In a case–control study, the Centers for Disease Control and Prevention (CDC) found that five factors were independently associated with the risk for HIV infection after percutaneous injury Table 1 [4]. Three of these factors (deep injuries, visible blood on the device before injury, and devices that were used in an artery or vein before injury are probably markers for exposure to a greater volume of infected blood. Exposure to blood from a source patient with preterminal acquired immunodeficiency syndrome [AIDS] was also associated with an increased risk for infection (patients were considered preterminal if they died within 2 months of the exposure). High virus titers, which are typically present in advanced HIV disease, and possibly phenotypic or genotypic differences among strains of virus in these patients may explain this finding [5, 6].

Use of zidovudine soon after exposure, the final factor that predicted infection in the CDC study, was associated with a 79% reduction in the odds of HIV transmission (odds ratio, 0.21 [CI, 0.10 to 0.60]). This observation is the first epidemiologic evidence for a benefit from antiretroviral treatment among health care providers exposed to HIV, and it has stimulated renewed interest in prophylaxis after exposure. Although effective treatments to prevent occupational infections are urgently needed, data from a retrospective case–control study alone are not sufficient to warrant issuing a new treatment standard. Ideally, treatment should be recommended on the basis of several criteria, including the biological plausibility of benefit, efficacy in suitable animal models and in clinical trials that can be generalized to the population being considered for treatment, safety, and cost-effectiveness [7, 8].

The efficacy of antiretroviral prophylaxis is biologically plausible. Local host defenses appear to be critical in deciding the outcome of transcutaneous exposure. Cutaneous dendritic cells normally take up viruses and present antigens to T lymphocytes, natural killer cells, and other immune effector cells; these cells are also believed to be the initial target cells in transcutaneous HIV exposure. Antigen-specific proliferation of T-lymphocytes, in response to HIV antigens, can be shown in lymphocytes obtained from exposed but uninfected health care providers [9]. Although a causal association between in vitro T-cell responsiveness and the outcome of exposure has not been proven, it is conceivable that the cellular immune system is important in aborting infection at this early stage. The process of initial virus uptake, antigen processing, and presentation to immune effector cells is not instantaneous and may take several hours or even days. Thus, a window of opportunity for intervention before local virus propagation or dissemination almost certainly exists. Early antiviral treatment may favor the host by minimizing the effective viral inoculum.

The results of experiments done in animal models to evaluate the efficacy of antiretroviral treatment after exposure are inconclusive. Zidovudine treatment after exposure can prevent retroviral infection in mice and cats, but data from these models are difficult to generalize to humans [10-13]. Most studies of prophylaxis with licensed antiretroviral drugs in primates have not shown efficacy, although treatment with zidovudine before exposure was effective in one study of infant macaques that were inoculated with a low titer of simian immunodeficiency virus [14]. The titer of virus used in most experiments involving primates is substantially higher than the amount that would be encountered in occupational exposure. In addition, the animals used in these experiments are usually inoculated intravenously, a method that would bypass potentially important cutaneous host defenses. Hence, absence of efficacy of prophylaxis in a primate model cannot be generalized to predict the outcome of percutaneous occupational exposure. Adverse consequences of treatment on the natural history of infection and delays in seroconversion have not been seen in animal experiments.

No randomized clinical trials of zidovudine for health care workers after exposure to HIV have been completed. A trial that is sufficiently powerful to prove efficacy is not feasible, given the low frequency of occupational infections [15, 16]. Zidovudine prophylaxis is effective in preventing transmission of HIV from infected women to their offspring [17]. Although it is not known exactly when or where zidovudine confers protection in the course of perinatal infection, the benefit cannot entirely be attributed to a reduction in maternal virus titers. The perinatal experience shows that zidovudine is an effective prophylactic drug in at least one clinical setting, but the relevance of this to occupational transmission is unclear.

The safety of zidovudine treatment among health care personnel after exposure to HIV has been evaluated in at least three studies [18-20]. In the only prospective, multicenter clinical trial of zidovudine treatment after occupational exposure [18], objective toxicity was uncommon among the more than 200 persons treated with relatively high dosages of zidovudine (200 mg five times a day for 4 weeks) [18]. Drug dosages did not have to be adjusted or discontinued because of clinical signs or laboratory evidence of toxicity. However, about one third of persons enrolled in this trial and in two surveillance studies [19, 20] stopped receiving zidovudine because they could not tolerate the side effects. Nausea, fatigue, headache, and gastrointestinal distress were frequent. These symptoms resolved when therapy was stopped, but the effect on otherwise healthy employees (including potential days of work lost) is a concern. Furthermore, these studies included few participants, are unlikely to have detected rare serious adverse events, and provide no data about delayed effects of treatment.

The cost-effectiveness of zidovudine prophylaxis after exposure has not been established. If it were possible to identify and treat only the small subset of persons at high risk for infection, costs would be small and would probably be offset by benefits (assuming that at least some efficacy exists). Probability models that distinguish high-risk exposure (exposure to a large volume of material with a high titer of virus) from low-risk exposure (exposure to a low volume of material with a low titer of virus) may be helpful in this regard [1, 4, 20, 21]. However, some degree of caution is appropriate because current estimates of risk are based on data from small samples of health care providers and have not been validated in prospective studies. Other factors that were not measured (such as injection of blood and quantitative HIV titer in the source patient) or that did not achieve statistical significance in the study by the CDC (such as use of gloves and the type and dimension of the needle) may eventually prove to be important in assessing risk [1, 4, 21]. One fact is evident: On average, 99.7% of health care providers who are exposed to HIV will not be infected, even if treatment is not provided. For these persons, treatment must be inexpensive and safe if it is to be accepted.

The criteria for establishing zidovudine treatment after exposure as a standard of care for health care providers have only been partially met. Nevertheless, data suggest that zidovudine prophylaxis is efficacious (although not 100% efficacious), serious short-term toxicity is rare, and it may be possible to identify and treat the subset of exposed persons at high risk. Thus, it might be reasonable to recommend zidovudine treatment after parenteral exposure to HIV, especially in the presence of factors associated with an increased risk for infection.

Unfortunately, the increasing prevalence of resistance to zidovudine among source patients is quickly rendering this recommendation obsolete. Several new antiretroviral agents have been approved for use in the past year. Resistance to these agents is likely to emerge with their long-term and widespread use, but most source patients currently have strains of HIV that are susceptible to the newer drugs. Use of zidovudine in combination with one or more new antiretroviral drugs is therefore a rational option for prophylaxis after exposure.

Lamivudine, a new reverse transcriptase inhibitor, is particularly attractive for use with zidovudine [22-25]. These drugs in combination appear to have synergistic antiviral activity, decrease the emergence of zidovudine resistance, and retain activity against many strains of HIV that are otherwise resistant to zidovudine [22-25]. The outcomes of clinical trials in HIV-infected patients show that adding lamivudine to zidovudine does not increase the incidence of side effects, but no data are available to predict the toxicity of this combination in otherwise healthy persons [22].

Protease inhibitors are also available for use in combination with reverse transcriptase inhibitors [25]. In patients with HIV infection, adding a protease inhibitor to antiretroviral treatment regimens can reduce serum virus titers to below the threshold of detection for at least several months. Indinavir appears to be the best licensed protease inhibitor for prophylaxis after exposure to HIV. It has potent antiviral activity and is generally well tolerated [25, 26]. It interacts with several drugs that might be used by otherwise healthy health care providers [26]. Approximately 2% to 3% of persons in clinical trials developed kidney stones that were attributed to indinavir, but protection can be offered by increasing fluid intake to at least 1.5 L per day. Saquinavir has poor bioavailability in its current formulation, but it is a reasonable alternative if indinavir cannot be taken [25]. Ritonavir also has excellent antiviral activity, but treatment must usually be instituted gradually to avoid adverse side effects. This strategy is not optimal for prophylactic treatment [25].

The efficacy of prophylaxis using a combination of antiretroviral drugs may be biologically plausible, but none of the other criteria for justifying a treatment recommendation have been met. Data are not yet available from animal models or clinical trials relevant to prophylaxis with combination regimens. Too few HIV-infected patients have been treated with the new drugs (alone or in combination) to allow any conclusions to be drawn about acute toxicity or to provide any perspective about potential reproductive or oncogenic effects. Combination treatment is expensive: It costs a pharmacy an estimated $860.00 to acquire a 4 week-course of zidovudine, lamivudine, and indinavir, and the cost of safety monitoring is substantial [25].

We are therefore left with a clinical conundrum. Given the moving target of antiviral resistance and the scarcity of relevant clinical data, how can we develop efficacious, safe, and feasible recommendations for treatment after exposure to prevent a rare but potentially fatal infection? To address this issue, the National Foundation for Infectious Diseases, in cooperation with the CDC, held a 2-day workshop on prophylaxis after exposure to HIV to gather information from a broad range of experts. After this meeting, the U.S. Public Health Service prepared new guidelines for managing occupational exposure to HIV [27]. The new guidelines differ substantially from the previous position of the U.S. Public Health Service, which was that available data were insufficient to allow a recommendation for or against treatment after exposure [28]. Treatment with zidovudine and lamivudine in combination is now recommended for most parenteral exposures. Addition of a protease inhibitor, such as indinavir, is advised for high-risk exposures and for persons in whom resistance to the reverse transcriptase inhibitors is suspected [27]. The International AIDS Society [29] has issued new recommendations for the use of antiretroviral agents that encourage treatment with antiretroviral drug combinations after exposure to HIV. Together, these new guidelines provide a strong incentive to expand access to prophylaxis after occupational exposure.

Investigations to evaluate the outcomes of new prophylactic regimens are already in progress. A prospective, multicenter clinical trial has begun to evaluate the safety of and compliance with treatment after exposure to HIV. The treatment protocol Table 2 includes 4 weeks of oral zidovudine (200 mg, three times a day) and lamivudine (150 mg, twice a day) for exposure that warrants prophylaxis (Table 3). Indinavir (800 mg orally, three times a day) is added when the source patient has received the two nucleoside analogues and if exposure is deemed to be especially hazardous (that is, if the exposure is to a large volume of material with a high titer of virus). Drug toxicity is monitored with a review of symptoms, a focused physical examination, and laboratory testing every 2 weeks during treatment (or more often if clinically indicated) and at regular intervals for 1 year after treatment.

The CDC is developing a surveillance system with which to monitor adverse effects associated with combination treatment regimens [27]. Clinicians are encouraged to enroll exposed health care providers in the anonymous registry when prophylaxis is initiated and to provide follow-up data about the treatment that was used, the adverse effects of the treatment, and compliance with the treatment regimen. A comprehensive surveillance program for monitoring occupational exposure and outcome of treatment is also being done in Italy and is planned in many other countries [2, 19].

To provide the best preventive care for health care providers who have been exposed to HIV, it is essential to have knowledgeable experts develop treatment protocols that can be rapidly updated as new data emerge [27, 29]. However, implementing the new guidelines presents difficult challenges. It is one thing to have a written protocol, but it is quite another to develop the ability to carry out this protocol in a timely manner, especially where exposure to HIV is infrequent.

The duration of the window of opportunity in which prophylaxis after exposure is effective is not known. Initiating therapy as soon as possible after the exposure (preferably within hours, not days) is presumed to be optimal. To ensure timely implementation of therapy, health care workers must be educated about the need for prompt reporting of exposures, must have access to efficient and reliable response system, and must be able to obtain drugs and instructions for drug use [1, 7, 8, 30]. In settings in which experience with the new antiretroviral agents is limited, a referral system should be established for obtaining expert clinical advice and follow-up care.

Most important is that decisions about prophylaxis after exposure to HIV must reflect the perspectives and needs of the exposed health care provider. Supportive counseling is perhaps the most important component of an effective program [30, 31]. Communication about the risk for infection and the rationale for antiviral treatment is difficult, especially during the acute crisis period that often follows exposure. Persons who express some interest in prophylaxis but cannot make a rational decision often benefit from starting treatment and then reconsidering their options when they feel more objective. A similar approach is useful when it is not yet known whether the source patient is infected with HIV.

Improving access to potentially life-saving prophylaxis is important for health care providers at risk for HIV infection. However, if prophylaxis after occupational exposure proves to be efficacious, where will we draw the line? Providing access to prophylaxis after accidental sexual exposure to HIV (for example, as a consequence of condom breakage) already seems to be a reasonable (and ethically justifiable) option in individual cases. The implications of a public health policy that advocates prophylaxis after sexual exposure to HIV are much more problematic. In this area, as in so many aspects of AIDS, advances inevitably bring new and more challenging questions.