The eight programs in this systematic review demonstrate that HIV screening can be initiated successfully in US PEDs. As an objective marker of success, six programs identified at least one patient with an HIV infection who otherwise would have gone undiagnosed, [13,14,15, 17,18,19]. Of these, five programs found a HIV seropositivity greater than 0.1% among those tested, [13, 14, 17,18,19] which is the CDC recommended threshold for conducting routine screening in a given healthcare setting [4]. In further defense of HIV screening in this setting, Mollen, et al. reported that 85% of PED providers surveyed supported HIV testing in the ED and 93% approved of the program implemented in the study [14]. Two other studies apart from the eight programs reviewed in this investigation also lend support for HIV screening in EDs. Of 114 adolescents (14–21 years-old) surveyed at a Philadelphia PED, 79% agreed with the statement that “Rapid HIV testing should be included in all trips to the ED” [20]. In Mehta et al.’s survey of 191 US PEM attending physicians, 84% indicated that they believed that ED HIV screening would increase HIV testing availability to adolescents [8].
Despite these successes, the PED HIV testing programs in this systematic review highlighted challenges that limited their effectiveness at identifying HIV cases, efficiency and reach. The proportions of adolescents and young adults approached for HIV screening ranged widely, although these estimates were not reported for all programs. This variation across programs is most likely related primarily to differences in methodology across the screening programs (i.e., who, what, when, where, and how of screening) and goals of the programs (i.e., why screening). A consequent common limitation was that many patients were not approached for screening, even if intended to be “routine” or “universal”, and despite using an “opt-out” approach. Furthermore, some screening practices resulted in differences in testing based on age, race and gender.
These differences in patient selection for screening may have contributed to different HIV positivity rates across EDs. Notably, two of the sites with HIV positivity below the CDC recommended threshold for screening at 0% and 0.06%. These sites had the nontargeted approach to screening- any age-eligible patient was offered testing, as opposed to the targeted programs which selected patients deemed to have higher likelihood to have HIV given their ED presentation for possible STI. The combination of more generalized risk in the denominator of patients tested at these sites, and also low levels of testing actually offered due to low clinician compliance (older patients were more frequently offered at one site, clinicians needed to collected written consent at another site) may have contributed to the low yields. Given the low rates of offering testing at these sites, it is uncertain whether the nontargeted approach is necessarily suboptimal in the PED and merits further research.
Consent processes also were a source of variation, which in turn could have affected screening acceptance. CDC recommended removing the barrier of separate, specific written consent for HIV testing in 2006 [4]. This landmark change in recommendations for HIV testing approaches may have contributed to the increase in opt-out HIV testing seen in three of the studies this review captured after 2006. In the studies with the opt-in approach, acceptance rates were slightly lower (49.4%, 60.7%, 74.7%) compared to the opt-out approach (42.7%, 72.6%, 86.7%). The CDC consent change also may have impacted the need to obtain parental consent in certain states, though the effect of parental consent on screening acceptance is unclear. In a study conducted before the recommendation, Mehta, et al. noted that requiring parental consent resulted in some patients not being tested [13]. In contrast, Rakhmanina, et al.’s study observed that parents/guardians accompanying the patient was associated with lower odds of opting-out of HIV screening [17]. However, the investigators noted that their presence affected patient disclosure of prior HIV testing and risk-taking behavior, which might have affected HIV testing. Further, 6% of adolescents in their study who had accepted testing were not tested because parents/guardians declined it. In contrast, Gutman, et al.’s study did not find any association of HIV testing acceptance with parent/guardian presence [19].
A common source of missed testing opportunities was due to placing the onus of offering testing on PED providers, who perhaps either lacked awareness of the screening program, forgot to offer screening to eligible patients, or did not have time to offer testing. For example, Minniear, et al. reported that only 22% of PED providers surveyed prior to implementation of the screening program were aware of the revised CDC guidelines on HIV testing, and all PED providers frequently forgot to offer testing to patients, with some providers disclosing that they felt uncomfortable offering screening [15]. However, solutions to this problem are possible, as evidenced by Bhatt, et al. showing that computerized prompts in the EHR resulted in a large increase in provider ordering of HIV testing [18]. As an alternative arrangement to PED providers initiating testing, some programs reported using dedicated research assistants, [13] health educators, [14] or nurses, [15] for identifying, approaching, testing and counseling patients.
HIV testing modalities varied across the studies, partly related to the continuing evolution in tests available: phlebotomy for conventional (laboratory-based) antibody testing; [12] oral fluid sampling for conventional (laboratory-based) antibody testing; [13, 14] rapid, oral fluid, point-of-care testing; [15, 17] fingerstick sampling for rapid, point-of-care testing; [16, 18] or fingerstick or phlebotomy sampling for 4th generation antigen/antibody testing [19]. The types of sampling techniques and tests employed impacted who could be tested and when tests results were received, and consequently when they could be acted upon for linkage to care. Earlier studies did not have access to the multitude of tests available currently for PED HIV testing programs, which permit not only selection based on product testing performance characteristics and price, but also according to sampling techniques, testing time, accuracy, turn-around time, and testing location (in the ED or laboratory based). Test choice in the eight studies in this review likely was decided by the funding source, and thus might not have been the optimal choice for that PED. When selection of testing was possible, there were trade-offs of ability to deploy the tests with accuracy and turn-around time.
Illustrating trade-offs in testing techniques is Gutman, et al.’s [19] program, for which blood sampling for the laboratory-based Architect HIV Ag/Ab Combo Test (Abbott Laboratories, Inc.) was either through phlebotomy or fingerstick sampling, which enabled testing for the large proportion of PED patients who did not undergo phlebotomy or have intravenous lines placed. The study authors indicated concerns about the accuracy of oral fluid sampling as a reason for using this test, as well as research suggesting that adolescents are more likely than adults to present with an acute HIV infection, which might go undetected using non-antigen tests [19]. However, a large fingerstick blood sample (400 µl) was required for this test, and sampling was performed by the study investigators, rather than ED nurses or ancillary staff. Due to the limitation of blood volume needed, 5.8% of samples had inadequate blood volumes obtained via fingerstick and could not be processed. Of the 462 patients who declined screening, 10.6% cited wanting to “avoid needles” as a reason, although the results do not indicate if this was related to phlebotomy or fingerstick sampling. Because it was a laboratory-based test with a resultant greater turnaround time than point-of-care tests, 11.7% of results were available within one hour, 63.7% within one to two hours, and 24.6% more than two hours. As a consequence, only 20.8% of patients received their results during their ED stay.
Ignoring considerations about ability to test as many patients as possible and as accurately as able, Gutman, et al.’s [19] program and others reveal how test choice affects receipt of results. Although the two early studies (Mehta, et al. [13] and Mollen, et al. [14]) used oral fluid sampling, testing was conventional or laboratory-based and not performed in rapid manner, so results were not available during the ED visit. Appointments were required to receive test results in follow-up about two weeks after the ED visit. As a consequence, Mollen, et al.’s loss to follow-up was 42% [14]. Although none of the eight studies reviewed reported being unable to secure follow-up for patients with a positive HIV test result, this problem is ubiquitous. Point-of-care rapid testing or testing with rapid results eliminate follow up appointments for receipt of test results and thus enable easier linkage to care.
The limitations of and challenges faced by these eight PED HIV screening programs can direct efforts for their implementation and for research. Feasibility no longer needs to be demonstrated. Instead, research should be directed at improving the methods of these programs, including interventions to increase their reach or coverage of patients in the PED who need to be tested for HIV, increasing testing acceptance, facilitating the performance of testing, decreasing time to test results, and ensuring patient receipt of test results. Further, optimal strategies to implement the programs in PEDs of all types commensurate with their resources and needs, as well as enabling their sustainability require attention to enable widespread and long-term success.
Potential strategies to increase PEDS implementation of HIV screening, as identified in the research covered in this review, involve leveraging provider education, the team care approach, investing in testing technology and EHR aids. Provider education should involve educational campaigns to inform providers of the prevalence of HIV in adolescents, the ability to test in the PED with linkages to care, and importantly, specific training of how to engage in conversations about sexual health. The team care approach has demonstrated success in sharing some of the time burden of additional screening and counselling conversations to members of the care team. Nurses can assist in screening questions and at certain institutions, health educators or case managers can assist in sexual health counseling and linkages to care, respectively, and lastly physician assistants and nurse practitioners can aid in follow-up calls to patients for whom positive screens result after the ED visit. As medical technology advances, the time and accessibility of point-of-care STI tests improve. Health systems' prioritization of patients’ sexual health by investing in faster rapid tests will increase the numbers of patients who can receive results during their ED visit, potentially increasing EDs willingness to incorporate HIV screening programs. Finally, the EHR can facilitate HIV screens by choice architecture of how the clinician interacts with HER by incorporating HIV screening in STI order sets as a pre-selected choice, reminding providers who do not order HIV that it can be order. Together these strategies to improve HIV screening merit further implementation science-informed research as to their effectiveness.
Limitations
As true for this type of investigation, this systematic review is limited by what has been published on this topic. Most of the studies were demonstration or pilot programs of a short duration. This limitation therefore precluded any long-term analysis on their effectiveness and sustainability. Furthermore, most programs were single-center studies that were highly heterogeneous in design and methodology. As a result, only a qualitative review of the studies could be conducted. All articles were analyzed from the lens of current-day knowledge of HIV testing and practices, and are therefore reflect the biases, knowledge and perspectives of the reviewers. In addition, all studies were focused in the United States, and findings may not apply to testing being performed in other countries. Although we performed manual searches of journal websites in addition to our database searches, there likely were journals we either did not search or failed to record that we had searched them. The former concern might have resulted in missed articles, although the likelihood of this occurrence is small given the database search. This potential effect on the analysis cannot be known.