ABSTRACT

OBJECTIVE:

To describe the frequency of behavioral problems and the internal consistency of the parent version of the Strengths and Difficulties Questionnaire (SDQ-P) in Amazonian preschool children during the covid-19 pandemic.

METHODS:

Data from the Maternal and Child Health and Nutrition in Acre (MINA-Brazil) study, a population-based birth cohort in the Western Brazilian Amazon, were used. The SDQ-P was applied in 2021 at the five-year follow-up visit to parents or caregivers of 695 children (49.4% of which were girls). This instrument is a short behavioral screening questionnaire composed of 25 items reorganized into five subscales: emotional symptoms, conduct problems, hyperactivity/inattention, peer relationship problems, and prosocial behavior. Cases of behavioral problems were defined according to the original SDQ cut-offs based on United Kingdom norms. Moreover, cut off points were estimated based on the SDQ-P percentile results of our study sample. Internal consistency was assessed by calculating Cronbach's alpha coefficient and McDonald's omega for each scale.

RESULTS:

According to the cut-offs based on our studied population distribution, 10% of all children had high or very high total difficulty scores, whereas it was almost twice when the original SDQ cut-offs based on United Kingdom norms, were applied (18%). Differences were also observed in the other scales. Compared to girls, boys showed higher means of externalizing problem and lower means of prosocial behavior. The five-factor model showed a moderate internal consistency of the items for all scales (0.60 ≤ α ≤ 0.40), except for total difficulty scores, which it considered substantial (α > 0.61).

CONCLUSIONS:

Our results support the usefulness of SDQ in our study population and reinforce the need for strategies and policy development for mental health care in early life in the Amazon.

DESCRIPTORS:
Psychometrics; Problem Behavior; Child Behavior Disorders; Behavior Rating Scale

INTRODUCTION

Estimates suggest that about 20% of all children in the world under the age of seven years experience some type of mental disorder¹1 Vasileva M, Graf RK, Reinelt T, Petermann U, Petermann F. Research review: a meta-analysis of the international prevalence and comorbidity of mental disorders in children between 1 and 7 years. J Child Psychol Psychiatry. 2021 Apr;62(4):372-81. https://doi.org/10.1111/jcpp.13261
https://doi.org/10.1111/jcpp.13261... . Thus, addressing mental health and precocious psychosocial development of children concerns both their chance of achieving a fulfilling life and the formation of human capital. Early childhood represents a particularly critical and vulnerable period as children in their early years of life respond to traumas and stress or express their emotions and feelings in specific ways unlike those of older children¹1 Vasileva M, Graf RK, Reinelt T, Petermann U, Petermann F. Research review: a meta-analysis of the international prevalence and comorbidity of mental disorders in children between 1 and 7 years. J Child Psychol Psychiatry. 2021 Apr;62(4):372-81. https://doi.org/10.1111/jcpp.13261
https://doi.org/10.1111/jcpp.13261... ,²2 Gil JD, Ewerling F, Ferreira LZ, Barros AJ. Early childhood suspected developmental delay in 63 low- and middle-income countries: large within- and between-country inequalities documented using national health surveys. J Glob Health. 2020 Jun;10(1):010427. https://doi.org/10.7189/jogh.10.010427
https://doi.org/10.7189/jogh.10.010427... .

The frequency of mental health disorders during childhood varies between countries. Data from children aged between 36 and 59 months from 63 countries showed that the frequency of suspected social-emotional delay is heterogeneous, ranging from 20% in high-income countries to 18%, 25%, and 32% in upper middle-, lower middle-, and low-income countries, respectively²2 Gil JD, Ewerling F, Ferreira LZ, Barros AJ. Early childhood suspected developmental delay in 63 low- and middle-income countries: large within- and between-country inequalities documented using national health surveys. J Glob Health. 2020 Jun;10(1):010427. https://doi.org/10.7189/jogh.10.010427
https://doi.org/10.7189/jogh.10.010427... .

The Strengths and Difficulties Questionnaire (SDQ) aims to track behavioral problems in children aged 2–17 years³3 Goodman R. The strengths and difficulties questionnaire: a research note. J Child Psychol Psychiatry. 1997 Jul;38(5):581-6. https://doi.org/10.1111/j.1469-7610.1997.tb01545.x
https://doi.org/10.1111/j.1469-7610.1997... . It is available in more than 40 languages and in three versions: for teachers (SDQ-T), parents/guardians (SDQ-P), and self-report (SDQ-S). It includes 25 questions about children's behaviors or abilities that are organized into five scales: emotional symptoms, conduct problems, hyperactivity, peer problems, and prosocial behavior.

SDQ results can be explored in two ways: as raw scores or categories, with cut-off points proposed by the authors of the instrument. Bands were established according to a large population-based study in the United Kingdom (UK), in which 80% of the children were classified as "close to average"; 10%, as "slightly raised"; 5%, as "high"; and 5%, as "very high" regarding behavioral problems in each scale (except for the prosocial scale, which is scored in reverse). However, considering the socioeconomic and cultural differences across countries, studies outside Europe suggested that bandings based on each sample distribution should be considered for comparisons⁴4 de Vries PJ, Davids EL, Mathews C, Aarø LE. Measuring adolescent mental health around the globe: psychometric properties of the self-report Strengths and Difficulties Questionnaire in South Africa, and comparison with UK, Australian and Chinese data. Epidemiol Psychiatr Sci. 2018 Aug;27(4):369-80. https://doi.org/10.1017/S2045796016001207
https://doi.org/10.1017/S204579601600120... .

More recently, as an alternative to the aforementioned five-scale model, a new three-factor solution has been proposed for low-risk or general population samples⁵5 Goodman A, Lamping DL, Ploubidis GB. When to use broader internalising and externalising subscales instead of the hypothesised five subscales on the Strengths and Difficulties Questionnaire (SDQ): data from British parents, teachers and children. J Abnorm Child Psychol. 2010 Nov;38(8):1179-91. https://doi.org/10.1007/s10802-010-9434-x
https://doi.org/10.1007/s10802-010-9434-... . This model reorganizes SDQ items into three subscales: internalizing problems (the sum of emotional and peer symptoms), externalizing problems (the sum of conduct and hyperactivity symptoms), and prosocial behavior⁵5 Goodman A, Lamping DL, Ploubidis GB. When to use broader internalising and externalising subscales instead of the hypothesised five subscales on the Strengths and Difficulties Questionnaire (SDQ): data from British parents, teachers and children. J Abnorm Child Psychol. 2010 Nov;38(8):1179-91. https://doi.org/10.1007/s10802-010-9434-x
https://doi.org/10.1007/s10802-010-9434-... .

The SDQ was validated in Brazil using a two-phase assessment in different local contexts to evaluate its retest reliability⁶6 Fleitlich-Bilyk B, Goodman R. Prevalence of child and adolescent psychiatric disorders in southeast Brazil. J Am Acad Child Adolesc Psychiatry. 2004 Jun;43(6):727-34. https://doi.org/10.1097/01.chi.0000120021.14101.ca
https://doi.org/10.1097/01.chi.000012002... –⁸8 Fleitlich B, Cortázar PG, Goodman R. Strengths and Difficulties Questionnaire (SDQ). Revista Infanto. 2000;8(1):44-50.. It is easy to implement and, when prior training is provided to users, a specialized professional is not required⁸8 Fleitlich B, Cortázar PG, Goodman R. Strengths and Difficulties Questionnaire (SDQ). Revista Infanto. 2000;8(1):44-50.. Furthermore, as the scientific community extensively uses the SDQ, results can be compared between national and international studies⁹9 Azevêdo PVB, Caixeta LF, Taveira DL, Giglio MR, Rosário MC, Rohde LA. Suggestive diagnosis of attention-deficit/hyperactivity disorder in indigenous children and adolescents from the Brazilian Amazon. Eur Child Adolesc Psychiatry. 2020 Mar;29(3):373-84. https://doi.org/10.1007/s00787-019-01356-y
https://doi.org/10.1007/s00787-019-01356... ,¹⁰10 Grasso M, Lazzaro G, Demaria F, Menghini D, Vicari S. The Strengths and Difficulties Questionnaire as a valuable screening tool for identifying core symptoms and behavioural and emotional problems in children with neuropsychiatric disorders. Int J Environ Res Public Health. 2022 Jun;19(13):7731. https://doi.org/10.3390/ijerph19137731
https://doi.org/10.3390/ijerph19137731... .

Although the SDQ has been applied in different Brazilian sites, it has been scarcely used in the remote areas of the country and its internal consistency has not been assessed. In the Brazilian Amazon, children in the most socioeconomically vulnerable families were disproportionately affected by SARS-CoV-2 infections¹¹11 Ferreira MU, Giacomini I, Sato PM, Lourenço BH, Nicolete VC, Buss LF, et al.; MINA-Brazil Working Group. SARS-CoV-2 seropositivity and COVID-19 among 5 years-old Amazonian children and their association with poverty and food insecurity. PLoS Negl Trop Dis. 2022 Jul;16(7):e0010580. https://doi.org/10.1371/journal.pntd.0010580
https://doi.org/10.1371/journal.pntd.001... . The great social inequalities and vulnerabilities require data on early childhood behavior in the Amazon. Therefore, this study describes the frequency of behavior disorders in preschool children during the COVID-19 restrictions. This study assessed this using the SDQ-P scales according to its original quadruple categorization and analyzed the internal consistency of the five- and three-solution models proposed by Goodman³3 Goodman R. The strengths and difficulties questionnaire: a research note. J Child Psychol Psychiatry. 1997 Jul;38(5):581-6. https://doi.org/10.1111/j.1469-7610.1997.tb01545.x
https://doi.org/10.1111/j.1469-7610.1997... and Goodman et al.⁵5 Goodman A, Lamping DL, Ploubidis GB. When to use broader internalising and externalising subscales instead of the hypothesised five subscales on the Strengths and Difficulties Questionnaire (SDQ): data from British parents, teachers and children. J Abnorm Child Psychol. 2010 Nov;38(8):1179-91. https://doi.org/10.1007/s10802-010-9434-x
https://doi.org/10.1007/s10802-010-9434-... , respectively, in a birth cohort in the Western Brazilian Amazon.

METHODS

Study Design, Site, and Population

The Maternal and Child Health and Nutrition in Acre, Brazil (MINA-Brazil Study), is a population-based birth cohort that was conducted in the municipality of Cruzeiro do Sul, Acre State, Western Brazilian Amazon. In 2021, Cruzeiro do Sul had an estimated population of 89,760 individuals (about 70% of whom lived in urban areas)¹²12 Instituto Brasileiro de Geografia e Estatística. Cidades e Estados: Cruzeiro do Sul. [cited 2023 Mar 26]. Available from: https://ibge.gov.br/cidades-e-estados/ac/cruzeiro-do-sul.html
https://ibge.gov.br/cidades-e-estados/ac... . The municipality represents the second largest municipal gross domestic product in Acre State and has a Human Development Index of 0.664¹²12 Instituto Brasileiro de Geografia e Estatística. Cidades e Estados: Cruzeiro do Sul. [cited 2023 Mar 26]. Available from: https://ibge.gov.br/cidades-e-estados/ac/cruzeiro-do-sul.html
https://ibge.gov.br/cidades-e-estados/ac... , which is below the Brazilian national average¹³13 United Nations Development Programme. Atlas do desenvolvimento humano dos municípios. 2022 [cited 2023 Mar 3]. Available from: https://www.br.undp.org/content/brazil/pt/home/idh0/atlas-do-desenvolvimento-humano/atlas-dos-municipios.html
https://www.br.undp.org/content/brazil/p... .

The Institutional Review Board of the School of Public Health, Universidade de São Paulo, approved this study (# 872.613, 2014; # 2.358.129, 2017). Informed consent forms were obtained from all participating mothers or children's parents or guardians (if mothers were aged < 18 years).

Baseline and Follow-Up Assessments

In 2015, mother-infant pairs were enrolled in the study either in public antenatal clinics or at birth in the Women and Children's Hospital of Juruá Valley, the only maternity hospital in Cruzeiro do Sul, as described elsewhere¹⁴14 Cardoso MA, Matijasevich A, Malta MB, Lourenco BH, Gimeno SG, Ferreira MU, et al.; MINA-Brazil Study Group. Cohort profile: the maternal and child health and nutrition in Acre, Brazil, birth cohort study (MINA-Brazil). BMJ Open. 2020 Feb;10(2):e034513. https://doi.org/10.1136/bmjopen-2019-034513
https://doi.org/10.1136/bmjopen-2019-034... . At baseline, 1,246 mother-infant pairs were eligible for follow-up visits. At birth, mothers' sociodemographic information was obtained, such as self-reported skin color (white, mixed, yellow, black, or amerindian); marital status (living with a partner or not); educational attainment (years of study); and beneficiary status of the Bolsa Família Program, a conditional cash transfer program directed to families living in poverty (yes or no)¹⁵15 Neves JA, Vasconcelos FA, Machado ML, Recine E, Garcia GS, Medeiros MA. The Brazilian cash transfer program (Bolsa Família): a tool for reducing inequalities and achieving social rights in Brazil. Glob Public Health. 2022 Jan;17(1):26-42. https://doi.org/10.1080/17441692.2020.1850828
https://doi.org/10.1080/17441692.2020.18... . Obstetric and perinatal data (birth weight in grams, gestational age, and type of birth) were collected from maternity records. After the baseline assessment, follow-up visits were carried out at ages one, two, and five years (Figure).

At the five-year follow-up visit, conducted during covid-19 restrictions from 15 January to 4 July, 2021, 695 out of the 1,240 eligible children (56.0%) were enrolled¹¹11 Ferreira MU, Giacomini I, Sato PM, Lourenço BH, Nicolete VC, Buss LF, et al.; MINA-Brazil Working Group. SARS-CoV-2 seropositivity and COVID-19 among 5 years-old Amazonian children and their association with poverty and food insecurity. PLoS Negl Trop Dis. 2022 Jul;16(7):e0010580. https://doi.org/10.1371/journal.pntd.0010580
https://doi.org/10.1371/journal.pntd.001... . Overall, less than 1% of the study children have reportedly attended childcare facilities since schools closed in early April 2020. All anthropometric measurements were conducted in duplicates using calibrated equipment and standardized procedures¹⁶16 World Health Organization. Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. World Health Organ Tech Rep Ser. 1995;854:1-452.. The children were barefoot, wore light clothes, and had no hairstyles. The stadiometers (to measure their height) were accurate to 0.1 millimeters and the scales (to measure the weight) had a 150 kg capacity and were accurate to 100 g (UM061, Tanita Corporation). The maximum allowed difference between measurements was 0.2 millimeters and 100 grams for height and weight, respectively. Z-scores were used according to the World Health Organization (WHO)¹⁷17 Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ. 2007 Sep;85(9):660-7. https://doi.org/10.2471/BLT.07.043497
https://doi.org/10.2471/BLT.07.043497... to generate anthropometric indices. We adopted the WHO Anthroplus software (version 1.0.4) to assess the nutritional status of population groups using its nutritional survey module. Stunting was defined as a height-for-age z-score lower than −2 standard deviation (SD), thinness as a body mass index-for-age lower than −2 SD, overweight as a body mass index-for-age greater than +1 SD and lower than +2 SD, and obesity as a body mass index-for-age z-score greater than +2 SD.

Children participating in the 5-year follow-up visit had similar perinatal characteristics (sex, gestational age, preterm birth, and birth weight) to those lost at follow-up (n = 514). However, they differed significantly in the proportion of children from the poorest families and mothers with ≤ 9 years of schooling (36.2 versus 44.1% and 26.4 versus 47.5%, respectively). A p < 0.01 was obtained for both these factors (χ² test).

The SDQ-P was included as part of the five-year follow-up. Trained interviewers completed the Portuguese version of the SDQ-P with parents or caregivers at the time of the interview. In total, 25 questions about the child's behavioral problems or strengths on a scale of 0–2 (0 = "Not True," 1 = "Somewhat True," and 2 = "Certainly True") were responded by parents/caregivers. According to their responses, a score from 0-2 was assigned accordingly. Each of the following five scales had five items: emotional symptoms, conduct problems, hyperactivity, peer problems, and prosocial behavior. The sum of the first four scales generated the total difficulty score (20 items). Overall, the higher the score, the greater the difficulty on the scale, except for prosocial behavior, which was reverse-scored (i.e., the higher the score, the better the prosocial behavior)¹⁸18 SDQINFO. Youth in mind: Scoring the SDQ. 2022 [cited 2023 May 10]. Available from: https://www.sdqinfo.org/py/sdqinfo/c0
https://www.sdqinfo.org/py/sdqinfo/c0... .

A similar procedure was adopted for the three-solution factorial model. Internalizing problems were considered the sum of emotional and peer symptoms, whereas externalizing problems, the sum of hyperactivity symptoms⁵5 Goodman A, Lamping DL, Ploubidis GB. When to use broader internalising and externalising subscales instead of the hypothesised five subscales on the Strengths and Difficulties Questionnaire (SDQ): data from British parents, teachers and children. J Abnorm Child Psychol. 2010 Nov;38(8):1179-91. https://doi.org/10.1007/s10802-010-9434-x
https://doi.org/10.1007/s10802-010-9434-... .

Statistical Analyses

Data were collected using tablets programmed with The Census and Survey Processing System (CSPro) 218 and transferred to Stata 17.0 (StataCorp, College Station, TX, USA) for statistical analysis. Supervisors routinely checked all information and provided feedback to research team members to correct inconsistencies whenever necessary.

Socioeconomic, obstetric, and perinatal characteristics included frequencies, means, and SD. Descriptive SDQ analysis for all children were conducted, who were divided by sex, including the mean score, SD, median, and interquartile intervals (25^th and 75^th percentiles).

The proportion of children was calculated according to the original instrument cut offs based on UK norms, classifying them as "close to average", "slightly raised", "high", and "very high" problems in each scale (except for the prosocial scale which can be classified as "close to average", "slightly lowered", "low", and "very low")¹⁸18 SDQINFO. Youth in mind: Scoring the SDQ. 2022 [cited 2023 May 10]. Available from: https://www.sdqinfo.org/py/sdqinfo/c0
https://www.sdqinfo.org/py/sdqinfo/c0... . Furthermore, cut-off points were estimated based on the SDQ distribution of our study sample. A similar methodology to that originally used by the authors of the instrument was followed based on a large UK population-based survey. Therefore, the estimated cut offs were elaborated to classify 80% of the children as having "close to average"; 10%, as "slightly raised"; 5%, as "high"; and 5%, as "very high" problems in each scale (except for the prosocial scale whose scoring is reverse, i.e., 80% were classified as "close to average"; 10%, as "slightly raised"; 5%, as "low"; and 5% as "very low").

The internal consistency of all scales was assessed using Cronbach's alpha coefficients and McDonald's omega (ω). A study conducted with the SDQ-P version and Dutch children aged from four to seven years suggested that McDonald's omega (ω) may be a more suitable alternative than Cronbach's alpha (α)¹⁹19 Stone LL, Otten R, Ringlever L, Hiemstra M, Engels RC, Vermulst AA, et al. The parent version of the Strengths and Difficulties Questionnaire: omega as an alternative to alpha and a test for measurement invariance. Eur J Psychol Assess. 2013;29(1):44-50. https://doi.org/10.1027/1015-5759/a000119.
https://doi.org/10.1027/1015-5759/a00011... . However, to enable comparability with other studies, both coefficients are shown.

RESULTS

Table 1 describes the sociodemographic, anthropometric, and perinatal characteristics of the 695 mother-child pairs at the five-year follow-up, in which children were aged between 58.5 to 71.9 months (average age 63.5 months, SD 2.71). Regarding nutritional status, only 2% of the children were classified as stunted and 1% as thin but 16% were overweight and 13% were obese, according to the WHO¹⁷17 Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, Siekmann J. Development of a WHO growth reference for school-aged children and adolescents. Bull World Health Organ. 2007 Sep;85(9):660-7. https://doi.org/10.2471/BLT.07.043497
https://doi.org/10.2471/BLT.07.043497... growth parameters. Overall, 49% of the children were girls and almost half of the families were Bolsa Família Program beneficiaries (45%). At birth, 7% were born with low weight (< 2,500 g) and 9% were premature; 50% of the deliveries were caesarean. Furthermore, the average maternal age was 25.7 years, with 21% of adolescents (< 19 years old). Most mothers self-declared non-white (88%), some lived in a different home than their partners (21%), had a low level of schooling (27% attended school for nine years or less), and were Bolsa Família Program beneficiaries (36%) at baseline.

Table 2 shows the distribution of "caseness" of SDQ data in the MINA-Brazil population according to the original score banding ("close to average", "slightly raised/lowered", "high/low", and "very high/very low"). This study observed frequency differences by comparing the proportion of children with behavior problems according to the original cut offs and those based on our study population distribution (Table 2). The behavior problem scale showed that the highest proportion of children was classified as having high or very high problems (UK norms, 25% versus MINA-Brazil, 11%); followed by emotional problems (UK norms 23% versus MINA, 7%); peer problems (UK norms 15% versus MINA, 7%); and hyperactivity (UK norms 13% versus MINA, 6%). Regarding prosocial behavior scale, 13% of children were classified as low or very low according to the UK norms and only 6% according to the MINA band. For the total difficulties scale, the proportion of children in the high or very high category totaled 10% when using the MINA banding, and almost twice (18%) when using the UK norms.

Table 3 shows the mean, SD, median and interquartile ranges, Cronbach's alpha, and McDonald's omega value for each scale considering the five-factor SDQ solution. Cronbach's alpha and McDonald's omega values indicated that the scales had moderate item internal consistency (0.60 ≤ α ≤ 0.40), except for the total difficulty score, which was considered substantial (α > 0.61). This research found similar results for McDonald's omega values, which ranged from 0.43-0.63.

Table 4 shows the mean, SD, median and interquartile ranges, Cronbach's alpha, and McDonald's omega values according to sex for each scale considering the three-factor SDQ solution. Cronbach's alpha showed that the internal consistency of the items was substantial for the externalizing problems scale (α = 0.69, ω = 0.70), and moderate for internalizing problems scale (α = 0.53, ω = 0.54).

Among the four scales, which comprised the total difficulties score, the hyperactivity scale had the highest mean, whereas the problems with peers scale had the lowest. Boys had higher means for the externalizing (mean: boys = 6.59, girls = 6.03), conduct (mean: boys = 2.48, girls = 2.08), and hyperactivity problems (mean: boys = 4.67, girls = 3.95) scales. Additionally, boys had a lower mean prosocial behavior (mean: boys = 8.32, girls = 8.89). Internalizing problems showed no differences between sexes.

DISCUSSION

Using the five-factor structure and original SDQ cut offs, we found that 25% of the children were considered as having high or very high problems as per the conduct scale, 23% as per the emotional scale, 15% as per the peer problems scale, 13% as per the hyperactivity scale, and 13% (low or very low) as per the prosocial behavior scale. The total difficulties scale considered 18% of the children as having high or very high problems. However, numbers greatly differed from the percentages when using cut offs based on study sample percentiles. Furthermore, our results showed that the SDQ scales had modest or substantial internal consistency in Amazonian pre-school children.

As expected, our results differed from studies conducted with children in other countries using the same cut off values. A Chinese study with 950 pre-schoolers aged 3-5 years found that 39% of participants had emotional problems; 27%, conduct problems; and 23%, hyperactivity problems²⁰20 Li S, Chen K, Liu C, Bi J, He Z, Luo R, et al. Dietary diversity and mental health in preschoolers in rural China. Public Health Nutr. 2021 May;24(7):1869-76. https://doi.org/10.1017/S1368980020003237
https://doi.org/10.1017/S136898002000323... , all higher than our results. However, a study with 800 Pakistani children aged 6-16 years showed that 23% of their sample had emotional problems; 27%, conduct problems; 11%, hyperactivity; 13%, peer problems; 3% low or very low prosocial behavior; and 16%, total behavioral difficulties²¹21 Malik TA, Siddiqui S, Mahmood A. Behavioural and emotional problems among school children in Pakistan: a telephonic survey for prevalence and risk factors. J Paediatr Child Health. 2019 Dec;55(12):1414-23. https://doi.org/10.1111/jpc.14429
https://doi.org/10.1111/jpc.14429... .

Overall, when comparing our results with those of other studies conducted in Brazil before the covid-19 pandemic, MINA-Brazil cohort participants showed greater behavioral disorders. Data from a Brazilian south cohort with 3,718 children showed that, at six years of age, 14.7 (16.1% boys; 13.1% girls) and 13.5% (13.0% boys; 14.0% girls) had high or very high levels of conduct and emotional problems, respectively²²22 Bauer A, Fairchild G, Halligan SL, Hammerton G, Murray J, Santos IS, et al. Harsh parenting and child conduct and emotional problems: parent- and child-effects in the 2004 Pelotas Birth Cohort. Eur Child Adolesc Psychiatry. 2022 Aug;31(8):1-11. https://doi.org/10.1007/s00787-021-01759-w
https://doi.org/10.1007/s00787-021-01759... . Another study conducted in southern Brazil with 152 school children aged 6-13 years observed higher behavior problems in children living near recycling sites exposed to urban waste than to those living more than 150 meters away²³23 Dumcke TS, Benedetti A, Selistre LD, Camardelo AM, Silva ER. Association between exposure to urban waste and emotional and behavioral difficulties in schoolchildren. J Pediatr (Rio J). 2020;96(3):364-70. https://doi.org/10.1016/j.jped.2018.11.014
https://doi.org/10.1016/j.jped.2018.11.0... . In that study, the authors defined ‘caseness’ as children on the borderline and abnormal problems categories (equivalent to the sum of "slightly raised," "high," and "very high" classifications in this study). Caseness frequency differed between children for total difficulties (24% in the exposed group vs. 10% in the control group) and was lower than MINA-Brazil data, which was 33.1% based on the same cut-offs. However, for some of the other SDQ subscales, our results were lower than those for the control group (i.e., hyperactivity and peer problems subscales).

When using a broader internalizing and externalizing model, we also observed differences between our results and those of other Brazilian studies before the covid-19 pandemic. Data from a cross-sectional study in southeastern Brazil with 825 children aged six to 11 years showed a higher frequency of internalizing problems (30.7%) than externalizing problems (18.3%)²⁴24 Fatori D, Brentani A, Grisi SJ, Miguel EC, Graeff-Martins AS. [Childhood mental health problems in primary care]. Cien Saude Colet. 2018 Sep;23(9):3013-20. Portuguese. https://doi.org/10.1590/1413-81232018239.25332016
https://doi.org/10.1590/1413-81232018239... . This contrasted with our results (16 and 17% for internalizing and externalizing problems, respectively, in the MINA-Brazil sample).

In this study, the SDQ was notably applied during the covid-19 pandemic, which has directly or indirectly caused unprecedented harm to the entire population, but especially to early childhood as it represents a vulnerable period in people's lives. Although children are not considered a group risk for SARS-CoV-2 infection, the pandemic worsened compromising factors for child development, such as parental stress, suspension of school activities, social isolation, lack of physical activity, and food insecurity, particularly in unstructured contexts²⁵25 Araújo LA, Veloso CF, Souza MC, Azevedo JM, Tarro G. The potential impact of the COVID-19 pandemic on child growth and development: a systematic review. J Pediatr (Rio J). 2021;97(4):369-77. https://doi.org/10.1016/j.jped.2020.08.008
https://doi.org/10.1016/j.jped.2020.08.0... . The covid-19 crisis in Brazil has been most dramatic across the Amazon, in which public health facilities were already operating near full capacity before the pandemic¹¹11 Ferreira MU, Giacomini I, Sato PM, Lourenço BH, Nicolete VC, Buss LF, et al.; MINA-Brazil Working Group. SARS-CoV-2 seropositivity and COVID-19 among 5 years-old Amazonian children and their association with poverty and food insecurity. PLoS Negl Trop Dis. 2022 Jul;16(7):e0010580. https://doi.org/10.1371/journal.pntd.0010580
https://doi.org/10.1371/journal.pntd.001... . An Italian study with 463 children aged 5 to 17 years showed that confinement measures and changes in daily routine during the covid-19 pandemic negatively affected parents' psychological dimensions, thus exposing children to a significant risk to their well-being²⁶26 Cusinato M, Iannattone S, Spoto A, Poli M, Moretti C, Gatta M, et al. Stress, resilience, and well-being in italian children and their parents during the COVID-19 pandemic. Int J Environ Res Public Health. 2020 Nov;17(22):8297. https://doi.org/10.3390/ijerph17228297
https://doi.org/10.3390/ijerph17228297... . This may explain the higher frequency of behavior problems in our study than expected according to other national and international studies.

We observed frequency differences when comparing the proportion of children with behavioral problems according to the original SDQ cut offs and those elaborated in this study based on our population distribution. In agreement, recent studies have suggested that using European norms in different contexts may fail to reflect local realities. Data from different continents support that, for clinical purposes, ‘caseness’ should be defined by in-country cut offs⁴4 de Vries PJ, Davids EL, Mathews C, Aarø LE. Measuring adolescent mental health around the globe: psychometric properties of the self-report Strengths and Difficulties Questionnaire in South Africa, and comparison with UK, Australian and Chinese data. Epidemiol Psychiatr Sci. 2018 Aug;27(4):369-80. https://doi.org/10.1017/S2045796016001207
https://doi.org/10.1017/S204579601600120... .

Based on eight population-based studies from seven countries (Bangladesh, Brazil, Britain, India, Norway, Russia, and Yemen) with data from more than 29,225 children aged 5-16 years, the authors concluded that SDQ cross-country differences fail to necessarily reflect comparable differences in disorder rates²⁷27 Goodman A, Heiervang E, Fleitlich-Bilyk B, Alyahri A, Patel V, Mullick MS, et al. Cross-national differences in questionnaires do not necessarily reflect comparable differences in disorder prevalence. Soc Psychiatry Psychiatr Epidemiol. 2012 Aug;47(8):1321-31. https://doi.org/10.1007/s00127-011-0440-2
https://doi.org/10.1007/s00127-011-0440-... . Therefore, interpreting frequencies and other results from cross-cultural comparisons of mental health requires caution, especially when assessments rely on brief questionnaires such as the SDQ²⁷27 Goodman A, Heiervang E, Fleitlich-Bilyk B, Alyahri A, Patel V, Mullick MS, et al. Cross-national differences in questionnaires do not necessarily reflect comparable differences in disorder prevalence. Soc Psychiatry Psychiatr Epidemiol. 2012 Aug;47(8):1321-31. https://doi.org/10.1007/s00127-011-0440-2
https://doi.org/10.1007/s00127-011-0440-... .

To investigate the measure of a construct by multiple items, many studies on social and behavioral problems have adopted Cronbach's alpha, including those using the SDQ-P. However, Cronbach's alpha may underestimate the true reliability of a construct. Therefore, McDonald's omega has been increasingly adopted and reported²⁸28 Deng L, Chan W. Testing the difference between reliability coefficients alpha and omega. Educ Psychol Meas. 2017 Apr;77(2):185-203. https://doi.org/10.1177/0013164416658325
https://doi.org/10.1177/0013164416658325... . Although many SDQ studies have found no significant differences between the two coefficients, some authors recommend using McDonald's omega¹⁹19 Stone LL, Otten R, Ringlever L, Hiemstra M, Engels RC, Vermulst AA, et al. The parent version of the Strengths and Difficulties Questionnaire: omega as an alternative to alpha and a test for measurement invariance. Eur J Psychol Assess. 2013;29(1):44-50. https://doi.org/10.1027/1015-5759/a000119.
https://doi.org/10.1027/1015-5759/a00011... . Our results concurred with previous studies. We observed higher McDonald omega coefficients than Cronbach's alpha coefficients for most SDQ scales, including when stratified by sex.

Our internal consistency analysis supported the potential usefulness of SDQ in the setting of this study. Results showed moderate internal reliability for all scales (0.60 ≤ α ≤ 0.40), except for total difficulty scores, which was substantial (α > 0.61). A study on Brazilian children aged 6 and 11 years found similar results: Cronbach's alphas ranged from 0.59-0.65 and 0.52-0.59 for conduct and emotional problems subscales, respectively²²22 Bauer A, Fairchild G, Halligan SL, Hammerton G, Murray J, Santos IS, et al. Harsh parenting and child conduct and emotional problems: parent- and child-effects in the 2004 Pelotas Birth Cohort. Eur Child Adolesc Psychiatry. 2022 Aug;31(8):1-11. https://doi.org/10.1007/s00787-021-01759-w
https://doi.org/10.1007/s00787-021-01759... .

In concurrence with other studies, our analysis showed that the peer problems scale had the worst internal consistency. A review of 26 studies showed that the peer problems scale items might fail to reflect the same construct as its alphas were the lowest for the SDQ²⁹29 Kersten P, Czuba K, McPherson K, Dudley M, Elder H, Tauroa R, et al. A systematic review of evidence for the psychometric properties of the Strengths and Difficulties Questionnaire. Int J Behav Dev. 2015;40(1):64-75. https://doi.org/10.1177/0165025415570647
https://doi.org/10.1177/0165025415570647... .

Strengths and Limitations

This study makes significant contributions to mental health assessments in early childhood in the Amazon during the covid-19 restrictions. It helps to comprehend the limitations and potential of the SDQ, a low-cost and easily applicable screening tool. It is worth mentioning that since data about behavior problems during childhood in the Brazilian north are rare, comparisons require careful interpretations. We also analyzed data from a large population-based birth cohort using validated instruments that attributed credibility to our findings. However, some important limitations of this study require consideration. We estimated SDQ-P cut offs based on the distribution of our own study population, which includes a narrow age range (58.5 to 71.9 months) and has very unique socioeconomic and cultural features. Thus, these results should not be generalized to other populations. Also, although our internal consistency analyses supports the usefulness of SDQ in our population, we recognize the existence of different possible analysis to explore other important psychometric characteristics of an instrument³⁰30 Vet HC, Mokkink LB, Mosmuller DG, Terwee CB. Spearman-Brown prophecy formula and Cronbach's alpha: different faces of reliability and opportunities for new applications. J Clin Epidemiol. 2017 May;85:45-9. https://doi.org/10.1016/j.jclinepi.2017.01.013
https://doi.org/10.1016/j.jclinepi.2017.... , which was not the focus of this study. Moreover, from the original MINA-Brazil participants enrolled at birth (n =1,240), the five-year assessment had a 56% retention rate (n = 695). Families of children lost to follow-up had a lower socioeconomic status and maternal education than those who attended the five-year follow-up. Thus, as the literature considers that a socially vulnerable background correlates with poor child development, the frequency of behavior disorders reported in this study could be even higher.

CONCLUSION

Our results regarding child behavior problems reinforce the need for mental health care in early life in the Amazon region. Furthermore, our internal consistency results agreed with those of other national studies, supporting the usefulness of the SDQ. However, interpreting our results and comparing them with those of other studies requires caution due to the cultural and contextual diversity across different settings.

Acknowledgments

We thank all women and children who have taken part in the MINA-Brazil Study and health professionals at the Maternity Hospital, Municipal Health Secretariat, and primary health care units of Cruzeiro do Sul. Members of the MINA-Brazil Working Group: Marly Augusto Cardoso (PI), Alicia Matijasevich, Bárbara Hatzlhoffer Lourenço, Jenny Abanto, Maíra Barreto Malta, Marcelo Urbano Ferreira, Paulo Augusto Ribeiro Neves (Universidade de São Paulo, São Paulo, Brazil); Ana Alice Damasceno, Bruno Pereira da Silva, Rodrigo Medeiros de Souza (Universidade Federal do Acre, Cruzeiro do Sul, Brazil); Simone Ladeia-Andrade (Instituto Oswaldo Cruz, Fiocruz, Rio de Janeiro, Brazil), Marcia Caldas de Castro (Harvard T.H. Chan School of Public Health, Boston, USA).

REFERENCES

Publication Dates

History