This study concerns the safety or risk of certain sexual practices among homosexually active men in regular or committed relationships. It revisits an HIV prevention strategy, ‘negotiated safety’ — a strategy where sexual partners in an HIV-seronegative concordant regular relationship agree to dispense with condoms for anal intercourse within their relationship while, at the same time, negotiating an agreement about sex outside the regular relationship. The major aim of the study is to test the safety of such negotiated agreements.
The notion of negotiation in the context of HIV risk avoidance was signalled in 1992 by Davies  and the term ‘negotiated safety’ was coined in 1993 by Kippax et al. . The argument advanced then and now is that dispensing with condoms is safe if the sexual partners: are in a regular relationship; are HIV antibody-negative and aware of each other's negative antibody status; and have reached a clear and unambiguous agreement about the nature of their sexual practice both within and outside their relationship, such that any sexual practice outside their relationship is safe, that is, precludes the possibility of HIV transmission. (Although it is acknowledged that other practices, such as unsafe drug injecting, may transmit HIV, this study concerns itself only with sexual transmission.)
There has been a great deal of debate, with some researchers [2,3] arguing that dispensing with condoms within a relationship should not be confused with relapse, whereas others have referred to negotiated safety as ‘negotiated danger’ . Some researchers have overlooked the necessity for negotiation of an agreement and have assumed that negotiated safety applies only within the context of a mutually exclusive HIV-seroconcordant monogamous relationship . Other researchers [4,6] have failed to recognize that negotiated safety depends not only on the concordance of HIV antibody, negative status but also on the negotiation of a contract or ‘clear agreement’ between the men with regard to sexual practices both inside and outside the regular or committed relationship. The term ‘negotiated safety’ has been misapplied by Dawson et al. , who blur the very important distinction between regular and casual relationships. Not all negotiation is safe and between casual partners can be particularly fraught.
The above debate has confused two sets of conditions — the necessary (or analytical) and the contingent (or empirical).
With regard to the analytical conditions, it is impossible for HIV transmission to occur between two HIV antibody-negative men. In principle, two HIV-negative men who eliminate all possibilities for HIV infection from outside their relationship are safe — they cannot infect each other. If such transmissions did occur, then we would indeed be witnessing what Stall and Ekstrand  refer to as ‘immaculate infections'.
The necessary relationship between HIV-negative seroconcordance and no transmission has been confused with the empirical. The empirical questions concern the certainty of knowing a partner's HIV-negative serostatus and the feasibility of two (or more) homosexually active men in a seronegative partnership eliminating all risks of HIV infection from outside.
There are a number of empirical tests of the safety or otherwise of a negotiated safety HIV avoidance strategy. The incidence of seroconversion in men who are in regular relationships where negotiated safety agreements are in place provides the strongest empirical test of negotiated safety. Another empirical test, albeit less powerful, is the frequency of unprotected anal intercourse outside HIV-seronegative regular relationships, within which unprotected anal intercourse may be practised. Unprotected anal intercourse outside regular relationships and factors associated with such unsafe sex are the focus of this study.
This study revisits negotiated safety using data from a recently recruited cohort of homosexually active men in Sydney, Australia (n = 1037). As the cohort has not been in existence long enough to apply the most powerful test of negotiated safety (the incidence of serocon-version) it focuses on men in regular relationships and examines the contexts in which men in regular relationships engage in unprotected anal intercourse (safely and unsafely). The test of the negotiated safety agreement examined in this study is the frequency of the occurrence of unprotected anal intercourse with casual partners outside the regular seroconcordant HIV-negative relationship.
A number of factors are likely to be important predictors of the occurrence of unprotected anal intercourse with casual partners outside such relationships. The factors examined in this study are as follows.
(i) The sexual partners’ knowledge that each is HIV antibody HIV-negative; (ii) the presence of an agreement between the partners, particularly about casual sex outside the relationship; (iii) the nature of the safety agreement; (iv) the nature of sexual practice within the regular relationship; (v) the length of the committed or regular relationship; (vi) the importance of anal intercourse; and (vii) the acceptability of condoms.
The first two factors were chosen as they are part of the meaning of negotiated safety. Awareness of one's own as well as one's partner's HIV-negative status is partly dependent on access to HIV testing. Patterns of testing vary from country to country and across subpopulations, so whereas testing among homosexual men in Sydney rose from 70% in 1986/1987 to 87% in 1991 , in a non-clinic sample obtained in England and Wales in 1988/1989 it was only 54% , and was 63% in a sample of homosexually active men recruited in England in 1991/1992 .
With the exception of our earlier study , few (if any) studies have reported on the presence of agreements reached between men in factor ii. Very few, if any studies, have reported on the nature of safety agreements in factor iii, or on the impact of factors iv, v and vi. With regard to the fourth factor many studies have found that unprotected anal intercourse is far more common in regular relationships than in casual partnerships independently of HIV test status or the presence of agreements [2,5,6,9].
The data reported here derive from the first two waves of men recruited into a cohort study (the Sydney Men and Sexual Health study) which focuses on the sexual practices of homosexually active men . Recruitment took place between November 1992 and February 1995. Men who had had any sexual contact with another man during the 5 years prior to recruitment and who either lived up to a 1 h car/train journey from Sydney, or regularly participated in Sydney gay community life were eligible for entry into the study.
The men were interviewed in a face-to-face setting at a time and place of their choice. The questionnaire was lengthy and took around 90 min to complete. The major topics covered in the questionnaire included sexual practice, contexts of sexual practice (including the nature of the sexual relationship), importance of anal intercourse, agreements between the men with regard to safe sexual practice, identification with the gay community, attitude to condoms, serostatus, and contact with the epidemic.
Variables were operationalized as follows:
I. ‘Agreements’ were assessed in terms of men's response to two questions: ‘Do you have a clear agreement with your regular partner(s) about anal sex within (and outside) your relationship(s)?’ Respondents who answered in the affirmative to either of these questions were asked, ‘What is that agreement?’
II. ‘Importance of anal intercourse’ was measured by a single item: ‘How important is anal intercourse as part of sex with men to you?’
III. ‘Attitude to condoms’ was measured in terms of three statements concerning their acceptability.
IV. ‘Gay community identification’ was measured in terms of a single question: ‘Do you see yourself, personally, as being part of the gay community?’ (v) ‘Contact with the epidemic’ was measured in terms of four items concerning knowing people who had died of AIDS and caring for those with HIV.
Homosexually active men were recruited using a variety of strategies including appeals through the gay media, posters, fliers and reply-paid contact cards distributed at gay venues, snowballing techniques, direct referrals from medical practitioners, and recruitment from earlier studies. At the time of interview, just under one-half of the total sample of 1037 men were in a regular relationship, and about one-half these men were in an HIV-negative seroconcordant relationship (23.3% of the total sample). As the test of safety is on sexual practice with casual partners outside the regular relationship and all data with regard to sexual practice refers to the 6 months prior to interview, the current analysis focuses on men who had been in a relationship for at least 6 months. Thirty-four per cent of the total sample (n = 354) had been in a regular relationship for 6 months or more.
Of these 354 men, 92% self-identified as gay or homosexual. Most of the men felt part of or identified with the gay community and most of them lived in an area referred to as gay Sydney or lived close to it (inner and eastern Sydney). The majority of men in the study were well educated with over one-half of the sample having received some form of tertiary education, and over 52% were engaged in professional or managerial occupations. The age of the men ranged from 17 to 69 years and the mean age was 34 years (Table 1).
A number of bivariate analyses were used to examine the following variables for their effect on the practice of unprotected anal intercourse with casual partners: demographic; attitudinal; sexual practice within regular relationships; context; and agreements reached between the men. A nominal alpha level of 0.05 was used. In the bivariate analyses, many of the variables that distinguished between those who engaged in unprotected anal intercourse with casual partners and those who did not were themselves interrelated. In order to investigate which of the variables were independently associated with the practice of unprotected anal intercourse with casual partners, a logistic regression analysis was carried out.
Seroconcordance and sexual practice
The first condition of negotiated safety is that the men are in an seroconcordant HIV-negative regular relationship. Of the 354 men in regular relationships, 181 reported being in a seronegative concordant regular relationship. These 181 men did not differ with respect to age, education, place of residence, or length of relationship from the 173 men in regular relationships not marked by HIV-negative seroconcordance (Table 1). The 181 men in seroconcordant HIV-negative relationships meet the first criterion of a negotiated safety strategy.
Within their regular relationship, 61.9% of these 181 men had engaged in unprotected anal intercourse at least once. When compared with men in non-concordant relationships, which we define here as discordant or unknown (where either one or both partners had unknown HIV status), the proportion of men engaging in unprotected anal intercourse within their relationship was greater in both for HIV-positive and -negative concordant partnerships (Table 2).
Many of the men in regular relationships also engaged in sex with casual partners. When the casual sexual practices of the men are examined, the following pattern emerges (Table 3). A small number of men in regular relationships, including those in concordant HIV-seronegative relationships, engaged in unprotected anal intercourse with their casual partners. Some other men had no casual partners, others did not engage in anal sex, whereas others engaged in only protected anal intercourse.
Of the men in HIV-seronegative concordant relationships, only 8.8% had engaged in unsafe sex with a casual partner. This is lower, but not statistically significantly lower, than the proportion of men engaging in unprotected anal intercourse with casual partners in seropositive or serodiscordant relationships (Table 3). The remainder of the men in seronegative concordant relationships had either not engaged in sex outside their relationship (39.2%) at least in the 6 months prior to interview, or their sexual behaviour outside their regular relationship was safe, that is, they had not engaged in anal intercourse (24.9%) or they had engaged only in protected anal intercourse (27.1%).
The strategy of dispensing with condoms within a regular sexual relationship, particularly if it was seroconcordant, was common in this Sydney cohort. The questions of whether it is a safe strategy, and what factors, if any, reduce the risk of HIV transmission were addressed with reference to the 181 men in HIV-negative seroconcordant regular relationships of 6 months or more duration. We also sought to determine what distinguishes the 165 men who did not engage in unprotected anal intercourse with a casual partner from the 16 men who did.
Negotiated safety test
In order to determine which men had successfully avoided the risk of HIV transmission from outside their relationship by not engaging in unprotected anal intercourse with a casual partner, a number of factors were examined. These factors, as discussed above, were as follows: the presence of an agreement between the partners, particularly about casual sex outside the relationship; the nature of the safety agreement; the nature of the sexual practice within the regular relationship; the length of the committed or regular relationship; the importance of anal intercourse; and the acceptability of condoms. A number of demographic variables was also examined.
None of the demographic variables was significantly associated with the practice of unprotected anal intercourse with casual partners, although ‘region’ approached significance (Table 4). Men who lived in what is generally known as ‘gay Sydney’ were more likely to engage in unprotected anal intercourse with casual partners than those who lived elsewhere (P = 0.06, two-tailed Fisher's exact test). The practice of unprotected anal intercourse with casual partners was not associated with ‘contact with the epidemic’.
Presence of an agreement about sex with casual partners
Of the 181 men in regular HIV-negative concordant relationships of 6 months or more duration, 80% had an agreement with their regular partner about the nature of their sexual practice outside their relationship with casual partners. Agreements were important: having a safety agreement (the second condition for negotiated safety) was predictive of ‘safer’ sex when compared with no agreement at all (Table 4).
Nature of the safety agreement
Not only was the presence of an agreement important, but the nature of that safety agreement was also predictive of the successful avoidance of unprotected anal intercourse with casual partners. The best agreement with regard to safe sex with casual partners was ‘no anal sex’: men who had this agreement with their regular partners did not engage in unprotected anal intercourse with casual partners (Table 4).
Sexual practice within the relationship
Whether the men engaged in anal intercourse with their regular partners and whether such sex was protected were not predictive of the safety of their anal intercourse with casual partners (Table 4).
Length of relationship (of over 6 months) appeared to have no effect on the practice of unprotected anal intercourse with casual partners (Table 4).
Strength of feelings about anal intercourse
Respondents who rated anal intercourse as very important were more likely to engage in anal intercourse with casual partners, but were not more likely to engage in unprotected anal intercourse than men who rated anal intercourse as only moderately or not very important (Table 4).
Attitude to condoms
On the other hand, attitude towards condoms as measured in terms of their acceptability did distinguish between the men. Men who found condom use acceptable were more likely to avoid unprotected anal intercourse with their casual partners (Table 4).
A logistic regression analysis including the above factors was carried out. Model reduction using backwards elimination resulted in a reduced model, as shown in Table 5. The presence of an agreement was the most important variable distinguishing between those men who were successful in their avoidance of unprotected anal intercourse with casual partners and those who engaged in unprotected anal intercourse. This variable accounted for 10% of the deviance when fitted alone.
The odds ratios were < 1.0 because unprotected anal intercourse with casual partners is less likely as the acceptability of condoms increases, and less likely where an agreement about sex outside the regular relationship is in place than where there is no agreement. Compared with having no agreement, an agreement to have anal intercourse only when using condoms does not significantly reduce the probability of unprotected anal intercourse with casual partners, whereas an agreement to have no sex outside the relationship or to have no anal sex outside the relationship significantly decreases the probability of engaging in unprotected anal intercourse with casual partners.
The findings of this study confirm the findings of others [2,5,6,9] that many men engage in unprotected anal intercourse within their regular relationships. This study shows that this is particularly true of men in seroconcordant relationships, whether HIV-positive or negative. The findings of this study also indicate that such a strategy may not be risky for men in HIV-seronegative regular relationships when a number of mutually agreed conditions are fulfilled — namely those that constitute an appropriate agreement about the nature of sex outside the regular relationship which is entered into by the partners in that relationship.
The avoidance of unprotected anal intercourse with casual partners was significantly related to and dependent on the presence of a safety agreement reached between the men. Men in regular relationships where agreements such as ‘no anal intercourse with casual partners’ are in place are at low risk of HIV infection. Attitudes to condom use were also important to the safety of the negotiated strategy, unacceptability of condoms being likely to work against the success of a negotiated agreement.
The avoidance of unprotected anal intercourse with casual partners was, however, independent of the major demographic variables, including age and education level reached. Nor was there a statistically significant difference between men living in gay Sydney and men from all other regions. Likewise, degree of contact with the epidemic, relationship length and feelings about the importance of anal sex were also unrelated to the success of negotiation. Given the small sample size and the associated lack of power, some caution is needed here. Although not statistically significant, unprotected anal intercourse was more likely among men who lived within the areas commonly associated with gay community in Sydney.
Further studies with larger samples are needed. Such studies should be prospective and should, ideally, include both parties in any negotiated safety agreement. Prospective studies examining seroconversion rates among those who had adopted negotiated safety agreements would provide the strictest test of the safety of such a strategy.
Negotiated safety is not a rare or uncommon strategy but a strategy adopted by a significant proportion of men, at least among this cohort of homosexual men, and the findings of this study must be placed in the context of a highly gay-identified Sydney sample with a strong gay community supporting it. Moreover, the results occurred in a context of high levels of testing. In the post-AIDS era, a term coined by Dowsett , the adoption of a negotiated safety agreement may work to support long-term maintenance of safe sex among homosexual men. For many homosexual men, the strategy of ‘condoms always’ may be difficult to adopt or sustain. Negotiated safety provides one strategy that may help a significant number of men maintain their own and their regular partners' HIV-negative serostatus. It is a strategy that appears to work equally well for men in regular relationships and with no casual partners (monogamous relationships), as for men who are in committed or regular relationships who have sex with casual partners (regular but not monogamous relationships). The strategy that works best for these latter men is to forgo anal intercourse with their casual partners.
It is important to note, however, that the negotiated safety strategy is not entirely free of risk. For men in non-monogamous relationships who make agreements about protected anal intercourse with casual partners there is a risk: a small proportion of the men in this study failed to keep this agreement. Furthermore, although not examined or discussed in this study, there may be a very small risk associated with agreements that do not discourage oral-genital intercourse with casual partners. It is, however, inappropriate to apply the term ‘negotiated danger’ to the strategy as a whole.
Proper discussion of negotiated safety has been stifled because there have been few empirical studies that allow researchers to distinguish the strategy of dispensing with condoms for anal intercourse within a regular relationship from the practice of unsafe anal intercourse, whether such behaviour is seen as a relapse into unsafe practice or the adoption of unsafe practice. The erroneous assumptions that negotiated safety can only be adopted within a monogamous relationship or that it can be negotiated between casual as well as regular sexual partners have also inhibited a full discussion.
Despite lack of a reasoned debate, a significant number of homosexual men are engaging in negotiated safety. Findings from a number of studies [2,5,6,9] show that ‘large’ minorities of men dispense with condoms within regular relationships. There is also some evidence that HIV transmission is occurring within regular relationships . So although the findings of the present study indicate that negotiated safety agreements can be and are kept, they clearly are not kept on all occasions or all the time. The strategy may be more successfully and widely used if well-funded education campaigns that deal with the issues of honesty, testing, trust, and talk between men are implemented. Agreements reached between men must be clear and unambiguous and the trust must not be misplaced.
Although not dealt with in this study, some men in seroconcordant HIV-positive regular relationships are also dispensing with condoms. This strategy is also in need of empirical examination and discussion, particularly because its adoption carries the risk of transmitting infection between sexual partners both within and outside any regular seroconcordant HIV-positive relationship.