Limited validity of parental recall on pregnancy, birth, and early childhood at child age 10 years

Abstract

Objective

Evidence on the validity of parental recall of early childhood behavior is lacking. Our aim was to examine the validity of parental recall at child age 10–12 years for maternal lifestyle during pregnancy, the birth characteristics, and early childhood behavior.

Study Design and Setting

The study population comprised 2,230 children and their parents. Children aged 10–12 years were recruited from elementary schools (response: 76.0%). Parents were asked to recall lifestyle during pregnancy, birth characteristics, and childhood behavior at age 4–6 years. Recalled data were compared with information registered by Preventive Child Healthcare (PCH) from birth onwards.

Results

For birth weight and gestational age, we found no systematic difference between recalled and PCH-registered data; 95% limits of agreement were ±1.2 pounds (600 g) and ±2.4 weeks, respectively. For maternal alcohol use during pregnancy and early childhood behavior problems, Cohen's kappas were low (0.03–0.11). Compared with PCH registration, parents tended to overreport at age 10–12 years. In contrast, kappa was high for maternal smoking during pregnancy (0.77).

Conclusion

Retrospectively collected information on lifestyle during pregnancy, birth, and early childhood behavior is sometimes biased, which limits its value in estimating the contribution of early-life adversity to health in later life.

Introduction

The period from conception until school age is considered to be extremely important for children's socioemotional development [1], [2], [3]. It can be a “window of opportunity” [4] and a “window of vulnerability,” depending on the outcome, and it can offer great opportunities to improve further life by reducing adverse environmental factors and/or empowering positive factors.

Early-life factors that are associated with future mental and physical health of the child concern either pregnancy, birth, or early childhood [5], [6], [7], [8], [9], [10], [11], [12], [13], [14]. Pregnancy factors are, for example, maternal smoking, alcohol use, drug abuse, medication, and disease during pregnancy. These have been shown to be risk factors for the development of psychopathology [5], [7], [8], [9], [11], growth retardation, respiratory problems, cardiovascular disease, and other health problems [6], [13]. Similarly, birth factors, such as low birth weight and premature birth, can lead to a number of health problems [8], [13]. Early childhood factors are, for example, toddler sleep problems [10], [14], early eating problems [15], [16], and preschool problems with social behavior [2], which are predictors for later mental health problems.

Ideally, the effects of early-life factors should be considered in follow-up studies in which information on these factors is collected from various sources (e.g., parents and professional assessment) in the period from conception to school age; follow-up extends to adolescent or even adult age, and attrition and loss-to-follow-up rates are low. However, most studies rely on retrospective data collection, which may introduce information bias. This may concern recall bias because of differences in validity of subject recall, marked by an over- or underreport of information. Or it may, for instance, be rumination bias, which occurs when people with a disease tend to think harder about their prior exposures than healthy people, causing them to systematically remember exposure differently [17]. This distorts the measurement of the association between exposure and disease. The magnitude and direction of this distortion are difficult to predict.

Many previous studies on early risk factors for adverse future mental health relied on retrospectively collected data and included a limited number of risk factors. The validity of data collected retrospectively needs to be examined before using them to estimate relationships with health or developmental outcomes. Regarding early-life factors, previous studies found good recall for maternal smoking during pregnancy [18], [19], [20], for gestational age [20], [21], [22], and for birth weight [19], [20], [23], [24], [25], [26], but not for alcohol use during pregnancy [18], [19]. However, most of these studies covered a limited period of time and did not consider a combination of variables. So far, no studies are available on the validity of parental recall of early childhood behavior and neither of pre- and perinatal factors.

The aim of the current study is to examine the validity and precision of recall of maternal lifestyle during pregnancy, birth characteristics, and early childhood behavior in a community-based sample. For this purpose, we compared data on prenatal and early-life characteristics collected at child age 10–12 years with data registered by Preventive Child Healthcare (PCH) from birth onwards.