Abstract
Intravaginal practices (IVPs) are associated with an increased risk of bacterial vaginosis and may play a role in HIV transmission. The objective of this study was to identify the importance of factors underlying the decision to engage in IVP using conjoint analysis; a novel statistical technique used to quantify health-related decisions. This study was a cross-sectional study. HIV-infected women in Zambia completed audio computer-administered self-interview questionnaires assessing demographic, risk factors and IVPs. Reasons for engaging in IVPs were explored using conjoint questionnaires. Conjoint analysis was used to identify the relative importance of factors for engaging in IVPs. Results of the conjoint analysis demonstrated that hygiene was the most important reason for engaging in IVPs (mean importance score = 61, SD = 24.3) followed by partner’s preference (mean importance score = 20, SD = 14.4) and health (mean importance score = 17, SD = 13.5). When making the decision to engage in IVPs, women rank the importance of hygiene, partner preference and health differently, according to their personal characteristics. The use of conjoint analysis to define the characteristics of women more likely to engage in specific practices should be used to develop tailored rather than standardised IVP interventions, and such interventions should be incorporated into clinical practice and women’s health programmes.
Keywords: Women, HIV, intravaginal practices, bacterial vaginosis, Africa, conjoint analysis
Introduction
Intravaginal practices (IVPs) are the introduction of products inside the vagina for hygienic, health or sexual reasons.1 Although women may perceive IVPs as a beneficial practice, IVPs are associated with numerous negative health outcomes, including the development of bacterial vaginosis (BV) and obstetric and gynecological complications. BV increases the risk of acquisition and transmission of sexually transmitted infections (STIs) and HIV, and therefore interventions to decrease IVPs may be important components of HIV prevention plans.2–5
Zambia is a sub-Saharan country severely impacted by the HIV epidemic and the prevalence of HIV among urban women is as high as 35%.6 IVPs are common and may play a role in exacerbating the HIV epidemic.7 IVPs among Zambians are strongly influenced by social and cultural beliefs, endorsed by men, and traditionally introduced to women during early adolescence by ‘Alengizis’ – marriage counsellors who guide young girls in preparation for becoming a woman.8–10 The primary reasons why women engage in IVPs are hygiene, perceived health benefits and men’s preferences. We have previously described that hygiene reasons appear to be the primary determinant for most IVP in Zambia.8,9 Due to their detrimental association with health, there is a need to develop culturally-tailored interventions to decrease or ameliorate IVPs in Zambia and in other sub-Saharan countries in which IVPs are prevalent.8,9,11,12 It is equally necessary to clarify the role of each of these reasons in making the decision to engage in IVPs.
Conjoint analysis (CA) has been used in market research to clarify and quantify preferences for different products. CA analysis is a technique that can be used to quantify the relative importance of different aspects of a decision by asking participants to rate or choose a product or service based on a set of characteristics of the product or service. In CA multiple characteristics are presented together (conjointly), and trade-offs must be made during the evaluation process. CA presents scenarios that are hypothesised to be closer to ‘real life’ than those assessed by asking each characteristic independently.13–15 Conjoint designs are increasingly being applied in HIV and other health research studies,16–22 and can be utilised to identify the relative importance of factors influencing health-related decisions, such as the decision to engage in IVPs. Using conjoint questionnaires, participants identify preferences between different combinations of attributes of a health decision, and CA then enables the quantification of the importance of each attribute in comparison with the others.13,14
This study sought to identify the relative importance of factors underlying the decision to engage in IVPs using CA, in order to develop culturally appropriate IVPs interventions for Zambian women. It was hypothesised that when comparing hygiene, male preferences and health, women would be most likely to engage in IVPs for hygienic reasons.8,9
Material and methods
Ethics statement
Institutional Review Board (University of Miami Miller School of Medicine) and Research Ethics Committee (University of Zambia) approvals were obtained prior to recruitment, assessment and any study-related interventions. Participants were provided with information about the study and assured of confidentiality of information and study records. Voluntary signed informed consent was obtained from every participant prior to participating in the study.
Study procedures
Study activities took place at a Community Health Center in urban Lusaka, Zambia. Data presented are part of a larger study to develop and test a culturally-tailored intervention to decrease IVPs and BV among HIV-infected women in urban Zambia. In order to develop questionnaires addressing IVPs, focus groups, key informant interviews and introductory meetings with the clinic staff and local Community Advisory Boards were conducted at two Community Health Centers. Participants were primarily referred by enrolled peers or clinic staff, and presented documentation of HIV infection at the time of enrollment. Documentation of HIV infection included clinic referral confirming the HIV status of a participant and was confirmed using antiretroviral registers and clinic files. Participants were women, 18 years of age or older, sexually active, engaging in receptive vaginal intercourse with men, engaging in IVPs in the prior month, HIV infected, receiving ART, not pregnant, not on contraceptive medications or with an intrauterine device, and living in the Lusaka Metropolitan Area.
Participants completed audio computer-administered self-interview (ACASI) questionnaires assessing demographic, sexual risk factors and IVPs. All audio and consent materials were provided in participants’ preferred local languages (English, Bemba, Nyanja).
Outcome measures
Demographics, sexual risk factors and medical history
This questionnaire included age, marital status, number of children, yearly income, educational level, current partner’s HIV status, mode of infection with HIV, number of partners in the prior month, sexual modalities (oral, anal sex), history of exchanging sex for money in lifetime and use of condoms in the prior month. Medical history was self-reported and included date of HIV infection and CD4 cells per microlitre.
Vaginal practices
This questionnaire assessed participants’ use of IVPs and was adapted from a previous study.9 The study coordinator described IVPs to the participants prior to completing the questionnaire as ‘the insertion of products inside the vagina for hygiene, health or sexual pleasure’, and clarified that IVPs did not refer to external vaginal cleansing. A picture of the female genital anatomy was shown to the participants to illustrate. The questionnaire addressed products used for IVPs as well as recency and frequency of IVPs. Questions to assess product used were on a dichotomous scale, ‘yes’ = 1, ‘no’ = 0, and each product was assessed individually. Products identified in the focus groups and key informant interviews as used for IVPs included water alone, soap, cloth or a sponge or a rag, herbs or flowers taken directly from the land, traditional medicines given by traditional healers, vinegar, salt, beer and yogurt. Questions assessing recency and frequency of use were scored using a Likert scale. The recency scale was scored, ‘within the last two days’ = 1, ‘within two days to one week’ = two, ‘within 1–2 weeks’ = 3, ‘within two weeks to one month’ = 4, ‘more than one month’ = 5. The frequency scale was scored, ‘daily’ = 1, ‘weekly’ = 2 and ‘monthly’ = 3.
Conjoint questionnaire
Reasons for engaging in IVPs were explored using a conjoint questionnaire developed from information collected from focus group discussions (FGD) with men and women and key informant interviews (KI) with marriage counsellors (Alengizis). Details of the FGD and KI have previously been described.8 Three common elements for engaging in IVPs were noted in both the FGD and KI: hygiene, partner preference and health. These elements were included as ‘attributes’ in the conjoint questionnaire. The study team then assigned different plausible ‘levels’ to each attribute based on previous research with this population. Levels assigned were positive (e.g., ‘IVPs are good for your hygiene’), neutral (e.g. ‘IVPs have no effect on your health’) and negative (e.g. ‘Your partner does not like that you engage in IVPs’). A set of nine profiles, including different combinations of attributes and levels, was generated using SPSS Statistics.23 Women were asked to rate how likely they would be to engage in IVPs for each profile presented, from 1, ‘very unlikely’ to 9, ‘very likely.’ In order to enhance comprehension, a pictorial depiction of each profile was presented to the participant in conjunction with the question, presented in both audio and text. The pictures represented a woman (suggesting hygiene preference), a health care provider (suggesting health preference) and a man (suggesting partner preference) with facial and body expressions indicating the different levels of response (positive, neutral or negative). Conjoint questions were grouped in sets of three and distributed between sets of IVP questions to reduce participant burden. Table 1 presents the conjoint questionnaire.
Table 1.
Questions used to assess the determinants for engaging in intravaginal practices (IVPs) using conjoint analysis (CA).
| Vaginal practices may have an effect on your HEALTH, your HYGIENE and the relationship with your PARTNER. Please rate from 1 to 9 (very unlikely to very likely) how likely would you be to engage in vaginal practices if they were: | |||||||||
|---|---|---|---|---|---|---|---|---|---|
| Are BAD for your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER does NOT LIKE them | Very unlikely | Somewhat likely | Very likely | ||||||
| Have NO EFFECT on your HEALTH | |||||||||
| Are GOOD for your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER LIKES them | Very unlikely | Somewhat likely | Very likely | ||||||
| Have NO EFFECT on your HEALTH | |||||||||
| Have NO EFFECT on your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER DOES NOT CARE | Very unlikely | Somewhat likely | Very likely | ||||||
| Have NO EFFECT on your HEALTH | |||||||||
| Are GOOD for your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER DOES NOT CARE | Very unlikely | Somewhat likely | Very likely | ||||||
| Are BAD for your HEALTH | |||||||||
| Have NO EFFECT on your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER LIKES | Very unlikely | Somewhat likely | Very likely | ||||||
| Are GOOD for your HEALTH | |||||||||
| Are BAD for your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER DOES NOT CARE | Very unlikely | Somewhat likely | Very likely | ||||||
| Are GOOD for your HEALTH | |||||||||
| Are BAD for your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER LIKES | Very unlikely | Somewhat likely | Very likely | ||||||
| Are BAD for your HEALTH | |||||||||
| Are GOOD for your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER DOES not like | Very unlikely | Somewhat likely | Very likely | ||||||
| Are GOOD for your HEALTH | |||||||||
| Have NO EFFECT on your HYGIENE | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 |
| Your PARTNER DOES NOT like | Very unlikely | Somewhat likely | Very likely | ||||||
| Are BAD for your HEALTH | |||||||||
Each profile was identified by including different combinations of attributes and levels using SPSS statistics and questions were created based on the different attributes of individual profiles.
Conjoint questions were grouped in sets of three and distributed between sets of IVP questions to reduce participant burden.
Statistical analysis
Descriptive analyses were performed to describe demographic, medical and sexual risk factors. Since CA cannot be performed when all of the profiles are given the same rating by a participant, the variability of profile ratings within participants was examined and those participants who did not vary in their ratings of profiles were not included in analyses of the conjoint survey. The nine different profiles and rating scales are illustrated in Table 1.
For those participants who can be analysed, CA is completed on an individual level (i.e. one analysis for each person), and the results are then averaged across the entire sample or compared between subgroups. In this CA, the likelihood of engaging in vaginal practices was modelled as the sum of a constant plus parameter estimates associated with the levels of the three attributes (i.e. hygiene, partner and health) appearing in the questionnaire. The outcome variables in the conjoint model were the participant ratings for the scenarios, and the categorical predictors were the attributes, with categories corresponding to the levels of each attribute, i.e. ‘positive, no effect, negative.’ Categorical variables were coded such that the estimate for each level represents deviation from the constant, and parameter estimates were obtained using ordinary least squares multiple regression techniques. Using the parameter estimates, importance scores were computed. This was done by first calculating the absolute value of the difference between the largest parameter estimate and the smallest estimate associated with the levels of each attribute. The range is then expressed as a percentage of the sum total of all ranges of all factors, and represents the importance of that factor relative to the other factors.13,14 Mean importance scores were calculated for the entire sample and transformed into a 1–100 scale. Importance scores were then compared across groups of women using independent sample t-tests. IBM SPSS Statistics for Windows version 21 was used for analyses.23 A p value <.05 was considered significant.
Results
Characteristics of study participants
One hundred and twenty-eight HIV-infected participants were enrolled. Two participants provided the same rating for all of the questions and were not included in the analysis. Demographic characteristics, medical history and risk factors are described in Table 2. Median age was 37.3 years. Most women had a very low income (three-quarters of women earned less than 150 US dollars per month) and about half of the participants had only primary education (46%). The majority of participants were married to a HIV-infected partner (68%). Ten percent had a history of exchanging sex for money or gifts. About half of the participants had been infected with HIV for more than one year but less than five years, and half self-reported CD4 counts over 500 cells/μl (47%).
Table 2.
Baseline socio-demographic characteristics, risk behaviours, medical history and intravaginal practices (IVPs) in the study population (n = 126), (number, %).
| Demographics | |
|---|---|
| Age in years (range) | 37.3 years (24–60) |
| Married or stable relationship | 126, 100% |
| Monthly income (approximately – US dollars) | |
| Less than USD150 | 98, 77.8% |
| More than USD150 | 28, 22.2% |
| Education (n = 113) | |
| Primary education | 58, 46% |
| Higher than primary education | 68, 54% |
| Sexual risk factors | |
| HIV-infected male partner | 86, 68.2% |
| Mode of HIV infection | |
| Sexual | 106, 84.1% |
| Unknown | 20, 15.9% |
| More than one partner in the prior month | 44, 34.9% |
| Oral sex in the prior month (n = 94) | 16, 17% |
| Anal sex in the prior month (n = 94) | 4, 4.3% |
| Exchange sex for money or gifts (lifetime) | 13, 10.3% |
| Condom use in every act in the prior month (n = 94) | 4, 4.2% |
| Medical history | |
| Time since HIV diagnosis | |
| Less than one year | 12, 9.4% |
| More than one year | 75, 58.6% |
| CD4 T cells per microlitre (n = 106) | |
| More than 500 | 49, 47% |
| One hundred to 500 | 49, 46.2% |
| Less than 100 | 7, 6.6% |
| Intravaginal practices | |
| Use of IVPs in the month prior to enrollment | 126 (100%) |
| Mean age of IVP initiation in years (range) | 15.8 years (8–41 years) |
| Last use of IVPs? | |
| In the last two days | 105, 83.4% |
| Two days to one week | 5, 3.9% |
| One to two weeks | 2, 1.6% |
| Two weeks to one month | 14, 11.1% |
| Products used for IVPs in the prior month | |
| Water use | 118, 93.6% |
| In the last two days (n = 117) | 109, 93.2% |
| Daily | 111, 94.1% |
| Weekly | 2, 1.7% |
| Monthly | 5, 4.2% |
| Soap | 89, 70.6% |
| In the last two days (n = 89) | 87, 97.8% |
| Daily | 83, 94.3% |
| Weekly | 2, 2.3% |
| Monthly | 3, 3.4% |
| Cloth | 68, 53.9% |
| In the last two days (n = 68) | 60, 88.2% |
| Daily | 48, 70.6% |
| Weekly | 12, 17.6% |
| Monthly | 8, 11.8% |
| Traditional medicines | 37, 29.3% |
| In the last two days (n = 37) | 25, 19.8% |
| Daily | 13, 38.2% |
| Weekly | 12, 35.3% |
| Monthly | 9, 26.5% |
| Herbs or flowers collected outdoors | 25, 19.8% |
| In the last two days (n = 25) | 22, 17.5% |
| Daily | 16, 66.7% |
| Weekly | 5, 20.8% |
| Monthly | 3, 12.5% |
Products listed are those that were used by more than 10% of the participants. Other products assessed and used by less than 10% of participants were: beer (8 or 6.3% of participants), lemon (12 or 9.5% of participants), salt (11 or 8.7% of participants) and vinegar (4 or 3.2% of participants).
IVPs
IVPs results are detailed in Table 2. By design, all women enrolled in our study engaged in IVPs and started these practices at a young age (median = 15 years, range 8–41). Most had engaged in IVPs in the prior two days (83%). Water, soap, a cloth, rag or sponge were most commonly used. Herbs collected outdoors were used by over half of the women and traditional medicines by over one-quarter. Other products such as vinegar, salt, beer had been used by 10% or less of the participants.
Determinants for engaging in IVPs
Importance scores for the different attributes (hygiene, partner preference and health) are illustrated in Table 3.
Table 3.
Association between importance scores for the different attributes (hygiene, partner preference and health) and demographic, medical, risk factors and intravaginal practices (IVPs) (n = 126).
| Hygiene preference Mean (SD) |
p Value | Partner preference Mean (SD) |
p Value | Health preference Mean (SD) |
p Value | |
|---|---|---|---|---|---|---|
| Mean importance score | 61.9 (24.3) | 20.7 (15.1) | 17.3 (13.9) | |||
| Demographics | ||||||
| Age more than 38 years | 0.841 | 0.669 | 0.908 | |||
| Yes | 61.4 (26.1) | 20.2 (13.3) | 17.1 (14.6) | |||
| No | 62.3 (23.0) | 20.2 (13.3) | 17.4 (13.5) | |||
| Married or in a stable relationship | 0.795 | 0.611 | 0.919 | |||
| Yes | 62.3 (24.3) | 20.3 (14.4) | 17.1 (14.2) | |||
| No | 61.0 (24.3) | 21.8 (17.2) | 17.4 (13.5) | |||
| Monthly income | 0.005 | 0.004 | 0.083 | |||
| Less than USD150 | 58.8 (25.0) | 22.9 (15.7) | 18.5 (14.5) | |||
| More than USD150 | 72.3 (17.9) | 13.7 (10.6) | 13.4 (11.3) | |||
| Education less than primary | 0.169 | 0.273 | 0.231 | |||
| Yes | 58.7 (24.6) | 22.3 (14.5) | 15.9 (13.9) | |||
| No | 64.7 (23.8) | 19.4 (15.7) | 18.9 (13.8) | |||
| Sexual risk factors | ||||||
| HIV-infected male partner | 0.186 | 0.098 | 0.620 | |||
| Yes | 62.8 (24.1) | 25.8 (17.2) | 18.7 (16.9) | |||
| No | 55.4 (25.6) | 20.0 (14.5) | 17.1 (13.5) | |||
| Oral sex in the prior month | 0.339 | 0.722 | 0.175 | |||
| Yes | 57.9 (25.3) | 21.7 (12.4) | 20.3 (15.8) | |||
| No | 64.6 (25.1) | 20.2 (16.4) | 15.2 (13.0) | |||
| History of exchanging sex for money or gifts (lifetime) | 0.445 | 0.587 | 0.043 | |||
| Yes | 57.0 (24.8) | 18.5 (16.5) | 24.3 (19.0) | |||
| No | 62.5 (24.2) | 20.6 (15.0) | 16.5 (13.1) | |||
| Condom use in every vaginal encounter in the prior month | 0.996 | 0.384 | 0.318 | |||
| Yes | 63.4 (28.1) | 27.2 (31.1) | 9.4 (7.6) | |||
| No | 63.3 (25.2) | 20.1 (14.8) | 16.4 (13.7) | |||
| Medical history | ||||||
| Time since HIV diagnosis | 0.394 | 0.941 | 0.160 | |||
| More than one year | 67.9 (24.3) | 20.7 (18.8) | 11.6 (10.4) | |||
| Less than one year | 61.4 (24.2) | 20.4 (14.8) | 17.8 (14.1) | |||
| CD4 counts (cells/microlitre) | 0.249 | 0.126 | 0.774 | |||
| Greater than 500 | 63.5 (22.3) | 19.0 (13.2) | 17.5 (13.9) | |||
| Less than 500 | 58.0 (25.3) | 23.7 (17.1) | 18.3 (13.9) | |||
| Intravaginal practices | ||||||
| Age of IVP initiation older than 16 years | 0.301 | 0.125 | 0.895 | |||
| Yes | 59.3 (23.3) | 23.1 (17.2) | 17.5 (13.0) | |||
| No | 63.8 (24.9) | 18.9 (13.3) | 17.1 (14.6) | |||
| Use of IVPs in the prior two days | 0.004 | 0.013 | 0.025 | |||
| Yes | 0.64 (0.22) | 0.19 (0.13) | 0.16 (0.12) | |||
| No | 0.49 (0.26) | 0.27 (0.19) | 0.23 (0.16) | |||
| Products used for IVPs in the prior month | ||||||
| Water | 0.008 | 0.015 | 0.052 | |||
| Yes | 63.6 (23.9) | 19.7 (15.0) | 16.6 (13.9) | |||
| No | 42.6 (20.1) | 31.8 (12.8) | 25.5 (12.8) | |||
| Soap | 0.237 | 0.179 | 0.552 | |||
| Yes | 63.6 (24.7) | 19.5 (15.2) | 16.8 (14.1) | |||
| No | 58.1 (23.1) | 23.4 (14.9) | 18.4 (13.6) | |||
| Cloth or fabric | 0.023 | 0.067 | 0.050 | |||
| Yes | 66.5 (22.4) | 18.4 (13.7) | 19.9 (15.3) | |||
| No | 56.7 (25.3) | 23.3 (16.3) | 15.0 (12.3) | |||
| Traditional medicines | 0.343 | 0.198 | 0.803 | |||
| Yes | 58.7 (25.5) | 23.5 (17.2) | 17.8 (13.6) | |||
| No | 63.2 (23.7)) | 16.6 (14.2) | 17.1 (14.1) | |||
| Herbs or flowers collected outdoors | 0.792 | 0.909 | 0.738 | |||
| Yes | 36.1 (25.6) | 20.4 (17.4) | 16.5 (14.3)) | |||
| No | 61.6 (24.0) | 20.8 (14.6) | 17.5 (13.9) | |||
Products listed are those that were used by more than 10% of the participants. Other products assessed and used by less than 10% of participants were beer, lemon, salt, vinegar and yogurt. Importance scores were compared across groups of women using independent sample t-tests.
Values in bold show the p value is less than 0.05 and is significant.
Two participants provided only one rating for all profiles and were not included in the analyses. Results of the CA (n = 126) demonstrated that hygiene was the most important reason for engaging in IVPs (mean importance score = 62, SD = 24.3) followed by partner’s preference (mean importance score = 20, SD = 14.4) and health (mean importance score = 17, SD = 13.5).
Importance scores were compared among group of women with different demographic and medical characteristics, risk factors and IVPs. As seen in Table 3, women with higher income, those that had used IVPs in the prior two days and those using water or cloths for IVPs placed more importance on hygiene. Women with lower income, those not did engage in IVPs in the prior two days and those not using water for IVPs placed more importance on partner preference. Women who had not engaged in IVPs in the prior two days, had a history of exchanging sex for money or gifts and used cloth for IVPs placed more importance on health. Finally, women who had engaged in IVPs most recently were more likely to rate hygiene as most important and women who had engaged in IVPs less recently placed partner preference or health as most important (see Table 3).
Discussion
This study examined the determinants of engaging in IVPs among women with HIV infection in Lusaka, Zambia, using CA to identify the relative importance of factors influencing the decision to engage in IVPs. In support of our previous studies in Zambia, we found hygiene to be the most important factor in the decision to engage in IVPs.9 In addition, women were found to rank the importance of hygiene, partner preference and health differently, according to their personal characteristics. Women who had engaged in IVPs more recently were more likely to rate hygiene as most important, in contrast with those who engaged in IVPs less recently, who were more likely to rate partner preference or health as most important. Women with a higher income rated hygiene as more important, in contrast with those with a lower income, who rated partner preference as most important.
Study findings have important implications for the development of interventions to reduce the prevalence of IVPs, and suggest the need for tailored rather than standardised interventions. In Zambia, IVP use begins at an early age and is typically practiced frequently using a variety of products.8 Although in general, hygiene appears to be the most important driver of IVPs; among subgroups, different factors emerged as potential predictors of the decision to engage in IVPs. Results indicate that women who believed that IVPs were associated with enhanced health had a history of exchanging sex for money and using a cloth for IVPs. This finding suggests an urgent need for interventions targeting sex workers regarding the health impact and limited benefits of IVPs. Similarly, among those women more recently engaging in IVPs, water or a cloth was most commonly used, and these women were more likely to believe it was associated with improved hygiene. The relationship with IVPs and BV has also been shown in other settings, in particular in the US where women engage in IVP with commercially available vaginal douches.24 The manual introduction of water into the vagina may also represent a type of douching in a setting in which douching products may not be affordable or available. However, previous studies have also found frequent use of IVPs and the use of cloth to be associated with BV,5,25 and regardless of the way the product is inserted in the vagina, women appear to engage in such practices as part of their routine daily hygiene. Future studies should explore interventions to reduce the frequency of IVPs and propose healthy and acceptable hygiene alternatives for all women. Messages regarding the negative impact of IVPs are primarily given by health care providers and focus on IVP engaged in for hygiene purposes. To have the greatest impact, IVPs prevention campaigns should have a broader message that includes hygiene, partner preference and health rather than focusing solely on hygiene.
Generalisation of the results of this study is limited by its small sample size. Due to the high number of statistical tests required, comparing individual importance scores between groups using t-tests may not be the optimal strategy for analysis of conjoint data. More sophisticated strategies to uncover different ‘types’ of decision makers, such as cluster analysis or latent class analysis, are often utilised with data resulting from a CA; however, the sample size in this study was too small to utilise these strategies. In addition, it is not feasible to include all aspects of a decision in a conjoint questionnaire, and thus conjoint analyses may not represent the full complexity of a decision-making process. Because conjoint questionnaires use hypothetical real-life scenarios applying combinations of factors rather than factors presented in isolation, the conjoint analytic technique represents a powerful tool for the assessment of health decision making and has been applied to other health settings.18 However, the current study did not include ‘holdout scenarios’ (profiles that are rated by participants but are ‘held out’ of the analysis to assess validity) in the questionnaire or other methods to test the overall validity of the survey.15 Future research should continue the use of the conjoint method with a larger sample and include an assessment of validity. In addition, due to the high prevalence of IVPs, using a recency/frequency scale with additional options (e.g. today or yesterday for recency and more than once a day or once a day for frequency) could be useful in future studies.
Both IVPs and BV have been shown to increase the risk of acquisition and transmission of STIs and HIV.2–5 As a culturally sanctioned behaviour, although IVPs are difficult to influence, public health initiatives to decrease the use of IVPs and its consequences may be an important addition to the HIV prevention toolkit.
The use of CA to define characteristics of women more likely to engage in specific practices should be used to develop tailored IVP interventions to be incorporated into clinical practice and women’s health programmes. CA can be a powerful tool to incorporate into HIV research studies addressing decision-making processes.
Acknowledgments
The Miami Center for AIDS Research (CFAR) at the University of Miami Miller School of Medicine (P30AI073961) provided mentorship. We thank all those in our research team at the University Teaching Hospital in Lusaka, community sites providing referrals and the women participating in this study. The opinions reflected in this report are those of the authors and do not necessarily reflect those of the funding agencies and participating institutions.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This study was funded by grant from the National Institutes of Health, K23HD074489.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
References
- 1.Hilber AM, Chersich MF, van de Wijgert JH, et al. Vaginal practices, microbicides and HIV: what do we need to know? Sex Transm Infect. 2007;83:505–508. doi: 10.1136/sti.2007.028597. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Holmes KSF, Stamm W, Piot P, et al. Sexually transmitted diseases. 4. New York: The McGraw-Hill Companies Inc; 2008. [Google Scholar]
- 3.Atashili J, Poole C, Ndumbe PM, et al. Bacterial vaginosis and HIV acquisition: a meta-analysis of published studies. AIDS. 2008;22:1493–14501. doi: 10.1097/QAD.0b013e3283021a37. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Bloom BS, Lee DW. Bacterial vaginosis and preterm birth. New Eng J Med. 1996;334:1338–1339. [PubMed] [Google Scholar]
- 5.Clark RA, Theall KP, Amedee AM, et al. Frequent douching and clinical outcomes among HIV-infected women. Sex Transm Dis. 2007;34:985–990. doi: 10.1097/OLQ.0b013e31811ec7cb. [DOI] [PubMed] [Google Scholar]
- 6.Minister of Health Z. Zambia Country Report. 2012. [Google Scholar]
- 7.Hilber AM, Francis SC, Chersich M, et al. Intravaginal practices, vaginal infections and HIV acquisition: systematic review and meta-analysis. PloS One. 2010;5:e9119. doi: 10.1371/journal.pone.0009119. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Alcaide ML, Chisembele M, Mumbi M, et al. Examining targets for HIV prevention: intravaginal practices in Urban Lusaka, Zambia. AIDS Patient Care STDs. 2014;28:121–127. doi: 10.1089/apc.2013.0309. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Alcaide ML, Mumbi M, Chitalu N, et al. Vaginal cleansing practices in HIV infected Zambian women. AIDS Behav. 2013;17:872–878. doi: 10.1007/s10461-011-0083-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Mbikusita-Lewanika M, Stephen H, Thomas J. The prevalence of the use of ‘dry sex’ traditional medicines, among Zambian women, and the profile of the users. Psychol Health Med. 2009;14:227–238. doi: 10.1080/13548500802270364. [DOI] [PubMed] [Google Scholar]
- 11.McClelland RS, Richardson BA, Hassan WM, et al. Improvement of vaginal health for Kenyan women at risk for acquisition of human immunodeficiency virus type 1: results of a randomized trial. J Infect Dis. 2008;197:1361–1368. doi: 10.1086/587490. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Masese L, McClelland RS, Gitau R, et al. A pilot study of the feasibility of a vaginal washing cessation intervention among Kenyan female sex workers. Sex Transm Infect. 2013;89:217–222. doi: 10.1136/sextrans-2012-050564. [DOI] [PubMed] [Google Scholar]
- 13.Sawtooth Software. [accessed 23 April 2015];Interpreting the Results of Conjoint Analysis. 2010 http://www.sawtoothsoftware.com/download/techpap/interpca.pdf.
- 14.Curry J. Paper series. Sawtooth Technologies Inc; 1996. [accessed 24 April 2015]. Understanding Conjoint Analysis. http://www.sawtoothsoftware.com/download/techpap/undca15.pdf. [Google Scholar]
- 15.Orme B. Assessing the Validity of Conjoint Analysis. [accessed 23 April 2015];Sawtooth Software Research Paper Series. 1997 http://www.sawtooth-software.com/download/techpap/assess2.pdf.
- 16.Holt BY, Morwitz VG, Ngo L, et al. Microbicide preference among young women in California. J Women Health. 2006;15:281–294. doi: 10.1089/jwh.2006.15.281. [DOI] [PubMed] [Google Scholar]
- 17.Lee SJ, Brooks R, Bolan RK, et al. Assessing willingness to test for HIV among men who have sex with men using conjoint analysis, evidence for uptake of the FDA-approved at-home HIV test. AIDS Care. 2013;25:1592–1598. doi: 10.1080/09540121.2013.793272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Bridges JF, Hauber AB, Marshall D, et al. Conjoint analysis applications in health–a checklist: a report of the ISPOR Good Research Practices for Conjoint Analysis Task Force. Value Health. 2011;14:403–413. doi: 10.1016/j.jval.2010.11.013. [DOI] [PubMed] [Google Scholar]
- 19.Bridges JF, Selck FW, Gray GE, et al. Condom avoidance and determinants of demand for male circumcision in Johannesburg, South Africa. Health Policy Plan. 2011;26:298–306. doi: 10.1093/heapol/czq064. [DOI] [PubMed] [Google Scholar]
- 20.Kinsler JJ, Cunningham WE, Nurena CR, et al. Using conjoint analysis to measure the acceptability of rectal microbicides among men who have sex with men in four South American cities. AIDS Behav. 2012;16:1436–1447. doi: 10.1007/s10461-011-0045-5. [DOI] [PubMed] [Google Scholar]
- 21.Reese PP, Tehrani T, Lim MA, et al. Determinants of the decision to accept a kidney from a donor at increased risk for blood-borne viral infection. Clin J Am Soc Nephrol. 2010;5:917–923. doi: 10.2215/CJN.08251109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Weaver J, Newman PA, Williams CC, et al. “Sisters, Mothers, Daughters and Aunties”: HIV vaccine acceptability among African, Caribbean and other Black women in Toronto. Can J Public Health. 2013;104:e413–e417. doi: 10.17269/cjph.104.3915. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.IBMcorporation. SPSS Statistics for Windows, version 21. Armonk, NY: Released 2012. [Google Scholar]
- 24.Brown JM, Hess KL, Brown S, et al. Intravaginal practices and risk of bacterial vaginosis and candidiasis infection among a cohort of women in the United States. Obstetr Gynecol. 2013;121:773–780. doi: 10.1097/AOG.0b013e31828786f8. [DOI] [PubMed] [Google Scholar]
- 25.Baisley K, Changalucha J, Weiss HA, et al. Bacterial vaginosis in female facility workers in north-western Tanzania: prevalence and risk factors. Sex Transm Infect. 2009;85:370–375. doi: 10.1136/sti.2008.035543. [DOI] [PMC free article] [PubMed] [Google Scholar]