University of Southern Denmark

Reduced Mortality After Oral Polio Vaccination and Increased Mortality After Diphtheria- Tetanus-Pertussis Vaccination in Children in a Low-Income Setting

Øland, Christian Bjerregård; Mogensen, Søren Wengel; Rodrigues, Amabelia; Benn, Christine S.; Aaby, Peter

Published in: Clinical Therapeutics

DOI: 10.1016/j.clinthera.2020.11.010

Publication date: 2021

Document version: Final published version

Document license: CC BY-NC-ND

Citation for pulished version (APA): Øland, C. B., Mogensen, S. W., Rodrigues, A., Benn, C. S., & Aaby, P. (2021). Reduced Mortality After Oral Polio Vaccination and Increased Mortality After Diphtheria-Tetanus-Pertussis Vaccination in Children in a Low- Income Setting. Clinical Therapeutics, 43(1), 172-184.e7. https://doi.org/10.1016/j.clinthera.2020.11.010

Go to publication entry in University of Southern Denmark's Research Portal

Terms of use This work is brought to you by the University of Southern Denmark. Unless otherwise specified it has been shared according to the terms for self-archiving. If no other license is stated, these terms apply:

• You may download this work for personal use only. • You may not further distribute the material or use it for any profit-making activity or commercial gain • You may freely distribute the URL identifying this open access version If you believe that this document breaches copyright please contact us providing details and we will investigate your claim. Please direct all enquiries to [email protected]

Download date: 30. Sep. 2021 Clinical Therapeutics/Volume 43, Number 1, 2021

Reduced Mortality After Oral Polio Vaccination and Increased Mortality After Diphtheria-tetanus-pertussis Vaccination in Children in a Low-income Setting

Christian Bjerregård Øland, MSc1;Søren Wengel Mogensen, MSc1,2; Amabelia Rodrigues, PhD1; Christine S. Benn, DMSc3,4; and Peter Aaby, DMSc1,3 1Bandim Health Project, Indepth Network, Bissau, Guinea-Bissau; 2Department of Math- ematical Sciences, University of Copenhagen, Copenhagen, Denmark; 3Research Centre for Vitamins and , Bandim Health Project, , Copenhagen, Denmark; and 4OPEN, Institute of Clinical Research, University of Southern Denmark/ Odense University Hospital, Odense, Denmark

ABSTRACT MV is given, DTP-vaccinated and MV-unvaccinated children had higher mortality (HR ¼ 2.76; 95% CI, Purpose: The diphtheria-tetanus-pertussis 1.36e5.59) than children who had received MV after (DTP) and oral polio vaccine (OPV) were introduced DTP, and among children who received DTP with in children 3 of 5 months of age in 1981e1983 in MV or after MV, DTP-only vaccination was Bandim, in the capital of Guinea-Bissau. Because DTP associated with a higher mortality than DTP with has been linked to deleterious nonspecific effects OPV (HR ¼ 6.25; 95% CI, 2.55e15.37). (NSEs) and OPV to beneficial NSEs, we followed up Implications: Because the 2 vaccines had differential this cohort to 3 years of age and examined the effects effects and the healthiest children were vaccinated first, of DTP with OPV on all-cause mortality and the selection biases are unlikely to explain the estimated interactions of DTP and OPV with the impact on child survival. OPV had beneficial NSEs, and vaccine (MV). administration of OPV with DTP may have reduced the Methods: DTP and OPV were offered at 3 monthly negative effects of DTP. (Clin Ther. 2021;43:172e184) community weighing sessions. Vaccination groups 2020 The Author(s). Published by Elsevier Inc. This were defined by the last vaccine received. We © is an open access article under the CC BY-NC-ND license compared overall mortality for different groups in (http://creativecommons.org/licenses/by-nc-nd/4.0/). Cox proportional hazards regression models, Key words: child mortality, diphtheria-tetanus- reporting hazards ratios (HRs) with 95% CIs. pertussis vaccine, DTP, , non-specific Findings: The study cohort included 1491 children effects of vaccines, oral polio vaccine. born in Bandim from December 1980 to December 1983. From 3 to 35 months of age, with censoring for MV, children vaccinated with DTP and/or OPV had higher mortality than both unvaccinated children INTRODUCTION ¼ e (HR l.66; 95% CI, 1.03 2.69) and OPV-only The effect of vaccines on overall survival had not been ¼ vaccinated children (HR 2.81; 95% CI, assessed in randomized clinical trials (RCTs) when the e 1.02 7.69); DTP-only vaccinated children had higher Expanded Program on Immunization was initiated in mortality than OPV-only vaccinated children (HR ¼ 3.38; 95% CI, 1.15-e9.93). In the age group of 3e8 months, before MV is administered, DTP- Accepted for publication November 11, 2020 https://doi.org/10.1016/j.clinthera.2020.11.010 only vaccination was associated with a higher 0149-2918/$ - see front matter mortality than DTP with OPV (HR ¼ 3.38; 95% CI, © 2020 The Author(s). Published by Elsevier Inc. This is an open access 1.59e7.20). Between 9 and 35 months of age, when article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

172 Volume 43 Number 1 C.B. Øland et al.

1974. The disease-protective effects were known; effects METHODS on survival were assumed to be proportional to the Demographic Surveillance burden of . Hence, the main interest was at Bandim Health Project (BHP) was started in 1978 in which age to vaccinate.1 However, subsequent studies an urban district. In 1978e1979, mortality among e of the introduction of measles vaccine (MV)2 6 children younger than 5 years was nearly 500 per suggested that the MV had beneficial nonspecific effects 1000.17 Malnutrition was assumed to be the main (NSEs) on child survival (ie, effects on survival not cause; BHP was initiated to determine why children explained by prevention of the vaccine-targeted disease).7 were malnourished.18 However, severe malnutrition The World Health Organization (WHO) recently was not evident, and to understand the high mortality, sponsored a review of the potential NSEs of the BCG we started population follow-up. Four health workers vaccine, diphtheria-tetanus-pertussis vaccine (DTP), identified pregnant women, encouraged women to and MV8,9 for child mortality in low-income countries. attend antenatal clinics, and followed up children BCG and MV had beneficial NSEs. The point estimate younger than 3 years with anthropometric for DTP was in the opposite direction.8 Although measurements. Each health care worker supervised 2 protective against the 3 target diseases, DTP apparently subdistricts; they kept lists of pregnant women and e increased susceptibility to unrelated .10 13 children younger than 3 years. BHP had no Other nonlive vaccines may also be associated with computerized registration system until 1990 but kept increased mortality.14,15 OPV was not examined. an A5 BHP card with weights and vaccination dates There are surprisingly few studies of OPV and DTP for each child. Growth cards were kept by the mother. and child survival in the medical literature. We have The study of nutritional status was planned by the data from 40 years ago when OPV and DTP were Swedish Agency for Research Collaboration and the introduced in Guinea-Bissau in the 1980s. Few sites Ministry of Health in Guinea-Bissau.16 have similar data, so we have gone back to examine the natural experiment of introducing DTP and OPV. Anthropometry In an urban area, weighing sessions were organized The health care workers arranged 3 monthly every 3 months to identify malnourished children. weighing sessions in each subdistrict. Mothers were When vaccines became available in June 1981, OPV notified before a community weighing. The following and DTP were offered from 3 months of age in morning, the child's weight was measured and noted connection with the weighing sessions. on the BHP card. Inadvertently, this created a natural experiment among 3- to 5-month-old children; some received Vaccinations vaccines when just 3 months old, whereas others were BHP organized MV campaigns in December 1979 3,19 nearly 6 months old before they were vaccinated. and December 1980. In June 1981, BHP started Thus, allocation was determined by birthdays and the to provide vaccines at the weighing sessions. A nurse dates of weighing sessions and not by selection biases. from the health center followed the weighing team In this natural experiment, DTP-vaccinated children and vaccinated eligible children. DTP and OPV were had higher mortality than children not yet vaccinated provided to children from 3 months of age and MV with DTP from 3 to 5 months of age.16 to children from 9 months of age. OPV was not We examine the effects of OPV and DTP up to 3 given at birth. Doses of DTP and OPV could be years of age in the same cohort.16 After 6 months of given with 1-month intervals, but because we only age, the unvaccinated group was increasingly arranged quarterly weighing sessions, most children composed of children who had not been vaccinated had longer intervals. In several periods, either OPV 16 because they were frail or malnourished or had or DTP was missing. BCG was rarely provided traveled to rural areas. Because most children were because nurses were not trained to administer subsequently vaccinated with MV, we also examined intradermal vaccinations. possible interactions between DTP and OPV with An expatriate nurse of the supervising field staff MV. Because there are plans to stop OPV globally in sometimes organized catch-up vaccination sessions 2024, we have particularly tried to assess the NSEs without weighing, but vaccinations were noted on of OPV on child survival. BHP cards. Both nurses and mothers thought that

January 2021 173 Clinical Therapeutics sick children should not be vaccinated; BHP cards Study Cohort often indicated that the child was sick, malnourished, We included children born December 3, 1980, to or orphan to explain why an eligible child had not December 31, 1983. The vaccination program started been vaccinated at a weighing session. June 2, 1981.16 Children who never attended any weighing session were not included as unvaccinated. Data Control We excluded orphans because they were not A computerized system became available in breastfed and likely to have different care. Children 1990e1991; weights and vaccination dates from the were included from 91 days of age if examined BHP cards were entered. For the present analysis, before 3 months of age. If first seen after 3 months, dates of visits, weights, and vaccination dates were they were included from the date seen (Figure 2). checked against the original cards (Figure 1). As DTP and OPV were not administered at other health described previously a few cards could not be found.16 centers; follow-up time therefore counted as

Figure 1. Flowchart of study population and children included in the analyses. Numbers in parentheses indicate number of . In the previous analysis,16 1452 children were registered before date of birth plus 183 days.

174 Volume 43 Number 1 C.B. Øland et al. unvaccinated until BHP administered DTP or OPV Statistical Analysis (Figure 2). Time as unvaccinated also came from DTP and OPV vaccinations started on June 2, 1981. children not seen at 3e5 months but seen before 3 Groups defined by the most recent vaccination(s) were months of age (Figure 2). Hence, DTP- and OPV- compared using Cox proportional hazards regression vaccinated and unvaccinated children were from the models with age as the underlying time; proportional same cohort born in Bandim; their randomization to hazard assumptions were tested using the Schoenfeld the DTP/OPV vaccination group or DTP/OPV not residuals test and visual inspection of the cumulative yet vaccinated group depended on birthdate, timing hazard ratios (HRs). A few children had received of weighing sessions, and traveling. The death of a BCG vaccine without documentation at the maternity traveling child could usually be discovered from other ward because they had a BCG scar but no members of the family who remained in the study area. vaccination card, and BCG vaccinations were The 3- to 5-month age group corresponds to the therefore ignored in the analyses. In a sensitivity natural experiment with unbiased allocation to analysis, we censored the children who had vaccination.16 After 6 months, most unvaccinated documented BCG vaccination. Although this reduced children were frail children who had been weighed the power of the study, it did not change the but not vaccinated or who had traveled. estimates (data not shown). Because of the lack of vaccines, some children To avoid survival bias, we used a landmark received DTP-only or OPV-only vaccination.16 approach20; hence, vaccination status was only Traveling patterns did not differ between children updated from the day the information was collected. who had received DTP1 and OPV1 or DTP1 only, We conducted 3 main analyses. First, we compared and these groups were equally likely to receive DTP and/or OPV-vaccinated with unvaccinated subsequent vaccinations (data not shown). children in the 3- to 35-month period; children were This cohort born between December 1980 and censored when known to have received MV. Second, December 1983 is entirely different from the cohort of we conducted an analysis between 3 and 8 months of children aged 6e35 months born between June 1978 and age before children receive MV. Third, we examined December 1980, which we have described elsewhere.13 whether DTP and OPV interacted with MV; this

Figure 2. Natural variation in the timing of vaccination.

January 2021 175 Clinical Therapeutics analysis included only the 9- to 35-month age group. children (HR ¼ 2.81; 95% CI, 1.02e7.69), and DTP- Because vaccine effects often differ by sex, we present only vaccination was associated with a higher main analyses by sex. mortality than OPV-only vaccination (HR ¼ 3.38; Control for confounders was conducted in the 3- to 95% CI, 1.15e9.93) (Table II). Adjusting for WAZ, 35-month age group. Subdistrict, ethnic group, and these HRs were 2.89 (95% CI, 1.06e7.94) and 3.33 twinning did not change results. Control for year of (95% CI, 1.13e9.79), respectively. birth slightly increased the HR of DTP-vaccinated Both analyses with OPV only as the comparator compared with unvaccinated children. There was no (Table II) failed the Schoenfeld test of the clustering of , and control for season did not proportional hazards assumption. Visual inspection change the estimates (data not shown). The WHO z of the cumulative HRs and the partitions presented in score for weight for age (WAZ) was used to assess Table II indicates that this could be explained by the nutritional status. However, we did not adjust for WAZ estimated HR for DTP and OPV versus OPV being in comparisons that involved unvaccinated children larger at ages 3e8 years than at ages 9e35 months because most unvaccinated children had traveled and although >1 in both intervals (HR not defined at had therefore not been weighed at a similar age as the 3e8 months because of no OPV deaths; HR ¼ 2.08; vaccinated children. In the comparison of vaccinated 95% CI, 0.74e5.85 at 9e35 months). Hence, the groups, the last observation of WAZ was carried cumulative HRs of the groups having received DTP forward if an observation was missing. and OPV and OPV only did not cross and stayed apart (Figure 4). RESULTS In the second analysis of children aged 3e8 months, DTP and OPV before MV, children vaccinated with DTP only (13 A total of 1184 children were included in the analysis deaths) had significantly higher mortality than children of the 6- to 35-month age group (Figure 1). The vaccinated with OPV only (0 deaths) (log-rank test: vaccination coverage is indicated in Supplemental P ¼ 0.006) (Table III). Furthermore, children Table I; 95% received DTP1 and OPV1 before 3 years vaccinated with DTP only had an HR of 3.92 (95% of age, but coverage for the third dose was CI, 1.78e8.62) compared with unvaccinated children approximately 80%. The groups did not differ with and a HR of 3.38 (95% CI, 1.59e7.20) compared respect to background factors, such as birthweight and with children vaccinated with DTP and OPV. weight before 6 months of age (Table I). At 6e8 months, WAZ was better for children who had MV Period received DTP and OPV, DTP only, or OPV only than In the third analysis, we followed up all children for children who remained unvaccinated after after 9 months of age when most children receive participating in a weighing session. After 9 months of MV (Table IV and Supplemental Figure 2). age, there was no clear difference in nutritional status Compared with children who received MV after measured by WAZ. Background factors after 12 DTP, DTP-vaccinated and MV-unvaccinated children months of age are listed in Supplemental Table II; had a HR of 2.76 (95% CI, 1.36e5.59). This unvaccinated children participated in fewer weighing negative effect may have been more pronounced for sessions, reflecting that they traveled more. girls (HR ¼ 5.13; 95% CI, 1.52e17.28) than boys In the first analysis, with censoring for MV, the DTP- (HR ¼ 1.76; 95% CI, 0.74e4.20) (test for and/or OPV-vaccinated children had a HR of 1.22 (95% interaction: P ¼ 0.16) (Supplemental Table IV). CI, 0.73e2.04) between 6 and 35 months of age Children with out-of-sequence vaccinations (ie, DTP compared with unvaccinated children. Between 3 and with MV or DTP after MV) tended to have higher 35 months, the HR was 1.66 (95% CI, 1.03e2.69) mortality than the MV after DTP children (Table IV (Table II): 2.13 (95% CI, 1.00e4.54) for girls and and Supplemental Table V). When we compared no 1.43 (95% CI, 0.78-2e59) for boys (Supplemental OPV with OPV among children who had DTP with Table II). The results for different age groups have or after MV, the HR was 6.25 (95% CI, been visually presented in Figure 3 (Supplemental 2.55e15.37); this effect may have been worse for Figure 1). DTP- and/or OPV-vaccinated children had girls, who had a HR of 13.31 (95% CI, 4.21e42.03) also higher mortality than OPV-only vaccinated (test of interaction: P ¼ 0.11) (Supplemental Table VI).

176 Volume 43 Number 1 aur 2021 January

Table I. Background factors for the different most recent vaccination groups observed up to 1 year of age.* y Factor Unvaccinated DTP Only (No MV) DTP and OPV (No MV) OPV Only (No MV) MV (n ¼ 400) (n ¼ 731) (n ¼ 225) (n ¼ 633) (n ¼ 155) z Birth WAZ, mean (SD) −0.18 (0.98) [313] −0.17 (1.02) [105] −0.26 (0.92) [281] −0.12 (1.25) [68] −0.18 (0.93) [156] [n] WAZ at <6 months of −0.37 (1.21) [685] −0.52 (1.36) [119] −0.38 (1.16) [394] −0.50 (1.35) [66] −0.86 (1.21) [8] age, mean (SD) [n] WAZ at 6e8 months of −1.07 (1.49) [77] −0.79 (1.27) [104] −0.65 (1.25) [344] −0.79 (1.22) [81] −0.78 (1.11) [75] age, mean (SD) [n] WAZ at 9e11 months of −0.98 (1.39) [42] −1.22 (1.15) [49] −1.11 (1.25) [117] −1.24 (1.36) [29] −0.87 (1.17) [359] age, mean (SD) [n] x Male sex, % 52.3 53.8 49.3 52.9 54.7 x Twin, % 2.5 4.6 1.9 4.4 1.3 x Ethnic group, % Pepel 48.7 45.1 50.0 48.5 50.7 Balanta 13.2 11.8 13.2 15.4 10.7 Other 38.1 43.1 36.8 36.0 38.7 Examination rate,|| 3.78 (728/192.8) 3.44 (74/21.5) 3.14 (273/87.0) 3.06 (41/13.4) 6.47 (48/7.4) (events/person-years)

DTP ¼ diphtheria-tetanus-pertussis vaccine; MV ¼ measles vaccine; OPV ¼ oral polio vaccine; WAZ ¼ weight for age z score. * Sample sizes are the number of children in the group, using only 1 observation per child but allowing a child to be included in all groups at once. y MV includes all children who have received MV and is therefore not necessarily the most recent vaccination. z WAZ is defined as the WAZ measured at birth or before 14 days of age. x The proportions of sex, twins, and ethnic group calculated before 9 months of age (where MV is scheduled). || Examination rate calculated as the number of all observations starting with the child being present divided by person-years. C.B. Ø ade al. et land 177 Clinical Therapeutics

DISCUSSION x Main Observations 5.29) 20.93) 47.96) 9.93) Children who received only the OPV had lower e e e e

z mortality than children who had received DTP only DTP Only

0.10 or DTP and OPV. Although unvaccinated children Only vs OPV ¼ were increasingly frail after 6 months of age, DTP P 3.38 (1.15

x vaccination was associated with slightly higher mortality than being unvaccinated. DTP vaccinated 7.69) 3.11) 0.75 (0.10 21.93) 2.18 (0.23 38.18) 6.00 (0.75 e e e e and MV-unvaccinated children had a nearly 3-fold z

OPV vs higher mortality than children having MV after 0.21 OPV Only DTP and/or

¼ DTP as their most recent vaccination. Nonlive DTP P was associated with negative NSEs,21 whereas live OPV and MV were associated with beneficial HR (95% CI) 3.70) 2.81 (1.02 30.37) 0.64 (0.13 2.86) 2.91 (0.39 3.57) 5.22 (0.71 31.87) effects. OPV reduced the negative effects of DTP e e e e e

DTP because children who received DTP and OPV had Only vs lower mortality than children who received DTP Unvaccinated only at 3e8 months of age before MV and when vaccines were given out of sequence with MV after 2.69) 2.00 (1.08 19.34) 2.75 (0.25 2.40) 0.74 (0.19 2.46) 1.46 (0.60 14.30) 9.54 (2.86 9 months of age. e e e e e oral polio vaccine. OPV vs ¼ Strengths and Weaknesses Unvaccinated We used 40-year-old data to assess the effects of OPV and DTP, when given separately and combined. 0.05). Few sites have data from when these vaccines were <

P first introduced in a low-income countries with high mortality. When introducing these vaccines, there were many natural experiments in which only 1 measles vaccine; OPV

¼ vaccine was given. Because the removal of OPV is planned, it is particularly important to understand y the nonspecific and specific effects of OPV. To ensure comparability, we included only children weighed in community examinations (Figure 2).16 DTP 100.

× and/or OPV vaccination was associated with higher hazard ratio; MV

¼ mortality than being DTP unvaccinated (HR ¼ 1.66; 95% CI, 1.03e2.69). This is bad news. First, the 5- fold increased mortality previously reported in 3- to 5-month-old children16 cannot be dismissed as a random result because of small numbers in that age

Mortality Rate (Deaths/Person-Years) group. DTP-vaccinated children had 1.7 times higher mortality than unvaccinated children even though the frail and malnourished remained unvaccinated. Second, the higher mortality in girls after DTP is unnatural because girls did not have higher mortality than boys in the prevaccination era.12 Unvaccinated DTP and OPV DTP Only OPV Only DTP and/or OPV DTP and/or The finding that, from 9 months of age, DTP-

diphtheria-tetanus-pertussis vaccine; HR vaccinated and MV-unvaccinated children had higher ¼ 35 2.6 (1/39.1) 5.8 (5/86.9) 6.9 (2/29.0) 9.0 (2/22.1) 6.0 (7/115.9) 2.38 (0.29 23 6.5 (7/107.2) 7.1 (13/183.6) 4.8 (3/62.0) 2.2 (1/45.3) 6.5 (16/245.6) 0.99 (0.41 mortality than children vaccinated with MV after 35 6.5 (26/403.0) 9.4 (50/532.8) 11.6 (19/163.3) 3.4 (4/116.8) 9.9 (69/696.1) 1.66 (1.03 11 9.1 (12/131.2) 11.9 (25/211.0) 14.2 (8/56.3) 2.4 (1/41.2) 12.3 (33/267.3) 1.27 (0.66 5 4.8 (6/125.3) 13.6 (7/51.3) 37.5 (6/16.0) 0 (0/8.2) 19.3 (13/67.3) 4.98 (1.73 * e e Children with no registered sex included. Mortality rate is presentedLog-rank as test (event/person-years) for equality. HR fails the Schoenfeld residual test of the proportional hazards assumption ( e e e All 3 24 12 Table II. Mortality rates and HRs by most recent vaccination and age (MV has been censored). 6 DTP * y z x Age Group, mo 3 DTP could possibly be explained by bias (eg, weak

178 Volume 43 Number 1 C.B. Øland et al.

Figure 3. Summary of comparisons of different vaccines at different ages. DTP ¼ diphtheria-tetanus-pertussis vaccine; HR ¼ hazard ratio; MV ¼ measles vaccine; OPV ¼ oral polio vaccine. children not being vaccinated with MV). However, Comparison With Previous Studies of DTP and following this logic, unvaccinated children should be OPV even weaker than DTP-vaccinated children and In the WHO-sponsored meta-analysis of the NSEs should have had higher HR compared with children of DTP, BCG, and MV, for child mortality before 5 vaccinated with MV after DTP. That was not the years of age, DTP was associated with a 38% (95% case (Table IV); unvaccinated children had only CI, −8%e108%)8 increase in mortality that was not slightly higher mortality than children vaccinated statistically significant. However, the WHO- with MV after DTP, and DTP-vaccinated children sponsored review included studies with survival bias. had higher mortality than unvaccinated children. Excluding studies with survival bias, DTP-vaccinated Most studies of vaccines and their impact on children had a 2-fold higher mortality (HR ¼ 2.00; survival report only 1 mortality estimate. However, 95% CI, 1.50e2.67).21 This unfortunate result is the dynamics of vaccination programs imply that the further strengthened by 2 recent studies of the unvaccinated group will become increasingly frail introduction of DTP to children aged 3e5 months with age because mothers and health care workers and 6e35 months; both found DTP to be associated are reluctant to vaccinate ill children. As a result, the with 2-fold higher mortality.13,16 Negative NSEs of HR changes over time. So, if the initial estimate was DTP were stronger for girls than for boys.13,16 negative,16 one might actually get a positive estimate The DTP estimates are consistent, but they may by waiting until only very frail children are left in the underestimate the harm from DTP: unvaccinated unvaccinated group, as done in a study from Cebu, children are inherently disadvantaged; they have the Philippines.22 We followed up DTP-vaccinated poorer WAZ scores, and their BHP card usually and unvaccinated children to see whether frailty bias indicates that the child is sick or malnourished. would eventually produce a beneficial HR for DTP, This study found evidence that OPV has beneficial but that did not happen. From 3 to 5 months of age, NSEs: OPV only was associated with a lower when the comparison was a natural experiment, mortality than DTP only, and OPV and DTP were DTP-vaccinated children had an HR 5 times that of associated with a lower mortality than DTP only unvaccinated children.16 Subsequently, the HR when given with or after MV. Only 2 previous decreased, but it continued to be at 1 among older studies have tested the NSEs of OPV: in an RCT in children despite frailty bias. Guinea-Bissau, OPV at birth reduced infant mortality

January 2021 179 Clinical Therapeutics

rural Guinea-Bissau,11 OPV was not used, and the HR for DTP-vaccinated versus DTP-unvaccinated children was 5.00 (95% CI, 0.63e39.7). From 1985 to 1987, DTP and OPV were coadministered, and the HR was 1.90 (95% CI, 0.91e3.97). In the 3- to 8-month age group in the present study, DTP only was associated with higher mortality than DTP and OPV (HR ¼ 3.38; 95% CI, 1.59e7.20). Likewise, when DTP was administered with MV or after MV, DTP only versus DTP and OPV was associated with higher mortality (HR ¼ 6.25; 95% CI, 2.55e15.37).

Interpretation Epidemiologic studies have found increased all-cause mortality associated with nonlive DTP vaccination Figure 4. Cumulative hazard ratios of different compared with reduced all-cause mortality associated vaccination groups. DTP ¼ diphtheria- with live OPV and MV that is not explained by tetanus-pertussis vaccine; OPV ¼ oral reduced mortality from polio or measles. The polio vaccine. beneficial NSEs of live vaccines, including BCG and vaccinia, may relate to epigenetic reprogramming of by 32% (range, 0%e57%) before the children the innate immune system, enhancing protection e received campaign OPV,23 and, in natural against unrelated infections.26 28 On the other hand, experiments, campaigns with OPV were associated nonlive vaccines may have the opposite effect, with 19% (range, 5%e32%) lower mortality.24 including tolerance and increased susceptibility to Hence, all the available studies point to OPV having unrelated infections.29,30 In an experimental study e beneficial NSEs.23 25 among young Dutch girls, diphtheria and tetanus In other words, DTP and OPV have contrasting toxoids and acellular pertussis (DTaP)einactivated effects. When DTP was first introduced in 1984 in polio vaccine (IPV) was associated with down- regulation of cytokine responses, which were abrogated if the child received BCG together with Table III. Mortality rates and HRs from 3 to 8 DTaP-IPV or BCG after DTaP-IPV.30 months of age by most recent vacci- After the WHO-sponsored review, WHO requested nation status (MV before 9 months has RCTs to settle the dispute about the importance of been censored). NSEs.8,31 The Strategic Advisory Group of Experts on Vaccination Mortality per 100 DTP-Only HR Immunization delegated the setting of priorities and Group Person-Years (95% CI) planning of RCTs to the Immunization and Vaccines (Deaths/ Related Implementation Research Advisory Committee; Person-Years) 6 years later nothing has happened in relation to assessing the potential negative NSEs of DTP in an RCT. DTP only 28.4 (13/45.8) Reference Hence, we need to triangulate all available data to < OPV only 0 (0/28.8) P 0.01* evaluate the impact of DTP. First, DTP would be e Unvaccinated 7.4 (15/203.7) 3.92 (1.78 8.62) expected to be associated with reduced mortality e DTP with OPV 8.5 (14/165.7) 3.38 (1.59 7.20) because inherent biases favor the DTP-vaccinated group.21,32 Second, all studies of DTP introduction DTP ¼ diphtheria-tetanus-pertussis vaccine; HR ¼ hazard have found a negative effect.13 Third, routine DTP is ratio; MV ¼ measles vaccine; OPV ¼ oral polio vaccine. * Log-rank test for equality. consistently associated with increased female mortality.33 Fourth, DTP after MV is associated with increased female mortality.8,10 Fifth, live high-titer

180 Volume 43 Number 1 C.B. Øland et al.

Table IV. Mortality rates and hazard ratios (HR) by disjoint vaccination groups exploring timing of vaccination and most recent vaccination, children aged 9e36 months of age.

Group (Most Recent Vaccines) Mortality Rate* HR (95% CI) HR (95% CI) (Deaths/Person-Years) With MV After With Unvaccinated DTP as Reference as Reference

MV after DTP (with or without OPV) 2.2 (10/448.0) Reference 0.53 (0.22e1.25) DTP with MV (with or without OPV) 4.7 (14/295.6) 1.96 (0.87e4.41) 1.03 (0.47e2.26) DTP after MV (with or without OPV) 2.9 (16/554.1) 1.44 (0.65e3.18) 0.76 (0.34e1.67) DTP with or without OPV (no MV) 8.5 (46/543.3) 2.76 (1.36e5.59) 1.45 (0.77e2.74) Most recent vaccine containing only 4.1 (3/73.5) 1.40 (0.38e5.12) 0.74 (0.21e2.61) live vaccines, MV, and/or OPV (no DTP) Unvaccinated (no DTP, MV, or OPV) 6.0 (12/199.3) 1.90 (0.80e4.49) Reference y DTP with or after MV (with or without OPV) 3.5 (30/849.7) 1.64 (0.80e3.36) 0.88 (0.43e1.77)

* Mortality rate presented as (event/person-years) × 100. y All groups represented in model with the 2 groups (DTP with MV [with or without OPV] and DTP after MV [with or without OPV]) combined. DTP ¼ diphtheria-tetanus-pertussis vaccine; HR ¼ hazard ratio; MV ¼ measles vaccine; OPV ¼ oral polio vaccine.

MV (HTMV) was given at 4e5 months of age, and vaccine12; however, when OPV campaigns were girls had a 2-fold increase in mortality if they conducted before children received early MV, early received nonlive vaccines (DTP or IPV) after MV did not reduce the mortality associated with HTMV.10 Hence, HTMV became associated with 2- DTP.40 Another strategy might be to give OPV with fold higher female mortality.10 Sixth, other nonlive DTP or MV after DTP.7,14 In addition, vaccines (pentavalent, hepatitis B virus, IPV, RTS,S, administration of BCG with DTP may reduce the and H1N1) are also associated with increased female adverse effects of DTP,30,41 which suggests that it is mortality even though girls did not have higher important to conduct RCTs of giving (or not) a e mortality in the prevaccination era.12,24,33 37 second dose of BCG with the third and last priming Seventh, a live vaccine given shortly after DTP may dose of DTP at 14 weeks of age. Such studies are reduce the negative NSEs.12 possible and could contribute to bringing the negative With this level of consistency among independent effects of DTP under better control. observations, it is likely that DTP has negative Because polio is about to be eradicated, the use of effects. To paraphrase Cornfield, “If only one OPV is planned to stop in 2024. Our findings suggest hypothesis can explain all the evidence, then the that withdrawing OPVs might increase child question is settled, even if the evidence is mortality.42 OPV only was associated with lower observational.”38 mortality than DTP only, and OPV with DTP reduced the negative effects of DTP only. This Implications possibility is supported by RCTs and observational More studies should investigate ways to reduce studies of routine OPV vaccinations and OPV negative effects of DTP and pentavalent. One campaigns.23,24,43 Hence, we need to mitigate the solution might be to replace the whole-cell pertussis potential negative effects of removing a vaccine with vaccine in DTP with the live pertussis vaccine that highly beneficial NSEs.42,44 For example, we need to has beneficial NSEs in animal studies.39 We have test whether more frequent use of MV and BCG can previously reported that live vaccines may reduce the replace the beneficial effects of OPV when it is negative effect if given shortly after a nonlive withdrawn.

January 2021 181 Clinical Therapeutics

CONCLUSIONS statistical analyses. The first draft was written by P. It is a sad conclusion that DTP is associated with Aaby and C. Bjerregård Øland; all authors increased female mortality. No data, without survival contributed to the final version of the paper. C. bias, contradict this statement. Although the WHO- Bjerregård Øland and P. Aaby will act as guarantors sponsored review concluded 6 years ago that DTP of the study. was associated with 38% (95% CI, −8%e108%) higher mortality and had completely different effects than BCG and MV,8 there has been no attempt to fi de ne how this issue should be resolved. REFERENCES DTP is the most commonly used vaccine, and the 1. Expanded Programme on Immunization. The optimal age possibility that it might increase mortality demands for measles immunization. Weekly Epidemiol Rec. 1982;57: that we urgently obtain more information about the 89e91. effect of DTP on all-cause mortality. 2. Aaby P, Bukh J, Lisse IM, Smits AJ. Measles vaccination and Given the beneficial NSEs of OPV, OPV should be child mortality. Lancet. 1981;ii:93. tested to see if it reduces the risk of severe COVID- 3. Aaby P, Bukh J, Lisse IM, Smits AJ. Measles vaccination and 19 infection.45 If such RCTs find that OPV reduces reduction in child mortality: a community study from e the risk of COVID-19, it may lead to a Guinea-Bissau. J Infect. 1984;8:13 21. 4. Aaby P, Bhuyia A, Nahar L, Knudsen K, de Francisco A, reconsideration of the current plans to stop using OPV. Strong M. The survival benefit of measles immunization may not be explained entirely by the prevention of measles FUNDING SOURCES disease: a community study from rural Bangladesh. Int J The present study and cleaning of the original data Epidemiol. 2003;32:106e115. were supported by a common grant from DANIDA 5. Holt EA, Boulos R, Halsey NA, et al. Childhood survival in and the Novo Nordisk Foundation. The work on Haiti: protective effect of measles vaccination. Pediatrics. e nonspecific effects of vaccines was supported by grant 1990;86:188 194. fl 104. Dan.8. f. from the Danish Council for 6. The Kasongo Project Team. In uence of measles vaccination on survival pattern of 7-35-month-old children Development Research, Ministry of Foreign Affairs, in Kasongo, Zaire. Lancet. 1981;i:764e767. Denmark, Novo Nordisk Foundation, and grant 7. Aaby P, Samb B, Simondon F, Coll Seck AM, Knudsen K, Health-F3-2011-261375 from European Union FP7 Whittle H. Non-specific beneficial effect of measles support for OPTIMUNISE. CSB held starting grant immunisation: analysis of mortality studies from developing ERC-2009-StG-243149 from the ERC. Research countries. Br Med J. 1995;311:481e485. Centre for Vitamins and Vaccines is supported by 8. Higgins JP, Soares-Weiser K, Lopez-Lopez JA, et al. grant DNRF108 from the Danish National Research Association of BCG, DTP, and measles containing vaccines Foundation. Peter Aaby, DMSc, held a research with childhood mortality: systematic review. BMJ. professorship grant from the Novo Nordisk 2016;355:i5170. Foundation. The funding agencies had no role in the 9. Strategic advisory group of experts on immunization. Week e study design, data collection, data analysis, data Epidemiol Rec. 2014;89:233 235. interpretation, or the writing of the report. 10. Aaby P, Jensen H, Samb B, et al. Differences in female-male mortality after high-titre measles vaccine and association with subsequent vaccination with diphtheria-tetanus- DISCLOSURES pertussis and inactivated poliovirus: reanalysis of West The authors have indicated that they have no conflicts African studies. Lancet. 2003;361:2183e2188. of interest regarding the content of this article. 11. Aaby P, Jensen H, Gomes J, Fernandes M, Lisse IM. The introduction of diphtheria-tetanus-pertussis vaccine and child mortality in rural Guinea-Bissau: an observational ACKNOWLEDGMENT study. Int J Epidemiol. 2004;33:374e380. Ø C. Bjerregård land and P. Aaby proposed the study. 12. Aaby P, Benn CS, Nielsen J, Lisse IM, Rodrigues A, Ravn H. P. Aaby collected the original data. A. Rodrigues is Testing the hypothesis that diphtheria-tetanus-pertussis responsible for the demographic surveillance system. vaccine has negative non-specific and sex-differential effects SWM and P. Aaby cleaned the data. C. Bjerregård on child survival in high-mortality countries. BMJ Open. Øland and S. Wengel Mogensen conducted the 2012;2, e000707.

182 Volume 43 Number 1 C.B. Øland et al.

13. Aaby P, Mogensen SW, Andersen A, 22. Chan GJ, Moulton LH, Becker S, Dis. 2019. https://doi.org/10.1093/ Rodrigues A, Benn CS. Evidence of Munoz~ A, Black RE. Non-specific cid/ciz246. increase in mortality after the effects of diphtheria-tetanus- 31. Pollard AJ, Finn A, Curtis N. Non- introduction of diphtheria-tetanus- pertussis vaccination on child specific effects of vaccines: plausible pertussis vaccine to children aged 6- mortality in Cebu, the Philippines. Int and potentially important, but 35 months in Guinea-Bissau: a time J Epidemiol. 2007;36:1022e1029. implications uncertain. Arch Dis Child. for reflection? Front Public Health. 23. Lund N, Andersen A, Hansen AS, 2017;102:1077e1081. 2018;6:79. et al. The effect of oral polio vaccine 32. Benn CS, Fisker AB, Rieckmann A, 14. Benn CS, Fisker AB, Rieckmann A, at birth on infant mortality: a Jensen AG, Aaby P. How to evaluate Sørup S, Aaby P. Vaccinology: time randomized trial. Clin Infect Dis. potential non-specific effects of to change paradigm? Lancet Inf Dis. 2015;61:1504e1511. vaccines: the quest for randomized 2020;20:e274ee283. 24. Andersen A, Fisker AB, Rodrigues A, trials or time for triangulation? Expert 15. Aaby P, Fisker AB, Bjorkman,€ et al. National immunization Rev Vac. 2018;17:411e420. Benn CS. Rolling out the RTS,S campaigns with oral polio vaccine 33. Aaby P, Ravn H, Fisker AB, et al. Is vaccine: to test or not to test reduce all-cause mortality: a natural diphtheria-tetanus-pertussis (DTP) the effect on mortality? BMJ. experiment within seven randomized associated with increased female 2020;368:l6920. trials. Front Public Health. 2018;6:13. mortality? A meta-analysis testing the 16. Mogensen SW, Andersen A, 25. Aaby P, Rodrigues A, Biai S, et al. hypotheses of sex-differential non- Rodrigues A, et al. The introduction of Oral polio vaccination and low case specific effects of DTP vaccine. Trans diphtheria-tetanus-pertussis and oral fatality at the paediatric ward in R Soc Trop Med Hyg. 2016;110(10): polio vaccine among young infants in Bissau, Guinea-Bissau. Vaccine. 570e581. an urban African community: a natural 2004;22:3014e3017. 34. Fisker AB, Biering-Sørensen S, experiment. EBioMedicine. 2017;17:192 26. Kleinnijenhuis J, Quintin J, Preijers F, Lund N, et al. Contrasting female- e198. et al. Bacille Calmette-Guerin induces male mortality ratios after routine 17. Smedman L, Aaby P, Lindeberg A, NOD2-dependent nonspecific vaccinations with pentavalent versus Zetterstrom R. Survival 0-6 years of protection from reinfection via measles and yellow fever vaccine: a age in a periurban community in epigenetic reprogramming of cohort study from Guinea-Bissau. Guinea-Bissau: a longitudinal monocytes. Proc Natl Acad Sci U S A. Vaccine. 2016;34:4551e4557. assessment. Ann Trop Pediatr. 1986;6: 2012;109:17537e17542. 35. Garly ML, Jensen H, Martins CL, 67e72. 27. Blok BA, Jensen KJ, Aaby P, et al. et al. Hepatitis B vaccination 18. Aaby P, Bukh J, Lisse IM, Smits AJ. Opposite effects of Vaccinia and associated with higher female than Measles mortality, state of nutrition, modified Vaccinia Ankara on trained male mortality in Guinea-Bissau: an and family structure: a community immunity. Eur J Clin Microbiol Infect observational study. Pediatr Infect Dis study from Guinea-Bissau. J Infect Dis. Dis. 2019;38:449e456. J. 2004;23:1086e1092. 1983;147:693e701. 28. Arts RJW, Moorlag S, Novakovic B, 36. Klein SL, Shann F, Moss WJ, et al. 19. Mogensen SW, Aaby P, Smedman L, et al. BCG vaccination protects RTS,S malaria vaccine and increased et al. Introduction of standard against experimental viral infection in mortality in girls. MBio. 2016;7. measles vaccination in an urban humans through the induction of 37. Aaby P, Garly ML, Nielsen J, et al. African community in 1979 and cytokines associated with trained Increased female-male mortality ratio overall child survival: a reanalysis of immunity. Cell Host Microbe. 2018;23: associated with inactivated polio and data from a cohort study. BMJ Open. 89e100 e5. diphtheria-tetanus-pertussis 2016;6(12), e011317. 29. Leentjens J, Kox M, Stokman R, et al. vaccines: observations from 20. Jensen H, Benn CS, Lisse IM, BCG vaccination enhances the vaccination trials in Guinea-Bissau. Rodrigues A, Andersen PK, Aaby P. immunogenicity of subsequent Pediatr Infect Dis J. 2007;26:247e252. Survival bias in observational studies influenza vaccination in healthy 38. Cornfield J. Principles of research. Am of the impact of routine vaccinations volunteers: a randomized, placebo- J Ment Defic. 1959;64:240e252. on childhood survival. Trop Med Int controlled pilot study. J Infect Dis. 39. Cauchi S, Locht C. Non-specific Health. 2007;12:5e14. 2015;212:1930e1938. effects of live attenuated pertussis 21. Aaby P, Ravn H, Benn CS. The WHO 30. Blok BA, de Bree LCJ, vaccine against heterologous review of the possible non-specific Diavatopoulos DA, et al. Interacting infectious and inflammatory effects of diphtheria-tetanus- non-specific immunological effects of diseases. Front Immunol. 2018;9:2872. pertussis vaccine. Pediatr Infect Dis J. BCG and Tdapf vaccinations: an 40. Fisker AB, Nebie E, Schoeps A, et al. 2016;35:1247e1257. explorative randomized trial. Clin Inf A two-centre randomised trial of an

January 2021 183 Clinical Therapeutics

additional early dose of measles Bangladesh. EBioMedicine. 2017;22: 44. Aaby P, Benn CS. Stopping live vaccine: effects on mortality and 173e180. vaccines after disease eradication measles antibody levels. Clin Infect 42. Aaby P, Benn C. Beneficial non- may increase mortality. Vaccine. Dis. 2017 Nov 21. https://doi.org/ specific effects of oral polio vaccine: 2020;38:10e14. 10.1093/cid/cix1033. implications for the cessation of 45. Chumakov K, Benn CS, Aaby P, 41. Aaby P, Andersen A, Ravn H, OPV? Clin Inf Dis. 2017;65:420e421. Kottilli S, Gallo R. Can existing live Khalequzzaman Md. Co- 43. Andersen A, Fisker AB, Rodrigues A, vaccines prevent COVID-19? Science. administration of BCG and et al. National immunisation 2020;368:1187e1188. diphtheria-tetanus-pertussis (DTP) campaigns with oral polio vaccine vaccinations may reduce infant may reduce all-cause mortality: an mortality more than the WHO- analysis of 13 years’ of demographic schedule of BCG first and then DTP. surveillance data from an urban A Re-analysis of demographic African area. Clin Infect Dis. 2020 Sep surveillance data from rural 19. https://doi.org/10.1093/cid/ ciaa1351/5909021 (in press).

EDITOR’S NOTE At the outset of the COVID-19 pandemic and while we await formal approval and widespread distribution of a specific vaccine, there has been a revival in the concept of the “collateral benefit” that e other live vaccines could offer against SARS-CoV2.1 3 This could be particularly useful in resource- limited settings where, due to disparities between rich and poor countries, the wait for COVID-19 vaccines may be several years.4 In this issue of Clinical Therapeutics, we feature a paper by Dr. Peter Aaby and co-authors that describes work to support this hypothesis. They studied children who received oral polio vaccine in the country of Guinea-Bissau and demonstrated a decreased overall mortality beyond the direct attributable effect of the vaccine. This research is difficult to perform and often met with concerns about confounding factors that could account for the effect.2 However, we commend the authors for their efforts and are pleased to feature this content for our readers. Ravi Jhaveri Co-Editor-in-Chief

REFERENCES 1. Rabin RC. Can an Old Vaccine Stop the New Coronavirus? New York Times; April 3, 2020. https://www.nytimes.com/2020/04/03/ health/coronavirus-bcg-vaccine.html. Accessed December 3, 2020. 2. Rabin RC. Old Vaccines May Stop the Coronavirus, Study Hints. Scientists Are Skeptical. New York Times; July 29, 2020. https://www. nytimes.com/2020/07/29/health/coronavirus-vaccines.html. Accessed December 3, 2020. 3. Williams S. How Some Vaccines Protect Against More than Their Targets. The Scientist; November 1, 2020. https://www.the- scientist.com/features/how-some-vaccines-protect-against-more-than-their-targets-68059. Accessed December 3, 2020. 4. Doucleff M. Developed Countries Plan To Start Vaccination Soon. What About The Rest Of The World? National Public Radio; December 3, 2020. https://www.npr.org/2020/12/03/942345198/developed-countries-plan-to-start-vaccination-soon- what-about-the-rest-of-the-wo. Accessed December 3, 2020.

Address correspondence to: Peter Aaby, DMSc, Statens Serum Institut, Bandim Health Project, Statens Serum Institut, Artillerivej 5, Copenhagen S 2300, Denmark. E-mail: [email protected]

184 Volume 43 Number 1 C.B. Øland et al.

APPENDIX

Supplementary Figure 1. Visual presentation of the mortality rates and the HR estimates for comparing different vaccines in different age groups. Mortality rates of the two groups compared are overlaid over each other (blue represent the comparator groups (i.e. DTP ± OPV or DTP-only) and black represent the baseline groups (i.e. unvaccinated or OPV-only)); mortality rate presented as (event/pyrs)*100; the hazard ratio esti- mates of the Cox proportional hazards model are then presented graphically in a log transformed axis; Log-rank test for equality in case one arm had no fatalities.

January 2021 184.e1 Clinical Therapeutics

Supplementary Figure 2. Visual presentation of the mortality rates and HR estimates for comparing different vaccines after introduction of measles vaccine. Mortality rate presented as (event/ pyrs)*100; Hazard ratio estimates of the Cox proportional hazards model are pre- sented graphically in a log transformed axis.

184.e2 Volume 43 Number 1 C.B. Øland et al.

Supplementary Table 1. Accumulated vaccination coverages up to 3 years of age.

Year of birth 1980 1981 1982 1983 1980e1983 N ¼ 32 N ¼ 433 N ¼ 354 N ¼ 398 N ¼ 1217

Median age [IQR] DTP1 182 (160e292) 184 (126e350) 157 (110e256) 134 (101e203) 155 (112e261) DTP3 503 (397e731) 519 (363e731) 386 (265e580) 353 (253e497) 412 (283e603) OPV1 182 (166e222) 152.5 (113e263) 155 (110e248.5) 175 (115e284) 161 (113e264) OPV3 301 (267e488) 386 (264e599) 362 (263e574) 498 (367e683) 425 (280e633) MV1 389 (344e713) 387 (305e570) 323 (285e433) 349 (301e445) 355 (298e487.5) BCG 296.5 (292e301) 66 (29e113) 34 (10e64) 33 (13e108.5) 39 (15e94) Vaccine coverage at 6 months of age. (censoring children ending observation time before 6 months of age) DTP1 48.39% 42.92% 54.94% 66.06% 54.09% DTP3 3.23% 2.59% 3.20% 7.51% 4.39% OPV1 48.39% 53.77% 56.10% 45.34% 51.56% OPV3 3.23% 3.07% 3.20% 3.11% 3.12% MV1 0.00% 2.83% 0.00% 0.26% 1.10% BCG 0.00% 14.15% 17.15% 24.35% 17.97% Vaccine coverage at 1 year of age. (censoring children ending observation time before 1 year of age) DTP1 86.21% 67.26% 75.87% 88.51% 77.07% DTP3 10.34% 16.50% 34.29% 42.53% 29.83% OPV1 89.66% 73.60% 79.37% 73.56% 75.69% OPV3 51.72% 30.46% 35.56% 16.95% 28.18% MV1 24.14% 32.23% 47.30% 47.70% 41.34% BCG 6.90% 14.47% 18.10% 26.15% 19.06% Vaccine coverage at 3 years of age. (censoring children ending observation time before 3 years of age) DTP1 95.83% 94.60% 96.97% 97.41% 96.26% DTP3 62.50% 75.18% 85.28% 86.30% 81.44% OPV1 95.83% 93.17% 95.67% 94.44% 94.40% OPV3 87.50% 73.02% 80.09% 78.15% 77.21% MV1 79.17% 80.58% 85.28% 91.48% 85.55% BCG 4.17% 16.91% 18.18% 31.11% 21.67%

Note: N denotes the total number of children in the group.

January 2021 184.e3 184.e4 lnclTherapeutics Clinical

Supplementary Table 2. Background factors for different most recent vaccination groups observed between 12 and 35 months of age.

Unvaccinated DTP-only [no MV] DTP + OPV [no MV] OPV-only [no MV] MV-after-DTP MV- with-DTP DTP-after-MV MV ± OPV N ¼ 86 N ¼ 94 N ¼ 186 N ¼ 48 N ¼ 314 N ¼ 229 N ¼ 439 No DTP N ¼ 41

Birth WAZ (SD) [N]a −0.02 (0.95) [25] −0.50 (1.21) [40] −0.22 (1.06) [74] −0.41 (1.35) [22] −0.23 (0.96) [146] −0.28 (1.03) [93] −0.18 (0.89) [172] −1.03 (1.22) [14] Mean WAZ (SD) [N] −1.44 (1.50) [45] −1.20 (1.31) [62] −1.17 (1.15) [132] −1.33 (1.30) [36] −0.83 (1.06) [270] −1.16 (1.17) [179] −1.04 (1.10) [337] −0.94 (1.48) [30] at 12e23 months of age Mean WAZ (SD) [N] −0.66 (0.59) [4] −0.94 (1.67) [17] −1.29 (1.22) [38] −1.07 (0.81) [11] −1.03 (0.99) [212] −1.17 (1.01) [124] −1.10 (0.96) [330] −1.12 (0.59) [7] at 24e35 months of age Male sex 59.3% 52.1% 49.5% 41.7% 50.3% 49.8% 51.3% 43.9% Twin 4.7% 4.3% 3.2% 4.2% 2.9% 3.1% 2.1% 4.9% Ethnic group Pepel 50.0% 51.1% 52.2% 60.4% 50.6% 53.3% 54.4% 61.0% Balanta 15.1% 13.8% 15.1% 12.5% 15.9% 11.8% 13.2% 4.9% Other 34.9% 35.1% 32.8% 27.1% 33.4% 34.9% 32.3% 34.1% Examination rateb 0.34 [46/136.2] 0.93 [57/61.4] 0.60 [120/199.1] 0.57 [26/45.7] 1.77 [201/113.5] 1.59 [172/107.9] 2.26 [237/104.7] 1.21 [29/24.0] [event/PYRS]

Note: N denotes the number of children having been in the group; only one landmark outside these definitions due to most recent vaccine being OPV with no DTP but having received MV (did not die); only using one observation per child but allowing a child to be included in all groups once. a Birth WAZ defined as the WAZ measured at or before 14 days of age. b Examination rate calculated as the number of all observations starting the observation with the child being present divided by PYRS in the specified age period. oue4 ubr1 Number 43 Volume aur 2021 January

Supplementary Table 3. Mortality rates and hazard ratios (HR) by most recent vaccination, sex and age; MV has been censored.

Girls Mortality rate [Deaths/Pyrs] HR (CI)

Age groups Unvaccinated DTP + OPV DTP-Only OPV-only DTP ± OPV DTP ± OPV vs. DTP-only vs. DTP ± OPV vs. DTP-only vs. Unvaccinated Unvaccinated OPV-only OPV-only

3e5 months 1.7 [1/58.4] 11.6 [3/25.8] 27.2 [2/7.4] 0 [0/4.1] 15.1 [5/33.1] 11.07 (1.23e99.31) 20.05 (1.74e230.62) P ¼ 0.43a P ¼ 0.33a 6e11 months 10.2 [6/58.7] 13.1 [14/107.2] 11.8 [3/25.5] 0 [0/19.6] 12.8 [17/132.7] 1.17 (0.46e2.98) 1.07 (0.27e4.29) P ¼ 0.10a P ¼ 0.15a 12e23 months 4.3 [2/46.3] 7.6 [7/92.0] 0 [0/23.6] 4.2 [1/24.0] 6.1 [7/115.6] 1.37 (0.29e6.61) P ¼ 0.30a 1.43 (0.18e11.63) P ¼ 0.29a 24e35 months 0 [0/15.5] 7.6 [3/39.7] 7.5 [1/13.3] 7.6 [1/13.2] 7.5 [4/53.0] P ¼ 0.28a P ¼ 0.24a 0.97 (0.11e8.64) 0.98 (0.06e15.61) 3e35 months 5.0 [9/179.0] 10.2 [27/264.7] 8.6 [6/69.8] 3.3 [2/60.9] 9.9 [33/334.4] 2.13 (1.00e4.54) 1.90 (0.67e5.43) 2.81 (0.67e11.71) 2.51 (0.51e12.45)

Boys Mortality rate [Deaths/Pyrs] HR (CI)

Age groups Unvaccinated DTP + OPV DTP-Only OPV-only DTP ± OPV DTP ± OPV vs. DTP-only vs. DTP ± OPV vs. DTP-only vs. Unvaccinated Unvaccinated OPV-only OPV-only

3e5 months 7.5 [5/66.9] 15.7 [4/25.5] 46.3 [4/8.6] 0 [0/4.1] 23.4 [8/34.2] 3.79 (1.16e12.39) 7.31 (1.86e28.76) P ¼ 0.33a P ¼ 0.19a 6e11 months 8.3 [6/72.5] 10.6 [11/103.8] 16.2 [5/30.8] 4.6 [1/21.6] 11.2 [16/134.6] 1.36 (0.53e3.48) 1.86 (0.57e6.09) 2.66 (0.35e20.09) 3.64 (0.42e31.15) 12e23 months 8.2 [5/60.9] 6.5 [6/91.6] 7.8 [3/38.4] 0 [0/21.3] 6.9 [9/130.0] 0.84 (0.28e2.51) 0.94 (0.22e3.93) P ¼ 0.22a P ¼ 0.19a 24e35 months 4.2 [1/23.7] 4.2 [2/47.2] 6.4 [1/15.7] 11.3 [1/8.9] 4.8 [3/62.9] 1.15 (0.12e11.03) 1.55 (0.10e24.77) 0.41 (0.04e3.93) 0.55 (0.03e8.81) 3e35 months 7.6 [17/224.0] 8.6 [23/268.1] 13.9 [13/93.6] 3.6 [2/55.9] 10.0 [36/361.7] 1.43 (0.78e2.59) 2.03 (0.97e4.26) 2.80 (0.67e11.64) 4.00 (0.90e17.73)c

Note: Mortality rate presented as (event/Pyrs)*100; a: log-rank test for equality; b: children with no registered sex included; c: HR fails the Schoenfeld residual test of the proportional hazards assumption (p < 0.05). C.B. Ø ade al. et land 184.e5 Clinical Therapeutics

Supplementary Table 4. Mortality rates and hazard ratios (HR) by disjoint vaccination groups exploring timing of vaccination and most recent vaccination, and sex, children aged 9e36 months of age.

Groups Sex Mortality rate HR (CI) HR (CI) [deaths/Pyrs] [With MV-after-DTP [With unvaccinated as reference] as reference]

MV-after-DTP (±OPV) Girl 1.3 [3/234.0] Ref. 0.39 (0.09e1.78) Boy 3.3 [7/214.0] Ref. 0.65 (0.23e1.84) DTP-with-MV (±OPV) Girl 4.5 [7/156.3] 3.18 (0.82e12.30) 1.25 (0.36e4.31) Boy 5.0 [7/139.3] 1.44 (0.50e4.10) 0.94 (0.34e2.62) DTP-after-MV (±OPV) Girl 2.9 [8/271.9] 2.53 (0.67e9.55)a 1.00 (0.29e3.40) Boy 2.8 [8/282.2] 0.97 (0.35e2.69) 0.64 (0.23e1.75) DTP ± OPV (no MV) Girl 9.1 [24/262.8] 5.13 (1.52e17.28) 2.02 (0.70e5.82) Boy 7.8 [22/280.4] 1.76 (0.74e4.20) 1.15 (0.51e2.59) Most recent vaccine containing Girl 5.2 [2/38.4] 3.07 (0.51e18.50) 1.21 (0.22e6.60) only Live vaccines, MV Boy 2.9 [1/34.3] 0.69 (0.08e5.65) 0.45 (0.06e3.62) and/or OPV (no DTP) Unvaccinated (No DTP, MV or OPV) Girl 4.7 [4/85.0] 2.54 (0.56e11.51) Ref. Boy 7.0 [8/114.3] 1.52 (0.54e4.29) Ref.

Note: Mortality rate presented as (event/Pyrs)*100; There were no registered sex for one landmark/sub-child (did not die). a HR fails the Schoenfeld test for the proportional hazards assumption.

184.e6 Volume 43 Number 1 C.B. Øland et al.

Supplementary Table 5. Mortality rates and hazard ratios (HR) by disjoint vaccination groups exploring timing of vaccination and most recent vaccination groups, and OPV, children aged 9e36 months of age.

Most recent vaccination(s) Mortality rate [Deaths/Pyrs] HR (CI)

1 MV-after-DTP and OPV 2.0 [9/445.6] Ref. 2 MV-after-DTP-only 41.9 [1/2.4] 18.72 (2.36e148.30) 3 DTP-with-MV and OPV 3.6 [10/278.8] 1.65 (0.67e4.07) 4 DTP-with-MV, no OPV 23.8 [4/16.8] 10.18 (3.12e33.20) 5 DTP-after-MV and OPV 2.6 [14/541.9] 1.41 (0.61e3.28) 6 DTP-after-MV, no OPV 16.4 [2/12.2] 8.22 (1.77e38.11) 7 DTP-only 7.0 [4/57.6] 2.48 (0.75e8.19) 8 DTP-only, OPV previously, no MV 3.3 [2/59.9] 1.30 (0.28e6.05) 9 DTP and OPV, no MV 9.8 [36/367.1] 3.52 (1.66e7.49) 10 OPV-only, DTP previously, no MV 6.8 [4/58.6] 2.65 (0.81e8.68) 11 OPV-only 0 [0/29.4] P ¼ 0.28a 12 MV and OPV, no DTP 8.0 [2/25.1] 3.06 (0.65e14.28) 13 MV, no DTP and OPV 5.4 [1/18.7] 2.44 (0.31e19.31) 14 Unvaccinated (No DTP, MV and OPV) 6.0 [12/199.3] 2.13 (0.88e5.17)

Note: Mortality rate presented as (event/Pyrs)*100; one landmark not included since the landmark had last vaccine VP but had MV and BCG and no DTP (did not die); using only the first dose of MV. a Log-rank test for equality.

Supplementary Table 6. Mortality rates and hazard ratios (HR) for out-of-sequence Vaccinations with MV and DTP, children aged 9e36 months of age

Groups from 9 to 36 months of age; Group Mortality rate HR (CI) most recent vaccination (deaths/100 person-years)

DTP-with-or-after-MV and OPV All 2.9 (24/820.8) Ref Female 2.6 (11/417.9) Ref Male 3.2 (13/402.9) Ref DTP-with-or-after-MV, no OPV All 20.7 (6/29.0) 6.25 (2.55e15.37) Female 38.6 (4/10.4) 13.31 (4.21e42.03) Male 10.7 (2/18.6) 2.92 (0.66e12.96)

January 2021 184.e7