Sex, Drugs, Intervention, and Research: From the Individual to the Social

Epidemiological estimates of the sexual risk behavior of drug users
have provided essential indicators to the current and future prevalence
of HIV transmission. An overview of recent research shows the majority
of drug injectors to be sexually active, low levels of reported
condom use, a significant minority of female injectors to be involved
in prostitution, relatively high levels of sexual mixing between drug
injectors and noninjectors, and only scant indications of sexual behavior
change. Epidemiological studies of risk, however, are unable to
yield the data required to understand the interaction between individual
risk behavior and social relationships. This is required if obstacles to
safer sex compliance and sexual behavior change are to be overcome,
and demands recognition of the influence and importance of social
context on the production of sexual risk behavior in future research
and intervention designs. In response, the paper explores the future
role of qualitative research in understanding the social relations of
“risk” and in contributing toward theoretical advancements in explanations
of risk perception and risk behavior. The paper concludes by
discussing the implications of this analysis for developing interventions
The advent of HIV infection and AIDS has encouraged public debate about
the most intimate of private behaviors. This debate has largely focused on the
sexual behavior and safer sex compliance of gay and bisexual men, women and
men involved in prostitution, and young people. In contrast, the primary focus
of research, intervention, and education targeting drug users has been injecting
behavior and-more specifically-the sharing of injecting equipment.
As mounting evidence indicates that drug injectors are changing their druginjecting
behavior in response to HIV and AIDS (Stimson, 1991), recent epidemiological
research has highlighted the importance of HIV-risk posed to
injecting drug users and their sexual partners through the sexual transmission
of HIV (Des Jarlais, 1992). In the United States, it is estimated that injecting
drug users are the source of HIV in at least three-quarters of heterosexually
transmitted cases of AIDS (Moss, 1987; Des Jarlais and Friedmaq1987). In
the United Kingdom, a drug-injecting partner is reported for over 60% of first
generation cases of heterosexual transmission (Evans et al., 1992). Concerns
predicting the “real heterosexual epidemic” to emanate from drug injectors
(Moss, 1987) have encouraged a “sexual re-awakening’’ for practitioners and
researchers working within the drug field. The recent British Governmenl
document and national health strategy Health of the Nation has fueled these
concerns by reaffirming the increasing significance of sexual transmission in
the future spread of HIV:
HIV is primarily sexually transmitted and prevention of infection depends
largely UPOR changes in sexual behaviour. (Department of
Health, 1992, p. 92)
This paper provides a brief overview of key research findings on the sexual
risk behavior of injecting drug users (IDUs) with the aim of discussing in
greater depth the role of future research and intervention in understanding and
responding to sexual risk behavior and sexual behavior change among drug
users and their sexual partners. It is argued first that there is a need for qualiiative
research to build upon current epidemiological understandings of sexual
risk so as to encompass understandings of the social relations and social context
of sexual behavior, and second that there is a concomitant need for interventions
to target social relationships (rather than simply individuals) so as to
overcome current obstacles to modifications in individual risk behavior and to
encourage wider social and community change.
DRUG TAKING AND SEXUAL RISK
Sexual Activity
Most studies of injecting drug use show the majority of IDUs to be sexually
active. One recent London study, for example, found 80% of drug injectors
to have had vaginal or anal sexual intercourse in the 6 months prior to interview,
and noted that two-thirds of IDUs had vaginal intercourse at least once
a week (Rhodes et al., 1994a). Despite the varying selection criteria and time
frames of measurement employed, other studies show similar proportions of
IDUs to be sexually active: 77% (Van den Hoek et al., 1990), 77% (Donoghoe
et al., 1989), 82% (Klee et al., 1990a), 86% (Coleman and Curtis, 1988).
Findings suggest that levels of reported penetrative sexual activity among
IDUs are comparable to those reported in the British adult population. One
recent study of heroin and cocaine users (IDU and non-IDU) found higher
levels of sexual activity than those in the adult population. In addition, the
average number of reported (noncommercial) sexual partners of IDUs in a 6-
month period (2.4 partners in London and 2.1 in Glasgow) have been found
to be slightly greater than comparative estimates in the British adult population
(Rhodes et al., 1993a).
Sexual behavior research among drug users prior to HIV infection and
AIDS focused primarily on the perceived pharmacological effects of drug use
on sexual activity. These studies suggested a reduction in sexual activity and
sexual interest to be associated with frequent opiate use (Mirin et al., 1980)
and an enhancement of sexual activity and interest to be associated with the use
of stimulants, such as amphetamines and cocaine (MacDonald et al., 1988).
Recent behavioral research undertaken in the context of HIV transmission has
supported an association between stimulant use and increased sexual activity
(Kall and O h , 1991; Fullilove et al., 1990; Chaisson et al., 1989). These
trends, however, are by no means consistent, and there remains considerable
uncertainty about their causal determinants (MacDonald et al., 1988; Washton,
1989; Marx et al., 1991).
Safer Sex Compliance
Most studies of injecting drug use show reported levels of condom use to
be comparable with those in the heterosexual population as a whole. They also
indicate greater likelihood and greater frequency of condom use with casual
partners than with primary partners. Recent findings in London, for example,
show that in a 6-month period two-thirds (68%) of drug injectors never used
condoms with primary partners and over a third (34%) never used condoms
with casual partners (Rhodes et al., 1994a). Other reports indicate that 79%
of injectors in Glasgow (Rhodes et al., 1993a) and 75% of injectors in the
West Midlands (Klee et al., 1990a) never use condoms. Safer sex compliance
with primary partners has been shown to be statistically associated with an
awareness of HIV positive antibody status (Van den Hoek et al., 1992). although
studies also indicate relatively high levels of continued sexual risk
behavior among HIV-positive IDUs (Rhodes et al., 1993b). Surveys of anonymously
tested saliva samples in London also show that the majority of HIVpositive
IDUs are unaware of their positive status (Donoghoe et al., 1993;
Rhodes et a]., 1993b).
A combination of commonsense assumption and research evidence suggests
that drug use has a disinhibitory effect on decision-making about sexual safety
and on safer sex compliance (see Rhodes and Stimson, 1994). While there
remains little comparative or conclusive research in this area, recent research
has associated higher levels of sexual risk behavior with increased severity of
drug dependence (Gossop et al., 1993), frequent amphetamine use, temazepam
and polydrug use (Klee et al., 1990b), and cocaine or crack use (Chaisson et
al., 1991).
Most United Kingdom studies conclude that condom use remains at insufficient
levels to prevent the potential for further sexual transmission of HIV
between drug injectors and their sexual partners, particularly given average
rates of partner change and the significant minority of injectors who also continue
to share used equipment with people other than their sexual partners
(Rhodes et al., 1993a).
Prostitution
There is an established overlap between an involvement in injecting drug
use and an involvement in female prostitution, Estimates in London show 14%
of women prostitutes attending sexually transmitted disease (STD) clinics (Day
et al., 1988) and 33% contacted through street outreach (Rhodes et al., 1991)
to inject drugs. Estimates elsewhere range from 25% (Kinnell, 1989) to 59%
(McKeganey and Barnard, 1992). Studies of injecting drug use also indicate
a high proportion of female injectors to be involved in prostitution: recent
estimates suggest 14% in London and 22% in Glasgow (Rhodes et al., 1993a:i.
There is little evidence of injecting drug use among male prostitutes (Bloor et
al., 1992).
As is the case with female sex workers (Day et al., 1988), female IDUs
involved in sex work report higher levels of condom use with paying partners
than with nonpaying partners. In Glasgow, female IDUs involved in prostitution
report almost 100% condom use with paying partners compared with 9%
“always” condom use with nonpaying primary partners and 22 % “always”
with nonpaying casual partners. In London, female IDUs involved in prostitution
report 70% “always” condom use with paying partners (Rhodes et al.,
1994b).
Estimates of HIV prevalence among women prostitutes have found higher
rates of prevalence among prostitutes with a history of injecting drug use, and
in European and North American countries evidence associates HIV transmission
among prostitutes with an involvement in injecting drug use rather than
with an involvement with prostitution per se (Padian, 1988; Van den Hoek et
al., 1988; McKeganey et al., 1992). In the absence of controlled studies designed
to assess the relative risks of sharing used injecting equipment and
sexual transmission of HIV, it is difficult to determine the epidemiology of
epidemic spread among drug-using and nondrug-using prostitutes. A recent
prevalence survey in London found HIV infection to be no higher among female
IDUs involved in prostitution than among IDUs not involved in prostitution
(13% compared with 14%) (Rhodes et al., 199413). This adds further
support to emerging evidence which suggests that prostitution per se is not
independently associated with HIV prevalence or HIV risk behavior.
Sexual Partners
Studies show a relatively high degree of sexual mixing between injecting
and noninjecting drug users: approximately half of the sexual partners of injectors
are estimated to be noninjectors while approximately half of injectors
report noninjecting sexual partners (Rhodes et a]., 1993a). The vast majority
of these partners are women. This is in part an artifact of injecting drug use
being a predominantly male activity and in part because male injectors show
specific preferences for noninjecting female partners (McKeganey and Barnard,
1992). Such preferences may also be more likely with primary (i.e., more
important longer term) partners than with casual partners (Rhodes et al.,
1994a). This poses increased sexual risks to the noninjecting sexual partners
of injectors, and in particular to female primary partners, for whom contact
with an injecting drug user may be their only significant risk factor. It is within
primary relationships that condom use is most infrequent, while a significant
minority (between 16% and 19%, Rhodes et al., 1993a) of injectors report both
primary and casual partners in a 6-month period.
There are few studies which have explicitly involved the sexual partners
of drug users in research. North American qualitative research indicates the
difficulties female sexual partners of injectors have in initiating and negotiating
strategies of protection not only for themselves but also for their partners
both with regard to safer sex and needle safety (Wermuth et al., 1992; Kane,
1991).
There are few longitudinal or cohort studies of sexual behavior change
among drug injectors, and in comparison to studies of injecting and sharing
practices, only scant indications of change. Although some studies point tci
reductions in the number of sexual partners and sexual encounters and increased
levels of reported condom use (Skidmore et al., 1989), these have been
limited changes, and some studies report either no change or increased levels,
of sexual risk behavior over time (Des Jarlais et al., 1992; Calsyn et al.,
1992). The lack of notable sexual behavior changes relative to changes in drugtaking
behavior among drug users probably relates to a combination of factors.
These include injectors’ own assessments of sexual risk as relative to drugrelated
risks (Jain et al., 1991), the problems experienced in translating knowledge
about sexual risk into action, the primary focus of research, intervention
and education agencies on modifications in injecting behavior, and the associated
notions of identity and responsibility which this has created and reinforced
within drug-injecting communities about sharing and drug-using practices
(Rhodes and Quirk, 1996).
RESEARCH: EXPLAINING THE INDIVIDUAL AND SOCIAL
DYNAMICS OF “RISK’
Current Epidemiological Explanation: Some Possible
Improvements
Conventional epidemiology is concerned with the study of the distribution
and determinants of disease (Barker and Rose, 1984). In the case of the public
health response to the HIV epidemic, epidemiological research has provided
essential indicators of the distribution and determinants of HIV disease among
injecting drug users. This has provided baseline quantitative indicators on levels
of sexual activity and sexual risk behavior among drug users and their sexual
partners and has laid the foundation for the development of a range of HIV
prevention and safer sex health promotion initiatives targeting changes in individual
sexual lifestyles.
As HIV transmission routes among injecting drug users shift from parenteral
to sexual routes as safer injection becomes more commonly adopted
(Schoenbaum et al., 1989), it is important that future epidemiological studies
of sexual risk behavior both clarify and improve indicators of the distribution
and determinants of sexual activity and sexual risk behavior. Current epidemiological
explanations of sexual risk among drug injectors can be improved
in four main ways.
First, there remains a need for greater comparability between studies in
measures of sexual risk behavior. At minimum, indicators need to include
measures of the frequency, type, and number of sexual partners; frequency and
type of penetrative and nonpenetrative sexual encounters; and frequency of
condom use and safer sex. Most previous studies have employed partial indicators
of sexual risk, usually as a component of investigations primarily concerned
with drug-taking practices. Reliable epidemiological indicators of HIV
risk need to measure the interaction between the frequency and type of drugrelated
and sex-related risk behavior. It is fundamental also that studies remain
comparable in the time-frames of measurement (i.e., period of recall) and
categories of measurement (i .e., continuous or dichotomous) employed
(Samuels et al., 1992), as well as in the definition of key “risk” variables (e.g.,
“prostitution,” “safer sex”).
One possible methodological development in the measurement of drugtaking
and sexual risk behavior is the use of retrospective and prospective selfcompletion
diaries. Sexual diaries have provided an effective and reliable means
of data collection among gay and bisexual men, minimizing problems of recall
and providing a detailed and time-coded description of sexual encounters
(Coxon, 1988). The feasibility and reliability of using diaries as a method of
data collection among drug users is largely unknown. The use of diary methods
may provide more accurate assessments of the interaction between drug
taking and sexual risk, allowing examination of the causal dynamics of the
relationship between drug use and sexual activity by gathering time-coded
behavioral data within specific drug use and sexual encounters. Such analyses
may also provide opportunities for delineating the dynamics of the pharmacological
relationship between drug use and sexual risk and the influence of interpersonal,
situational, and social context on the perceived and experienced
effects of drug use on sexual behavior.
Second, improvements can be made to sampling designs. The majority of
studies of HIV prevalence and HIV risk behavior among drug users draw on
highly selective samples drawn primarily from drug user treatment and agencybased
populations (Samuels et al., 1992). The majority of drug users, however,
remain out of contact with treatment and helping services (Frischer, 1992), and
a number of studies have indicated higher levels of HIV prevalence and higher
levels of drug-related HIV risk behavior among nontreatment populations
(Lampinen et al., 1991; Donoghoe et al., 1993). Less is known about the
degree to which “hidden” populations of drug users also engage in sexual
transmission behaviors, and the influence (if any) of drug treatment on health
behavior changes not directly related to drug use such as sexual health is unclear.
The desire to change HIV transmission behavior, however, may be an
equally important determinant of changes in drug taking and sexual behavior
than the influence of drug treatment and helping services per se, while those
engaging in higher levels of risk related to drug use may also have a propensity
for higher levels of risk behavior as a whole. Future epidemiological research
needs to simultaneously recruit samples from a variety of drug user
treatment and nontreatment settings, with the aim of investigating the possibility
of bias in indicators of risk which rely primarily on clinic and agency-based
samples (Alcabes et al., 1992).
Third, improvements can be made to study design. The majority of studies
employ cross-sectional study designs with retrospective measures of risk, and
there are few longitudinal or cohort studies of sexual risk behavior and sexual
behavior change among drug users. While there are practical and methodological
difficulties inherent in cohort study designs among drug users (Vlahov and
Polk, 1988; Samuels et al., 1992), if future studies are to yield data on the
problems and possibilities of sexual behavior change, ideally these should simultaneously
include longitudinal as well as cross-sectional designs.
Fourth, improvement can be made to current epidemiological measurements
of “risk behavior.” Current studies invariably take the individual as the
unit of analysis. Because HIV infection is a behavioral disease, its progression
is not random or uniform but subject to much variation and change. Reliable
epidemiological indicators of the distribution of HIV risk and HIV spread require
a measure not just of the frequency and type of risk behavior in individuals
but also a measure of the interaction and epidemiological efficiency of
mixing patterns between individuals (Vlahov et al., 1990; Samuels et al.,
1992).
In addition to the individual, it is equally important that future epidemiological
study takes as its unit of analysis “social units” of drug injectors, defined
in terms of the social and epidemiological ties and connections between
individuals which are relevant for investigating HIV transmission. These social
units range from particular relationships between individuals as in the case
of friendship or sexual dyads to wider social relationships of individuals as with
drug dealing and friendship networks. Since it is the interaction between individuals
which determines HIV transmission, shifting the unit of analysis toward
“social units” to determine the epidemiological efficiency of these inter -
actions enables greater reliability in assessments of risk and in estimates of
cpidemic spread.
One priority for future research in this area includes an assessment of the
assortative and disassortative* sexual mixing patterns among drug injectors and
the sexual partners of drug injectors. While current research has highlighted
an increased risk of HIV infection to the noninjecting sexual partners of injec-
tors, there are difficulties in estimating the associated risks of sexual transmission
to noninjecting heterosexual populations without concomitant knowledge
of the sexual mixing patterns of both injectors and their sexual partners.
Theorizing Risk: The Limits of Individualism
Conventional epidemiology remains locked into a conception of risk which
is restricted to the individual (Tannahill, 1992). As noted above, most epidemiological
study takes the individual as its unit of analysis and most explanation
and prediction is based entirely on measures of individual risk behavior
(see Table 1). Such explanations are limited because they remain blind to a
variety of other social and cultural processes which influence the ways in which
individuals behave, and thus also the ways in which epidemics spread. As
recently suggested, the social impact and significance of HIV-risk can only be
understood and explained by “filling-in and questioning the empty categories
of epidemiological prediction” (Kane, 1991, p. 1037).
It is in this context that it is important to note that epidemiology and epidemiological
approaches have largely framed the focus and parameters of lay
and professional understanding about HIV infection and AIDS (Herdt and
Lindenbaum, 1992). As noted by Berridge, epidemiologists have played the
lead part in defining and ordering the disease and in giving it a name
(Berridge, 1992). It is important to recognize that it is epidemiological understandings
and categories of “risk behavior” and of “risk groups” that have
informed and defined the boundaries of psychosocial behavioral research investigating
the determinants of individual risk behavior and lifestyle.
Psychosocial models of research and health behavior based upon individualistic
lifestyle notions of risk are often inadequate to address the complex
social realities of risk acceptability, risk perception, risk assessment, and behavior
change (see Table 1). These models, which emphasize the “health beliefs”
(Becker, 1974), “self-efficacy’’ (Bandura, 1977), and motivations and
skills (Joseph et al., 1988) of individuals to behave in certain ways, recognize
a cognitive decision-making process in risk perception and risk behavior but
fail to adequately capture either their social dimensions or their complexity.
While providing pointers to behavioral intention, this is often devoid of social
and cultural explanation or understanding (Romer and Hornik, 1992; Ingham
et al., 1992). Such research has been found to have a limited capacity and
utility in either predicting or explaining health beliefs about sexual risk and
sexual behavior change (Rosenstock et al., 1988; Bloor et al., 1992; Montgomery
et al., 1989).
Current epidemiological and psychosocial theorizing on risk perception and
behavior is based on the assumption of individual rationality (Rhodes, 1995).
Dominant theoretical approaches to understanding health behavior-such as
derivatives of the theories of Reasoned Action, Planned Behavior, and the
Health Belief Model (see; Ingham et al., 1992; Becker, 1974; Fishbein and
Azjen, 1975; Azjen, 1988)-view “risk-taking” as the result of an individual’s
rational decision-making based on the perceived costs and benefits of risk action.
At their crudest, such models of explanation assume a single rationality
of choice-making about risk (Rhodes, 1995). Choices to avoid risk, which are
frequently spoken of in health promotion discourses as the “healthy choices, ”
are seen as “reasoned” choices, as exemplars of rational rather than irrational
behavior. This often demeans explanations of continued risk behavior among
IDUs to the rather dubious scientific categories of “unreasoned” behavior and
cognitive malfunction.
Recent attempts to move beyond “single rationality” theories of risk perception
toward theories of “situated rationality” overcome some of these limitations
but clearly remain inadequate (Bloor, 1995; Rhodes, 1995). These theories
recognise that rationality is inextricably linked to the specific situations and
contexts in which choices about risk are made. They go as far as to allow a
plurality of rationalities and thus move beyond a one-dimensional matrix of
“cost and benefit” where cost is synonymous with risky actions and benefit is
synonymous with their avoidance. “Situated rationality” theories of risk behavior
may help explain, for example, why condom use by IDUs with casual
partners has increased over time while condom use with primary partners has
remained relatively constant. Our current qualitative work on sexual negotiation
indicates that in some cases HIV-negative IDUs may continue to have
unprotected penetrative sex with their HIV-positive partners, despite knowledge
and understanding of the proximity and susceptability of risk (Rhodes and
Quirk, 1996). “Situated rationality” theories would posit that these decisions
about risk are reasoned by individuals on the basis of costs and benefits which
are situation and context dependent (e.g., where loss of trust or intimacy may
be perceived to be of greater cost than the risk of HIV).
But as we noted above, “situated rationality” theories remain inadequate
to explain the social realities of risk behavior. As with other derivatives of
theories of “reasoned action,” they assume that decisions about risk action are
calculated. While in some instances this may be the case, this does not recognize
the habituation of risk behavior (Bloor, 1995). Many of the behaviors
in which IDUs routinely engage, whether deemed “risky” by themselves or by
social scientists, are everyday behaviors which occur in a mundane or unspectacular
fashion, often without individual “decisions,” “choices,” or “calculations”
having to be made. Because such theories are ostensibly theories of
individual cognition, they are unable to recognize that individual rationalities
and perceptions are socially organized:
If a group of individuals ignore some manifest risks, it must be because
their social network encourages them to do so. Their social
interaction presumably does a large part of the perceptual coding on
risks. (Douglas, 1986, p. 67)
The individualism of current theories of risk provide limited understandings
of risk behavior. “Risk” is neither perceived nor understood by individuals
as a neutral category but is socially and culturally organized and acted upon
(Douglas, 1986, 1992; Douglas and Wildavsky, 1983; Bloor, 1995, Rhodes,
1995; Hart and Boulton, 1995). This means that epidemiological understandings
and measures of risk-particularly when applied in survey-based research-
often lack social and cultural specificity and appropriateness. They fail
to account for the ways in which risk is socially and culturally defined and the
ways in which individual understandings of risk are socially and culturally
mediated (see Table 1). Because individual rationalities are based on wider
socially organized boundaries of explanation and meaning, “risks” are not simply
or only “calculated” by individuals and neither is risk action necessarily
individually “chosen” or “decided” upon. It is for this reason that the notion
of risk taking is both inaccurate and misleading. The “choices” which current
psychosocial research paradigms assume to be “taken” by individuals are COMcomitantly
determined by a combination of social, cultural, and economic factors.
What social scientists often view and measure as being individual volktion
may sometimes not be “choice” at all (e.g., cases of sexual persuasion,
“negotiation,” and coercion, or unsafe sex for money). Contemporary explanations
of HIV and sexual risk behavior provide little notion or measure of
“the social.” Parrallels can be made with anthropological critiques of how
theorizing on risk has tended toward the “deculturing” of individuals:
Expert risk analysis takes as its decision-making unit the individual
agent, excluding from the choice any moral or political feedback that
he may be receiving from his surrounding society. The rational agent
of theory is decultured. (Douglas, 1986, p. 67)
Rather than focusing exclusively on the psychological determinants of individuals
in risk-related encounters, there is a need for future research to recognize
how individual perceptions of risk and individual capabilities to control
risk-related encounters are relative, both to wider peer-group, social, and community
norms and to situational and structural context (see Table 1). The fundamental
aim of such research is to focus on the pattern of interaction between
risk behavior and social relationships and less on individual decision-making
and the risk behavior patterns themselves (McKeganey and Barnard, 1992).
Investigating the relationship between individual risk perception and social
context encourages an understanding of the obstacles to individual behavior
change. This demands a shift in direction toward a more qualitative action
oriented research paradigm suited to investigating the social contexts and social
relations of drug use and sexual activity (Rhodes and Stimson, 1994).
This new paradigm of HIV risk research aims to build upon current epidemiological
and sociological understanding of sexual risk behavior among
drug users in an attempt to provide pragmatic support to developments in health
promotion and intervention. It is our contention that it is timely and important
that a new paradigm of sexual risk research requires more than evidence of risk
behavior and of behavior change: it also needs to investigate and influence the
process of change.
Understanding the Social Processes of Risk
We have suggested that epidemiological research is of incontestable importance
in mapping the future determinants of epidemic spread, and that this
contribution is best invested in an understanding the interactive nature of risk
and risk behavior. While epidemiological and psychosocial study is suited to
mapping the determinants and distribution of individual risk perception and
behavior cross-sectionally and longitudinally, it currently lacks the descriptive
capabilities to understand the social processes which determine the ways in
which perceptions and behaviors are produced. It is unable to appreciate the
subjective nature of the objects of study.
What is needed is a move toward an interactive paradigm of research
which is inclusive of sociological and anthropological methodology and explanation.
At the outset this demands a shift from conventional epidemiological
approaches toward a “social epidemiology” which aims to classify the determinants
of disease and illness on the basis of their social and economic origins
(Paterson, 1981; Scott-Samuel, 1989; Tannahill, 1992). The role of sociological
and anthropological research in this context is twofold.
First, it aims to describe the personal and social meanings attached by
individuals and by groups of individuals to specific behaviors categorized as
“risky” by the epidemiologist. This means describing behavior in the context
of the meanings participants themselves have ascribed to their behavior. Behaviors
attributed “risky” by the epidemiologist are thus to the qualitative sociologist
part of a wider structure or culture of behaviors and associated meanings,
which to participants themselves are often viewed and experienced as “normal,”
rational, even mundane (Schwartz and Jacobs, 1979). This means understanding
risk behavior in the context of drug users’ everyday lives:
For the IV drug using subculture in particular, the risks associated
with AIDS transmission overlap with a constellation of risks about
which we know little. The concept of risk-taking as a common and
meaningful dimension of the lives of IV drug users has so far been
hidden behind an externally-constructed pastiche of risk behaviours
specific only to AIDS. (Connors, 1992, p. 591)
Second, it aims to understand the processes by which individuals come to
attach meaning to “risk behaviors” and the ways in which individuals interact
with wider systems or structures of knowledge and influence about HIV, risk,
and health. This means investigating the social and cultural production of
knowledge about HIV-related risk (of which epidemiology and dominant scientific
discourse is a part) in the light of other situational and environmental
factors which influence health behavior. Overall, the aim is to understand the
“reciprocal effects of social settings upon individuals and of individuals upon
social settings” (Schwartz and Jacobs, 1979, p. 9) with the objective of determining
the problems and possibilities of reducing drug and sex-related harm.
The move toward such a research paradigm demands fundamental shifts
in contemporary thinking about social problems. As noted above, conventional
epidemiology has played the key role in identifying and defining HIV risk and
in influencing how policy and health interventions should best respond. The
need to view HIV infection and HIV risk as socially constructed problems requires
untangling the processes which have been key to “inventing” HIV and
AIDS (Patton, 1990). One of the challenges of the second decade of AIDS is
to bring about fundamental shifts in how research aims to reconstruct the social
realities of health behavior and everyday life among populations affected
by HIV transmission. As has been observed in the field of risk perception:
A very significant body of work views risk perception as an individual
and not as a social phenomenon. . . . It seems that the neglect of culture
is so systematic and so entrenched that nothing less than a large
upheaval in the social sciences would bring about a change. (Douglas,
1986, p. 1)
Understanding Drug Taking and Sexual Risk
A closer inspection of current research explanations of the relationship
between drug taking and sexual risk demonstrates the importance of viewing
sexual risk behavior as a socially organized interaction. Here we use the example
of crack and cocaine use (see Rhodes and Stimson, 1994, for a full
discussion).
There is mounting epidemiological evidence which shows there to be an
association between the use of crack and cocaine and increased levels of reported
sexual activity and sexual risk behavior (Wolfe et al., 1990; Fullilove
et al., 1990; Chitwood and Comerford, 1990; Chaisson et al., 1989, 1991).
This reflects contemporary concerns that stimulant drugs (and in particular,
cocaine and crack) have disinhibiting effects on safer sex compliance:
The danger of crack lies in its potential to promote high-risk sexual
behavior through which AIDS can be contracted [italics added].
(Bowser, 1989, p. 539)
More than a cursory glance of the epidemiological literature reveals that
there are many studies which show no such associations or which show such
associations to be complicated by an interaction of social, situational, and
material factors (Hartgers et al., 1991; Wolfe et al., 1990, 1992; Inciardi,
1989). While studies show associations between cocaine and crack use and HIV
positivity, there are few studies of sexual transmission and few which show
these to be causal associations in people without a history of injecting drug use
(Marx et al., 1991; Chaisson et al., 1991).
While the current epidemiological picture remains blurred (see Marx et al.,
1991, for a review), it is becoming increasingly clear that sexual risk behavior
among crack and cocaine users is determined by a range of social, situational,
and cultural factors which often remain peripheral to the vision of
epidemiology. Two examples help to demonstrate this.
First, ethnographic research has shown the importance of social and group
norms in influencing individuals’ perceptions, expectations, and accounts of the
effects of crack on sexual behavior and performance (Inciardi, 1989; Carlson
and Siegal, 1991). This means that there is often a subcultural “mythology”
attached to crack and sexuality which informs individual and group expectations
and understandings of the effects of crack and cocaine on sex-related
behavior. While sexual behavior in crack-related encounters may vary depending
on an interaction of social, situational, and material factors (see below),
individuals often make sense of such experiences in a limited number of ways
in the light of shared knowledge about what is legitimate (i.e., “normal”)
behavior. It is important to study both the processes by which knowledge is
socially organized and the ways in which individuals interact with this body
of knowledge to make sense of their own behavior:
We may make more sense of people’s explanations, especially when
given in social contexts, if we . . . acknowledge that, as accounts,
common-sense explanations often serve to excuse and justify, and not
merely to explain. (Hewstone, 1989, p. 37)
Second, ethnographic research has shown that in some crack-related settings
it is normal and legitimate for the drug to be exchanged for (often unsafe)
“sexual favors” which are often initiated and performed by men as “degradation
rituals” (Carlson and Siegal, 1991). These encounters are often not
viewed or understood as “prostitution” by the participants concerned but are
seen as a necessary or usual component of the drug deal and of the crack-related
encounter. While this may be documented as being of epidemiological importance
given the increased sexual risks associated with crack use, this cannot
fully explain the processes which determine such events. Such sexual
encounters are determined not simply by an interaction between pharmacology
and individual psychology but by a complex interaction between the individual
and the social which determines both the economics and currency of drug and
sex exchanges.
Future research which aims to investigate and influence sexual behavior
change among drug users requires an understanding of the social, cultural, and
material exchange “value” of behavior and the ways in which such values and
meanings limit the predictive and explanatory effectiveness of rational and
decision-making models of individual behavior.
INTERVENTION: FROM INDIVIDUAL TO SOCIAL CHANGE
HIV Prevention and the Limits of Individual Change
While ethnographic research points to indications of large-scale and community
changes in drug-injecting behavior (Burt and Stimson, 1993), HIV
prevention programs targeting drug users have found greater difficulty in promoting
and achieving sexual behavior change (Table 2). In recognizing the
inadequacy of interventions based on biomedical notions of individual lifestyle
and sexual behavior change (Ehrhardt, 1992), the challenge for safer sex health
promotion is to both create and nurture a collective and social responsibility
about sexual behavior and sexual health.
The recent advocation and adoption of community-based HIV prevention
strategies targeting harder-to-reach drug users may provide the foundation and
stimulus for such a response (Rhodes, 1993, 1994a, 1994b). Current UK interventions
are predominantly focused toward the individual client and toward
achieving individual behavior change. These initiatives attempt to contact drug
users, the majority of whom are out of contact with existing health services,
with the aim of enabling them with the means to make safe choices about drugtaking
and sexual behavior.
The effectiveness and efficiency of current UK models of community-based
HIV prevention have recently come under critical review. Recent evaluation
has raised important questions about the limitations of community-based interventions
which work within a mode of health education dominated by an individualistic
focus. Evaluation of the UK syringe exchange schemes, for example,
has indicated the inherent limitations of the approach in encouraging and
sustaining behavior change among drug injectors in the community and in
social environments where risk behavior is actually produced (Stimson et al.,
1991; Stimson and Donoghoe, 1996). Despite the availability of injecting
equipment though syringe exchanges, “choices” about whether to share such
equipment are also influenced by particular social relationship dynamics (e.g.,
between sexual partners), social desirability, and social acceptability. Qualitative
research in Glasgow, for example, notes differences in patterns of sharing
among women and men where sharing was found to be a “socially embedded
behaviour which [was] responsive to the many rights and obligations” within
social relationships (Barnard, 1993).
While outreach and extra-agency work enables education in situ-within
the social environments where risk behaviors are produced-this also largely
remains targeted toward individuals with the aim of encouraging “self-empowerment”
on a client-centred basis (see Table 2). Evaluation of street-based
outreach questions the utility of such an approach, pointing to wider social and
material factors (e.g., peer group norms, housing and welfare needs) which can
impede the effective promotion and adoption of changes in individual lifestyle
and sexual risk behavior (Rhodes and Holland, 1992).
This confirms ethnographic and behavioral research which shows the importance
of social and peer group norms and of situational and social setting
in shaping behavior change (Rhodes and Hartnoll, 1996). If intervention is to
build effectively on the findings of recent research, greater emphasis must be
placed on the targeting of networks and communities as objects and as agents
of change rather than individuals and individual risk behavior alone (Friedman
et al., 1992, 1994; Rhodes, 1993, 1994b; Stimson et al., 1994). This is necessary
as a first step to creating the social relations in which individuals can
exercise “choices” about their health behavior.
Toward Social Network and Community Change
There are few UK interventions which explicitly attempt to encourage
collective or community change among drug users, and there remains considerable
inexperience in using the appropriate intervention methods to achieve
these aims. UK interventions have much to learn from their international counterparts
(largely US and Australian) reportedly effective in encouraging community
change (Wiebel, 1988; Friedman et al., 1992; Friedman et al., 1994;
Trotter et al., 1993) and much to learn from studies of community participation
and organization in health promotion as a whole (Rogers and Shoemaker,
1971; Rogers, 1983; Bracht, 1990; Tones et al., 1990; Freire, 1972).
In understanding the possibilities for initiating and reinforcing change in
social networks or communities of drug users, it is useful to draw on the established
theories and practices of community development and, in particular,
communication and diffusion of innovations (Rogers and Shoemaker, 197 1 ;
Rogers, 1983). Communication of innovations theory provides important guiding
principles which govern the conditions necessary for change and the likelihood
of change being adopted. There are essentially four principles (Tones
et al., 1990): the characteristics of communities govern the need and desire for
change; the ownership of, and identification with, an innovation (or intervention)
by a community governs the likelihood of adopting and sustaining change;
the process of change is governed by the homophily* existing between community
leaders and change agents; and the process of change is governed by
the characteristics and perceived consequences of change.
To move beyond the limitations of interventions targeting individuals as
agents of change, interventions first require knowledge of the characteristics
and structure of drug-using social networks. This is necessary so as to understand
and monitor the possibilities for diffusion, of the “processes by which an
innovation is communicated through certain channels over time among the
members of a social system” (Rogers, 1983, p. 5). Analysis of social network
structure thus requires epidemiological and sociological mapping of the nature
and structure of social relationships within social networks. As noted by Scott
on the subject and method of social network analysis:
Relations are not the properties of agents, but of systems of agents;
these relations connect pairs of agents into larger relational systems.
(Scott, 1991, p. 3)
This means describing the “contacts, ties, and connections” between individuals
within a network with the aim of delineating the channels of communication
and influence for targeted innovations. It is for this reason that the
role of outreach worker or peer educator often overlaps with the role of community
ethnographer and that many outreach programs in the United States
have developed simultaneously in the light of ongoing ethnographic intervention-
based research (Wiebel, 1988, 1996; Feldman and Aldrich, 1990; Grund
et al., 1996). An ethnographic description of drug-using special networks thus
consists of “a body of qualitative measures of network structure” (Scott, 1991,
p. 3) delineating the “specific type of relation linking a defined set of persons”
(Knoke and Kuklinski, 1982, p. 12).
More particularly, it is important to gauge the extent to which specific
networks or communities of drug users are homogeneous. The extent as well
as the specific nature of connections within networks clearly influences the
feasibility for diffusion of communications. Each individual drug user can be
seen to have a multitude of ties into a number of overlapping ego-centered
networks: it is the job of ethnography to determine which particular ties into
which particular networks are relevant for targeting as potential channels of
diffusion into group-centred networks. The potential that drug-dealing networks
hold for communication of innovations, for example, may differ from the
potential that drug-using, friendship, or sexual networks have. The heterogeneity
within as well as across social networks and communities of drug users
across time and space must be considered the first obstacle to targeting social
networks as a way of instrumenting “community change” (Rhodes, 1993). In
contrast to a greater developed sense of collective social and political identity
among communities of gay men, for example, identities, “ties and connections”
within social networks of drug users may be more functional than ideological,
and perhaps more imagined than real (Rhodes, 1994a). Interventions encouraging
community change within drug-using networks thus need to build upon
existing social ties, norms, and values by first identifying perceived needs for
change and second by creating and nurturing a sense of collective identity and
shared responsibility about innovation and change in individual and collective
health behavior.
Available evidence points to the effectiveness of interventions targeting
peer influence as a method of facilitating collective action and community
change. Among gay men, research has demonstrated the importance of collective
action in first creating the social and cultural conditions necessary for
sexual behavior change and second in influencing and reinforcing the validity
and efficacy of continued changes in sexual behavior (Ehrhardt, 1992; Hart and
Boulton, 1995). A number of studies show greater sexual risk reduction
changes among gay men who are socially integrated into existing gay social
networks than among gay men who are not (Kippax et al., 1992; Freeman et
al., 1992). Studies also show that greater sexual risk reduction changes are
reported among gay men who receive social and peer support when attempting
changes in their sexual behavior and condom use (Kelly et al., 1990,
1992). Crucially, controlled comparative evaluation shows greater sexual risk
reduction to be achieved among those targeted by peer group organizing within
preexisting social networks than by conventional individually-targeted health
education alone (Kelly et al., 1992).
While drug-using subcultures are often characterized by social relationships
which appear unconducive to the creation and reinforcement of collective social
responsibilities (Friedman et al., 1990), recent research has demonstrated
both the normative importance of sharing in drug users’ social and material
relationships (McKeganey and Barnard, 1992) and of peer support in influencing
behavioral norms and behavior change (Friedman et al., 1992). Friedman
et al. (1992), for example, note the importance of peer support from the drugusing
and nondrug-using friends and relatives in changing drug users’ sexual
behavior (Abdul-Quader et al., 1989) and condom use (Sotheran et al., 1989),
while condom use among female sexual partners of drug users (Tross et al.,
1992) and among women enrolled in methadone treatment programs (Ramos
et al., 1992) have also been found to be associated with peer support and
endorsement.
It is well established that longer-term more experienced drug users often
initiate and “educate” new recruits into appropriate drug-taking behavior (Des
Jarlais et al., 1989). Future interventions might begin by targeting such individuals
within specified social networks with the aim of encouraging them to
impart health recommendations to new recruits into drug use and to individuals
new to their social networks and social circles.
Examples of Social Network and Community Change
Since 1987, the US National Institute of Drug Abuse (NIDA) has funded
a number of demonstration outreach projects targeting drug injectors and their
sexual partners. Of key importance has been the development of “Peer Driven”
and “Indigenous Leader Models” of outreach (Wiebel, 1988, 1996; Broadhead
and Heckathorn, 1994; Koester, 1992) and recent moves toward social network
interventions (Trotter et al., 1993). One of the best established demonstration
projects targeting community changes among injecting drug users is the Chicago
AIDS Community Outreach Intervention Project (Wiebel, 1988, 1996).
Developed initially as a method to intervene and control community outbreaks
of heroin use, it combines epidemiological indicators of risk behavior with
community-based ethnography as a way of designing and implementing appropriate
intervention in the community. The project employs a sequence of strategies
to identify and execute appropriate intervention targets with the overall
aim of facilitating collective change (Wiebel, 1988).
First, the use of qualitative and ethnographic methods and of ethnographers
are outreach workers helps identify community norms and values attached to
health behaviors. Second, the use of former and current drug users as outreach
workers and ethnographers facilitates access to target populations and communication
with target drug users. Third, the repeating of outreach contacts using
a series of complementary risk reduction messages at different locations
maximizes health recommendation exposure and reinforces its content. Fourth,
and most significantly, the targeting of key individuals and their subsequent recruitment
as AIDS Prevention Advocates to enhance and impart health recommendations
to their friends and peers helps encourage socially responsible
beliefs and opinions about health behavior, and, over time, generates a collective
response to behavior change.
The Chicago outreach project can be seen to embody many of the classic
heath promotion strategies developed in communications of innovations theory
(Rogers and Shoemaker, 1971; Rogers, 1983). It is based on sound epidemiological
and ethnographic assessment of the structure and shared norms of identified
social networks, and it employs ideas of homophily in the use of indigenous
outreach workers/ethnographers and key community leaders as peer
educators. This gives indication of the likelihood of change being adopted and
of the possibilities for reinforcing and sustaining the process of change overtime.
The project has been effective in encouraging risk reduction changes over
time and recently has been associated with a declining incidence of new HIV
infections among target populations (Wiebel et al., 1994).
The Chicago project can be seen as a model intervention where ethnographic
observation informs intervention strategy and response (Wiebel, 1996).
One other model intervention, recently developed in Connecticut, provides an
example of a social network intervention where the preexisting structure of
social networks is defined less by a priori ethnographic research than by IDUs
and their peers themselves (Broadhead and Heckathorn, 1994; Grund et al,
1996). The East Connecticut Outreach Project (ECHO) employs ethnography
to make initial contact with IDUs and to develop and implement appropriate
intervention messages, but thereafter encourages IDUs, by a coupon system of
peer-referral, to contact and educate their peers (see Grund et al., 1996, for
a description of intervention methods). All IDUs receiving education from their
peers are encouraged to do likewise and educate peer contacts of their own and
to make contact with the core outreach team for assessments of peer education
given and received. Whereas the Chicago project builds up a picture of the
structure and connections within drug-using social networks by ongoing ethnographic
observations, the ECHO project begins to identify the structure of
preexisting networks by the connections which are made between peers involved
in the outreach intervention.
This means that no a priori assumptions are made as to what a network
is or of how it best operates with regard to the communication of health interventions.
While the Chicago project operates within the paradigm of “Indigenous
Leader Models” of intervention aiming to identify which IDUs within
a network have “leadership” status or potential, the ECHO project operates
within a paradigm of “Peer Driven Models” of intervention which aim to identify
and exploit preexisting channels of communication and influence as they
“naturally occur” within social networks (Broadhead and Heckathorn, 1994;
Grund et al., 1996). The Chicago project aims to diffuse communication within
networks by identifying which individuals appear to have most influence in
maintaining or “policing” network norms, while the ECHO project aims to
diffuse communication by a cue-system based on preexisting power and organi
zational structures (peer to peer rather than “peer educator” to peer). Without
altering the dynamics and nature of communication flow within networks, the
ECHO project aims to encourage a system of “group mediated social control”
where groups or networks of people “police” themselves (see Heckathorn,
1990, for a theoretical outline of Group-Mediated Social Control, and Broadhead
and Heckathorn, 1994, for a description of its application to HIV outreach).
While the ECHO project remains in its developmental stages (6 months
implementation at the time of writing), preliminary findings from process evaluation
suggest that it is well received by IDUs (Grund et al., 1996).
Social network interventions may also be planned and developed on the
basis of systematic social network analysis. Work undertaken by Trotter and
colleagues points to the pragmatic value of an intimate and synergistic linkage
between social network analysis and intervention (Trotter et al., 1993). Such
research has shown how ethnographic research undertaken among key target
populations may lead to a realization that there exists a luck of cohesion within
and across bounded groups of drug users. Exploration of the interplay between
such groups produced a closer understanding of the extent and nature of connections
within different “types” of drug-using networks, which ranged from
“closed networks” displaying an absence of social interaction between members
to “open networks” where membership was based on acquaintance and
acquiescence (Trotter et al., 1993). Social network interventions require an understanding
not simply of whether connections between drug-using individuals
and groups exist but of the nature and suitability of these connections for
the feasibility and diffusion of interventions.
At the time of writing there are few published evaluations of “peer education”
or social network interventions encouraging group-mediated change
among drug users. Ethnographic field research has shown the value of understanding
preexisting channels of communication and influence within drug-using
social networks when targeting key individuals as peer educators or health
advocates. As well as drug dealers and drug users, key individuals selected as
peer educators have included local shopkeepers, bar workers, and managers of
shooting galleries (Oeullet et al., 1991; Murphy and Waldorf, 1991). The
comparative value of peer or “indigenous leader” interventions against those
which tend to be “peer-driven” remains unknown. It is clear, however, that
future HIV prevention interventions need to make fundamental shifts toward
encouraging group rather than individually-mediated change within social networks
of drug users (Rhodes and Hartnoll, 1996).
While little is known about the scope, feasibility, and effectiveness of peer
interventions among drug users, preliminary research findings suggest that
interventions targeting the changing of community norms may do more to
change risk behavior than interventions targeting changes restricted to individuals.
Friedman et al. (1992) reported on the preliminary findings from a community
organizing initiative emphasizing collective identity and participation
among drug injectors in Brooklyn, which they show achieved greater levels of
risk reduction in sexual and drug-taking behavior than street-based individually-
focused outreach (see also Jose et al., 1996).
Evaluation of syringe exchange in Rotterdam shows a higher return rate
of equipment, a higher retention rate of attendance, and higher levels of risk
reduction from an intervention recommending collective change and encouraging
drug users to take care of their friends and peers than from a similar intervention
recommending individual change (Grund et al., 1992). Similar findings
are reported from peer-based syringe exchange projects in South Australia,
which shows that peer-based programs were more effective in distributing
equipment and in reaching new injectors than nonpeer-based programs (Herkt.
1993).
Peer education and peer endorsement may be particularly important in the
context of behaviors most private and subject to most social and public policing,
such as sexual behavior (Ehrhardt, 1992). The targeting of peer influence
as a method of initiating and reinforcing change in drug users’ social and
sexual relationships can be viewed as a first step toward encouraging a process
of community and collective change. It also may be considered the first
step toward providing the foundation for community mobilization and organization
among drug users in defining and controlling collective norms and values
about health behavior and in confronting the social constraints which marginalize
equity to public health.
CONCLUSION
An understanding of the obstacles to individual behavior change requires
an understanding of the social context of risk behavior. Without confronting
the obstacles to behavior change it is unlikely that opportunities for change will
be created. This demands an approach to research and intervention which recognizes
both an epidemiology and a sociology of risk behavior. It is timely for
a shift in the direction and emphasis of most research and intervention designs
toward a conception of risk behavior and behavior change which encompasses
and combines a vision of the social as well as the individual. If intervention
is to be effective in changing individual risk behavior, then it must also be
effective in changing the social context of risk behavior. It is timely for research
and intervention to consider the problem of the social and of social
change. This is one of the challenges of the second decade of AIDS. When
facing this challenge, drug users themselves may prove to be the most helpful
advocates of innovation and change.