XML Template (2015) [30.3.2015–6:51pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150036/APPFile/SG-STDJ150036.3d
(STD)
[1–5] [PREPRINTER stage]
Int J STD AIDS OnlineFirst, published on April 2, 2015 as doi:10.1177/0956462415578953
Original research article
Sex workers, condoms, and mobility among men in Uzbekistan: implications for HIV transmission
International Journal of STD & AIDS 0(0) 1–5 ! The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0956462415578953 std.sagepub.com
Alex Smolak1, Nabila El-Bassel1, Anne Malin1, Assel Terlikbayeva2 and Salkynai Samatova1
Abstract Uzbekistan has one of the fastest growing HIV epidemics in the world. In this study, men who reported mobility were compared to non-mobile men in regard to paying for sex and condom use during paid sex. The sample included 2333 men between the ages of 15 and 59 in Uzbekistan. Generalised linear models were used to assess the relationship between mobility, paid sex, and condom use while adjusting for possible confounders. Of the total sample, 103 (4.4%) reported engaging in paid sex and 43 (42%) reported using condoms while engaging in paid sex. Mobile men were found to have more than three times the odds of paid sex than non-mobile men (OR: 3.209; 95% CI: 2.481, 4.150; p < 0.001). Mobile men were not significantly different from non-mobile men in terms of condom use; however, unmarried men were found to have six times the odds of not using a condom when compared to married men during paid sex (OR: 6.411; 95% CI: 2.502, 16.425; p ¼ 0.004). Only one of the men who reported paid sex also reported using a condom with their spouses at last intercourse. The findings contribute to understanding mobility and HIV risk, and have important implications for HIV prevention interventions.
Keywords HIV, migration, mobility, condom, sex worker, Uzbekistan, Central Asia Date received: 11 July 2014; accepted: 3 March 2015
Introduction Uzbekistan contains one of the fastest growing HIV epidemics globally. HIV prevalence in Uzbekistan was estimated at 0.2 in 2013.1 The spread of HIV has been especially acute in Uzbekistan. Only 51 cases had been diagnosed prior to 1998; there were a total of 800 registered cases in 2001 and more than 3500 in 2003,2 and an estimated 35,000 in 2013.1 The collapse of the Soviet Union and the related political, economic, and public health infrastructure changes form the context of Uzbekistan’s HIV epidemic.3 Sex work has been identified as a global driver of HIV. HIV infection among female sex workers (FSWs) is estimated at 4.4–28% in Uzbekistan and appears to be playing an increasing role in HIV transmission.1,4,5 The HIV epidemic in Central Asia has been primarily driven by injection drug use, and injection drug use is not uncommon among FSW.6,7 Few FSWs insist that their clients wear condoms and many men refuse to do so.3 Condom use among FSW Uzbekistan has been self-reported at 30.6%.8
According to the International Organization for Migration, labour migration has been increasing over recent years since Uzbekistan gained independence from the Soviet Union.9 Male labour migration was estimated at 504,582 persons for 2010.10 Internal migration has been reshaping the country, as rural populations migrate to urban areas in search of opportunities. A study in Uzbekistan’s capital Tashkent found that 80.5% of internal migrants were from rural areas and composed primarily of young men.11 These types of migration patterns could contribute to increased sexual risk behaviour. Mobility has been associated with HIV risk on a global scale. Short-term mobility has been linked to partnership concurrency.12 Mobility among FSW was found to 1 Global Health Research Center of Central Asia, Columbia University, New York, NY, USA 2 Global Health Research Center of Central Asia, Almaty, Kazakhstan
Corresponding author: Alex Smolak, Global Health Research Center of Central Asia, Columbia University, 1255 Amsterdam Avenue New York, New York 10027, USA. Email: [email protected]
Downloaded from std.sagepub.com at University of Otago Library on November 15, 2015
XML Template (2015) [30.3.2015–6:51pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150036/APPFile/SG-STDJ150036.3d
(STD)
[1–5] [PREPRINTER stage]
2
International Journal of STD & AIDS 0(0)
negatively impact FSW ability to negotiate condom use.13 In neighboring Kazakhstan, a study found that longer duration trips and greater frequency of trips from participant’s hometown were respectively associated with increased odds of multiple partners and paid sex.14 The purpose of this paper is to (1) describe the prevalence of paid sex among men in Uzbekistan; (2) describe condom use during paid sex; (3) examine the relationship between mobility and paid sex; and (4) examine mobility and condom use during paid sex. These aims were guided by an ecosystems perspective,15 as sexual risk behaviour was examined within the individual’s context. In this case, the primary context was mobility. This paper provides new knowledge on mobility, paid sex, and condom use in a country that is currently a hotspot for HIV transmission. Two hypotheses were tested: (1) men who are mobile will have higher odds of paid sex when compared to non-mobile men; (2) men who are mobile will have lower odds of condom use during paid sex when compared to non-mobile men.
This study is an analysis of data collected by the Demographic and Health Survey (DHS). The Uzbekistan DHS is a nationally representative survey of randomly selected respondents using a stratified multistage sampling strategy and was most recently collected in 2002. The sample consists of 2333 men between the ages of 15 and 59. The data have been de-identified and therefore this study is not considered human subjects research.
assess the association between mobility and paid sex and condom use among men who engaged in paid sex. Two GLM were used: (1) to test the association of mobility and paid sex and (2) to test the association of mobility and condom use among men who engaged in paid sex. Several potential confounders were adjusted for in each of the models. The potential confounders were selected based on the literature and balanced with limitations resulting from a smaller sample size in the second model as some variable categories became sparsely populated. The first model adjusted for age, education, marital status, ethnicity, and employment status. The second model adjusted for age, marital status, and employment status. Multiple imputation (MI) with 10 imputations corrected for 34% missing data for the condom use during paid sex variable. Sinharay et al.16 found that MI was appropriate for up to 40% missingness. All variables included in the first model were used in the imputation. MI may also correct for non-response bias when questions pertain to a sensitive topic. The odds ratios were nearly the same with both MI data and raw data, but all p values were significant with raw data. We think this analysis was misleading as there is likely a degree of similarity between the participates who did not answer, ‘did you use a condom last time you paid for sex?’ As such, these missing data were not ignorable; however, MI does not assume that nonresponse is ignorable and would produce valid answers for the missing data under the presented condition.17 Therefore, we reported findings with MI as these findings are more likely to be representative.
Measures
Results
Paid sex was measured by asking ‘‘have you ever paid for sex?’’ This question was followed by asking ‘‘how long ago was the last time you paid for sex?’’ Only participants who paid for sex in the last 12 months were included in the paid for sex group as this inclusion criteria allowed for the analysis to be within the same timeframe as the mobility variable. Men who paid for sex more than 12 months ago did not fit the inclusion criteria for the paid for sex group. Mobility was measured by asking, ‘in the last 12 months, have you ever traveled away from your home community and slept away?’ Condom use among men who engaged in paid sex was measured by asking ‘did you use a condom last time you paid for sex?’ The category married included married and cohabitating individuals. Demographic variables were measured categorically and dichotomously and are displayed as such in the tables.
Socio-demographic characteristics of the total sample, 79% were Uzbek, 93% were Muslim and were fairly well distributed over age and educational categories. Nearly half were married and employed. Fourteen percent were mobile. Paid sex was reported by 103 (4.4%) men, and 58% of those men reported not having used a condom during paid sex. Further details are found in Table 1. The first GLM tested the association of mobility and paid sex while adjusting for possible confounders. Mobile men were found to have more than three times the odds of paid sex than non-mobile men (OR: 3.209; 95% CI: 2.481, 4.150; p < 0.001). When compared to married men, unmarried men were found to have nearly six times the odds of engaging in paid sex (OR ¼ 5.847; CI: 4.464, 7.659; p < 0.001). Other significant predictors associated with paying for sex that were adjusted for in this model included age and education. The results of this first GLM are displayed in Table 2. The second GLM tested the association of mobility and condom use among men who reported paid sex
Methods
Data analysis Descriptive statistics included frequencies. Generalised linear models (GLM) with logit links were used to
Downloaded from std.sagepub.com at University of Otago Library on November 15, 2015
XML Template (2015) [30.3.2015–6:51pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150036/APPFile/SG-STDJ150036.3d
(STD)
[1–5] [PREPRINTER stage]
Smolak et al.
3
Table 1. Sample characteristics (Total n ¼ 2333, engaged in paid sex n ¼ 103).
Table 2. Summary of generalised linear models (GLMs) results for mobility and paid sex (n ¼ 2333).
Variable
Predictor
Total n (%)
Engaged in paid sex n
Gender Male 2333 (100) Female 0 (0) Age 15–18 360 (15.4) 19–24 431 (18.5) 25–30 383 (16.4) 31–40 590 (25.3) 41þ 569 (24.4) Ethnicity Uzbek 1850 (79.3) Russian 135 (5.8) Other 348 (14.9) Education Secondary or less 1113(47.7) PTU/SPTU 310 (13.3) Technikum 466 (20.0) University 444 (19.0) Married Yes 1520 (65.2) No 813 (34.8) Religion Muslim 2172 (93.1) Non-Muslim 161 (6.9) Working Yes 1398 (59.9) No 934 (40.1) Ever tested for HIV Yes 266 (11.4) No 2067 (88.6) Mobility Yes 348 (14.9) No 1985 (85.1) Used condom with spouse (n ¼ 1520) Yes 17 (1.2) No 1503 (98.8) Paid for sex Yes 103 (4.4) No 2230 (95.6) Used condom during paid sex (n ¼ 103) Yes n/a No n/a
103 (100) 0 (0) 7 33 18 28 17
(6.8) (32.0) (17.5) (27.2) (16.5)
79 (76.7) 7 (6.8) 17 (16.5) 40 7 23 33
(38.8) (6.8) (22.3) (32)
56 (54.4) 47 (45.6) 92 (89.3) 11 (10.7) 64 (62.1) 39 (37.9) 11 (10.7) 92 (89.3) 36 (35) 67 (65) 1 (1.8) 56 (98.2) 103 (100) 0 (0) 43 (41.7) 60 (58.3)
while adjusting for possible confounders. This model found that mobility was not significantly associated with condom use during paid sex. Age and employment status were also not significant. However, marital status was highly significant. Unmarried men were found to
Mobility Yes No Age 15–18 19–24 25–30 31–40 41þ Education University Technical Vocational Secondary or less Ethnicity Uzbek Other Russian Religion Muslim Non-Muslim Tested for HIV No Yes Married No Yes Employed Yes No
OR
95% CIs for OR
p
3.209 –
2.481 – 4.150 –
(STD)
[1–5] [PREPRINTER stage]
Int J STD AIDS OnlineFirst, published on April 2, 2015 as doi:10.1177/0956462415578953
Original research article
Sex workers, condoms, and mobility among men in Uzbekistan: implications for HIV transmission
International Journal of STD & AIDS 0(0) 1–5 ! The Author(s) 2015 Reprints and permissions: sagepub.co.uk/journalsPermissions.nav DOI: 10.1177/0956462415578953 std.sagepub.com
Alex Smolak1, Nabila El-Bassel1, Anne Malin1, Assel Terlikbayeva2 and Salkynai Samatova1
Abstract Uzbekistan has one of the fastest growing HIV epidemics in the world. In this study, men who reported mobility were compared to non-mobile men in regard to paying for sex and condom use during paid sex. The sample included 2333 men between the ages of 15 and 59 in Uzbekistan. Generalised linear models were used to assess the relationship between mobility, paid sex, and condom use while adjusting for possible confounders. Of the total sample, 103 (4.4%) reported engaging in paid sex and 43 (42%) reported using condoms while engaging in paid sex. Mobile men were found to have more than three times the odds of paid sex than non-mobile men (OR: 3.209; 95% CI: 2.481, 4.150; p < 0.001). Mobile men were not significantly different from non-mobile men in terms of condom use; however, unmarried men were found to have six times the odds of not using a condom when compared to married men during paid sex (OR: 6.411; 95% CI: 2.502, 16.425; p ¼ 0.004). Only one of the men who reported paid sex also reported using a condom with their spouses at last intercourse. The findings contribute to understanding mobility and HIV risk, and have important implications for HIV prevention interventions.
Keywords HIV, migration, mobility, condom, sex worker, Uzbekistan, Central Asia Date received: 11 July 2014; accepted: 3 March 2015
Introduction Uzbekistan contains one of the fastest growing HIV epidemics globally. HIV prevalence in Uzbekistan was estimated at 0.2 in 2013.1 The spread of HIV has been especially acute in Uzbekistan. Only 51 cases had been diagnosed prior to 1998; there were a total of 800 registered cases in 2001 and more than 3500 in 2003,2 and an estimated 35,000 in 2013.1 The collapse of the Soviet Union and the related political, economic, and public health infrastructure changes form the context of Uzbekistan’s HIV epidemic.3 Sex work has been identified as a global driver of HIV. HIV infection among female sex workers (FSWs) is estimated at 4.4–28% in Uzbekistan and appears to be playing an increasing role in HIV transmission.1,4,5 The HIV epidemic in Central Asia has been primarily driven by injection drug use, and injection drug use is not uncommon among FSW.6,7 Few FSWs insist that their clients wear condoms and many men refuse to do so.3 Condom use among FSW Uzbekistan has been self-reported at 30.6%.8
According to the International Organization for Migration, labour migration has been increasing over recent years since Uzbekistan gained independence from the Soviet Union.9 Male labour migration was estimated at 504,582 persons for 2010.10 Internal migration has been reshaping the country, as rural populations migrate to urban areas in search of opportunities. A study in Uzbekistan’s capital Tashkent found that 80.5% of internal migrants were from rural areas and composed primarily of young men.11 These types of migration patterns could contribute to increased sexual risk behaviour. Mobility has been associated with HIV risk on a global scale. Short-term mobility has been linked to partnership concurrency.12 Mobility among FSW was found to 1 Global Health Research Center of Central Asia, Columbia University, New York, NY, USA 2 Global Health Research Center of Central Asia, Almaty, Kazakhstan
Corresponding author: Alex Smolak, Global Health Research Center of Central Asia, Columbia University, 1255 Amsterdam Avenue New York, New York 10027, USA. Email: [email protected]
Downloaded from std.sagepub.com at University of Otago Library on November 15, 2015
XML Template (2015) [30.3.2015–6:51pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150036/APPFile/SG-STDJ150036.3d
(STD)
[1–5] [PREPRINTER stage]
2
International Journal of STD & AIDS 0(0)
negatively impact FSW ability to negotiate condom use.13 In neighboring Kazakhstan, a study found that longer duration trips and greater frequency of trips from participant’s hometown were respectively associated with increased odds of multiple partners and paid sex.14 The purpose of this paper is to (1) describe the prevalence of paid sex among men in Uzbekistan; (2) describe condom use during paid sex; (3) examine the relationship between mobility and paid sex; and (4) examine mobility and condom use during paid sex. These aims were guided by an ecosystems perspective,15 as sexual risk behaviour was examined within the individual’s context. In this case, the primary context was mobility. This paper provides new knowledge on mobility, paid sex, and condom use in a country that is currently a hotspot for HIV transmission. Two hypotheses were tested: (1) men who are mobile will have higher odds of paid sex when compared to non-mobile men; (2) men who are mobile will have lower odds of condom use during paid sex when compared to non-mobile men.
This study is an analysis of data collected by the Demographic and Health Survey (DHS). The Uzbekistan DHS is a nationally representative survey of randomly selected respondents using a stratified multistage sampling strategy and was most recently collected in 2002. The sample consists of 2333 men between the ages of 15 and 59. The data have been de-identified and therefore this study is not considered human subjects research.
assess the association between mobility and paid sex and condom use among men who engaged in paid sex. Two GLM were used: (1) to test the association of mobility and paid sex and (2) to test the association of mobility and condom use among men who engaged in paid sex. Several potential confounders were adjusted for in each of the models. The potential confounders were selected based on the literature and balanced with limitations resulting from a smaller sample size in the second model as some variable categories became sparsely populated. The first model adjusted for age, education, marital status, ethnicity, and employment status. The second model adjusted for age, marital status, and employment status. Multiple imputation (MI) with 10 imputations corrected for 34% missing data for the condom use during paid sex variable. Sinharay et al.16 found that MI was appropriate for up to 40% missingness. All variables included in the first model were used in the imputation. MI may also correct for non-response bias when questions pertain to a sensitive topic. The odds ratios were nearly the same with both MI data and raw data, but all p values were significant with raw data. We think this analysis was misleading as there is likely a degree of similarity between the participates who did not answer, ‘did you use a condom last time you paid for sex?’ As such, these missing data were not ignorable; however, MI does not assume that nonresponse is ignorable and would produce valid answers for the missing data under the presented condition.17 Therefore, we reported findings with MI as these findings are more likely to be representative.
Measures
Results
Paid sex was measured by asking ‘‘have you ever paid for sex?’’ This question was followed by asking ‘‘how long ago was the last time you paid for sex?’’ Only participants who paid for sex in the last 12 months were included in the paid for sex group as this inclusion criteria allowed for the analysis to be within the same timeframe as the mobility variable. Men who paid for sex more than 12 months ago did not fit the inclusion criteria for the paid for sex group. Mobility was measured by asking, ‘in the last 12 months, have you ever traveled away from your home community and slept away?’ Condom use among men who engaged in paid sex was measured by asking ‘did you use a condom last time you paid for sex?’ The category married included married and cohabitating individuals. Demographic variables were measured categorically and dichotomously and are displayed as such in the tables.
Socio-demographic characteristics of the total sample, 79% were Uzbek, 93% were Muslim and were fairly well distributed over age and educational categories. Nearly half were married and employed. Fourteen percent were mobile. Paid sex was reported by 103 (4.4%) men, and 58% of those men reported not having used a condom during paid sex. Further details are found in Table 1. The first GLM tested the association of mobility and paid sex while adjusting for possible confounders. Mobile men were found to have more than three times the odds of paid sex than non-mobile men (OR: 3.209; 95% CI: 2.481, 4.150; p < 0.001). When compared to married men, unmarried men were found to have nearly six times the odds of engaging in paid sex (OR ¼ 5.847; CI: 4.464, 7.659; p < 0.001). Other significant predictors associated with paying for sex that were adjusted for in this model included age and education. The results of this first GLM are displayed in Table 2. The second GLM tested the association of mobility and condom use among men who reported paid sex
Methods
Data analysis Descriptive statistics included frequencies. Generalised linear models (GLM) with logit links were used to
Downloaded from std.sagepub.com at University of Otago Library on November 15, 2015
XML Template (2015) [30.3.2015–6:51pm] //blrnas3.glyph.com/cenpro/ApplicationFiles/Journals/SAGE/3B2/STDJ/Vol00000/150036/APPFile/SG-STDJ150036.3d
(STD)
[1–5] [PREPRINTER stage]
Smolak et al.
3
Table 1. Sample characteristics (Total n ¼ 2333, engaged in paid sex n ¼ 103).
Table 2. Summary of generalised linear models (GLMs) results for mobility and paid sex (n ¼ 2333).
Variable
Predictor
Total n (%)
Engaged in paid sex n
Gender Male 2333 (100) Female 0 (0) Age 15–18 360 (15.4) 19–24 431 (18.5) 25–30 383 (16.4) 31–40 590 (25.3) 41þ 569 (24.4) Ethnicity Uzbek 1850 (79.3) Russian 135 (5.8) Other 348 (14.9) Education Secondary or less 1113(47.7) PTU/SPTU 310 (13.3) Technikum 466 (20.0) University 444 (19.0) Married Yes 1520 (65.2) No 813 (34.8) Religion Muslim 2172 (93.1) Non-Muslim 161 (6.9) Working Yes 1398 (59.9) No 934 (40.1) Ever tested for HIV Yes 266 (11.4) No 2067 (88.6) Mobility Yes 348 (14.9) No 1985 (85.1) Used condom with spouse (n ¼ 1520) Yes 17 (1.2) No 1503 (98.8) Paid for sex Yes 103 (4.4) No 2230 (95.6) Used condom during paid sex (n ¼ 103) Yes n/a No n/a
103 (100) 0 (0) 7 33 18 28 17
(6.8) (32.0) (17.5) (27.2) (16.5)
79 (76.7) 7 (6.8) 17 (16.5) 40 7 23 33
(38.8) (6.8) (22.3) (32)
56 (54.4) 47 (45.6) 92 (89.3) 11 (10.7) 64 (62.1) 39 (37.9) 11 (10.7) 92 (89.3) 36 (35) 67 (65) 1 (1.8) 56 (98.2) 103 (100) 0 (0) 43 (41.7) 60 (58.3)
while adjusting for possible confounders. This model found that mobility was not significantly associated with condom use during paid sex. Age and employment status were also not significant. However, marital status was highly significant. Unmarried men were found to
Mobility Yes No Age 15–18 19–24 25–30 31–40 41þ Education University Technical Vocational Secondary or less Ethnicity Uzbek Other Russian Religion Muslim Non-Muslim Tested for HIV No Yes Married No Yes Employed Yes No
OR
95% CIs for OR
p
3.209 –
2.481 – 4.150 –
