Abstract
Observational studies suggest association between low concentrations of omega‐3 family fatty acids and greater risk for post‐partum depression (PPD). The objective was to investigate the effect of unbalanced dietary intake of omega‐6/omega‐3 ratio >9:1 in the prevalence for PPD. The study comprises a prospective cohort with four waves of follow‐up during pregnancy and one following delivery. PPD was evaluated according to the Edinburgh Post‐partum Depression Scale (PPD ≥ 11) in 106 puerperae between 2005 and 2007, in Rio de Janeiro, Brazil. Independent variables included socio‐demographic, obstetric, pre‐pregnancy body mass index (BMI) and dietary intake data, which were obtained by means of a food frequency questionnaire in the first trimester of pregnancy. Statistical analysis involved calculation of PPD prevalence and multivariate Poisson regression with robust variance. PPD prevalence amounted to 26.4% [n = 28; confidence interval (CI) 95%: 18.0–34.8], and higher prevalences of PPD were observed in women who consumed an omega‐6/omega‐3 ratio >9:1 (60.0%) and in those with pre‐pregnancy BMI <18.5 kg/m2 (66.7%). These variables held as factors associated to PPD in the multivariate model, elevating the chances of occurrence of the outcome in 2.50 (CI 95%: 1.21–5.14) and 4.01 times (CI 95%: 1.96–8.20), respectively. Analyses were adjusted for age, schooling, pre‐pregnancy BMI, lipids consumption and time elapsed since delivery. It verified an association between omega‐6/omega‐3 ratio above 9:1, the levels recommended by the Institute of Medicine, and the prevalence of PPD. These results add to the evidence regarding the importance of omega‐6 and omega‐3 fatty acids in the regulation of mental health mechanisms.
Keywords: depression, post‐partum, pregnancy, omega‐3, schooling, body mass index
Introduction
Post‐partum depression (PPD) is a serious public health problem (Cruz et al. 2005). Among the most commonly observed symptoms are intense sadness, frequent crying, lack of motivation, lowered interest for eating or for self‐care, lack of concentration and a low interest for the newborn (Baker et al. 2005). Children of depressed mothers bear greater risks of developing psychopathologies, behavioural and developmental problems, delay in the cognitive and emotional development (Tam et al. 2002; Zinga et al. 2005).
It is often observed that PPD symptoms are neglected by the woman, producing an underestimation of the prevalence, that makes it harder to track and treat the disease adequately (Baker et al. 2005; Zinga et al. 2005). In Brazil, PPD prevalence varies between 13.4% (Santos et al. 1999) and 37.1% (Cruz et al. 2005), while results from international studies revealed prevalences between 10 and 15% (Boyce & Hichey 2005; Freeman et al. 2006).
Recent evidence from observational studies and clinical trials have evaluated the possible role of omega fatty acids consumption in the aetiology of PPD and suggested an association between the low level of an essential long‐chain fatty acid of the omega‐3 family, precursor to the docosahexaenoic acid (DHA), following pregnancy and the occurrence of PPD (Freeman et al. 2006; Freeman et al. 2008; Golding et al. 2009).
A conceptual framework in which polyunsaturated fatty acids (PUFAs) are thought to influence depression follows a bio‐psychosocial model of depression. This model is encompassed within the class of the diathesis–stress model, and is based on the hypothesis that certain biological characteristics (genetic traits or disturbances in the biochemical, neuroendocrine or immune systems), interact with psychological characteristics, such as cognitive style, that make people vulnerable to depression upon exposure to negative life events or stressful experience (Schotte et al. 2006; Sontrop 2007).
According to this model, the experience of depression does not result from a single cause but from the culmination of events originating from a variety of causal pathways. Support for this model comes from experimental evidence demonstrating that cognitive‐behavioural therapy combined with antidepressant treatment is more effective at treating depression than either form of treatment alone (Arnow & Constantino 2003). It is important to recognize that the hypothesis that PUFAs may be a contributory factor in the pathophysiology or pathogenesis of depression represents part of a complex bio‐psychosocial framework (Sontrop 2007).
Few are the studies regarding dietary consumption in PPD. Thus, the main objective of this study is to evaluate the association between an unbalanced dietary intake ratio between the omega‐6 and the omega‐3 fatty acids above 9 in the first trimester of pregnancy and the prevalence of PPD.
Key messages
-
•
Post‐partum depression (PPD) is a serious health problem for underprivileged Brazilian women.
-
•
Low pre‐pregnancy body mass index may act as a risk factor for PPD.
-
•
A high dietary ratio of omega‐6 to omega‐3 polyunsaturated acids during the first pregnancy trimester increased the prevalence of PPD.
Methods
This investigation originated from a study with a prospective cohort design performed in the state of Rio de Janeiro, Brazil. Enrolment of pregnant women in the cohort was open during 20 months (July 2005–March 2007). The study comprised five waves of follow‐up: on the 8th–13th, 19th–21st, 26th–28th and 36th–40th gestational week, and with at least 30 days post‐partum. At those times, interviews were conducted with collection of socio‐economic, food consumption, obstetric, lifestyle and anthropometric data. The interviews were conducted during appointment days at a public health centre of the Brazilian Unified Health System. The interviewers were trained for administering the questionnaires. The women were weighed on a digital scale (Filizzola PL 150, Filizzola LTDA, São Paulo, Brazil) and stature was measured with a portable stadiometer (Harpenden Inc., England) by trained interviewers, following standardized criteria (Lohman et al. 1988). The pre‐gestational nutritional status was assessed by pre‐gestational body mass index (BMI = weight/stature2), which was measured at the first wave of follow‐up, time limit for the definition of the pre‐gestational nutritional diagnosis with measured weight (Institute of Medicine 2009).
Only women who met the following eligibility criteria at the time of enrolment were selected to participate in the study: (1) being between the 8th and 13th week of gestation; (2) being between 18 and 40 years of age; (3) being free from any chronic diseases such as hypertension and diabetes; and (4) not presenting twin pregnancy.
Two hundred fifty‐five women receiving prenatal care at the public health centre were recruited to participate in the study. In the present investigation were included data of PPD from 106 women who completed the follow‐up. From these 106, 5 were considered outliers (values of energy intake outside the range ± 3 SD), and were not included in the analyses regarding PPD prevalence according to dietary data and in the unadjusted and adjusted models (final n = 101). Some variables presented missing values in baseline (8th–13th weeks of gestation) as follows: monthly income (n = 1), skin colour (n = 3), menarche age (n = 1) and dietary data (n = 21). Some variables were collected in other follow‐up waves, moments when some losses of follow‐up were already present, like stressful events (available number of observations = 88) and marital problems during pregnancy (available number of observations = 67).
With the objective of evaluating the presence of selective losses in the final follow‐up, comparisons regarding the distribution of monthly income, age, schooling, self‐reported skin colour, marital status, smoking status, alcohol consumption, menarche age, parity, pre‐gestational BMI, energy intake adequacy and omega‐6/omega‐3 ratio variables were made between the women who completed follow‐up and those who did not participate in the fifth wave of follow‐up. Proportions were compared using chi‐squared test. For dichotomous variables, Fisher exact test was employed, and Yates corrected otherwise.
PPD was considered as a dependent variable and its evaluation was performed by means of the Edinburgh Post‐partum Depression Scale (EPDS). This scale consists of a self‐report instrument composed of 10 items, with options for scoring between 0 and 3 according to the presence and the intensity of the depressive symptom (Cox et al. 1987). Several studies have already employed the EPDS as this is an easy scale to understand and apply, not to mention its effectiveness in tracking PPD (Tam et al. 2002; Otto et al. 2003; Boyce & Hichey 2005; Cruz et al. 2005; Moraes et al. 2006; Flores‐Quijano et al. 2008; Freeman et al. 2008; Gausia et al. 2008). In a study conducted in Brazil for the validation of the Portuguese version of EPDS, the instrument presented an internal consistency with a Cronbach's alpha coefficient of 0.80 (Santos et al. 1999).
EPDS was applied in the fifth wave of follow‐up, at least 30 days following delivery. The participants were encouraged to complete the questionnaire by themselves, but they had access to the help of a trained interviewer in case any explanation was necessary or to assist them in reading the items and filling out the answers in case any of the women were not able to do so independently. Studies within the Brazilian population revealed that scores ≥11 presents a better predicting power, combining higher sensitivity and specificity (Santos et al. 1999). In a recent validation study of EPDS for the Brazilian population, Santos et al. (2007) have also shown that scores ≥11 had higher sensitivity (83.8%) and specificity (74.7%) for screening moderate and severe PPD, in comparison with scores ≥10, which presented 82.6% and 65.4% of sensitivity and specificity, respectively. For this reason scores ≥11 was adopted in the present investigation.
The main exposure investigated in this study was the dietary intake in the first trimester of pregnancy, which was assessed by means of a Food Frequency Questionnaire (FFQ). The previously validated FFQ employed encompasses a list with 81 items and 8 frequency options. It is a robust method and largely accepted in epidemiological studies (Willett 1994; Sichieri & Everhart 1998). An FFQ questionnaire was applied in the second and third trimester, but losses of follow‐up constrained its use.
The pregnant women's dietary intake was converted into weight or volume according to each food type. Subsequently, total energy intake and diet composition were analysed based on a worksheet developed and adapted for the present study in microsoft ® office excel 2003 (Microsoft Corporation, Richmond, VA, USA). This worksheet was based on data contained in the Brazilian Food Composition Table (Tabela Brasileira de Composição de Alimentos 2004) and on the United States Development Agency foods composition table (United States Department of Agriculture 2006). The consumption of daily macro‐ and micronutrients in g was estimated, as well as the ratio omega‐6/omega‐3.
The following variables were considered in the analyses: time elapsed since delivery (defined as the number of days between delivery and measurement of the EPDS); socio‐economic – monthly income (in quintiles of American dollars); socio‐demographic – age (18.0–19.9; 20.0–29.9; 30.0–41.0 years), schooling (≤9; >9 years), self‐reported skin colour (white, mulatto, black) and marital status (married or stable partnership, single); lifestyle – smoking status (yes, no) and alcohol consumption (yes, no); obstetric – age at menarche (≤13, >13 years), type of delivery (vaginal, caesarean) and parity (zero, 1–2, ≥3); infant outcomes – prematurity (gestational age measured through last menstrual cycle <37 weeks) and low birthweight (birthweight <2500 g); pre‐gestational nutritional status classified based on the BMI according to the American Institute of Medicine, as follows: underweight <18.5 kg/m2, normal weight 18.5–24.9 kg/m2, overweight 25.0–29.9 kg/m2 and obese ≥30.0 kg/m2 (Institute of Medicine 2009); and dietary intake in the first pregnancy trimester: energy intake adequacy – per cent (<80; 80–120; >120), lipids consumption – g (≤92.0, >92.0) and omega‐6/omega‐3 ratio intake (≤9:1, >9:1). This ratio was employed based on the isolated recommendations for omega‐6 and omega‐3 from Institute of Medicine (2004), which are 13.0 g and 1.4 g, respectively. The food intake energy adequacy was calculated according to the recommended energy for age, weight and estimated level of physical activity, considering an adequate gestational weight gain (Institute of Medicine 2009).
Additionally, possible confounding factors in the relationship between consumption of omega‐6/omega‐3 and PPD were included in the analysis: problems in the marital relationship during pregnancy (yes, no) and occurrence of stressful events during pregnancy (yes, no). This information was obtained through the question, ‘Did you suffer any stressful event during pregnancy?’, evaluated according to the women's perception. Some examples reported include relationship problems with husband and lack of money to buy food items.
Statistical methods
Double typing in the epi‐info ® version 6.04b (Center for Disease Control and Prevention, Atlanta, GA, USA) was performed. Analyses were conducted employing stata version 9.2 (Stata Corporation, College Station, TX, USA). Initially, the distribution of each co‐variable of the sample was investigated with the objective of defining the best categories for the calculation of PPD prevalence. The prevalences were compared by means of the chi‐squared test for proportions. Continuous variables were compared using student t‐test whenever the variables had normal distribution and Mann–Whitney for those with skewed distributions, stratifying according to the presence or not of PPD. In the bivariate analysis, unadjusted prevalence ratio (PR) was used as effect measure, with a confidence interval of 95% (CI 95%), and the Wald test was employed to verify the association. As the outcome presents a binary distribution and prevalences above 10%, the procedure of multivariate analysis chosen was the Poisson regression with robust variance (Zou 2004). The variables were selected for multivariate analysis based either on P‐values <0.20 in the bivariate analysis or when defined as part of a previous established theoretical model. The implementation of this strategy avoids the exclusion of potentially important variables and also controls for residual confounding (Rothman et al. 2008). In the present analysis, variables such as age, schooling, time elapsed since delivery and lipid consumption were forced into the model as they were considered to be part of our theoretical model and thus adjust variables. The association was evaluated with the Wald test and for the quality of the adjustment of the final model of Poisson regression, the goodness‐of‐fit test was employed.
Results
From the 255 pregnant women who enrolled the cohort, 106 finished the follow‐up and, thus, have data regarding PPD. The final follow‐up rate was 41.6% (106/255). The comparisons performed between losses to follow‐up and those that completed the study revealed that there are no significant differences in the distribution of variables such as monthly income, age, schooling, self‐reported skin colour, alcohol consumption, parity, pre‐pregnancy BMI, dietary intake adequacy and in the omega‐6/omega‐3 ratio. Smokers present lower follow‐up rate, as well as women who were married or had a stable partnership (Table 1).
Table 1.
Frequency distribution for selected variables between losses and complete follow‐up and final follow‐up rate for Brazilian women aged 18–41 years in Rio de Janeiro, 2005–2007
| Variables | Initial observations (n) | Losses to follow‐up (n) | Complete follow‐up (n) | Final follow‐up rate (%) | P‐value* |
|---|---|---|---|---|---|
| Monthly income (dollars) † | |||||
| <205.00 | 51 | 33 | 18 | 35.3 | |
| 205.00–335.00 | 62 | 39 | 23 | 37.1 | |
| 335.00–442.00 | 38 | 21 | 17 | 44.7 | |
| 442.00–670.00 | 50 | 25 | 25 | 50.0 | |
| ≥670.00 | 53 | 31 | 22 | 41.5 | 0.711 |
| Age (in years) | |||||
| 18.0–19.9 | 38 | 24 | 14 | 36.8 | |
| 20.0–29.9 | 155 | 94 | 61 | 39.4 | |
| 30.0–41.0 | 62 | 31 | 31 | 50.0 | 0.388 |
| Schooling (years) | |||||
| ≤9 | 125 | 78 | 47 | 37.6 | |
| >9 | 130 | 71 | 59 | 45.4 | 0.253 |
| Skin colour (self‐reported) ‡ | |||||
| White | 64 | 35 | 29 | 45.3 | |
| Mulatto | 152 | 93 | 59 | 38.8 | |
| Black | 36 | 19 | 17 | 47.2 | 0.647 |
| Marital status | |||||
| Married or stable partnership | 198 | 125 | 73 | 36.9 | |
| Single | 57 | 24 | 33 | 57.9 | 0.006 |
| Smoking status | |||||
| Yes | 30 | 23 | 7 | 23.3 | |
| No | 225 | 126 | 99 | 44.0 | 0.032 |
| Alcohol consumption | |||||
| Yes | 31 | 22 | 9 | 29.0 | |
| No | 224 | 127 | 97 | 43.3 | 0.173 |
| Menarche age (in years) § | |||||
| ≤13 | 180 | 107 | 73 | 40.5 | |
| >13 | 74 | 41 | 33 | 44.6 | 0.577 |
| Parity (number of birth) | |||||
| 0 | 128 | 66 | 62 | 48.4 | |
| 1–2 | 111 | 71 | 40 | 36.0 | |
| ≥3 | 16 | 12 | 4 | 25.0 | 0.101 |
| Pre‐pregnancy BMI (kg/m2) ¶ | |||||
| <18.5 | 14 | 8 | 6 | 42.8 | |
| 18.5–24.9 | 156 | 87 | 69 | 44.2 | |
| ≥25.0 | 85 | 54 | 31 | 36.5 | 0.579 |
| Energy intake adequacy (%)**, †† , ‡‡ | |||||
| >80 | 13 | 7 | 6 | 46.2 | |
| 80–120 | 22 | 14 | 8 | 36.4 | |
| >120 | 199 | 112 | 87 | 43.7 | 0.907 |
| Omega‐6/omega‐3 ratio intake**, †† , ‡‡ | |||||
| ≤9:1 | 208 | 117 | 91 | 43.8 | |
| >9:1 | 26 | 16 | 10 | 38.5 | 0.678 |
Fisher exact test P‐value refers to chi‐square for proportions for all dichotomous variables. For non‐dichotomous variables, Yates correction was applied.
† One missing value for monthly income.
‡ Three missing values for skin colour in baseline and one with complete follow‐up.
§ One missing value for menarche age in baseline.
Body mass index (BMI) classification according to the American Institute of Medicine 2009 (2009).
**Data from Food Frequency Questionnaire regarding the first trimester of pregnancy.
†† Twenty‐one missing values for dietary variables in baseline.
Five outliers for the dietary variables (values for energy intake outside the range ± 3 SD).
The prevalence of PPD observed in the sample was 26.4% (n = 28; CI 95%: 18.0–34.8). The greater prevalences of PPD were observed in women with pre‐pregnancy BMI <18.5 kg/m2 (66.7%), in women with an unbalanced dietary intake in the first trimester regarding the omega‐6/omega‐3 ratio (60.0%) and in those with schooling ≤9 years (36.2%) (Table 2).
Table 2.
Number of cases (n) and prevalence of post‐partum depression (PPD)
| Variable | N | n | Prevalence of PPD (%) | P‐value* |
|---|---|---|---|---|
| Monthly income (dollars) † | ||||
| <205.00 | 18 | 5 | 27.8 | 0.980 |
| 205.00–335.00 | 23 | 5 | 21.7 | |
| 335.00–442.00 | 17 | 6 | 35.3 | |
| 442.00–670.00 | 25 | 6 | 24.0 | |
| ≥670.00 | 22 | 6 | 27.3 | |
| Age (years) | ||||
| 18.0–19.9 | 14 | 3 | 21.4 | 0.880 |
| 20.0–29.9 | 61 | 18 | 29.5 | |
| 30.0–41.0 | 31 | 7 | 22.6 | |
| Schooling (years) | ||||
| ≤9 | 47 | 17 | 36.2 | 0.049 |
| >9 | 59 | 11 | 18.6 | |
| Skin colour (self‐reported) ‡ | ||||
| White | 29 | 8 | 27.6 | 0.885 |
| Mulatto | 59 | 16 | 27.1 | |
| Black | 17 | 3 | 17.6 | |
| Marital status | ||||
| Married or stable partnership | 73 | 19 | 26.0 | 0.999 |
| Single | 33 | 9 | 27.3 | |
| Smoking status | ||||
| Yes | 7 | 2 | 28.6 | 0.999 |
| No | 99 | 26 | 26.3 | |
| Alcohol consumption | ||||
| Yes | 9 | 2 | 22.2 | 0.999 |
| No | 97 | 26 | 26.8 | |
| Menarche age (in years) | ||||
| ≤13 | 73 | 19 | 26.0 | 0.999 |
| >13 | 33 | 9 | 27.3 | |
| Type of delivery | ||||
| Vaginal | 60 | 14 | 23.3 | 0.506 |
| Caesarean | 46 | 14 | 30.4 | |
| Parity (number of birth) | ||||
| 0 | 62 | 17 | 27.4 | 0.880 |
| 1–2 | 40 | 10 | 25.0 | |
| ≥3 | 4 | 1 | 25.0 | |
| Prematurity (gestational age < 37 weeks) § | ||||
| Yes | 10 | 3 | 30.0 | 0.724 |
| No | 95 | 25 | 26.3 | |
| Low birthweight (birthweight < 2500 g) § | ||||
| Yes | 6 | 1 | 16.7 | 0.999 |
| No | 99 | 27 | 27.3 | |
| Marital problems during pregnancy ¶ | ||||
| Yes | 11 | 4 | 36.4 | 0.719 |
| No | 57 | 16 | 28.1 | |
| Stress during pregnancy** | ||||
| Yes | 66 | 21 | 31.8 | 0.167 |
| No | 20 | 3 | 15.0 | |
| Pre‐pregnancy BMI (kg/m2) †† | ||||
| <18.5 | 6 | 4 | 66.7 | 0.156 |
| 18.5–24.9 | 69 | 15 | 21.7 | |
| ≥25.0 | 31 | 9 | 29.0 | |
| Energy intake adequacy (%) [Link] , [Link] | ||||
| >80 | 6 | 2 | 33.3 | 0.902 |
| 80–120 | 8 | 1 | 12.5 | |
| >120 | 87 | 23 | 26.4 | |
| Lipids consumption (g) [Link] , [Link] | ||||
| ≤92.0 | 52 | 10 | 19.2 | 0.172 |
| >92.0 | 49 | 16 | 32.6 | |
| Omega‐6/omega‐3 ratio intake [Link] , [Link] | ||||
| ≤9:1 | 91 | 20 | 22.0 | 0.017 |
| >9:1 | 10 | 6 | 60.0 |
*Fisher exact test P‐value refers to chi‐square for proportions for all dichotomous variables. For non‐dichotomous variables, Yates correction was applied. †One missing value for monthly income. ‡One missing value for skin colour. §One missing value for prematurity and low birthweight.¶Thirty‐eight missing values for marital problems during pregnancy (variable collected in the fifth follow‐up wave).**Twenty missing values for stress during pregnancy (variable collected in the fifth follow‐up wave). ††Body mass index (BMI) classification according to the American Institute of Medicine (2009). ‡‡Data from Food Frequency Questionnaire regarding the first trimester of pregnancy. §§Five outliers for the dietary variables (values for energy intake outside the range ± 3 SD).
The median energy intake in the first trimester of pregnancy was 3423 kcal/day, of which 512.8 g were carbohydrates, 94.2 g were lipids and 139.1 g were proteins. The medians of consumption for linoleic and linolenic acids were 15.4 and 2.1 g, respectively, with a ratio between them of 7.5 g (Table 3). The major dietary sources that contributed to omega‐6 and omega‐3 PUFAs intake were crackers, fried potato, sausage and chicken and rice, beans and low fat cow meat, respectively (results not shown). The mean pre‐pregnancy BMI was 23.8 kg/m2. The women from the sample presented a mean age of 26.1 years, income of $511.40 (in American dollars), and schooling of 9.3 years (Table 3).
Table 3.
Medians [interquartile range (IQR)] and means (±SD) of dietary daily intake and selected variables according to presence of post‐partum depression
| Variable* | n | Medians (IQR) or means (±SD † ) | Post‐partum depression | ||||
|---|---|---|---|---|---|---|---|
| Yes | No | P‐value | |||||
| n | Medians (IQR)or means (±SD) | n | Medians (IQR) or means (±SD) | ||||
| Energy (kcal) ‡ , § | 101 | 3423.6 (2621.9–4252.0) | 26 | 3379.3 (2393.6–4741.8) | 75 | 3423.6 (2656.6–4242.8) | 0.950 |
| Carbohydrates (g) ‡ , § | 101 | 512.8 (384.7–597.1) | 26 | 466.4 (334.1–588.3) | 75 | 506.0 (392.1–599.1) | 0.494 |
| Lipids (g) ‡ , § | 101 | 94.2 (68.4–116.7) | 26 | 97.9 (68.2–128.5) | 75 | 88.8 (68.4–109.5) | 0.207 |
| Proteins (g) ‡ , § | 101 | 139.1 (100.7–185.2) | 26 | 149.9 (102.8–190.3) | 75 | 138.0 (98.9–179.9) | 0.822 |
| Omega‐6 family fatty acids (18:2) (g) ‡ , § | 101 | 15.4 (11.1–18.6) | 26 | 16.1 (10.6–19.0) | 75 | 14.3 (11.1–18.4) | 0.492 |
| Omega‐3 family fatty acids (18:3) (g) ‡ , § | 101 | 2.1 (1.4–2.6) | 26 | 2.0 (1.2–2.6) | 75 | 2.0 (1.4–2.6) | 0.792 |
| Omega‐6/omega‐3 ratio ‡ , § | 101 | 7.5 (6.7–8.2) | 26 | 8.0 (6.9–8.8) | 75 | 7.4 (6.6–8.1) | 0.040 |
| Age (years) | 106 | 26.1 (5.7) | 28 | 25.8 (5.6) | 78 | 26.2 (5.8) | 0.767 |
| Monthly income (dollars) ¶ | 105 | 511.4 (421.7) | 28 | 534.8 (405.7) | 77 | 502.9 (429.6) | 0.733 |
| Schooling (years) | 106 | 9.3 (3.0) | 28 | 8.8 (3.0) | 78 | 9.4 (3.0) | 0.294 |
| Pre‐pregnancy BMI (kg/m2) | 106 | 23.8 (4.3) | 28 | 23.9 (5.2) | 78 | 23.7 (3.9) | 0.923 |
| Time elapsed since delivery (days) | 106 | 42.8 (20.4) | 28 | 41.2 (20.1) | 78 | 43.4 (20.6) | 0.852 |
Median values and P‐value according to Mann–Whitney test for the variables: energy, carbohydrates, lipids, proteins, omega‐6 family fatty acids, omega‐3 family fatty acids and omega‐6/omega‐3 ratio; and mean values and t‐test for the variables: monthly income, schooling, pre‐gestational body mass index (BMI) and time elapsed since delivery.
† SD, standard deviation.
‡ Data from the Food Frequency Questionnaire regarding the first trimester of pregnancy.
Five outliers for the dietary variables (values for energy intake outside the range ± 3 SD).
¶One missing value for monthly income.
In the bivariate analysis, greater prevalences ratio of PPD were observed in women with ≤9 years of schooling (PR = 1.94; CI 95%: 1.00–3.74), pre‐pregnancy BMI <18.5 kg/m2 (PR = 3.07; CI 95%: 1.48–6.33) and with food intake in the first trimester for the omega‐6/omega‐3 ratio >9 : 1 (PR = 2.73; CI 95%: 1.44–5.18) (Table 4). The variables stress during the pregnancy and consumption of lipids in the first trimester of pregnancy presented a P‐value less than 0.20, and followed to the multivariate model. Lipids consumption in the first trimester of pregnancy lost significance but was kept in the model because it was considered an important variable in the theoretical model once it interferes directly in the consumption of omega‐6 and omega‐3. Stress during pregnancy was removed from the model because it has excessive missing values, constraining the potentiality of the final model.
Table 4.
Unadjusted prevalence ratio and 95% confidence intervals (CI 95%) for post‐partum depression*
| Variable | Prevalence ratio | CI 95% | P‐value |
|---|---|---|---|
| Monthly income (dollars) | |||
| <205.00 | 1.00 | ||
| 205.00–335.00 | 0.78 | 0.26–2.30 | 0.657 |
| 335.00–442.00 | 1.27 | 0.47–3.42 | 0.635 |
| 442.00–670.00 | 0.86 | 0.31–2.41 | 0.780 |
| ≥670.00 | 0.98 | 0.36–2.71 | 0.972 |
| Age (years) | |||
| 18.0–19.9 | 1.00 | ||
| 20.0–29.9 | 1.38 | 0.47–4.06 | 0.562 |
| 30.0–41.0 | 1.05 | 0.32–3.50 | 0.932 |
| Schooling (years) | |||
| ≤9 | 1.94 | 1.00–3.74 | 0.048 |
| >9 | 1.00 | ||
| Skin colour (self‐reported) | |||
| White | 1.00 | ||
| Mulatto | 0.98 | 0.48–2.03 | 0.963 |
| Black | 0.64 | 0.19–2.10 | 0.462 |
| Marital status | |||
| Married or stable partnership | 1.00 | ||
| Single | 1.05 | 0.53–2.07 | 0.893 |
| Smoking status | |||
| Yes | 1.09 | 0.32–3.69 | 0.893 |
| No | 1.00 | ||
| Alcohol consumption | |||
| Yes | 0.83 | 0.23–2.96 | 0.773 |
| No | 1.00 | ||
| Menarche age (in years) | |||
| ≤13 | 1.00 | ||
| >13 | 1.05 | 0.53–2.07 | 0.893 |
| Delivery type | |||
| Vaginal | 1.00 | ||
| Caesarean | 1.30 | 0.69–2.46 | 0.413 |
| Parity (number of birth) | |||
| 0 | 1.10 | 0.19–6.33 | 0.918 |
| 1–2 | 1.00 | 0.17–5.98 | 0.999 |
| ≥3 | 1.00 | ||
| Prematurity (gestational age < 37 weeks) | |||
| Yes | 1.14 | 0.41–3.13 | 0.597 |
| No | 1.00 | ||
| Low birthweight (birthweight < 2500 g) | |||
| Yes | 0.53 | 0.06–4.78 | 0.533 |
| No | 1.00 | ||
| Marital problems during pregnancy | |||
| Yes | 1.30 | 0.53–3.16 | 0.569 |
| No | 1.00 | ||
| Stress during pregnancy | |||
| Yes | 2.12 | 0.70–6.42 | 0.183 |
| No | 1.00 | ||
| Pre‐pregnancy BMI (kg/m2) † | |||
| <18.5 | 3.07 | 1.48–6.33 | 0.002 |
| 18.5–24.9 | 1.00 | ||
| ≥25.0 | 1.34 | 0.65–2.72 | 0.426 |
| Time elapsed since delivery (days) | 0.99 | 0.98–1.01 | 0.518 |
| Energy intake adequacy (%) [Link] , [Link] | |||
| >80 | 2.67 | 0.30–23.24 | 0.375 |
| 80–120 | 1.00 | ||
| >120 | 2.11 | 0.32–13.80 | 0.434 |
| Lipids consumption (g) [Link] , [Link] | |||
| ≤92.0 | 1.00 | ||
| >92.0 | 1.70 | 0.85–3.39 | 0.133 |
| Omega‐6/omega‐3 ratio intake [Link] , [Link] | |||
| ≤9:1 | 1.00 | ||
| >9:1 | 2.73 | 1.44–5.18 | 0.002 |
*Prevalence ratio and 95% CI were estimated through Poisson regression. P‐value for Wald test. †Body mass index (BMI) classification according to the American Institute of Medicine (2009). ‡Data from the Food Frequency Questionnaire regarding the first trimester of pregnancy. §Five outliers for the dietary variables (values for energy intake outside the range ± 3 SD).
It was observed that, of the variables studied, the dietary intake in the first trimester for the omega‐6/omega‐3 ratio >9 : 1 and the pre‐pregnancy BMI below 18.5 kg/m2 remained as factors associated with PPD, increasing the chances for occurrence of the outcome for women 2.50 times (CI 95%: 1.21–5.14) and 4.01 (CI 95%: 1.96–8.20), respectively. The estimations were adjusted for age, schooling, time elapsed since delivery and lipid consumption (Table 5). The final sample size in the multivariate analysis was 101 subjects.
Table 5.
Multivariate Poisson model for factors associated to post‐partum depression* (n = 101)
| Variable | Unadjusted prevalence ratio | Adjusted prevalence ratio | CI 95% | P‐value |
|---|---|---|---|---|
| Age (in years) | ||||
| 18.0–19.9 | 1.00 | 1.00 | ||
| 20.0–29.9 | 1.38 | 1.21 | 0.40–3.65 | 0.730 |
| 30.0–41.0 | 1.05 | 1.03 | 0.31–3.38 | 0.961 |
| Schooling (in years) | ||||
| ≤9 | 1.94 | 1.80 | 0.89–3.65 | 0.102 |
| >9 | 1.00 | 1.00 | ||
| Pre‐pregnancy BMI (kg/m2) † | ||||
| <18.5 | 3.07 | 4.01 | 1.96–8.21 | <0.001 |
| 18.5–24.9 | 1.00 | 1.00 | ||
| ≥25.0 | 1.33 | 1.23 | 0.57–2.65 | 0.602 |
| Time elapsed since delivery (days) ‡ | 0,99 | 1.00 | 0.98–1.02 | 0.969 |
| Lipids consumption (g) [Link] , [Link] | ||||
| ≤92.0 | 1.00 | 1.00 | ||
| >92.0 | 1.70 | 1.68 | 0.88–3.20 | 0.112 |
| Omega‐6/omega‐3 ratio intake [Link] , [Link] | ||||
| ≤9:1 | 1.00 | 1.00 | ||
| >9:1 | 2.73 | 2.50 | 1.21–5.14 | 0.013 |
*Prevalence ratio and 95% confidence intervals (CIs) were estimated through Poisson regression. P‐value for Wald test. †Body mass index (BMI) classification according to the American Institute of Medicine (2009). ‡Continuous variable included on the final model. §Data from the Food Frequency Questionnaire concerning the first trimester of pregnancy.¶Five outliers for the dietary variables (values for energy intake outside the range ± 3 SD). In the multivariate model all variables are adjusted for each other.
Discussion
The prevalence of PPD was 2.5 times greater among Brazilian women whose dietary ratio of omega‐6/omega‐3 PUFAs in the first trimester of pregnancy was greater than 9 : 1. This ratio represents the division between the specific recommendations of the Institute of Medicine (2004) for omega‐6 and omega‐3, which are, respectively, 13.0 and 1.4. The results remained statistically significant even when the analysis were adjusted for the effect of several confounding factors such as age, schooling, pre‐pregnancy BMI, lipid consumption and time elapsed since delivery. These results corroborate the data from an observational study where the increase in DHA concentration in the post‐partum, was significantly lower in women with PPD (Otto et al. 2003).
In this study, we investigated the median difference in the consumption of omega‐6 and omega‐3, and the ratio between these nutrients, according to presence of PPD. None of the nutrients evaluated separately showed association with PPD. However, upon evaluating the omega‐6 and omega‐3 family fatty acids ratio, we verified that the median of the omega‐6/omega‐3 ratio consumed in the first trimester of pregnancy was significantly higher (P = 0.040) for women with PPD. Jacka et al. (2004) did not observe any significant difference for the means of omega‐3 fatty acid consumption between depressed and non‐depressed in a sample of Australian women and Sontrop et al. (2008) revealed an inverse relationship between omega‐3 family fatty acids intake and depressive symptoms in single and smoker pregnant women. More recently, Golding et al. (2009) observed association between low omega‐3 fatty acid intake and high levels of depressive symptoms during pregnancy.
The evidence of the omega intake effect during pregnancy in PPD prevention are still scanty, although some studies sustain the hypothesis that there is a causal relationship between DHA deficiency and the occurrence of depressive episodes (Otto et al. 2003; Sontrop & Campbell 2006). Fatty acids of the omega‐3 family, especially the linolenic acid (ALA), precursor of the DHA (Sontrop & Campbell 2006), are utilized in the neuronal development of the fetus, a process that poses a high demand of this fatty acid. Thus, its levels decreased gradually throughout the pregnancy, especially in the last trimester, and remain low until the puerperium because of the release of DHA through the maternal milk (Otto et al. 2003; Freeman et al. 2006). The reduction in DHA levels in the maternal organism is even higher when associated to the low consumption of fish, such as sardines, salmon, mackerel and tuna – its main sources – as humans' conversion capacity from ALA to DHA is limited (Sontrop & Campbell 2006). The maternal nervous system might become unprotected because, in these conditions, there is a greater production of docosapentaenoic acid‐omega‐6 [synthesized from the linoleic acid (LA)], and the DHA‐omega‐3 is substituted by the former in the neuron's myelin sheath, yet without the same efficiency, altering the patterns of synapses and maybe causing PPD (Simopoulos 2002; Otto et al. 2003).
PPD consists of a public health concern. The prevalence of PPD in the present study was 26.4%, lower than the one reported by Cruz et al. (2005) in a study with women monitored by the Family Health Program of the Brazilian government (37.1%), greater than the one observed by Moraes et al. (2006) – 19.1% and similar to the one observed by Tannous et al. (2008) – 20.7% – in women residing in Porto Alegre, a metropolis in the south of Brazil. The prevalences observed in the present study and in other Brazilian samples were similar to the ones observed in samples of puerperae from other countries such as Bangladesh (Gausia et al. 2008) – 22%, and Mexico (Flores‐Quijano et al. 2008) – 24.5%, and greater than those observed in studies in developed countries (Boyce & Hichey 2005; Freeman et al. 2006). Despite having a frequency 5– 10 times greater than that of gestational diabetes and being as frequent as hypertension during pregnancy, the early detection of PPD is not part of the routine in most health institutions (Troppil et al. 2005).
It is important to link the high prevalences of PPD with one of the main characteristics of nutritional transition, in regard to changes in eating habits (Bermudez & Tucker 2003). In general, diets flow from a healthier pattern to a high consumption of industrialized foods, with higher proportions of LA in relation to the ALA (Martin et al. 2006). It is known, however, that before the phenomenon of nutritional transition, this ratio was much lower (around 1:1 or 2:1), with a higher consumption of fatty acids from the omega‐3 family. Several studies have already described this ratio in the population's diet, and the values from 10:1 to 20:1 were observed (Simopoulos 2002; Martin et al. 2006).
Recent studies based on randomized clinical trials (RCT) revealed the potential effect of the omega supplementation in the prevention of PPD and have already evaluated the effectiveness of supplementation from the omega‐6 and omega‐3 fatty acids family, during pregnancy, in the attempt of improving the concentrations of DHA in the mothers and in the umbilical cord of their children (Su et al. 2003; Groot et al. 2004; Freeman et al. 2008; Doornbos et al. 2009). The effect of an unbalanced dietary intake in the omega‐6/omega‐3 ratio in increasing the risk of PPD is in a similar line of evidence as the studies cited above, although we are aware that RCT are much less prone to confounding as our observational study, so, less importance should be given to observational studies rather than to RCT. The results from RCT were inconclusive and given the seriousness of PPD and the availability of proven treatment, it is very important to state, that at this moment, there are no evidence that PUFAs supplementation will alleviate PPD. This put the issue as a field of study that deserves to be further investigated. Until recently, the strongest evidence supporting a protective effect of omega‐3 PUFAs on depression came from RCTs where participants were administered omega‐3 PUFAs concurrently to antidepressants (Sontrop & Campbell 2006). However, this is refuted by a trial by Carney et al. (2009) demonstrating no effect of omega‐3 PUFAs in combination with sertaline on depression in patients with coronary heart disease. It is also important to consider that negative results may be less likely to be published, which may lead to a positive‐result publication bias.
Women with BMI below 18.5 kg/m2 presented four times greater PPD prevalences. We might hypothesize that women with low BMI may present a dietary composition different from that of the rest of the sample, poorer in fatty acids, leaving them unprotected from PPD. A study conducted in Pakistan to evaluate the persistence of prenatal depression during the first year of post‐partum did not find associations between low BMI and the continuance of depressive symptoms (Rahman & Creed 2007).
We observed in our data a slightly higher energy intake when compared with other studies (Siega‐Riz et al. 2002; Lacerda et al. 2007). This result might be part of a drawback of the FFQ, which consists in the potential for overestimation. We opted to employ an FFQ during the first pregnancy trimester because it comprises a group with very scarce information. This procedure can estimate individual's usual intake (Willett 1994), recent intake (Castro et al. 2006), past intake (Bunin et al. 2001) and intake from long (Sichieri & Everhart 1998) or short intervals (Bunin et al. 2001).
This study adds to a body of research suggesting a relationship between PUFAs and depression but have some limitations. The first one consists in the lack of dietary data in reasonable number of women for the second and third pregnancy trimesters. The reduced size of the sample due to a high loss to follow‐up is another limitation. Although the losses were excessive, significant differences were not observed between women who completed the follow‐up and those regarded as losses, except for those who smoke and for women married or with stable partnership. The fact that the study could not exclude women with pre‐existing depression during pregnancy is another limitation. In this regard, we need to be careful in drawing conclusions based on temporality and causation.
A greater prevalence of PPD was observed in Brazilian women with an unbalance in the dietary intake of the omega‐6 and omega‐3 ratio in the first trimester of pregnancy, even when this relationship was adjusted for confounding factors. PPD is a serious public health concern, both in developed and in developing countries such as Brazil, and multi‐professional preventive approaches are needed.
Source of funding
Funding for this study was provided by the Brazilian National Research Council through the Call for Projects CT SAÚDE/MCT/MS/CNPq, no. 505346/2004‐6. G. Kac is a research fellow for the Brazilian National Research Council.
Conflicts of interest
The authors declare that they have no conflicts of interest.
Acknowledgements
We would like to thank Professor Elisa Maria Lacerda Aquino (Institute of Nutrition Josué de Castro. Federal University of Rio de Janeiro) for her assistance on preparing the worksheet regarding the food composition database and Professor José Eduardo Corrente (Institute of Bioscience, University of São Paulo) for his assistance in the data analysis.
References
- Arnow B.A. & Constantino M.J. (2003) Effectiveness of psychotherapy and combination treatment for chronic depression. Journal of Clinical Psychology 59, 893–905. [DOI] [PubMed] [Google Scholar]
- Baker L., Cross S., Greaver L., Wei G. & Lewis R. (2005) Prevalence of postpartum depression in a Native American population. Maternal and Child Health Journal 9, 21–25. [DOI] [PubMed] [Google Scholar]
- Bermudez O.I. & Tucker K.L. (2003) Trends in dietary patterns of Latin American populations. Cadernos de Saúde Pública 19, S87–S99. [DOI] [PubMed] [Google Scholar]
- Boyce P. & Hichey A. (2005) Psychosocial risk factors to major depression after childbirth. Social Psychiatry and Psychiatric Epidemiology 40, 605–612. [DOI] [PubMed] [Google Scholar]
- Bunin G.R., Gyllstrom M.E., Brown J.E., Kahn E.B. & Kushi L.H. (2001) Recall of diet during a past pregnancy. American Journal of Epidemiology 154, 1136–1142. [DOI] [PubMed] [Google Scholar]
- Carney R.M., Freedland K.E., Rubin E.H., Rich M.W., Steinmeyer B.C. & Harris W.S. (2009) Omega‐3 augmentation of sertraline in treatment of depression in patients with coronary heart disease: a randomized controlled trial. Journal of the American Medical Association 302, 1651–1657. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Castro M.B.T., Kac G. & Sichieri R. (2006) Padrão de consumo alimentar em mulheres no pós‐parto atendidas em um centro municipal de saúde do Rio de Janeiro, Brasil [Dietary patterns among postpartum women treated at a municipal health center in Rio de Janeiro, Brazil]. Cadernos de Saúde Pública 22, 1159–1170. [DOI] [PubMed] [Google Scholar]
- Cox J.L., Holden J.M. & Sagovsky R. (1987) Detection of postnatal depression: development of the 10 Item Edinburgh Postnatal Depression Scale. British Journal of Psychiatry 150, 782–786. [DOI] [PubMed] [Google Scholar]
- Cruz E.B.S., Simões G.L. & Faisal‐Cruzi A. (2005) Rastreamento da depressão pós‐parto em mulheres atendidas pelo Programa de Saúde da Família [Post‐partum depression screening among women attended by the Family Health Program]. Revista Brasileira de Ginecologia e Obstetrícia 27, 181–188. [Google Scholar]
- Doornbos B., Goor S.A., Dijck‐Brouwer D.A.J., Schaafsma A., Korf J. & Muskiet F.A. (2009) Supplementation of a low dose of DHA or DHA+AA does not prevent peripartum depressive symptoms in a small population based sample. Progress in Neuro-Psychopharmacology & Biological Psychiatry 33, 49–52. [DOI] [PubMed] [Google Scholar]
- Flores‐Quijano M.E., Córdova A., Contreras‐Ramírez V., Farias‐Hernandes L., Tolentino M.C. & Casanueva E. (2008) Risk for postpartum depression, breastfeeding practices, and mammary gland permeability. Journal of Human Lactation 24, 50–57. [DOI] [PubMed] [Google Scholar]
- Freeman M.P., Hibbeln J.R., Wisner K.L., Brumbach B.H., Watchman M. & Gelenberg A.J. (2006) Randomized dose‐ranging pilot trial of omega‐3 fatty acids for postpartum depression. Acta Psychiatrica Scandinavica 113, 31–35. [DOI] [PubMed] [Google Scholar]
- Freeman M.P., Davis M., Sinha P., Wisner K.L., Hilbben J.R. & Gelenberg A.J. (2008) Omega‐3 fatty acids and supportive psychotherapy for perinatal depression: a randomized placebo‐controlled study. Journal of Affective Disorders 110, 142–148. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Gausia K., Fisher C. & Oosthuizen J. (2008) Magnitude and contributory factors of postnatal depression: a community‐based cohort study from a rural sub district of Bangladesh. Psychological Medicine 39, 999–1007. [DOI] [PubMed] [Google Scholar]
- Golding J., Steer C., Emmett P., Davis J.M. & Hibbeln J.R. (2009) High levels of depressive symptoms in pregnancy with low omega‐3 fatty acid intake from fish. Epidemiology 20, 1–6. [DOI] [PubMed] [Google Scholar]
- Groot R.H.M., Hornstra G., Houwelingen A.C. & Roumen F. (2004) Effect of a‐linolenic acid supplementation during pregnancy and neonatal polyunsaturated fatty acid status and pregnancy outcome. American Journal of Clinical Nutrition 79, 251–260. [DOI] [PubMed] [Google Scholar]
- Institute of Medicine (2004) Dietary Reference Intakes for Energy, Carbohydrate, Fiber, Fat, Fatty Acids, Cholesterol, Protein, and Amino Acids. National Academy Press: Washington, DC. [DOI] [PubMed] [Google Scholar]
- Institute of Medicine (2009) Weight Gain During Pregnancy: Reexamining the Guidelines Nutrition during Pregnancy. National Academy Press: Washington, DC. [Google Scholar]
- Jacka F.N., Pasco J.A., Henry M.J., Kotowicz M.A., Nicholson G.C. & Berk M. (2004) Dietary omega‐3 fatty acids and depression in a community sample. Nutritional Neuroscience 7, 101–106. [DOI] [PubMed] [Google Scholar]
- Lacerda E.M.A., Kac G., Cunha C.B. & Leal M.C. (2007) Consumo alimentar na gestação e no pós‐parto segundo cor da pele no município do Rio de Janeiro [Food intake during pregnancy and postpartum according to skin color in Rio de Janeiro, Brazil]. Revista de Saúde Pública 41, 985–994. [DOI] [PubMed] [Google Scholar]
- Lohman T.G., Roche A.F. & Martorell R. (1988) Anthropometric Standarization Reference Manual. Human Kinetics Books: Champaign, IL. [Google Scholar]
- Martin C.A., Almeida V.V., Ruiz M.R., Visentainer J.E.L., Matshushita M., Souza N.E. et al (2006) Ácidos graxos poliinsaturados ômega‐3 e ômega‐6: importância e ocorrência em alimentos [Omega‐3 and omega‐6 polyunsaturated fatty acids: importance and occurrence in foods]. Revista de Nutrição 19, 761–770. [Google Scholar]
- Moraes I.G.S., Pinheiro R.T., Silva R.A., Horta B.L., Sousa P.L.R. & Faria A.D. (2006) Prevalência de depressão pós‐parto e fatores associados [Prevalence of postpartum depression and associated factors]. Revista de Saúde Pública 40, 65–70. [DOI] [PubMed] [Google Scholar]
- Otto S.J., Groot R.H.M. & Hornstra G. (2003) Increased risk of postpartum depressive symptoms is associated with slower normalization after pregnancy of functional docosahexaenoic acid status. Prostaglandins, Leukotriens and Essential Fatty Acids 69, 237–239. [DOI] [PubMed] [Google Scholar]
- Rahman A. & Creed F. (2007) Outcome of prenatal depression and risk factors associated with persistence in the first postnatal year: prospective study from Rawalpindi, Pakistan. Journal of Affective Disorders 100, 115–121. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Rothman K.J., Greenland S. & Lash T.L. (2008) Modern Epidemiology. Lippincot Williams & Wilkins: Boston, MA. [Google Scholar]
- Santos M.F.S., Martins F.C. & Pasquali L. (1999) Escala de auto‐avaliação de depressão pós‐parto: estudo no Brasil [Post‐natal depression self‐rating scales: Brazilian study]. Revista de Psiquiatria Clínica 26, 90–95. [Google Scholar]
- Santos I.S., Matijasevich A., Tavares B.F., Barros A.J.D., Botelho I.P., Lapolli P. et al (2007) Validation of the Edinburgh Postnatal Depression Scale (EPDS) in a sample of mothers from the 2004 Pelotas birth cohort studies. Cadernos de Saúde Pública 23, 2577–2588. [DOI] [PubMed] [Google Scholar]
- Schotte C.K., Van Den B.B., De Doncker D., Claes S. & Cosyns P.A. (2006) Bio‐psychosocial model as a guide for psychoeducation and treatment of depression. Depress Anxiety 23, 312–324. [DOI] [PubMed] [Google Scholar]
- Sichieri R. & Everhart J.E. (1998) Validity of Brazilian food frequency questionnaire against dietary recalls and estimated energy intake. Nutrition Research 18, 1649–1659. [Google Scholar]
- Siega‐Riz A.M., Bodnar L.M. & Savitz D.A. (2002) What are pregnant women eating? Nutrient and food group differences by race. American Journal of Obstetrics and Gynecology 186, 480–486. [DOI] [PubMed] [Google Scholar]
- Simopoulos A.P. (2002) The importance of the ratio of omega‐6/omega‐3 essential fatty acids. Biomedicine & Pharmacotherapy 56, 365–379. [DOI] [PubMed] [Google Scholar]
- Sontrop J.M. (2007) Prenatal and Postnatal Depressive Symptoms in Relation to Fish Consumption and Intake of Omega‐3 Fatty Acids. Faculty of Graduate Studies, University of Western Ontario: London. [Google Scholar]
- Sontrop J. & Campbell M.K. (2006) ω‐3 poplyunsatured fatty acids and depression: a review of the evidence and a methodological critique. Preventive Medicine 42, 4–13. [DOI] [PubMed] [Google Scholar]
- Sontrop J., Avison W.R., Evers S.E., Speechley K.N. & Campbell M.K. (2008) Depressive symptoms during pregnancy in relation to fish consumption and intake of n‐3 polyunsaturated fatty acids. Paediatric and Perinatal Epidemiology 22, 389–399. [DOI] [PubMed] [Google Scholar]
- Su K.P., Huang S.Y., Chiu C.C. & Shen W.W. (2003) Omega‐3 fatty acids in major depressive disorder. A preliminary double‐blind, placebo‐controlled trial. European Neuropsychopharmacology 13, 267–271. [DOI] [PubMed] [Google Scholar]
- Tabela Brasileira de Composição de Alimentos (TACO) (2004) Núcleo de Estudos e Pesquisas em Alimentação. Universidade Estadual de Campinas [NEPA/Unicamp]. Tabela Brasileira de Composição de Alimentos [TACO]: versão 1. NEPA/Unicamp: São Paulo. [Google Scholar]
- Tam L.W., Newton R.P., Dern M. & Parry B.L. (2002) Screening women for postpartum depression at well baby visits: resistance encountered and recommendations. Archives of Women's Mental Health 5, 79–82. [DOI] [PubMed] [Google Scholar]
- Tannous L., Gigante L.P., Fuchs S.C. & Busnello E.D.A. (2008) Postnatal depression in Southern Brazil: prevalence and its demographic and socioeconomic determinants. BMC Psychiatry 8, 1–8. doi: 10.1186/1471‐244X‐8‐1. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Troppil J., Riutcel T.L. & Nalesnik S.W. (2005) Early intervention for perinatal depression. American Journal of Obstetric and Gynecology 192, 1446–1448. [DOI] [PubMed] [Google Scholar]
- United States Department of Agriculture (USDA) (2006) Agricultural Research Service USDA National Nutrient Database for Standard Reference, Release 19. Nutrient Data Laboratory Home Page, Available at: http://www.ars.usda.gov/ba/bhnrc/ndl
- Willett W.C. (1994) Future directions in the development of food‐frequency questionnaires. American Journal of Clinical Nutrition 59, 171–174. [DOI] [PubMed] [Google Scholar]
- Zinga D., Phillips S.D. & Born L. (2005) Depressão pós‐parto: sabemos os riscos mas podemos preveni‐la? [Postpartum depression: we know the risks, can it be prevented?]. Revista Brasileira de Psiquiatria 27, S56–S64. [DOI] [PubMed] [Google Scholar]
- Zou G. (2004) A modified poisson regression approach to prospective studies with binary data. American Journal of Epidemiology 159, 702–706. [DOI] [PubMed] [Google Scholar]