A cross-sectional study was undertaken of a broad, representative and stratified sample of the adult population (from 18 to 75 years old) who live in the Andalusian Community. A company which specializes in health surveys undertook data gathering in respondents’ homes. All of the interviewers attended a one-week training course given by the head researchers (JC), and they showed that they had sufficient knowledge of interview techniques and filling out scales and questionnaires. The teaching techniques included talks, role-playing between interviewers and scoring video recordings of interviews undertaken by experts and volunteers. Sampling took place by using different levels of successive stratification: a) proportional stratification between 2 geographical areas, the east and west of Andalucía; b) stratification according to population density in the cities in each one of both geographical areas; c) stratification based on the population in each one of the 8 Andalusian provinces; d) simple randomization methodology was used within each province to select from one to 5 municipalities in each type of location (urban, intermediate and rural). Finally, the final sampling areas were selected by using the same simple random assignation method while taking into account age and sex quotas, census sections and the districts within each locality. The final sampling units were therefore individuals of both sexes aged from 18 to 75 years old who were interviewed after selecting one of every 4 consecutive homes within the predetermined street routes in the districts and census sections previously selected at random. The necessary sample size was calculated by estimating a 2% prevalence of mental disorders with an exactitude of ±0.5%, with confidence intervals of 95% (CI 95%) and for an estimated effect size of 1.5.

SAD was diagnosed using the Spanish version25 of the MINI international neuropsychiatric interview.26 This is a short structured diagnostic interview (of questions with Yes(No answers) on Axis I psychiatric disorders compatible with the Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (DSM-IV)27 and the tenth edition of the International Classification of Diseases (CIE-10).28 The MINI is divided into 16 modules that are identified by letters, each one of which corresponds to diagnostic category. The MINI interview was administered during the PISMA-ep study in the way originally described by the authors as valid, that is, in order. More specifically, it started with section A (major depressive episode) and ending in P (antisocial personality disorder). Nevertheless, to undertake this work the SAD module was specifically included, corresponding to section G and its corresponding items (Figure 1). This diagnosis is classified as current SAD, and the questions refer to the last month. If the answer is No, the diagnosis of SAD is ruled out. If the answer is Yes, the respondent has to answer three more questions to confirm the diagnosis. To select the patient as a possible case, they have to answer Yes to all four questions. The MINI interview has been used in many different cultures,29–31 and it has display satisfactory psychometric properties in each language, with kappa values that are in concordance with other interviews such as the Composite International Diagnostic Interview (CIDI)32 or the Structured Clinical Interview for DSM-IV Disorders, Patient Edition (SCI)33, which in the majority of cases are above the value of 0.7. Its inter-evaluator reliability has also been shown to be high, with a correct level of sensitivity and a low rate of false positives when it is used, as it was in this study, in a healthy population living within the community.

Fig. 1.

Diagnosis of social phobia using module G of the MINI international neuropsychiatric interview.25

(0.23MB).

In total, 70.8% of the households selected at first did not respond after 4 attempts or lacked respondents within the age and sex quotas of the study, so that they had to be replaced by the next available household on the predetermined route. A total of 5,496 households were therefore finally contacted for this study, of which 4,507 completed the interview (a response rate of 83.7%) (Figure 2).

Fig. 2.

PISMA-ep cohort sample.

(0.15MB).

Rates of error in data input were far lower than 1% in the 8 Andalusian provinces. Detailed information on the methodology and protocol of the PISMA-ep study have been published in the scientific literature.15

Study of the data commenced by calculating the prevalence of SAD, using a 95% CI. Descriptive analysis of the independent variables included calculations of frequency, percentage, average and standard deviation.

The association between SAD and the other selected independent variables was calculated using the Chi-squared test (2 tails) and the Student t-test, depending on data distribution. To finish, a multivariable regression model was used with a logistic regression technique in which all of the variables that had been found to be significant in the bivariate analysis were included once (the “enter” method). The level of statistical significance for all of the analyses was set at lower than 0.05. Data were stored and processed using version 26 of the Statistical Package for the Social Sciences (SPSS).

The study response rate was 83.7%, as 4507 of the 5496 who were invited completed the study (n=4507). The monthly prevalence of SAD in our population sample was 1.1% (95% CI: 0.8-1.4). Table 2 shows the prevalence of SAD for each one of the main sociodemographic and clinical variables together with its corresponding bivariate analysis. To summarise, the prevalence of SAD was significantly higher (Chi-squared; P=.042) in women (1.4%) than it was in men (0.8%), in younger subjects (Student t; P=.020), in respondents who were neither married nor living together vs. those who were married (1.6 vs. 0.8%; Chi-squared; P=.020), in those with poorer social and personal functioning as determined by score lower than 70 on the PSP scale (6.8% vs. 0.8%; Chi-squared; P<.001), in subjects who had suffered childhood abuse (Chi-squared; P<.001) that was physical (3.8% vs. 0,9%), psychological (5.4% vs. 0.7%) or sexual (10.4% vs. 1.0%) or who had suffered stressful event in the previous six months (Student t; P<.001). Higher levels of SAD prevalence were also found in connection with clinical variables such as having a probable personality disorder (Student t; P=.015) or paranoia (Student t; P<.001) and health variables such as nicotine dependency (Student t; P=.038) and personal state of health (Student t; P<.001).

A multivariate model was developed for association with SAD based on 2 demographic risk variables (female sex [OR=1.134; 95% CI=1.6-2.1; P=.697] and a younger age [OR=0.974; 95% CI=0.98-0.99; P=.019]), 3 psychosocial variables (a lower level of independence [OR=0.702; 95% CI=0.5-0.9; P=.024], having suffered abuse during childhood [OR=2.441; 95% CI=1.3-4.7; P=.007] or having suffered threatening life events in the previous 6 months [OR=2.424; 95% CI=1.0-5.6; P=.04]) and 3 clinical variables (probable personality disorder [OR=2.736; 95% CI=1.0-5.6; P=.005], a higher level of neuroticism [OR=1.492; 95% CI=1.2-1.8; P<.001] and a higher score for paranoia [OR=1.042; 95% CI=1.0-1.1; P<.001]). Figure 3 shows the findings of multivariate analysis with the odds ratios and the CI for each one of the model variables, as well as its statistical significance. Lastly, we also explored a multivariate model to study the association between SAD and the other psychiatric disorders assessed by the MINI. SAD was thereby found to be associated as a comorbidity independently of age and sex with major depression (OR=11.322; 95% CI=5.9-21.8; P<.001), panic disorder (OR=3.962; 95% CI=1.7-9.2; P=.001) and alcohol abuse (OR=4.097; 95% CI=1.4-12.1; P=.011) (Figure 4).

Fig. 3.

Results of the variables associated independently and in multivariate form with SAD.

(0.14MB).

Fig. 4.

Results of psychiatric disorders associated independently and in multivariate form with SAD.

(0.13MB).