The WHO recommends exclusive breastfeeding (EBF) for the first 6 months of life(1). The infant should not receive any liquid (water, teas or juices), semi-solid or solid food before reaching 6 months of age, except medicines and supplements(2). In low- and middle-income countries, only 37 % of infants <6 months of age are exclusively breastfed, and this prevalence is even lower in high-income countries(Reference Victora, Bahl and Barros3). Data from four national surveys collected through 24-h dietary recalls showed that EBF prevalence in Brazilian infants <6 months of age has increased from 2·9 % in 1986 to 36·6 % in 2013, with a statistically significant increase in each decade up to 2006 and a posterior stabilisation until 2013(Reference Boccolini, Boccolini and Monteiro4). Similarly, among infants <2 months of age, there was an increase of 44 % in EBF prevalence between 1986 and 2006, and a reduction of 0·3 % between 2006 and 2013(Reference Boccolini, Boccolini and Monteiro4). In Pelotas, a city in southern Brazil, data from three population-based birth cohorts showed that the prevalence of EBF at 3 months of age increased from 7·0 % in 1993 to 45·0 % in 2015(Reference Santos, Barros and Horta5).
Breastfeeding brings numerous benefits to the health of both mother and child(Reference Victora, Bahl and Barros3,Reference Victora, Horta and Loret de Mola6–Reference Marseglia, Manti and D’Angelo11) . The UNICEF and WHO recommend that countries should monitor breastfeeding rates at least every 5 years in order to detect negative trends and identify the need to direct resources for the strengthening of policies and programmes to promote breastfeeding(12). Large national health surveys, such as the Multiple Indicator Cluster Survey (MICS)(13) and the Demographic and Health Survey (DHS)(14), which are responsible for collecting and disseminating accurate national nutrition and health data in low- and middle-income countries, as well as country-specific surveys, such as the National Health and Nutrition Survey (PNSN)(15) in Brazil, are applied in the form of interviews, with the aid of structured questionnaires, with information about EBF usually obtained by the use of 24-h dietary recalls.
Despite the importance of the accuracy of information collected and the availability of studies that have assessed the validity of maternal recall of breastfeeding duration(Reference Li, Scanlon and Serdula16–Reference Barbosa, Oliveira and Zandonade18), there are only few studies that have assessed the validity of maternal recall of duration of EBF(Reference Burnham, Buczek and Braun19–Reference Mazariegos, Slater and Ramirez-Zea21). Thus, the current study aimed to assess the validity of maternal information, obtained 12 months after childbirth, on the practice of EBF at 3 months of age among mothers of the Pelotas 2004 Birth Cohort in southern Brazil.
Methods
This study was conducted with data collected at birth and at 3- and 12-month follow-ups of the Pelotas 2004 Birth Cohort. All newborns in 2004, with at least 500 g of birthweight or 20 weeks of gestational age, whose mothers lived in the urban area of the municipality and in Jardim América, a neighbourhood adjacent to Pelotas belonging to a neighbouring municipality (Capão do Leão), were enrolled to the cohort. With a refusal rate of 0·8 %, 4231 newborns from the five maternity hospitals in the city were included in the cohort. Follow-up rates at 3 and 12 months were 95·7 and 94·3 %, respectively. Further methodological details about the cohort are available in previous publications(Reference Santos, Barros and Matijasevich22,Reference Santos, Barros and Matijasevich23) .
For the present investigation, only newborns of single births were considered. A total of 3700 mothers provided information about the patterns of breastfeeding at 3- and 12-month follow-ups. Maternal report on EBF obtained at 3 months of age was used as the gold standard for validation analysis. Initially, mothers were asked if the infant received breast milk. Then the mothers were questioned about the introduction of other foods by means of the following questions: ‘When has <INFANT’S NAME> begun to eat regularly (being presented one by one the foods in the following list): cow’s milk, powdered milk, tea, juice, water, mashed fruits, soup, mashed vegetables, porridge, egg and other food?’. Regular consumption was considered when the food had been offered at least twice, on different days, in the last week. The age of introduction of each food was recorded in months and days. EBF was recorded when breastfed children were not fed any other liquid, semi-solid or solid foods up to 3 months of age.
The variable to be validated was obtained at 12 months after childbirth when the mother was asked about the age of food introduction. The question used was: ‘Now, I am going to tell you a list of liquids and foods and let me know if you have started giving them to <INFANT’S NAME>. When I say “started”, I want to know if <INFANT’S NAME> receives or received that liquid or food every day, or almost every day of the week. If you have already started giving it, I want to know when you have started: cow’s milk, powdered milk, tea, juice, mashed fruit, soup, mashed vegetables, porridge, yogurt, bread or cracker, egg (yolk), egg (white), meat, bean broth, bean grain, pasta, legumes/vegetables (in pieces), and other foods’. The age of food introduction was collected in months and days from the first ingestion of each food. When the mother reported that no other liquid, semi-solid or solid food but breastmilk was introduced up to the age of 3 months, the breastfeeding pattern at 3 months was recorded as EBF.
In order to characterise the sample, the following information about the infant at birth was used: gestational age (complete weeks), sex (male or female) and weight (<2500 or ≥2500 g). In addition, the following maternal characteristics were used: age at delivery (subsequently categorised as ≤19, 20–29 or 30–46 years); self-reported skin colour (white, black or other); completed years of schooling at delivery (later categorised into 0–4, 5–8, 9–11 and ≥12 years); socioeconomic level based on the criteria of the Research Companies Brazilian Association (ABEP), which uses schoolarity of the family head and household assets, categorised into economic classes A/B (wealthiest), C or D/E (poorest)(24); parity (number of children born alive or dead); maternal smoking during pregnancy (at least one cigarette per day in any trimester of gestation); marital status (without or with partner); self-reported depressive symptoms during pregnancy (‘During pregnancy, did you have depression or nervous problem’? no; yes, treated; yes, untreated); type of delivery (vaginal or C-section); previous experience with breastfeeding (yes or no); and number of antenatal consultations. At the 3-month follow-up, the mothers were asked if they had returned to work after childbirth (yes or no) and if they were currently smokers (yes or no).
Statistical analyses were conducted in Stata, version 14.2 (StataCorp). To verify the sensitivity, specificity, positive (PPV) and negative predictive values (NPV), the statistical command ‘diagt’ was used. These parameters were first calculated for the entire sample and then after stratifying the sample according to the independent variables. Accuracy was calculated based on the following formula: (true positives + true negatives)/whole sample. The 95 % CI was calculated for all the estimated parameters. True and false positive and negative values were defined as follows:
True positives: mothers who reported EBF both at the 3-month (gold standard) and 12-month follow-ups;
False negatives: mothers who reported EBF at the 3-month follow-up, but in the 12-month follow-up they reported having introduced any liquid, semi-solid or solid food before the age of 3 months;
True negatives: mothers not reporting EBF at the 3- and 12-month follow-ups;
False positives: mothers not reporting EBF at the 3-month follow-up as well as, at the 12-month follow-up, not having introduced any liquid, semi-solid or solid food before the age of 3 months.
Results
At both follow-ups (3 and 12 months), the majority of mothers were between 20 and 29 years of age; self-reported white skin colour; lived with a partner; had two or more children; were non-smokers during or after pregnancy; had previous experience with breastfeeding; attended eight or more antenatal consultations; had a vaginal delivery; and had not returned to work after childbirth (Table 1). Approximately 20 % of the mothers reported depressive symptoms during pregnancy, and most received no treatment for depression. The proportion of mothers with lower years of schooling (<9 years) was higher at the 12-month (52·2 %) than at the 3-month follow-up (42·7 %), whereas the proportion from socioeconomic classes D/E was higher at the 12-month (44·3 %) than at the 3-month follow-up (38·6 %) (Table 1). As for infant characteristics (Table 1), the prevalence of low birthweight (LBW) (<2500 g) and preterm birth were 5·6 and 7·8 %, respectively, at the 3-month follow-up, and 9·5 and 12·6 % at the 12-month follow-up.
Table 1 Maternal and child characteristics of the Pelotas 2004 Birth Cohort at 3 and 12 months postpartum – Pelotas, Rio Grande do Sul, Brazil
EBF, exclusive breastfeeding; SES, socioeconomic status; ABEP, Research Companies Brazilian Association.
* Mothers who reported about exclusive breastfeeding at 3 and 12 months postpartum.
The prevalence of EBF by maternal report at the 3-month follow-up (gold standard) was 27·8 % (95 % CI 26·4, 29·3). At the 12-month follow-up, the maternal report provided a prevalence of EBF at 3 months of 49·0 % (95 % CI 47·4, 50·6) (data not presented in table). Maternal recall of EBF at 3 months at the end of the first year after childbirth showed a sensitivity of 98·3 % (95 % CI 97·4, 99·0), specificity of 70·0 % (95 % CI 68·2, 71·7) and accuracy of 77·9 % (95 % CI 76·6, 79·2). PPV and NPV were 55·8 % (95 % CI 53·4, 58·1) and 99·1 % (95 % CI 98·6, 99·5), respectively (Table 2).
Table 2 Validity of maternal recall on exclusive breastfeeding (EBF) at 3 months, obtained at 12 months after childbirth, compared with the gold standard (EBF at 3 months, obtained at 3-month follow-up) – Pelotas 2004 Birth Cohort, Rio Grande do Sul, Brazil
PPV, positive predictive value; NPV, negative predictive value.
* PPV and NPV at an EBF prevalence of 27·8 %.
The parameters of validation estimated according to maternal and newborn characteristics are described in Table 3. Among the different variables, sensitivity remained nearly 98·0 %, while specificity ranged from 66·4 to 81·8 %. PPV was higher among non-adolescent mothers, those who lived with a partner (58·1 %), those with ≥12 years of formal education (74·4 %), those who belonged to socioeconomic level A/B (69·4 %), those who did not smoke during pregnancy (59·4 %) or at the 3-month follow-up (59·3 %), and those who attended eight or more antenatal care consultations (61·6 %), as well as among mothers of infants with birth weight ≥2500 g (57·0 %) or ≥37 weeks of gestational age (57·7 %).
Table 3 Validation of maternal recall on exclusive breastfeeding at 3 months, obtained at 12 months after childbirth, according to maternal and infant characteristics – Pelotas 2004 Birth Cohort, Rio Grande do Sul, Brazil (n 3700)
PPV, positive predictive value; NPV, negative predictive value; SES, socioeconomic status; ABEP, Research Companies Brazilian Association.
Discussion
The present study tested the validity of maternal recall on EBF at 3 months of age obtained 12 months after childbirth. The probability of maternal recall at the end of first year of child’s life correctly identifying mothers who reported EBF at 3 months after delivery was 98·3 %. The specificity at 12 months was lower, with a 30 % rate of false-positive responses, possibly indicating that mothers – despite being aware that EBF is advocated up to the first 6 months of infant’s life – reported having offered it even when they did not. The accuracy of almost 80 % is less than perfect or substantial but can be considered moderate(Reference Fischer, Bachmann and Jaeschke25), indicating that maternal recall on EBF at 3 months of age collected after 12 months of childbirth is a valid measure deserving application in surveillance studies.
The measurement of EBF is complex, as rates may vary according to the definition, measurement period, instrument of assessment and even child’s age(Reference Khanal, Lee and Scott26). Although maternal recall on breastfeeding has been widely used in research, its validity and reliability have been questioned, because it is a memory-dependent information(Reference Barbosa, Oliveira and Zandonade18). However, a literature review of articles published between 1966 and 2003 in English has found that maternal report on initiation and duration of any breastfeeding would be accurate and reliable, although the recall of breastfeeding duration might become less accurate as the period of recall increases(Reference Li, Scanlon and Serdula16). On the other hand, the validity and reliability of maternal recall of EBF duration might be less accurate(Reference Li, Scanlon and Serdula16).
More recently, a survey conducted in the United States has shown that maternal recall on any breastfeeding duration in the first year of child’s life verified 6 years after childbirth had a high intraclass correlation coefficient (ICC = 0·84) against the gold standard obtained monthly by maternal report until 12 months of age(Reference Amissah, Kancherla and Ko17). In Brazil, researchers have found a high agreement (ICC = 0·92) between maternal recall on any breastfeeding duration collected when the child was 2 years of age compared with reports collected monthly in the first 3 months of child’s life (gold standard)(Reference Barbosa, Oliveira and Zandonade18), which is consistent with our findings.
In terms of EBF, an investigation has found that at 2 years after delivery, mothers overestimated EBF up to 6 months of age as compared with the gold standard measured weekly up to 1 month of life and monthly up to the age of 6 months(Reference Burnham, Buczek and Braun19). A study in Guatemala tested two self-reported instruments to assess EBF in infants aged 3 months using the dose-to-mother deuterium oxide turnover (DMDOT) technique as the reference method. The prevalence of EBF was 50 % according to the current feeding practice reported, 61 % by the 24-h dietary recall and only 36 % when using DMDOT. The sensitivity to detect EBF from the mother’s report was 92 % (95 % CI 62, 99), but from the 24-h dietary recall was 100 % (95 % CI 72, 100). The specificity for both methods was lower – 74 % (95 % CI 51, 89) for reported current feeding practice and 61 % (95 % CI 39, 79) for the 24-h dietary recall(Reference Mazariegos, Slater and Ramirez-Zea21).
In Sri Lanka, two methods to collect retrospective data on EBF up to 6 months of age were tested: (1) based on an event calendar (date of introduction, frequency of use and quantity of specific food items) and (2) mother’s recall at 9 months after childbirth(Reference Agampodi, Fernando and Dharmaratne20). The gold standard was obtained through prospective data collected since birth. The authors have reported 100 % sensitivity for both methods, specificity of 26·2 % (95 % CI 17·9, 36·8) for an event calendar, and specificity of 75·0 % (95 % CI 64·5, 83·2) for mother’s recall at 9 months after childbirth(Reference Agampodi, Fernando and Dharmaratne20). The way the questions are framed may affect the accuracy of response from the mother. Asking the time of complementary feeding introduction (as in our study) may provide different results than asking the duration of EBF.
Some studies have reported differences in the validity of maternal recall on breastfeeding according to sociodemographic characteristics(Reference Amissah, Kancherla and Ko17,Reference Burnham, Buczek and Braun19) . In our study, however, the sensitivity remained stable and around 98 % at the different maternal and child characteristics evaluated. On the contrary, PPV varied according to maternal age, economic class, schooling, smoking, number of antenatal care consultations attended, birthweight and gestational age, thus reflecting differences in the prevalence of EBF among these groups of mothers. Youngest mothers, those under socioeconomically disadvantaged conditions as well as smokers presented the lowest rates of EBF at 3 months. Such findings are in agreement with the results of Amissah et al. from the USA(Reference Amissah, Kancherla and Ko17).
Our study has few strengths and limitations. The main strength is it being a population-based study. In addition, the gold standard information was collected near the moment of its occurrence. The mean age of infants when the follow-up occurred was 3·0 (sd 0·1) months, thus reducing the probability of recall bias. Among the limitations, it is possible that mothers taking part in a birth cohort study tend to remember more promptly of the events that occurred in the child’s life than non-participating mothers, thus compromising the external validity of our results. Another point to consider is the information bias arising from the maternal knowledge about the recommendation of EBF for the first 6 months of infant’s life. As a result, false-positive responses may be present in our findings at both follow-ups, thus increasing the estimated prevalence of EBF at 3 months and decreasing the specificity of maternal recall at 12 months.
Conclusion
This study contributes valuable data to epidemiological research on maternal recall of EBF. Information relating to EBF at 3 months of age obtained from mothers 12 months after childbirth showed almost 80 % accuracy. The information remained valid even after the sample was stratified by newborn weight, gestational age and several maternal characteristics.
Acknowledgements
Acknowledgements: This article is based on the data reported by the study ‘Pelotas 2004 Birth Cohort’ conducted by Postgraduate Programme in Epidemiology at Universidade Federal de Pelotas, with the collaboration of the Brazilian Public Health Association (ABRASCO). The WHO, the National Support Program for Centers of Excellence (PRONEX), the Brazilian National Research Council (CNPq), the Brazilian Ministry of Health and the Children’s Mission supported the 2004 Birth Cohort study. Financial support: This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior – Brazil (CAPES), finance code 001. A.M., A.J.D.B. and I.S.S. are supported by CNPq. Conflict of interest: The authors have no conflicts of interest directly relevant to the content of this article. Authorship: I.S.S., A.M., T.M.S. and B.C.S. conceived and designed the study. B.C.S., B.O.C.P., D.D.G., F.S.S., M.D.D., P.L., P.W. and U.M.R. analysed, interpreted the data and drafted the manuscript. I.S.S., A.M., A.J.D.B., F.C.B. and T.M.S. critically revised the manuscript. All authors approved the final version of the manuscript for submission. Ethics of human subject participation: This study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving study participants were approved by the Research Ethics Committee of the Faculty of Medicine of the Federal University of Pelotas. All interviews were performed only after obtaining written informed consent from participating mothers.
