Prevalence of headache and associated factors among adolescents: results of a population-based study

: Introduction: Cephalalgia is one of the most common somatic complaints related to health problems in childhood and adolescence. Objective: To measure the cephalalgia prevalence in adolescents from the city of São Paulo, Brazil, and associated factors. Methods: This is a cross-sectional population-based study, carried out in 2015, with 539 adolescents of both sexes, aged between 15 and 19 years. The information was collected in a household survey, and the participants were selected from probabilistic sampling. Frequencies, χ 2 test and logistic regression analysis were used in the study, and significance level was 5%. Results: the estimated prevalence of cephalalgia was 38.2% (95%CI 33.8 – 42.7), and 7.8% (95%CI 5.6 – 10.7), migraine. The associated factors for cephalalgia were: female sex (OR = 2.2; 95%CI 1.4 – 3.4), Common Mental Disorder (OR = 2.8; 95%CI 1.7 – 4.9), vision impairment (OR = 2.6; 95%CI 1.6 – 4.2), besides back pain (OR = 2.2; 95%CI 1.3 – 3.5), sinusitis (OR = 2.0; 95%CI 1.2 – 3.4) and incomplete elementary education (OR = 3.0; 95%CI 1.6 – 5.6). Conclusion: The prevalence of headache among adolescents in the city of São Paulo represented more than 1/3 (one third) of this population. The main associated factors were sex, low schooling and the following comorbidities: common mental disorder and vision impairment.


INTRODUCTION
Cephalalgia is one of the most common complaints among children and adolescents. A systematic review study estimates that the prevalence is 58.4% among people aged less than 20 years 1 . In another study, regarding pain among adolescents aged 15 years, about 30% of them reported headache; 30%, back pain; and 20%, stomachache 2 .
In Brazil, a study analyzed the prevalence rates of cephalalgia. Only six studies were found, all involving adults and none in the city of São Paulo; the mean prevalence of cephalalgia was 70.6%, and migraine, 15.8% 3 .
Studies have shown the association between cephalalgia in adolescents with sex 1 , age 1 , parental anxiety 4 , mood disorders 5 and region of household 1 .
Since cephalalgia is a subjective phenomenon of pain, it is hard to diagnose and to treat. It is also considered as a major health issue, which can lead to worsened quality of life, reducing the capacity for work, study and leisure 6 . For adolescents, there are variables that make the sickness process more complex, with the addition of hormonal changes related to varied social adaptations and changes in levels of autonomy and responsibility, when adulthood is closer 7 .
The objective of this study was to estimate the prevalence and associated factors to cephalalgia among adolescents in the city of São Paulo.

Palavras
This survey is constituted of a sample composed of people aged between 12 years and older. We used the stratified probability sampling, including selection in two stages: • census sectors; • households 8 .
The study domains were the regions and the interviewees in the age groups of 12 to 19 years; 20 to 59 years; and 60 years of age or older. For purposes of statistical inference, each individual of the sample was associated with a sampling weight, composed of three components: • design weight, which considers the sampling fractions of the two stages of selection; • non-response adjustment; • post-stratification, which adjusts the distribution of the sample by sex, age group and region of the household, according to the distribution of the population in the city and according to the population estimation 8 .
In this study, 539 (97.3%) of the adolescents aged between 15 and 19 years were selected, with responses about cephalalgia.
The considered dependent variable was cephalalgia, whose measurement was obtained through the answer to the question: "Do you usually have migraine or headache?". Those who answered "yes" to the question were asked to define if they suffered from migraine and/ or headache. Independent variables were: • Sociodemographic variables: sex, age, race/ethnicity, and schooling; • Variables related to health status and life style: self-perceived health 9 , nutritional status 10 , smoking, alcohol consumption and use of contraceptives, only for women; • Variables related to chronic conditions: all self-reported ones were considered and tested; • Variables related to emotional condition: we selected those who answered "yes" to eight or more questions in the E block of ISA Capital 2015. This block is composed of questions from the Self-Reporting Questionnaire 20 (SRQ-20), instrument containing 20 questions, in which eight affirmative responses selected people with common mental disorders (CMD), revalidated by Gonçalves et al. 11 .
A descriptive analysis was carried out with proportions, 95% confidence interval (95%CI) and frequencies. The χ 2 test was used to test the difference between proportions. Then, the variables presenting p < 0.20 in the bivariate analysis for the multiple logistic regression analysis were selected, with odds ratio (OR) and 95%CI. Of these, the ones in which p < 0.05 remained in the model. The adjustment of the logistic regression model was assessed by the Hosmer-Lemeshow test.

REV BRAS EPIDEMIOL 2020; 23: E200067
In all of the analyses, the sample design effect and the weighting for the analysis of surveys based on complex designs were considered, using the software Stata 14 (StataCorp LP, College Station, United States).
All participants or respective tutors signed the Informed Consent Form, which explained the study objectives and the information that would be requested; the confidentiality of the information was guaranteed. The research protocol was approved by the Ethics Committee from the Epidemiology Department of the School of Public Health, at Universidade de São Paulo -Report n. 1.420.473.
Characteristic of the interviewees: 50.9% were male, and 49.1% were female, therefore, the proportion was similar. Besides, 45% reported having white skin, and 54.1% had completed elementary school ( Table 1).
Of the univariate analysis, sex and schooling (Table 2), nutritional status, alcohol consumption, health status characterization, sinusitis, back pain, vision impairment and CMD were selected (Table 3).
In the univariate logistic regression, the variables associated with cephalalgia were: being female, having incomplete elementary school, being obese, having CMD, vision impairment, back pain and sinusitis.

REV BRAS EPIDEMIOL 2020; 23: E200067
To verify the predictive capacity of the logistic regression model, the Hosmer-Lemeshow test was used and indicated that adolescents present with 99.6% of chances to have cephalalgia at the presence of these factors.

DISCUSSION
The estimated prevalence of cephalalgia, of 38.2% (95%CI 33.8 -42.7) in adolescents (aged from 15 to 19 years), in the city of São Paulo, is lower than the global estimation. Abu-Arafeh et al. 1 , in a systematic review, presented the prevalence of cephalalgia among children and adolescents (aged from 0 to 20 years) as 58.3% (95%CI 58.1 -58.8). Wöber-Bingöl 12 , also in a systematic review from carried out from 1990 to 2013, with four studies in Brazil, estimates the prevalence of 54.4% (95%CI 43.1 -65.8). One of the reasons why cephalalgia as a bit below expected is that the question in ISA Capital 2015 referred to the usual cephalalgia, understood as a frequent symptom. In some studies, however, the question was different, mentioning cephalalgia without any further specifications, or not mentioning the exact question 12 .
Wöber-Bingöl 12 estimates the prevalence of migraine of 9.1% (95%CI 7.1 -11.1) among children and adolescents. In this study, we found the estimated prevalence of migraine of 7.8% (95%CI 5.6 -10.80). The result is similar to the global estimation, which may have occurred because this diagnosis is usually medical.
In the systematic review by Abu-Arafeh 1 , women were more prone to having a headache (OR = 1.5, 95%CI 1.4 -1.6) than men; this increased chance can also be confirmed in other studies, carried out by different researchers, at different age groups [13][14][15] . In this study, with adolescents from the city of São Paulo, in 2015, the estimated presence of cephalalgia was also higher among women in relation to men (OR = 2.1; 95%CI 1.4 -3.2). Among the several deductions regarding this association, the literature highlights the report of the relation with female sexual hormones, which seem to be one of the main elements that explain the differences of cephalalgia between genders 14,15 . Even though there are several studies approaching cephalalgia among students, we did not identify studies in which schooling is an associated factor. In a study by Barros et al. 16 , there was higher prevalence of chronic conditions among those with lower schooling. In this study, the estimated prevalence of cephalalgia among adolescents in the city of São Paulo with incomplete elementary school was 52.8% (95%CI 40-8 -64.4), which may indicate a relation with the vulnerability and social exclusion factors. The relation between cephalalgia and back pain was identified in studies by the World Health Organization (WHO), besides others 6,17,18 . This association is still little understood, being mainly related to stress and musculoskeletal conditions, especially in the cervical region. Ashina et al. 18 indicate that the relation between back pain and cephalalgia was OR = 2.1, which is the same result as the one found in this study.
Jain et al. 19 did not identify an association between refraction correction in the glasses and reduced cephalalgia. Another study showed that the latter is more associated to the use of electronic devices, such as computers and cell phones, than to refraction errors 20 . In this study, there was an association between vision impairment and cephalalgia, with OR = 2.6 (95%CI 1.5 -3.9).
The isolated identification of CMD factors, such as depression 21,22 , insomnia 23 and anxiety 24 , is common in studies in which they are associated with cephalalgia. In this study, we used the SRQ-20 questionnaire, which qualifies psychic suffering. The association between suspicion of CMD and cephalalgia in our study was significant, with OR = 4.3 (95%CI 2.7 -7.0).
One of the study limitations was memory bias, because, when questioned, the interviewees may have forgotten to mention an event. There may be potential for verification bias, with the possibility of incorrect classification regarding an individual, a number or an attribute in a different category than the one it should be attributed to. The household selection did not include the population living in census sectors located in the rural area, the homeless population and institutionalized patients. The choice of specific estimations for significant associations in the simple and multiple analyses can show the overestimation in the odds ratio, in comparison to prevalence ratios, but it is known that both measurements of association can be used 25 .
95%CI: 95% confidence interval; CMD: common mental disorder. Six factors (being female, having CMD, back pain, vision impairment and sinusitis, and having incomplete elementary school) associated with cephalalgia allow to use them as predictors of this condition. The 99.5% chance of an adolescent presenting with cephalalgia in the presence of these factors can support clinical decisions and the control of cases, especially in primary care.
It is important to highlight that diagnosed and poorly treated conditions in adolescence have great potential of aggravation in adulthood.

CONCLUSION
The prevalence of cephalalgia in adolescents in the city of São Paulo in 2015 was 38.2%, representing more than 1/3 (one third) of this population. The associated factors were being female, having low schooling, having CMD, vision impairment, back pain and sinusitis. Understanding cephalalgia as a public health issue makes us think of ways to interpret its origins, associated factors and coping strategies, which may influence new ways to prioritize and organize health care.