Lower accuracy of estimated delivery date using last menstrual period recall and related factors of adverse events during childbirth: A sub-analysis of a birth cohort study in Makassar, Indonesia

Study Design

This study is a sub-analysis of the Indonesian Birth Cohort Study, an ongoing, dynamic, longitudinal study in urban Makassar, Indonesia, initiated in 2022. A prospective cohort design was employed, including socio-economic, demographic, and pregnancy-related variables, to assess factors related to adverse delivery events and the accuracy of EDD estimation based on LMP recall, with a focus on adverse outcomes.

Samples/Participants

This sub-analysis used data from 171 pregnant women with singleton pregnancies and complete records of both LMP and birth outcomes. These participants were drawn from the larger parent study, which initially enrolled 205 pregnant women during their first and second trimesters. Eligibility criteria ensured that only those with accurate and comprehensive LMP data and birth records were included in this sub-study to allow for a focused evaluation of the association between LMP-based EDD and adverse outcomes during delivery.

Variable Definitions

Adverse events were defined as any unexpected negative occurrence during childbirth, including both preventable and unpreventable events (Skoogh et al., 2021). These events included factors related to access to health care, birth attendance, and the medical condition of the mother and newborn. Specific examples of adverse events were complications due to inadequate medical facilities or staff, prolonged labor, emergency cesarean section, postpartum hemorrhage, birth asphyxia, neonatal infection, breech presentation, seizures, premature rupture of membranes, cord entanglement, placenta previa, retained placenta, hypertension, and any other unexpected event requiring hospitalization. We also included prehospital delivery, a condition where the delivery occurs unexpectedly and in an unplanned setting, such as on the way to a health facility or a location without immediate access to medical care (Beaird et al., 2023). Participants were considered to have experienced an adverse event if they reported at least one during delivery.

Data Analysis

Data were analyzed using Stata version 14.0. Descriptive statistics, including frequencies and percentages, were used to summarize categorical variables. For continuous variables, means and standard deviations were reported to describe normally distributed data. Relationships were assessed using chi-square tests for bivariate analysis and Poisson regression for multivariate analysis, with statistical significance set at p <0.05 and <0.001 and 95% confidence intervals (CIs).

Ethical Considerations

Ethical approval for the parent cohort study and this sub-analysis was obtained from the Research Ethics Review Board of the Faculty of Public Health, Hasanuddin University (Ref. No. 10424/UN4.14.1/TP.01.02/2021). Written informed consent was secured from all participants prior to the study, ensuring that their rights and confidentiality were protected.

Participants Characteristics

A total of 171 pregnant women were included in this sub-analysis, and the descriptive characteristics of the participants are presented in Table 1.

The study population had a wide range of demographic and pregnancy-related characteristics. Most mothers (75.44%) were aged 20–35 at the time of delivery, 13.45% were under 19, and 11.11% were 36 or older. In terms of education, the majority of participants had completed tertiary education or senior high school (40.94%), while only 0.58% had no formal education. Employment status showed that most respondents (85.38%) were not employed.

More than half of the participants reported previous contraceptive use (50.88%), and most deliveries were at mother and child hospitals (69.01%). Spontaneous vaginal birth was the predominant mode of delivery (69.59%), and the infant sex favored males (57.89%). Adverse birth events were reported by most participants (66.67%), with preterm birth affecting almost half (48.54%) of the newborns. The previous birth interval was mostly short (≤18 months) at 49.71%. Interestingly, 50.88% of current pregnancies were unintended. Birth weight outcomes showed a majority of normal birth weight infants (85.96%) with lower rates of low birth weight (13.45%) and macrosomia (0.58%). Regarding reproductive history, a significant proportion of women were multiparous (69.01%) and multigravidas (71.93%).

Factors Related to Adverse Events During Childbirth

Various factors associated with adverse events during delivery are shown in Table 2. The categorized ADD-EDD difference showed a significant association (p <0.001), with the highest incidence of adverse events occurring when the difference was ≤-15 days (93.62%). Maternal age did not show a significant association (p = 0.563), although women over 36 years of age, considered at risk, had a higher incidence (73.68%). Education levels did not significantly correlate with adverse events (p = 0.830). Employment status also had no significant association (p = 0.723).

History of contraceptive use was not significantly associated with adverse events (p = 0.537), as 60.92% of women with a history of contraceptive use experienced adverse events. Place of birth also showed no significant association (p = 0.886), with the highest incidence occurring in general hospitals (71.43%). However, the mode of delivery was significantly associated with adverse events (p = 0.014), with 76.92% of cesarean births experiencing adverse events compared to 57.14% of normal births. Newborn sex had no significant effect on adverse events (p = 0.624), with little difference between male (61.62%) and female (62.58%) births. The interval between pregnancies did not significantly affect adverse events (p = 0.220), with a short interval (<18 months) showing an incidence rate of 68.24%.

Unintended pregnancy had no significant effect on the incidence of adverse events (p = 0.537), although the incidence rate of unwanted pregnancy was 65.48%. For birth weight, although not significantly associated (p = 0.122), there was a higher incidence of adverse events in low-birth-weight infants (78.26%). In terms of parity, there was no statistically significant difference, with slight variation between the primipara (62.26%) and multipara (63.56%) categories. Similarly, in the gravidity category, there was no significant relationship (p = 0.290), with differences between the primigravida (60.42%) and multigravida (64.23%) categories.

Table 3 shows the Poisson regression analysis of factors influencing adverse events during delivery. Two primary variables were analyzed: mode of delivery and ADD-EDD difference. For the mode of delivery, spontaneous vaginal birth was used as the reference, while cesarean section showed an estimate of 2.03 ± 1.14 with a relative risk (RR) of 2.51 (95% CI: 1.03-6.15, p = 0.043) and ARR of 1.11 (1.02-1.22, p <0.001), indicating a statistically significant increased risk of adverse events compared with spontaneous vaginal birth.

The ADD-EDD difference was significantly associated with adverse events in several time intervals. A difference of ≤-15 days (RR: 32.35, 95% CI: 8.59-121.75, p <0.001) and a difference of -14 to -8 days (RR: 6.23, 95% CI: 2.38-16.29, p <0.001) both showed a substantial increase in the risk of adverse events, and these associations remained significant after adjusting for covariates, with ARR of 1.49 (95% CI: 1.35-1.65, p <0.001) and 1.33 (95% CI: 1.19-1.49, p <0.001), respectively. While a difference of 8 to 14 days was not statistically significant in the unadjusted model (RR: 2.35, 95% CI: 0.75-7.32, p = 0.139), it became marginally significant after adjustment (ARR: 1.20, 95% CI: 1.00-1.44, p = 0.046).

Additionally, a difference of ≥15 days (RR: 11.60, 95% CI: 2.41-55.68, p = 0.002) was associated with a significantly increased risk of adverse events, which remained significant even after adjustment (ARR: 1.37, 95% CI: 1.20-1.58, p <0.001).

Figure 1 shows the kernel density of the difference between ADD and EDD. The mean difference was -7.15 days, indicating that, on average, deliveries occurred 7.15 days earlier than expected for the study population. While the mean difference is considered a normal range, a significant gap was observed in the distribution of the ADD-EDD difference, with deviations of the ADD occurring 109 days before and 45 days after the EDD.

Summary of the Findings

Our analysis centered on evaluating the accuracy of self-reported LMP in predicting EDD and identifying sociodemographic and pregnancy-related characteristics that might serve as risk factors for adverse events during delivery. The absence of significant associations with several sociodemographic and reproductive factors in this study suggests that these variables may not be primary determinants of such outcomes in this population. However, the mode of delivery and categorized ADD-EDD differences were identified as significant predictors of adverse events. Our results indicate a statistically significant increased risk of adverse events associated with cesarean sections compared to spontaneous vaginal births, aligning with previous research highlighting the potential complications of cesarean deliveries (Sandall et al., 2018). Furthermore, the analysis of the ADD-EDD difference revealed a complex relationship between LMP-based EDD accuracy and adverse events.

Women with a substantial ADD-EDD discrepancy, particularly those with an earlier ADD (differences ≤-15 and -14 to -8 days), were at significantly higher risk of adverse events. While a difference of 8 to 14 days did not reach statistical significance, a difference of ≥15 days was associated with increased risk. This suggests a potential threshold effect, where wider ADD-EDD discrepancies are more strongly linked to adverse events. We assume that the impact of misestimated EDD is related to the woman’s birth preparedness. Previous studies have shown that ill-formed travel plans for health facilities (Devkota et al., 2020; Exavery et al., 2014; Iftikhar ul Husnain et al., 2018; Pfeiffer & Mwaipopo, 2013), inadequate financial savings for delivery (Devkota et al., 2020; Iftikhar ul Husnain et al., 2018), and unexpected delivery (Exavery et al., 2014) were significantly associated with a reduced likelihood of health facility delivery. This finding underscores the importance of accurate EDD estimation in identifying pregnancies at increased risk and implementing appropriate preventive measures.

Furthermore, our findings indicate a systematic failure to accurately estimate EDDs in our study population, with deliveries occurring, on average, almost a week earlier than predicted based on LMP. Although the mean difference is considered a normal range, the wide range of EDD accuracy, with ADDs occurring up to 109 days prior to and 45 days after the EDD, indicates that relying on LMP alone for EDD estimation is unreliable. This discrepancy is particularly critical as both preterm and post-term births are associated with higher risks of neonatal complications and mortality. This result is aligned with existing studies that consistently reported that LMP dating can result in substantial misclassification of preterm and post-term deliveries if compared to advanced methods. For instance, a study conducted in South Africa reported that LMP-based dating resulted in a significantly greater number of misclassified preterm and post-term deliveries compared to early ultrasound and a smartphone app. The preterm birth rate was 11.4% by LMP, 1.9% by EUS, and 3.4% by the smartphone app (Majola et al., 2021). Similarly, research conducted in Tanzania indicated that LMP-based EDD resulted in preterm and post-term birth rates that were two and five times higher, respectively, compared to those determined by ultrasound, with 17.0% preterm and 17.6% post-term by LMP compared to 7.7% preterm and 3.4% post-term by ultrasound (Nielsen et al., 2021). Therefore, infants classified as preterm based on ultrasound but as a term based on LMP are at a higher risk of adverse outcomes, including increased infant mortality and NICU (Neonatal Intensive Care Unit) admissions, compared to those classified as preterm based on LMP but as a term based on ultrasound (Morken et al., 2016).

The inaccuracy of LMP-based EDD can also lead to mistimed antenatal care and obstetric interventions, which are crucial for managing high-risk pregnancies and ensuring timely delivery in health facilities. For example, a study conducted in Zanzibar revealed that a significant overestimation of EDD based on the LMP was associated with a reduced likelihood of delivering at a health facility, which in turn increased the risk of adverse maternal and neonatal outcomes (Fulcher et al., 2020). Furthermore, applying LMP without considering menstrual irregularities can further exacerbate these inaccuracies. A study underscored that medical professionals frequently neglect to consider the necessity for cycle length correction when employing Naegele’s rule, resulting in significant discrepancies in EDD calculation (Parikh & Pandia, 2011). The inaccuracy of LMP is further evidenced by difficulties in recall, which are particularly common among women with irregular menstrual cycles or lower educational levels. It has been demonstrated that a significant proportion of women experience difficulty in accurately recalling their LMP (Kullinger et al., 2018). This can result in errors in gestational age estimation due to assumptions about cycle regularity, ovulation timing, and other factors (Kullinger et al., 2018). For instance, in Bangladesh, only 53% of women were able to recall their LMP, and this recall was significantly associated with higher educational levels and calendar literacy (Sarker et al., 2020). In Brazil, the sensitivity of LMP in estimating preterm birth rates was 65.6% in São Luís and 78.7% in Ribeirao Preto, with positive predictive values of 57.3% and 73.3%, respectively, indicating substantial errors in LMP-based dating (Medeiros et al., 2015).

Implications of the Study

Overall, this study provides strong evidence that inaccuracies in LMP-based EDD estimation are associated with a higher risk of adverse events during delivery, underscoring the urgent need for more reliable EDD methods to improve maternal and neonatal health. While the results show significant associations, further research is needed to clarify the causal mechanisms linking mode of delivery, EDD-ADD discrepancies, and adverse events.

Limitations

We recognize that the broad definition of adverse events may lead to misinterpretation of the results. Additionally, the study was conducted in Urban Makassar, which may limit the generalizability of the findings to rural populations or different regions, where healthcare access and demographic factors may differ. Future studies should consider potential confounding factors and conduct a more expansive sample size and settings to strengthen our understanding of these complex relationships.