate the number of persons with identified HIV risk factors (more ...)
We implemented a provider activation program that includes academic detailing and social marketing [25
]. This approach recognizes that the engagement of providers and the use of multiple modalities are necessary to achieve and sustain the transformation of group norms and maximize quality improvement [24
]. The goal of these activities was to engage providers and influence their attitudes, skills and habits regarding offering HIV testing to at-risk patients. We also used these interactions to reinforce more formal educational efforts. Particular attention was paid to: increasing providers' sense of responsibility for ordering HIV tests, emphasizing the use of streamlined pre-test counseling, reinforcing the use of telephonic notification of negative test results, and assuring providers of the availability of assistance for notification of persons with positive test results.
The academic detailing component of the project involves multiple presentations by clinical champions (physician and nursing staff), supplemented by project staff to the primary care team meetings and educational sessions. We have specifically targeted primary care clinic leaders as local, organizational opinion leaders [25
]. We used social marketing with providers to reinforce the importance of changing their practice regarding HIV testing, and further motivate them to do so. Social marketing entails the development of a shared buy-in to the overall goal of the behavior change and is predicated on social exchange theory
, which borrows from social and behavioral science doctrines in emphasizing the client/patient/provider as the starting point [37
]. The social marketing element includes regular informal discussions of the basis for and benefits of increased rates of HIV testing by project staff during frequent ad hoc visits to the primary care clinic and presentations to sub-station and clinic leadership.
As per the precepts of social marketing theory [37
], we have undertaken: audience segmentation
for focused detailing to nurses, mid-level providers and physicians; channel analysis
to optimize the setting and materials for these audience segments; goal orientation
to keep stakeholders focused on why they are involved (i.e., emphasizing the differing tasks by nurses [pre-test counseling ] and physicians [order entry]); and process tracking
to monitor progress and provide feedback for refinement and revision of strategies (i.e., through audit-feedback and through formative evaluation), which can be considered an intervention [38
Based on survey responses by physicians, mid-level providers, nurses, and case managers, we also have developed provider education materials to supplement the face-to-face training conducted by project staff that focus on preparing providers to use the reminders effectively, making providers aware of HIV risk factors not captured by the reminder (i.e., multiple unprotected sexual contacts), and increasing provider comfort and abilities to provide pre- and post-test HIV counseling.
In our step 4/phase 1 project, all social marketing has been performed by senior project staff. Using the insights gained in the two-station pilot project, as we progress down the pipeline to a step 4/phase 2 small-scale, multi-station evaluation we will rely on a train-the-trainer model to activate local champions. Project staff will support the local champions with regular visits (monthly for the first 3 months of local implementation, and then quarterly), weekly telephone conference calls and e-mail support.
Delivery System Changes
We have strived to ensure that all providers are trained to use a recent revision of the VA HIV Consent Form [39
]. This new document includes all the necessary elements of pre-test HIV counseling, and thus facilitates the consent process for healthcare providers who lack specific training regarding the performance of HIV counseling. We also have encouraged nurse-based rather than physician-based pre-test counseling [40
]. Nurses perform many educational, health promotion, and disease prevention tasks as well as physicians [41
]. Organizational changes that shift responsibilities from physicians to other personnel, most often nurses, are effective in improving preventive care. Moreover, we have encouraged providers to use a streamlined HIV counseling process that covers all the required elements of HIV pre-test counseling and reduces the time of pre-test counseling to 2–3 instead of 10–15 minutes, with some counseling lasting as long as 35 minutes [42
]. Similarly, we have reduced the logistical challenges of post-test HIV counseling [16
]. Given the gravity of the information, post-test counseling for persons with new positive HIV test results strongly warrants face-to-face counseling. In contrast, we have alerted providers that for patients with a negative result, post-test counseling can be very brief and can be done via the telephone [43
]. To ensure compliance with post-test counseling requirements, we have distributed sample scripts for transmitting the results of the test.
In addition to being theory-informed, these interventions are informed by empirical evidence provided by studies in urgent care clinics, emergency departments, and STD (sexually transmitted disease) clinics that show increased testing rates and patient receipt of test results after implementation of structural changes, such as improved staff training in pre- and post-testing screening, introduction of streamlined counseling, and substitution of telephonic post-test counseling in place of a required return visit for face-to-face notification [43
The intervention program is being put in place for one year in the primary care clinics of stations A and B in VISN 22. The three remaining stations in VISN 22 (stations C, D and E) served as controls. These facilities each provide care to 37,000 – 80,000 veterans per year. Facilities were assigned to the active or control arms by convenience. All facilities, except for one of the controls, consisted of an inpatient center plus one or more geographically dispersed sub-stations in which primary care and specialty services, including mental health and substance abuse treatment programs, were provided by academic staff physicians, post-graduate medical trainees, and mid-level providers. In addition, these facilities also provided primary care in other sub-stations staffed solely by non-academic physicians and mid-level providers. At the remaining control facility, care was provided only in outpatient sub-stations by providers who generally did not have an academic affiliation.
The decision support aspects of the intervention and policies regarding performance of HIV consenting and counseling have been implemented at all sub-stations and all clinics at the active facilities. However, the provider activation component of the program (i.e., academic-detailing and social marketing) has been fully implemented only in the primary care clinics at the two largest sub-stations at Facility A (out of a total of 11 sub-stations) and at the two largest sub-stations at Facility B (out of a total of 6 sub-stations); these sub-stations account for 46% and 69% of patients seen in primary care clinics at Facilities A and B, respectively. The other, smaller and geographically distant sub-stations differ in that e-mail and telephonic outreach largely replace personal outreach to promote academic detailing; all other tools (i.e., audit/feedback, provision of printed materials such as e-mail communications, pocket cards, posters and flyers, and removal of organizational barriers) are the same at all stations. The audit feedback program is directed at all providers in every primary care clinic.
The primary endpoints of this step 4/phase 1 two-station pilot project are the effect of the implementation program on the rates of resolution of the HIV clinical reminder and of HIV testing in patients with identified HIV risk factors. A multi-level, logistic regression analysis of the HIV testing rate will be done to adjust for the covariates at patient, provider, and sub-station levels – and for clustering.
We have obtained information regarding all inpatient and outpatient patient encounters within VISN 22 from a pre-established network database. For patients seen in outpatient clinics before and/or after the intervention, we obtained relevant laboratory tests, diagnosis codes, and health factors to determine if they were at increased risk for HIV.
Data regarding non-VA HIV testing, refusal of HIV testing, and incompetence for HIV testing were extracted from the standardized VA clinical reminder software package. Patients were defined as having been tested for HIV if there was documentation of HIV testing done within the VA healthcare system. Veterans were defined as having been evaluated for HIV if there was electronic documentation of prior HIV testing within the VA or elsewhere, patient's refusal to be tested, or patient's incompetence to consent for testing. Information regarding prior HIV testing within the VA was obtained through the VA electronic laboratory records, whereas information regarding outside testing, test refusal, and incompetence to consent was collected through responses to the HIV testing clinical reminder.
In addition to the data abstracted from the VISN 22 database, we also obtained a list of primary care providers classified into provider types (senior staff physicians, mid-level providers, physician assistants, and post-graduate medical trainees) from the primary care administration staff at Facilities A and B. The data on provider types were used to compare HIV testing and evaluation performance across different types of primary care providers.
As previously noted, we fully implemented the provider activation module only at the largest sub-stations at the intervention stations. This design allows us to assess whether this module, which is the most labor intensive component of our implementation strategy, independently contributes to improvement in the rates of HIV screening and testing. In addition, we are conducting formative evaluations to further refine our program and assess the organizational factors that determine the generalizability [38
]. The overall aim is to better understand the influences that have an impact on the success of the implementation program by identifying contextually relevant factors (i.e., facilitators and barriers) and assessing the degree to which behaviors that led to improved testing performance become part of routine practice [38
]. Semi-structured interviews with key informants will provide qualitative data regarding the effectiveness of the mode used for providing audit/feedback, the usefulness and usability of the testing reminder, and the efficiency of the consenting/counseling process. Interview questions will employ rapid ethnographic assessment methods to explore the ecological context of HIV testing [44
To assess the degree to which behaviors leading to improved testing performance are institutionalized (i.e., become embedded in standard operating procedures), leading clinicians and administrators at each sub-station will complete a Level of Institutionalization survey. The instrument measures four sub-systems that support routine use of an innovation: production
, where it must be integrated with other routine clinical services; maintenance
, where employees must support it; supportive
, where it must have a stable source of funding; and managerial
, where it must be assigned to a specific service, staff must have written job descriptions, and performance is required to be measured and reported [45
Finally, we are generating a comprehensive analysis of the workload and implementation costs of HIV screening and testing programs, using Business Case Modeling, a method for constructing data-driven models that forecast costs under varying specified conditions that support managerial or technical decision-making. This is warranted as the models of the cost-effectiveness of HIV screening do not address the upfront costs of implementing screening programs across differing clinic settings [2