Postexposure Prophylaxis After Nonoccupational HIV Exposure: Title and subTitle BreakClinical, Ethical, and Policy Considerations

BECAUSE PRESS REPORTS often describe protease inhibitors as breakthroughs in the treatment of human immunodeficiency virus (HIV) infection,^{1 - 3} patients are increasingly inquiring whether these drugs might prevent HIV infection after exposure to the virus through sex or injection drug use.⁴ In San Francisco, Calif, 2 clinics have been established to provide postexposure prophylaxis (PEP) after high-risk exposures.^{5 - 6}

The Centers for Disease Control and Prevention (CDC) has issued recommendations for PEP after some occupational exposures to HIV.⁷ If the risk of HIV transmission through certain sexual or injection drug practices is of the same order of magnitude as those occupational exposures, PEP would seem to be indicated in these circumstances as well.⁸ The ethical principle of fairness requires that clinically similar patients be treated similarly, regardless of the route of HIV exposure.^{9 - 10 (pp201-211)} The issue of PEP after exposure through sex or injection drug use is important because these modes of HIV transmission are far more frequent than occupational transmission.

There are no data on the effectiveness of PEP in nonoccupational settings, and definitive studies are unlikely because of the large sample sizes required and the ethical obstacles to a placebo-controlled trial. However, clinicians are accustomed to making decisions under conditions of uncertainty. By using the best available epidemiologic information, clinicians can estimate the risk of HIV transmission for any given exposure and compare it with risks for which the CDC recommends occupational PEP.

In this article, we first review existing data on PEP and the CDC occupational PEP recommendations. Drawing on these recommendations, we present a framework for assessing the risk of transmission and develop criteria for recommending PEP for nonoccupational exposures. We then use a series of paradigmatic clinical cases to explore factors that clinicians should consider when deciding whether to recommend PEP for a particular patient. Finally, we respond to several concerns about nonoccupational PEP and assess PEP as part of an overall HIV prevention strategy. Previous authors have discussed the rationale for PEP for nonoccupational exposures and recommended specific regimens.^{8 ,11} This article focuses on the practical questions clinicians face when deciding in which situations and settings to recommend PEP.

The principal evidence on the efficacy of PEP in occupational settings is a multicenter case-control study of health care workers (HCWs).^{7 ,12} This study compared the PEP use of 31 HCWs who became HIV infected following an occupational exposure to HIV-infected blood with that of 679 control HCWs who were exposed but remained seronegative. In the unadjusted analysis, there was no evidence of the efficacy of PEP, usually 3 to 4 weeks of zidovudine treatment. However, after statistical adjustment for potential confounders, there was a 79% reduction in the odds of infection for those who received PEP. Of course, case-control studies have greater potential for bias than randomized controlled trials. Additional support for PEP comes from studies of the prevention of perinatal transmission^{13 - 15} and some animal studies.^{16 - 18}

Extrapolating from these studies, the CDC produced its current guidelines for occupational PEP.⁷ The guidelines are based on stratifying occupational HIV exposures by the type of exposure (percutaneous, mucosal, or cutaneous), the type of body fluid, and the disease status of the HIV-positive source patient.

In all cases, the CDC recommends informing the HCW of the option of PEP. The CDC "recommends" PEP only for percutaneous exposures that involve large volumes of blood and/or blood containing a high HIV titer. The agency suggests "offering" PEP for most other percutaneous exposures, as well as for mucous membrane exposures involving visible blood, tissue, or certain potentially infectious fluids such as semen or vaginal secretions. The CDC does not recommend PEP for other mucous membrane exposures or the great majority of cutaneous exposures.⁷ These recommendations do not address the issue of nonoccupational exposures.

The probability of HIV transmission is a function of 3 factors: the frequency of exposure, the probability that the source is HIV positive, and the probability of transmission if the source is infected.

While repeated exposures are infrequent in the occupational setting, they are common with sexual contact or drug injection. Continuing exposures that would in effect require ongoing prophylaxis should be distinguished from sporadic exposures that can be treated with discrete courses of PEP.

In the occupational setting, the HIV status of the source patient is often known or can be readily determined. In contrast, in sex or drug exposures, the source may not be available or the HIV status may be unclear. In some instances the source patient may be known to be HIV positive. In others, the clinician may have to estimate the likelihood that the source is HIV positive by the seroprevalence in similar patients. An estimate of seroprevalence can be made from local research studies or the CDC's National HIV Seroprevalence Surveys,¹⁹ which collect information on sentinel hospital inpatients, drug treatment entrants, and Job Corps enlistees.

The risks of occupational HIV transmission have been fairly well delineated. Percutaneous exposures involving body fluid from an HIV-infected source carry a risk of transmission of about 0.25%.²⁰ Exposures with large amounts of blood, high titers of HIV, or deep injury are most risky. For mucous membrane exposures, the risk is about 0.09%. Although HIV seroconversion following cutaneous exposure has not been reported in a prospective study,²⁰ such transmission does occur.

The risk after nonoccupational exposures is less certain. For men who have sex with men, the risk of HIV transmission following a single act of receptive anal intercourse with an HIV-positive partner has been estimated at 0.5% to 3%,²¹ greater than the risk from a percutaneous occupational exposure. For discordant heterosexual couples, it is estimated that 0.1% of acts of penile-vaginal intercourse lead to transmission.²² The risk is believed to be higher for male-to-female transmission than vice versa.^{23 - 24} While oral sex can transmit HIV, the risk appears to be too low to quantify with precision.²² Factors associated with higher risk of sexual transmission include high virus titer, viral subtype, mucosal trauma, and concomitant genital tract infection.^{25 - 28}

Quantifying the risks after needle sharing is still more difficult because injection drug users (IDUs) may also have sexual risks for HIV infection. One expert estimates the risk as 1% (range, 0.4%-3%) per shared injection with an HIV-positive partner, if that partner is beyond the acute phase of HIV infection (written communication, D. C. Des Jarlais, MD, January 31, 1997). Mathematical modeling studies suggest a risk of 0.67%.²⁹ This risk is higher than the rate after a percutaneous occupational exposure to HIV-infected fluid.

In sum, receptive anal intercourse and sharing injection drug equipment with an HIV-infected partner probably carry a risk of HIV transmission at least as great as the risk that the CDC believes warrants offering PEP in the occupational setting.

Table 1 presents a conceptual framework for considering PEP for a patient who reports sexual or needle-sharing exposure to HIV. Because there are no clinical data on the benefits of PEP after such exposures, we extrapolate from existing data on the benefits of PEP after occupational exposure. Although drug injection differs from percutaneous injury in that it is usually intravenous and transfers substantially more blood,³⁰ the implications of these differences for PEP efficacy are unclear.

Similar to the CDC's occupational PEP recommendations, we recommend PEP after a sporadic exposure with a risk of infection of about 0.30% or greater. If the risk of infection is moderate, in the range of 0.10% to 0.30%, we do not believe the evidence currently warrants a recommendation for PEP in all cases. Rather, we believe that all persons exposed should be informed of the risks and benefits of PEP, so that decisions can be individualized, taking into account patient preferences. In lower-risk situations, clinicians should inform patients of PEP but not recommend it (see definitions of terms in Table 1). Continuing exposures are discussed under "Case 2" below. The CDC guidelines for occupational PEP suggest using a combination of 2 nucleoside analogs, zidovudine and lamivudine, for most exposures for which PEP is indicated, with a protease inhibitor, indinavir, added for the highest-risk exposures.⁷ A similar approach to nonoccupational PEP regimens has been proposed¹¹ with the highest-risk exposures defined as receptive anal intercourse or syringe sharing with an HIV-infected partner who has an elevated viral load.

The following hypothetical cases illustrate the factors that a clinician should consider when a patient inquires about nonoccupational PEP. We concur with the CDC that all exposed patients need to be informed that PEP exists. Ethically, respect for the autonomy of the exposed patient requires the disclosure of an unproven but plausible use of available drugs to try to prevent a potentially fatal (although infrequently transmitted) infection. Maintaining the confidentiality of patients requesting PEP is crucial. Merely requesting PEP may stigmatize the patient or lead to discrimination in health insurance or employment.

Case 1: Sporadic Exposure Through Sharing Injection Equipment With an HIV-Positive Partner.— A heterosexual IDU who had a negative HIV test result 3 months ago presents in a New York City emergency department. He has been injecting cocaine for 2 years but reports that he has not shared syringes for over a year. Four hours earlier, because he did not have his own syringe with him, he used a syringe that had been used recently by an HIV-positive friend.

This sporadic high-risk behavior carries an HIV transmission risk similar to or higher than an occupational needlestick with an HIV-infected syringe. In this scenario, we would recommend PEP, together with counseling on safer drug injection practices, referral to ongoing primary care and drug treatment programs, and baseline and follow-up HIV testing.

If the source's HIV status is unknown, the probability he or she is HIV positive can be estimated by the local IDU HIV seroprevalence. In New York City, where that seroprevalence is 41% to 48%,^{31 - 32} we would recommend PEP. However, in a city such as Columbus, Ohio, where the IDU HIV seroprevalence is 0.5%,³² the risk of transmission is about 2 orders of magnitude lower, and we would not recommend PEP.

Similar considerations would have applied had the patient presented soon after a single episode of receptive anal intercourse with an HIV-positive partner. However, had the patient instead reported oral sex, we would not recommend PEP because of the low risk of transmission.

Case 2: Continuing Exposure Through Sex, Source Serostatus Unknown.— A 19-year-old, HIV-negative female sex worker presents at a primary care clinic. Although she "mostly practices safe sex," several times a month she has vaginal intercourse without a condom, "because the customers pay a lot more for it." She denies ever injecting drugs and asks for something to prevent HIV infection from her most recent exposure, which occurred 4 hours previously.

This is a continuing exposure. Because the CDC recommends that PEP continue for 28 days after occupational exposures,⁷ an individual who is exposed several times a month would, in effect, need continuous prophylaxis. Because the toxic effects, effectiveness, and cost-effectiveness of continuous PEP are not known and may be unfavorable, we believe that continuous prophylaxis should be formally evaluated in a clinical trial before it can be recommended. This patient should receive HIV counseling and testing and be referred to state-of-the-art risk reduction programs.

Case 3: Sporadic Exposure Through Sex, Source Serostatus Unknown.— A heterosexual man who tested negative for HIV 2 years ago attends college in Nebraska. Until 3 months ago, he had a regular female partner who was also HIV negative, but he has not been sexually active since. Last night, however, he had unprotected vaginal intercourse with a female student. He comes to the college health center asking for "something to prevent AIDS." He denies injection drug use or sex with men.

This is a sporadic exposure to a source of unknown serostatus. The HIV seroprevalence among college students is estimated to be 0.2%.³³ The risk of transmission is at least 2 orders of magnitude lower than with occupational mucous membrane exposure to HIV-positive blood, a circumstance in which the CDC recommends offering PEP. We would inform the patient about PEP but would not recommend it. Safer sex counseling would be essential. If the patient insists on PEP, he and his partner should ideally be tested for HIV, and his prophylaxis discontinued if she is HIV negative.

Cases of sporadic unprotected heterosexual vaginal intercourse with a known HIV-positive partner would fall into a medium-risk zone. The risk is similar to those occupational exposures for which the CDC suggests offering PEP. We therefore believe that under these circumstances the clinician should inform the patient of the risks and benefits of PEP and prescribe it only if the patient requests it. When individualizing decisions about PEP, practitioners should consider factors that may increase HIV infectivity, such as the presence of bleeding, trauma, or sexually transmitted diseases.

Clinicians caring for rape survivors may recommend PEP even though the risk of HIV transmission may not be as high as in cases 1 and 2. The risk of HIV transmission after rape may be greater than from consensual sex because of genital and rectal trauma and bleeding,^{34 - 35} exposure to multiple assailants, or exposure through multiple receptive sites. Psychological trauma is common among female rape survivors and may include anxiety over becoming pregnant or acquiring a sexually transmitted disease such as HIV.³⁶ Postexposure prophylaxis for HIV might help survivors gain a sense of control and decrease their anxiety. If the assailant is apprehended and can be tested for HIV, prophylaxis could be discontinued if the test result is negative. Prophylaxis for sexually transmitted diseases and postcoital contraception are routinely offered to rape survivors^{34 ,37 - 38} and provide a precedent for HIV PEP.

The perceived risk of acquiring HIV infection is a key motivation for adopting and maintaining risk reduction behaviors.^{39 - 41} The availability of PEP could attenuate this perceived risk and lower motivation to avoid HIV exposure. Recommending PEP may be interpreted as a "double message": on the one hand, it is critical to avoid HIV exposure because it is incurable and fatal, while on the other hand, some risk taking is acceptable because PEP is available. There is some early evidence that "disinhibition" might occur. Four of 22 gay men in San Francisco indicated that, at the time of a high-risk exposure, they had thought about PEP to prevent infection.⁴² We do not believe that these concerns justify withholding PEP at the present time, although they will require ongoing monitoring.

An analogous situation is emergency "morning-after" contraception. Although opponents feared that it would encourage sexual risk taking, there is no evidence that nonemergency contraceptive use rates declined after morning-after contraceptives were approved.^{43 - 45}

The HIV prevention message will need to integrate PEP with behavioral risk reduction in a more complex message: "The best ways to avoid getting HIV are sexual abstinence or a mutually monogamous sexual relationship and not injecting drugs. The next best ways are to practice safer sex and use clean syringes for each injection. However, if exposure does occur, prompt treatment may prevent HIV infection in some situations. Taking this treatment can be difficult, has side effects, is not appropriate for all exposures, and may not work." When patients seek PEP, clinicians need to promote behavioral risk reduction through counseling and referrals for drug treatment to minimize the potential for disinhibition.

The risk of disinhibition because of the availability of PEP is serious because disinhibition may lead to an increase in new infections rather than a decrease, as illustrated in this simplified calculation. It is estimated that the annual number of incident HIV infections among gay men in San Francisco is 336 and that the prevalence in this population is 30.5%.⁴⁶ Assuming that all transmission among gay men is caused by sexual activity, this is equivalent to 110,163 unprotected sexual acts (336/[transmission rate of 0.01 × prevalence of 0.305]), if unprotected acts occur as often with HIV-positive as HIV-negative partners. If unprotected acts increased as a result of PEP by only 10% to 121,179 and 10% of unprotected acts led to the administration of PEP (equivalent to 33 persons a day, 7 days a week), the total number of HIV infections would increase to 340. Thus, even if PEP prevents some HIV transmission, a modest increase in the number of unprotected sexual acts could actually lead to an increase in HIV transmission. (Among those presenting for PEP, 30 HIV infections would be prevented [121,179 × 0.1 presenting for PEP × 0.305 × 0.01 × 0.8 PEP efficacy] and 7 persons would receive PEP but still become infected [121,179 × 0.1 × 0.305 × 0.01 × 0.2]. Among those not presenting for PEP, 333 infections would occur [121,179 × 0.9 × 0.305 × 0.01] for a total of 340 infections [333 + 7]. This assumes that referral rates would not differ and that HCWs would not prescribe differentially according to actual partner HIV status, which often will not be known.)

Most clinicians can empathize with colleagues who have suffered a needlestick from an HIV-positive patient and would probably be willing to prescribe PEP for them. In contrast, some HCWs may be reluctant to provide PEP to individuals whose sexual activities or drug use they deem irresponsible or self-destructive. However deep these feelings, practitioners have an ethical obligation to act in the best interests of the patient and to address the patient's clinical needs, regardless of the cause of the patient's clinical problems.^{10 (pp36-43)} Clinically, it is irrelevant whether the patient is an "innocent victim" or a "knowing participant." Health care workers are accustomed to treating injured patients who were driving while intoxicated and caring for patients with coronary artery disease who are smokers.

Animal studies suggest that PEP may not be effective after 24 to 36 hours, and the CDC therefore recommends starting PEP within 2 hours of occupational exposure, although the agency also recommends PEP for some individuals presenting 36 hours or more after exposure.⁷ Many HCWs who sustain a needlestick are aware of PEP, and some hospitals facilitate rapid access to PEP after occupational exposures on a 24-hour basis.⁴⁷

In contrast, it may be difficult to provide timely nonoccupational PEP. Sites for care may not be open around-the-clock, and clinicians may not be aware of the latest developments in this rapidly evolving field. Persons most likely to benefit from PEP may be poorly informed and estranged from the health care system. Outreach programs would be needed if PEP is to have a significant public health impact.

In 1 study, 31% of HCWs who began prophylaxis with zidovudine after occupational exposures failed to complete the regimen because of adverse effects.⁴⁸ More adverse effects can be expected with combination antiretroviral therapy. Protease inhibitors have significant adverse effects in HIV-positive patients, ranging from gastrointestinal intolerance and peripheral neuropathy with ritonavir to nephrolithiasis with indinavir.⁴⁹ Adverse effects acceptable to those infected with a fatal disease may be intolerable to healthy patients taking prophylaxis for an infection they are unlikely to acquire.⁵⁰ Poor adherence may lead to PEP failure and to the emergence of resistant HIV strains.⁵¹

Problems with adherence should not preclude offering PEP; instead, they demand strategies to improve adherence. Helpful strategies include educating the patient about the importance of taking medications as prescribed, addressing potential barriers to adherence, and helping patients identify personal daily routines as cues. Psychological distress, which is associated with nonadherence among persons with HIV,⁵² is likely to be present in persons presenting for PEP and needs to be addressed through counseling. Special packaging of the full course of PEP and home or telephone follow-up might also improve adherence.⁵³

Cost-effectiveness will depend on the likelihood that the source patient is infected, the risk of transmission, and the effectiveness of PEP. Currently, a 28-day course of zidovudine and lamivudine costs about $500.⁵⁴ It has been estimated that HIV testing and requisite office visits would add $592 to the drug costs.⁵⁵ To calculate cost-effectiveness, we assume an HIV-positive source, a risk of transmission of 1% (the approximate risk of seroconversion from receptive anal intercourse with an HIV-positive partner), and PEP effectiveness of 80% (the effectiveness of occupational PEP with zidovudine).⁷ The cost of preventing an HIV infection under these conditions is $136,500, less than the estimated lifetime cost of $195,188 to treat an HIV-infected person.⁵⁶ Assuming that life will be prolonged by only 10 years if PEP is effective, the cost per year of life saved will be $13,650, considerably less than the $50,000 per quality-adjusted life-year saved usually considered cost-effective.⁵⁷ Thus, we believe that insurance plans should cover PEP for sporadic high-risk exposures comparable in risk with those occupational exposures for which the CDC recommends prophylaxis.

Under other assumptions, PEP becomes less cost-effective. Adding a protease inhibitor for the highest-risk exposures will increase the cost of a month of PEP by $500⁵⁴ and increase the cost of preventing an infection to $199,000. Moreover, if the source patient's HIV status is unknown, cost-effectiveness declines further. For example, if the HIV seroprevalence is 10%, the prevalence among IDUs and men who have sex with men in many large US cities,³¹ triple therapy would cost $2 million per life saved. These rough calculations do not include the prevention of secondary infections.

Because of the large number of exposures, the total cost for nonoccupational PEP will be much higher than for occupational PEP and may strain available resources. While such an effort may be appropriate in HIV epicenters, in many regions PEP may divert large amounts of scarce resources from other worthy, often more cost-effective prevention interventions. Even in San Francisco, the existing PEP clinics served only 202 people in their first 10 months of operation.⁵⁸

Clinicians may feel uncomfortable discussing interventions that insurance may not cover. However, clinicians need to provide sufficient information to allow patients to make informed choices about options; some patients may choose to pay for PEP out-of-pocket. Practitioners also need to act in the patient's best interests, by making a recommendation regarding PEP and advocating for the patient if PEP is indicated but not covered by the insurer.^{10 (pp297-299)}

Recent advances in the understanding of the transmission of HIV now present clinicians with new opportunities and challenges. While recommendations for occupational PEP have been developed, there are no corresponding guidelines for the far more common nonoccupational exposures. Because definitive data on the efficacy of nonoccupational PEP will likely never be generated, physicians need to extrapolate from the known efficacy of occupational PEP. When making decisions in specific cases, physicians will need to take into account epidemiologic data on the risk of transmission, as well as complex ethical, behavioral, and social issues.

Since acceptance of this article, the CDC has published 2 documents relevant to PEP. The first proposed an algorithm for assessing whether occupational PEP is indicated and recommended a somewhat more aggressive approach to occupational PEP than in CDC's previous publications.⁵⁹ The second was a Public Health Service (PHS) Statement on nonoccupational PEP, which concluded that "PHS cannot definitively recommend for or against antiretroviral agents in these situations because of the lack of efficacy data."⁶⁰