HOXWORTH, TAMARA PhD*; SPENCER, NANCY E. MSPH*; PETERMAN, THOMAS A. MD, MSc†; CRAIG, THELMA BS*; JOHNSON, SOLWAZI BS*; MAHER, JULIE E. PhD‡
WITHOUT CURES OR VACCINES for HIV, prevention depends on the practice of safer behaviors by infected and at-risk persons and on the use of effective targeted prevention interventions. Partner notification (PN) is one such intervention. PN began in the United States in the 1930s as a method for controlling the spread of syphilis. PN for HIV involves informing the sex and needle-sharing partners of HIV-infected persons about their potential exposure, counseling them on how to reduce risk of exposure, and offering healthcare services, including testing and treatment. PN provides a way to identify and target persons at high risk of acquiring or transmitting infection for risk-reduction counseling. 1–4
A number of studies have examined the efficacy of HIV PN in identifying, locating, testing, and counseling the partners of HIV-infected persons. 1,2,5,6 A few studies have suggested that PN can reduce high-risk behaviors. Wykoff et al 7,8 found that notified partners decreased their mean number of sex partners (80–82% reduction in positive partners; 50–54% reduction in negative partners) and increased condom use (to about 80% by positive partners and 69% by negative partners). Pavia et al 9 found the incidence of sexually transmitted infections (STIs) among partners decreased following notification (from 19.4 to 2.8 STI episodes per 1000 person-years), but the difference was not statistically significant.
Few studies to date have examined how PN influences partnerships of infected persons and their partners. 3,10 However, if HIV PN increases partnership dissolution and new partnership formation, it could increase HIV transmission in the community. One study examined the effect of serostatus disclosure on relationships. 11 Schnell et al 11 evaluated the disclosure of HIV status to a main sex partner and the impact on the relationship for men who have sex with men. Of the 89% who disclosed their serostatus to partners, 70% of the seronegative men and 82% of seropositive men reported that their relationship remained as strong as ever after 6 months. For seronegative men, those who did not disclose were more likely to be single at the return visit. This pattern was similar but not statistically significant for the small number of seropositive men.
To determine if PN influences HIV-risk behaviors and changes in partnerships, the current study examined dissolution of sexual partnerships, formation of new sexual partnerships, and changes in sexual behaviors in ongoing partnerships after PN.
Recruitment for this study targeted three groups of people: (1) index persons who were HIV-infected who had been interviewed to identify their partners for notification; (2) partners who were notified by the health department that they had been exposed to HIV; and (3) controls who were at high risk for HIV infection and who received negative test results and counseling at a metropolitan Denver HIV counseling and testing site. Five subjects originally enrolled as partners tested HIV-positive and were reclassified as index persons after they were interviewed to identify their partners for notification. Recruitment was conducted from September 1998 through August 1999, and follow-up interviews were completed by the end of April 2000.
Subjects were recruited from the following pools of potentially eligible persons: 254 HIV-infected persons who were provided PN and 108 partners who were notified in the metropolitan Denver area during the recruitment period, as well as approximately 1200 HIV-negative high-risk persons who presented to the counseling and testing site during the recruitment period. To enroll similar numbers of subjects in each study group, twice a week for half a day (the day of the week was varied systematically) during the study period when indexes and partners were being enrolled, a study staff identified control persons eligible for participation in the study.
Eligible subjects were aged 18 years or older, English- or Spanish-speaking, and not incarcerated; had practiced vaginal or anal sex within the past 90 days; and had received PN (for index persons and partners) or HIV counseling services (for controls) no more than 30 days before administration of the baseline questionnaire. Potential study subjects were first contacted by a health worker, who forwarded the names of eligible and interested individuals to a study interviewer. The interviewer called them to explain the study in greater detail and invite them to enroll. If the individual wanted to enroll, an interview was scheduled, an informed consent form was signed, and a baseline interview was conducted.
Interviewers administered 30-minute questionnaires at baseline and at 3 and 6 months. The baseline questionnaire captured information on behaviors occurring during the 3 months before PN or (for the control group) 3 months before their HIV test and risk-reduction counseling. Follow-up questionnaires asked about behaviors within the previous 3 months. Interviews were conducted at a location selected by the subject (e.g., home, office, park, health department office, or clinic). The same study interviewer conducted both baseline and follow-up interviews when possible. Questions covered sexual activities (i.e., type, frequency, and condom use), substance use (i.e., alcohol, marijuana, or needle use), and partnership outcomes (i.e., continuing and new sexual partnerships, abstinent partnerships, and dissolved partnerships). Subjects identified all sex partners within the previous 3 months and answered many questions about each partnership. For confidentiality purposes, subjects’ sex partner information was recorded on questionnaires with use of only first names, nicknames, code names, or initials. To confirm that the sex partners reported on questionnaires were the same ones seen by state health workers, name identifiers, age, birth date, date of last sexual encounter, and partnership duration were also used.
The PN given before the study to index persons and their partners was conducted according to the Colorado Definitions, Standards, and Criteria for Health Department Delivery of Partner Counseling and Referral Services. 12 To ensure confidentiality, when partners were notified about their possible exposure they were not told the name of the index patient who reported having contact with them. Similarly, index persons were not told which partners were reached during the notification process. Some notified partners had only one sex partner who could be the index person, others may have learned the identity through discussion with their partners, and still others may have guessed. Our classification was based on comparison of the nickname or initials reported in the interview with the records maintained by the PN staff.
Formative evaluation and pilot testing were conducted for enrollment methods, questionnaires, props (e.g., calendar and response cards), incentives, and data collection methods. Study interviewers had prior training and experience in discussing sensitive information in an open, honest, and discreet way. Subjects were assured that study information was confidential, was protected by physical security and Colorado law, and would not be shared with the health worker who had provided the PN. At the conclusion of each interview, subjects received a $15 certificate to a local grocery store to compensate them for their time and effort. This study was approved by institutional review boards at the Centers for Disease Control and Prevention, the Colorado Department of Public Health and Environment, and Denver Health and Hospitals.
Most of the analyses focused on what happened to partnerships that were reported at baseline; however, characteristics of individual subjects at baseline and their numbers of and total new sex partners over time were also examined. Analyses of partnerships examined partnership dissolutions and sexual behaviors practiced within the partnership. The analyses of sexual behaviors within partnerships used only partnerships with a baseline and at least one follow-up observation. When there were two follow-up observations, the last observation was analyzed.
Partnerships were classified into five groups according to the type of subject (listed first) and the type of partner reported by the subject (listed second): (1) index persons describing partnerships with partners who were notified (index with notified partner); (2) index persons describing partnerships with partners who were not notified (index with nonnotified partner); (3) notified partners describing partnerships with an index person (notified partner with index); (4) notified partners describing partnerships with people who were not index persons or notified partners (notified partner with other person); and (5) persons receiving testing and counseling at an HIV testing site reporting their partnerships (controls). Thus, there were some partnerships in which both persons had received PN; some were described by the index, others by the partner. When both an index and the partner were enrolled in the study, to avoid duplication, only the index's description of the partnership was included.
Chi-square tests were conducted to assess subgroup differences in proportions of partnership dissolutions, and analysis of variance was used to assess subgroup differences in the mean numbers of new and total sex partners. Both tests were used to analyze sexual behaviors.
To identify dissolved partnerships, at each follow-up, subjects were asked specifically about the partners they had named in the previous interview: “In the last 3 months, did you have sex with _____?” If the answer was yes, the interviewer proceeded to ask questions about sexual behaviors and relationship factors regarding that partnership. If the answer was no, the interviewer asked if the subject still did any socializing with the partner. From these answers, relationships were coded as continuing sexual, continuing but nonsexual, or dissolved.
To study the sexual behaviors in partnerships, the analysis focused on the sexual behaviors of partnerships reported at baseline but not specific sexual behaviors of new partnerships that formed after baseline. Partnerships were classified as (1) having had vaginal or anal sex, (2) having had oral sex only, or (3) abstaining from vaginal, anal, and oral sex. For partnerships that had vaginal or anal sex, subjects were asked whether condoms were always, sometimes, or never used; the proportion of sex episodes that were protected; and the number of sex episodes that were unprotected. Measures of condom use were considered indicative of behavior, whereas the number of unprotected sex acts was considered a better indicator of risk of exposure.
Between September 1998 and August 1999, 830 individuals were screened for study eligibility, and 556 (67%) were eligible (Table 1). Sixty-one percent of those ineligible reported no vaginal or anal sex in the prior 3 months. Of those eligible, 260 (47%) were interested in talking to a study interviewer about possibly enrolling, 276 (50%) declined enrollment, and 20 (4%) were deemed unsuitable by the screener on the basis of concerns about mental health status, intoxication, or impaired cognitive function. Of the 260 who expressed interest in the study, 58 (11% of eligible) later refused or were never located. A total of 202 (38% of eligible) individuals enrolled in the study. Persons were more likely to enroll if they were index persons (62%) or notified partners (65%) rather than controls (26%). At least one follow-up interview was completed by 165 (82% of those enrolled). Those 165 were more likely than eligible persons to be female (28% versus 6%;P < 0.001), people of color (46% versus 20%;P < 0.001), married (14% versus 5%;P = 0.001), and heterosexual (39% versus 16%;P < 0.001).
The 165 subjects reported a total of 293 partnerships at baseline. Nine partnerships were reported twice: once by an index patient and once by the notified partner. The partners’ reports were deleted, leaving 284 unique partnerships at baseline. Of those, 125 partnerships had sex after PN or counseling and testing (CT) or were reported as “still involved” at the follow-up visit. Those 125 partnerships were used in analyses of sexual risk behaviors in ongoing partnerships. Of the 125 partnerships, 19 were index persons describing a partnership with a notified partner, 7 were notified partners describing a partnership with an index patient, 17 were index persons describing a partnership with a nonnotified partner, 8 were notified partners describing a partnership with someone other than the index patient, and 74 were control partnerships.
Index persons and controls were more likely than notified partners to be male, unmarried, and homosexual (Table 2). Index persons and notified partners were more likely than controls to be people of color, have only one sex partner, and have a history of receiving PN.
Regardless of who was describing the relationship, partnerships between index persons and notified partners were more likely than partnerships with nonnotified persons to involve living together and to be reported as main partnerships (Table 3). Notified partners who described their partnerships with index patients were more likely than other groups to be in partnerships that had children. Partnerships between index persons and nonnotified partners and partnerships reported by controls were of shorter duration than the other partnerships.
At baseline, the 165 subjects reported having 284 unique partnerships in the 3 months before notification (or testing, for controls). By the 6-month interview, 185 partnerships (65%) had dissolved (Table 4). Most (64%) of the dissolved partnerships had already dissolved before PN or HIV testing and counseling (88 of those were reported as “one-night stands”). Regardless of who was describing the relationship, dissolutions between index persons and notified partners (33%–36%) occurred less frequently than dissolutions in other partnerships (65%–81%). However, since most dissolutions occurred before PN or CT, the differences mainly reflect differences present at baseline. Partnership dissolution after PN or CT was similar for all groups. None of the dissolutions that occurred after PN were reported to be caused by the PN experience, but eight (that dissolved by 3 months) were said to be influenced by fears and mistrust related to HIV status.
During follow-up of ongoing partnerships, there were no major differences observed between groups in the percentages that had vaginal or anal sex (Table 5). However, condom use during every episode of vaginal or anal sex was reported at follow-up more frequently by index persons with notified partners (80%) and notified partners describing their partnership with an index patient (100%) than by index persons with nonnotified partners (50%), notified partners with other persons (38%), or controls (30%). The proportion of partnerships in which condoms were never used actually increased slightly for index persons with nonnotified partners and controls. Three index persons (16%) with notified partners reported vaginal or anal sex with zero condom use; all were in seroconcordant partnerships of two HIV-infected persons.
At baseline, there were no differences in the mean proportion of vaginal and anal sex episodes in which condoms were used. However, at follow-up, more episodes were protected for index persons with notified partners (85%) and notified partners reporting partnerships with index persons (100%) than in other groups (P = 0.002). Also at follow-up, there was generally less unprotected sex reported in partnerships involving an index patient, although these differences were not statistically significant.
There was no evidence that PN led to new partnership formation (Table 6). All groups averaged less than one new sex partner during the follow-up period. All groups but one reduced their numbers of total sex partners. Controls had the highest mean numbers of new (0.8) and total sex partners (1.5) at follow-up, but differences were not statistically significant.
This study examined the effect of PN on subsequent sexual behaviors and partnership dissolution and formation. Partnerships where both persons received PN were less likely to break up or acquire new partners and more likely to use condoms at follow-up.
Before or immediately after PN or CT, there were high numbers of dissolved partnerships in all five groups (159; 56%), with fewer in partnerships where both parties received PN. However, partnerships in which both persons received PN were more likely to involve main partners living together and to be of longer duration, which may have contributed to the lower frequency of dissolutions. After PN or CT, the frequency of dissolutions lessened and there were no differences between subgroups. It appears that the experience of PN did not bring about any more dissolutions than would be expected in a comparison group.
Partnerships between index persons and notified partners were more likely than other partnerships to change sexual behaviors that might transmit HIV. Changes in the percentages of partnerships reporting any vaginal or anal sex were similar across subgroups. However, at follow-up, only three of 26 partnerships between index persons and notified partners reported sex without a condom, and these were all partnerships in which both partners were already infected. Index persons were less likely to increase condom use with nonnotified partners, but the average number of unprotected sex acts reported for this group decreased from 2.7 at baseline to 0.4 at follow-up. This suggests that some infected persons changed their behaviors to reduce transmission, even if their partners were not notified.
PN did not appear to increase new partnership formation. Controls reported the most new and total sex partners, which may be related to the finding that controls at baseline were less likely than index persons or notified partners to be in partnerships with main partners, to be married, and/or to be living together. Four groups, including controls, reduced their total numbers of sex partners, but this may represent regression to the mean because subjects were selected after they had received PN or CT. Thus, at baseline, they may have been more sexually active than usual.
Several challenges were encountered in conducting this research. First, recruitment required the effort and collaboration of outside staff members, including disease intervention specialists, HIV test site counselors, and disease registry staff. The 82% follow-up rate is a tribute to the interviewers’ ability to locate subjects and maintain their ongoing trust. Second, the classification of partnerships required time-intensive record-searches to verify that the sex partners reported on questionnaires were the same ones seen by state health workers. This was difficult because, for confidentiality purposes, subjects’ sex partners were recorded on study questionnaires by only first names, nicknames, code names, or initials. We believe our links are accurate on the basis of these identifiers and other linking variables, including age, birth date, date of last sexual encounter, and partnership duration.
Third, the ethical implications of withholding services to HIV-infected persons constrained our choice of control group. Instead of using HIV-infected persons who had received delayed or no PN as control subjects, we used HIV-negative persons receiving testing and counseling. Although we attempted to select a comparison group highly similar to our index persons, subgroup differences in study outcomes may have been influenced by differences in race, partnership commitment, baseline levels of risk behavior, and HIV infection status. Fourth, subjects were different from persons who declined enrollment in gender, marital status, and sexual orientation, and the resulting study sample was a small fraction of all persons receiving PN or HIV counseling and testing. Finally, although our study screened 352 index persons or partners, we were able to enroll only 103 and follow-up on 86, a relatively small cohort in which to study changes in sexual behavior.
Despite the limitations, this and a similar study 10 are the first to examine the effects of PN on both partnership dissolutions and sexual behaviors. We conclude that PN was not associated with increases in partnership dissolutions or acquisition of new sex partners. Partnerships between index persons and notified partners were, following PN, more likely to involve use of condoms than were other partnerships. These findings suggest that PN can help reduce HIV transmission within the community.