Why don't physicians test for HIV? A review of the US literature

Abstract

Objective: 

In its 2006 HIV testing guidelines, the Centers for Disease Control and Prevention (CDC) recommended routine testing in all US medical settings. Given that many physicians do not routinely test for HIV, the objective of this study was to summarize our current understanding of why US physicians do not offer HIV testing.

Design: 

A comprehensive review of the published and unpublished literature on HIV testing barriers was conducted.

Methods: 

A literature search was conducted in Pubmed using defined search terms. Other sources included Google, recent conference abstracts, and experts in the field. Studies were divided into three categories: prenatal; emergency department; and other medical settings. These categories were chosen because of differences in physician training, practice environment, and patient populations. Barriers identified in these sources were summarized separately for the three practice settings and compared.

Results: 

Forty-one barriers were identified from 17 reports. Twenty-four barriers were named in the prenatal setting, 20 in the emergency department setting, and 23 in other medical settings. Eight barriers were identified in all three categories: insufficient time; burdensome consent process; lack of knowledge/training; lack of patient acceptance; pretest counselling requirements; competing priorities; and inadequate reimbursement.

Conclusion: 

US physicians experience many policy-based, logistical, and educational barriers to HIV testing. Although some barriers are exclusive to the practice setting studied, substantial overlap was found across practice settings. Some or all of these barriers must be addressed before the CDC recommendation for routine HIV testing can be realized in all US medical settings.

Introduction

The Centers for Disease Control and Prevention (CDC) estimated that 925 000–1 185 000 individuals were living with HIV or AIDS during 2003 in the United States ^[1]. Approximately one quarter of these individuals were, however, unaware of their HIV-positive status, many because they had not been HIV tested ^[1]. Many individuals unaware of their infection remain undiagnosed until they present with an AIDS-related condition or a low CD4 cell count ^[2–4], when treatment is less effective and health outcomes less favourable ^[2,5]. Furthermore, individuals with undiagnosed HIV infection are more likely to transmit the virus than individuals with diagnosed HIV, because HIV risk behaviors are reduced after an individual learns their HIV-positive status ^[6].

The high proportion of undiagnosed and late diagnosed HIV is a reflection of insufficient HIV testing. In the United States in 2002, 38% of adults reported ever being tested for HIV and only 10% had been tested in the past 12 months ^[7]. In 2005, 53% of New York City residents aged 18 years and older reported ever being tested for HIV (New York City Community Health Survey, 2005, unpublished data) despite the 1993 and 2001 recommendations of the CDC for routine screening in medical settings of individuals aged 13–64 years in areas such as New York City where HIV seroprevalence exceeds 1% ^[8,9].

Because many individuals with HIV infection remain undiagnosed, the CDC recently broadened its HIV testing guidelines to include routine HIV screening for all individuals aged 13–64 years in all US healthcare settings ^[10]. For the first time, an ‘opt-out’ model of testing was recommended to facilitate implementation of the guidelines, in which patients are informed that an HIV test will be performed unless they decline ^[10]. Without this and other changes, however, the CDC 2006 recommendations are not likely to result in widespread routine HIV testing given the experience following the earlier CDC recommendations for routine testing in high prevalence settings.

Barriers to HIV testing potentially exist at both the patient level and at the provider level, but existing literature suggests that patients are largely accepting of HIV testing ^[11]. Therefore, much of the failure to expand testing despite guidelines appears to stem from physicians' hesitancy to test. Understanding why physicians do not offer HIV testing routinely will prove critical in the successful implementation of the new CDC recommendations. In addition, it is vital to characterize barriers to testing by different practice settings because of differences in physician training, practice environment, and patient populations. Guidelines for prenatal testing have historically been different, with recommendations dating back to 1995 for all pregnant women to undergo counselling and voluntary HIV testing ^[12]. Similarly, past CDC recommendations have focused specifically on emergency department settings, making this an important location to examine separately ^[8,9]. Therefore, we conducted a literature review to identify physician barriers to offering an HIV test, and to determine if these barriers differ among different practice settings.

Methods

A comprehensive search of published and unpublished research was conducted. Pubmed was the primary search service used, with Google also used to broaden the types of sources collected. In addition, the 2006 International AIDS Conference and 2006 Conference on Retroviruses and Opportunistic Infections programme abstracts were reviewed for relevant US studies. Once sources were identified, the reference lists of each were examined for additional sources. Finally, researchers studying physician practices and attitudes towards HIV testing were contacted on the basis of referral from CDC staff to gather research the literature search might not have uncovered.

Only sources that addressed physician testing barriers were included. Sources that only examined patient barriers were excluded, as were studies that described institutional and physician characteristics associated with higher testing rates, and studies conducted outside the United States. We chose to exclude non-US studies because of differences in healthcare systems and policies between the United States and other countries and because of our interest in obtaining information needed to implement successfully the new CDC HIV testing recommendations for the United States.

Any study with evidence addressing physician barriers was included, regardless of source (e.g. published or unpublished), methodology (e.g. poor response rate or weak study design), practice setting, or type. This was done in an effort to gather all existing evidence for all barriers. All studies were categorized into one of three practice settings: prenatal; emergency department; or other medical settings. The other medical settings classification served as a catchall for all settings but prenatal or emergency department (e.g. internal medicine, sexually transmitted disease clinics, adolescent clinics).

Each barrier identified in a study was abstracted. No barrier was excluded in the summary because of a low reported frequency. In general, no attempt was made to change or collapse barriers across studies, except when it was clear that the same barrier was being named in different studies (e.g. insufficient time and not enough time). In addition, lack of space, staff, or money were combined and relabelled ‘lack of resources’ because of the inherent relationship between the three.

Results

We identified 17 studies that ascertained a total of 41 physician barriers to HIV testing. Table 1 presents a summary of each study, including the study design and population, response rate, and how barriers were ascertained. A supplementary table with a complete summary of study results including all barriers identified and the proportion of respondents that reported the barrier can be found on-line (http://aidsonline.com).

Table 1: Summary of studies included in the review by practice setting.

Prenatal providers

Nine studies in the prenatal setting were identified ^[13–21]. Six of the studies asked respondents specifically to identify barriers to HIV testing ^{[13,14,17,19–21]}. Two others elicited barriers on HIV education and counselling ^[15,16]. Segal ^[18] conducted a survey to determine physician attitudes on mandatory HIV testing in pregnant women, and gleaned information on barriers from supplementary comments submitted with the survey.

Twenty-four barriers were cited in the prenatal setting. The most frequently cited barrier was insufficient time, with seven of nine studies naming it as a barrier ^[13–18,21]. In those seven studies, 2–82% of responding study participants identified it as a barrier. The next most frequently cited barrier (identified in six of the studies) was the perception that the patient population was low risk (13–56% of respondents) ^{[13,15–17,20,21]}. Consent process was cited in five of the studies (3–90% of respondents) ^{[13,14,19–21]}.

Five other prenatal testing barriers were cited in four studies: provider lack of knowledge/training (27–82%) ^[15–17,21]; lack of patient acceptance (4–84% of respondents) ^{[14,16,19,21]}; fear/concern of offending the patient (5–82% of respondents) ^[16,19–21]; concern about treating an HIV-positive patient (1–82% of respondents) ^[16,19–21]; and pretest counselling requirements (3–90% of respondents) ^[14,19–21]. Two barriers were cited in three studies: late entry into prenatal care (13–84%) ^[19–21]; and concern about informing an HIV-positive patient (1–82%) ^[19–21]. Four barriers were also named in only two studies: language (15 and 92%) ^[20,21]; lack of resources (29 and 69%) ^[15,17]; low prevalence/incidence (17 and 65%) ^[14,16]; and inadequate reimbursement (2% and 79%) ^[20,21]. An additional 10 barriers were named in only one study, which are listed in Fig. 1.

Fig. 1: Venn diagram of barriers by practice setting. ACOG, American College of Obstetricians and Gynecologists; ED, emergency department; STD, sexually transmitted disease.

Emergency department providers

Three reports were identified in the emergency department setting as shown in Table 1^[22–24]. One report is a review article addressing the feasibility of HIV counselling, testing and referral in the emergency department setting ^[24]. In that article, Rothman ^[24] cited barriers from an unpublished work that was not found in the current review. That particular study described a focus group of 21 emergency department physicians representing a variety of locales and types of hospitals. Hardwicke and Malecha ^[23] presented a list of barriers identified by medical personnel working in emergency departments. Fincher-Mergi et al.^[22] conducted a survey in which barriers to testing were ascertained only from those respondents not offering HIV testing to all emergency department patients presenting with a sexually transmitted infection.

In all, 20 barriers were described, with only one source providing point estimates ^[22]. Concern about patient follow-up, insufficient time to test, and patient confidentiality concerns were cited in all three reports (with point estimates of 51, 19, and 17%, respectively). Three other barriers were cited by two reports: pretest counselling requirements (45% of respondents) ^[22,23]; language barriers ^[23,24]; and cultural barriers ^[23,24]. An additional 14 barriers were cited in only one study, which are listed in Fig. 1.

Providers in other medical settings

The review of other medical settings provided five reports (Futterman et al., unpublished data; Goetz et al., unpublished data and Jain et al., unpublished data ^[25,26] (Table 1). Futterman et al. (unpublished data) conducted in-depth phone interviews with healthcare providers in high-risk communities. These healthcare providers were encouraged to share what barriers they encountered when trying to incorporate HIV testing into their practice. Three studies conducted surveys of providers in which barriers were obtained (Goetz et al., unpublished data; Jain et al., unpublished data) ^[26]. Finally, the study conducted by Myers et al. was a survey to ascertain whether a separate written consent form was a barrier to testing; no information was collected on other barriers ^[25].

Those five reports named a total of 23 barriers. Insufficient time was cited as a barrier in four of the five studies (55 and 71% of respondents; Futterman et al., unpublished data; Goetz et al., unpublished data and Jain et al., unpublished data ^[25]). Provider lack of knowledge/training (13 and 45% of respondents) and patient confidentiality concerns (13 and 42% of respondents) were cited in three of the five studies (Futterman et al., unpublished data; Jain et al., unpublished data) ^[26]. Three barriers were cited by two studies: fear/concern of offending the patient (one point estimate of 26%; Futterman et al., unpublished data; Jain et al., unpublished data); low-risk patient population (one point estimate of 2%; Goetz et al., unpublished data; Jain et al., unpublished data); and posttest counselling requirements (no point estimates; Futterman et al., unpublished data; Goetz et al., unpublished data). An additional 17 barriers were cited in only one study, which are listed in Fig. 1.

Similarities and differences across practice settings

Figure 1 summarizes the barriers identified in the review. The eight barriers named in all three practice categories are listed in the center area, including insufficient time, consent process, lack of knowledge/training, language, lack of patient acceptance, pretest counselling requirements, competing priorities, and inadequate reimbursement. Other barriers, such as informing an HIV-positive patient, institutional costs, having a low-risk patient population, fear/concern of offending the patient, patient confidentiality concerns, posttest counselling requirements, cultural barriers, concern about patient follow-up, lack of HIV-related referral networks, and testing not considered appropriate were cited by two but not all three categories.

Discussion

We performed a comprehensive review to identify barriers commonly cited by physicians as a reason for not testing patients for HIV. Overall, the body of literature on this subject was lacking; many studies had high refusal rates, substantial methodological limitations, and less than half were published in the peer-reviewed literature. Despite these limitations and the differences in the three physician groups, a core group of identified barriers emerged (Fig. 1). These can be divided into policy level barriers, logistical barriers, and educational barriers (Table 2). Whereas some of the barriers identified are universal to any prevention-oriented healthcare activity (e.g. insufficient time and competing priorities), many are specific to HIV (e.g. burdensome consent process and pretest counselling requirements).

Table 2: Potential interventions to reduce the eight core physician barriers to routine HIV testing.

Policy level barriers include the consent process, pretest counselling requirements, and inadequate reimbursement. Separate written consent and pretest counselling were standards established at a time when HIV was untreatable and highly stigmatized. HIV is now treatable and the stigma has decreased. Although separate consent and pretest counselling were once necessary, they now appear to be two of the major barriers to a physician offering an HIV test. It might appear that one solution to overcoming these barriers would be to have more staff dedicated to HIV testing and counselling. Parallel clinical staff for HIV testing have been in place for more than a decade, but the large number of infected but undiagnosed individuals remains. It is therefore evident that many at-risk individuals are not reached by traditional counselling and testing centers.

There are other options to address the policy barriers identified in all three practice settings, as shown in Table 2. Streamlined pretest counselling has been shown to increase testing three to fivefold in the Veterans Health Administration hospitals (Goetz et al., unpublished data). Similarly, Futterman et al.^[27] found a 133% increase in testing rates among community health centers incorporating a streamlined process for counselling and testing. In New York City, some clinics have implemented the screening of a video in the waiting room designed to address pretest counselling requirements, shifting the burden away from the physician ^[28]. To alleviate the barrier of the burdensome consent process, oral consent or opt-out consent can be adopted ^[10]. Stringer et al.^[29] showed testing rates significantly increased in the prenatal setting when they began notifying all women that an HIV test would be performed as part of their routine battery of prenatal tests, and introduced a specific consent that needed to be signed to decline an HIV test. In New York, there is a legislative push to change the informed consent requirement for HIV testing from separate written consent to documented oral consent. Finally, to address inadequate reimbursement, New York State Department of Health implemented increased reimbursement for HIV testing in emergency department settings for Medicaid participants.

Logistical barriers to HIV testing include insufficient time, competing priorities, and language barriers. Insufficient time and competing priorities are issues that every physician faces and are not specific to HIV testing ^[30]. These barriers could be mitigated with some of the policy changes mentioned above. In addition, creating understandable and user-friendly materials for physicians to distribute to patients is a simple intervention to decrease the time burden. An American College of Obstetricians and Gynecologists study showed that giving prenatal providers a laminated fact sheet on HIV testing and a ‘prescription-like tear-off pad’ with information for their patients dramatically decreased the perceived problem of insufficient time among physicians ^[13]. A tool for overcoming competing priorities is establishing reminder systems for HIV testing, as has been established for other care areas such as immunizations and prescribing practices ^[31–34]. Such reminders could be incorporated into electronic medical records as clinical decision support systems to prompt physicians to order HIV testing when indicated.

Language barriers can prevent a patient from having access to HIV testing when translation services are inadequate. Language barriers can be addressed in multiple ways. Many states and local health departments provide consent forms and educational materials in numerous languages. Having these available in physician offices is necessary for physicians serving a diverse population. In addition, multilingual staff or on-site translators would help eliminate language barriers. When those resources are unavailable, healthcare clinics and organizations can contract with a telephone interpretation service.

The educational barriers identified in all three practice settings include lack of patient acceptance and lack of physician knowledge/training. Lack of patient acceptance is a perceived barrier because patient acceptance rates have actually been found to be as high as 100% in the prenatal setting and 91% in the hospital setting ^[11]. Physicians need to be informed that patients will be receptive to HIV testing if asked.

Lack of physician knowledge and training is a barrier that could be addressed with continued education of the medical community. HIV testing requirements and recommendations change over time, and it is the responsibility of the physician, health departments, and professional groups to disseminate this information appropriately. Educational materials and training on HIV testing could be available to physicians as are educational modules on infection control training and child abuse recognition.

The major limitation of this review derives from the sparse literature available. As a result, we decided to include all studies identified regardless of quality. Most of the traditional reports are from the prenatal literature, in which simplified, universal HIV screening with flexible informed consent was first recommended by the US Public Health Service in 1995 ^[12,35]. Therefore, the barriers to HIV testing have been better studied in that practice setting. Research on barriers in the other practice settings is limited and few studies have been published in peer-reviewed journals. An additional limitation arises from the nature of conducting a review of existing research. Interpretation of this summary is limited by the various data collection methods of the original studies and by the lack of clear descriptions in some studies of how barriers were ascertained from the respondents (Table 1).

The United States is far from being ready and able to implement fully routine HIV testing in medical settings. The 2006 CDC recommendations, with clear endorsement of opt-out testing, are an important step forward. An opt-out model with its reduced pretest counselling and oral consent for testing would address the barriers to testing most frequently identified in our review (i.e. insufficient time, consent process, and pretest counselling requirements). Because many of the identified barriers require policy level intervention, each jurisdiction will need to do a local assessment and develop its own plan to move towards this national goal. Action on some or all of these barriers will be needed to implement routine HIV testing fully throughout medical settings in the United States.

Sponsorship: This study was partly supported by an appointment to the Applied Epidemiology Fellowship Program administered by the Council of State and Territorial Epidemiologists (CSTE) and funded by the Centers for Disease Control and Prevention (CDC) cooperative agreement U60/CCU007277.