Introduction

Religion influences various health outcomes including mortality, depression, and drug use [13] as a key factor that shapes behavior within many communities [4]. Early in the HIV epidemic, religion was considered detrimental to prevention efforts, as some religious communities considered AIDS to be a punishment for sin and opposed the public promotion of condom use and forms of sex education [57]. More recently, theologies of harm reduction have emerged within some religious groups [8] and religious organizations have become major HIV intervention partners in multiple countries [5].

Research on religion and HIV emphasizes the role of religion as a resource for people living with HIV, and suggests that religion is useful in helping individuals cope and find meaning [9, 10]. Less research is available on pathways through which religion may influence HIV risk behavior. Previous research that suggests religious tenets and practices impact HIV risk has focused on differences in HIV transmission across religious affiliations [11]. A number of studies have examined religious affiliation and HIV in Africa [11, 12], but none have examined associations between religious affiliation or religiosity and HIV risk worldwide. The limited emphasis on how religion may be a risk or protective factor, as well as an avenue for HIV prevention, may be related to contentions between religion and HIV preventative approaches described above. This may be shifting; however, as prevention efforts increasingly incorporate the importance of structural influences on risk, including gender inequalities, stigma, community networks, and behavioral norms, which may be influenced by religion. Attention to structural influences on risk highlights cultural, political, economic, and religious factors that may be related to risk behaviors, and emphasizes the importance of understanding unique risk contexts [13, 14]. Innovative approaches to HIV prevention that draw from local contexts through utilizing religious venues, partnerships, and frameworks (including, for example, an emphasis on specific values such as destiny, compassion, and purity) have demonstrated promise in reducing HIV risks [15].

A review of research to date is needed to consider how religion may be associated with HIV risk and protection and pathways by which religion may influence risk. This systematic review considers studies published on religion and sexual HIV risk since 1988. For the purposes of this paper religion is defined as including two primary components: (1) religious affiliation, and (2) a level of adherence to, participation in, influence of, or identification with a set of beliefs, referred to as “religiosity.” The review examines study methodologies, definitions and conceptualizations of religion, locations in which studies were conducted, and sexual risk outcomes associated with religion. We also discuss mechanisms that explain these associations and gaps in research on the relationship between religion and sexual HIV risk.

Methods

A total of 761 abstracts were reviewed through electronic searches of PubMed (n = 395), PsycInfo, Medline, and Biological Abstracts (n = 96), Sociological Abstracts (n = 56), and Web of Science (n = 214). Searches were conducted in February 2012, considering sources from 1985 through 2012. The search included three key areas: (1) HIV or AIDS, (2) sexual risk, and (3) religion (or religiosity, religious, church, Muslim, Islam, Christian, Christianity). Duplicate abstracts were excluded, as were reviews, dissertations, or abstracts focused on outcomes unrelated to sexual HIV risk.

Of the 157 remaining articles, six were excluded because they consisted of data presented in other papers in the sample. An additional 11 were excluded as they measured outcome variables not directly related to sexual HIV risk. Three articles were excluded for not specifying either a religious affiliation or a definition of religiosity. Data from each study was entered into SPSS version 20, after which frequencies and distributions were examined. We then conducted Chi square goodness of fit tests to identify differences between groups of studies (Fig. 1).

Fig. 1
figure 1

Article selection

Results

Study Methodologies

The 137 studies included in this review were published between 1988 and 2012. Most studies employed cross-sectional designs (n = 113, 82.5 %). Twenty-two were longitudinal studies and two were randomized controlled trials. Four studies (2.9 %) included mixed methods or a qualitative component. Nearly one fourth (n = 33, 24.1 %) of the studies included national samples. Sample size varied from 33 to 216,000 individuals, with a mean of 4,714 and a median of 880. Four studies considered countries as the unit of analysis rather than individuals. Nearly equal numbers focused on adult populations only (n = 52, 38.0 %) and youth or young adults only (n = 53, 38.7 %); the latter included research with secondary school and college students. Another 23.4 % (n = 32) of studies considered mixed samples, including youth and adults, or a participant age range beginning below and exceeding 18. When age was not specified, the sample was considered to include only adults. Most studies were conducted with both women and men (n = 85, 62.0 %), while 28 studies (20.4 %) reported results solely for women and 24 (17.5 %) only included men.

Definition of Religion

Religion was conceptualized differently across studies. Most studies only examined religious affiliation (n = 57). Another group of studies (n = 48) only measured participant religiosity, without describing the religious affiliation of the sample. Finally, 32 studies measured both religious affiliation and religiosity.

Study Locations

Most studies were conducted in Africa (n = 61, 44.5 %), and North and South America (n = 53, 38.7 %), with 47 of the latter (34.3 %) conducted in the United States (US). Less commonly considered regions included Europe (n = 8, 5.8 %), the Western Pacific (n = 7, 5.1 %), South-East Asia (n = 4, 2.9 %), and the Middle East (n = 2, 1.5 %). Two studies were conducted in multiple regions (1.5 %).

Sexual Outcomes Associated with Religion

Because of differences in how religion and religiosity were presented, risk outcomes are detailed according to studies which describe: (1) religious affiliation only (n = 57), (2) religiosity only (n = 48), and (3) both religiosity and religious affiliation (n = 32).

Religious Affiliation and Sexual HIV Risk (n = 57 Studies)

Studies that only compared participants by religious affiliation differed from those that included religiosity in multiple ways. For example, the former were more likely to be conducted outside the US, with 61.1 % of articles conducted outside the US only measuring religious affiliation, as compared to only 4.2 % of studies conducted within the US (p < .001). Most studies conducted in Africa only compared outcomes of participants according to religious affiliation (n = 44, 72.1 %). Each of the four articles that considered countries rather than individuals as the unit of analysis described countries by their predominant religious denomination and did not describe religiosity.

Fifty-seven studies examined how sexual HIV risk varied according to religious affiliation, without measuring religiosity. As observed in Table 1, of 57 studies, 42 included participants who were designated as Muslims, 32 as Christians (including 3 as Orthodox), 27 as Catholics, 24 as Protestants, and 13 as ‘Traditional’ [11] or ‘Indigenous’ (two), all of which were conducted in Africa. Forty-one had designations beyond these categories, primarily ‘other’ or ‘other and none’ but nine studies identified other Christian faiths, five included the Hindu religion, three included Judaism, and Buddhist, Animist, and Agnostic affiliations were included in one study each. This review will compare studies according to the affiliations described.

Table 1 Religious affiliation only (n = 57)

Study outcomes included an examination of how affiliation is associated with condom use (n = 24), HIV positive status or prevalence (n = 19), having multiple partners or a partner outside of the primary relationship (n = 17), sexual initiation (n = 10), having sex with a commercial sex worker (n = 9), sexual experience (n = 6), STI prevalence or symptoms (n = 4), and other outcomes (n = 9), including general risk indicators. Over half of the studies in this group measured multiple sexual risk outcomes (n = 29, 52.7 %).

Fifty-one of the 57 studies found a particular religious affiliation to be associated with reduced sexual risk (n = 31), increased risk (n = 10), or both reduced and increased risk (through different pathways) (n = 10) for acquiring HIV. Six studies found no association between affiliation and sexual HIV risk (see Table 1). Protective effects were found within the following affiliations: Muslim (n = 16), Christian (n = 4), Protestant (n = 3), Catholic (n = 2), Muslim and Hindu (n = 2), Muslim and Christian (n = 1), Muslim and Traditional (n = 1), Evangelical (n = 1), and among those with any affiliation (n = 1). The remaining articles found risk associated with the Muslim (n = 4), Christian (n = 3), ‘Traditional’ (n = 2), or Catholic (n = 1) affiliations.

Of the ten studies reporting both protection and sexual risk as associated with a religious affiliation, five described an association of sexual risk and protection with being Muslim, four with being Christian, and one described protection as associated with Christianity and risk as associated with any affiliation. Of these ten studies, five reported that affiliation was protective in regards to some outcomes but was associated with lower rates of condom use. Among the other five, two found affiliation to be associated with lower prevalence but higher risk, one found the same affiliation to be associated with later sexual initiation and fewer sexual partners but a higher likelihood of paying for sex, and two identified some affiliations to be associated with lower risks and others with higher risks. One study also identified Islam to be associated with a lower prevalence among men but higher prevalence among women (see Table 1).

Religiosity and Sexual HIV Risk (n = 48 Studies)

Articles exploring differences in sexual risk by religiosity without specifying the religious affiliation of the sample were predominantly conducted in the US, with 72.9 % (n = 35) of studies conducted in the US only examining religiosity. Studies with youth or young adults only were also more likely to only consider religiosity (p < .001) and to be conducted in the US (p < .01) when compared to studies with mixed or adult only samples.

Forty-eight papers examined how sexual risk for HIV varied according to religiosity without describing the religious affiliation of the sample. Religiosity was defined in various ways. The most frequently used marker of religiosity was attendance or participation in religious services (n = 28). Other commonly used indicators included the importance of religion in respondent’s lives (n = 11), how religious they were (n = 7), how religion played a role in their upbringing or childhood (n = 4), and the influence of religion on their behavior (n = 4). Other indicators were also common (n = 19), with nine using religiosity scales adapted or obtained from various sources. Two of the nine articles cited the same scale [152]. More than half of the studies in this group included more than one indicator of religiosity (n = 35). A few also reported measuring religion but did not specify the name of the affiliation, or referred to a denomination without specifying that the sample identified with this affiliation.

Sexual HIV risk outcomes analyzed in studies focused on religiosity included condom use (n = 21), having multiple partners (n = 14), sexual initiation (n = 9), a general sexual risk indicator or composite (n = 15), and other outcomes (n = 14), with 20 studies considering multiple of these outcomes. Of those in the ‘other’ category, two included accessing commercial sex, one considered HIV seropositive status, and no studies considered STI status or symptoms.

Findings are reported in Table 2. Most studies found religiosity to be protective, with participants who were more religious experiencing lower levels of sexual risk (n = 30). Another eight studies found religiosity to have no association with risk, five found religiosity to be associated with increased sexual HIV risk, and three cited religiosity as being associated with both an increase and decrease in risk. Of the three that found both risk and protective results, one found religiosity to be associated with risk for some men and protection for women; another found attendance to be associated with abstinence, but among sexually active participants, attendance was associated with lower levels of condom use; and the third found a religious upbringing to be associated with a lower likelihood of sexual initiation as well as lower sexual literacy. Four of the five studies that identified religiosity as a risk factor found religiosity to be associated with lower levels of condom use. Two studies did not fit this framework of religiosity relating to risk or protection: one reported an association between ‘what one’s religion says about gay sex’ and risk behavior [16] and another described increased relevance of religion among participants who had not had sex [17].

Table 2 Religiosity Only (n = 48)

Both Religiosity and Affiliation (n = 32 Studies)

Thirty-two studies included measures of both religious affiliation and religiosity. As presented in Table 3, 19 studies presented results regarding participants who were Catholic. An equal number included Muslim (n = 13) and Protestant (n = 13) participants, and nine included Christian participants. Most studies (n = 25) included other categories as well, such as ‘none’ or no religion (n = 10), ‘other’ (n = 9), other Christian denominations (n = 8), Jewish (n = 4), Buddhist (n = 2), ‘African traditional’ (n = 1), and Animist (n = 1).

Table 3 Religiosity and Affiliation (n = 32)

Religiosity was measured by frequency of attendance or participation in religious services (n = 17), the importance of religion in participants’ lives (n = 8), how religious they considered themselves to be (n = 6), the influence of religion on their behavior (n = 6), and the role of religion in their upbringing (n = 2). Many studies (n = 15) considered other indicators, and 17 included more than one measure of religiosity. Five included different measures used by other researchers.

Sexual risk behaviors considered in these studies included condom use (n = 13), sexual initiation (n = 10), having multiple or extra partners (n = 8), a general sexual risk indicator (n = 4), or another indicator (n = 13). Thirteen studies included more than one indicator.

Most studies in this group described findings related to religiosity (n = 31), and ten also reported differences according to religious affiliation. Religiosity was found to be associated with lower levels of sexual HIV risk in 22 studies, associated with higher risk in three, associated with both increased and decreased risk in one, and was not significant in five studies. Results on the influence of religious affiliation were mixed, with reduced levels of sexual risk associated with being Catholic (n = 2), non-Catholic (n = 2), Christian or Orthodox Christian (n = 2), Muslim (n = 1), and non-Muslim (n = 1). Risk was associated with changing Christian denominations in one study, and in another, having a ‘traditional’ Jewish family background was associated with both risk and protection when compared a secular upbringing (see Table 3).

Differences Across Study Categories

Differences were apparent across the three conceptualizations of religion studied (affiliation only, religiosity only, both affiliation and religiosity). Whether a study considered religious affiliation or religiosity was primarily determined by study region. Studies conducted outside the US examined differences in sexual HIV risk outcomes according to religious affiliation, with nearly three quarters of studies in Africa considering participants only by affiliation, compared to less than 5 % of studies in the US. Alternately, a large group of studies (n = 48), most of which were conducted in the US, explored religiosity without describing the sample’s religious affiliation.

Among all studies that examined religiosity (n = 80), findings varied according to study region, participant age group, and the risk outcome considered. Approximately two thirds of studies reporting results regarding religiosity in Africa (n = 18) and the Americas (n = 46), and all studies in the Western Pacific (n = 7) and the Middle East (n = 2) found higher religiosity to be associated only with reduced sexual risk. Of the five studies reporting religiosity results in Europe, only one found higher religiosity to be associated with reduced risk.

Studies with youth were more likely to measure religiosity only and to be conducted within the US, when compared to studies with mixed or adult only samples. Additionally, studies tended to use different sexual risk outcomes according to participant age. Across the 137 studies, sexual initiation was more likely to be included as a measure of sexual risk among youth and young adult samples (32.1 %, 17 out of 53 studies) when compared to mixed age or adult only samples (14.3 %, 12 out of 84, p < .05).

Over half of the studies that included condom use as an outcome found religiosity to be associated with lower levels of sexual risk (57.6 %, or 19 out of 33) while 71.1 % of studies (32 out of 45) that did not include condom use found religiosity to be protective. This difference was not significant as the .05 level (p = .154). Most studies that found religiosity to be associated only with increased sexual risk examined condom use as an outcome (77.8 %, or 7 out of 9), with six of these studies identifying an association between religiosity and lower rates of condom use, or feelings that condoms are ineffective.

Mechanisms of Sexual HIV Risk

Half of the studies (n = 68) included some description of the mechanisms that link religiosity or religious affiliation to sexual risk. Mechanisms were minimally described, generally without theoretical support. Most commonly, religion and religiosity were described as influencing sexual HIV risk through behavior norms or belief systems related to sexuality and sexual risk (see Table 4). Thirty-five studies described behavioral norms as influencing sexual risk. For example, studies attributed lower rates of sexual risk to an emphasis on premarital virginity or abstinence [1821], restrictions on alcohol consumption [22, 23], norms of monogamy [24, 25], or to less tolerance of sexual freedom [26]. A few studies that did not find religion to have a protective effect on risk also identified behavioral norms as explaining associations, for example, attributing no association between religion and risk to the promotion of condom use at similar levels across affiliations [27] and citing deterioration in sexual sanctions due to the influx of Christianity (in rural Thailand) [29].

Table 4 Mechanisms (n = 68)

Studies emphasizing the mechanism of belief suggested religion leads to lower levels of sexual risk through decisions to avoid high-risk behaviors [31, 32]. Some studies ascribed risk to disregarding religious beliefs, suggesting that a failure to follow religious teachings led to risky behavior [33], while others pointed to risks of adhering to beliefs, positing that following religious convictions against condom use and planned sexual activity led to unprotected, unplanned sex [34].

Thirteen studies pointed to the role of social organization and social influence. For example, protective effects of religiosity were attributed to social support and connection that comes through participation and interaction in organized religious activities [3537], fears of disclosing condemned activities [38], high social control [12], or living in an area with less exposure to alcohol and nightclubs [39].

Nine studies cited circumcision as a factor explaining differences or similarities in risk across affiliations; and an additional nine studies described another mechanism, such as communication between partners, gender roles, polygamy, alcohol use, or education.

Discussion

This systematic review suggests that both religiosity and religious affiliation influence sexual HIV risk in various ways. More than 40 % of studies in this review considered religious affiliation but not measures of religiosity. Approximately one third of studies addressed religiosity but did not report affiliation, and half of the studies reviewed did not examine potential mechanisms. Of those that described mechanisms linking religion and sexual HIV risk, major mechanisms identified included (1) behavior norms, (2) beliefs, and (3) social influence.

Results relating to religious affiliation suggest that engagement in sexual risk may vary according to religious affiliation. Protective effects were found most often among groups of people affiliated with Islam, however, most of these studies utilized HIV prevalence or seropositive status as the sexual risk outcome (n = 14), rather than indicating specific sexual risk behaviors. Although this leaves little evidence as to what aspects of religious affiliation may be associated with particular sexual risk behaviors, 35 of the studies that only examined religious affiliation also incorporated discussion on possible mechanisms between affiliation and risk. Most of the studies that found Islam to be protective ascribed this relationship to the role of behavior norms or higher rates of circumcision among Muslim men when compared to those of other affiliations. Studies identifying affiliation to be associated with risks suggested risks resulted from restrictions on behaviors, lower levels of adherence to teachings, traditions influencing the ability of women to avoid risk, and polygamy.

Studies that only measured religious affiliation were predominantly conducted outside of the US. Researchers may be interested in risk outcomes associated with denominations less commonly found in the US. For example, there may be an interest in studying differences between Muslims and Christians in locations where large groups of people identifying with each affiliation reside, such as in Africa. Without attention to other aspects of religion such as levels of participation and adherence, beliefs, and practices, this research provides only a cursory understanding of how religion influences behavior.

Studies only considering religiosity were conducted primarily in the US. Researchers may be less likely to consider religious affiliation in the US if they assume the reader is aware of the sample’s religious affiliation or if they consider affiliation differences to be less relevant in predicting behavior than religiosity. When the sample’s affiliation is not measured or described, study findings are limited. The large number of affiliations present in the US may present difficulties for researchers, but these can be addressed through preliminary screening or utilization of measures conducted with similar groups of people. As the research included in this review suggests affiliation is associated with differences in sexual risk outcomes, it is critical to measure and report affiliation in addition to religiosity.

The lower likelihood of observing religiosity to have a protective effect among studies conducted in Europe may suggest that religiosity is less prevalent, or that it has a weaker effect, in different parts of the world. However, study numbers by location are insufficient to reach conclusions about varying levels of religiosity or religiosity’s impact.

Studies on sexual risk in the US appear to be more narrowly focused on religiosity among younger samples when compared to studies conducted elsewhere that tended to include adult or mixed age samples and measure only affiliation. Additionally, studies conducted among youth and young adults only were more likely to utilize sexual initiation as a measure of risk, as sexual initiation is a less meaningful indicator of sexual risk for populations of adults who are sexually active.

Studies that found religiosity to be associated with higher rates of sexual HIV risk were most likely to conceptualize risk as lower rates of condom use. Although the smaller number of studies that identified a link between religiosity and risk tended to include condom use as the outcome, most studies that considered condom use as an outcome found religiosity to be protective in regards to sexual HIV risk overall. Some of these studies examined multiple outcomes without reporting findings between religiosity and condom use. Others described an association between religion and condom use but did not identify potential mechanisms. Of the studies that discussed mechanisms linking religion and condom use, studies that identified a protective effect pointed to positive social influences resulting from religious leaders, a desire to adhere to social norms, and fears about HIV. Studies identifying religiosity to be associated with risk (lower rates of condom use) pointed to opposition from religious groups towards condom sales as well as beliefs linking sexual behavior to punishment. Although the relationship between religiosity and condom use appears to be context specific, additional reporting of findings and research on this topic is needed to identify common pathways between religious influence and condom use.

A limited number of studies examined mechanisms between religion and sexual risk outcomes. Most often these studies attributed the influence of religion on sexual behavior to religious behavior norms, beliefs, and social support. Behavior norms were most often considered the pathway through which religion influenced sexual risk outcomes, suggesting that belonging to a religious community coincides with social expectations and group influence on sexual norms. Studies citing beliefs suggest that an individual’s appropriation of religious convictions (i.e. regarding monogamy or abstinence) influenced their individual risk behaviors. The emphasis of other studies on social support pointed to the influence of social control, supports and expectations, communication, networks, and fears of social isolation.

Specific religious practices that may form causal pathways between religion and risk were also cited by some studies (particularly those on affiliation only), including circumcision, polygamy, and alcohol use. Other studies pointed to aspects of religiosity such as a sense of spirituality, divine pressures, moral reasoning, or altruism and a concern for others as potentially influencing risk behaviors. Some suggested tradition and societal expectations were more pertinent than religious influence or that the influence of religion was minimal when compared to other cultural influences such as the media.

While the explanations put forward compose a list of mechanisms that could be understood through various theoretical frames relating to gender and power, social capital, and decision making, additional research is needed to demonstrate pathways through which religion influences risk. Adamczyk and Hayes [40] explored differences in premarital and extramarital sex across religious affiliation in multiple countries and found that the lower rate of premarital sex among married Hindus and Muslims was not explained by an earlier marriage age. Furthermore, lower rates of extramarital sex among nations with a higher percentage of Muslims were not explained by restrictions on female mobility. Conducting in depth analysis of HIV and religion in Africa, Trinitapoli and Weinreb [151] explore mechanisms between religious affiliation, religiosity, and sexual HIV risks, suggesting (1) abstinence is more common among people who attend religious services more regularly and is primarily explained through attitudes about acceptable sexual behavior, but does not vary by affiliation; (2) monogamy varies by affiliation, is more likely among those who attend regularly, and can be explained by the mechanism of social control or monitoring from religious leaders; and (3) condom use is not strongly effected by affiliation, and religiosity does not impact condom use or attitudes towards condoms. They did find, however, that positive religious leader attitudes towards condoms coincided with higher levels of condom use among adherents. Additional research that considers individual and community level religious influences could help disentangle ways in which religion impacts sexual risk in various contexts.

Implications for Future Research

Future research on religion and sexual HIV risk needs to: (1) incorporate measures of both religious affiliation and religiosity; (2) consider specific sexual risk outcomes in addition to HIV prevalence; (3) determine which measures of religiosity are most relevant to understanding sexual risk contexts for particular groups of people; (4) more thoroughly examine mechanisms by which religion may influence sexual risk behaviors; and (5) conduct qualitative research to further examine how religious affiliation and religiosity are linked to specific sexual risk behaviors.