Article Text

Download PDFPDF

Original article
Changes in high-risk sexual behaviour among Estonian adults between 1996 and 2006
  1. Hedi Liivlaid1,
  2. Anneli Uusküla2
  1. 1Department of Health Statistics, National Institute for Health Development, Tallinn, Estonia
  2. 2Department of Public Health, University of Tartu, Tartu, Estonia
  1. Correspondence to Hedi Liivlaid, National Institute for Health Development, Department of Health Statistics, Hiiu 42, Tallinn 11619, Estonia; hedi.liivlaid{at}


Objective To describe and compare sexual behaviour patterns in 1996 (before the explosion of the HIV epidemic in 2000) and 2006, and the correlates of high-risk sexual behaviour in Estonia.

Methods Data on selected sexual behaviour characteristics were obtained by interviewing a probability sample of the adult population in Estonia in 1996 and 2006.

Results Data were available from 2441 sexually experienced women and 1994 men, and 3005 women and 2812 men in the year 2006. There were no changes in the proportion of sexually experienced adults, frequency of intercourse, or same gender sexual relationships. There was an increase in sex under the age of 16 (women: 8.4% vs 14.1% adjusted OR (AOR) 2.05 (1.67 to 2.53); men: 21.9% vs 28.0% AOR 1.58 (1.35 to 1.84)). From 1996 to 2006, the proportion reporting intercourse with a casual partner in the previous 12 months decreased significantly for women (10.7% vs 6.0% AOR 0.54 (95% CI 0.41 to 0.72)) and men, (24.0% vs 16.5% AOR 0.71 (0.59 to 0.85)) in parallel with a decrease in those reporting never using condoms during intercourse with a casual partner (for women: 49.3% vs 19.2% AOR 0.27 (0.14 to 0.51); and men: 35.4% vs 20.2% AOR 0.37 (0.25 to 0.56)). In 2006, the odds for not always using condoms in intercourse with casual partners increased significantly with age (eg, ≤24 vs ≥45: AOR 3.46 (1.68 to 7.13)), and were higher among those widowed/divorced/separated (AOR 2.35 (CI 1.20 to 4.60)).

Conclusions From 1996 to 2006, a significant reduction in high-risk sexual behaviour has occurred in Estonia across all age groups and genders.

  • Sexual Behaviour
  • Hiv
  • Sexual Health
  • Reproductive Health
  • Prevention

Statistics from


Sexual behaviour is not only of basic biological importance, but also of central social importance.1 Population patterns of sexual behaviour are major determinants of conception rates, sexually transmitted infections (STI), including HIV transmission, and other sexual health outcomes.

Examining patterns of individual knowledge and behaviour over time is critical to understanding the response to the AIDS epidemic at the population level, and the need for future interventions aimed at prevention. Among gay men in the USA in the 1980s, the revelation of the AIDS epidemic resulted in huge decreases in the proportion of men with multiple partners.2–4 Changing sexual behaviour is not only limited to high-risk population groups. General population studies from the USA,5 Switzerland,6 France7 and Sweden,8 have shown temporal changes in sexual behaviour over time: lower levels of risk behaviour (reducing numbers of partners, using condoms more frequently, selecting partners more carefully) earlier in the course of the epidemic.

The HIV epidemics in several countries in Eastern Europe, which are currently concentrated among injecting-drug users (IDUs), are at risk of spreading to the general population. Characteristics of the sexual networks of IDUs/their main sexual partners, and sexual behaviour of the general population, will probably determine the further generalisation of the HIV epidemics.9 However, data on sexual behaviour among the general population and, therefore, indications of any changes from these countries is scant.

In the early 2000s, Estonia witnessed an explosive injection drug-use-driven HIV outbreak that has led to the highest prevalence of HIV infection among adults aged 15–49 years in Europe (1.2% (1%–1.5%) in 2009).10 State-financed national programmes for HIV/AIDS prevention in Estonia started in 1992, with one of the key activities being preventive work among the general population (awareness campaigns), access to free and confidential HIV testing, and access to free HIV care/antiretroviral treatment. Further, the combination of increased attention on school sexual education programmes and youth-friendly services, has had a demonstrable effect on reducing adverse health outcomes in Estonia.11 ,12

This analysis was undertaken to provide population estimates of sexual behaviour patterns in 1996 (before the HIV epidemic), to compare them with estimates from 2006 (several years after the explosive HIV epidemic initiation in 2000), and describe the correlates of high-risk sexual behaviour.


The Estonian Health Interview Surveys (EHIS) 1996 and 2006 are stratified (age, gender, residency: urban/rural) probability sample surveys of the general adult population. Details of the methodology have been published elsewhere.13 ,14 Briefly, interviews were conducted with respondents aged 15–79 years in 1996, and 15–84 years in 2006. The response rates were 84.3 and 60.2 in 1996 and 2006, accordingly, which is in line with other general population surveys conducted in Europe.15 ,16 Trained interviewers conducted face-to-face interviews in respondents’ homes. The survey instrument used in both surveys was based on previously performed national surveys,17–19 WHO,20 and selected modules recommended by Eurostat.21 In both surveys, respondents were asked about their sexual health and behaviour. For the current analysis, only questions with identical wording in both surveys were used.


Respondents were asked if they had ever had sexual intercourse during their lifetime and within the last 4 weeks. The type of the sexual intercourse (oral, vaginal, anal) or the gender of the partners (opposite or same gender) was not specified in detail. Those reporting being sexually experienced (defined as those reporting ever having had sexual intercourse) were questioned further about their age at the time of first sexual intercourse and protection used to avoid pregnancy/STI at first intercourse (‘Did you or your partner do or use anything to avoid pregnancy/sexually transmitted diseases?’). Data on sexual activity and contraceptive use in the last 4 weeks was also collected. Use of hormonal contraceptives (pills, tablets, transdermal, emergency contraception), intrauterine devices, sterilisation or condoms were considered as effective contraception methods for the current analysis, while reliance on periodic abstinence (depending on the menstrual cycle), interrupted intercourse, chemical contraceptives (foam), and cervical cap were not. Respondents were also asked about intercourse with a casual partner and condom use in such intercourse during the last 12 months. Intercourse with a casual partner was defined as other than main relationships.

Respondents were asked if they had ever paid or received payment for sexual intercourse, had ever had intercourse with a partner of same gender, and if they had ever been sexually harassed.

Other data used in this analysis include respondents’ gender, age, residence, ethnicity, marital status, educational level and income. These measures were used to assess associations between high-risk sexual behaviour and sociodemographic factors.

Given the aims of the study, we restricted the analysis to sexually experienced (reporting ever having had sexual intercourse) respondents aged 15–79 years.

Statistical analysis

No formal sample size calculations were made for the original studies in 1996 and 2006. The sample size decisions for the original samples were based on good practice in neighbouring countries,22 and financial considerations (ie, an increase in sample size in 2006). However, a posthoc sample size calculation for the main outcome measure of the current analysis (the prevalence of those reporting casual partners) assuming a margin of error for the prevalence of 4%, and 95% CI for the estimate indicates that a sample size of 339 in 1996 and 236 in 2006 would have been needed. Given that the actual numbers of respondents included in the analysis (491 in 1996, 420 in 2006) exceeded the calculated samples’ size numbers, we consider that it is still appropriate to present the results of the analysis.

Descriptive statistics, including mean, median, SD, IQR and range were used for continuous variables. For categorical variables, percentages and absolute (n) frequencies are presented. Student's t test was used for continuous variables, and χ2 test for categorical variables to explore differences between groups. The data were weighted to adjust for the unequal probabilities of selection, so that the sample was broadly representative of the Estonian population. Complex survey analyses were used to take account of the weighting and stratification of the data.

ORs and 95% CI were used to compare estimates from EHIS 1996 and EHIS 2006. Adjusted ORs (AORs) and 95% CI were calculated using a logistic regression model, adjusting for age, gender, education, income and marital status. In the multiple logistic regression, unweighted regression estimates were used, including stratification factors as explanatory terms in the regression model.23 ,24 p Values of less than 0.05 were considered statistically significant.

Factors associated with high-risk sexual behaviour were assessed by the use of two variables:

  1. Reporting intercourse with a casual partner during the last 12 months (yes/no), and

  2. Reporting condom use during intercourse with a casual partner during the last 12 months.

Reporting not always using condoms in intercourse with a casual partner was used as an outcome variable in an analysis of the individual characteristics associated with high-risk sexual behaviour. The responses to condom use during intercourse with a casual partner were dichotomised into ‘more risky’ (combining responses were: never, sometimes and usually using condoms during intercourse with a casual partner), and ‘less risky’ (those reporting always using condoms during intercourse with a casual partner).

Associations between the measure of risky sexual behaviour, sociodemographic and lifestyle factors were assessed for sexually active respondents by means of multiple logistic regression analysis. Factors with p<0.1 in bivariate analysis were entered into a multivariate logistic regression model to adjust for potential confounding. The measures of association were expressed as OR and AOR with corresponding 95% CI.

Data were analysed using the statistical package STATA 11.1. The survey was approved by the Tallinn Medical Research Ethics Committee, and informed consent was obtained from all respondents before their inclusion to the study.


The overwhelming majority of the adult population interviewed in 1996 and 2006 were sexually experienced: 95.3% and 96.2% (p=0.11) of women, and 93.1% and 93.7% (p=0.34) of men in 1996 and 2006, respectively. With this restriction, the sample in the year 1996 consisted of 2441 females and 1994 males; and in the year 2006 it was 3005 females and 2812 males.

The proportion of male respondents was 45.8% and 45.0% in 1996 and 2006, respectively (p=0.55) (table 1). The mean age of female respondents was 43.7 years (SD=15.5) in 1996 and 44.8 years (SD=17.2) in 2006 (p=0.01). The mean age of male respondents in 1996 was 40.6 (SD=14.7), and in 2006 it was 40.5 (SD=15.2) p=0.82). Across the samples, close to half the respondents were married (56.3% in 1996, 46.0% in 2006; p<0.01), but there was a slight increase in the proportion of those reporting cohabiting over the decade (11.1% in 1996 and 18.4% in 2006; p<0.01). About half the respondents reported having received secondary education (ie, up to 12 years of education) (60.3% in 1996, 49.6% in 2006; p<0.01). There was an increase in the proportion of respondents who reported having university education (14.6% in 1996, 35.5% in 2006; p<0.01), and a decrease in the proportion of respondents who had only lower secondary (ie, up to 9 years) or less education (25.1% in 1996, 14.9% in 2006; p<0.01).

Table 1

Sociodemographic characteristics of sexually experienced participants in the Estonian Health Interview Survey (EHIS) 1996 and EHIS 2006, Estonia

Sexual behaviour

Among the sexually experienced respondents, the median age of first sexual intercourse was 19.0 years (IQR 4) in 1996 and 2006 for women, and 18.0 years (1996: IQR 4; 2006: IQR 3), respectively, for men. The proportion of respondents having sex before the age of 16 years was higher in 2006 than in 1996 for both women (AOR=2.05 95% CI 1.67 to 2.53; table 2) and men (AOR=1.58 95% CI 1.35 to 1.84; table 2).

Table 2

Selected sexual behaviour characteristics by gender reported in the Estonian Health Interview Survey in 1996 and 2006

The data showed a marked increase in the use of measures against pregnancy/sexually transmitted infection at first sexual intercourse. The proportion of respondents reporting taking precautions against pregnancy/sexually transmitted infection during their first sexual intercourse increased significantly between 1996 and 2006 among both women (19.8% vs 34.7%; AOR=2.13 95% CI 1.84 to 2.48) and men (19.1% vs 36.8%; AOR=2.62 95% CI 2.22 to 3.08).

In 2006, there were no significant gender differences in reporting same gender sexual intercourse (men vs women: 0.6% vs 0.6%; p=0.77). However, there were significant gender differences in reporting ever being sexually harassed (men vs women; 0.8% vs 6.9%; p<0.01), and being paid/receiving payment for sex (men vs women: 7.0% vs 0.4%; p<0.01). Experience of sexual harassment, reporting same gender sexual partners, and paying or receiving payment for sexual intercourse showed no changes over time.

There were no changes in the proportions of people reporting being sexually active within 4 weeks of the interview. Across both genders, the proportion of respondents using effective contraception during the last 4 weeks increased between 1996 and 2006. Although these changes are statistically significant, they are very modest on an absolute scale. However, there was an almost twofold increase in reported condom use among those sexually active within the last 4 weeks (table 2).

Risky sexual behaviour

Men were more likely to report having had intercourse with a casual partner in the preceding 12 months than women (in 2006: 16.5% vs 6.0%; AOR* 3.33 (2.80 to 3.94); p<0.01). However, the proportion of respondents reporting intercourse with a casual partner within the preceding 12 months has decreased significantly for both, women (10.7% vs 6.0%; AOR 0.54 95% CI 0.41 to 0.72) and men (24.0% vs 16.5%; AOR 0.71 95% CI 0.59 to 0.85) (table 2). The reported decrease during intercourse with a casual partner was statistically significant in all except the two oldest age groups (table 3).

Table 3

High-risk sexual behaviour during the last 12 months by age group, Estonian Health Interview Surveys (EHIS), 1996 and 2006

In 2006, more than half (58.3%) the respondents reported not always using condoms during intercourse with a casual partner in the preceding 12 months (table 3), and there were no gender differences in the proportion reporting not always using condoms during intercourse with a casual partner (men vs women: 57.8% vs 60.0%; OR 1.11 (0.70 to 1.74); p=0.66) (table 4). Nevertheless, from 1996 to 2006, the proportion reporting never using a condom during intercourse with a casual partner decreased for both, women (49.3% vs 19.2%, AOR 0.27 95% CI 0.14 to 0.51) and men (35.4% vs 20.2%; AOR 0.37 95% CI 0.25 to 0.56) (table 2). A decrease in not always using condoms during intercourse with a casual partner was observed across all age groups (table 3).

Table 4

Sociodemographic factors associated with reporting not always using condoms during intercourse with a casual partner in the last 12 months, Estonian Health Interview Survey 2006

In bivariate analysis, not always using condoms during intercourse with a casual partner was associated with older age (p<0.01), and marital status (p<0.01). After adjustment (for age and marital status), age and marital status remained significantly associated with not always using condoms during intercourse with a casual partner. The odds for reporting unprotected intercourse with a casual partner increased significantly with age (eg, ≤24 vs ≥45: 41.9% vs 67.5%; AOR=3.46 95% CI 1.68 to 7.13), and were higher among those widowed/divorced/separated (AOR=2.35 95% CI 1.20 to 4.60) (table 4).


To our knowledge, this is the first study to report on sexual behaviour in the general population before and at the time of the HIV epidemic in an Eastern Europe country facing an injection drug-use-driven HIV epidemic. The HIV epidemic in Estonia emerged against a background of radical changes in political, economic and social realities following the collapse of the communist system.

Based on our data from consecutive population-based surveys, there have been no significant changes in the reporting of the sexual behaviours studied in the general population (such as the proportion of sexually experienced adults, or the frequency of sexual intercourse) between 1996 and 2006.

Importantly, we documented significant declines in some high-risk sexual behaviour (reporting intercourse with a casual partner, not always using condoms in relationships with a casual partner) affecting both genders and across age groups. Between the two surveys, we recorded decreases in not always using condoms in intercourse with a casual partner across all age groups. Besides declines in risk behaviour, our analysis documents the increased use of effective contraceptive methods in the population. This information suggests that sexual health promotion messages may have had some impact. Since the early 1990s, there have been significant changes in policy and provision of sexual health promotion activities in Estonia. During the Soviet period (ie, pre-1991), sexual and reproductive health issues were primarily the responsibility of gynaecologists and midwives. In 1991, the first Youth Counselling Centre (YCC) was opened in Estonia, and today there is a network of 17 YCCs working on common principles (integrated approach to problems, providing counselling in medical, psychological as well as social matters), and serving over 25 000 young people each year. Further, in 1996, the school curricula incorporated a new compulsory subject starting from the fourth grade, called ‘human studies’. This subject included courses in health, relationships, psychology and family.25 Since the 2000s, new sexual health and HIV prevention initiatives (the National Programme of Reproductive Health, 2001–2009 and Estonian National HIV and AIDS Strategy, 2006–2015, respectively) are in operation.

Our findings are also in agreement with the continuous decline in the numbers of new bacterial STI reported in Estonia over the period.26 However, the lowest sexual risk behaviour occurred among the youngest respondents in both 1996 and 2006; that is, in 2006, less than half those aged ≤24 years reported not always using a condom during intercourse with a casual partner, but two-thirds of those aged ≥35 years did so. Increasing age and marital status, or being widowed, divorced or separated, were independent correlates for engaging in unprotected intercourse with a casual partner.

By using probability sampling techniques, large-scale behavioural surveys can provide robust estimates of the prevalence of behaviours in the population. However, our study has some limitations. The large number of participants implies stable estimates and statistical power. However, non-response can lead to bias in estimates of sexual behaviour characteristics. It is possible that respondents and non-respondents differ in terms of sexual behaviour characteristics. The 1996 survey achieved a response rate of 84.3% in comparison with the considerably lower response rate of 60.2% in 2006. Decreased participation in health research over the period has been reported by other authors.27 ,28 Differentials in non-response across years can affect the validity of comparisons provided. However, we have used analytic methods (weighting, adjusting) to facilitate the appropriateness of the comparisons. The results of the study may have been affected by measurement bias (recall, social desirability biases). To reduce the bias and strengthen the validity of the responses, the interviews used carefully phrased, non-judgemental and formal questioning. Further, interviews were conducted by trained interviewers in an environment familiar to the respondents, and as with the potential non-response bias, such biases would not be expected to undermine year-to-year comparisons. Our data analysis is based on the EHIS studies that did not specifically focus on sexual behaviours. Given the instrument design, we were not able to use some measures traditionally used to characterise sexual behaviour (ie, numbers of sexual partners), or describe some sexual behaviours in detail (eg, separating buying and selling sex), and the question on sexual harassment (‘have you ever been sexually harassed?’) used a rather broad term which could include a range of experiences in terms of HIV risk, however, we included it in our analysis in the absence of more specific questions about non-consensual intercourse or rape. In addition, we were unable to assess respondents’ knowledge or attitudes regarding HIV and changes in knowledge or attitudes. However, there is data from HIV awareness, attitude and behavioural studies documenting an increase in knowledge about HIV transmission and effective methods for protection against STI among those aged 25–29 years in the mid 2000s.29

Our results are in line with data from other European countries (Switzerland, Sweden) that have shown changes in partner change and condom use over time in response to, or at the time of, HIV epidemics.6 ,7 ,30 We expect people to respond to the HIV epidemic, because sex in a world with HIV carries a risk of death and, as a result, a risk of lost future utility.31 Examination of trends in self-reported behavioural change provides an opportunity to indirectly evaluate interventions (educational and other prevention efforts) aimed at reducing high-risk behaviours. However, disentangling the effects of the interventions, the impact of HIV and time trends is difficult, and beyond the reach of the current analysis. For example, in several instances, it has been discovered (retrospectively) that substantial change away from high-risk behaviours occurred before the formal health education/prevention campaigns actually began.32 ,33

In conclusion, our results show that from 1996 to 2006, a significant reduction in some high-risk sexual behaviours across age groups and genders, has occurred in the general population of Estonia. Our results underscore the importance of effectively targeting both younger and older individuals to reappraise risk perception and promote healthy sexual behaviours. Standardisation in terms of measurement of the most important variables for the population-based monitoring of behavioural changes will advance our understanding of the effectiveness and impact of population-based HIV prevention programmes.

Key messages

  • From 1996 to 2006, covering the period before and after the HIV epidemic emergence in Estonia, in the general population:

    • there have been no significant changes in basic sexual behaviour characteristics (proportion of sexually experienced adults, frequency of sexual intercourse);

    • there has been a significant reduction in high-risk sexual behaviour (intercourse with casual partners, and condom use in these relationships) across genders and age groups;

    • there has been a significant increase in reported condom use across genders.

  • Examining patterns of individual behaviour over time is critical for interpretation of the response to and the need for future interventions aimed at HIV prevention.


The authors thank the Institute for Health Development in Estonia for the opportunity to work with the data, and Mall Leinsalu, Allan Puur, Esa Läära for useful discussions in the process of manuscript revision.



  • Contributors HL, AU designed the structure of the manuscript and the required statistical analyses. HL conducted the statistical analysis. AU and HL interpreted the data and wrote the first draft of the manuscript. Both authors contributed to revising the manuscript and have approved it.

  • Funding This research was supported by Target Financing of Estonian Ministry of Education and Research grant SF0180060s09.

  • Competing interests None.

  • Ethics approval Tallinn Medical Research Ethics Committee.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.