SEARCH

SEARCH BY CITATION

Keywords:

  • birthweight;
  • reliability;
  • validity;
  • uganda
  • Poids à la naissance;
  • fiabilité;
  • validité;
  • Ouganda
  • peso al nacer;
  • Fiabilidad;
  • Validez;
  • Uganda

Abstract

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Objectives  To assess the reliability of maternally recalled birthweight and size in Entebbe, Uganda.

Methods  The study population comprised 404 mothers, who were participants in the Entebbe Mother and Baby Study (EMaBS). Mothers were recruited to EMaBS during antenatal care, maternal characteristics were recorded during pregnancy, and birthweight was recorded at delivery. Four to seven years after delivery, mothers were asked to recall the child’s birthweight and size. Their responses were compared with the birthweight recorded in the EMaBS database.

Results  Of 404 interviewed mothers, 303 (75%) were able to give an estimate of birthweight and for 265 of these EMaBS data on recorded birthweights were available. Women who were educated and whose children had low birth order were more likely to be able to give an estimate: 37 (14%) recalled the exact recorded birthweight; a further 52 (20%) were accurate to within 0.1 kg of the recorded weight. On average, mothers overestimated birthweight by 0.06 kg (95% CI: 0.00–0.13 kg, P = 0.04). Recalled and recorded birthweights showed moderate agreement with an intraclass correlation coefficient of 0.64. Four hundered mothers gave an estimate of birth size: the sensitivity and specificity of recalled birth size for classifying low birthweight were 76% (95% CI: 50–93%) and 70% (95% CI: 65–75%), respectively.

Conclusions  Mothers’ recall of birthweight was not precise but in absence of other data, recall of birthweight and size may have some value in epidemiological studies in these settings.

Objectifs:  Evaluer la fiabilité du poids et de la taille à la naissance dont se souviennent les mères à Entebbe, en Ouganda.

Méthodes:  La population étudiée comprenait 404 mères qui participaient à l’Etude Mère et Bébé de Entebbe (EMaBS). Les mères ont été recrutées pour EMaBS pendant les soins prénataux, les caractéristiques maternelles ont été enregistrées pendant la grossesse et le poids de naissance a été enregistréà l’accouchement. 4 à 7 ans après l’accouchement, il a été demandé aux mères de se rappeler le poids et la taille de naissance de l’enfant. Leurs réponses ont été comparées aux poids à la naissance enregistrés dans la base de données EMaBS.

Résultats:  Sur 404 mères interrogées, 303 (75%) étaient en mesure de donner une estimation du poids à la naissance et pour 265 de celles-ci des données enregistrées sur EMaBS pour le poids à la naissance étaient disponibles. Les femmes instruites et celles dont les enfants avaient un poids de naissance faible étaient plus susceptibles d’être en mesure de donner une estimation: 37 (14%) se sont rappelées du poids de naissance exact enregistré; 52 (20%) étaient exacts à 0,1 kg près du poids enregistré. En moyenne, les mères ont surestimé le poids de naissance de 0,06 kg (IC95%: 0.00–0.13 kg, P = 0.04). Les poids à la naissance souvenus et enregistrés ont montré une concordance modérée avec un coefficient de corrélation intra classe de 0.64. 400 mères ont donné une estimation de la taille à la naissance: la sensibilité et la spécificité de la taille à la naissance souvenus pour la classification du poids à la naissance était de 76% (IC95%: 50–93%) et 70% (95% CI: 65–75%), respectivement.

Conclusions:  Le souvenir des mères sur le poids à la naissance n’était pas précis, mais en l’absence d’autres données, le souvenir du poids et de la taille à la naissance peut avoir une certaine valeur dans les études épidémiologiques dans ces zones.

Objetivos:  Evaluar la fiabilidad del recuerdo materno del peso y el tamaño al nacer en Entebbe, Uganda.

Métodos:  La población de estudio incluyó a 404 madres, participantes del estudio de madres e hijos de Entebbe (Entebbe Mother and Baby Study - EMaBS). Las madres fueron reclutadas en EMaBS durante el seguimiento prenatal, se recogieron datos sobre las características maternas durante el embarazo, y el peso al nacer se registró en el momento del parto. Entre 4-7 años después del parto, se pidió a las madres que recordaran el peso y la talla de su hijo al nacer. Sus respuestas se compararon con el peso al nacer registrado en las bases de datos de EMaBS.

Resultados:  De 404 madres entrevistadas, 303 (75%) fueron capaces de dar un peso al nacer aproximado, y de estas, los datos registrados sobre el peso al nacer estaban disponibles en EMaBS para 265. Las mujeres que tenían educación y cuyos niños habían tenido un bajo peso al nacer tenían una mayor probabilidad de dar una aproximación: 37 (14%) recordaban el peso al nacer exactamente igual al registrado; otras 52 (20%) eran precisas dentro de 0.1 kg del peso registrado. En promedio, las madres sobreestimaban el peso al nacer en 0.06kg (95% IC: 0.00 a 0.13 kg, P = 0.04). Los pesos al nacer recordados y registrados tenían una concordancia moderada con un coeficiente de correlación interclase del 0.64. 400 madres dieron una aproximación del tamaño al nacer: la sensibilidad y especificidad del tamaño al nacer recordado, para clasificar un bajo peso al nacer, eran del 76% (95% IC: 50–93%) y 70% (95% IC: 65–75%), respectivamente.

Conclusiones:  El recuerdo de las madres del peso al nacer no era preciso, pero en ausencia de otros datos, el recuerdo del peso y tamaño al nacer podría tener algún valor en estudio epidemiológicos en estos emplazamientos.


Introduction

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Birthweight is an important predictor of future growth patterns (Hindmarsh et al. 2008) and of mortality and morbidities later in life (Barker et al. 1989; Gofin et al. 2000; Godfrey & Barker 2001). It is also vital in assessment of population health status (Gofin et al. 2000). Records of birthweight are seldom available to researchers investigating disease aetiology in developing countries (Walton et al. 2000; Catov et al. 2006). Maternally recalled birthweight is often the only available source of birthweight information for use in retrospective epidemiological studies, and this may introduce information bias.

In developed countries, several studies have examined concordance between the birthweight recalled by the mother and the recorded birthweight and have shown that maternally recalled birthweight is a good proxy for recorded weight (Gofin et al. 2000; Walton et al. 2000; Tate et al. 2005; Van Gelder & Roeleveld 2011). However, in developing countries, there is limited information on accuracy of maternally reported birthweight and birth size. A study in Brazil reported that mothers accurately recalled birthweight 12 months after delivery but that this accuracy decreased with time after birth (Araújo et al. 2007). In Taiwan, mothers over reported birthweight even within a few months after delivery (Li et al. 2006). In Cameroon, maternal recall was very poor (Mbuagbaw & Gofin 2010), whereas in Kenya, mother’s recall of low birthweight (<2.5 kg) was very good (Mung’ala-Odera & Newton 2001). Given this variability, we have taken the opportunity provided by our birth cohort (the Entebbe Mother and Baby Study; EMaBS) to assess the reliability of maternally recalled birthweight and the validity of maternally recalled birth size and their determinants in Uganda.

Methods

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Between April 2003 and November 2005, the EMaBS birth cohort was established to investigate the effect of antihelminthic treatment during pregnancy on the offspring’s response to immunisation and on susceptibility to infectious diseases. Two thousand five hundred and seven women attending antenatal care at Entebbe hospital were enrolled into the trial. Full details of the trial design and procedures are described elsewhere (Elliott et al. 2007). Babies delivered in Entebbe Hospital were weighed immediately after birth using scales graduated in 0.1 kg units (Fazzini SRL, Vimodrone, Italy) and recorded to the nearest 0.1 kg. For babies delivered elsewhere, birthweight was recorded as it appeared on the child health card. Birthweight was available for 1964 of the 2345 live births in the cohort (Ndibazza et al. 2010). The children are currently being followed up, with regular visits to the clinic both for scheduled and illness visits.

From 21 September to 8 December 2010, we interviewed sequentially the mother of each child who attended the study clinic. Children were 4–7 years old at the time of interview. Mothers were asked whether they still possessed the child health card showing the birthweight record of the child. Without reference to the health card, mothers were asked to recall the birthweight of their child and to give a categorical estimate of the birth size of the child (small, normal or large). These data were linked with antenatal and delivery information from the EMaBS database, thus allowing for comparison of recalled and recorded birthweight.

Reliability of maternal recall of birthweight was assessed by calculating the mean difference between recalled birthweight and recorded birthweight and conducting a paired t-test. The intraclass correlation coefficient was calculated as a measure of the agreement between reported and recalled birthweight. Recorded birthweight was categorised into low birthweight (<2.5 kg), normal birthweight (2.5–4.0 kg) and large birthweight (>4.0 kg). Sensitivity and specificity of a mother’s perception of small birth size in detecting low birthweight babies and of a mother’s perception of large birth size in detecting large birthweight babies were calculated.

Logistic regression was used to examine factors associated with mother’s recall of birthweight. Two binary outcomes were considered: first, ability to recall any numerical estimate of birthweight; second, ability to recall birthweight to within 0.1 kg of the recorded weight. Explanatory factors considered were mother’s age, education and socio-economic status, child’s birth order, gender, recorded birthweight and the child’s age at the time of this study. Multivariable analysis was used to adjust for the possible confounding effect of factors that were crudely associated with the outcome.

Results

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

Between 21 September and 8 December 2010, 404 mothers were interviewed. Mothers who were interviewed were on average slightly older, were less likely to be primigravidae and had attended more routine study visits, than the remaining mothers enrolled in the EMaBS cohort whose children did not attend the clinic during this study period. Their children were less likely to have been born at home. Of the 404 children whose mothers were interviewed, 204 (51%) were male and 200 (49%) were female, with a mean age of 5.7 years (range, 4.5–7.5 years). One hundred and ninety-seven (49%) had attended the clinic because of illness, and 207 (51%) had attended for a routine visit. The average age of mother at the time of delivery of the study baby was 25 years (range, 15–45 years), and 356 (88%) said they still had the child health card. Three hundred and three (75%) of the women were able to give an estimate of birthweight; of the remaining 101 women who were unable to give an estimate of birthweight, 11 had delivered at home, and thus, birthweight is unlikely to have been measured (although one woman who delivered at home did give an estimate of birthweight). Characteristics of those who recalled and did not recall birthweight are shown in Table 1. Women who gave an estimate for birthweight were more likely to be educated, and their children were more likely to be of low birth order. There was a crude association between younger maternal age and ability to give an estimate of birthweight, but maternal age and birth order were associated, and multivariable analyses suggested that the association between age and ability to give an estimate of birthweight was mediated through birth order (Table 1).

Table 1. Comparison of maternal and child characteristics between mothers who gave an estimate of birthweight and those who did not
CharacteristicTotal mothers interviewed N = 404Number (%) mothers who estimated birthweightCrude OR (95% CI) P-valueAdjusted OR (95% CI)* P-value*
  1. *Multivariable model included mother’s age at birth of child, mother’s education and birth order, the adjusted estimate for mother’s age is interpreted as the independent effect of age that does not act through birth order, controlling for mother’s education.

  2. †One missing value.

  3. ‡Restricted to the 333 children for whom a record of birthweight was available.

Mother’s age at birth of child (years)
 15–197259 (82%)10.004 [trend]10.95
 20–24153121 (79%)0.83 (0.41–1.70)1.46 (0.63–3.39)
 25–2910073 (73%)0.60 (0.28–1.26)1.50 (0.54–4.13)
 30+7950 (63%)0.38 (0.18–0.81)1.19 (0.37–3.79)
Mother’s education†
 None133 (23%)0.12 (0.03–0.48)<0.0010.11 (0.03–0.44)0.002
 Primary183132 (72%)11
 Secondary169138 (82%)1.72 (1.04–2.85)1.36 (0.80–2.31)
 Tertiary3829 (76%)1.24 (0.55–2.81)0.97 (0.41–2.30)
Birth order
 18776 (87%)1<0.001 [trend]10.02 [trend]
 210380 (78%)0.50 (0.23–1.10)0.51 (0.21–1.22)
 3–413198 (75%)0.43 (0.20–0.91)0.40 (0.15–1.05)
 ≥58349 (59%)0.21 (0.10–0.45)0.23 (0.07–0.74)
Sex of child
 Male204149 (73%)10.36  
 Female200154 (77%)1.24 (0.79–1.94)
Age of child (years)
 48462 (74%)10.68  
 5175133 (76%)1.12 (0.62–2.04)
 611991 (76%)1.15 (0.61–2.20)
 72617 (65%)0.67 (0.26–1.72)
Recorded birthweight (kg)‡
 <2.51713 (76%)10.94  
 2.5–4302241 (80%)1.22 (0.38–3.86)
 >41411 (79%)1.13 (0.21–6.17)

Analysis of the agreement between recorded and maternally recalled birthweight was restricted to 333 (82%) of the 404 interviewed mothers who gave birth in Entebbe hospital. Sixty-eight of these women were unable to recall their child’s birthweight, leaving 265 mother–child pairs with both a recalled and a recorded birthweight. The mean (standard deviation; range) of recalled and recorded birthweights were 3.28 kg (0.68 kg; 1.50–6.40 kg) and 3.21 kg (0.50 kg; 1.50–5.50 kg), respectively: on average, mothers overestimated the birthweight by 0.06 kg (95% CI: 0.00–0.13 kg, P = 0.04, paired t-test). Agreement between recalled and recorded birthweight was moderate (intraclass correlation coefficient 0.64, Figure 1). Only 37 (14%) of mothers recalled their child’s birthweight exactly as recorded; a further 52 (20%) recalled the birthweight to within 0.10 kg of the recorded value. Of the covariates considered, none was associated with accurate recall, or with the difference between recalled and recorded birthweight.

image

Figure 1.  The relationship between recalled and recorded birth weights.

Download figure to PowerPoint

All but four of the 404 mothers gave a response regarding the size of the baby at birth. Thirty-five (9%) described their baby as large, 237 (59%) as normal and 128 (32%) as small. Reported size was associated with recorded birthweight (P < 0.001): the mean (SD) recorded birthweights for the recalled large, normal and small size groups were 3.73 kg (0.60 kg), 3.31 kg (0.40 kg) and 2.92 kg (0.49 kg), respectively. The sensitivity and specificity of mother’s recall of small size for low birthweight babies were 76% (95% CI: 50–93%) and 70% (95% CI: 65–75%), respectively, while the sensitivity and specificity for detecting large birthweight babies were 57% (29–82%) and 94% (91–97%), respectively.

Discussion

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

This study from Uganda is one of a very small number of studies in sub-Saharan Africa to have assessed mothers’ recall of birthweight and birth size. Many mothers (25%) could not recall any numerical estimate of birthweight but almost all gave an approximate birth size. Those who gave an estimate of birthweight were not very accurate as to the precise figure, but there was moderate agreement between recalled and recorded birthweight. This was consistent with findings from the Netherlands by Jaspers et al. (2010) who found maternally recalled birthweight was not very accurate. Studies from the UK showed better maternal recall, with over 92% recalling birthweight to within 0.1 kg of recorded birthweight (Tate et al. 2005) and 85% to within 0.22 kg of recorded birth (Walton et al. 2000).

Studies by Rice et al. (2007), Gofin et al. (2000), O’Sullivan et al. (2000), Tate et al. (2005), Walton et al. (2000) and Jaspers et al. (2010) reported no mean difference between mothers’ recalled birthweight and recorded birthweight. In this study, we found there was a tendency of mothers to overestimate birthweight, and this finding was consistent with results from Taiwan reported by Li et al. (2006), but in contrast to findings from Denmark where mothers underestimated the birthweight (Adegboye & Heitmann 2008).

In this community, mothers’ concerns at birth are viability, absence of congenital anomalies and child’s sex and so little emphasis is placed on birthweight. Therefore, it is perhaps not surprising that many mothers in our study were unable to give a numerical estimate of weight. Ability to recall any numerical estimate of birthweight increased with education and decreased with the birth order of the child. However, we did not identify any maternal or child factors that were associated with accurate recall of birthweight. Similar results were reported by O’Sullivan et al. (2000), McCormick and Brooks-gunn (1999) and Olson et al. (1997). However, Tate et al. (2005) found that birth order, birthweight and socio-economic status influenced accurate recall of birthweight.

Mothers’ recall of birth size was more robust, allowing classification of babies as low birthweight with sensitivity and specificity of 76% and 70%, respectively (compared to a sensitivity of 60% and specificity of 93% seen in Cameroon (Mbuagbaw & Gofin 2010).

Possible sources of error and bias in this study were considered. Only one child per mother was enrolled into the EMaBS, and to be part of this analysis, the mother and child pair had attended the study clinic together; thus, it is unlikely that the mother gave data on any non-EMaBS sibling. Mothers who participated in this study were on average slightly older and had attended more routine study visits than members of the EMaBS cohort who were not included; and only 3% of their children had been born at home compared to EMaBS (Ndibazza et al. 2010) and community (Tann et al. 2007) estimates of 11%. Thus, some elements of our source population are under-represented. However, the differences in characteristics are not large; thus, any impact is likely to be minimal.

In summary, mother’s recall of birthweight was not precise but in absence of other data, mother’s recalled birthweight and birth size have some value for epidemiological studies, as long as it is not crucial to know the exact birthweight. Clinicians and researchers using maternally recalled birthweight should be cautious when using such information. Recalled birth size should be used only where recorded birthweight is not available.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References

We would like to thank the study participants and the staff of Entebbe hospital and MRC/UVRI Uganda Research Unit on AIDS. The work was funded by the Wellcome Trust.

References

  1. Top of page
  2. Abstract
  3. Introduction
  4. Methods
  5. Results
  6. Discussion
  7. Acknowledgements
  8. References
  • Adegboye AR & Heitmann B (2008) Accuracy and correlates of maternal recall of birthweight and gestational age. BJOG115, 886893.
  • Araújo CLP, Dutra CL C & Hallal P C (2007) Validity of maternal report on birth weight 11 years after delivery : the 1993 Pelotas Birth Cohort Study, Rio Grande do Sul State, Brazil. Cad, Saude Publica, Rio de Janeiro23, 24212427.
  • Barker DJ, Osmond C, Winter PD, Margetts B & Simmonds SJ (1989) Weight in infancy and death from ischaemic heart disease. Lancet334, 577580.
  • Catov JM, Newman AB, Kelsey SF et al. (2006) Accuracy and Reliability of Maternal Recall of Infant Birth Weight Among Older Women. Annals of Epidemiology16, 429431.
  • Elliott AM, Kizza M, Quigley MA et al. (2007) The impact of helminths on the response to immunization and on the incidence of infection and disease in childhood in Uganda: design of a randomized, double-blind, placebo-controlled, factorial trial of deworming interventions delivered in pregnancy and early childhood. Clinical trial4, 4257.
  • Godfrey KM & Barker DJ (2001) Fetal programming and adult health. Public Health4, 1624.
  • Gofin R, Neumark Y. & Adler B (2000) Birthweight recall by mothers of Israeli children. Public Health114, 161163.
  • Hindmarsh PC, Geary MPP, Rodeck CH et al. (2008) Factors Predicting Ante- and Postnatal Growth. Pediatric Research, 63, 9102.
  • Jaspers M, Meer GD, Verhulst FC et al. (2010) Limited validity of parental recall on pregnancy, birth, and early childhood at child age 10 years. Journal of Clinical Epidemiology, 63, 5191.
  • Li C, Wei J, Lu T et al. (2006) Mothers tended to overreport categorical infant birth weight of their children. Journal of Clinical Epidemiology, 59, 91325.
  • Mbuagbaw L & Gofin R (2010) Can recall of birth size be used as a measure of birthweight in Cameroon?Paediatric and Perinatal Epidemiology, 24, 3389.
  • McCormick MC & Brooks-gunn J (1999) Concurrent Child Health Status and Maternal Recall of Events in Infancy. Pediatrics, 1040, 681.
  • Mung’ala-Odera V & Newton CRJC (2001) Recall of Perinatal Events by Mothers Living in Rural Kenya. Epidemiology, 12, 6.
  • Ndibazza J, Muhangi L, Akishule D et al. (2010) Effects of deworming on maternal and perinatal outcomes in Entebbe, Uganda: a randomized controlled trial. Clinical Infectious Diseases, 50, 140.
  • Olson JE, Shu XO, Ross JA, Pendergrass T & Robison LL (1997) Medical Record Validation of Maternally Reported Birth Characteristics and Pregnancy-related Events : a Report from the Children ‘s Cancer Group. American Journal of Epidemiology, 145, 867.
  • O’Sullivan JJ, Pearce MS, Parker L & Infirmary V (2000) Parental recall of birth weight: how accurate is it?Journal of Abnormal and Social Psychology, 82, 2203.
  • Rice F, Lewis A, Harold G et al. (2007) Agreement between maternal report and antenatal records for a range of pre and peri-natal factors: the influence of maternal and child characteristics. Early Human Development, 83, 7504.
  • Tann CJ, Kizza M, Morison L et al. (2007) Use of antenatal services and delivery care in Entebbe, Uganda: a community survey. BMC Pregnancy and Childbirth, 7, 23.
  • Tate AR, Dezateux C, Cole TJ & Davidson L (2005) Factors affecting a mother’s recall of her baby’s birth weight. International Journal of Epidemiology, 34, 8695.
  • Van Gelder MMHJ & Roeleveld N (2011) Early Human Development Validation of maternal self-report in retrospective studies. Early Human Development87, 4344.
  • Walton KA, Murray LJ, Gallagher AM, Cran GW, Savage MJ & Boreham C (2000) Parental recall of birthweight: a good proxy for recorded birthweight?European Journal of Epidemiology, 16, 3796.