The Transformation of Partner Notification
- Reprints or correspondence: Dr. Richard Rothenberg, Department of Family and Preventive Medicine, Emory University School of Medicine, 69 Butler St. SE, Atlanta, GA 30303 (rrothen{at}emory.edu).
Abstract
Since 1996, several traditional epidemiological studies of program efficacy have done little to alter the fixed opinions of promoters or detractors of partner notification. Modeling studies have appeared that confirm the value of the contact-tracing approach in locating and treating persons who may be important in transmission. These provide a theoretical base for a number of empirical studies that have applied social network methods to classical partner-notification approaches, which have demonstrated the dense networks and geographic clustering of persons involved in sexually transmitted disease transmission and have provided justification for focusing efforts on small groups that may be critical to epidemicity or its maintenance. These initial studies would appear to warrant a broad-based inclusion of network concepts into programmatic activities, with careful monitoring and analysis of the impact they have on both the traditional indices of partner notification as well as the larger effect on disease transmission.
Partner notification (PN), as practiced in sexually transmitted disease (STD) control programs, is a multistep process that begins with eliciting information about sexual contacts from a person infected with an STD. Those partners are then notified that they are at risk for being infected, are assisted in seeking medical care, and, if infected, are treated and interviewed. Although implemented under a variety of names and with numerous variations (contact tracing, case investigation, partner services referral, self-referral, patient referral, and provider referral), the basic process has been unchanged for 50 years.
The value of PN continues to bedevil public health workers and investigators [1]. Many of these people are deeply committed to the process simply because it is enshrined in STD control activities. Others stress its value to individual infected persons, who can help friends get treatment and prevent their own reinfection. Still others emphasize the epidemiological and research value of the process. But a substantial number of voices have raised the question of proven efficacy and point to the lack of demonstrable influence of PN on disease transmission.
Since the appearance of several recent reviews of PN [1–3], a number of publications have taken a view of the process of PN, a view that incorporates concepts in social network analysis [4, 5]. Some of these concepts have been acknowledged in STD control for years, but none has played a central role.
The network-informed approach posits that, in areas of heightened transmission, a social network with sexual interactions will contain persons infected from various sources. An infected person who happens to be interviewed will thus have others in his or her sexual or social environment who are infected but not necessarily because of sexual contact with the interviewee. Thus, the friends, associates, and others who are part of a social milieu may be important to pursue, in addition to direct sexual contacts. This hypothesis leads to an expanded approach to PN—one that includes assessment of nonsexual contacts, selective screening of associates in the field, construction of the actual networks of interacting individuals (known in social network parlance as “connected components” [groups of people wherein everyone is connected to everyone else by a path of some length]), and visualization and geographic placement of networks as tools for deciding on next epidemiological steps.
In the present article, we review publications during the period from 1996 through early 2000 (table 1), noting the more traditional approaches that have been taken and the newer activity involving a social network approach that has emerged in the evaluation of PN.
Table 1
Table of evidence from information on partner notification (PN) published, 1996–2000.
Traditional PN Approaches
Program descriptions. Although a number of descriptions of program activity have been published in recent years [6–8, 13, 15, 16], they were in the minority. Such descriptions detail the well-known difficulties of actually performing PN, particularly in suboptimal settings. The enthusiasm for attempting to bring PN to third-world environments is considerable but is tempered by the lack of resources, the newness of the process in many cultures, and specific objections that relate to required training [6, 30–32]. In the developed world, several approaches have been offered to organizing program services for STDs, including PN [33, 34], and an updated program operations guide for STD control and prevention is currently being prepared by the Centers for Disease Control and Prevention (CDC; J. Wasserheit, personal communication).
Evaluating PN. During recent years, only a few studies of the effectiveness of traditional PN have been published. In general, the results were similar to those of studies done in past years [1]. Faxelid et al. [9] demonstrated that individual counseling of patients in a Zambian setting improved the contact index to 1.8, compared with 1.2 in the uncounseled group. Levy and Fox [21] demonstrated that injecting drug users were encouraged to name partners with the help of an outreach team that assisted in their notification. In a study of PN for syphilis [14], a CDC team determined that 3 intervention groups—(1) self-referral, (2) provider referral by a disease intervention specialist, and (3) provider referral by a disease intervention specialist who took blood samples in the field—were no different in their final outcome. The number of partners brought to examination in each group varied from 0.61 to 0.67. In a parallel study of HIV PN [19], the original study design was abandoned because of considerable crossover in group assignment, but the investigators were able to show that the cost per new HIV infection identified was $2200. As noted, the low contact indices in these 4 studies, the substantial logistical problems encountered, and the absence of a linkage of the activity to interruption of disease transmission have been characteristic of studies in past years.
Commentaries and surveys. Actual studies of PN have been augmented in recent years by specific inquiries into the particulars of the process as well as a number of editorials and articles of opinion. Several investigators have documented the current practices of clinics in the United Kingdom [15, 35, 36] and agreed that practices are diverse and continue to hinder an evidence-based approach to evaluation of the process. In the United States, a time-allocation study of the elements of PN in public clinics revealed a similar diversity and lack of systematization in the approach [37]. In the private sector, it has been noted that clinicians and their patients are often entirely unaware of the process [38]. Some of those attempting a broader assessment of the value of PN have been positive about the process in general settings [39], for the prevention of pelvic inflammatory disease (PID) [40], for use with drug injectors [41], and in developing countries with appropriate attention to context [32]. Others have been more reserved, often citing a lack of information about cost effectiveness [42, 43] and stressing the need for better information about the practice itself and its effect on outcomes [44].
Models and constructs. Only a few authors have focused on the value or effects of PN from the perspective of transmission modeling. Ghani et al. [31] used a Monte Carlo simulation (a process whereby a large number of networks are generated on the basis of values of network parameters chosen stochastically by computer) to create a “true” network picture and sampled from it using alternative sampling strategies (individual reports, snowball sampling, and contact tracing). In a complex analysis, they demonstrated that all 3 methods produce biases in the estimation of true network properties (mixing properties and network structure). In general, snowball sampling and contact tracing provided less biased estimates of mixing matrices, and contact tracing may be the preferred approach because of logistic features. Several other investigators have studied the cost effectiveness of PN. Rahman et al. [28] estimated that $4930 would be gained per year of life saved through application of PN for HIV in Japan. Their analysis, however, made assumptions about the effectiveness of the practice that might be hard to justify from published data. Howell et al. [27] used a model to apply cost effectiveness to the evaluation of alternative PN strategies for preventing PID in women. They found that early diagnosis and treatment of the female sex partners of infected men prevented more cases and saved more money than the strategy of interviewing infected women for their male sex partners. This strategy, however, did not consider the differential contribution of some men with long-term asymptomatic infection to the overall transmission of chlamydia [23].
Network Considerations
During the past several years, a number of published articles have attempted to examine the network attributes of the PN process and to use network concepts to enhance these procedures. Some of these have incorporated network concepts without overt reference to the field. For example, Jordan et al. [17], in a brief communication, demonstrated that asking infected persons about their network (as opposed to their contacts) may be of value in finding new cases. In 1 focus group, persons were asked to identify others they knew who might be practicing unsafe sex. Of 30 persons identified, 17 were HIV-positive and 9 had been previously unaware of their infection. In a second focus group, persons were asked to identify persons they thought might be HIV-positive. Of 15 such persons identified, 11 were HIV-positive and 8 had been previously unaware of their infection. Although the communication provides little information about the actual persons involved, the use of such techniques for case finding and targeting services deserves further exploration, because it uses social (as well as sexual or drug-using) contacts who do not bear the traditional PN relationships to the reporting patient.
Epidemic and endemic transmission. A similar nontraditional path was used by Rothenberg et al. [18], who investigated an outbreak of syphilis among teenage girls and boys in a suburban community. They found that interviewing important persons—not necessarily sex contacts—identified through network connections or other ethnographic means was a valuable adjunct to case finding. In fact, interviewing persons with or without syphilis in such a network had the same yield of positive contacts. In addition, they demonstrated that traditional notions of source and spread cannot be used in a network in which cases have overlapping contact with multiple other cases. Finally, they showed that the prevention of cases could be better documented through network connections that showed multiple exposure to syphilis in persons who were epidemiologically treated before onset of clinical or serologic evidence of disease. In a follow-up to that outbreak investigation, Rothenberg et al. [25] conducted a 6-month demonstration project in a high-morbidity area in inner-city Atlanta and demonstrated that the use of the larger social network perspective can be of considerable value in identifying new cases and in securing follow-up with typically difficult-to-reach people. In all, 48 cases of syphilis were identified in a group of 396 people (a 6-month cumulative incidence of 12.6%), 275 of whom could be connected to each other in a single component. They estimated that ∼30% of the cases found during this period could be attributed to network connections (friends, associates, or ethnographically connected persons) obtained outside the usual realm of PN. For example, 8 cases were connected directly to a single person through the process of network ascertainment, although none of them had been identified as a partner by that person. The study did not, however, provide information on the longer term impact of the process on the interruption of disease transmission. In a study of syphilis transmission in Louisiana, Rosenberg et al. [20] indicated the importance of central persons in the network, whom they felt were “core transmitters.” Their diagram of network activity was consistent with low-level endemic transmission. Using network ascertainment, Stoner et al. [45] demonstrated clear differences in the networks of persons infected with Nisseria gonorrheae or Chlamydia trachomatis. Incarceration, crack use, and a continuing drug problem were far more common in the networks of persons with gonorrhea.
Mixing matrices. Continuing the theme of the use of network-related concepts, Aral et al. [22] demonstrated that persons with gonorrhea and chlamydia have differing mixing matrices (the frequency with which groups with given characteristics have contact with each other). In particular, they showed that the differences in mixing patterns between the 2 infections centered on social characteristics (ethnicity, age, and sex) rather than on risk behavior, such as number of partners. They demonstrated as well the importance of links (bridges) between high- and low-prevalence populations. Such information is of importance in identifying groups at greater risk for specific infections on the basis of their network associations. Using a different approach, Laumann and Youm [26] analyzed data from the National Health and Social Life Survey to examine mixing patterns and self-reported STDs. They concluded that the general higher prevalence of some STDs among the black population might have a network explanation: a propensity for assortative mixing (like with like) with regard to ethnicity and for disassorative mixing with regard to level of sexual activity (highly active with less highly active).
Geographic clustering. Another form of network association—spatial proximity—was highlighted by Zenilman et al. [24], who demonstrated remarkable geographic proximity of dyads with gonorrhea, confirming the neighborhood nature of disease transmission and the geographically cohesive characteristics of core groups. These observations using quantitative measurement of distance confirmed earlier assertions, based on more traditional PN analysis, that gonorrhea was a neighborhood disease [46]. The availability of simpler geographic information systems now provides a method for the more intense investigation of this aspect of network associations as well as other factors associated with the conduct of network studies [47]. Such observations may be of considerable value in focusing control efforts on specific locations.
Network structure. In an article on the case-finding results from an intensive chlamydia control effort, Zimmerman-Rogers et al. [23] demonstrated that the usual interview periods, as promulgated by the CDC, will reveal most of the contacts but miss a potentially important group, men with long-term asymptomatic infection, who might be of considerable importance. They point out that traditional case-finding intervals may ignore the network implications of such persons in maintaining endemicity. In further exploration of these data, this group demonstrated that concurrency of sexual partnerships is the primary factor in predicting that a person is a transmitter of chlamydia [10]. Concurrency, which is strictly defined as sexual interactions that occur during the same time frame with 2 partners, is a key measure of network structure that, in modeling analyses, has been shown to bear an important relationship to transmission [48, 49]. In the analysis of chlamydia data, concurrency was more strongly associated with transmission than was the actual number of partners, thus providing empirical evidence for the theoretical and model-based importance of such structure.
In keeping with the recent appearance of network-motivated PN studies, those who display greater enthusiasm for the process do so from a network perspective [50–53]. These authors have stressed the larger perspective that network-informed PN provides, the insights into disease transmission, and the potential for a variety of newer interventions that may be network-based. It is important to recognize, however, that these studies represent some early efforts to inject network concepts into the traditional PN practices. Consideration of mixing patterns of persons at risk, of the geographic contiguity of current and potential sex partners, and of notions of social structure all appear to add to an understanding of the dynamics of disease transmission. It is clear that a more coherent and concerted approach is required to investigate the value of PN in the network context and to establish the basis for further intervention activity (table 2). The empirical data on which such understanding is based are derived directly from the PN approach that has been part of STD control for many years. Such analysis requires the addition of information about networks—social partners, drug-using partners, acquaintances, ethnographically connected individuals, and possibly even family members—and the use of network analytic tools to provide the broader context for assessing the intensity of relationships and the propensity for transmission.
Table 2
Elements of a coordinated approach to investigating the network characteristics of sexually transmitted disease (STD) transmission.
These additions are obviously important in a research context but are just as important for ongoing programs. Ultimately, changes in the modus operandi for disease-control efforts that use PN rest with their acceptance in the field, and field attempts at incorporation are critical for such acceptance. These initial studies would appear to warrant a broad-based inclusion of network concepts into programmatic activities, with careful monitoring and analysis of the impact they have both on the traditional indices of PN as well as the larger effect on disease transmission.
- © 2002 by the Infectious Diseases Society of America
References
This Article
- AbstractFree
- » Full Text (HTML)Free
- Full Text (PDF)Free
Classifications
Services
Citing Articles
Google Scholar
PubMed
Related Content
Share
- Email this article
-
CiteULike
Connotea
Delicious
Facebook
Google+
Mendeley
Twitter
Published on behalf of
IDSA Members: For your free access to this journal, log in via the IDSA members area.
The Journal
Impact Factor: 8.186
5-Yr impact factor: 7.898
For The Media
For Authors
- Services for authors
- Self-archiving Policy
- IDSA Conflict of Interest policy
- Instructions for Authors
- ICMJE Form
- Submit Now!
Open access options for authors visit Oxford Open
This journal enables compliance with the NIH Public Access Policy
Corporate Services
Alerting Services
Related IDSA Publications
