Jacob B. Hjelmborg

Objective Child mortality and stillbirth rates (SBR) remain high in low‐income countries but may be underestimated due to incomplete reporting of child deaths in retrospective pregnancy/birth histories. The aim of this study is to compare stillbirth and mortality estimates derived using two different methods: the method assuming full information and the prospective method. Methods Bandim Health Project's Health and Demographic Surveillance Systems (HDSS) follows women of reproductive age and children under five through routine home visits every 1, 2 or 6 months. Between 2012 and 2020, we estimated and compared early neonatal (ENMR, <7 days), neonatal (NMR, <28 days), and infant mortality (IMR, <1 year) per 1000 live births and SBR per 1000 births. Risk time for children born to registered women was calculated from birth (the method assuming full information) versus date of first observation in the HDSS (the prospective method), either at birth (for pregnancy registration) or registration. Rates were calculated using the Kaplan–Meier estimator and compared in generalised linear models allowing for within‐child correlation obtaining relative risks (RR). Results We registered and followed 29,413 infants (1380 deaths; 1459 stillbirths) prospectively. An additional 164 infant deaths and 129 stillbirths were registered retrospectively and included in the method assuming full information. The ENMR was 24.5 (95%CI: 22.6–26.4) for the method assuming full information and 25.8 (23.7–27.8) for the prospective method, RR = 0.96 (0.93–0.99). Differences were smaller for the NMRs and IMRs. For SBRs, the estimates were 53.5 (50.9–56.0) and 58.6 (55.7–61.5); RR = 0.91 (0.90–0.93). The difference between methods became more pronounced when the analysis was limited to areas visited every 6 months: RR for ENMR: 0.91 (0.86–0.96) and RR for SBR: 0.85 (0.83–0.87). Conclusions Assuming full information underestimates SBR and ENMR. Accounting for omissions of stillbirths and early neonatal deaths may lead to more accurate estimates and improved ability to monitor mortality.

Background: The time during which there is an increased risk of death for cancer survivors was evaluated in a large twin study, which allows for matching on shared components such as age, genes, and socioeconomic factors in childhood. Methods: By use of data from Danish registers, time to death from initial cancer was studied prospectively in twins in two different settings. The twins were diagnosed with at least one cancer in the period 1943 to 2011. Setting I included 5,680 same-sex twin pairs aged 6 and over, while Setting II included 3,218 twin individuals from age 70 and over. The study provides comparisons within twin pairs and across birth cohorts, age at diagnoses, and time at diagnosis. Results: In 2001 to 2011, the 5-year mortality risk for a twin surviving cancer after the age of 70 was twofold that of the co-twin, regardless of sex and zygosity, and it was 1.5-fold if the twin survived the initial 9 months. After 5 to 6 years, the mortality risk corresponded to that of the co-twin. In previous decades, the excess hazard risk was considerably higher for both older and younger cohorts. There were no indications of change in relative survival across old birth cohorts. Conclusions: This large twin study suggested that for a cancer-treatment survivor diagnosed at age 70 or later, the additional mortality risk was largely absent 5 years later, by which time the survival relative to the co-twin was 60%. Impact: Elevated mortality risk after cancer is offset after 5 to 6 years.