Physicians in primary care settings, emergency departments, and sexually transmitted disease clinics must be prepared to evaluate, treat, and counsel patients with recent sexual exposure to HIV. Several tasks should be accomplished during the initial visit (Table 1). First, appropriate candidates for postexposure prophylaxis should be identified and offered antiretroviral treatment. Second, HIV testing must be performed to identify persons who are already infected and need long-term antiretroviral therapy. Finally, interventions to prevent future transmission of HIV must be initiated [8, 9].

Rationale for Postexposure Treatment

The theory underlying the use of postexposure prophylaxis is that antiretroviral treatment instituted immediately after exposure to HIV may abort infection by inhibiting local HIV replication and allowing the host's immune defenses to eradicate the virus inoculum. Although no direct evidence shows that postexposure treatment prevents infection after sexual exposure, this is biologically plausible given the efficacy of treatment after transcutaneous occupational exposure and the similarities between the immune responses to transcutaneous and transmucosal exposures [10-12].

Identification of Candidates for Treatment

Identification of candidates for postexposure treatment requires assessment of the type of sexual exposure, the HIV status of the patient and the patient's partner, the patient's attitude toward safer sex, and the length of time since the exposure (Table 1).

Type of Sexual Exposure

Few studies have assessed the per-episode risk for HIV infection with specific sexual practices [13-16]. The probability is highest with unprotected receptive anal intercourse (0.008 to 0.032) [13]. The risk is higher with receptive vaginal intercourse (0.0005 to 0.0015) than with insertive vaginal intercourse (0.0003 to 0.0009) [14-16]. No per-contact estimates of risk with insertive anal intercourse or with oral intercourse have been published, although seroconversion as a result of oral sex has been documented [17-19].

Even for sexual acts for which the average per-exposure risk has been quantified, the average risk may not apply to a specific encounter. In some cases, HIV infections result from few contacts; in others, HIV infection does not occur despite many contacts [15, 16]. Many source and host factors influence the likelihood of HIV infection [20].

HIV Status of the Patient

Preventive treatment is given on the basis of the assumption that the patient is not already HIV infected. Many persons at risk have never been tested [21, 22], however, and others will have had unsafe sex since 6 months before their last negative test result. For these reasons, an HIV antibody test should be done during the initial visit. With conventional HIV assays, most infected persons will have detectable antibodies 4 to 6 weeks after exposure and almost all will seroconvert by 6 months [23]. If the patient reports having had multiple exposures in the recent past, an HIV viral load test should be performed to detect primary HIV infection [19]. Testing for HIV should not prevent the immediate initiation of prophylaxis. Rapid HIV tests are extremely sensitive but are not specific enough to definitively diagnose pre-existing HIV infection [24, 25]. Patients found to be HIV infected at baseline should be prescribed an optimal drug regimen for primary or long-standing HIV infection [26].

Assessment of Sexual Partner (Source of Infection)

The HIV status of the sexual partner may not be known if he or she has never been tested or has engaged in unsafe behaviors since 6 months before the last negative test result. If the partner is willing to be tested, treatment can be started and then stopped if the partner's HIV antibody test result is negative. If the partner is unavailable or unwilling to be tested, treatment decisions should be based on the likelihood that the partner is HIV infected. Factors to consider include the partner's HIV risk behaviors and the prevalence of HIV and AIDS in the partner's community [1].

If the partner is HIV infected, additional information, including stage of illness and results of recent HIV viral load tests, can improve treatment decisions. Persons with advanced disease [2] and high viral loads are more likely to transmit HIV. Because the correlation between plasma and genital fluid viral titer varies [27, 28], transmission may still occur even if the source has a nondetectable serum viral load. A history of antiretroviral treatment can identify cases in which drug resistance is more likely so that the prophylactic regimen can be adjusted [29]. Care of the source patient should include counseling on how future risks can be avoided, both with the partner who has presented for care and with other partners.

Several probable cases of HIV transmission during rape have been reported [30, 31], and postexposure prophylaxis of HIV infection should always be considered and recommended when it is indicated [32, 33].

Patient Attitude toward Safer Sex

Many motivated persons diligently practice safer sex but are accidentally exposed to HIV when a condom breaks. Reported per-episode failure rates with condoms are high (0.9% to 4.5%) [34, 35]. In contrast, some persons may seek postexposure prophylaxis in the hope that it will enable them to maintain high-risk behaviors without being infected.

Between these two extremes, patients present with various risk histories. Some have an isolated episode of unsafe sex while high on substances. Others may have many unsafe exposures with partners of unknown HIV status but seek care because of exposure to HIV through a partner who is known to be infected. Because it is difficult to assess a patient's motivation for practicing safer sex in the short time frame necessary to initiate postexposure treatment, clinicians should err on the side of offering treatment for an exposure but should emphasize the importance of safer sexual behavior in the future.

Recommendations for Postexposure Prophylaxis

We recommend postexposure prophylaxis after unprotected receptive and insertive anal and vaginal intercourse with a partner who is or is likely to be HIV infected (Table 2). Prophylaxis should also be offered for receptive fellatio with ejaculation, although the lack of per-contact estimate for this behavior makes it more difficult to weigh the benefits against the risks. We do not recommend prophylaxis after other sexual exposures, such as cunnilingus or receptive fellatio without ejaculation, unless factors are present that increase the likelihood of transmission (such as exposure to menstrual blood). We recommend treating only isolated exposures or exposures in a patient who is willing to practice safer behaviors in the future. Finally, immediate treatment is optimal, and treatment should be initiated within 72 hours (the cut-off suggested for occupational exposures [6]).

Treatment Regimen

Until further data are available, the treatment regimen should be modeled after that used for occupational exposures (Table 3) [36, 37]: zidovudine and lamivudine for 4 weeks. An alternative regimen is stavudine and didanosine, but zidovudine is the only drug for which efficacy data in the setting of postexposure treatment are currently available. Addition of a protease inhibitor may be indicated if the source patient has advanced HIV disease or is known to have a high HIV viral load (>50 000 copies/mL). Addition of a protease inhibitor should also be considered if the source partner has been previously treated with one or both of the nucleoside analogues in the recommended two-drug regimens. If the partner is experienced with numerous antiretroviral drugs, individualization of the regimen is necessary. The importance of adhering to the regimen must be emphasized.

Of the available protease inhibitors, nelfinavir or indinavir should be used. Although there is more experience with indinavir, this drug must be taken on an empty stomach; this requirement may decrease compliance. Indinavir use is also often complicated by kidney stones or other urologic symptoms [38], although this can be minimized by having patients drink at least 48 ounces of fluid a day. Nelfinavir commonly causes diarrhea, which is usually well controlled with over-the-counter diarrheal medications.

Some HIV experts routinely prescribe triple therapy for prophylaxis after sexual exposure. We do not favor this as a routine approach because use of a third drug increases the risk for side effects, complicates the regimen (which may decrease adherence), and increases the cost of treatment. Although triple therapy is more effective than two nucleoside analogues in reducing viral replication in HIV-infected persons [39], the same may not apply to prophylaxis. Persons with long-standing HIV infection have billions of viral particles in their bodies; in contrast, the viral inoculum immediately after sexual exposure is very small and a single drug may therefore be effective. However, patients who have had multiple exposures and do not seek care until close to the 72 hour cut-off will probably have higher viral loads. Transmission of HIV strains that are resistant to zidovudine [40, 41], didanosine [42], or lamivudine have been documented [43], but transmission of resistance from partners who have not been receiving these medications is unlikely.

Baseline laboratory evaluation detects abnormalities that would affect drug choices or dosing and establishes a baseline so that abnormalities are not incorrectly attributed to drug therapy (Table 3). Treatment should not be delayed until results are obtained. Women should have a urine pregnancy test. If pregnancy is suspected or documented, the woman should be counseled about the potential risks that antiretroviral treatment presents to the fetus. Zidovudine is the only drug for which extensive safety data during pregnancy is available; it seems to be safe even in the first trimester. We have limited experience with lamivudine and indinavir in the second and third trimesters. Women who have not used contraception should be offered emergency contraception [44]. Victims of sexual assault require additional evaluation and counseling [45].

Persons with sexual exposures to HIV should also be assessed for the presence of other sexually transmitted diseases, including hepatitis B, hepatitis C, gonorrhea, syphilis, and chlamydial infection (Table 3). Treating sexually transmitted diseases is an effective HIV prevention strategy because several of these infections facilitate HIV transmission [46-48]. Patients who do not have antibodies to hepatitis B should be immunized [49].

Counseling and Prevention Interventions

Postexposure prophylaxis allows persons who are at high risk for HIV infection to interact with the medical profession. Although the impact of physician counseling on the reduction of HIV risk has not been demonstrated, clinician counseling has been shown to be successful in reducing other risk behaviors [50, 51].

Counseling is particularly important in the context of postexposure prophylaxis. This prophylaxis may encourage unsafe behavior during treatment (because patients feel protected) or after treatment (because patients think that they can obtain another course of treatment).

All patients, regardless of HIV status, should receive risk reduction counseling [8, 52-54]. Targeting specific behaviors and skill building is more effective than providing informational health education [55, 56]. For example, a person whose exposure is caused by a broken condom may benefit from advice about the selection and proper use of condoms. Counseling should address risks with the source partner and with other partners. Because ongoing prevention counseling is more effective than a single session [57, 58], referral to multisession programs where available is advised [56, 59, 60]. This may be especially useful for women who need support to improve condom negotiation skills [56, 61].

For substance users, referral for substance treatment, including needle exchange (if available) for injection drug users, should be a high priority [62]. Postexposure prophylaxis is warranted for drug users who have used HIV-contaminated equipment [7] as well as persons with sexual exposure.

Follow-up Care and Surveillance for Primary HIV Infection

At follow-up visits, adverse reactions to drug treatment should be evaluated. Monitoring should include complete blood count, liver function tests, amylase evaluation (if didanosine or indinavir is used), glucose evaluation (if indinavir or nelfinavir is used), and urinalysis (if indinavir is used). Patients should receive HIV antibody testing at 6 weeks, 3 months, and 6 months.

Patients should be advised to see their clinician if they develop fever, fatigue, pharyngitis, rash, or other symptoms of acute HIV seroconversion [19, 63]. Diagnosing primary infection has emerged as a high priority because treatment during this period may decrease the total body burden of virus and delay or prevent the onset of immunodeficiency [19, 64, 65]. Diagnosing primary infection is also important to ensure that patients who need long-term therapy do not receive only a 4-week prophylactic regimen. Because primary infection is a time of high viral load [19, 63, 66] and infectivity [67], prompt treatment during this period may decrease viral load and thereby decrease the chance that the patient will transmit HIV to others [19, 68, 69].

Although viral load tests can detect primary HIV infection as soon as 10 to 14 days after exposure, false-positive results occur in 2% to 3% of cases with the currently available branched-DNA test (D Chernoff. Personal communication). Positive test results, especially those at low numbers of copies, should be repeated and confirmed with antibody testing.

Cost of Treatment

The average wholesale cost of a two-drug postexposure regimen is approximately $500 [70]. Adding a protease inhibitor adds $500 to $600. Initial and follow-up visits, including laboratory evaluation, cost about $600 [71].

Two analyses indicate that postexposure prophylaxis is cost-effective for sexual exposures, especially exposures that are most likely to result in HIV infection [71, 72]. The marginal cost-effectiveness of zidovudine prophylaxis after receptive anal intercourse with a known HIV-infected partner was approximately $16 000 per seroconversion averted, whereas the cost of prophylaxis after insertive vaginal intercourse with a known HIV-infected partner was approximately $1.7 million per infection averted [71]. Similarly, in a model that included combination therapy and the medical costs of HIV disease, postexposure treatment of persons exposed through receptive anal intercourse saved money as long as the probability of the partner being infected was 0.50 or greater [72]. Both analyses assume that the efficacy of postexposure prophylaxis for sexual exposures is the same as that for occupational exposures (79%). However, a cost-effectiveness model of postexposure treatment in the occupational setting that assumed a likelihood of seroconversion of 0.005 found that postexposure prophylaxis was cost-effective even if its efficacy was only 40% [73].

These analyses indicate that postexposure treatment is cost-effective, but the question of who will pay remains, especially because the costs for a community could be prohibitive. Given the low rate of HIV infection resulting from a single exposure, the 41 000 new infections per year in the United States [1] translate to an astronomical number of exposures that could require prophylaxis. The proportion of persons with exposures who will seek treatment within the narrow time frame is not known. Because few persons at high risk for HIV infection have private insurance [74, 75], the costs of treatment will be borne by public insurance programs (such as Medicaid) and local health departments.

Repeated Exposures

Patients with repeated HIV exposures pose difficult practical and ethical dilemmas. Persons committed to safer sexual practices will sometimes relapse [76, 77], and it would be unfair to deny them potentially effective therapy. Others may seek continuous or episodic treatment as a way of maintaining or increasing unsafe behavior. Such persons must be counseled that repeated exposures are likely to result in HIV infection because postexposure treatment is probably not 100% effective. Although physicians are not obligated to provide harmful or ineffective treatment, they should not turn away patients who repeatedly put themselves at risk for HIV infection. Such patients desperately need other types of help. Prevention counseling, mental health services, or substance treatment services are more likely to address their needs than are repeated courses of antiretroviral therapy.

Public Health Implications of Postexposure Prophylaxis

We focus on the risks and benefits of treatment for individual persons who seek care after an exposure. However, the potential risks and benefits for the community as a whole must also be considered. In particular, existing HIV prevention efforts could be undermined if persons initiate or resume unsafe sexual practices because they expect postexposure treatment to be protective. To prevent an increase in unsafe behavior, public health messages must emphasize that the use of condoms and the avoidance of high-risk behaviors are the most effective methods of preventing infection. Postexposure treatment is a backup in case of failure of primary methods.

It is also possible that the availability of postexposure treatment will strengthen existing prevention efforts. Postexposure treatment may motivate those persons who have the highest risk for HIV to seek care; once engaged, they may be referred to ongoing prevention programs.

Conclusions

Although we do not know whether postexposure prophylaxis is effective, HIV-prevention interventions decrease the frequency of HIV risk behaviors [59, 60]. Physicians can play a critical role in decreasing HIV incidence by providing postexposure treatment with an emphasis on prevention of future high-risk exposures.

Dr. Gerberding: UCSF/SFGH Epidemiology and Prevention Interventions Center, San Francisco General Hospital, University of California, San Francisco, Box 1372, San Francisco, CA 94143.