Mortality Risk Among Frail Neonates and Maternal BCG Vaccine Scar Status: Observational Study From Guinea-Bissau

Abstract

Background

Maternal priming with the Bacille Calmette-Guérin (BCG) vaccine has been associated with reduced offspring mortality rates. We investigated this association in a cohort of frail neonates.

Methods

We performed an observational study within a randomized BCG trial conducted at the neonatal intensive care unit (NICU) in Guinea-Bissau from 2015 to 2017. At NICU admission and after informed consent, the maternal scar status was evaluated by visual inspection before neonates were randomized 1:1 to receive BCG + oral polio vaccine immediately or at hospital discharge. Stratified by maternal scar status, we assessed overall in-hospital and postdischarge mortality rates through 42 days of age in Cox proportional hazards models providing adjusted mortality rate ratios (aMRRs).

Results

Overall, 62% of mothers (903 of 1451) had a BCG vaccine scar. During NICU admission, the mortality risk was 1.7% (15 of 903) for neonates born to mothers with a scar versus 3.3% (18 of 548) for those born to mothers with no scar; the aMRR for maternal scar versus no scar was 0.53 (95% CI, .26–1.05), 0.39 (95% CI, .13–1.05) for unvaccinated and 0.70 (95% CI, .26–1.87) for vaccinated neonates.

Conclusions

This small study indicates that maternal BCG vaccine might be associated with reduced all-cause NICU mortality rate. If confirmed elsewhere, this finding would have substantial ramifications for global health.

Bacille Calmette-Guérin (BCG) vaccine, maternal BCG scar, NICU mortality, nonspecific effects of vaccines, vertical priming, maternal BCG priming, perinatal mortality

A series of observational studies and randomized controlled trials (RCTs) have associated BCG vaccination with beneficial nonspecific effects, substantially reducing all-cause mortality rates in low-income countries with high infectious disease prevalence [1–8]. In 2014, a World Health Organization–commissioned review of nonspecific effects concluded that BCG vaccine halves child mortality rates and encouraged further research [9]. Even in a high-income country with lower prevalence of infectious diseases, contrasting BCG versus no BCG at birth in a large-scale RCT revealed that if the mother had also been BCG vaccinated, the benefit of being randomized to neonatal BCG was substantial [10].

To further explore the novel finding of maternal—or “vertical”—BCG priming effects, the Bandim Health Project (BHP; www.bandim.org) initiated a series of maternal scar studies in the low-income and high-infection prevalent country Guinea-Bissau [11–14]. We found that for children born to mothers with a scar, childhood BCG scars were associated with a 66% (33%–83%) lower mortality risk when children with a scar were compared with those with no scar; there was no effect of child BCG vaccine scar if the mother had no scar [11]. Similarly, presence of a maternal scar was associated with a 60% (4%–83%) reduction in all-cause mortality rate by 6 weeks of age in a retrospective analysis [12]. Finally, in a large cohort of healthy BCG-vaccinated newborns, presence of a maternal scar was associated with fewer fatal infections, such as sepsis, by 6 weeks of age [13]. Most of the available data sets to examine the effect of maternal BCG vaccine involve BCG-vaccinated newborns, since the policy is to provide BCG at birth. In a 2021 analysis from the rural areas of Guinea-Bissau, however, the maternal scar status was evaluated for pregnant women, and having a maternal scar was associated with a trend toward fewer adverse pregnancy outcomes, the maternal scar versus no scar hazard ratio being 0.78 (95% CI, .59–1.01) [14].

With the a priori hypothesis that maternal BCG vaccination would reduce the overall mortality risk, especially for BCG-vaccinated neonates, we here report the first prospective study evaluating all-cause mortality effects of maternal priming with BCG. Our analysis was conducted within a unique cohort of frail neonates admitted to the neonatal intensive care unit (NICU), who were randomly allocated to receive BCG at NICU admission or at discharge (the usual practice) [15].

METHODS

Setting

The main trial was conducted in the NICU of Hospital Nacional Simão Mendes (HNSM), located in the capital, Bissau. The hospital’s maternity ward, Guinea-Bissau’s principal birthplace, has approximately 7000 deliveries per year. BHP maintains a routine data collection system to register births, vaccinations, admissions, and admission outcomes at the HNSM maternity and pediatric wards [4, 15–19].

Study Design

This prospective study was based on data collected from 2015 to 2017 within an RCT initiated in 2013 that compared the overall health effects of providing neonatal vaccines immediately at NICU admission versus at discharge, described in detail elsewhere [15]. Briefly, the main outcome was all-cause mortality rate during NICU admission. Known risk factors for NICU admission are birth weight <1500 g, Apgar score ≤3, single motherhood, and cesarean delivery [17]. Exclusion criteria were weight on admission <1250 g and Apgar score <2 at birth [15].

The recommended vaccination schedule at birth in Guinea-Bissau is BCG and oral polio vaccine (OPV), and the usual practice is coadministration at discharge from HNSM. Thus, NICU-admitted neonates were randomized 1:1 to BCG + OPV immediately versus BCG + OPV at hospital discharge, per the standard of care. The BCG strains used in the trial were BCG-Denmark in 2013–2016 and BCG-Japan in 2016–2017, owing to a BCG-Denmark production halt [15].

Ethical Approval

The study protocol was approved by the Guinea-Bissau Health Ministry’s Research Coordination Committee (reference no. CNES-2013-0054) and given consultative approval by the Central Danish Ethical Committee (case no. 1303771-1). A subsequent protocol revision was equally approved by both ethical committees (nos. CNES-2014-001 and 1303771-2). The trial was conducted in accordance with the Helsinki Declaration ethical standards, and a data and safety monitoring board oversaw the trial. Free healthcare consultations and essential drugs were provided to all infants invited to participate in the study, which was registered at ClinicalTrials.gov (NCT01989026) on 20 November 2013.

Enrollment

Among all neonates admitted to the NICU during the trial, 86% were eligible for participation [15]. Enrollment procedures occurred the morning after admission to the NICU, where mothers or guardians were provided written study information in Portuguese and a verbal explanation of the study in the local language Creole and invited to ask questions. If the mother or guardian of the newborn gave written informed consent to participate, the newborn was enrolled in the main trial. At enrollment, we collected socioeconomic data and recorded the maternal mid–upper-arm circumference along with newborn weight and twinning status.

The study group became aware of the possible importance of maternal BCG priming for offspring outcomes in July 2015 and conducted a training course for the study assistants. As part of inclusion procedures, we initiated assessments of the maternal BCG vaccine scar status by visual inspection of both upper arms on 10 July 2015, when 1750 neonates had been enrolled in the trial.

Follow-up

The weight and vital status for enrolled neonates was monitored daily, and follow-up continued at the adjacent pediatric ward for neonates transferred there. At hospital discharge, we ensured that control neonates received the recommended BCG + OPV. After discharge, the neonate and family were provided home transport by BHP, and a map was drawn to facilitate follow-up home-visits, which were conducted at 3 days after discharge and at 2, 6, and 12 months of age, applying the same sequence of visits as in previous trials [1–3].

Statistical Analyses

Mortality rate ratios (MRRs) by maternal BCG vaccine scar status (scar vs no scar) were estimated in Cox proportional hazards models, with age as the underlying time variable. Age was thus inherently controlled for. Tests of proportionality of hazards were computed using Schoenfeld residuals.

We tested associations between baseline inclusion characteristics and maternal BCG scar status, using Kruskal-Wallis and Pearson χ² tests. The maternal level of schooling (no education, primary school, secondary school, high school, or university) differed significantly between the maternal scar and no maternal scar cohorts and was therefore included in a multivariate Cox model containing maternal BCG scar, sex, and the outcome variable and providing adjusted MRRs (aMRRs). For 9 mothers (0.6% of the cohort) with missing information, the level of schooling was imputed as no education.

We present mortality risk estimates by maternal scar status during NICU admission, overall (main outcome) and by offspring BCG randomization allocation (secondary outcome). A senior pediatrician blinded to both randomization allocation and maternal scar status evaluated NICU deaths and assigned a probable cause of death, which was used to differentiate between causes of death in a supplementary analysis. Furthermore, estimates including follow-up time after NICU discharge (when all neonates had been BCG vaccinated) and through 42 days of age are presented (secondary outcome). This approach was chosen to evaluate maternal BCG effects in both BCG-vaccinated, unvaccinated, and combined cohorts and further to eliminate interference from other childhood vaccines administered from 42 days of age, in accordance with previous studies [7, 12, 16, 20].

For the NICU analysis, neonates contributed risk time until they died, were lost to follow-up, or were discharged, whichever came first. In the analysis including postdischarge follow-up time, neonates contributed risk time until they died, migrated, or reached 42 days of age, whichever came first. Because neonatal BCG has previously been shown to be beneficial in low-birth-weight (LBW; birth weight <2500 g) neonates [3, 7] and because there might be differences between BCG strains [21], we conducted supplementary analyses assessing effects of maternal BCG scars by LBW status (yes/no) and by BCG strain. All analyses were intention-to-treat analysis, performed overall and by sex using StataIC 16 software (StataCorp), and all estimates are reported with 95% confidence intervals (CIs). The sample size for the present study was pragmatic, based on the number of neonates enrolled in the main trial from when the potential importance of maternal BCG priming was recognized and the collection of maternal scar data initiated, and until the main trial was halted [15].

RESULTS

Between October 2013 and August 2017, 3353 neonates were enrolled in the main trial; 1750 of these were enrolled before we initiated maternal scar assessments. For the remaining 1603 neonates, the maternal scar status was captured for 90% of the mothers (1451 of 1603); 62% (903 of 1451) had a BCG vaccine scar. Mothers with a BCG scar had received more schooling, a higher proportion was literate (P < .001 for both comparisons), and the proportion of male neonates recruited in the study was 50% (452 of 903) for mothers with a BCG scar versus 57% (311 o f548) for mothers with no scar (P = .01) (Table 1). Six infants (4 with and 2 with no maternal scar) corresponding to 0.4% were lost to follow-up during the NICU admission.

Table 1.

Baseline Inclusion Characteristics at Admission to Neonatal Intensive Care Unit by Maternal BCG Scar Status

Characteristic	Neonates Included, % (No. Included/Total No.)a		P Valueb
	Maternal BCG Scar	No Maternal BCG Scar
Total	62 (903/1451)	38 (548/1451)	…
Maternal characteristics
ȃMaternal age, median (IQR), y	27 (23–30)	27 (20–33)	.45c
ȃMUAC, median (IQR), mm	274 (252–298)	274 (250–298)	.67c
ȃResidence in BHP HDSS	13 (115/903)	14 (77/548)	.47
ȃResidence in rural area	29 (261/903)	31 (168/548)	.48
ȃNo maternal schooling	24 (214/897)	36 (198/545)	<.001
ȃMother is literate	75 (676/896)	63 (342/542)	<.001
Infant characteristics
ȃCesarean delivery	72 (652/901)	69 (374/545)	.13
ȃApgar score (1 min after birth), mean (SD)	6.7 (2.1)	6.5 (2.0)	.12c
ȃLBW (<2500 g)	31 (279/903)	31 (172/548)	.85
ȃBirth weight, median (IQR), g	2854 (2330–3360)	2836 (2250–3365)	.94c
ȃInclusion weight, median (IQR), g	2753 (2240–3250)	2732 (2150–3235)	.82c
ȃRecruited on day of birth	27 (247/903)	24 (131/548)	.15
ȃMale sex	50 (452/903)	57 (311/548)	.01
ȃTwinning	21 (185/901)	17 (90/545)	.10

Abbreviations: BHP, Bandim Health Project; HDSS, Health and Demographic Surveillance System; IQR, interquartile range (25% to 75% quartile); LBW, low birth weight; MUAC, mid–upper arm circumference; NICU, neonatal intensive care unit; SD, standard deviation.

Data represent % Included (No. Included/Total No.), unless otherwise specified. The same inclusion supervisor was responsible for all inclusions, including the assessment of maternal BCG scar status.

P value based on Pearson χ² test unless otherwise specified.

P value based on Kruskal-Wallis test.

All-Cause Mortality Risk During NICU Admission by Maternal BCG Scar Status

For the entire cohort of both vaccinated and unvaccinated neonates, the overall mortality risk during NICU admission was 1.7% (15 of 903) if the mother had a scar and 3.3% (18 of 548) if not, the aMRR for the maternal scar/no scar comparison being 0.53 (95% CI, .26–1.05) (Table 2). By cause of death, the aMRR was 0.92 (95% CI, .08–10.8) for birth complications, 0.80 (95% CI, .21–3.02) for infectious diseases, 0.38 (95% CI, .12–1.13) for prematurity or respiratory insufficiency, and 0.39 (95% CI, .09–1.80) for unknown causes (Table 2).

Table 2.

Mortality Risk During Neonatal Intensive Care Unit Admission by Maternal BCG Scar Status, Overall and by Randomization Allocation, Stratified by Sex

Neonate Group	Mortality Rate, % (Deaths, No./No. Enrolled) [Mortality Rate per PY; PY]		MRR (95% CI)a	aMRR (95% CI)b
	Maternal BCG Scar	No Maternal BCG Scar
All neonates admitted to NICU
ȃOverall	1.7 (15/903) [1.1; 13.6]	3.3 (18/548) [2.2; 8.4]	0.50 (.25–.99)	0.53 (.26–1.05)
ȃMale	2.0 (9/452) [1.2; 7.2]	3.9 (12/311) [2.6; 4.6]	0.49 (.20–1.16)	0.51 (.21–1.21)
ȃFemale	1.3 (6/451) [0.9; 6.4]	2.5 (6/237) [1.6; 3.7]	0.55 (.18–1.71)	0.56 (.18–1.77)
ȃBy cause of death
ȃȃBirth complications	0.2 (2/903)	0.2 (1/548)	1.10 (.10–12.2)	0.92 (.08–10.8)
ȃȃȃMale	0.2 (1/452)	0.3 (1/311)	0.58 (.04–9.28)	0.54 (.03–8.88)
ȃȃȃFemale	0.2 (1/451)	0.0 (0/237)	NA	NA
ȃȃPrematurity or respiratory insufficiency	0.6 (5/903)	1.6 (9/548)	0.34 (.11–1.02)	0.38 (.12–1.13)
ȃȃȃMale	0.9 (4/452)	1.6 (5/311)	0.55 (.15–2.05)	0.57 (.15–2.14)
ȃȃȃFemale	0.2 (1/451)	1.7 (4/237)	0.13 (.01–1.20)	0.15 (.02–1.39)
ȃȃInfectious diseases	0.6 (5/903)	0.7 (4/548)	0.75 (.20–2.81)	0.80 (.21–3.02)
ȃȃȃMale	0.4 (2/452)	1.0 (3/311)	0.38 (.06–2.41)	0.39 (.06–2.50)
ȃȃȃFemale	0.7 (3/451)	0.4 (1/237)	1.89 (.19–18.7)	2.12 (.21–21.4)
ȃȃUnknown cause	0.3 (3/903)	0.7 (4/548)	0.45 (.10–2.00)	0.39 (.09–1.80)
ȃȃȃMale	0.4 (2/452)	1.0 (3/311)	0.45 (.08–2.70)	0.39 (.06–2.35)
ȃȃȃFemale	0.2 (1/451)	0.4 (1/237)	0.52 (.03–8.26)	0.41 (.02–6.64)
Neonates admitted to NICU and allocated to immediate BCG + OPV
ȃOverall	1.8 (8/441) [1.2; 6.7]	2.8 (8/287) [1.7; 4.7]	0.69 (.26–1.83)c	0.70 (.26–1.87)
ȃMale	2.2 (5/231) [1.3; 3.7]	3.8 (6/157) [2.5; 2.4]	0.60 (.18–1.96)	0.62 (.19–2.04)
ȃFemale	1.4 (3/210) [1.0; 3.0]	1.5 (2/130) [0.9; 2.3]	1.02 (.17–6.13)	1.19 (.20–7.23)
Neonates admitted to NICU and allocated to control group
ȃOverall	1.5 (7/462) [1.0; 6.9]	3.8 (10/261) [2.8; 3.6]	0.37 (.14–.96)	0.39 (.15–1.05)
ȃMale	1.8 (4/221) [1.1; 3.5]	3.9 (6/154) [2.7; 2.2]	0.41 (.11–1.45)	0.41 (.11–1.48)
ȃFemale	1.2 (3/241) [0.9; 3.4]	3.7 (4/107) [2.8; 1.4]	0.32 (.07–1.43)	0.31 (.07–1.39)

Abbreviations: aMRR, adjusted mortality rate ratio; CI, confidence interval; MRR, mortality rate ratio; NA, not applicable; NICU, neonatal intensive care unit; OPV, oral polio vaccine; PY, person-years.

Cox proportional hazards model (crude).

Cox proportional hazards model adjusted for level of maternal schooling and sex (overall estimate).

Test of the proportional hazards assumption using Schoenfeld residuals: P = .02.

All-Cause Mortality Risk During NICU Admission by Maternal BCG Scar Status, Stratified by Offspring Vaccination Status

Among neonates randomized to vaccination with BCG + OPV at NICU admission, the overall mortality risk during NICU admission was 1.8% (8 of 441) if the mother had a scar and 2.8% (8 of 287) if not, the aMRR for the scar/no scar comparison being 0.70 (95% CI, .26–1.87) (Table 2). For the neonates randomized to control (vaccination at NICU discharge), the overall mortality risk during NICU admission was 1.5% (7 of 462) if the mother had a scar and 3.8% (10 of 261) if not, the aMRR for the scar/no scar comparison being 0.39 (95% CI, .15–1.05) (Table 2). There was no significant difference in the effect of maternal BCG vaccination within randomization strata (P-value for same effect = .42).

All-Cause Mortality Risk During NICU Admission and Post-NICU Discharge and Through 6 Weeks of Age (Includes Only Vaccinated Follow-up Time)

The unvaccinated control group of the main trial was vaccinated at discharge from the NICU, and for the combined vaccinated cohort, the mortality risk through 6 weeks of age was 1.9% (17 of 896) for neonates born to mothers with a BCG vaccine scar and 3.0% (16 of 538) for those born to mothers with no scar, the corresponding overall aMRR being 0.74 (95% CI, .37–1.48) (Table 3).

Table 3.

Mortality Risk Through 42 Days of Age by Maternal BCG Scar Status and Offspring BCG Vaccination Status, Overall and by Sex

Neonate Groupa	Mortality Rate, % (Deaths, No./No. Enrolled) [Mortality Rate per PY; PY]		MRR (95% CI)b	aMRR (95% CI)c
	Maternal BCG Scar	No Maternal BCG Scar
BCG-vaccinated neonatesd
ȃOverall	1.9 (17/896) [0.2; 80]	3.0 (16/538) [0.4; 46]	0.63 (.32–1.25)	0.74 (.37–1.48)
ȃMale	2.5 (11/448) [0.3; 40]	4.3 (13/305) [0.5; 26]	0.56 (.25–1.24)	0.61 (.27–1.38)
ȃFemale	1.3 (6/448) [0.2; 40]	1.3 (3/233) [0.2; 20]	1.06 (.27–4.45)	1.28 (.32–5.15)
Both BCG-vaccinated and unvaccinated neonates
ȃOverall	2.7 (24/903) [0.3; 87]	4.7 (26/548) [0.5; 49]	0.53 (.31–.93)	0.59 (.34–1.04)
ȃMale	3.3 (15/452) [0.3; 43]	6.1 (19/311) [0.7; 28]	0.52 (.26–1.02)	0.55 (.28–1.09)
ȃFemale	2.0 (9/451) [0.2; 44]	3.0 (7/237) [0.3; 21]	0.64 (.24–1.73)	0.70 (.26–1.88)

Abbreviations: aMRR, adjusted mortality rate ratio; CI, confidence interval; MRR, mortality rate ratio; PY, person-years.

Both groups include neonates during neonatal intensive care unit (NICU) admission and after NICU discharge, through 42 days of age.

Cox proportional hazards model (crude).

Cox proportional hazards model adjusted for level of maternal schooling and neonate sex.

Deaths among neonates allocated to BCG during NICU admission and after discharge as well as control neonates who were BCG vaccinated at discharge from the NICU.

Overall Mortality Risk From Enrollment to 42 Days of Age (Includes Both Vaccinated and Unvaccinated Follow-up Time)

The overall mortality risk during NICU admission and after discharge through 42 days of age was 2.7% (24 of 903) if the mother had a scar and 4.7% (26 of 548) if the mother had no scar, the aMRR for the scar/no scar comparison being 0.59 (95% CI, .34–1.04) (Table 3). By birth weight, among 451 LBW neonates, the aMRR was 0.60 (95% CI, .30–1.18). For 1000 neonates with normal birth weight, the aMRR was 0.48 (95% CI, .18–1.39). By strain of BCG administered, there were 834 neonates that received BCG-Denmark; 60% (498 of 834) were born to mothers with a BCG vaccine scar, and the overall aMRR was 0.57 (95% CI, .22–1.51). Among 600 neonates that received BCG-Japan, 66% (398 of 600) were born to mothers with a BCG scar, and the overall aMRR was 0.94 (95% CI, .34–2.59) (see Supplementary Table 1).

All-Cause NICU Mortality Risk by Main Trial Randomization Allocation (BCG vs Control)

The overall MRR for the BCG vs. control comparison in the subgroup with maternal scar assessments was 0.90 (95% CI, .45–1.78). For neonates born to mothers with a BCG vaccine scar, the immediate BCG versus no-BCG MRR was 1.21 (95% CI, .44–3.34), compared with 0.65 (.25–1.64) for those born to mothers with no BCG scar (P for same effect = .37, Supplementary Table 2).

DISCUSSION

Main Findings

Maternal BCG vaccine scars appeared to mainly affect the risk of death from prematurity or respiratory insufficiency during NICU admission, and while maternal scars were associated with a trend for protection from death in the unvaccinated neonates, effects were comparable within randomization strata and not statistically significant overall. The unvaccinated control group received BCG + OPV at discharge from the NICU, after which the beneficial effects of maternal priming were less pronounced.

Strengths and Weaknesses

To our knowledge, this is the first prospective study to assess all-cause survival data by maternal BCG vaccine scar status within a frail cohort enrolled in an RCT with a strict protocol for the provision of neonatal vaccines. The control group was thus not vaccinated without our knowledge.

We initiated the assessments of maternal BCG vaccine scars in 2015 when we became aware of the possible importance. Enrollment into the trial was more than halfway completed at this time point, and the NICU mortality risk had declined during the course of the trial and was substantially lower (<3%) than expected a priori (12%) [15], factors that reduced study power. We furthermore conducted several subgroup analyses of outcomes, which were assessed both before and after discharge, but we did not adjust estimates for multiple testing. Results from this explorative observational study should therefore be interpreted with caution.

In our data, maternal BCG vaccine scars were associated with the maternal education level, consistent with 2 BHP studies from the same period [13, 14]. Similarly, an HNSM maternal scar study also reported a greater age spread for mothers with no scar, and their infants had a slightly lower inclusion weight [13]. For the inclusion weight, the difference (21 g) in the present study was not significant, but even small differences in weight can be vital in LBW (<2500 g) and very low-birth-weight (<1500 g) neonates. In our supplementary analysis, however, effects of maternal scar tended to be more pronounced in normal-birth-weight rather than LBW neonates, and mothers with no scar were not more likely to deliver LBW infants. Residual confounding cannot be ruled out. Speaking against this possibility, for BCG-vaccinated newborns, we note that both in the NICU and after discharge when the control group had been BCG vaccinated, there was a consistent trend for a beneficial effect of maternal scars in male neonates only, an observation consistent with data published in 2021 [13]. If the effects had been caused by an undetected confounder associated with the maternal scar status and mortality risk, it would likely have affected mortality risk equally in both sexes.

Our team conducted daily follow-up in the NICU and the adjacent pediatric ward. Causes of death during admission were assessed by a senior pediatrician blinded to the neonatal and maternal vaccination status. However, our data regarding causes of deaths were limited by the availability of diagnostic tools (eg, no laboratory results or blood cultures) and sometimes insufficient clinical data. Hence, we also encourage caution in interpreting the cause-of-death data.

Two separate BCG strains were used in the main trial, but the overall mortality risk was also declining during the study, and substantial improvements at the NICU occurred in the last part of the study, which used BCG-Japan [15]. Any deductions regarding differential effects of maternal BCG vaccination by offspring BCG strain are therefore likely confounded by chronology, and there were no differences in maternal scar effects across 3 BCG strains used in our larger maternal scar sister study [13].

BCG vaccination does not always result in the development of a scar, and the determinants for scar development are the BCG strain used and the vaccination technique, not host factors [6, 22]. In a study from rural Guinea-Bissau involving >15 000 BCG-vaccinated children, only 52% developed a BCG vaccine scar [23], the lowest scar prevalence reported in the literature [5].

Mothers with no scar in the present cohort will represent a proportion that was BCG vaccinated but did not develop a scar and a proportion that never received BCG. The relative proportions are unknown, since maternal BCG vaccination data for the cohort is not available. Likewise, the coverage of other childhood vaccines in the cohort’s mothers is unknown. If maternal BCG vaccination not leading to a scar is also protective against NICU mortality risk, the effects we report may have been reduced by misclassification bias. Alternatively, maternal BCG vaccine scars might be associated with other characteristics that influence early-life survival without being the protective factor itself.

We reviewed available covariates and did not detect other factors associated with having a BCG vaccine scar aside from maternal educational status. Control for maternal educational status did not change estimates much. The fact that the same pattern of sex-differential effects of maternal BCG among the BCG-vaccinated offspring was found both in the present study and another study, published in 2021 [13] speaks against an overarching unmeasured confounder being the explanation that best fits the available data. It is necessary for the hypotheses generated by the present study to be rigorously tested, from the epidemiological and immunological angle and preferably also within other settings to further pinpoint the relative importance for offspring outcomes.

Consistency With Previous Findings

The hypothesis that maternal BCG vaccination influences offspring outcomes came from studies finding a particularly beneficial effect of measles vaccination when given in the presence of maternal measles antibody [24, 25]. In a concurrent Danish trial, we therefore investigated whether neonatal BCG was particularly beneficial when given to BCG-vaccinated rather than BCG-unvaccinated mothers, which turned out to be the case [10]. Based on these observations, we initiated the present data collection with the hypothesis that maternal BCG vaccination would enhance the beneficial effect of BCG provided to the offspring. There are 2 additional observational studies from Guinea-Bissau focusing on maternal scars from the same period: the sister study from the HNSM maternity ward [13] and a study from rural Bissau [14]. Both studies report maternal scar prevalences of 60%–64% in accordance with the prevalence reported in the present study.

The study from rural Guinea-Bissau involved 1320 pregnant women who had their scar status assessed during pregnancy, which allowed for an assessment of perinatal outcomes. That study reported a trend for fewer adverse pregnancy outcomes associated with maternal BCG vaccine scars; this trend was evident for stillbirths and early neonatal deaths but not miscarriages [14]. Enrollment in the present study occurred later—after the child was born—precluding any assessment of impacts on adverse pregnancy outcomes, but we identified indications of protection against perinatal deaths due to prematurity.

Our analysis is the first to present prospective data on maternal priming effects strictly controlled for offspring vaccination status. During the NICU admission, neonates were monitored daily with minimal loss to follow-up, and the unvaccinated control group was vaccinated only at discharge, after which we continued to monitor effects. This “controlled crossover” RCT design enabled us to report maternal scar effects by offspring vaccination status, providing a unique 2 × 2 possibility to study the effects of maternal priming among both BCG-vaccinated and unvaccinated neonates.

In a 2021 retrospective analysis of a cohort of LBW neonates that had been randomized to BCG versus no BCG vaccine at discharge from the hospital, the maternal scar effect estimate was slightly larger than in the present study [12]. That retrospective cohort featured a mix of BCG-vaccinated and unvaccinated neonates, where crossover would have occurred naturally during follow-up. Effects of maternal priming appeared more enhanced in the BCG-vaccinated group, but the study included a total of only 23 deaths and was underpowered to thoroughly investigate subgroup effects by vaccination status and sex. Outcomes by maternal BCG vaccination status have now been reported for 6 BHP cohorts with different study designs and thus differences in underlying confounding structures, yet maternal BCG scars have consistently been associated with beneficial overall effects on perinatal and infant health [10–14].

While possible immunological pathways for nonspecific effects of neonatal administration of BCG include “trained innate immunity,” induction of emergency granulopoiesis, and/or heterologous T-cell immunity [26–29], the immunological pathways behind maternal priming effects in offspring remain to be identified. One possibility was suggested in a Ugandan study, where the presence of maternal BCG vaccine scars in BCG-vaccinated offspring was strongly associated with an enhanced offspring proinflammatory immune profile at 1 and 6 weeks after neonatal BCG vaccination [30]. In a study from the United Kingdom, it was reported that neonates born to BCG-vaccinated parents have a 4-fold higher T-helper 17 (CD4⁺interleukin 17⁺) cell population as a percentage of the total CD4⁺ T-cell population [31]. The importance of the induction of T-helper 17 cells by parental BCG priming for clinical outcomes is unknown, however.

Implications

BCG is among the world’s most widely used vaccines, with >120 million infants vaccinated per year and total vaccinations exceeding 4 billion [32]. Vaccines are often delayed for various logistic reasons [33], and vaccination often does not result in a scar owing to inadequate vaccination technique and less effective strains [5, 6, 16]. A series of RCTs has demonstrated substantial beneficial effects on neonatal mortality and morbidity rates after BCG vaccination [1–4, 7], and observational studies have demonstrated marked beneficial effects associated with BCG vaccine scars [5, 6, 9]. An emerging series of studies, including the present one, indicate that the presence of a maternal BCG scar is associated with marked beneficial health effects in the offspring. This emphasizes that there is a strong rationale to ensure provision of immunogenic BCG vaccines with adequate vaccination technique as early as possible after birth, to produce high BCG scar rates at the population level.

Conclusions

In a cohort of frail neonates admitted to the NICU, the presence of a maternal BCG vaccine scar tended to be associated with protection from death. This extends the findings of recent studies that point to marked beneficial overall effects for neonatal outcomes associated with maternal priming with BCG, warranting further studies.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Acknowledgments. We thank all participants and their families. We also thank the staff at Hospital Nacional Simão Mendes for their dedicated work, including nurses Luis Camala, Paulo Mendes, and Abdalaha Umaro Cande for treating the neonates in the hospital’s neonatal intensive care unit.

Author contributions. M. B. A. and C. S. B. were the principal investigators and guarantors of the main trial. F. S. B., P. A., T. R. K., M. B. A., and C. S. B. designed the study. F. S. B., I. S., I. M., N. A., and M. B. A. supervised the data collection and data entry. F. S. B. conducted the statistical analyses and wrote the first draft of the article, and all authors approved the final manuscript.

Disclaimer. The funding agencies had no role in designing the study, data collection, data analysis, or the decision to publish.

Financial support. This work was supported by the Karen Elise Jensens Fond; Augustinus Fonden, the Else & Mogens Wedell Wedellborgs Fond, Fonden til Lægevidenskabens Fremme, the Research Center for Vitamins and Vaccines (CVIVA; including support to F. S. B. [PhD grant] and M. B. A.), the University of Southern Denmark (PhD grant to F. S. B.), The European & Developing Countries Clinical Trials Partnership (EDTCP, postdoctoral grant RIA2020EF-3049 to F. S. B.), Novo Nordisk (professorship grant to P. A.), and the European Research Council (starting grant ERC-2009-StG-243149 to C. S. B.). CVIVA was supported by the Danish National Research Foundation (grant DNRF108).

Data sharing. Deidentified participant data with a data dictionary can be shared after approval of a data sharing proposal sent to C. S. B. ([email protected]).

Author notes

Potential conflicts of interest. All authors: No reported conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.