A controlled, randomized, single-blinded trial was carried out with usual-risk pregnant women. The study was conducted in a municipality in northeastern Brazil with a population of 271,926 inhabitants, a Municipal Human Development Index (MHDI) score of 0.694 and divided administratively into seven health districts. Registro Brasileiro de Ensaios Clínicos (Brazilian Registry of Clinical Trials) (ReBEC): RBR-649bhb.
2.1. Sample Selection and Group Allocation
The groups were formed from Basic Health Units (BHUs), where family health teams with a dentist work. These are health services that offer primary healthcare. Random sampling was performed without replacement from a list of 20 BHUs selected for the study using BioEstat version 5.3 (Instituto Mamirauá, Tefé, Brazil) for composition of the two groups; ten BHUs in the intervention group (IG) and ten BHUs in the control group (CG) would follow routine prenatal care (
Figure 1). This selected study design avoids the risk of bias from the study experiment, respecting Consort’s recommendation [
17].
The participation of pregnant women was conditioned to registration in the public health information system (e-sus) and prenatal follow-up in the selected BHU. The selection criteria were as follows: willingness to participate in the intervention meetings; to be in the first trimester of pregnancy or, at most, in the first prenatal consultation of the second trimester in the recruitment period, which took place from April to June 2018. Exclusion criteria were related to health, including psychiatric problems, HIV, autoimmune diseases and illicit drug use, and to participation, including a lack of two consecutive or three alternate collective activities. Abortion, change of the drawn area or maternal death were characterized as losses.
For the sample calculation, the following parameters were used: 95% confidence interval, frequency of oral diseases in pregnant women of 21% [
18], difference of 40% in the frequency of these diseases between the groups after the intervention and a sampling power of 80%. The sample size of pregnant women eligible to participate in the study was 198, divided into two groups. However, the final analysis was performed with 146 pregnant women, with 58 belonging to the intervention group (IG), considering the protocol’s inclusion criteria, and 88 pregnant women for the control group (CG); the losses and reasons are exposed in the flowchart (
Figure 1).
Data gathering was performed from April 2018 to January 2019 with the participation of research assistants trained by four nutrition students, four nursing students and two dentists (n = 10) to level and apply the pilot test of instruments and collection. The research assistants attended on the days scheduled for prenatal consultation of the IG and CG at the BHU in order to identify the pregnant women who met the requirements at the first moment and to continue the study.
2.2. Intervention Description
The integrated care intervention for pregnant women was designed and validated by the consensus of specialists. It consisted of actions of healthcare, promotion and surveillance within the scope of primary care considering the entire gestational period until the beginning of the puerperium.
Pregnant women classified as having intermediate or high risk in oral health received a home visit from a community health worker (CHW) to reinforce the need to attend the BHU for dental treatment and guidance on oral hygiene care. To increase intervention adherence and follow-up, the research assistants made a telephone call to each patient five days before the collective actions (CAs) meeting to confirm the date and emphasize the importance of maintaining follow-up.
The assistance action consisted of a dental treatment plan registered in the medical records and a “warning” on the pregnant woman’s card of the need for dental care at the BHU.
Concomitantly with dental care, in the IG, the pregnant women participated in biweekly meetings for health promotion actions (Cas) using the conversation wheel technique. This technique, which provides a space for dialogue and reflection of health problems, enables integrated healthcare strategies and contributes to health promotion [
19,
20].
In the conversation wheels, the pregnant woman’s card, vaccines and routine prenatal examinations were observed and questions were asked about complications in the period.
A program was elaborated with 14 Cas, lasting for an average of 60 min each, divided by trimester.
1st trimester of pregnancy: Five CAs were carried out, addressing the following themes: physiological, emotional changes and nutritional risks in pregnancy; harm of drugs (smoking, alcoholic beverages and other legal and illegal drugs); oral hygiene guidance and practice; healthy eating, obesity prevention, diabetes, hypertension and use of sugars and sweeteners; exchange of experiences between pregnant women.
2nd trimester of pregnancy: Four CAs were carried out, covering the following themes: gingival and periodontal disease during pregnancy; the importance of daily personal and oral hygiene; self-examination of the mouth (prevention of oral cancer and other diseases); oral hygiene guidance and practice; food hygiene and rational use of sugar and sweeteners; social and labor rights for pregnant women.
3rd trimester of pregnancy: Five CAs were carried out, addressing the following themes: labor, family planning, exams and baby vaccines; exclusive breastfeeding and baby’s oral health; prevention of caries, oral hygiene and care for baby’s utensils; responsible parenting and accident prevention; oral hygiene guidance and practice.
Every two months, during the intervention period, the main researcher visited the BHU nurses for complementary information about the pregnant women (IG and CG) regarding complications to define the follow-up in the study.
A questionnaire was applied to all participants, from both groups, in a private room or in the BHU dental office.
The questionnaire was composed of five parts: 1—General Identification, including identification of the BHU, data gathering date, name of the participant, address, telephone contact, gestational period and name of the community health agent; 2—Sociodemographic Factors, including maternal age in years, skin color/ethnicity, marital status and how many people live in the house; 3—Socioeconomic Data, including educational level (categorized as <5 years of study, incomplete primary school (from 1st grade to 8th grade), complete primary school (from 9th grade to 2nd grade of high school), incomplete high school and higher education (3rd year of high school and incomplete higher education) and complete higher education); family income based on the minimum wage in force in 2018 (261.36 USD) [
21] (≤1 salary, 1–2.9 salaries, ≥3 salaries); home water supply (indoor household water from the public network, piped water from a well or a spring, water from a neighbor or other); 4—Behavioral Factors, including smoking (yes/no); alcohol (yes/no); daily tooth brushing (yes/no); food and access to dental services; 5—Self-perception and impacts on oral health.
To stratify the oral health risk, a clinical examination was carried out in the dental office (using procedure gloves and wooden spatula) of each basic health unit at two points: after the questionnaire was applied at the beginning of the study and during the puerperal visit by auxiliary BHU dentists and the main researcher. The oral health risk of both groups was stratified, which served as a baseline and was included in the medical record. The risk stratification took into account biological criteria (diabetes, hypertension, pregnant woman, use of alcohol and/or tobacco, adolescent) and socioeconomic and dental information (unemployed or not, toothache in the last six months, number of teeth with caries injury, mouth sore, gingival bleeding, need for specialized treatment); each item was assigned a score, the sum of which was evaluated at the end, being classified as low risk (0–10 points), medium risk (11–30 points) and high risk (over 30 points) [
22,
23]. All participants were informed of their risk classification; pregnant women who were members of the CG were instructed to seek dental care at the BHU and the IG had already started the dental intervention at the BHU.
Information on the occurrence of complications in childbirth and the babies’ data (gestational age, birth weight and complications) were collected in the BHU and/or in the puerperal consultation with the participants of both groups.
Maternal and child factors included the following: prematurity (term ≥37 weeks; preterm <37 weeks); maternal complications in the puerperal pregnancy period—no, with complication (hypertension, gestational diabetes, pre-eclampsia, postpartum hemorrhage); child weight at birth in grams—normal weight ≥2500, low weight <2500; baby complications—without complications, with complications (large for gestational age (LGA), neonatal death).
2.3. Analysis Plan
Descriptive analysis and frequency distribution were used to describe the sociodemographic, behavioral, maternal and child variables and oral health risk. Descriptive analyses were performed using the Jasp 13.1 program (Free Version, University of Amsterdam, Amsterdam, The Netherlands).
A network analysis was used to assess the association between sociodemographic, behavioral, maternal and child oral health factors and group (intervention and control). The betweenness, closeness and strength indicators have been reported; variables with higher betweenness values are more sensitive to changes from interventions and can act as a hub, connecting other pairs of variables in the network. A variable with a high proximity value (closeness) will be quickly affected by changes anywhere in the network and can also affect other parts. The strength indicator is essential for understanding which variables have the most robust connections in the current network pattern [
24].
The Fruchterman–Reingold algorithm was applied; therefore, the data were shown in the relative space, in which the variables with stronger permanent statistics together and those with less strongly applied variations repelled each other. To improve network accuracy, we used the pairwise Markov random fields model. The algorithm adds a penalty “L1” (regularized neighborhood regression). The adjustment is estimated by a less complete selection and contraction operator (Lasso) that controls the sparse network [
25].
Network analysis uses regularized algorithms of least absolute shrinkage and selection operator (LASSO) to obtain the precision matrix, which, when standardized, represents the associations between the variables present in the network. For a better visualization of the weight matrix, the network is presented in a chart that includes the variables (nodes) and relationships (lines). Blue or green color represents positive associations, and red represents negative associations. The thickness and intensity of the colors represent the magnitude of the associations. The R studio Q GRAPH package was used [
24].
The extended Bayesian information criterion (EBIC) was observed to select the lambda of the regularization parameter [
26]. EBIC uses a hyper parameter (y) that determines how much the EBIC selects sparse models. The value of y was determined as 0.25 (range from 0 to 0.50), which is a more parsimonious value when there are exploratory networks, as in the present study.
The ethical aspects of the Declaration of Helsinki were respected [
27]. This study was approved by the Research Ethics Committee at the Integral Medical Institute Professor Fernando Figueira—IMIP/PE, under no. 1.744.599, and the University of Juazeiro do Norte—Unijuazeiro/CE, under no. 1.802.276.
