Menu
Menu
Menu
Menu
Background: Live attenuated vaccines have been observed to have particularly beneficial effects for child survival when given in the presence of maternally transferred immunity (priming). We aimed to test this finding and furthermore explore the role of paternal priming. Methods: In an exploratory, retrospective cohort study in 2017, parental Bacillus Calmette-Guérin (BCG) scars were assessed for infants from the Bandim Health Project (BHP) who had participated in a 2008–2013 trial of neonatal BCG vaccination. Parental scar effects on mortality were estimated from birth to 42 days, the age of the scheduled diphtheria-tetanus-pertussis (DTP) vaccination, in Cox proportional hazard models adjusted with Inverse Probability of Treatment Weighting. Findings: For 66% (510/772) of main trial infants that were still registered in the BHP area, at least one parent was located. BCG scar prevalence was 77% (353/461) among mothers and 63% (137/219) among fathers. In the first six weeks of life, maternal scars were associated with a mortality reduction of 60% (95%CI, 4% to 83%) and paternal scars with 49% (-68% to 84%). The maternal scar association was most beneficial among infants that had received BCG vaccination at birth (73% (-1% to 93%)). Although priming was less evident for paternal scars, having two parents with scars reduced mortality by 89% (13% to 99%) compared with either one or none of the parents having a scar. Interpretation: Parental BCG scars were associated with strongly increased early-life survival. These findings underline the importance of future studies into the subject of inherited non-specific immunity and parental priming. Funding: Danish National Research Foundation; European Research Council; Novo Nordisk Foundation; University of Southern Denmark.
The Bandim Health Project (BHP) maintains a Health and Demographic Surveillance System (HDSS) in six urban districts in Bissau, Guinea-Bissau. Births are registered by BHP staff at the country’s main hospital and during regular trimonthly home-visits. Routine vaccinations are provided at three HDSS health centres in accordance with the WHO-recommended immunization schedule. The original RCT design has been described in detail previously [2]. Briefly, the trial was designed to test the effects of early BCG vaccination (intervention group) versus delayed BCG vaccination (control group, standard practice) on neonatal mortality among LW infants, with infant mortality as a secondary outcome. Between 2008 and 2013, 4159 infants weighing <2500 g were recruited at four hospitals and the three HDSS health centres. Infants were randomised (1:1) to 0·05 mL intradermal BCG-Denmark vaccine (Statens Serum Institut) (intervention group, ‘BCG-at-birth’) or standard practice; that is, mothers were encouraged to have their infant BCG vaccinated at the local health center when the infant had gained weight (control group, ‘delayed BCG’). All infants received oral polio vaccine (OPV) at inclusion and we conducted follow-up visits at three days after enrolment and at 2, 6, and 12 months of age (Fig. 1). The family of infants who died were visited three months after the death by a specially trained field assistant to conduct a WHO/INDEPTH verbal autopsy [11]. Study design of the original trial and flowchart of study participants. In the original trial infants were randomized to BCG-at-birth or standard practice (‘delayed’ BCG). The BCG vaccination at 6 weeks in the control group is an indication from when most children in the control group would start receiving BCG, not the age to which the control group was randomized. Abbreviations: BCG, bacillus Calmette-Guérin; BHP, Bandim Health Project; HDSS, Health and Demographic Surveillance System. During the original trial, which included 4159 infants from all parts of greater Bissau, we had not been aware of the possible interaction between maternal and infant BCG, and no parental BCG information had been collected. For the present retrospective cohort study, all infants from the original trial that had lived within the HDSS area at inclusion (N = 1258) were sought for in the HDSS system in February 2017. Families of infants who had not moved according to the latest census information (N = 722) were visited, including parents of infants who had died during the original trial. We did not perform a sample size calculation as it was impossible to estimate the proportion of parents that could be reached and would be willing to participate. We conducted up to three home visits to find both parents. Parents who were present, received an explanation about the present study being an extension of the original BCG trial. After informed oral consent, BCG scars were assessed and measured with a ruler by one of two field assistants. Field assistants were unaware of the hypothesis regarding parental BCG scars. We have previously shown that a BCG scar is an indicator for a correctly applied BCG vaccine[10,12,13] and its use, instead of reported BCG vaccination, overcomes possible problems with recall bias. While the original trial reported effects of BCG on the all-cause neonatal mortality for historical reasons, our primary outcome was all-cause mortality for the period up to 42 days of age, before the infants would have reached the age where other routine vaccines are administered. We have done so in subsequent RCTs evaluating BCG versus no BCG ({"type":"clinical-trial","attrs":{"text":"NCT02504203","term_id":"NCT02504203"}}NCT02504203) and RCTs evaluating effects of different BCG strains ({"type":"clinical-trial","attrs":{"text":"NCT03400878","term_id":"NCT03400878"}}NCT03400878 and {"type":"clinical-trial","attrs":{"text":"NCT04383925","term_id":"NCT04383925"}}NCT04383925). The scheduled diphtheria-tetanus-pertussis (DTP) vaccination at six weeks of life has been associated with negative effects on overall female survival and might distort the effects of early BCG [14,15]. Furthermore, the majority of the infants in the delayed BCG group will not have received BCG by this age. Secondary outcome was all-cause mortality between 42 days and 1 year of age. Infant mortality rates for 0–42 days and 42 days-1 year were compared according to maternal and paternal scar status in Cox proportional hazard models, providing mortality rate ratios (MRRs). Two children (1 with both maternal and paternal scar, 1 only paternal scar) were enrolled in the trial later than 42 days after birth and were thus only included in the 42 day-1 year analysis. Time from randomization to all cause death was our main outcome and survival time was censored at migration out of HDSS area, or end of follow-up (42 days or 1 year). Clustering of twins was adjusted for by using robust standard errors. The proportional-hazards assumption was assessed graphically and tested using Schoenfeld residuals. In case there were no deaths in one of the groups, a univariate log-rank test was performed. Models were adjusted using stabilised Inversed Probability of Treatment Weighting (sIPTW). Probability of the infant's parent to have a BCG scar was predicted from a logistic regression including age, maternal schooling, electricity, and indoor toilet based on a directed acyclic graph (DAG) for this study (Supplementary Figure 1). Weights were stabilised using the marginal probability for exposure in the numerator. The assessment of correctly weighted samples is summarised in Supplementary Table 1 [16]. Effect modification by BCG randomization and sex was determined by including an interaction term to the analyses. Sensitivity analyses for outcome were performed using not 0–42day mortality, but neonatal mortality as the outcome, as well as censoring infants in the delayed BCG group at the day of BCG receipt. Sensitivity analyses for statistical adjustment included using different models for the prediction of the probability of exposure for sIPTW; 1) addition of place of inclusion and weight at inclusion, 2) addition of the subgroup variable (BCG randomization/sex), 3) prediction within the subgroups. The latter two analyses were to ensure sIPTW did not bias estimates due to possible different effects within subgroups, as detailed by Izem and colleagues [17]. Statistical analyses were performed using Stata 12 (Stata Corp, College Station, Texas). The original protocol was approved by the Guinean Ministry of Health's Research Coordination Committee and consultative approval was obtained from the Danish Central Ethical Committee and contained a statement on participants being approached for future studies. The parental scar follow-up amendment was approved by the same instances. Both the parental scar follow-up ({"type":"clinical-trial","attrs":{"text":"NCT03020147","term_id":"NCT03020147"}}NCT03020147) and the original study ({"type":"clinical-trial","attrs":{"text":"NCT00625482","term_id":"NCT00625482"}}NCT00625482) were registered at clinicaltrials.gov. The manuscript adheres to the STROBE guidelines. The funder of the study had no role in study design, data collection, data analysis, data interpretation, or writing of the report. All authors had full access to all the data in the study and had final responsibility for the decision to submit for publication.
