Introduction
What is new?
Key findings- •
For 94% of women, gestational age based on reported expected delivery date was within 1 week of hospital-based gestational age.
- •
88% of children who were born preterm according to hospital data were also correctly identified as such by maternal report.
- •
Our study showed higher percentage of agreement and narrower limits of agreement than previous studies.
- •
Agreement varied with various indicators of social position. Investigators need to take this into account when basing a study on maternal report of gestational age.
Preterm birth is one of the most important determinants of neonatal and infant mortality [1], [2] and is strongly associated with child health and development [3]. As a result, gestational age is often considered in epidemiological studies as a pregnancy outcome, a risk factor for outcomes occurring in childhood and adulthood, or as a confounder.
Many research studies, such as maternity surveys [4], [5] and follow-up studies of preterm children [6], [7], rely on retrospective maternal report of gestational age when hospital records with results from ultrasound scan are unavailable or are too expensive or time-consuming to access. Asking women about gestational age or expected date of delivery (EDD) is often the simplest and cheapest way of obtaining the information, and it is therefore relevant to know how reliable maternal report of gestational age is.
In studies of the accuracy of maternal recall of perinatal events, it has generally been found that recall depends on the type of event, time since the event, and on how the question is formulated [8]. Some studies have looked at maternal recall of gestational age among recall of other perinatal factors [9], [10], [11], [12], [13], [14], but few have focused on gestational age alone and attempted to describe which factors influence the accuracy of recall of gestational age [15], [16].
Several studies have compared maternal report of gestational age with gestational age in medical records. High intraclass correlations (ICC) have been reported ranging from 0.64 [12] to 0.90 [14] and high kappa coefficients ranging from 0.56 [15] to 0.90 [14], but exact agreement in weeks has often been found to be low. In a recent Danish study of 678 women, the ICC was high (0.76), but only 42% of participating women reported exact gestational age, although 94% reported gestational age within 2 weeks of that recorded on the birth registration [15]. In a study of 208 women in Taiwan, the correlation coefficient was high (0.93), but only approximately two-thirds of women correctly reported exact gestational age [11].
The present study uses data on 8,058 women from the Millennium Cohort Study (MCS) to measure agreement between maternal report of gestational age 9 months after delivery and routine hospital data to describe which perinatal and sociodemographic factors influenced agreement and to examine how the potentially observed disagreement influenced classification of preterm birth.