Associations between abdominal pain symptom dimensions and depression among adolescents

2.1 Sample

In September 2010 through June 2011 adolescents from the municipality of Tromsø in Northern Norway were invited to participate in a cross-sectional population based study that included all pupils in first-year high school (11th school year) in both academic and vocational educational programmes from all high schools in the study area. Each participant completed a comprehensive health questionnaire in the Department of Research at the University Hospital of North Norway.

A total of 1117 students from five high-schools were invited and of these 1038 students participated (participation rate = 92.9%). Participants that were 18 years or older (n = 77) were excluded from the analysis since this was a study of only adolescents, leaving a final sample of 469 girls and 492 boys, aged 15–17 years (mean = 16.1).

The study was approved by the Regional committee for medical and health research ethics in North Norway. Written informed consent from the participants was obtained prior to inclusion.

2.2 Measurement of depression

The Short Mood and Feeling Questionnaire (SMFQ) was used to measure depressed mood. SMFQ is a 13-item screening tool for depression, which has been validated in both clinical samples and in the general population [2,3]. The Norwegian translated full version of the Mood and Feeling Questionnaire (34 items) has been validated against other measures of depression [34] and the SMFQ has also been used in an previous Norwegian epidemiological studies of depression among adolescents [24]. SMFQ sum scores above ≥11 have a sensitivity of 0.86 and specificity of 0.87 for detecting clinical depression in adolescents [35]. Consequently, participants scoring above this cut-off were operationally defined as cases suffering from depression, but without subsequent clinical depression diagnostics. The SMFQ does not include questions about somatic symptoms or pain.

2.3 AP and IBS case definition

A broad definition of AP was used in order to include as many participants as possible, enabling further subgroup classification with respect to pain frequency and duration, and other dimensions of AP (see below for details). Therefore, participants reporting AP during the past two months that occurred at least once a month were defined as AP cases.

The revised Irritable Bowel Syndrome criteria from 2006 (Rome III IBS criteria) were used to identify participants with IBS symptoms [30]. Adolescents were classified as IBS cases if they reported weekly abdominal pain or discomfort during the past two or more months and two or more of the following associated bowel symptoms at least 25% of the time: (1) relief with defecation (2), change in stool frequency (3) and/or change in stool form.

Both the AP and IBS case definitions were solely based on self-reported symptoms, not including self-reported clinical diagnosis or medical diagnostic examinations. Adolescents who fulfilled neither AP nor IBS criteria were defined as controls in the statistical analyses.

2.4 AP symptom dimensions

AP cases were further sub-classified along the following symptom dimensions: (1) how long they had abdominal pain; (2) how often they experienced abdominal pain or discomfort; (3) how long each painful episode lasted; (4) whether their symptoms were limited to a single abdominal site (below, around or above the bellybutton) or whether the symptoms were widespread and included two or all of these sites; and (5) the average intensity of their abdominal pain, rated on a 0–10 numeric rating scale (NRS), with anchors “No pain” and “The most intense pain imaginable”. The AP intensity results were sub-classified as mild (NRS 1–3), moderate (NRS 4–6) and severe (NRS 7–10) due to the somewhat skewed distribution and few cases with the very lowest or highest ratings. Additional classification was made with respect to associated bowel symptoms that are integral to Rome III IBS criteria: (6) softer stools and/or more frequent stools (diarrhoea), (7) harder stools and/or less frequent bowel movements (constipation), and (8) improvement with defecation and (9) a combination of two or more the bowel symptoms as defined in the IBS diagnostic criteria. Finally, in a separate part of the questionnaire, the participants were asked whether they had persistent pain that had lasted 3 months or longer. AP cases were classified as having co-morbid chronic pain (10) if they responded “yes” to this question, and indicated any site other than the abdomen as the location for this pain. The study questionnaire covered questions on specific pain localization including pain sites in the musculoskeletal and uro-genital system and head, but further sub-classification of co-morbid pain symptoms and other chronic pain disorders (e.g. headache) was not included in this study.

For an overview of the group categorizations of the above dimensions, see Table 3.

2.5 Data analysis

The statistical analysis was performed using the IBM SPSS version 20 software. Results with p < 0.05 are reported as statistically significant. Univariate analysis of differences in sex, parental education, and the prevalence of depression between AP and IBS groups and controls were analyzed with χ²-tests. Associations between depression and IBS and AP were analyzed using stepwise logistic regression, adjusting for differences in sex, parental educational level and co-morbid chronic pain in the second step. Finally, analysis of associations between depression and AP symptom dimensions were performed among individuals with AP only. This analysis was performed in two steps: first, the relationship between depression and each symptom dimension was tested individually, using logistic regression and controlling for sex. Next, all symptom dimensions that were statistically significant in the first analysis were entered in a multivariate logistic regression. This two step approach was used due to the large number of predictors tested in a relatively small sample. Similar multivariate analyses were not performed within the participants reporting symptoms of IBS, due to the small sample size of this group.

Multi-collinearity statistics were computed with a multi-linear regression model of the continuous SMFQ sum score, including all the AP symptom dimensions as independent variables.

Possible interaction effects between sex and IBS, AP and the AP dimensions in the association with depression are reported when statistical significant.

3.1 AP and IBS case description

AP was reported by 27.0% of the participants and 8.2% met the Rome III IBS criteria as shown in Table 1. Approximately half of the AP and IBS cases had symptoms lasting one year or more (49.3% and 50.3%, respectively), and the majority had symptoms more than three months (81.9% and 86.6%, respectively), thus the remaining cases had symptoms during the last two to three months. The sample prevalence of depression and co-morbid pain were 11.5% and 17.9% respectively. Both depression and co-morbid pain were more common in AP and IBS compared to controls (p < 0.01). Low back pain was the most common co-morbid chronic pain site in the full sample (10.2%), and was reported more frequently by both AP and IBS groups (AP 15.2%, IBS 16.9% vs. controls 8.2%, p < 0.01 both comparisons). Both AP and IBS were more common among girls than boys (AP girls 34.5% vs. boys 19.7%, p < 0.01 and IBS girls 10.3% vs. boys 6.2%, p < 0.05) as were the prevalence of co-morbid chronic pain (girls 22.6% vs. boys 13.5%, p < 0.01) and depression (girls 15.9% vs. boys 7.3%, p < 0.01).

3.2 Associations of AP and IBS with depression

Adolescents with IBS and AP had three times higher odds of depression compared to controls, as shown in Table 2. These results remained significant after controlling for sex and co-morbid chronic pain, both independent predictors of depression. Parental education did not emerge as a significant predictor in the analysis. The magnitude of the adjusted results were comparable for IBS and AP (OR = 2.4 and 2.5 respectively).

3.3 Associations between AP symptom dimensions and depression

Analysis of relationships between the individual symptom dimensions and depression among AP cases are shown in Table 3 (Step 1). Sex was not found to be a significant predictor in these analyses (OR = 1.9, CI = 1.0–3.7). However, as the results suggested a trend, sex was included in the second step below. Adolescents with severe pain had more than four times higher odds of depression than mild pain cases, and almost four times higher than those with moderate pain (OR = 3.9, CI = 1.9–8.3, p < 0.01). Individuals with an episode frequency of several times per week had significantly higher risk of depression than cases with monthly pain (p < 0.01), but the difference compared to adolescents with pain once a week was not significant. Individuals reporting episode duration greater than four hours had also a higher risk of depression than those with symptoms less than one hour, but the differences were non-significant compared to episodes lasting less than four hours. Abdominal pain distribution was the single strongest predictor, with OR = 6.0 for depression compared to adolescents with localized abdominal pain, with OR = 4.9 in two sites (CI = 2.3–10.8, p < 0.01) and OR = 9.4 in three sites (CI = 3.1–28.3, p < 0.01). We found no significant differences in depression for localized pain across the three abdominal pain sites. None of the bowel symptom indices were significantly associated with depression.

AP symptom dimensions that were found to be significantly associated to depression in the first analyses were included in a multiple logistic regression analysis as shown in Table 3, Step 2. Severe pain intensity, widespread abdominal pain distribution and co-morbid chronic pain remained the only significant predictors of depression. The odds ratio for co-morbid pain was unchanged (OR = 3.3) from Step 1. Associations between depression and abdominal pain severity and pain distribution were only slightly attenuated. Multiple abdominal pain sites remained the strongest single predictor of depression in the multivariate model (OR = 5.5, p < 0.01). The same AP subgroup had the highest depression prevalence as shown in Fig. 1. Pain frequency and episode duration, both significantly associated with depression when considered individually, did not emerge as independent predictors when other symptom dimensions were included in the analysis.

Fig. 1

Prevalence of depressive symptoms in abdominal pain as function of symptom severity.

No collinearity problems were found in the multi-linear regression analyses of the SMFQ sum score and the AP symptom dimensions, with variance inflation factors below 1.5 for each independent AP symptom dimension variable.

4.1 AP and IBS prevalance

Both AP and IBS are common complaints among adolescents [7,11,16,43]. In this cross-sectional study, more than one in four participants reported monthly or more frequent AP, half of them with symptoms more than one year, and nearly one third of the AP cases met the criteria for IBS. Both AP and IBS were more common among girls than among boys, as is the case for most pain conditions [17,29]. Direct comparison of the prevalence of AP with previous studies is difficult, due to the diversity of AP definitions and consequent estimates. However, our findings lie well within the range reported by other studies of recurrent abdominal pain [7]. The IBS prevalence in this study was lower compared to earlier studies of adolescents [11,42,43], but higher than in one community study of pre-pubertal children [13].

4.2 Associations of AP and IBS with depression

The high prevalence of AP ad IBS reported by adolescents should not be taken as an indication that these complaints are inconsequential, as is illustrated by the high rates of depression in these groups. It is notable that compared to controls; the odds ratio for depression was identical for AP and IBS, after adjusting for sex, parental education and co-morbid pain. This was unexpected, as AP was a broader category than IBS with respect to pain frequency (at least weekly in IBS vs. at least monthly in AP) and not dependent on associated bowel symptoms. This suggests that the risk of depression does not depend on the specific diagnosis, but rather on symptom dimensions that are common across various types of AP, as is discussed in the following paragraph. For both IBS and AP, co-morbid chronic pain and female sex were significant independent predictors of depression in the full sample analyses, which is consistent with a previous AP study [22]. In contrast, parental level of education had no appreciable association with depression, a finding that harmonizes with some earlier reports, but not with others [26,32]. These diverging results in the literature may be a consequence of different operational definitions of socio-economic status, and our finding does not give grounds to conclude that depression is unrelated to other socio-economic indices in this age group.

4.3 Associations between AP symptom dimensions and depression

In the analysis of symptom dimensions within the AP group, we could identify pain at more than one abdominal site, severe abdominal pain intensity and co-morbid non-abdominal chronic pain as independently associated to depression, after adjusting for sex and other AP symptom dimensions. Previous studies have partly described comparable results, but have analyzed symptom dimensions individually, rather than in an overall analysis as was done here [6,16,33,38,41]. To our knowledge, no studies have analyzed abdominal pain distribution in adolescents with AP and how this symptom dimension is related to mental health. However, it is known that chronic multi-site pain is an important risk-factor for mental health problems, daily life restrictions and reduced quality of life, compared to single-site chronic pain [1,4,5,15,21,25,27]. Thus one interpretation might be that increased risk of depression among those with widespread abdominal pain is part of a continuum where widespread pain in general is a greater mental stress-factor than more localized, and possibly more manageable, chronic pain. Our results indicate that diffuse visceral pain is strongly associated with depression and it is remarkable that pain distribution in the relative small area represented by the abdomen has such a profound association with mental health (OR = 5.5 for multi-site vs. single site AP cases). This interpretation is consistent with our finding that co-morbid chronic pain at non-abdominal loci also is independently associated to depression, with comparable magnitudes as shown in Table 2 (OR = 2.9 vs. 2.5 after adjustments for sex and parental level of education). How the distribution of co-morbid somatic pain is related to depression remains unanswered in this study. Finally, if AP is considered to be a stressor that increases risk of depression, it is to be expected that this effect will increase with increasing pain intensity, as found here (OR = 4.0 for severe vs. mild AP). However, our study is cross-sectional and uninformative about causality. While there is evidence that pain may be considered a stressor increasing risk of subsequent depression [4,21,39], there is also evidence that negative affect is a risk factor for the development of chronic pain [5,8,12,28]. Most likely, this relationship is bidirectional and that pain and depression are mutually reinforcing. There is also evidence of shared patho-physiological mechanisms underlying both symptoms, as is described in the medical literature [19,23].

4.4 Generalizability of the study results

In contrast to most other studies of patients with AP and depression we have included a population of unselected adolescents, also including sub-clinical AP cases. Our results are therefore representative of a broader spectre individual with AP. Most likely, the same associations are also generalizable to IBS, since AP is the cardinal symptom of IBS. In particular is it important to note that pain frequency and associated bowel symptoms, which are the symptom dimensions distinguishing IBS from AP in this study, had no independent predictive value for depression in the multivariate analysis. Thus the associations of AP and IBS with depression are not only quantitatively similar, as described above, but also appear to be qualitatively similar with respect to the symptom dimensions driving this association.

4.5 Study limitations

A limitation of this study was that the classification of AP and IBS relied on self-reported symptoms only, and did not include clinical examinations and diagnostics. For this reason we could not distinguish cases with underlying organic disease, or referred somatic pain (e.g. low back pain) from cases with functional gastrointestinal disorders. However, as the vast majority of AP cases are functional disorders, and IBS is the most common of these [9,10,18], the number of misclassified cases is likely to be low. Thus we believe that the study results are valid for most adolescents with AP and IBS symptoms in this population, with acceptable low risk of misclassification. To what degree AP was confounded with menstrual pain among the female adolescents was furthermore not answered in this study. However, as non-sex interaction effects were found, this unlikely to be a major confound in the analysis. Finally, it should also be noted that the lack of association between AP duration and depression may be explained by the overall short symptom duration reported in our sample (50% less than one year). However, similar results have been described in an earlier report [16], albeit without controlling for additional pain dimensions.