Health information technology interventions enhance care completion, engagement in HIV care and treatment, and viral suppression among HIV-infected patients in publicly funded settings

Abstract

Background The National HIV/AIDS Strategy (NHAS) emphasizes the use of technology to facilitate coordination of comprehensive care for people with HIV. We examined the effect of six health information technology (HIT) interventions in a Ryan White-funded Special Projects of National Significance (SPNS) on care completion services, engagement in HIV care, and viral suppression.

Methods Interventions included use of surveillance data to identify out-of-care individuals, extending access to electronic health records to support service providers, use of electronic laboratory ordering and prescribing, and development of a patient portal. Data from a sample of electronic patient records from each site were analyzed to assess changes in utilization of comprehensive care (prevention screening, support service utilization), engagement in primary HIV medical care (receipt of services and use of antiretroviral therapy), and viral suppression. We used weighted generalized estimating equations to estimate outcomes while accounting for the unequal contribution of data and differences in the distribution of patient characteristics across sites and over time.

Results We observed statistically significant changes in the desired direction in comprehensive care utilization and engagement in primary care outcomes targeted by each site. Five of six sites experienced statistically significant increases in viral suppression.

Discussion These results provide additional support for the use of HIT as a valuable tool for achieving the NHAS goal of providing comprehensive care for all people living with HIV. HIT has the potential to increase utilization of services, improve health outcomes for people with HIV, and reduce community viral load and subsequent transmission of HIV.

Introduction

In 2010, President Barack Obama introduced the USA National HIV/AIDS Strategy (NHAS). Its goals are to: (1) reduce the number of people who become infected with HIV; (2) increase access to care and optimize health outcomes for people living with HIV; (3) reduce HIV-related health disparities; and (4) achieve a more coordinated national response to the HIV epidemic.¹ The strategy articulates measurable action steps and sets 5-year quantitative targets based on evidence-based approaches to achieving the goals it outlines. The explicit intent of the strategy is to provide a roadmap for responding to the epidemic among all public and private stakeholders; its guiding principles include accountability and science-driven decision making.

The NHAS can require high standards because there is now strong evidence that the viral suppression that comes with early, consistent engagement in high-quality HIV care results in the best patient outcomes.² The NHAS contains targets to increase the proportion of all HIV-infected individuals in the USA who are aware of their serostatus, linked to care within 3 months of diagnosis, continuously engaged in HIV care, and ultimately achieve suppression of the virus.³ Achieving all of these targets would help end the epidemic because suppression of viral load is important for individual patient health and for the prevention of transmission of HIV in the population as a whole.⁴

Unfortunately, there are numerous barriers to optimal engagement in care.^5–12 They encompass structural challenges (eg, lack of housing), insufficient financing (eg, lack of insurance),^13,14 personal and cultural characteristics (eg, beliefs about the health system, racism) co-morbidities (eg, mental illness and substance abuse), stigma, fear of confidentiality violations, and healthcare provider attitudes.^5–24 Research suggests that both individual and system-level interventions have the potential to improve care engagement and the quality of HIV medical and support services.^1,3,25

To date, few studies have explored to what extent sharing patient information across geographically disparate HIV surveillance, primary care, and support service organizations can enhance linkage to care, retention and adherence to care and treatment, the quality of core and support services, as well as health outcomes for people with HIV. This process, however, is key to achieving aspects of the NHAS. For example, the strategy's implementation plan calls for enhanced collaboration among providers, improved assessments and measurements of health outcomes, and the use of community level measures (eg, community viral load) to track progress toward NHAS goals across populations in a region.²⁶ Effectively meeting these objectives requires technologies and practices that ensure that providers in the communities are coordinating the services that they deliver to patients.

Health information technology (HIT) interventions, including the electronic transfer of information between organizations (health information exchange), extending access to information to additional providers within organizations, and sharing of information with patients has the potential to enhance engagement in comprehensive HIV care. HIT interventions have been previously implemented in the context of other diseases to link public health surveillance programs to primary care services, laboratories and pharmacies to primary care, and primary and specialty care.^27–31 HIE has the potential to close many of the gaps that lead to sub-optimal care for people with HIV by, for example, improving the referral and tracking of patients among services, identifying patients who have fallen out of care in a health system, and allowing providers to coordinate services to ensure that each individual receives the comprehensive set of services that he or she needs.

To test the potential value of HIT for supporting better-coordinated care in publicly funded settings, the Health Resources and Services Administration (HRSA) sponsored an initiative to develop HIT interventions at clinical care sites serving patients eligible for the Ryan White HIV/AIDS Program (RWP). The RWP was enacted by Congress to ensure access to quality HIV care and treatment for those who cannot afford it, provision of support services (eg, transportation and housing) for those who experience challenges and/or obstacles in entering and remaining in care, and coordination of care (eg, medical case management) for those who have co-occurring conditions that impact the effectiveness of HIV care.^32,33 RWP providers typically deliver a constellation of services to support status awareness, linkage, and retention. Each of the demonstration sites in the initiative implemented systems that supported engagement in care by targeting gaps they had identified among patients in their clinical settings. This paper describes the outcomes of the initiative in terms of the aims that each of the sites set out to achieve.

Methods

Description of the initiative

One component of the RWP is the Special Projects of National Significance (SPNS) program, which provides grants to fund innovative models of care and supports the development of effective delivery systems for HIV care. In 2007, SPNS funded a 4-year initiative in six demonstration sites to enhance and evaluate existing health information electronic network systems for people living with HIV/AIDS in underserved communities. These six demonstration sites were: the Bronx-Lebanon Hospital Center in the Bronx, New York; the City of Paterson Department of Human Services, New Jersey; the Duke University Center for Health Policy in Durham, North Carolina; the Louisiana State University Health Services Center in New Orleans, Louisiana; the New York Presbyterian Hospital in New York; and the St. Mary Medical Center Foundation in Long Beach, California. Additionally, the University of California, San Francisco (UCSF) Center for AIDS Prevention Studies was funded as the Evaluation and Support Center (hereafter referred to as the ‘Center’) to conduct a cross-site evaluation of the HIT interventions and provide technical assistance and support to the demonstration sites. The protocol for the cross-site evaluation was approved by the Committee for Human Research at UCSF.

HIT interventions and targeted outcomes

Each of the six demonstration sites implemented one or more HIT interventions in order to facilitate comprehensive care and enhance engagement in HIV medical services. These interventions and the characteristics of patients within each demonstration site have been previously detailed,^34–36 but are briefly summarized here:

• Site 1 developed an electronic health record (EHR) with summary comparison reports that was shared across all health service providers. This shared record facilitated the development of a quality improvement framework in which activities were implemented to increase targeted prevention services. Health record alerts were also formulated to help providers identify and intervene with patients who had not received needed clinical services.

• Site 2 developed a structured patient summary within its EHR that included highlighted alerts to identify needed clinical services. The hospital also provided case managers with access to these patient summaries to facilitate coordination of support services and reinforce engagement in HIV care. This site targeted ‘high need patients,’ defined as patients who had detectable viral loads despite multiple medical and social support service visits across multiple social service providers. This site anticipated that coordination of care would result in utilization of fewer redundant support services.

• Site 3 provided support service providers with access to a regional medical center EHR to facilitate coordination of support services and reinforce engagement in HIV care. This site anticipated that coordination of care would result in higher utilization of needed support services.

• Site 4 created continuity of care patient summaries. Patients were then given access to this information through a patient portal to facilitate engagement in HIV care. Patients were also able to provide this information to external providers to facilitate coordination of services. This site anticipated that access to information would increase utilization of needed care and support services.

• Site 5 linked the state surveillance branch and EHRs in publicly funded health facilities. It created an alert whenever a patient known to the surveillance branch to be out-of-care for HIV presented for services in an emergency room and other (non-HIV) healthcare settings. Providers at the care site then acted on the alert to facilitate the patient's re-engagement in HIV care.

• Site 6 implemented HIT to automate electronic laboratory ordering and prescribing to reduce the time needed to access these services and to enhance engagement in HIV care.

The comprehensive care utilization and engagement in care outcomes targeted by the interventions are summarized in table 1.

Table 1:

Outcomes and desired direction targeted by HIT interventions

Bold/gray indicates directly targeted outcomes; light grey indicates indirectly targeted outcomes; + and – indicate the hypothesized direction of the relationship.

ART, antiretroviral therapy.

Four sites implemented HIT intervention directly targeted to enhance utilization of comprehensive care. These included one intervention (site 1) to increase preventive screenings, two interventions (sites 2 and 3) to increase access to information among support service providers to facilitate the coordination of support services, and one intervention (site 4) to increase access to information among patients to facilitate coordination of care and support services. All six sites implemented HIT interventions that were intended to either directly or indirectly increase engagement in HIV care. Two interventions directly targeted engagement in care. These included one intervention (site 5) that employed data from surveillance and other systems to link or re-engage HIV-infected individuals in care, and one site (site 6) that implemented automated prescribing and ordering to increase retention in HIV care and appropriate utilization of antiretroviral therapy. Four interventions indirectly targeted engagement in care. These included two interventions that created alerts for patients who had not received needed clinical services (sites 1 and 2), and three interventions that expanded access to information for support service providers (sites 2 and 3) and patients (site 4).

Data collection

The Center collected de-identified quantitative electronic patient record data from each demonstration site for the 6-month period preceding the implementation of the HIT intervention and also for each 6-month period thereafter through the end of the project. A serial cross-sectional study design was implemented whereby, after each data reporting period, the sites provided data from a simple random sample of at least 100 patients from their patient population. The dates of the first 6-month reporting period (the baseline) varied between sites due to different implementation timelines, and ranged from May 2008 to December 2009. As a result, by the end of the project date, the sites had submitted unequal numbers of periods of data; those with an earlier implementation date submitted more periods of data than those with a later implementation date. The present analyses are based on the combined data from all sites and all data reporting periods (N = 7354; range among sites of 1019–1490 across four to six reporting periods).

Measures

Agencies in the USA are required to submit to HRSA standardized data for each client who receives services supported by RWP funds. The information is recorded in the agency's annual Ryan White Services Report (RSR). The data elements collected for the present evaluation followed the data specifications of the RSR (http://hab.hrsa.gov/manageyourgrant/clientleveldata.html). However, to better capture the changes in patient outcomes at the sites, the data were collected at 6-month intervals instead of the RSR's usual 12-month frequency. Since the data were drawn exclusively from HIV care clinics, the entire sample was considered to be linked to HIV care. The subsets of the RSR's data elements utilized in the present analyses are explained below.

• Preventive care: Individual binary variables were created to capture which of six routine preventive screenings—syphilis, chlamydia, gonorrhea, alcohol and drug use, oral health, mental health—were performed for a client during each 6-month reporting period (0 = the screening was not performed in the period; 1 = the screening was performed in the period). These six variables were summed for each period to create an index outcome variable.

• Support services: Individual binary variables were created to capture which of 10 support services—non-medical case management, treatment adherence counseling, health education/risk reduction, psychosocial support, transportation, outreach, referral for health care/supportive services, housing assistance, emergency financial assistance, and food bank or home-delivered meals—the client had utilized during each 6-month reporting period (0 = the service was not utilized in the period; 1 = the service was utilized in the period). These variables were then summed for each period to create an index outcome variable.

• Engagement in HIV care: Engagement in HIV care was a binary outcome defined as at least one outpatient/ambulatory care visit or at least one laboratory test—CD4 or viral load—during a 6-month reporting period (0 = no visit or laboratory test in the period; 1 = at least one visit or laboratory test in the period).

• Prescription of antiretroviral therapy: A binary outcome variable was created to record if the client had been prescribed highly active antiretroviral therapy (HAART) at any time during a reporting period (0 = not prescribed HAART; 1 = prescribed HAART).

• Undetectable viral load: Undetectable viral load was defined as less than 75 copies/mL. A binary outcome variable was created to record if the client had undetectable viral load at the last screening during a reporting period (0 = viral load was detectable; 1 = viral load was undetectable).

Analyses

We used logistic and linear regression to assess changes over time in the delivery of preventive care (health screenings in the 6-month reporting period), utilization of support services, engagement in HIV care, prescription of antiretroviral therapy, and laboratory values reflecting undetectable viral load. Generalized estimating equations were modeled to estimate the linear trend in each outcome between baseline and the end of follow-up while controlling for correlation of data within sites. Final models were weighted to prevent the undue influence of data from sites with a higher number of records (weight assigned to a given site = 100/number of data records from that site). In addition, inverse probability weighting was employed to adjust for differences in the distribution of patient characteristics across sites and over time.

Results

Improvement in utilization of comprehensive care services

We observed statistically significant improvements between baseline and the end of follow-up in the utilization of targeted comprehensive care services in three of the four sites that targeted these outcomes (table 2).

Table 2:

Change in comprehensive care services between baseline and the end of follow-up by site

Bold typeface indicates values of p<0.05.

‘b’ indicates estimated values; dark gray shading indicates directly targeted outcome.

At site 1, the estimated mean number of preventive screenings since diagnosis increased from 1.5 to 2.8 and the estimated mean number of preventive screenings during the 6-month reporting period increased from 0.5 to 2.8. At site 2, the estimated mean number of support services utilized by high needs patients decreased from 1.6 to 0.3, while at site 3, the estimated mean number of support services utilized increased from 0.8 to 1.1. At site 4, we anticipated a significant increase in the estimated mean number of support services, but instead observed a significant decline in the estimated mean number of support services from 5.8 to 4.2.

Improvement in engagement in HIV care

We observed statistically significant increases between baseline and the end of follow-up in all targeted engagement in care outcomes (figure 1 and table 3).

Figure 1:

(A) Proportion of patients retained in care during each reporting period. (B) Proportion of patient prescribed antiretroviral therapy during each reporting period. (C) Proportion of patients with viral load <400 copies/mL during each reporting period.

Table 3:

Change in odds of engagement in HIV care between baseline and end of follow-up by site

Bold typeface indicates values of p<0.05.

Dark gray highlight indicates directly targeted outcome; light gray highlight indicates indirectly targeted outcome.

ART, antiretroviral therapy; VL, viral load.

Care engagement at site 5 increased from 65% to 83%. At site 6, the proportion of patients engaged in HIV care increased from 90% to 92%, while the proportion prescribed antiretroviral therapy increased from 73% to 92%. We observed less consistent results in care engagement outcomes indirectly targeted by the HIT interventions. Four sites indirectly targeted engagement in care and prescription of antiretroviral therapy by making patient health record data available to support services providers (sites 1, 2, 3, and 4) and/or providing alerts for patients who had not received needed clinical services (sites 1 and 2). Each of these sites showed a statistically significant decrease in the appropriate utilization of HIV care (from an estimated 99% to 96%, 98% to 89%, 99% to 96%, and 100% to 91%, respectively). When we examined prescription of antiretroviral therapy, only site 3 demonstrated a statistically significant increase in antiretroviral therapy prescription (from an estimated 74% to 89%). Site 4 showed a statistically significant decrease in the prescription of antiretroviral therapy (from an estimated 84% to 75%), while sites 1 and 2 showed no change in the prescription of antiretroviral therapy (from an estimated 89% to 88% and 83% to 83%, respectively).

Five of the six demonstration sites showed a statistically significant increase in undetectable viral load. The proportion of patients with an undetectable viral load during each reporting period increased from an estimated 53% to 68% at site 1, 48% to 63% at site 3, 22% to 28% at site 4, 3% to 11% at site 5, and 54% to 66% at site 6. Only site 2 showed a statistically significant decrease in the proportion of patients with an undetectable viral load (from an estimated 62% to 46%).

Discussion

The sites participating in this initiative employed HIT interventions to enhance HIV comprehensive care utilization and primary HIV medical care engagement outcomes. We observed a statistically significant improvement in outcomes directly targeted by the HIT intervention(s) in five of the six demonstration sites. These interventions included use of surveillance data to identify out-of-care individuals, use of electronic laboratory ordering and prescribing to facilitate appropriate utilization of HIV care and treatment, and providing EHR access to support service providers to facilitate coordination of care. The intervention health information that patients could share with providers (site 4) was not associated with improved utilization of HIV comprehensive care. Although previous literature has described implementation of HIT to link public health surveillance programs to primary care, laboratories and pharmacies to primary care, and primary to specialty care,^27–31 this is the first study to demonstrate an association between HIT interventions and engagement in HIV care.

We observed less consistent findings for outcomes that were not directly targeted by the HIT interventions. Several HIT interventions indirectly targeted retention in HIV care and appropriate antiretroviral therapy by providing access to patient health information to support service providers and/or creating alerts for patients who had not received needed services. Only one of these demonstration sites showed an increase in the appropriate utilization of antiretroviral therapy during the intervention period. These results highlight the importance of targeting specific outcomes for change. Each of these HIT interventions created a mechanism to share data across geographically distinct providers. However, to meaningfully change outcomes, providers must then utilize the new information. An HIT system appears to be most successful at altering HIV care engagement when it is paired with quality improvement goals that are specifically tied to linkage and retention outcomes.

The findings in this paper correspond well with related findings that we reported from parallel qualitative interviews conducted with providers at the sites.³⁶ Specifically, we observed differences in the degree to which clinic procedures changed as a result of the HIT interventions. At most sites, practices and procedures were reconfigured to make use of the information now available through the technologies. However, this was not universally true. For example, at site 4 the intervention did not have much influence on the behaviors of providers. Although patients made the information available through the portal, the providers found it challenging to incorporate this information into their workflows. This may explain in part why the intervention did not impact coordination of care. Similarly, at site 6, providers made use of much of the new technology, but did not always review electronic laboratory results, which may explain the failure at that site to impact certain screening related outcomes.

To our knowledge, this is the first time that HIT interventions have been associated with improved health outcomes for people with HIV, and notably with increased viral suppression, which is critical for optimal health outcomes. In fact, few previous studies have examined whether HIT interventions are associated with subsequent patient health outcomes and these focused only on elimination of errors to protect patient safety.³⁰ We observed statistically significant increases in the proportion of patients with undetectable viral load at five of the six demonstration sites after initiation of the HIT interventions. Patients at site 2 were significantly less likely to experience undetectable viral load after implementation of a shared patient summary. However, this is likely due to the fact that this site continued to recruit ‘high need patients’ throughout the initiative. Patients who had detectable viral load despite high utilization of medical and support services, were enrolled in the demonstration site project to facilitate better coordination of care. Once these individuals were able to achieve viral suppression, they graduated from the demonstration site program. Therefore, this site, by definition, did not expect to retain patients who experienced viral suppression.

The results of this initiative must be interpreted within a limited context. First, because this initiative consisted of six demonstration projects implemented at the health facility or system level, we are not able to isolate the HIT interventions implemented as part of this initiative from other interventions or temporal trends. We cannot rule out the possibility that events outside of the initiative contributed to the increased frequency of undetectable viral load. However, provider workflows and procedures changed substantially as a result of these interventions.³⁶ Therefore, we have reasons to believe that the interventions contributed to meaningful changes in clinical practices and other outcomes. Second, because the care and treatment of people with HIV is complex, we are not able to identify the specific mechanisms through which these HIT interventions effect viral suppression. However, in both cases, the consistency of the findings across sites suggests the success of these interventions and the need for additional, more rigorous research in this area.

These results suggest that HIT interventions may improve the delivery of directly targeted care completion outcomes within participating RWP settings, and also more distal patient health outcomes, including viral suppression. Our results provide additional support for the use of HIT as a valuable tool to meet the NHAS goal of providing comprehensive care and treatment for all people living with HIV in the USA.¹