TOOMEY, KATHLEEN E. MD, MPH*; PETERMAN, THOMAS A. MD, MSc*; DICKER, LINDA W. PhD*; ZAIDI, AKBAR A. PhD*; WROTEN, JACK E. BS†; CAROLINA, JERRY BA‡
SURGEON GENERAL Thomas Paran promoted partner notification in 1936 for the control of syphilis in the United States.1,2 Anecdotal evidence suggests that partner notification can be effective in reducing the transmission of sexually transmitted diseases (STDs). Partner notification has been credited with decreasing the incidence of gonorrhea by 13%,3 and partner notification efforts are commonly intensified in response to an increase in syphilis.4 Although considered a cornerstone of care for syphilis and gonorrhea,1,2 the effectiveness of partner notification for human immunodeficiency virus (HIV) infection is less certain because partners are not cured and intervention with high‐risk partners may not necessarily reduce further transmission.5 Partner notification for HIV infection was instituted in part because of a “duty to warn,” as a means of identifying infected persons and in hopes that partners might change behavior to avoid acquiring or transmitting HIV infection.6 Although there is still no cure for HIV infection, the benefits of early intervention and new drug therapies make the rationale for partner notification and the identification of new HIV infections more compelling.
Little is known about the cost or effectiveness of different partner notification strategies. A recent review of the evidence found seven studies that compared different approaches to partner notification for STD,7 including one for HIV. In that randomized trial for HIV, only 46% of the eligible HIV‐positive index patients participated.8 Provider referral for 39 index patients located 78 partners (2 per index patient); patient referral for 35 index patients located 10 partners (0.3 per index patient). The study provided no information on the relative costs.
Our study was designed to compare four HIV partner notification strategies by measuring the cost and effectiveness of each strategy to locate and test partners. The strategies ranged from instructing patients to notify partners themselves to immediate health department notification, including taking blood samples in the field if necessary. However, our attempt at a randomized controlled trial was unsuccessful because of frequent crossover of partners into a strategy other than the one assigned by randomization. Here, we discuss the problems encountered with the trial and our analysis of the cost and effectiveness of provider referral for HIV infection.
We studied all persons referred for partner notification because of a new diagnosis of HIV infection in three areas (Broward County [Ft. Lauderdale] and Tampa, FL and Paterson, NJ). Patients who obtained their HIV diagnosis outside the three participating STD clinics were included in the study only if they were referred to one of the clinics for partner notification. A concurrent study of partner notification for syphilis was conducted in these clinics; that study is reported elsewhere.9
Trained disease intervention specialists interviewed patients who had a positive test for HIV to identify sex partners possibly exposed to HIV in the preceding year. Partner notification was to be carried out by one of four methods. (1) “Patient referral” involved counseling and role play about how the patient should notify their partners and advise them to visit the clinic for testing. Patients were given referral cards and asked to have their partners present them when they came to the clinic. (2) “Contract referral” involved contracting by disease intervention specialists with the index patients to notify named partners within 3 days; patients were told that after 3 days the disease intervention specialist would notify any partners that had not yet presented to the clinic. (3) “Provider referral, field notification” immediately involved the disease intervention specialist, who would collect names and identifying information about named partners, find them in the community, and ask the partners to come to the clinic for HIV testing. (4) “Provider referral, field blood” was similar to “provider referral, field notification” except that the disease intervention specialist could take a blood sample in the field, thus eliminating the need for a clinic visit by the partner.
The main objective was to compare the costs and the numbers of partners who came for HIV testing under each strategy. A secondary objective was to evaluate the cost and effectiveness of partner notification for index patients who had differing demographic characteristics.
Persons who had both syphilis and HIV were assigned to intervention methods 2 to 4 according to the randomization strategy for the syphilis study.9 Persons with HIV infection alone were randomly assigned to intervention methods 1 through 4.
Data were evaluated in an ongoing way to ensure consistent randomization. Early in the study, resistance to randomization among staff was identified by a break in randomization for specific categories. This occurred when the disease intervention specialist thought a particular strategy might work better for a given index patient. When the data were analyzed further, we detected larger than expected crossover between study arms; no arm represented what would have occurred if the health department had adopted that strategy exclusively. This crossover reflected the fact that often partners were named by multiple patients and therefore had been previously notified by other investigators through a different strategy. We therefore eliminated from further analysis all data from intervention method 1 (patient referral) and combined data from methods 2 through 4 as an assessment of provider referral partner notification.
Costs were considered under three broad categories: interviewing index patients, interviewing partners, and program overhead. Index‐patient interview costs were based on the minutes the disease intervention specialist spent in posttest counseling, reinterviewing (if necessary), and travel time plus mileage costs to reinterview. Partner interviewing costs were based on minutes the disease intervention specialist spent searching records, telephoning partners, and searching or counseling in the field. The cost of the disease investigator's salary per minute was determined by dividing the investigator's annual salary by 47 weeks per year, 5 days per week, 8 hours per day, and 60 minutes per hour. Mileage costs for travel associated with index patient and partner searches were calculated at 20 cents per mile. Mileage related to the index patients was not measured until March 1992, so mileage for index patients before that date was assumed to be the average number of miles measured for all index patients from that site. Overhead included the supervisory and support staff salaries, telephone expenses, and rent for the space devoted to partner notification. The total overhead was multiplied by the fraction of STD and HIV partner notification time that was devoted to HIV partner notification. Because we excluded subjects from the first arm from the analysis, we also reduced overhead by the fraction of disease intervention specialist time that had been spent with patients from Arm 1.
From December 1990 through February 1993, there were 1,399 patients diagnosed as having HIV infection in one of the study clinics or referred to the clinics for partner notification. The patients were randomly assigned to Group 1 (n = 329), Group 2 (n = 363), Group 3 (n = 359), or Group 4 (n = 348). Of these, 168 had both syphilis and HIV infection.
Many partners were notified by using strategies not permitted by their study arm. Although Group 1 was intended to be exclusively patient referral, 97 partners of patients assigned to Group 1 had been actively sought by a disease intervention specialist, including 84 who had a field visit. Of the Group 1 partners that were located, pregnant women were only slightly more likely to have had health department intervention than women who were not pregnant (62% of 8 pregnant partners versus 51% of 181 female partners who were not pregnant). In addition, 118 of the persons assigned to Groups 2 or 3 had blood samples taken in the field, despite the intention to limit that option to persons in Group 4.
A total of 1,070 index patients were offered provider referral: Broward enrolled 730 during 25 months (December 1990 through December 1992), Tampa enrolled 247 during 16 months (September 1991 through December 1992), and Paterson enrolled 103 during 18 months (September 1991 through February 1993). Of those enrolled, 47 (4.4%) were not interviewed about their partners because they could not be located for the interview (n = 45), they refused to be interviewed (n = 1), or there was a clinic error (n = 1).
The 1,070 index patients reported having had 8,633 total partners in the preceding year; information was sufficient to initiate a search for 1,290 (15%) of these partners (Figure 1). Nearly one half of the partners sought (n = 548) had lived with the index patient in the preceding year. The disease intervention specialists located 1,035 partners via record search or by personal contact; 248 had previously tested positive, 158 were located by record search only, and 629 were told of their exposure, including 69 who refused testing and 560 who were tested. Of those tested, 438 were negative and 122 were positive; 21 of those who tested positive had a previous negative test result and the other 101 had not been previously tested.
The disease intervention specialists spent an average of 107 minutes working with each index patient: posttest counseling (48 minutes), interview and reinterview (27 minutes), and travel time to interviews (32 minutes). The average time spent investigating partners per index patient was 90 minutes: record search (11 minutes), telephone contact time (6 minutes), and field visit time (73 minutes). The total time (recorded by disease intervention specialists) was 227,487 minutes. If the intervention specialists are considered to work 60 minutes per hour, 8 hours per day, 5 days a week, and 47 weeks per year, then the intervention specialist time required to carry out HIV provider referral amounted to two full‐time equivalents. The time spent by each intervention specialist was multiplied by that person's annual salary to arrive at the total cost of $53,393 for the disease intervention specialists.
The disease intervention specialists drove an average of 23 miles to interview the index patient and 28 miles per index patient to interview their partners. At 20 cents per mile for the 54,069 miles, the cost estimate for travel is $10,814.
We calculated the average time required to do each step in provider referral by dividing the total time spent on that step by the number of times that step was required in the investigations. Field visits, when they were done, took an average of 98 minutes (Table 1).
The total cost estimate for the provider referral in this study was $268,425: disease investigation specialists' time ($53,343), mileage ($10,814), and overhead ($204,218) (see Appendix for details on overhead). Thus, the cost was calculated to be $251 per index patient with HIV infection (n = 1,070), $427 per partner notified (n = 629), $479 per partner tested (n = 560), or $2,200 per partner newly discovered to have HIV infection (n = 122).
Variability in the costs and the likelihood of finding or testing a partner depended on the age, race, sex, and number of partners reported by the index patient, but differences were not great enough to suggest specific targeting strategies (Table 2). When the index patient was reported from a jail, partner notification took a little more time per partner tested, but this comparison is based on only 25 patients from a jail. All persons diagnosed with HIV in the study clinics were included in the study, but we do not know how many persons were diagnosed with HIV at other sources that referred patients to the clinics for partner notification.
The index patients reported up to 750 partners but could provide locating information for only a few of their partners. The number of partners located for each index patient ranged from none to 8: none (409 index patients), 1 (422), 2 (155), 3 (53), 4 (17), 5 (10), 6 (3), and 8 (1). The number of partners tested for HIV infection was lower, ranging from none to 5: none (665 index patients), 1 (295), 2 (77), 3 (22), 4 (10), and 5 (1). The number of partners with new positive HIV tests ranged from none to 2: none (953 index patients), 1 (112), and 2 (5).
We were unable to analyze our study as a randomized controlled trial because many partners assigned to one strategy received an intervention as though they had been assigned to another strategy. This crossover occurred in part because the partners of an index patient assigned to one approach may have been previously reported by another index patient and assigned to a different strategy. Our data collection system did not allow us to determine how often this happened. We could link index patients to multiple partners, but we could not link partners to multiple index patients. Many partners of index patients who were assigned to patient referral received intervention by the health department. In a few instances this might be attributed to duty‐to‐warn laws that were interpreted as mandating provider referral for spouses and pregnant women. In other instances, health department staff used whatever means necessary to notify partners because of the importance staff placed on referral for a given partner. The practice of taking blood samples in the field from all persons who “refused” to come to the clinic for testing was widely accepted by study staff, although it was not permitted in the protocol. Some thought that these persons would be analyzed as “failures” of the arm to which they were assigned, but we chose not to include this as part of the analysis because some of them might have visited the clinic a few days later if a blood sample had not been drawn in the field.
The benefits of partner notification are not easily quantified. Although partner notification found 122 persons who did not know they were HIV infected, partner notification accomplishes more than just case finding. The program notified 629 persons that one of their sex partners was infected with HIV and led to HIV testing of 560 of these exposed partners‐but partner notification is more than just counseling persons at high risk. Because the duty to warn the partners is also important, costs per person tested via partner notification programs cannot simply be compared with costs per person tested via screening or other outreach programs. However, partner notification was initiated for only 15% of the 8,633 partners reported by the index patients. Legally, duty to warn may not apply to the partners without identifying information because insufficient information was provided to locate them, but considering that 85% of partners have not been reached, the relative prevention impact of partner notification on HIV transmission in the community may be small. Cluster investigations in high‐risk settings such as crack houses may provide an opportunity to identify these unnamed high‐risk partners.
Partner notification has been criticized because some of the high‐risk partners most likely to transmit infection are least likely to be identified for notification.10 Our study supports that view: Steady partners were more likely to be sought than casual partners (584 of 1,290 partners that were sought had lived with the index patient in the preceding year). By contrast, other investigators contend that the overlap between partnerships means that most partners in the community would eventually be notified even if partner notification reached only a small percentage of each index patient's total sex partners.11 Finally, although some studies reported behavior change by partners reached via partner notification programs,12 no published studies measured the impact of partner notification on HIV transmission in the community.
Our cost estimates illustrate some of the difficulties in estimating the costs for interventions. The disease intervention specialists kept meticulous records on the amount of time spent in each aspect of partner notification. Our approach of multiplying their salary per minute times the minutes recorded for partner notification actually leads to an underestimate of the true cost because our calculations do not include the time that the disease intervention specialists spend waiting to interview a patient, attending staff meetings, or training nor did our calculations include staff time between tasks. Using the disease intervention specialists' diaries, we calculated that their total time commitment for HIV partner notification was 2 full‐time equivalents, whereas the time commitment of the overhead personnel for the duration of the study was estimated as 4.5 full‐time equivalents. During the study we estimated overhead personnel costs by asking how many management and support personnel worked on HIV partner notification. We cannot retrospectively gather the number of disease intervention specialists devoted to HIV partner notification because the number of employees and their time allocations changed frequently.
Costs for provider referral have been described in other articles. With many contributing costs and many ways of measuring cost, the estimates from different studies are not exactly comparable.13 The costs for a series of investigations starting from one index patient in South Carolina in 1987 were based on the salaries and travel costs for 450 hours of employee time during an 8‐week investigation.12 The total was $100 per partner tested or $810 per HIV‐positive partner identified. In Colorado in 1988, costs were calculated on the basis of the disease investigator's time to interview the index patient, locate and refer partners, and document their activities.14 Supervisors, travel, and supplies were included in their total of $84 per index patient or $1,625 per newly identified HIV‐positive person. In New Jersey from June 1988 through May 1989, costs included salaries for 10 persons, transportation, and office equipment, for a total of $2,260 per partner identified.15 In Utah from 1988 through 1990, costs included an estimated 60% of the staff time from the counseling and testing program, travel, training, laboratory costs, and a percentage of this total to cover office support and overhead. In that study, 279 partners were tested and 39 new infections were found, for a total cost of $373 per partner counseled or $3,205 per new infection identified.16
We found no major differences in the costs of partner notification or the likelihood of locating partners on basis of the demographic characteristics of the index patient. We do not have information on the partners for whom investigations were not initiated, so we cannot compare the likelihood of finding partners on the basis of characteristics of the partner. However, because 45% of the partners sought had lived with the index patient within the past year, we suspect that the partners from long‐term relationships are easier to find.
Other studies suggest that provider notification is superior to patient referral for notifying partners. According to the only trial of provider notification for HIV, provider referral contacted 2 partners per index patient compared with 0.3 for patient referral, and provider referral identified 0.23 infected partners per index patient compared with 0.14 for patient referral.8 Studies on syphilis and gonorrhea found conflicting results but generally favored provider referral over patient referral.7 The only cost data available were for gonorrhea; they suggested that provider referral costs four to eight times more than patent referral per partner with a positive culture. A review of the partner notification literature reported that only limited broad conclusions about the effectiveness of various approaches could be drawn from the available evidence.7
The effectiveness of HIV partner notification can be evaluated at different levels, including the number of partners notified, tested, or infected. From a public health perspective, the number of new HIV infections prevented is the most important outcome, but this outcome is not easily measured. Behavior change has been reported among persons who learn they have a positive test result.17–19 Partners who test negative may also change their behavior after notification, particularly if they receive quality client‐centered counseling.20 In one study, partners of gay men increased condom use from “no use” by anyone to at least “some use” by 69% after notification.12 Translating these reported behavior changes to actual HIV cases prevented is difficult. Comparing the value of partner notification with other prevention strategies is further complicated by the ethical obligation to warn persons of a potential threat to their health. Additional research can (and should) inform decision makers about the potential prevention benefits of partner notification, but the value of fulfilling the ethical obligation remains difficult to quantify.
Appendix: Overhead Costs
Partner notification overhead for Broward comprised personnel, 2.85 full‐time equivalents at $87,555 per year (5% of a program manager, GS‐13; 50% of two office managers, GS‐12; 45% of two supervisors, GS‐9/11; 45% of two state supervisors); rent, $54,647 for 2,295 square feet at $23.81 per square feet (5% of 132 square feet, 50% of 108 square feet, 50% of 120 square feet, 45% of two 100‐square foot rooms, 45% of two 90‐square foot rooms, a supervisor office at 100 square feet, an interview room at 85 square feet, eight interview rooms at 64 square feet, a lot room at 114 square feet, and 85% of a field office at 1,403 square feet); and telephone, $12,000. This total overhead for partner notification ($154,202) was multiplied by the fraction of partner notification that was devoted to HIV (0.45) and the fraction of the disease intervention specialist's time recorded for Arms 2 to 4 of the study (0.894). The annual overhead for HIV partner notification was $62,035, or $129,241 for the 25 months included in the study.
For Tampa, overhead costs comprised personnel, 3.7 full‐time equivalents at $120,955 per year (20% of an office manager, GS‐12; and 70% of five supervisors, GS‐9/11); rent, $7,374 for 578 square feet at 12.76 per square feet (20% of 99 square feet, 70% of five 61‐square foot rooms, one clerk room at 170 square feet, and 80% of the disease intervention specialist room at 350 square feet); and telephone, $2,520. This total overhead for partner notification ($130,849) was multiplied by the fraction of partner notification that was devoted to HIV partner notification (0.37) and the fraction of disease intervention specialist time that was devoted to Arms 2 to 4 (0.87). The annual total over‐head for HIV partner notification was $42,120, or $56,160 for the 16 months included in the study.
For Paterson, overhead costs comprised personnel, 1.15 full‐time equivalents at $35,622 per year (70% of a supervisor, 10% of a gonorrhea investigator, 10% of an HIV clerk, and 25% of a senior clerk); although no rent was reported, it was estimated at $13 per square feet for a cost of $1,352 (70% of 100 square feet, 10% of 90 square feet, 25% of 100 square feet); and telephone, $1,900. This total overhead for partner notification ($38,874) was multiplied by the fraction of partner notification devoted to HIV partner notification (0.35) and the fraction of the disease intervention specialist time that was devoted to Arms 2 to 4 (0.922). The annual total overhead for HIV partner notification was $12,545, or $18,817 for the 18 months included in the study. Cited Here...
1. Rothenberg RB, Potterat JJ. Strategies for management of sex partners. In: Holmes KK, Mardh PA, Sparling PF, Wiesner PJ, eds. Sexually Transmitted Diseases. 2nd ed. New York: McGraw-Hill, 1990:1081-1086.
2. Bayer R, Toomey KE. HIV prevention and the two faces of partner notification. Am J Public Health 1992; 82:1158-1164.
3. Woodhouse DE, Potterat JJ, Muth JB, Pratts CI, Rothenberg RB, Fogel JS II. A civilian-military partnership to reduce the incidence of gonorrhea. Public Health Rep 1985; 100:61-65.
4. Engelgau MM, Woernle CH, Rolfs RT, Greenspan JR, O'Cain M, Gorsky RD. Control of epidemic early syphilis: the results of an intervention campaign using social networks. Sex Transm Dis 1995; 22:203-209.
5. Fenton K, Peterman TA. HIV partner notification: taking a new look. AIDS 1997;11:1535-1546.
6. Toomey KE, Cates W, Jr. Partner notification for the prevention of HIV infection. AIDS 1989: 3(suppl 1):S57-S62.
7. Oxman AD, Scott EAF, Sellors JW, et al. Partner notification for sexually transmitted diseases: an overview of the evidence. Can J Public Health 1994; 85(suppl 1):S41-S47.
8. Landis SE, Schoenbach VJ, Weber DJ, et al. Results of a randomized trial of partner notification in cases of HIV infection in North Carolina. N Engl J Med 1992; 326:101-106.
9. Peterman TA, Toomey KE, Dicker LW, Zaidi AA, Wroten JE, Carolina J. Partner notification for syphilis: a randomized, controlled trial of three approaches. Sex Transm Dis 1997; 24:511-518.
10. Andrus JK, Fleming DW, Harger DR, et al. Partner notification: can it control epidemic syphilis? Ann Intern Med 1990; 112:539-543.
11. Potterat JJ, Spencer NE, Woodhouse DE, Muth JB. Partner notification in the control of human immunodeficiency virus infection. Am J Public Health 1989; 79:874-876.
12. Wykoff RF, Heath CW, Jr, Hollis SL, et al. Contact tracing to identify human immunodeficiency virus infection in a rural community. JAMA 1988; 259:3563-3566.
13. Holtgrave DR, Qualls NL, Curran JW, Valdiserri RO, Guinan ME, Para WC. An overview of the effectiveness and efficiency of HIV prevention programs. Public Health Rep 1995; 110:134-146.
14. Spencer NE, Hoffman RE, Raevsky CA, Wolf FC, Vernon TM. Partner notification for human immunodeficiency virus infection in Colorado: results across index case groups and costs. Int J STD AIDS 1993; 4:26-32.
15. Crystal S, Dengelegi L, Beck P, Dejowski E. AIDS contact notification: initial program results in New Jersey. AIDS Educ Prev 1990; 2:284-295.
16. Pavia AT, Benyo M, Niler L, Risk I. Partner notification for control of HIV: results after 2 years of a statewide program in Utah. Am J Public Health 1993; 83:1418-1424.
17. Higgins DL, Galavotti C, O'Reilly KR, et al. Evidence for the effects of HIV antibody counseling and testing on risk behaviors. JAMA 1991; 266:2419-2429.
18. Choi KH, Coates TJ. Prevention of HIV infection. AIDS 1994; 8:1371-1389
19. Cleary PD, Van Devanter N, Rogers TF, et al. Behavior changes after notification of HIV infection. Am J Public Health 1991; 81:1586-1590.
20. Kamb ML, Bolan G, Zenilman J, et al. Does HIV/STD prevention counseling work? Results from a multi-center randomized trial (Project RESPECT) (Abstract O134). Presented at the International Congress of Sexually Transmitted Diseases, Seville, Spain, October, 1997.