A follow-up prevalence study was carried out in two middle-sized cities in the State of Rio de Janeiro, Brazil, to test the relationship of social determinants with pregnancy outcomes and oral health measures . All pregnant women enrolled in a fixed cohort who sought prenatal care at the four main public health care units administered by the National Health Care System ("Sistema Unico de Saude - SUS") were selected and invited to participate in this study. They were a representative sample of 95% of the women who were pregnant during the study period in both cities.
The sample size was estimated as 1059 subjects based on the prevalence of 59.5% of the impact of oral health on quality of life, considering OHIP>1 [32,40] to detect a 5% of the differences between groups, with a significance level of 5% and power of 95% . A study with 20% of losses during follow-up required 1270 participants.
Primary data were collected through face-to-face individual structured interviews between October 2008 and December 2009. The information was obtained at baseline (first trimester of pregnancy) and during the 30 days postpartum period (follow-up).
The selection criteria were women in the first trimester of pregnancy and living at their current address for at least 12 months. The latter criterion was used because social networks and social support tend to be stable after some months. First, the interviewers inspected the medical notes and chose pregnant women according to the selection criteria. All eligible pregnant women were invited to participate. They were informed about the objectives of the study. One of the interviewers requested their participation. After obtaining their consent, the women were interviewed. The study was approved by the Committee of Ethics and Research of the National School of Public Health - ENSP/FIOCRUZ (protocol no. 158/06).
Definition of occupational context
The main exposure was the occupational context, which was considered to be composed of different characteristics of way of life and characteristics related to occupational status.
Groups of comparison
Participants were allocated to two groups: 1. the work-based social network group were employed women with paid work. 2. the home-based social network group were women with no paid work, housewives or unemployed women. Measures of social support and social network were evaluated to characterize the occupational context.
Social network and social support measures
Social networks was considered as the "web" of social relationships surrounding the individual as well as their characteristics, or groups of people who have contact with, or with some form of participation . The questionnaire used to assess social networks consisted of 5 questions concerning the person's relationship with family and friends, and their participation in social groups. The instrument has adequate psychometric properties for the Brazilian population [43, 44]. Social support was considered as a system of formal and informal relationships through which individuals receive emotional support, material or information to cope with stressful emotional situations . Social support was evaluated using a questionnaire consisting of 19 items comprising five dimensions of functional social support: material (4 questions - provision of practical resources and support material), emotional (3 questions - physical expressions of love and affection), emotional (4 questions - expressions of positive affection, understanding and feelings of confidence), positive social interaction (4 questions - availability of people to have fun or relax), and information (4 questions - availability of people to obtain advice or guidance) . For each item, the women indicated how often they experienced each type of available support: never, rarely, sometimes, often or always. This questionnaire had good reliability for the Brazilian population .
The impact of oral health on quality of life
The outcome was the impact of oral health on quality of life, which reflects the perception of people about dysfunction, discomfort and disability related oral conditions. The validated version of Oral Health Impacts Profile (OHIP-14) for Brazilian population was used to evaluate the experience of impact on oral health on quality of life in the preceding 6 months [32, 40]. OHIP-14 is composed of 14 items, aggregated in 7 dimensions (two items per dimension) as following: functional limitation (items 1 and 2), physical pain (items 3 and 4), psychological discomfort (items 5 and 6), physical disability (items 7 and 8), psychological disability (items 9 and 10), social disability (items 11 and 12) and handicap (items 13 and 14). The overall score was computed by additive method, which is the sum of the individual scores of all items. For each item, the score varied from 0 to 4: "never" = 0, "hardly ever" = 1, "occasionally" = 2, "often" = 3, and "very often" = 4. A high score indicates a negative influence of oral health on quality of life.
The covariates were demographic and socioeconomic characteristics, health related behaviors previous and during pregnancy, dental pain in the last 6 months and number of teeth (<10 teeth versus ≥10 teeth). Demographic data were maternal age, ethnicity and number of children.
Socioeconomic characteristics were marital status, educational level (years of schooling), familial income, head of the family, housing conditions and social class. In this study the term social class refers to the social and economic factors that influence what position(s) individuals and groups hold within the structure of society . A standard social class classification commonly used in Brazil was used . This is an economic classification based on market power comprising a group of specific indicators such as number of bathrooms, number of full-time domestic servants, number of cars owned by the family, possession of domestic items such as television sets, radio sets, VCRs, vacuum cleaners, washing machine, fridges, freezers; and level of education of the head of household. A set of points is assigned to these indicators and a final score defines the socioeconomic groups; A (highest), B, C, D, and E (lowest). Those with the highest scores represented the highest socioeconomic groups.
The health behaviors, assessed before pregnancy, were smoking, cigarette consumption and alcohol consumption. In addition, the Brazilian version of T-ACE questionnaire, based on 5 questions concerning self-perception of drinking habits, was used to assess risky alcohol drinking before pregnancy .
The interviewers were trained to conduct structured and standardized interviews. After training the interviewers, a pilot study was performed to test understanding and layout of questionnaires. Examiners interviewed 40 pregnant selected women at the same health care units of the main study but who were not included in the main study.
Data collection was performed by 20 trained interviewers and four fieldwork supervisors. The baseline was conducted in the prenatal health care units to collect occupational context data, social network, social support, demographic and socioeconomic characteristics, number of teeth and health related behaviors. During the baseline interview different strategies were established to reduce the losses to follow-up. First, two telephone numbers were requested. Second, the full current address was registered, including the zip code. Third, contact telephone numbers of the fieldwork supervisors were provided for all women. They were requested to telephone one of the supervisors when admission to the maternity unit or discharge from it was arranged. In addition, they were asked to report if they moved home or changed their telephone number.
The follow-up study was performed in the post partum period immediately after the delivery to collect data on the impact of oral health on quality of life and dental pain in the last 6 months. The interview was conducted in the maternity hospital wards or at the mother's house up to 30 days after discharge. Women who moved home were excluded. In addition, those who had a miscarriage (pregnancy interrupted before the 20th gestational week) or abortion were not re-interviewed.
All variables were computed for each participant and then for each group. The normal distribution of continuous variables was tested using the Kolmogorov-Smirnov test. Since the continuous variables were not normally distributed, the comparison of groups was performed by Mann-Whitney test. Categorical variables were analyzed by Chi-square test.
Internal consistencies for the OHIP scale and its domains were evaluated by the Cronbach's α coefficient. Cronbachs' α removing each domain of the OHIP were also assessed.
The relationship between occupational context and the impact of oral health on quality of life was tested using multinomial logistic regression. The sample was categorized into 3 groups according to the prevalence and the median (the median of OHIP = 3) of the number of impacts of OHIP: OHIP = 0 (No impact); OHIP 1-3 (Scores from 1 to 3); OHIP ≥4 (Scores ≥4). In addition, the sample was grouped concerning the dimensions of social support. Subjects with low levels of impacts were those with scores equal to zero, moderate level subjects were those between zero and the median, and high level subjects were those above the median.
First, a comparison was made between social support dimensions and types of social networks in the work-based and home-based groups. Social support and social network variables that were statistically different between occupational context groups were included in the bivariate analysis. The crude Odds Ratio (OR) and Confidence Intervals of 95% were calculated between occupational context and covariates and OHIP groups. Second, multinomial logistic regression was performed to obtain adjusted OR of occupation context, affectionate support, positive social interaction and social network/friends with OHIP adjusted for age, ethnicity, family income, schooling, marital status and social class (Model 1).
To test the statistical significance of interaction between occupational context and potential modifying factors (social support dimensions and social network) the occupational context and covariates were first added to the regression model. After that, the interaction terms 'occupational context X affectionate support','occupational context X positive social interaction' and 'occupational context X social network/friends' where added to the model (Model 2). Model 1 (without interaction terms) and Model 2 (with interaction terms) were compared using Likelihood Ratio tests.
All statistical analyses were performed using the SPSS (Statistical Package for Social Sciences, version 13.0). The significance level for all analysis was 5% (P = 0.05).