Determinants of Safer-Sex Behaviors of Brothel Female Commercial Sex Workers in Jakarta, Indonesia.

The first case of AIDS in Indonesia was found in 1987, and by December 1998 the official number of reported HIV-1-positive cases was 819, of which 227 were full-blown AIDS cases (Ministry of Health, 1998). There is no doubt that these numbers represent an undercounting; however, the available surveillance system in Indonesia did not allow us to approximate the HIV-1 seroprevalence in this country. Since heterosexual transmission of HIV-1 is the predominant mode of transmission in Indonesia, female commercial sex workers constitute one of the communities at high risk to become infected with and to transmit the HIV-1 virus.

Yearly surveys on the prevalence of gonorrhea and syphilis among brothel and nonbrothel female sex workers in Jakarta repeatedly showed high prevalence of both diseases (18-25% and 5-7%, respectively) (Gunawan, 1997; Van der Sterren, Murray, & Hull, 1995). Furthermore, the 1992 and 1994 HIV-1 seroprevalence surveys in female sex worker communities in Jakarta also indicated an increase in HIV-1 prevalence from 0.3% to 0.6% (Dinas Kesehatan DKI Jakarta, 1994). These are indications that sex workers in Jakarta practice unsafe sex behaviors.

A study in East Java province (population 33 million) showed that 7% of men aged 15 to 60 years had ever had sex with a sex worker (Linnan, Kestari, & Kambodji, 1995). Nearly all major cities in Indonesia have one or more brothel complexes (lokalisasi) and in Jakarta there are 8 illegal brothel complexes that have been in business for years, with approximately 10,000 female sex workers. These established brothel complexes are evidence of a large client base among the general population. Focusing AIDS prevention efforts on the female sex worker communities and the clients is, therefore, justifiable. The current culturo-political situation in Indonesia, however, still prevents a public safer-sex campaign (KOMPAS, November 7, 1995); therefore, considerable efforts targeted at brothel communities (i.e., the female sex workers themselves, the managers, and the clients) are more feasible and could have a substantial effect on the predicted AIDS epidemic in this country.

Studies have shown that inconsistent condom use is ineffective in reducing the risk of STDs or AIDS infection (Ford & Wirawan, 1996; Sawanpanyalert, Ungchusak, Thanprasertsuk, & Akarasewi, 1994; Taha et al., 1996; Zenilman et al., 1995). Therefore, although it has also been calculated that in a place where HIV-1 prevalence is still low, any increase in condom use will somewhat reduce the risk of infection (Fineberg, 1988), consistent condom use is the ultimate behavioral change. Accordingly, this study focused on consistent use during sexual intercourse.

The government policy toward female sex workers is mainly to rehabilitate and resocialize them (Dinas Sosial DKI Jakarta, 1993; Jones, Sulistyaningsih, & Hull, 1995). Throughout the country, there are 22 rehabilitation centers for sex workers run by either the national or the provincial government. The Jakarta Social Welfare Office (referred to as "the Office") runs one such center in Kramat Tunggak, Northern Jakarta. Here, female sex workers and brothel managers are still allowed to carry out their business, albeit under some regulations and restrictions.

Since the Office exercises substantial control over the Kramat Tunggak brothel complex, its policies and programs for the sex workers, managers, and clients, or the lack thereof, may be influential determinants of the sex workers' safer-sex practice. The office provided a monthly health service to the sex workers, which the women were required to attend. Those who refused to come, however, were not penalized. Although the Office had no well-planned STD/AIDS-related educational programs in addition to this health program, condoms were occasionally mentioned in their regular talks, which were attended each time by about 100 sex workers and a few brothel managers.

Other powerful decision makers in a brothel complex are the brothel managers and the clients (Swaddiwudhipong, Chaovakiratiping, Sift, & Lerdlukanavonge, 1990). Client-related factors, such as types of clients, clients' attitudes toward condom use, number of clients, and price per sexual encounter may be important determinants for the sex workers' consistent condom use (Mhalu et al., 1991; Picketing, Quigley, Hayes, Todd, & Wilkins, 1993).

In addition, certain sociodemographic factors, such as age, educational attainment, hometown, and years of working as a sex worker, may also influence safer-sex behaviors (Pickering et al., 1993; Wilson, Sibanda, Mboyi, Msimanga, & Dube, 1990).

METHODS

Participants

Our study population was the female sex worker community in Kramat Tunggak, Northern Jakarta. Since Kramat Tunggak is a unique place, a brief description is necessary.

It has a total area of 11.5 hectares (approximately 28.4 acres) and is semi-isolated from the surrounding residential neighborhoods by a two-meter-high brick wall. Officially, only the sex workers are allowed to live there; the brothel managers are only permitted to come on a daily basis, and no children are allowed to enter. In reality, however, many of the brothel managers live there with their families and bodyguards. The complex is guarded by official civil guards who are posted at the two main entrance gates, which abut two spacious parking lots available for Kramat Tunggak guests only. On average, each brothel manager employs 5 to 10 sex workers; usually these women have to pay the managers for their room, water, and electricity. The women also have to share their earnings with their managers, as the clients pay the women directly. The percentage varies according to each brothel, but 25% for the managers is the average proportion (Sedyaningsih-Mamahit, 1997). The Office is located across the street from the brothel complex. The rehabilitation and resocialization programs include some vocational training classes, such as literacy, sewing, and cooking, which are held in an adjacent building.

Measures

This study used the AIDS Risk Reduction Model (ARRM) (Catania, Kegeles, & Coates, 1990) with the inclusion of elements of the Health Belief Model (HBM) (Janz & Becker, 1984) and Bandura's concept of self-efficacy theory (Bandura, 1989). The questionnaire was developed through several stages with the help of experts and the use of several other researchers' questionnaires as models and/or comparisons (Basuki, 1991; Rahardjo, 1992; Wingood & Case, personal communication, 1994). It is based on theories and models used in developing questions, both general and behavioral (Ajzen & Fishbein, 1980; Bandura, 1977, 1989; Catania et al., 1990; Fowler & Mangione, 1990; Janz & Becker, 1984), and on preliminary qualitative research conducted at the study site in June and July, 1993. The questionnaire assessed the following: (a) sociodemographics, (b) occupational-related information, (c) STD/AIDS-related behavioral information, and (d) other health risk behaviors.

To improve accuracy, questions about condom use referred to the previous two weeks only. Condom use was initially measured as a percentage (i.e., the number of clients who used condoms divided by the number of all clients). This was based on the assumption--supported by the preliminary study--that most sex workers had only one intercourse per client. The term client encompassed all males who received sex services from the sex workers, including occasional clients (tamu), regular clients (kenalan), and the women's lovers (gendak).

Procedure

Respondents for the survey were randomly chosen (using a random number list) from the sex worker name list recorded by the Office in April, 1995. At that time, 1,600 women and 228 brothel managers were officially registered in Kramat Tunggak. The only exclusion criterion for our survey was if the sex workers had been trained by Yayasan Kusuma Buana (YKB), a nongovemment organization that had given about 80 sex workers a 3-day intensive STD/AIDS training course in 1994 (Sasongko, personal communication, 1995). Ten women, ages 20 to 30, conducted the face-to-face interviews. Using an available area map, our team visited the first 500 selected sex workers in their brothels during the daytime. Those who refused to participate and those who were not found after two visits were dropped, and other names were selected randomly to replace them using the random number list. Most of the sex workers who agreed to participate fully in the study signed or fingerprinted a written consent; only a few agreed verbally.

After one and a half months (May-June 1995), data were obtained from 459 survey respondents. The initial response rate was 63%: About 5% of the nonrespondents refused to participate, while the rest (32%) were either not present or were not recognized in that brothel (the original Office name list was handwritten and sometimes hard to read).

To estimate the reproducibility of data on consistent condom use, a 2 week test-retest reliability assessment was conducted on a random subset of the sample (N = 46), while their validity was estimated by comparing them with condom use data obtained in 2-week diaries, filled out by a small number (N = 40) of participants randomly chosen from the survey participants. For two weeks, these women were asked to afix a green sticker with a man's picture on it for every client she had sex with; a red sticker with the picture of a heart on it for every sex act with a lover; and a yellow sticker with a picture of a condom on it beside the stickers of the clients and lovers when they used a condom in intercourse. They were also asked to collect their used condom wrappers to be matched with their condom use records in the diaries.

Test-retest reliability analysis showed that the sex workers' self-reported condom use showed a moderate reproducibility, with Spearman correlation estimated as 0.38 (p [is less than] .04). The relative validity--comparing self-reported data with diary data--was also moderate. The Spearman correlations were estimated as 0.61 (p [is less than] .004) and 0.52 (p [is less than] .02).

As we realized that the sex workers might be telling us what they believed we wanted to hear, we also conducted an extensive qualitative study. This was carried out between April and November of 1995, mainly by the primary investigator. Respondents were chosen using predetermined criteria, (such as age, sex, and size of brothel) from among the sex workers, the brothel managers and bodyguards, the clients (convenience samples), the vocational trainers, and government officers (see Sedyaningsih-Mamahit, 1997 for findings from this qualitative study).

Statistical analyses

Data were initially recorded in Epi-Info (Center for Disease Control, 1990). Univariate, bivariate and multivariate analyses were carried out using STATA (STATA, 1993). Our main outcome variable was consistent condom use. Condom use was classified as 0 (never), 1 (seldom), 2 (often), or 3 (always) and for the final analysis, into 1 for always and 0 for others. The association of consistent condom use with other variables was estimated by odds ratios in logistic regressions. The sample size varied because of missing data: Condom use was reported only by those indicating vaginal sex during the previous two weeks.

Condom use reproducibility and validity was assessed by Spearman correlation coefficients between self-reported and retest data, and between self-reported and diary data.

RESULTS

Descriptive Analysis

Sociodemographic characteristics. In general, the characteristics of our study samples, the sex workers who refused to become respondents, and the entire Kramat Tunggak sex worker population (data from the May, 1993 census) are quite similar. Table 1 indicates that we surveyed more educated sex workers as respondents (4.7% or n = 18 attended senior high school), compared to the nonrespondents (1.5% or n = 4), a very likely scenario, as it is likely that they have more self-confidence.

Table 1. Comparisons Among Nonrespondents, Respondents, and the Sex Worker Population in Kramat Tunggak (KT), by Sociodemographic Characteristics.

                                                  Sample
                                      Respondents     Nonrespondents
Characteristics                       n = 459 (%)      n = 270(a)(%)

Age (in years)(c)
  [is less than or equal to] 20          99 (22%)           51 (19%)
  21 - 25                                208 (45)           136 (50)
  > 25                                   152 (33)            83 (31)
Education Attained(c,d)
  No school                                32 (7)              4 (1)
  Elementary
   ([is less than or equal to] 6
   years)                                362 (79)           251 (94)
  Jr high school (> 6-9 y)                 44 (9)             11 (4)
  Sr high school (> 9-12 y)                21 (5)              4 (1)
Marital Status(c)
  Never married                           56 (12)                 NA
  Divorced/widowed                       379 (83)                 NA
  Married                                  24 (5)                 NA
Years working in KT(c,d)
  0 - 1 years                            274 (60)           207 (77)
  1 - 2 years                            108 (23)            53 (19)
  > 2 years                               77 (17)             10 (4)
Hometown
  West Java & Jakarta                    241 (53)           156 (58)
  Central Java &
   Yogyakarta                            143 (31)            80 (30)
  East Java & elsewhere                   75 (16)            34 (12)

                                              Whole
                                           Population
Characteristics                         n = 1977(b) (%)
Age (in years)(c)
  [is less than or equal to] 20              595 (30%)
  21 - 25                                     917 (46)
  > 25                                        465 (24)
Education Attained(c,d)
  No school                                    174 (9)
  Elementary
   ([is less than or equal to] 6
   years)                                    1688 (85)
  Jr high school (> 6-9 y)                      99 (5)
  Sr high school (> 9-12 y)                     16 (1)
Marital Status(c)
  Never married                               694 (35)
  Divorced/widowed                           1283 (65)
  Married                                        0 (0)
Years working in KT(c,d)
  0 - 1 years                                 983 (50)
  1 - 2 years                                 615 (31)
  > 2 years                                   379 (19)
Hometown
  West Java & Jakarta                        1097 (56)
  Central Java &
   Yogyakarta                                 597 (30)
  East Java & elsewhere                       283 (14)

(a) Kramat Tunggak Name List, April 1995.

(b) Kramat Tunggak Census, May 1993.

(c) Difference between respondents and the whole population significant at p < .05 (Chi-square test),

(d) Difference between respondents and non-respondents significant at p < .05 (Chi-square test).

The government policy regarding marital status may explain the difference between our respondents and the general Kramat Tunggak population. This policy only allows divorcees, widows, and nonvirgin unmarried women ("holed girls" or gadis bolong is the official term for these women) to work as sex workers in Kramat Tunggak: Virgins and married women are prohibited. It was no surprise, therefore, to see that no married women were officially registered; nevertheless, interviews revealed that married women did work in Kramat Tunggak (5.2% or n = 24).

Data on the number of years worked in Kramat Tunggak show that we missed more new sex workers; again, a very likely scenario since they were probably afraid of us. More than 80% (n = 382) of the sex workers had worked for 2 years or less in Kramat Tunggak. This finding was not only in accordance with the Kramat Tunggak census, but also with data on other brothel complexes in Jakarta (Basuki, 1991).

Occupational-related characteristics. The respondents' occupational-related characteristics are displayed in Table 2. Most of the sex workers in Kramat Tunggak came from poor families, so it was not a surprise to learn that about 63% (n = 291) were motivated by reasons that included economics. Motivation was determined using an open-ended question, in which the women could relate their personal stories. In Indonesia, where being a sex worker is highly stigmatized, it is interesting to see that 13% (n = 60) of the women frankly stated that they chose, and were not forced into, sex work. Nearly 15% (n = 67) also mentioned that they enjoyed working as sex workers. The remaining sex workers' motivations to enter the job included disharmony with their significant men and other stressful conditions (24% or n = 108). Previous experience of being raped was not a commonly reported "push" factor: Only 5 % (n = 25) of the sex workers reported being raped before entering prostitution.

Table 2. Measures of Occupational-Related Characteristics of The Sex Worker Respondents in Kramat Tunggak

Variable                                                n = 459 (%)
Occupational-Related Characteristics
  Age when started working as sex worker
    Range                                               11 - 32.0 y
    Mean                                                     22.0 y
    SD                                                        3.6 y
  Reason works as sex worker
    Economic reason only                                   188 (41)
    Combination of economic & other reasons                103 (22)
    Dispute with husband/boyfriend, being sold/
      cheated, and other stressful conditions              108 (24)
    Tempted by others and self-motivated                    60 (13)
  Monthly income as sex worker(a)
    [is less than or equal to] $ 45.00(b)                     8 (2)
    > $ 45.00 - $136.00                                    130 (28)
    > $136.00 - $ 227.00                                   130 (28)
    > $ 227.00 - $ 454.00                                  145 (32)
    > $ 454.00                                              46 (10)
Types of Sexual Services (previous 2 years)
  Vaginal sex
    Never                                                   27  (6)
    Seldom                                                   7  (1)
    Sometimes                                                8  (2)
    Often                                                   30  (7)
    Always                                                 387 (84)
  Anal sex
    Never                                                 459 (100)
  Oral sex
    Never                                                  452 (96)
    Seldom                                                    2 (1)
    Sometimes                                                 2 (1)
    Often                                                     2 (1)
    Always                                                    2 (1)
  Manual sex (i.e., hand masturbation)
    Never                                                  388 (85)
    Seldom                                                   43 (9)
    Sometimes                                                15 (3)
    Often                                                    11 (2)
    Always                                                    2 (1)
Number of clients(c) (within previous 2 weeks)
    None                                                     27 (6)
    <7                                                     264 (58)
    7-14                                                    31 (26)
    15-21                                                    39 (8)
    >21                                                       9 (2)
  Condom use with clients(d)
   (within previous 2 weeks)
    Never                                                  108 (25)
    Seldom                                                  96 (22)
    Often                                                   74 (17)
    Always                                                 154 (36)
Related to Safer-Sex Practice(d)
  Self-efficacy in using condoms
    Low                                                    202 (46)
    High                                                   230 (53)
  Experience in negotiating condom use
    Low                                                    281 (65)
    High                                                   151 (35)
  Experience in using condoms for family planning
    Yes                                                      11 (3)
    No                                                     421 (97)
  Perception of clients and managers
   attitudinal barriers
    Low                                                    172 (40)
    High                                                   260 (60)

(a) $1.00 = Rp. 2200.00 (1995). (b) The average wage of a woman factory worker. (c) Includes occasional and regular customers, and lovers. (d) N = 4322 (excludes the sex workers who had no clients).

More than half the women had tried different jobs before entering the sex industry; working in a factory was the most popular job. From the qualitative study we found that not only did these jobs yield much less money (the average wage was $45.00 per month), but the working hours were also much longer than commercial sex work.

Although only 59% (n = 269) of the women adopted at least one measure to prevent pregnancy, 12.4% (n = 57) had induced abortion. The seemingly low number of unwanted pregnancies may be due to the sex workers habit of drinking traditional herbs, or to pelvic inflammatory disease (PID) as a complication of repeated STDs.

Features of the sex workers sexual behaviors showed their practices over the previous two years and some over the previous two weeks. In line with others unpublished findings (Basuki, 1991; Rahardjo, 1992), we too found that vaginal sex was the most preferred type of sex, and no one reported having anal sex. The 5.9% (n = 27) who said they had not had vaginal sex over the previous two weeks were sex workers who for various reasons did not receive any clients during that time.

More than half of the women had had fewer than 7 clients in the previous two weeks (this finding was similar to that from another study of a different Jakarta brothel complex by Basuki, 1991). This was due to the fact that many clients come just to drink beer and to dance, and many of the women only gave sexual service to regular customers or lovers, who would visit and pay on a regular basis.

One percent (n = 5) of the women said that they had never seen a condom before; all of them were new to the job. An estimated 36% (n = 154) of the sex workers said that they required their clients to use a condom all the time, and 25% (n = 108) said they never required their partners to use condoms when engaging in vaginal intercourse.

Predictors of Consistent Condom Use

Before adjusting for other factors, several variables were significantly associated with consistent condom use. In the multivariate analysis, however, some of those associations became statistically insignificant. In multivariate logistic regression analysis, all statisttically significant factors (p [is less than] .05) from the bivariate analyses--as well as other factors that we thought were conceptually important in predicting consistent condom use--were initially included in the model. Hence, we were testing the direct association between factors from the three different stages of the AIDS Risk Reduction Method and consistent condom use. Variables that remained significant--as well as factors that are conceptually important--were kept in the final model. Thus, we ended up with one final model to predict consistent condom use (see Table 3).

Table 3. Estimated Odds Ratio From Multiple Logistic Regression Predicting Consistent Condom Use Among Sex Workers in Kramat Tunggak.

Predictive Variable(a)                        Consistent condom use
                                               OR(b)    95% CT(c)

Sociodemographic Factors
   Age (years)
       [is less than or equal to] 20(d)         1.00
       21 - 25                                  1.31   0.68 - 2.53
       > 25                                     1.40   0.69 - 2.82
   Education attained
       No school(d)                             1.00
       Elementary                               0.98   0.39 - 2.49
       Jr High School                           0.46   0.14 - 1.50
       Sr High School                           0.42   0.09 - 1.86
   Hometown
       West Java & Jakarta(d)                          1.00
       Central Java & Yogyakarta                1.81   1.04 - 3.15
       East Java & elsewhere                    1.08   0.54 - 2.14
   Years working in KT
       [is less than or equal to] 1(d)          1.00
       1 - 2                                    1.06   0.60 - 1.88
       > 2                                      0.18   0.07 - 0.43
Sex Workers' Personal Determinants
   Self-efficacy in carrying out the
    intention to do safer sex
       Low(d)                                   1.00
       High                                     1.61   0.94 - 2.75
   Experience in negotiating condom use
       Low(d)                                   1.00
       Med                                      1.76   0.75 - 4.12
       High                                     5.15   2.16- 12.25
   Experience in using condoms for
    family-planning purposes
       Nod                                      1.00
       Yes                                      9.13   1.85 - 45.08
External Factors
   Perception of clients' and managers'
    attitudinal barriers towards condom use
       Low(d)                                   1.00
       Med                                      0.35   0.20 - 0.62
       High                                     0.13   0.06 - 0.27
   Exposure to govt.'s programs
       No(d)                                    1.00
       Yes                                      0.59   0.36 - 0.97

Predictive Variable(a)
                                                p

Sociodemographic Factors
   Age (years)
       [is less than or equal to] 20(d)
       21 - 25                                 .42
       > 25                                    .35
   Education attained
       No school(d)
       Elementary                              .96
       Jr High School                          .20
       Sr High School                          .25
   Hometown
       West Java & Jakarta(d)
       Central Java & Yogyakarta               .04
       East Java & elsewhere                   .83
   Years working in KT
       [is less than or equal to] 1(d)
       1 - 2                                   .84
       > 2                                     .0001
Sex Workers' Personal Determinants
   Self-efficacy in carrying out the
    intention to do safer sex
       Low(d)
       High                                    .08
   Experience in negotiating condom use
       Low(d)
       Med                                     .19
       High                                    .0001
   Experience in using condoms for
    family-planning purposes
       Nod
       Yes                                     .007
External Factors
   Perception of clients' and managers'
    attitudinal barriers towards condom use
       Low(d)
       Med                                     .0001
       High                                    .0001
   Exposure to govt.'s programs
       No(d)
       Yes                                     .04

(a) Initially, we included all statistically significant variables from the bivariate analyses in the multiple logistic regression analysis to follow the hypothesis model (i.e., hometown, length of time working in Kramat Tunggak, knowledge of STDs, knowledge of AIDS, condom acceptance and condom rejection, self-efficacy in using condoms, intention to perform safer sex the next day, self-efficacy in carrying out the intention, previous experience in using [for family planning purposes] and negotiating condoms, consistent condom provision by the managers, and perceived clients' and managers' attitudinal barriers towards condom use), as well as other factors that were conceptually important for predicting condom use (i.e., age, education attained, and attending the government's programs). Variables with p < .05--as well as others that were conceptually important--were kept in the final model, (b) OR = odd ratio, (c) CI = confidence interval, (d) Reference.

Sociodemographic characteristics. Independent of the other variables in the model, sex workers from Central Java and Yogyakarta were more likely to practice consistent condom use than their colleagues from either West or East Java. The odds that these women practiced consistent condom use were nearly two times their colleagues (OR = 1.8; CI = 1.03 - 3.15).

Data showed that the longer the sex workers worked in Kramat Tunggak, the less they would practice consistent condom use. The "longest term" women were about six times less likely to practice consistent condom use when compared to others (OR = 0.18; CI = 0.07 - 0.43). From the qualitative study we learned that the longer-term women usually had regular clients or lovers. Since these were also the clients least likely to use condoms, we suspected that this might be the cause. Another reason may be that these women felt they were experienced enough to select which clients were healthy and which were not, although their concept of healthy was actually "clean in appearance."

Another interesting fact was that the higher the women's previous monthly income, the less they would practice consistent condom use (p for trend test [is less than] .002). A similar result was found for the variable number of clients over the previous two weeks: The more clients the women had, the less they practiced consistent condom use (p for trend test [is less than] .006).

How much clients pay for sex has been found to be a factor that determines sex workers' condom use (Mhalu et al., 1991; Picketing, Quigley, Hayes, Todd, & Wilkins, 1993). In Kramat Tunggak, however, the prices for different sex services (i.e., short time and overnight) were more or less fixed.

Unfortunately, our questionnaire provided limited data about the sex workers' current economic status. Since our income data only referred to the previous month and prices were fixed, this information was more of an indication of the number of clients served by the women than of their economic status. Our income data were positively correlated to the number of clients served: The Spearman correlation was 0.39 (p [is less than] .0001). Regression analysis also indicated that every increase of 7 clients was associated with an income difference of Rupiah (Rp.) 240,530.00 (approximately $109.00), or about Rp. 34,000.00 per client (approximately $15.00), which is similar to results of the qualitative study (i.e., the prices for short time and overnight sex were Rp. 15,000.00, or about $7.00, and Rp. 40,000.00, or about $18.00 respectively). Currency values are based on exchange rates at time of this study.

In the bivariate analysis, the number of clients over the previous two weeks also showed a negative association with consistent condom use: The more clients the women had, the less they practiced consistent condom use (p for trend test [is less than] .006). Therefore, we used only one variable, such as the number of clients, in the multivariate logistic regression.

In the multivariate model, the negative association between number of clients and condom use persisted. The cross-sectional study design, however, did not allow us to estimate a temporal relationship between them. Our qualitative study indicated that many sex workers, in their anxiety over losing their clients, were reluctant to insist that their clients use condoms. We assumed it was more likely that less consistent condom use caused the number of clients to increase, rather than the other way around. As the number of clients was also positively correlated with the number of years of working in Kramat Tunggak (Spearman rho: 0.12, p [is less than] .009), it could also be that many of those clients were either the regulars and/or the lovers, who were less likely to use condoms. Since the reverse causal path was a likely reason for this association, we decided to exclude the number of clients from the final model.

The sex workers' personal determinants. In the final model, knowledge about STDs and AIDS, perception of susceptibility and severity of the diseases, attitudes, and self-efficacy in using condoms did not significantly predict consistent condom use. On the other hand, women who had ever used condoms for family planning purposes were 9 times more likely to use condoms consistently (OR = 9; CI = 1.85 - 45.08), and the women most experienced in negotiating condom use were 5 times more likely to practice it consistently than the least experienced ones (OR = 5; CI = 2.16 - 12.25).

However, in a separate multiple linear regression model with experience in negotiating condoms as the outcome variable and other factors as independent variables, we found that this experience was significantly predicted by knowledge of STDs (p [is less than] .0001), positive beliefs about condoms (p [is less than] .0001), and self-efficacy in using condoms (p [is less than] .0001).

External factors. In the final model, one external factor was significantly associated with consistent condom use: Women who perceived the clients' and managers' rejection of condoms as high were eight times less likely to practice consistent condom use than women who perceived the rejection to be low (OR = 0.12; CI = 0.06 - 0.27). It was difficult to separate the clients' attitudes from the managers' in our data; however, the fact that the managers consistently providing condoms in their brothels did not significantly increase the women's consistent condom use might mean that the client factor was the more important one.

Since the number of women who received treatment and/or examination from the government monthly mobile service was too small, this variable was dropped, and only the variable of gaining AIDS knowledge from the government's talks was used to represent the influence of government programs. This variable had a negative impact on the sex workers' consistent condom use (OR = 0.58).

Further analysis showed that this variable was negatively confounded by multiple factors, such as years of working in Kramat Tunggak and education. As both were inversely associated, it seemed that the longer the women worked in Kramat Tunggak and the higher their level of education, the less likely they would attend the government talks. However, after controlling for these confounding factors, attending the talks was still inversely associated with the women's consistent condom use, suggesting that the government program is ineffective.

DISCUSSION

We have conducted a behavioral survey among brothel-based female sex workers in Kramat Tunggak, the largest and only official brothel in Jakarta. A large proportion ([+ or -] 28%) of participants were randomly chosen from among the population living and working in the complex, implying representativeness of the data obtained. Realizing that participants might give answers just to please the researchers, we have also conducted an extensive qualitative study to verify certain aspects of the participants' sex behaviors, especially regarding their condom use.

We found that only 36% (n = 154) of the participants reported always using condoms during the previous two weeks, 25% (n = 108) never used condoms at all, and the rest (39%) used condoms occasionally. The pattern of condom use in this community was not consistent over time: The reproducibility of these data in test-retest interviews was moderate, and when validated with diary data, showed a moderate correlation. Our qualitative study indicated that this inconsistency was mainly due to real inconsistency in condom use practice, and not because participants lied to us.

After adjusting for sociodemographic variables, participants' consistent condom use was significantly and directly associated with their experience in using condoms for family planning purposes and their experience in negotiating condoms with clients in previous times. On the other hand, significant inverse associations were found with the women's length of time as sex workers in Kramat Tunggak, their perceived rejection of clients and managers, and government talks as the source of their AIDS/STD knowledge.

Focusing efforts to modify the above factors may substantially change the women's condom use behavior, and may in turn reduce the spread of HIV-1 infection in this community. Trying to work within the existing system, we suggest interventions that place the programs' providers as main actors, involving the sex workers, the clients, and the brothel managers.

Although officials publicly state that condom use is promoted in areas with a high rate of prostitution to prevent men from getting AIDS, during the study we found that the government only had two programs to support condom use: the occasional talks and the monthly health services. In order to promote condom use and other safer-sex behaviors of the sex workers, the government should develop more aggressive and effective health services and education programs for the sex workers, the brothel managers, and the clients.

More regular and systematic talks to smaller groups, preferably based in only one brothel, with more discussions that involve everyone in the brothel including the managers, may be a more constructive way of communicating STD/AIDS knowledge and prevention measures to the sex workers. As condom negotiating and technical skills are important predictors for the women's consistent condom use, they should be included in the training curriculum. Moreover, as previous experiences in using condoms with clients or for family-planning purposes were very significant in promoting consistent condom use, it is best that the government not give sporadic talks or short, intensive condom-training programs, but should have more lengthy regular classes, similar to their other vocational programs. This way, the sex workers will have a chance to build their self-efficacy in using condoms by practicing their negotiating skills, and to revise them using friends' comments as inputs.

In our opinion, a daily clinic located inside the Kramat Tunggak complex would be more effective than a monthly mobile health service: Data showed that only a few sex workers got STD examination/treatment from the current monthly service. It should be staffed by doctors, nurses, and other health counsellors who are female, and its services should be both user-friendly and out-reaching. This clinic should primarily address the sex workers' health concerns, which do not necessarily mean STD problems. Only by doing this will the clinic gain the trust of the sex workers.

The biggest hindrance in setting up the clinic will be the government's reluctance to provide official health services for sex workers out of fear of public protest, since this may be misinterpreted as an act legitimizing prostitution (Sihombing, personal communication, 1995). Nevertheless, if the government is to work toward the women's social rehabilitation and not just oversee a quasi-legal brothel, it must begin to focus on returning healthy women to the community. A health clinic within Kramat Tunggak is a necessary first step in this direction.

Another significant factor that hinders the sex workers consistent condom use is the managers' and clients' attitudes toward condom use. To overcome this, the government should have a formal policy on condom use in Kramat Tunggak (and other brothels in Jakarta). In most of the brothels in Kramat Tunggak, one can see signs such as "No guns or sharp weapons allowed," or "Sorry, for security's sake, we will knock on your door every half hour." Managers posted these signs to comply with government safety regulations. If managers put up signs stating that condom use is a must in the brothel, the sex workers will then be legally empowered to negotiate condom use with their clients. The managers could also support this by either providing free condoms or by selling condoms in their brothels.

Direct programs targeted to clients in the general population are still difficult to launch in Indonesia, since this act will be interpreted as enhancing promiscuity. Intervention programs for these men can, therefore, only be conducted in brothel communities and in STD clinics; both still fail to attract the government's current attention. With regard to the brothel communities, billboards and posters to encourage condom use should be put up in brothel complexes. Condom booths, where the clients can buy condoms as well as receive free leaflets about STDs, AIDS, and condoms, should be placed in strategic places inside the complexes.

Finally, our study has shown that combining a behavioral survey with a qualitative study enables us to understand the data better. In the future, a survey among the general population surrounding brothel complexes should also be conducted to understand their involvement with the sex workers inside. The results may be used as a basis for expanding the safer-sex information campaigns outside of the brothel complexes.

REFERENCES

Ajzen, I., & Fishbein, M. (1980). Understanding attitudes and predicting social behavior. Englewood Cliffs, NJ: Prentice-Hall, Inc.

Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall, Inc.

Bandura, A. (1989). Perceived self-efficacy in the exercise of control over AIDS infection. In V. M. Mays (Ed.), Primary prevention of AIDS: Psychological approaches. Newbury Park, CA: Sage Publications.

Basuki, E. (1991). Perilaku berisiko tinggi terhadap AIDS pada kelompok wanita tuna susila Kecamatan Pasar Rebo Jakarta Timur [AIDS-related high risk behaviors among female sex workers in Pasar Rebo district, Eastern Jakarta]. Unpublished manuscript.

Catania, J. A., Kegeles, S. M., & Coates, T. J. (1990). Towards an understanding of risk behavior: An AIDS Risk Reduction Model (ARRM). Health Education Quarterly, 17, 53-72.

Centers for Disease Control and Prevention. (1990). Epi-Info 5.0. Atlanta, GA: CDC Press.

Dinas Kesehatan DKI Jakarta [Jakarta Health Provincial Office]. (1988-1995). Archives Year 1995. Jakarta, Indonesia.

Dinas Sosial DKI Jakarta [Jakarta Social Welfare Provincial Office]. (1993, December). Himpunan peraturan tentang Panti Rehabilitasi Wanita Tuna Susila Dinas Dosial DKI Jakarta [Collections of rules and regulations on sexworker rehabilitation programs). Jakarta, Indonesia: Dinas Sosial DKI Jakarta.

Fineberg, H. (1988). Education to prevent AIDS: Prospects and obstacles. Science, 239, 592-596.

Ford, K., & Wirawan, D. N. (1996). Behavioral interventions for reduction of sexually transmitted disease/HIV transmission among female commercial sex workers and clients in Bali, Indonesia. AIDS, 10, 213-222.

Fowler, Jr., F. J., & Mangione, T. W. (1990). Standardized survey interviewing: Minimizing interviewer-related error. Newbury Park, CA: Sage Publications.

Gunawan, S. (1997). Sexually transmitted disease control programs in Indonesia. Paper presented at the 4th International Congress on AIDS in Asia and the Pacific. Manila, The Philippines.

Indonesian Health Ministry (November 7, 1995). Pemerintah tak promosikan penggunaan kondom [Government will not promote condom use]. KOMPAS.

Janz, N. K., & Becker, M. H. (1984). The Health Belief Model: A decade later. Health Education Quarterly, 11, 1-47.

Jones, G. W., Sulistyaningsih, E., & Hull, T. H. (1995). Prostitution in Indonesia. Canberra, Australia: The Australian National University Press.

Linnan, M., Kestari, M., & Kambodji, A. (1995). Adult sexual behavior and other risk behaviors in East Java: Behavioral surveys from urban, peri-urban and rural areas of East Java. Unpublished manuscript.

Mhalu, F., Hirji, K., Ijumba, P., Shao, J., Mbena, E., Mwakagile, D., Akim, C., Senge, P., Mponezya, H., Bredberg-Raden, U., & Biberfeld, G. (1991). A cross-sectional study of a program for HIV infection control among public house workers. Journal of Acquired Immune-Deficiency Syndrome, 4, 290-296.

Ministry of Health, Republic of Indonesia: Subdirectorate of STD, AIDS and Yaws Control, Directorate General of Communicable Disease Control & Environmental Health. (1998). Monthly Report of December, 1998. Jakarta, Indonesia.

Pickering, H., Quigley, M., Hayes, R. J., Todd, J., & Wilkins, A. (1993). Determinants of condom use in 24,000 prostitute/client contacts in The Gambia. AIDS, 7, 1093-1098.

Rahardjo, H. (1992). Isu dan pemahaman AIDS terhadap penghuni di lokalisasi WTS di Kramat Tunggak [Knowledge of AIDS among female sex workers in Kramat Tunggak]. Unpublished manuscript.

Sawanpanyalert, P., Ungchusak, K., Thanprasertsuk, S., & Akarasewi, P. (1994). HIV seroconversion rates among female commercial sex workers, Chiang Mai, Thailand: A multi cross-sectional study. AIDS, 8, 825-829.

Sedyaningsih-Mamahit, E. R. (1997). Clients and brothel managers in Kramat Tunggak, Jakarta, Indonesia: Interweaving qualitative with quantitative studies for planning STD/AIDS prevention programs. Southeast Asian Journal of Tropical Medicine and Public Health, 28, 513-524.

Stata Corporation (1993). Reference Manual 1,2,3. College Station, TX: Stata Press.

Swaddiwudhipong, W., Chaovakiratiping, C., Siri, S., & Lerdlukanavonge, P. (1990). Sociodemographic characteristics and incidence of gonorrhea in prostitutes working near the Thai-Burmese border. Southeast Asian Journal of Tropical Medicine & Public Health, 21, 45-52.

Taha, T. E. T., Canner, J. K., Chiphangwi, J. D., Dallabetta, G. A., Yang, L. P., Mtimavalye, L. A. R., & Miotti, P. G. (1996). Reported condom use is not associated with incidence of sexually transmitted diseases in Malawi. AIDS, 10, 207-212.

Van der Sterren, A., Murray, A., & Hull, T. H. (1995). A history of sexually transmitted diseases in the Indonesian archipelago since 1811. Canberra, Australia: Australian National University Press.

Wilson, D., Sibanda, B., Mboyi, L., Msimanga, S., & Dube, G. (1990). A pilot study for an HIV prevention programme among commercial sex workers in Bulawayo, Zimbabwe. Social Science & Medicine, 31, 609-618.

Zenilman. J. M., Weisman, C. S., Rompalo, A. M., Ellish, N., Upchurch, D. M., Hook III, E. W., & Celentano, D. (1995). Condom use to prevent incident STDs: The validity of self-reported condom use. Sexually Transmitted Diseases, 22, 15-21.

Manuscript accepted November 23, 1998

Endang R. Sedyaningsih-Mamahit
Communicable Disease Research Center, Jakarta, Indonesia

Steven L. Gortmaker
Harvard School of Public Health, Boston

Supported by the government of Indonesia through the Overseas Training Office - National Development Planning Board, Jakarta, Indonesia.

We are grateful to Dr. K. Heggenhougen, Dr. D. Hunter, Dr. G. Wyshak, and D. Halstead for their contribution to this work. Special thanks go to the team: Dama, Tini, Vita, Tuti, Sofie, Nila, Ipur, Titing, Nia, Mega, Lana, Hikmah, Mar, and Slamet. Finally, we are indebted to all participants in the field, especially the women of Kramat Tunggak.

Address correspondence to Dr. Endang R. Sedyaningsih-Mamahit, DPH; JI. Pendidikan Raya III Blok J-55, Duren Sawit; Jakarta 13440; Indonesia; e-mail: [e-mail].