5.9. Results of multivariate analyses (Papers II, III, IV)

5.9.1. Factors associating with frequent attendance

The aim of multivariate analyses was to find out which factors explain patients’ frequent attendance in the health centre. Using frequent attendance as a dependent variable, the factors that were significantly connected with frequent attendance in bivariate analyses were included in the logistic regression analyses as potential explanatory variables. These dichotomized factors were basic education (over / under 12 years), occupational status (working / not working), poor self-rated health (health rated as very good, rather good or satisfactory / health rated rather poor or poor), chronic somatic disease (no / yes), chronic mental disorder (no / yes), psychological distress (non-distressed / distressed), somatization (non-somatizing / somatizing) and hypochondriacal beliefs (no / yes). Age and gender were included as potential confounding variables in the analyses.

The odds ratios (OR) and their 95% confidence intervals (95% CI) based on the logistic regression analyses are shown in Table 7 before and after adjustment for chronic somatic disease. After controlling for chronic somatic disease, it was found that a low occupational status, poor self-rated health, mental disorders and hypochondriacal beliefs associated with frequent attendance.

5.9.1.1. Chronic somatic disease and hypochondriacal beliefs as predictors of frequent attendance

When these factors (low occupational status, poor self-rated health, mental disorders, hypochondriacal beliefs and chronic somatic disease) were included in the same model and a stepwise regression analysis was carried out, it turned out that chronic somatic disease and hypochondriacal beliefs were the only significant predictors of frequent attendance.

Patients with a chronic somatic disease were more likely to be frequent attenders (OR 5.6; 95% CI 2.6–12.5; P < 0.001) than patients without a chronic somatic disease. Hypochondriacal beliefs associated significantly with frequent attendance (OR 3.4; 95% CI 1.6–7.2; P 0.002).

Table 7. Association of various psychosocial and health factors with frequent attendance. Frequent attendance as the dependent variable and psychosocial factors as covariates one by one, adjusted for age and sex. The odds ratios (OR) and 95% conficence intervals (CI) are from logistic regression analyses, without and after adjusting for chronic somatic disease (method: Enter).

VariableWithout adjustment Adjusted for chronic somatic disease
OR95% CIPOR95% CIP
Low basic education 3.11.0–9.60.052 3.00.9–14.70.075
Low occupational status3.11.4–7.2 0.0072.91.2–6.90.018
Low self-rated health4.72.1–10.5 < 0.0013.0 1.3–7. 00.012
Chronic somatic disease5.82.7–12.5 < 0.001N.A.  
Chronic mental disorder2.91.1–7.80.0312.81.0–7.60.049
Psychological distress2.21.1–4.10.0181.70.9–3.30.115
Somatization2.11.0–4.10.0391.80.9–3.60.109
Hypocondriacal beliefs3.61.7–7.50.0013.4

1.6–7.2

0.002
Hypochondriacal beliefs and somatization interactive term10.02.2–45.30.0039.42.1–43.30.004
P-values by Wald’s tests.

Table 8. Association of various psychosocial and health factors with somatization among frequent attenders. Dichotomized somatization variable as the dependent variable and psychosocial and health factors as covariates one by one, adjusted for age and sex. The odds ratios (OR) and 95% conficence intervals (CI) are from logistic regression analyses (method: Enter).

VariableOR95% CIP
Low occupational status2.00.6–7.30.291
Low self-rated health15.3 1.7–134.10.014
Chronic mental disorder4.61.5–14.1 0.008
Hypocondriacal beliefs4.21.6 –11.10.004
P-values by Wald’s tests.

5.9.1.2. The interaction effect of hypochondriacal beliefs and somatization on frequent attendance

In order to find out the simultaneous interaction effect of hypochondriacal beliefs and somatization on frequent attendance, the interaction term was included in the stepwise logistic regression model with the somatization main effect term and chronic somatic disease, age and gender as other covariates. The analysis revealed that the interaction effect term and chronic somatic disease were significant explanatory factors of frequent attendance. The interaction effect (simultaneous occurrence of somatization and hypochondriacal beliefs) was a significant risk factor for frequent attendance (P 0.006) when the main effect term of somatization was included in the model and age, gender and chronic somatic disease were controlled for (Paper III).

5.9.2. Factors associating with frequent attenders’ somatization

5.9.2.1. Hypochondriacal beliefs and poor self-rated health as predictors of frequent attender’s somatization (Paper III)

The potentential explanatory factors significantly connected with frequent attenders’ somatization in the bivariate analyses were included in the logistic regression analyses. The dichotomized somatization variable was included in the analyses as a dependent variable. The dichotomized potential explanatory variables were: occupational status (working / not working), poor self-rated health (health rated as very good, rather good or satisfactory / health rated as rather poor or poor), chronic mental disorder (no / yes) and hypochondriacal beliefs (no / yes). Age and gender were included as potential confounding variables in the analyses.

The results showed that poor self-rated health, hypochondriacal beliefs and chronic mental disorders were connected as significant explanatory factors with frequent attenders’ somatization (Table 8). When modelling frequent attenders’ somatization, these factors were included in the same logistic regression model and a stepwise logistic regression analysis was carried out. It was found out that poor self-rated health (OR 5.7; CI 1.9–17.0; P 0.002 and hypochondriacal beliefs (OR 3.2; 95% CI 1.1–9.4; P 0.036) were significant predictors of frequent attenders’ somatization.

5.9.2.2. Poor marital communication as a predictor of female frequent attenders’ somatization

According to bivariate analyses, poor marital communication tended to associate with female frequent attenders’ somatization. The logistic regression analyses revealed an association between poor marital communication and somatization among female frequent attenders. When somatization was analyzed as a dependent variable, the odds ratio (OR) of being a somatizer was 5.8 (95% CI 1.3–25.1; P 0.019) among the female frequent attenders having the lowest quartile scores of marital communication, when the other quartiles were considered as a reference category, after controlling for age and chronic diseases. Among male frequent attenders, no such association existed.

Somatizing male frequent attenders assessed their marital communication as significantly better than somatizing female frequent attenders. Because of that, marital communication appeared to be a non-significant predictor of frequent attenders’ somatization in the total study group, and it was thus not included in the multivariate model.