Early childhood caries: Are maternal psychosocial factors, decision-making ability, and caries status risk indicators for children in a sub-urban Nigerian population?

Objective: Early childhood caries (ECC) is caries in children below the age of 72 months. The aim of the study was to determine the association of maternal psychosocial factors (general anxiety, dental anxiety, sense of coherence, parenting stress, fatalism, social support, depressive symptoms, and executive dysfunction), decision-making abilities, education, income and caries status with the prevalence and severity of ECC among children resident in Ile-Ife, Nigeria. Methods: A dataset of 1549 mother–child (6–71-months-old) dyads collected through examinations and a household survey, using validated psychometric tools to measure the psychosocial factors, were analyzed. The DMFT for the mothers and the dmft for the child were determined. The association between maternal psychosocial factors, education, income, and decision-making ability, the prevalence of maternal caries, and the prevalence of ECC was determined using logistic regression analysis. Results: The prevalence of maternal caries was 3.3%, and the mean (standard deviation-SD) DMFT was 0.10 (0.76). The ECC prevalence was 4.3%, and the mean (SD) dmft was 0.13 (0.92). There was no significant difference between the prevalence and severity of maternal caries and ECC by maternal age, education, income, or decision-making abilities. There was also no significant difference in maternal caries, ECC prevalence and ECC severity by maternal psychosocial factors. The only significant association was between the prevalence of caries in the mother and children: children whose mothers had caries were over six times more likely to have ECC than were children with mothers who had no caries (AOR: 6.67; 95% CI 3.23–13.79; p < 0.001). Conclusion: The significant association between ECC and maternal caries prevalence suggests that prenatal oral health care for mothers may reduce the risk for ECC. This was a secondary analysis of a dataset of 1549 mother–child dyads collected through a household survey and dental examination between December 2018 and January 2019 to determine the association between maternal psychosocial factors and ECC. The sample recruited children 6–71 months old who were living with their biological parents or legal guardians at Ife Central Local Government Area of Osun State, Nigeria and who were both at home at the time the data was collected. Using a multi-stage sampling technique. Children were excluded from study participation if they had medical conditions that could increase the risk of ECC such as HIV infection. Study participants were recruited through a multi-stage sampling technique which required sampling 10% of the 700 enumeration centers in the Local Government Area; sampling every other household on each street in the eligible enumeration area; and sampling eligible child and mother dyad in each household. The study’s sampling and recruitment methods have been reported in detail prior studies [33, 34]. Maternal data obtained included mother’s age at last birthday (≤ 29 years, 30–39 years, and ≥ 40 years), education (no formal education, primary school only, secondary school only, or post-secondary school), and income (≤ 18,000 ($50) per month, 18,001–30,000 ($84) per month, 30,001–60,000 ($168) per month, and > 60,000 per month) [35]. The age of the children was recorded in months and translated into years during the data analysis. The age was dichotomized into 0–2-years-old and 3–5-years-old based on evidence that suggests different ECC risk profiles for the two age groups [36, 37]. Maternal decision-making ability was determined by a ‘yes’ or ‘no’ response to three questions: Which person usually decides on her healthcare? Which person usually decides on large household purchases? Which person usually decides on visits to family or relatives? [38]. The 7-item Generalized Anxiety Disorder-7 scale [39], validated for use in Nigeria [40], was used to measure general anxiety. The Generalized Anxiety Disorder-7 score is calculated by assigning scores of 0–3 to the response categories, ranging from ‘not at all’, to ‘nearly every day.’ Overall scores of 5, 10, and 15 are the cut-off points for mild, moderate, and severe anxiety, respectively. This was measured with the Modified Dental Anxiety Scale [41], validated for use in Nigeria [42]. The scale is calculated by assigning scores of 1–5 to the response to a 5-item Likert scale questionnaire ranging from ‘not anxious’ to ‘extremely anxious.’ Overall scores range from 5–25. Scores 19 and above indicate high dental anxiety, while scores lower than 19 indicate low dental anxiety. Six items of the 19-item Parenting Stress Index [43], used in the Detroit Dental Health Project [44] and validated for use in the Nigerian population [45], were used to measure maternal stress. Possible scores for each item ranged from 1 to 5 for responses from ‘never’ to ‘almost.’ Total scores range from 6 to 30, with higher scores reflecting more frequent experiences of stress due to parenting role [46]. Scores below the 15th percentile are regarded as low stress; 16th to 80th percentile as normal stress; 81st to 84th percentile as borderline stress; and higher than the 84th percentile as high stress. For logistic regression analysis, low and normal stress levels were combined, as were borderline and high stress levels. This was measured with the Sense of Coherence-13 scale [47, 48] on a 7-point Likert scale adapted for use in Nigeria [49]. Possible scores ranged from 7 to 9, with higher scores indicative of a better sense of coherence. The scores were divided into lower-than-median (low coherence) and equal-to and greater-than-median (high coherence). The median score was 67. The 20-item Centre for Epidemiologic Studies and Depression Scale, developed by Radloff [50] and validated for use in Nigeria [51], was used to determine the level of depressive symptoms. Each item in the scale was assigned scores of 0–3, depending on the frequency of symptoms per week, with the total score ranging from 0 to 60. Scores of less than 15 indicate no depressive symptoms; 15–21 indicate mild to moderate depressive symptoms; and 21–60 indicate major depressive symptoms. For logistic regression analysis, the scores were dichotomized into normal and mild/moderate/major depressive symptoms. This was assessed with the 20-item Modified Dysexecutive questionnaire [52, 53], validated for use in Nigeria [54]. Each of the 20 statements was assigned a score of 0–4 on a 5-point scale ranging, from ‘Never’ to ‘Very often.’ Possible scores ranged from 0–80 with higher scores suggesting worsening dysfunction. For this study, executive dysfunction was dichotomized into low executive dysfunction (scores lower than the median) and high executive dysfunction (scores higher than or equal to the median). The median score was 32. The 12-item 7-point Multidimensional Scale of Perceived Social Support scale [55], validated for use in Nigeria [56], was used to measure perceived emotional support from family, friends, and significant others. Scores of 1.0–2.9 indicate low support; 3.0–5.0 indicate moderate support; and 5.1–7.0 indicate high support. For the logistic regression analysis, the scores were dichotomized to low/moderate and high support. The modified Powe Fatalism Inventory, which measures maternal fatalistic beliefs about ECC [57] and was validated for use in Nigeria [54], was used to measure perception of fatalism. Responses to the nine questions ranged from ‘strongly agree’ to ‘strongly disagree’ on a Likert-like scale with scores from 5–1, respectively. The scores were dichotomized to low (scores lower than the median) and high (scores higher than or equal to the median) fatalism around the median score of 25. Five calibrated dentists with intra-examiner agreement and inter-examiner agreement Cohen’s kappa coefficient values higher than 0.80 determined the maternal and child caries status using the Decayed-Missing-Filled teeth (DMFT) index and the decayed-missing-filled teeth (dmft) index, respectively using the World Health Organisation’s criteria [58]. Caries was present if the DMFT/dmft scores were greater than 0 and absent if the DMFT/dmft scores were 0. Children were examined either sitting on their mother’s lap or on a chair, under natural light, with plain dental mirrors. Mothers were also examined after their child, sitting on a chair under natural light, with plain dental mirrors. For both mother and the child, the teeth were not dried before examination, but gross debris was cleared with gauze where necessary. The examination of the teeth was done in an orderly manner from one tooth or tooth space to the adjacent tooth or tooth space. The association of maternal and ECC prevalence with maternal sociodemographic characteristics (age, education and income), psychosocial status (general anxiety, dental anxiety, sense of coherence, parenting stress, fatalism, social support, depressive symptoms, and executive dysfunction) and decision-making factors was determined with Chi-square and Fisher’s Exact tests. Analysis of variance and independent sample t-test were used to compare the maternal mean DMFT and the children’s mean dmft values among and between categories of the independent factors. Multivariable binary logistic regression analysis was used to determine the maternal factors associated with the prevalence of ECC using a series of models in line with the hierarchical theoretical model proposed by Nunes et al. [59] for determining the risk indicators for ECC, was used for developing the regression analysis for the study. Model 1 included maternal age, income and educational status and protective psychosocial factors (sense of coherence and social support). Model 2 comprised maternal age, income and educational status, and psychosocial risk factors (fatalistic belief, dental anxiety, general anxiety, depressive symptoms and executive dysfunction). Model 3 contained maternal age, income and educational status, and decision-making on health access. Model 4 comprised maternal age, income and educational status, and maternal caries. Lastly Model 5 contained all factors with p ≤ 0.20 from Models 1–4. The adjusted odds ratios (AORs) and their 95% confidence intervals were calculated. Multicollinearity diagnostics, with correlation matrix of coefficients, were used with each logistic model to identify independent variables that were strongly correlated (r > 0.5); in Model 3, two decision-making variables—decisions on household purchase and family visit—were dropped from the model because of multicollinearity. Statistical analyses were conducted with Stata/SE 14.0 for Windows (2015). Statistical significance was inferred at p ≤ 0.05.