The current research is an observational study of clinical sample, based on data collected from the standard electronic records used during the first interview conducted with patients who search for an Addiction Unit (PROAD, in the Portuguese acronym) for drugs dependence treatment.
The studied sample is one of convenience, comprehending all adult patients from both sexes who searched for PROAD (Addiction Unit of the Department of Psychiatric – Universidade Federal de São Paulo - UNIFESP) for treatment of their substance dependence, from January 2008 to May 2016. Only those with moderate or severe dependence of alcohol or cocaine or both according to DSM-IV (diagnostic and statistical manual of mental disorders) were included. Although part of data collection was performed before the release of DSM-IV criteria of dependence, the records of the patients had enough information for us to stablish their diagnosis according to DSM-IV. The use of tobacco was not considered an exclusion criterion.
The following excluding criteria were consiered: (1) Records with unanswered sexual behavior questionnaire; (2) Patients who declared not having practiced sex in the last six months.
From the total amount of 711 patients records in the period, 318 files were of patients with exclusive use of alcohol and alcohol with inhaled or smoked cocaine were selected for this study, and the final sample comprised 245 individuals after checking the excluding criteria, which corresponds to 77% from the selected records.
Considering the outcome “sexual risky behavior” in the two groups (alcohol users and alcohol + cocaine users), the total simple size (N = 245), the proportion of the outcome in each group (10 and 20%), significance level of 5% and sampling ratio 2.5 (175/70), the sample power was 94%. Considering the association analysis between the outcome and other variables (alcohol and alcohol + cocaine groups and covariates), the sample size was close to N = 100 + 20i (i = 7, counting the numerical and dichotomous as variable 1 and the number of dummy variables (number of categories − 1) for categorical variables with more than two categories). The most recommended formula is N = 100 + 50i .
Data collecting procedure
The first procedure at the institution is patient screening, with only the patient and the screener where the individualized therapeutic planning is carried out. It is a standard structured interview, with the use of instruments following an order, conducted by doctors or nurses, which lasts nearly one hour.
The main instruments used at the first appointment and in the current study address socio-demographic characteristics, identification of sexual abuse in childhood, practices related to sexual behavior and identification of psychoactive substances used in the last year.
The patients undergo the treatment voluntarily and the collected data is confidential. The screener explains that neither answering the questions presented in the instruments nor undergoing treatment is mandatory and the patients are free to stop the interview at any moment they wish to do so. Since the institution is devoted to conducting study and research, at the first interview the patients are informed that their authorization concerning the use of the unidentifiable data may be requested, provided such data is useful for a specific clinical study. It is important to highlight that the treatment is in no way harmed by the patients’ refusal in any phase.
The variables of the present study were organized in three blocks: socio-demographics, substance use, and sexual behavior.
The following socio-demographic data available at the screening were selected: age, sex, marital status, schooling and working.
Schooling was divided in a scalar non-dichotomized variable, the categories being “illiterate or up to 8 years of study”, “9 to 12 years of study” and “over 12 years of study”. Although unbalanced in number of years studied, this categorization was used due to data dispersion, so that no category had much more subjects than the others. As to working the results were categorized dichotomously with the possible answers being “yes” and “no” to the question whether the patient was practicing remunerated activity when the screening was conducted. Marital status was categorized as “single” and “married”; patients who identified themselves as widows/widowers and separated were included in the single category; the ones who lived with their partners out of wedlock were considered married.
Use of substances
The instruments include questions on use of substances and were developed at PROAD based on the questionnaire issued by the World Health Organization (WHO) in 2008 to assess the use of drugs at primary care . The questionnaire presents the possible answers “yes” or “no” to the use in the past year and the affirmative answer for more than one substance is possible. According to the use of substances in the past year, the users were classified in two groups: exclusive use of alcohol and use of alcohol with inhaled or smoked cocaine.
Sexual behavior and sexual abuse in childhood
The current study selected two questions to analyze sexual risk behavior: “how many people have you had sex in the last six months?” and “how frequent have you worn condom in the last six months during sexual intercourse?”, which led to the possible answers: “never”, “few times”, “most of the times” and “always”. Such answers were afterwards dichotomized as “inconsistent use” versus “consistent use”. Based on this two information a new variable was created to analyze the composite outcome, considering as risky sexual behavior in the past six months just the inconsistency in the use of condom and the sexual practice with more than one partner in the past six months.
Besides the above mentioned questions the following were included in the analyses: “have you ever had sex for money or drugs?”, “in the last six months have you had sex under the effect of alcohol or other drugs?” and “did you suffer any kind of sexual abuse in childhood?”
It is important to state that these questions were developed by the Addiction Unit facility and they do not comprise a validated instrument developed to make an in-depth analysis of risk sexual behaviors. All questions were self-reported. We believe that self-report was the best way to obtain these information, as sexual behavior and history of sexual abuse might be sensitive topics for most people and the presence of an interviewer could have led the subjects to not give the correct answers, in order to avoid judgments.
According to the aim, the study conducted the exploratory analysis of the association between the outcome variable “risky sexual behavior” and the group of substances users (alcohol exclusively and alcohol with inhaled or smoked cocaine) and the other variables, followed by a multivariate analysis.
A descriptive statistical analysis of the variables according to their nature, numeric or categoric, was carried on throughout the sample. The categoric variables were described by relative and absolute frequency, and the numeric variable “age” (in years) was described by central tendency measures (mean, medium) and variability (minimum, maximum, standard deviation, interquartile range) and tested in relation to adhesion to normal distribution by the Kolmogorov-Smirnov test.
The study on the association between risky sexual behavior and the group of users and the other categoric variables was conducted through the Pearson’s chi-square test (X2). In order to study the association between the numeric variable age and the outcome variable “risky sexual behavior” the Mann-Whitney U test was conducted, since “age” was not normally distributed between the two outcome categories.
The multivariate analysis was chosen as logistic regression analysis . The selection of the variables to compose the logistic regression model was based on the results obtained by the exploratory analysis: the variables selected were those which presented descriptive level (p value) inferior to 0.20 during the association test. To ensure the absence of association between the independent variables, thus avoiding multicollinearity in the model, Cramer’s V test was conducted. The test was interpreted according to Pett . No variable presented values superior to 0.50 to that coefficient (data not shown).
Once selected the explanatory variables, the univariate logistic analysis was conducted, presenting the Odds Ratio (OR) values, p value and confidence interval (CI 95%). Subsequently was carried on multivariate logistic regression, obtaining adjusted Odds Ratio (ORa), adjusted p value (p valuea) and adjusted confidence interval (CI 95%a).
The program SPSS 22.0 was used throughout the analysis in all the phases described above; the relevancy level of 5% was established for all the tests.
The study project was assessed and approved by Research Ethics Committee of ‘Universidade Federal de São Paulo’ (Protocol n. ° 0798/2016). The active patients included in the study received the information about it and signed the Informed Consent Form. Three attempts were made to contact by phone patients who were no longer undergoing treatment and they were informed of the present study and invited to sign the form if they wished to do so. Nevertheless, since it was necessary to work retroactively in revising patients’ registers the contact was not successful. In such cases the Ethical Committee authorized the use of data without the Informed Consent Form, but the researchers were compromised to keep privacy of the collected data remained. Considering the likely vulnerability of the patients and also their possible involvement in potential illegal activities, the data bank as well as the medical records consultation were made anonymous and secret.