Thousands of health care workers are potentially exposed to human immunodeficiency virus (HIV) each year via occupationally acquired needlesticks. The Centers for Disease Control and Prevention (Atlanta, GA) advise health care workers who experience a high-risk occupational exposure from an HIV-infected patient to begin receiving multidrug antiretroviral postexposure prophylaxis (PEP) as soon as possible, preferably within 36 h after exposure. Although the need to prescribe antiretroviral postexposure prophylaxis in a timely fashion is common, few data exist regarding the efficacy and optimal regimen for prophylaxis to prevent transmission. Our objectives were to examine the limited human and animal data on postexposure prophylaxis, to elucidate the factors that affect the choice of 2 versus 3 drugs as the optimal prophylactic drug regimen, and to place these findings within a mathematical framework to help guide the prescription of PEP.
An estimated 400,000 health care workers in the United States sustain occupationally acquired percutaneous injuries annually, putting these workers at risk of acquiring bloodborne pathogens, including HIV . Transmission of HIV via needlestick is rare; it is estimated at a rate of 0.3% per exposure in pooled analyses of exposed health care workers. Transmission via mucous membranes is rarer, at an estimated rate of 0.006% per exposure [2, 3]. However, the Centers for Disease Control and Prevention's (CDC; Atlanta, GA) voluntary reporting system has recorded 57 cases of HIV infection/AIDS that were thought to have definitely been acquired by occupational exposure through December 2001, as well as 138 possible cases [4, 5]. In addition, as many as 180,000–280,000 people in the United States are unaware of their HIV-positive serostatus and may unknowingly serve as sources of viral transmission to health care workers . Among hospitalized patients in urban medical centers, the seroprevalence (as estimated by routine HIV testing) may be as high as 3.8% .
Although occupational HIV transmission is uncommon, needlestick injuries are frequent and require rapid decision making about whether postexposure prophylaxis (PEP) is required, and, if so, the most appropriate PEP regimen. Our objective was to review the available literature regarding 2- versus 3-drug PEP, to offer a model-based approach to optimizing PEP decisions in the face of limited data, and to offer future directions for PEP research.
PEP data and guidelines. PEP has been administered in animal studies of simian immunodeficiency virus (SIV), as well as in a single, small, retrospective case-control study of health care workers treated with zidovudine monotherapy [8–10]. Studies of macaques demonstrate that treatment initiated ⩽24 h after inoculation and extended through 28 days is effective in preventing SIV transmission . The duration of treatment in these studies is an important factor in preventing transmission; 1 week of PEP is shown to be less effective than 4 weeks . The single human case-control study, completed in 1994, describes 33 health care workers who became seropositive after an exposure, compared with 665 exposed control subjects who did not have seroconversion. In addition to identifying high-risk sources of exposure, such as visibly contaminated needles or advanced disease in the source patient, this study showed that PEP monotherapy with zidovudine diminishes the risk of transmission by ∼79% (95% CI, 48%–94%) . Observational studies of occupational and community PEP and the CDC registry address short-course (4-week) antiretroviral prophylaxis [2, 10–16]. Using these results, and extrapolating from the perinatal prevention literature, the CDC developed empirical guidelines for treating occupational HIV exposure (table 1) . These recommendations suggest either 2- or 3-drug prophylaxis on the basis of the perceived severity of the exposure and the stage of illness of the source patient [11, 18].
Few data exist regarding the efficacy and long-term safety of PEP regimens in health care workers. However, several studies of short-term follow-up of health care workers and patients receiving nonoccupational PEP address the extent of toxicity and adherence with PEP regimens (table 2). Although the sample sizes are small, there appears to be a greater incidence of side effects and discontinuation of the 3-drug regimen, compared with the 2-drug regimens [12, 13, 15, 16, 19, 20]. There are no published data that address the efficacy of 2-drug and 3-drug PEP regimens compared with one another or compared with zidovudine monotherapy.
PEP in practice: how are the guidelines used? In practice, PEP may be overused. Of >4000 calls to a national clinicians' PEP hotline, experienced clinicians responding to calls recommended stopping or not starting PEP in 58% of consultations. Furthermore, 7% of callers reported receiving PEP after an event that was not a true exposure . Despite guidelines that frequently recommend 2-drug PEP, the most commonly prescribed regimens in the United States and in Europe use 3 drugs [13, 22]. Individual preferences and emotions may ultimately guide treatment decisions . Although the perception of health care workers is that a 3-drug regimen is the more conservative option, toxicity leading to premature discontinuation may threaten the overall efficacy of a 3-drug regimen.
Is 3-drug PEP better than 2-drug PEP? Since 1996, using at least a 3-drug regimen of combination antiretroviral therapy has become the standard of care for the treatment of HIV infection, improving overall outcomes for infected patients [24, 25]. The goal of treatment of HIV infection is sustained HIV RNA suppression . Recent clinical experience has shown that, to achieve suppression, therapy should attack the virus at multiple sites of action—thus, the attraction of 3-drug regimens .
However, the goal of preventing transmission differs from that of treatment . After a needlestick, the intent is to prevent small amounts of virus from establishing infection in lymph nodes, a rare event even in the absence of prophylaxis . Although the study of health care workers that used zidovudine for PEP was conducted before the emergence of significant antiretroviral resistance, it suggests that even a single drug used for 28 days is relatively effective in neutralizing wild-type virus transmitted by a needlestick . Because single-drug PEP is so effective, and because an important cause of PEP failure is inadequate treatment duration, one underestimated priority in prescribing PEP is ensuring course completion.
The CDC recommends the use of a 3-drug regimen for high-risk exposures, such as those from deep, penetrating sticks with hollow-bore needles involving patients with very high levels of circulating virus. However, 3-drug regimens, which often include a protease inhibitor, have been associated with more-frequent side effects and decreased adherence to treatment [12–14, 16, 19, 26]. The most common short-term toxicities reported in the PEP registry of 492 health care workers included nausea/vomiting, fatigue, headache, and diarrhea. The potential risks include serious adverse events attributed to nevirapine in 3-drug regimens, including 2 cases of fulminant hepatitis, one of which required liver transplantation . Nonadherence is more frequently seen in PEP recipients than in patients who are being treated with antiretroviral therapy for known HIV infection. For example, Quirino et al.  compared the tolerability of 3-drug PEP with that of 3-drug therapy for HIV-infected patients and found that rates of antiretroviral toxicity were 6 times higher among PEP recipients (57%) than among HIV-infected patients (7%), and the rate of discontinuation of therapy among PEP recipients was 8-fold higher than that among HIV-infected patients. Thus, perhaps the accepted notion that a 3-drug regimen is preferred for HIV treatment should be challenged in the setting of prophylaxis.
The impact of drug resistance on the PEP decision. The only published case-control study of PEP collected data through 1994, before the advent of multidrug antiretroviral regimens and the widespread emergence of drug-resistant virus . Antiretroviral resistance among patients infected with HIV has been increasing in North American cities. The frequency of high-level resistance to ⩾1 drug has been estimated to be 12% ; among patients in San Francisco, it has been reported to be as high as 27% . Data on the treatment of drug-resistant HIV infection suggests that virus load suppression is possible but is generally delayed, compared with treatment of wild-type virus . Although the implications of primary drug resistance on the efficacy of PEP are unknown, there are ⩾21 documented failures of PEP among health care workers in which some of the HIV strains showed decreased drug sensitivity [11, 18, 30]. It makes intuitive sense that, in areas where drug resistance is common, a third drug in the regimen would help ensure that at least 1 drug is active against the potentially transmitted virus. In accordance with CDC guidelines, if the source patient's virus is known or suspected to be resistant to an individual drug, “the selection of drugs to which the source person's virus is unlikely to be resistant is recommended; expert consultation is advised” [18, p. 27]. PEP providers are faced with determining where individual situations fit in the spectrum of exposure severity and how to factor potential viral resistance into prescribing decisions.
One Approach to Pep Decision Making
The balance between drug toxicity and the attendant discontinuation of drug therapy on the one hand, and drug resistance and the potential need for more potent PEP regimens on the other, makes the optimal prophylaxis strategy unclear. Because occupational HIV transmission is rare, randomized clinical trials to elucidate the most appropriate PEP regimen are not feasible, as evidenced by the lack of any published data on transmission reduction since 1994. One way to explore this balance is by using decision modeling to synthesize the overall benefits of 2- and 3-drug PEP regimens for occupational HIV exposures.
We designed a decision model to compare the number of work-related HIV transmissions for every 100,000 needlesticks from an infected source (figure 1). The decision model (designed using DATA, version 3.5; TreeAge Software) (figure 1) begins with an occupational exposure to a known HIV-infected patient and assesses the number of infections under each of 3 strategies: no PEP, 2-drug PEP, and 3-drug PEP (Appendix 1). Two-drug prophylaxis consists of 2 nucleoside reverse-transcriptase inhibitors (zidovudine and lamivudine); 3-drug PEP consists of zidovudine and lamivudine plus a protease inhibitor (indinavir or nelfinavir). The virus from the source patient may be susceptible or resistant to the PEP regimen that is used. Each PEP recipient may experience drug toxicity and discontinuation of drug therapy. The rate of PEP discontinuation reflects the proportion of people who stop the drug before completing a full 4-week course; this may be because of drug toxicity, regimen complexity, or the perception by the PEP recipient that PEP is unnecessary. Finally, the efficacy of 2-drug and 3-drug therapy is included to determine the likelihood of preventing HIV infection. We define efficacy as the percentage by which PEP decreases the likelihood of HIV transmission. Model outcomes are the number of HIV transmissions per 100,000 health care workers after high-risk needlesticks from an infected source.
One Approach to Dealing with Data and Uncertainty
Model input parameters are shown in table 2. Parameters such as the improved efficacy of 3-drug PEP versus 2-drug PEP are unavailable in the literature. One advantage to using a modeling framework is the ability to explore alternative parameter estimates (based initially on expert opinion) using sensitivity analyses, which vary input values to explore their influence individually and collectively on the stability of the model results. Given the lack of guidance on how to balance the impact of key parameters, we can use the decision model to examine variables, such as PEP efficacy, prevalence of background resistance, toxicity, and medication discontinuation, and to establish thresholds at which the optimal PEP strategy might change.
When we apply the available data in the literature to the model described above, we find that, without PEP, 300 cases of HIV transmission would occur per 100,000 high-risk needlesticks (i.e., a transmission rate of 0.3%). Compared with no PEP, 3-drug PEP decreased transmission to 108 cases per 100,000 high-risk needlesticks, a reduction of 192 cases. However, 2-drug PEP was the preferred strategy, with an expected HIV transmission rate of 105 cases per 100,00 high-risk needlesticks—3 fewer cases than with 3-drug PEP.
Toxicity and discontinuation. We varied the rate of toxicity for 2-drug and 3-drug PEP in one-way sensitivity analyses and found the toxicity threshold at which the preferred strategy changes. As the 3-drug regimen becomes less toxic, it becomes the optimum choice. For example, 2-drug PEP is preferred when toxicity rates were 59% for 2-drug PEP but 66% for 3-drug PEP . However, the 3-drug regimen was favored if 3-drug PEP had a toxicity rate of <62%. We performed similar analyses on the rate of discontinuation. For example, 2-drug PEP is preferred when 21% of people taking 2-drug PEP and 28% of people taking 3-drug PEP did not complete the course . However, 3-drug PEP was favored if its discontinuation rate decreased to <26% or, conversely, if the 2-drug PEP discontinuation rate increased to >24%.
Efficacy and resistance. The interplay between drug resistance and the incremental increase in efficacy of 3-drug PEP over 2-drug PEP was also important. When the source patient virus was susceptible to the drug regimen, 3-drug PEP needed to be at least 83% effective (i.e., a transmission rate of 0.051%) to become the favored treatment option, compared with the estimated 79% efficacy (i.e., a transmission rate of 0.063%) for the 2-drug regimen. In the face of only drug-resistant virus, when 3-drug PEP was more than 61% effective (i.e., a transmission rate of 0.117%), compared with 50% efficacy for the 2-drug PEP (i.e., a transmission rate of 0.15%), the 3-drug regimen was favored. Examining the overall efficacy (for both drug-susceptible and drug-resistant virus) of 3-drug versus 2-drug PEP, 3-drug PEP was the preferred regimen when it was 25% more effective than 2-drug PEP. This means, for example, that 3-drug PEP was the best option when 3-drug PEP was 93% effective (i.e., a transmission rate of 0.021%) and 2-drug PEP was 74% effective (i.e., a transmission rate of 0.078%). If >15% of the source population harbored resistant virus (base case, 10%), the 3-drug regimen was favored.
Efficacy, discontinuation, and resistance. This model also allows exploration of the interplay of multiple variables and of the impact these estimates may have on the PEP decision. We used the model to explore the interaction between rate of antiretroviral resistance, efficacy of 3-drug PEP, and the probability of discontinuing 3-drug PEP (figure 2). We found that the increased efficacy of 3-drug PEP is offset by the increased rate of discontinuation with the 3-drug regimen. For example, when the prevalence of drug resistance was 10%, at a baseline increased efficacy of 81% (vertical line), 3-drug PEP was the preferred option only if it was discontinued <25% of the time. As the efficacy of 3-drug PEP increases, greater rates of discontinuation could be acceptable and still favor the use of 3-drug PEP.
We also examined how alternative prevalences of drug-resistant virus alter the PEP decision with respect to efficacy and discontinuation rates. At the same efficacy of 81%, the 3-drug regimen remained favored if the prevalence of drug resistance was increased to 18%, even with a discontinuation rate of 28% . At lower prevalences of drug resistance—for example, 4% —2-drug PEP was favored more often, because the increased discontinuation rates for 3-drug PEP were less likely to be offset by any potential efficacy gained by the third drug.
Future Directions for Pep Research
Although combination antiretroviral therapy has improved viral suppression in patients infected with HIV  and prevents maternal-fetal transmission of the virus , published data regarding the utility of prophylactic antiretroviral therapy are for zidovudine alone. The incremental ability of 2-drug and 3-drug PEP to prevent HIV transmission caused by occupational exposures, compared with the effect of zidovudine alone, has never been prospectively studied. In contrast to current practices of using 3-drug PEP most often, this decision model suggests that 2-drug PEP may, in many circumstances, lead to fewer cases of HIV transmission after a high-risk occupational exposure. This is because it is likely to be almost as effective, it is better tolerated, and regimens are more often completed, compared with a 3-drug PEP regimen.
Research on the relative efficacy of 2-drug versus 3-drug PEP for occupational exposures would help elucidate whether the increased toxicity is outweighed by improved efficacy. In addition, compiling occupational health data on toxicity experienced by health care workers, as well as on the rates of follow-up and nonadherence with PEP, would refine our estimates. Similarly, this model suggests that an important predictor of the optimal PEP regimen is the background prevalence of antiretroviral resistance; once the prevalence of drug resistance increases to >15%, 3-drug PEP is favored. PEP providers should be aware of local viral resistance rates as an important factor and should choose PEP regimens accordingly.
In the absence of available data, we can use modeling to explore the different variables that interact in this complex decision. However, such modeling exercises must be interpreted in the context of their limitations. This model does not allow for a change of regimen should toxicity occur, nor does it allow for a regimen choice based on drug resistance patterns of the source patient's virus. Data on the role of drug resistance in decreasing drug efficacy are limited. We may have overestimated the absolute number of transmissions by assuming that all exposures are high-risk; however, the absolute number of cases is less relevant than the relative number of cases in determining the optimal strategy. The prevalence of drug resistance locally—another important parameter for determination of the best strategy in the model—is often not available. Animal models and studies of nonoccupational PEP may help to elucidate PEP efficacy in the face of drug resistance.
Although occupational transmission of HIV is rare, the need to make a rapid decision about PEP on the basis of the best available data is a common and pressing problem [11, 21]. PEP providers should be clear that the goal of PEP is different than that of HIV treatment, and they should not focus solely on HIV suppression, but also on completion of a 4-week regimen, which has proven to be critical in animal studies [8, 9]. In the absence of source patient information, knowledge of local drug resistance prevalence may assist in the decision of when to add a third drug to a PEP regimen. This analysis is performed on the population level and should never replace important information about the source patient. The individual source patient's history of antiretroviral therapy and HIV genotype should be sought urgently, if available, because this information is critical to guiding the optimal PEP regimen. In addition, attention to anticipatory counseling and treatment for medication toxicity may improve adherence rates and, therefore, the overall efficacy of PEP .
In conclusion, we have reviewed current data regarding the optimal use of HIV PEP, the CDC prophylaxis guidelines, and the important factors worthy of consideration in the PEP prescribing decision. Motivated by uncertainty in the guidelines, we have used a modeling approach to explore the balance in the PEP decision between the risk of HIV transmission, PEP toxicity, PEP discontinuation, and the prevalence of antiretroviral resistance. Under many conditions, the benefit of completing a full course of a 2-drug PEP regimen exceeds the benefit of adding a third antiretroviral drug. Two-drug PEP regimens, especially in areas of low antiretroviral resistance, should continue to be considered reasonable.
We used observational studies of occupational and community PEP and the CDC HIV PEP registry to include in the model risks of drug toxicity and therapy discontinuation (table 2). The efficacy of PEP to prevent transmission was 79% with zidovudine monotherapy in a small case-control trial . The transmission probability was calculated by multiplying the transmission probability without PEP (0.3%) by (1 - efficacy). We obtained the background prevalence of antiretroviral resistance in a given source patient from the literature [28, 29, 34–37], which included a study of patients who were the source of occupational exposures at 7 sites in the United States during 1998–1999 , as well as patients who were recently infected with HIV .
In the absence of data, we made the following assumptions in the base case of the analysis.
1. The efficacy of 2-drug PEP is the same as that observed for zidovudine monotherapy (i.e., 79%) .
2. The efficacy of preventing transmission from an incomplete course of PEP was equivalent to that for no PEP (i.e., there was no partial benefit to a short course, as demonstrated in animal models) .
3. The probability of toxicity reflects side effects reported most commonly in the literature and excludes the risk of the rarest adverse events, such as fulminant hepatic failure .
4. The probability of the source patient having any single class of drug resistance is independent of the strategy chosen.
5. In the circumstance of drug resistance, drug efficacy in preventing transmission is decreased (table 2).
We thank Robert T. Morris and Heather E. Smith, for their technical assistance, and A. David Paltiel, for his critical review of the manuscript.
Financial support. National Institute of Allergy and Infectious Diseases (grants 1K23AI01794, RO1AI42006, and P30AI42851).