Data for: CONGENITAL HEART DISEASE IN CHILDREN: OROFACIAL MYOFUNCTIONAL ASPECTS, EATING BEHAVIOR AND FACIAL TEMPERATURE

1. Introduction Patients with heart disease will be compared with healthy patients (control group) regarding thermographic variables, speech condition (altered or not), assessment of orofacial conditions and questions about eating behavior. Firstly, regarding the orofacial conditions questionnaire (AMIOFE), multiple linear regression models were adjusted (GOLDBERG, M.A. and CHO, H.A.) to compare the three groups: control, cyanotic heart disease and acyanotic heart disease; controlling for the variables gender and age. The models were adjusted for all questions and also for the totals by category (appearance / posture, mobility and functions) and overall total. For the 12 thermographic variables, multiple linear regression models were also adjusted to compare the three groups, controlling for age and gender. T-tests were also performed to compare the means between groups for each of the variables. Regarding food consistency, a multiple linear regression model was adjusted for the total questionnaire with explanatory variables group, age and gender. To compare the eating behavior categories, a chi-square homogeneity test (BUSSAB, Wilton, MORETTIN, P. A.) was performed to verify if the distributions of the categories are different between groups. To analyze the speech condition (altered or not), a logistic regression model (NETER, J., WASSERMAN, W., KUTNER, MH and LI, W.) was adjusted to verify whether the disease group, the Age and gender are associated with the likelihood of impaired speech. Finally, to study the associations between the different variables evaluated in the study, Pearson's linear correlations were calculated between the thermographic variables, eating behavior and the AMIOFE questionnaire totals and tests were performed to verify whether such correlations were significantly different from zero. The software used for analysis was RStudio version 3.5. 2. Analysis of the AMIOFE questionnaire In order to compare patients with acyanotic and cyanotic heart disease with patients in the control group, multiple linear regression models were adjusted for all questionnaire questions and also for totals by category and overall. In these models, gender and age were also included as explanatory variables. The intercept in all tables corresponds to the expected value of the variable for newborn boys in the control group. It is important to highlight that it was not possible to construct the model for the variables lip commissures, hard palate volume, jaw slip, food leakage, choking and uncoordinated jaw movements because these variables always presented the same value for all patients evaluated. .

1. 引言
本研究将对比心脏病患者与健康受试者（对照组）的热成像变量、言语状况（是否异常）、口面状况评估结果以及饮食行为相关问卷数据。
首先，针对口面状况问卷（AMIOFE），本研究构建了多元线性回归模型（GOLDBERG, M.A. 与 CHO, H.A.），用于对比对照组、发绀型心脏病组与非发绀型心脏病组这三组人群，并在模型中控制性别与年龄作为混杂变量。模型分别针对问卷所有条目、分类维度总分（外观/姿势、活动度与功能）以及总总分进行构建。
针对12项热成像变量，本研究同样构建了多元线性回归模型以对比三组人群，同样控制年龄与性别作为混杂变量。同时针对每一项变量，采用t检验对比组间均值差异。
针对食物质地相关问题，本研究以分组、年龄与性别作为解释变量，构建了针对问卷总分的多元线性回归模型。为对比饮食行为分类的组间差异，本研究采用卡方齐性检验（BUSSAB, Wilton, MORETTIN, P. A.）验证各组间饮食行为分类的分布是否存在差异。
为分析言语状况是否异常，本研究构建了逻辑回归模型（NETER, J.、WASSERMAN, W.、KUTNER, MH 与 LI, W.），以验证疾病分组、年龄及性别是否与言语受损风险存在关联。最后，为探究本研究中各项评估变量间的关联，本研究计算了热成像变量、饮食行为与AMIOFE问卷总分之间的皮尔逊线性相关系数，并通过检验验证此类相关系数是否显著异于零。本研究采用RStudio 3.5版本软件进行数据分析。
2. AMIOFE问卷分析
为对比非发绀型心脏病患者、发绀型心脏病患者与对照组受试者，本研究针对问卷所有条目、分类维度总分以及总总分分别构建多元线性回归模型。此类模型同样将性别与年龄作为解释变量纳入其中。所有表格中的截距项，对应对照组新生男婴的变量预期值。需特别说明的是，针对口角、硬腭体积、下颌滑动、食物渗漏、呛咳以及下颌运动不协调这几项变量，由于所有受试患者的该项指标取值均完全一致，无法构建对应的回归模型。