Psychosomatic complaints and sense of coherence among adolescents in a county in Sweden: a cross-sectional school survey

The major causes of mortality and morbidity among adolescents have, during the last decades, shifted from infectious to behavioural aetiologies [1]. This shift is mostly due to risk taking behaviours among adolescents that lead to, among others, injuries and drug and alcohol misuse [2]. Mental health problems are also common among young people. The mental health problems can range from fully developed psychiatric diseases to learning disabilities, behavioural problems in school, speech- and language problems and psychosomatic complaints (PSC) [3]. To better understand the development of psychosomatic complaints among adolescents, it was decided to perform a population-based study, which is to our knowledge, the first to examine the relation between PSC and sense of coherence (SOC) among adolescents.

The distribution of psychosocial background factors is shown in Table 1. Approximately the same proportion of adolescents had families in which both parents and the child came from abroad (7,0%) as were born in Sweden with foreign parents (5,9%). Roughly one out of four (27%) of the adolescents lived in block of flats and three out of ten (33%) came from families with separated parents. Parental unemployment was reported by 14–15% of the adolescents.

The distribution, by gender and age, of the reported symptoms is described in Table 2. There was a gender difference in the symptoms reported, with the boys reporting fewer and less frequent symptoms than the girls. The most common symptoms among the girls were headache and feeling irritated, while the boys reported feeling irritated and having sleeping problems.

The distribution of psychosomatic symptoms related to psychosocial background factors varied with gender, as can be seen in Table 3. According to the severe PSC symptom dichotomization (>1 SD) there were significantly more girls affected by PSC, (χ² = 179.322, df = 1, p < 0.001) Among the boys, 9% have regular PSC as compared to 25% among the girls. In our study we found that severe PSC (> 1 S.D. from mean) among both boys and girls was significantly related to all the psychosocial background factors, i.e. ethnicity, residential type, family constellation and parental employment.

Besides the difference in the reported prevalence of symptoms, there was also a gender difference in the SOC scores. The girls scored lower than the boys in both age groups. The sixteen-year-old boys had a mean SOC score of 61.10 (S.D. = 12.14) while the nineteen-year-old boys had a mean score of 62.5 (S.D. = 12.33). The girls had a mean score of 59.0 (S.D. = 13.41) for the sixteen year olds and the nineteen-year-old girls had a mean score of 58.5 (S.D. = 12.01).

A bivariate correlation procedure computing Pearson's correlation coefficient or Pearson's r to determine the covariance between the psychosomatic symptoms was done. There was a weak association, between the psychosomatic symptoms, with values ranging between .489 and .281. The symptoms grouped themselves as follows: headache to stomach ache, feeling irritated to nervousness and problem sleeping to dizziness.

Table 4 illustrates that the stronger the SOC the lesser proportion of adolescents that can be classified as having severe PSC and vice versa (p < 0.001 among both sexes). Almost one out of ten boys and just over one out of four girls reported having severe PSC. In addition, in each SOC-subgroup girls had more than double proportions of severe PSC than boys. In the group of adolescent with the weakest SOC, more than 21% of the boys reported to have severe PSC in comparison to the more than 51% of the girls.

The one-way ANOVA (post hoc Tukey's test) showed that there was a significant association between the SOC quartiles and the psychosomatic index as shown in Table 5.

Table 6 shows the logistic analyses that were performed using the two models. In the first model, model A, SOC was tested in relation to PSC, and in model B, adjustment for psychosocial background factors was made. In the logistic model A, we found that the category of adolescents that reported many and severe psychosomatic complains was significantly related to SOC-quartiles. The same pattern was found when the model was adjusted for psychosocial background factors; model B. All psychosocial background factors had a non-significant relation too psychosomatic complaints when adjusted for SOC, except among girls from blocks of flats who had an increased risk for psychosomatic complaints (p = 0.046 OR 1.3) (not shown in figure). As shown in the table, adolescent boys with weak SOC had an increased risk of 14.7 times (unadjusted) to 16.4 times (adjusted) for experiences of severe PSC in comparison with boys with strong SOC (the reference category). The corresponding elevated risk for girls with weak SOC was 9.9 (adjusted) and 10.5 (unadjusted). Our SOC-models explained 16–20% of severe psychosomatic complaints present among the 16- and 19-year-old adolescents.

The results from our study showed that there was a statistically significant association between PSC and SOC, and also that lower SOC was associated with a higher prevalence of PSC. Adolescent boys with a weak SOC had an increased risk for severe psychosomatic symptoms, 15 times higher than boys with a high SOC. The corresponding risk-elevation among girls was 10 times. Conversely, higher SOC was associated with a lower prevalence of PSC. The findings in our study suggest that a strong SOC can help adolescents choose a coping strategy that is appropriate for the situation and thereby may prevent them from developing PSC, which is in line with other authors findings [21, 22].

Our results also indicate that the psychosocial background factors used in this study do not in any meaningful way explain the variation in psychosomatic complains, when controlling for SOC. Instead, the reverse pattern was found in which the SOC variable out-ruled the traditional psychosocial risk factors in our model. Consequently, this does not exclude the possibility that there are other psychosocial factors that could influence or explain the variation in PSC.

In an earlier study it was found that 25% of the children in the Nordic countries had at least one PSC occurring every week or every other week, with a higher prevalence among girls [23]. The most common complaints were headache and stomachache. It also showed that there are several factors related to PSC. These factors can be found at the child, family and socio-economic level [13]. In our study, we also found the same pattern with headache and stomachache as the most common complaints, of gender difference, with more PSC among girls. The proportion of adolescents that reported PSC was similar to the proportion found in the aforementioned study. In an unpublished study, we found that the proportion of girls with PSC increased from 14 to 21% during a 10-year period, while there was no increase in PSC among the boys.

Total SOC scores are normally lower for women than for men. The reason for this is not clear. However it has been suggested that the SOC scale could be based on a way of reasoning that is more relevant for women than men, as men and women have different coping strategies. Men tend to be more aggressive and presumptuous, while women are more reflective and emotional [24].

McSherry and Holm showed in their study that not only are low SOC individuals more psychologically distressed prior to a stressful situation, but they also maintained these higher levels of distress after stressful experience as well [25, 26]. Although the levels of distress showed significant pre and post increases for all three SOC groups, the low SOC group began and ended the experiment as the most distressed. This could be a part of the explanation why individuals with low SOC develop more psychosomatic symptoms, as the results from our study have shown.

Geyer discussed in his paper the statistical relationship between SOC and symptoms/disease [27]. He concluded that due to the simultaneous assessment of variables, it is open to debate whether a low SOC has some effect on the probability of falling ill or whether it is the other way around.

Our results indicate that SOC has an effect on the development of psychosomatic symptoms, as almost no psychosocial background factors explained any difference, when controlling for SOC. Instead, all psychosocial factors had significant relation to PSC if not controlled for SOC.

SOC could be interpreted as an autonomous resource contributing to a favourable development of subjective state of health. SOC data should; however, be regarded as complementary to and not a substitute for information already known to be associated with increased risk of future ill health [27]. Other authors report that SOC was not found to be a buffering variable [28].

All these measures are self-reported and therefore subject to reporting bias. Self-reported data are also considered to have questionable validity in many cases, especially survey questions about socially unacceptable behaviour. Questions about somatic and psychological symptoms are also difficult to validate in a cross-sectional study with a potential for under- or over-reporting that could lead to an inflation of the estimates of PSC in our study. Complementary interviews would have been preferred, but this was not feasible due to the large sample size and the confidential nature of the survey. There were no missing data analyses performed.

The strengths of this study are the large sample, the high response rate and the use of commonly used and tested measures. As the data was sampled during a lesson-hour under the surveillance of a teacher, the situation was controlled. However, with an external non-response rate of 16% (820 students) and an internal non-response rate that varied between 1 to 3% the students not attending the school during the period the survey was conducted may introduce a response bias. Non-respondents may also be more likely to suffer from PSC, hence leading to less statistical power in detecting this potential response bias.

The data from the adolescents in this study represent only one county in Sweden and this may limit our ability to make generalisations. The county does not, however, differ from other counties with regard to socio-economic or demographic factors.

Our results only represent relations among variables and not causations. The relations presented are associative, given that conclusions regarding the directions of cause and effect must be considered tentative. It is possible that a person who developed severe PSC has weakened ability to develop a strong SOC or coping ability, or that there are other, yet unknown factors that might be related to both SOC and PSC. These factors could be a genetic predisposition or environmental or cultural influences, as well as other psychosocial factors than those used in our study. As this study is one of the first studies to examine the association between SOC and PSC in an adolescent population, both a replications of this study and these analyses in different settings would be needed as is longitudinal study to follow the secular trend of PSC in adolescents.

In summary our study shows that adolescents with weak SOC report more psychosomatic symptoms than adolescents with a strong SOC. We also found that SOC ruled out traditional psychosocial risk factors, indicating that SOC is an intervening variable that mediates psychosocial risks in relation to PSC.