The aim of the study was to describe the use of nonoccupational postexposure prophylaxis (NPEP) in Australia, and to estimate the number of HIV infections that its use prevented.
The aim of the study was to describe the use of nonoccupational postexposure prophylaxis (NPEP) in Australia, and to estimate the number of HIV infections that its use prevented.
We conducted a population-based observational cohort study of people who presented to antiretroviral prescribers in Eastern Australia, and reported a high-risk nonoccupational exposure to HIV, in 1998–2004. Prescribers collected data at baseline, 4 weeks and 6 months. Data collected included details of HIV exposure, drug regimens and HIV serostatus.
The great majority of the 1601 participants were male (95%) and presented after male homosexual exposure (87%). Only 32% of exposures were to HIV-positive sources. Two antiretroviral drugs were prescribed after 48% of events, and three or more drugs after 52% of events. The median time to receipt of NPEP was 23 h. Side effects were reported by 66% of participants. No case of NPEP failure in an adherent individual was identified. It was estimated that 0.9–9.2 HIV infections had been prevented. This compared with a total of 1138 newly acquired HIV infections notified in the geographical area covered by the study.
In Australia, NPEP has been widely prescribed and is mainly targeted at high-risk exposures. Although there were no identified failures of NPEP, it is likely that only a small proportion of new HIV infections in the study area were prevented. NPEP may be a valuable preventive intervention for an individual, but it can only play a minor role in HIV prevention at the population level unless targeting can be further improved.
Nonoccupational postexposure prophylaxis (NPEP) remains a controversial method of HIV infection prevention. Only a relatively small number of countries have published guidelines recommending its use [1–5]. Available guidelines are inconsistent with respect to the recommended antiretroviral regimen, including the choice of two vs. three drugs, the type of exposures that are eligible, and whether or not the source is required to be HIV positive [6,7]. Implementation of NPEP on a more extensive scale has been impeded by concerns about its efficacy, inappropriate use, toxicities and potential undermining of behavioural HIV infection prevention strategies [8–10].
In Australia, guidelines recommending the use of NPEP after high-risk exposures were released in New South Wales (NSW) in 1998 , and national guidelines were published in 2001 . The guidelines have been widely promoted among doctors and among people at high risk of HIV infection, predominantly homosexual men. Levels of awareness of its availability in this population are relatively high . In this report, we describe the Australian experience of NPEP and examine its public health impact in a setting where its use is well accepted and relatively widespread.
We conducted an observational cohort study of people prescribed NPEP in the Eastern states of Australia between 1998 and 2004.
In Australia, prescription of antiretroviral medication is limited to HIV specialist medical practitioners and to a relatively small number of other medical practitioners who are registered as antiretroviral prescribers. In 1998, all registered prescribers in NSW were included in the study. Prescribers in Queensland were included from 1999, and those in Victoria and the Australian Capital Territory from 2001 as NPEP became available in each jurisdiction.
Participants were eligible for this study whether or not they elected to take NPEP.
The national guidelines recommended two-antiretroviral-drug NPEP in most situations. Under the guidelines, three-drug NPEP was reserved for high-risk exposures to known HIV-positive sources when the source was known to have advanced HIV disease, or had an HIV viral load greater than 20 000 HIV-1 RNA copies/mL, or had known antiretroviral drug resistance to drugs from at least two classes .
The study was approved by the University of New South Wales Human Research Ethics Committee and by local area health authority ethics committees.
Prescribers completed data collection forms at the time of initial presentation, and at 4 weeks and 6 months later.
At the initial presentation, information collected included participant demographics, previous HIV test results, history of previous NPEP prescription and whether or not a baseline HIV test was performed. Details of the exposure included whether contact had been anal, vaginal, oral or other sexual exposure. Condom use was recorded, and whether or not breakage or slippage had occurred. Information was collected on percutaneous exposures including the reuse of injecting equipment and the depth of skin penetration. Information on the source, as reported by the patient, included gender, HIV serostatus and prior HIV infection risk behaviour. At 4 weeks after the exposure, another HIV antibody test was recommended, and adherence and side effects were assessed at the clinic visit. At 6 months after the exposure, results of all available HIV antibody tests since the exposure were collected.
To ascertain the proportion of all NPEP prescriptions that were included in the study, data from administrative records concerning drug cost reimbursement were obtained from 11 area health services in New South Wales between April 2001 and March 2002, and from all areas of Queensland between June 2000 and October 2002, and compared with numbers of NPEP prescriptions recorded through the study in the corresponding areas and time periods.
Original data were coded and entered into an Oracle8 database. Statistical analysis was performed using stata version 8.0 (Stata Corp., College Station, TX, USA). All P-values are two-tailed.
Four-week and 6-month follow-ups were defined as a visit at least 26 and 165 days after the initial consultation, respectively. The number of HIV infections expected in the study population was estimated by taking the sum over all participants of the product of the per-contact transmission probability, based on the reported type of exposure, multiplied by the probability that the source was HIV positive. The per-contact transmission probabilities were based on published estimates from prospective studies conducted in industrialized countries. Estimates from modelling or cross-sectional studies were not considered. These estimates were 0.82% [95% confidence interval (CI) 0.24–2.76%] for the receptive partner in unprotected anal intercourse (UAI) , 0.1% for both male and female partners in unprotected vaginal intercourse (95% CI 0.05–0.17%) , 0.0% (95% CI 0.0–0.04%) for receptive fellatio , 0.29% (95% CI 0.13–0.70%) for needle stick injury , and 0% (95% CI 0–0.28%) for mucosal exposure . As there are no published estimates of risk of insertive anal intercourse with a known HIV-positive partner, we assumed that the transmission probability was the same as for the insertive partner in vaginal intercourse.
The probability of the source being HIV positive was taken as 100% when the source was reported as positive by the participant. In all other cases, published estimates of HIV seroprevalence in appropriate populations were used. For male-to-male sex, the prevalence of HIV infection among gay men was based on self-reported HIV status as recorded in Australian community-based cross-sectional studies in 2003–2005 [17–20]. These estimates were 5.1% in Canberra, 6.4% in Brisbane, 9.1% in Melbourne and 14.2% in Sydney. Average seroprevalence weighted for state of residence was calculated. For heterosexual contacts, the estimated prevalence of HIV infection in 15–49-year-olds (0.2%)  was assumed. For injecting drug user sources, the prevalence of 1.1% measured among attendees at needle and syringe exchange programmes was used .
Of 1601 participants who were enrolled during the study period, 97% (1552) received NPEP and are the subject of the analyses reported here. During the period in which cost records were obtained, comparison of study NPEP prescriptions with drug cost reimbursement records showed that 52% of participants who were recorded as having received NPEP were enrolled. Participants were overwhelmingly male (95%) and had a median age of 33 years. Previous receipt of NPEP was reported by 217 participants (14%). The great majority (88%) reported a previous HIV test, and baseline HIV testing at the initial consultation was performed in 92% of subjects (Table 1).
|Age [median (IQR)]||33 (27–39)|
|Year of enrolment|
|2004 (until 31 May)||226 (14.6)|
|Received PEP before||217 (14)|
|Three times||7 (3.2)|
|≥Four times||7 (3.2)|
|Not received before||1302 (83.9)|
|Previous HIV test|
|HIV test at baseline|
Follow-up data at 4 weeks or later were available for 74% of participants. Of the 406 with less than 4 weeks of follow-up, 24(6%) had withdrawn because they had determined that their source was HIV negative. Follow-up data at 24 or more weeks postexposure were available in 26% of cases.
Most participants who received NPEP identified their source as a homosexual man (86%), an injecting drug user (4.5%) or a homosexual man with a history of injecting drug use (2.0%).
Participants most frequently reported male homosexual contact (87%). Among these, the most commonly reported exposures were unprotected receptive (55%) or insertive (32%) anal intercourse or both (7.8%). The source was known to be HIV positive in 31% of these exposures. Most risk episodes were reported as being with new partners, including 48% described as a casual partner and 36% as an anonymous partner. A further 12% were reported as being with a regular partner.
Among the 87 people (5.6%, of whom two-thirds were women) with heterosexual exposure, the source was known to be HIV positive for 31% of men and 57% of women (see Table 3 below). Most commonly (43%), exposure was to a regular partner, but 30% reported a casual partner and 15% an anonymous partner.
|Exposure type||Per-contact risk†|
(%, 95% CI)
|Exposures to HIV-positive|
|Exposures to HIV-unknown sources||Total infections|
|Anal receptive||0.82 (0.24–2.76) ||232||1.90 (0.56–6.40)||624||0.62 (0.18–2.10)||2.52 (0.74–8.50)|
|Anal insertive||0.1 (0.05–0.17) ||162||0.16 (0.081–0.28)||266||0.032 (0.016–0.055)||0.19 (0.10–0.34)|
|Receptive oral||0.0 (0.0–0.05) ||19||<0.01¶||27||<0.01¶||<0.01¶|
|Other||0.0 (0.0–0.28) ||11||<0.01¶||13||<0.01¶||<0.01¶|
|Male to female||0.1 (0.05–0.17) ||33||0.033 (0.017–0.056)||25||<0.01¶||0.033 (0.02–0.06)|
|Female to male||0.1 (0.05–0.17) ||9||<0.01¶||20||<0.01¶||<0.01¶|
|Injecting drug use||0.29 (0.13–0.70) ||10||0.029 (0.013–0.07)||8||1.1%||<0.01¶||0.029 (0.013–0.07)|
|Needle injury||0.29 (0.13–0.70) ||15||0.044 (0.020–0.11)||56||0.2%||<0.01¶||0.044 (0.020–0.11)|
|Other||0.29 (0.13–0.70) ||8||0.023 (0.010–0.056)||7||0.2%||<0.01¶||0.023 (0.010–0.056)|
|Other#||0 (0.0–0.28) ||3||<0.01¶||4||<0.01¶||<0.01¶|
|Total||502||2.19 (0.70–7.02)||1050||0.65 (0.20–2.16)||2.84 (0.90–9.18)|
Among participants with percutaneous exposures (6.9% of all exposures), the majority (67%) were nonoccupational, accidental needlestick injuries (NSI). Of those with NSI, most (78%) were unaware of the HIV status of the source and the majority (74%) reported either moderate or deep injuries with a hollow bore needle. Twenty-one (20%) of the percutaneous exposures involved reuse of injecting equipment. A considerably higher proportion of injecting drug users (52%) than people who sustained other NSIs (22%) reported that their source was known to be HIV positive. The remaining 13% of the percutaneous exposures were a heterogeneous group, with almost half (47%) presenting after assaults or physical altercations involving lacerations.
NPEP was prescribed within 72 h of exposure in 94% of participants and within 24 h in 52%. The median time to receipt of NPEP was 23 h (interquartile range 12–42 h).
Slightly over half of NPEP prescriptions (808; 52%) were for three or more antiretroviral drugs (Table 2). The most frequently prescribed regimens were the dual combination of lamivudine and zidovudine (38%), followed by the three-drug combinations of lamivudine, zidovudine and nelfinavir (27%) and tenofovir, lamivudine and stavudine (10%). The prescription of nevirapine declined markedly over time, from 34% in 2000 to no prescriptions in 2004. Prescription of tenofovir increased from none prescribed until 2003, to 36% of prescriptions in 2003 and 28% of prescriptions in 2004.
|Drug combination||Number prescribed|
|% of total|
|Zidovudine and lamivudine||586||37.9|
|Tenofovir and lamivudine||82||5.3|
|Stavudine and lamivudine||58||3.7|
|Other two-drug regimens||11||0.7|
|Zidovudine, lamivudine and nelfinavir||423||27.3|
|Tenofovir, stavudine and lamivudine||160||10.3|
|Zidovudine, lamivudine and nevirapine||54||3.5|
|Zidovudine, lamivudine and indinavir||43||2.8|
|Stavudine, lamivudine and nelfinavir||22||1.4|
|Stavudine, lamivudine and nevirapine||22||1.4|
|Zidovudine, lamivudine and efavirenz||8||0.5|
|Zidovudine, lamivudine and kaletra||8||0.5|
|Other three-or-more-drug regimens||68||4.4|
Among those 1146 participants who returned for 4-week follow-up, 80% completed the initially prescribed regimen and 5.8% completed a modified regimen. The most common reason for discontinuation was side effects (69% of those who discontinued permanently). Overall, 66% of participants who returned for the 4-week follow-up reported side effects. Nausea was most commonly reported (69% of those with side effects), followed by diarrhoea (44%), headache (29%) and vomiting (9.6%). There were few serious side effects. Two participants developed Stevens–Johnson syndrome within 14 days after initiating a regimen containing nevirapine. One participant who received a regimen containing indinavir developed renal colic probably related to nephrolithiasis. Being prescribed a combination of three or more drugs was associated with a greater incidence of side effects at any severity level (P<0.0001), compared with receipt of two drugs.
No new HIV infections related to treatment failure were observed. Overall, there were 10 new HIV diagnoses in people who were referred for NPEP. Seven were diagnosed at baseline and were therefore unrelated to the reported risk exposure. Three participants were found to have seroconverted after having been prescribed NPEP. All three had a documented negative HIV result at least 4 weeks after exposure. One participant seroconverted 3 months after being prescribed zidovudine, lamivudine and nelfinavir. His medical practitioner reported that he was poorly compliant and had ongoing risk behaviours. Another seroconverted 110 days after initiation of stavudine, didanosine and lamivudine, and the third 311 days after initiation of zidovudine, lamivudine and nelfinavir. Both had completed the prescribed regimen but were believed by the prescribing doctors to have had other HIV infection risk episodes after completion of NPEP.
The weighted average seroprevalence of HIV in source homosexual men was 12.2%. The range of the total number of HIV infections estimated to have been prevented by use of NPEP in the study period was 0.9–9.2 (Table 3). If NPEP use had been restricted only to those who reported receptive anal sex with a source known to be HIV positive, the range of expected HIV infections averted would have been 0.56–6.4 infections.
In Australia, use of NPEP has become widespread, but remains highly targeted towards homosexual men, with little prescription outside this group. This usage reflects the epidemiology of HIV in Australia . The vast majority of prescriptions occurred after high-risk exposures which qualified for NPEP according to local guidelines . Side effects were common but were generally mild to moderate and none was irreversible. Although 10 people were newly diagnosed with HIV infection, seven were men found to be HIV positive at baseline. The other three seroconverted more than 4 weeks after their exposure, and ongoing risk behaviour after the completion of NPEP was reported in all these cases. It appears that no new HIV infections related to NPEP failure occurred. We estimate that 1–9 HIV infections were prevented by the provision of NPEP.
As nearly three-quarters of eligible study participants did not return for follow-up at 6 months after exposure, our data cannot be seen as definitive evidence of the efficacy of NPEP. However, there are three reasons suggesting that it is unlikely that we missed NPEP failures. First, all the 850 people who had an HIV test at 4 or more weeks after exposure were HIV negative at that time, and the sensitivity of the HIV serological test is such that the vast majority of NPEP failures would have been detected then. Secondly, those who returned for the 4-week follow-up tended to be at higher risk than those lost to follow-up (data not shown). Thirdly, homosexual men in Australia report frequent testing for HIV, and testing is most frequent in high-risk men , so it is likely that they would have re-presented for testing to their doctors, and been notified to the study, had they seroconverted.
This study represents one of the largest population-based studies of NPEP. Australia was among the first countries in the world to recommend the use of NPEP, and awareness of the availability of NPEP among homosexual men in 2004 ranged from 46 to 66% in cross-sectional studies in Eastern State capitals . In an ongoing observational community-based cohort study of HIV-negative homosexual men in Sydney, 2–7% of men used NPEP each year . Levels of use of NPEP among at-risk individuals in Australia, and the impact of NPEP on the epidemic, are probably as high as anywhere in the world. It was therefore disappointing that only 1–9 infections were potentially prevented. However, as only about half of those who received NPEP in the study region were enrolled, it might be that twice that number were actually prevented by NPEP. Also, if the exposures were of higher than average risk, as a result of the presence of transmission risk cofactors such as high viral load and sexually transmissible infections, or if a higher than average proportion of sources were HIV positive, then it is possible that we have underestimated the number of infections averted. However, to put the number of infections possibly averted in perspective, surveillance systems in the areas covered by this study reported 3768 new HIV diagnoses and 1138 newly acquired HIV infections during the study period .
There were at least three major contributions to the relatively limited population-level impact of NPEP in this population. First, the HIV status of the source was known in only 32% of participants, suggesting that a relatively large number of NPEP prescriptions were unnecessary. In other reports of use of NPEP, the reported proportion of sources who were known to be HIV positive was similarly low, ranging from 30 to 47% [1,2,25]. In a Swiss study, a policy of active contacting and HIV testing of source persons reduced exposure to unnecessary NPEP by 73% amongst those whose source person could be contacted and tested . However, in our study, 33% of the contacts were reported as anonymous and an additional 44% casual, so it is unlikely that active contact tracing could have had a large effect in reducing NPEP prescribing. Secondly, only 15% of exposures, but 68% of prevented infections, were related to the highest risk exposure, receptive UAI with a known HIV-positive partner. This raises the question of whether NPEP prescription after anal intercourse between homosexual men should be restricted to the receptive partner or to cases in which the partner is known to be HIV positive. In fact, in our data, there already appears to have been a degree of selection towards presentation after receptive intercourse. Of all UAI episodes that led to NPEP, nearly one in five (18%) episodes were after receptive UAI with a known HIV-positive partner. In contrast, in a community-based cohort study of HIV-negative homosexual men in Sydney, only 6% of all reported UAI episodes that could have led to HIV transmission were reported to be receptive with a known HIV-positive partner . Whether this selection towards prescription of NPEP after receptive intercourse with an HIV-positive partner occurred because men were more likely to present to their doctor after such an event, or whether doctors were more likely to reserve NPEP for such a situation, is unknown. Thirdly, because of the relatively low transmission probabilities for HIV, even for the highest risk exposures, the great majority of NPEP prescriptions will always be for people who would not have become HIV infected anyway.
In a setting where so few HIV infections were prevented, safety is of paramount importance. Although the majority of NPEP recipients (66%) reported side effects, only a small number of side effects were reported as severe. The occurrence of side effects was more common in those prescribed three or more drugs, suggesting that the safety of NPEP could be improved by reserving three-drug prescriptions for highest risk exposures. However, a triple-nucleoside regimen with much improved safety and tolerability has recently been described . The occurrence of severe side effects, in particular the cases of Stevens–Johnson syndrome associated with nevirapine and nephrolithiasis associated with indinavir, highlights the imperative to balance the risk of HIV infection against the potential toxicity of the individual antiretroviral agents. The frequent occurrence of subjective drug-related toxicities in this study confirms the need to restrict the prescription of NPEP to those only at high risk of HIV infection after exposure.
The relatively small population-level impact of NPEP that we have estimated should not be seen as an argument against NPEP availability. An NPEP programme is but one part of a comprehensive HIV infection prevention programme , and the provision of NPEP offers an opportunity for HIV infection prevention counselling to high-risk individuals. The lack of identified cases of HIV infection attributable to NPEP failure after NPEP use suggests that NPEP may be an effective preventive intervention against HIV infection at an individual level.
The National Centre in HIV Epidemiology and Clinical Research is funded by the Australian Government Department of Health and Ageing. GlaxoSmithKline and the NSW Health Department provided additional financial support for this project.
The authors would like to acknowledge the work of research assistants Patty Correll, Belinda O'Sullivan and Wei Zheng. We would also like to thank the participants who consented to the collection of information.
We also wish to acknowledge the following medical practices, doctors and research nurses who enrolled participants for the study, and completed the data forms.
In New South Wales: 16 Oxford Street, Darlinghurst (Anthony Tedesco); 229 Oxford St, Darlinghurst (Marilyn McMurchie, Mark Kelly, Michael Kidd and Craig Rodgers); 407 Bourke Street, Darlinghurst (David Baker, William Genn, Janet Kidd, Robert Macfarlane, Hugh Macleod and Ankita Roy); 5 Bligh Street, Tamworth (Miriam Grotowski and Pam Rooney); 66 Oxford St, Darlinghurst (Ben Hanson); Albion Street Clinic, Surry Hills (Fraser Drummond, Brian Hughes, Harry Michelmore and Don Smith); Bigge Park Centre, Liverpool (Rodney Lovett, John Quin and Ruhand Steele); Burwood Road Practice, Burwood (Nicholas Doong); Coffs Harbour Sexual Health (David Ellis); Darlinghurst Medical Centre (Dennis Rhodes); Hawkesbury Sexual Health Services (Deborah Couldwell); Holden Street Sexual Health Centre, Gosford (Deborah Allen, Kym Collins and Cheryn Palmer); Holdsworth House General Practice (David Austin, Kate Bessey, Mark Bloch, Andrew Gowers and Dick Quan); John Hunter Hospital Byrne House (Michael Boyle); Kirketon Road Centre, Darlinghurst (Melanie Andromas, Marianne Jauncey, John Kearley, Tim Shortus and Hester Wilce); Livingstone Road Sexual Health, Marrickville (Damian Conway); Lismore SHAIDS (Steven Davies, Brian Mulhall and Gillian Smith); Nepean Sexual Health Clinic (Tong Liang); Paramatta Sexual Health (Katrina Lagios and John Moran); Port Kembla Sexual Health (Simon Marlton); Prince of Wales Hospital (Jeff Post); Quay Street Medical Centre (Ben Anderson); Royal North Shore Hospital, Clinic 16 (Derek Chan, David Crawford, Archie Darbar, Linda Dyan, Elizabeth Hamlyn, Ann Hofmeyer, George Kotsiou, Catriona Ooi, Claudia Whyte and Sebastian Van Hal); Royal Prince Alfred Hospital (Louise Evans); Sydney Sexual Health Centre (Chris Bourne, Ellen Chan, Marcus Chen, Damien Conway, Basil Donovan, Rene Francia, Janine Haran, Jo Lusk, Anna McNulty, Heidi Spillane, Julia Trayer, Catriona Ooi and Nathan Ryder); St George Hospital (Chris Carmody and Pam Konecy); Taylor Square Private Clinic, Darlinghurst (Neil Bodsworth, Rachel Burdon, John Byrne, John Ewan, Samuel Hay, Robert Finlayson, George Nisyrios, Cathy Pell, Ross Price, Matthew Routley, Matt Shields, Louise Tomlins, Emmanuel Vlahakis and John White); St Vincent's Hospital, Darlinghurst (David Cooper, Andrew Carr, Fraser Drummond and John McAllister).
In Queensland: 38 Gladstone Road (Leonie Todhunter); AIDS Medical Unit (Elizabeth Christensen, Ken Clare, John Patten and Hugo Ree); Blackall Terrace Clinic, Nambour (David Sowden); Brisbane Sexual Health Clinic (Gale Bearman, Natalie Edmiston, Bella George, Margaret Mobbs, David Orth, Diane Rowling and Tracy Schrader); Brunswick Street Medical Centre (Kate Evans, Graham Lister and Dominique Wells); The Doll's House Clinic, Cairns (David Bradford); Nambour Sexual Health (David Sowden); Gold Coast Sexual Health (John Chuah, Sally Jacobs and Maree O'Sullivan); Princess Alexandra Hospital Sexual Health Clinic (David Jardine and Jackie Mein); Townsville General Hospital (Robert Norton); Toowoomba Health Clinic (John Hooper).
In the Australian Capital Territory: Canberra Sexual Health (Frank Bowden, Sarah Martin, Vanita Parekh and Alexandra Tyson); Interchange General Practice (Janelle Hamilton, Tuck Meng Soo and Claire Willington).
In Victoria: Alfred Hospital (Kate Cherry, Ben Cowie, Suzanne Crowe, Anne Drake, Phillip du Cros, Lloyd Eisendal, Beng Eu, Andrew Fuller, Margaret Hellard, Ben Howden, Jenny Hoy, Stephen Kent, Chris Lemoh, Sharon Lewin, Lyn-Li Lim, Graeme Maclaren, Anne Mijch, John Mills, Richard Moore, Norman Roth, Joe Sasadeusz, Kasha Singh, Denis Spelman, John Spicer, Ivan Straton, Alan Street, Irani Therarajan, Olga Vukovic, Steven Wesselingh, Ian Woolley and Edwina Wright).