Campaign participation in 2006 was high. Among the 5600 children (69%) present at the first visit after the campaign, 90% were said to have participated (5015/5600).
In the primary analysis with follow-up for 12 months, 579 children died during the year after index dates, 164 after the campaign and 208 and 207 during the same period the previous years. Censoring 9 deaths due to accidents retained 161 deaths after the campaign (mortality rate = 23.8 per 1000 person-years) and 409 in the 2 precampaign years (mortality rate = 30.7 per 1000 person-years; Fig. 1). The mortality rate ratio (MRR) comparing the year after the campaign with the 2 previous years was 0.78 (0.65–0.93) (Table 2). Adjusting for maternal education and maternal age had little effect on the estimate, the adjusted MRR (aMRR) being 0.80 (0.66–0.96). The reduction was statistically significant in girls [aMRR = 0.74 (0.56–0.97)] but not in boys [aMRR = 0.86 (0.66–1.11)]; the interaction between sex and campaign was not statistically significant (P = 0.44). In children aged 6–11 months, the aMRR after the campaign was 0.92 (0.63–1.34); in children aged 1–4 years of age, it was 0.76 (0.61–0.94), the effect being similar in children aged 1–2 years [aMRR = 0.77 (0.60–1.00)] and children aged 3–4 years [aMRR = 0.75 (0.49–1.13)].
During the first 6 months of follow-up, the aMRR was 0.79 (0.63–0.98) for all eligible children; from 6–12 months, aMMR was 0.86 (0.60–1.24) (data not shown). Changing the timescale to time since the index date yielded essentially the same estimates when mortality rates were compared in a Cox proportional hazards model: aMRR = 0.79 (0.66–0.95); 0.84 (0.65–1.08) in boys and 0.74 (0.57–0.97) in girls (Fig. 2).
Cause of Death
Guinea-Bissau experienced a measles epidemic in 2003 and 2004.23 On the basis of interviews conducted after the death of the child, 15 deaths in the 2004 cohort were classified as due to measles. No child was said to have died from measles in the 2005 and 2006 cohorts. Censoring follow-up time of the children who died from measles on the date of death, the aMRR comparing the 2006 cohort with the 2004 and 2005 cohorts was 0.83 (0.69–1.00). When limiting the analyses to the 2006 and 2005 cohorts in which no measles deaths were registered, the aMRR comparing the 2006 cohort with the 2005 cohort was 0.79 (0.64–0.98) (Table, Supplemental Digital Content 3, http://links.lww.com/INF/C268). Nonmeasles febrile illness was reported as the main symptom for 68% of the deaths (Table, Supplemental Digital Content 4, http://links.lww.com/INF/C269), the MMR being 0.82 (0.66–1.03). Diarrheal disease was the second most common cause, accounting for 10% of the deaths and was associated with an MRR of 0.71 (0.40–1.26).
Effect of MV Campaign by Routine MV and DTP-3 Status
Among children aged 1–4 years at the time of the index date, 5946 (84%), 6134 (85%) and 6276 (86%) had a visit within the follow-up period in the 2004, 2005 and 2006 cohorts, respectively. The vaccination cards were seen for 42% (2487/5946), 41% (2535/6134) and 44% (2775/6276) in the 3 cohorts, respectively; the coverage for routine MV was 91%, 91% and 84%, respectively. Among children whose vaccination card was seen, the aMRR was 0.59 (0.36–0.99) for children who had received both routine and MV campaign and 0.97 (0.38–2.52) for children who had received no routine MV and only MV campaign (P for interaction = 0.37; Table 3). The effect was particularly strong among children who had received DTP-3 and therefore were unlikely to receive DTP during follow-up, the MRR being 0.49 (0.28–0.87) in the unadjusted and 0.50 (0.28–0.88) in the adjusted model (Table, Supplemental Digital Content 5, http://links.lww.com/INF/C270). In routine measles-vaccinated children, the effect of the campaign tended to differ by DTP-3 vaccination status for girls [MRR = 0. 40 (0.17–0.91) among DTP-3-recipients, but MRR = 1.81 (0.40–8.18) in girls who had not received DTP-3 (P = 0.07 for same effect)] while there was no difference in boys (Table, Supplemental Digital Content 5, http://links.lww.com/INF/C270).
Deaths with only the month of death known were coded as having occurred on the 15th of the month. The proportion of deaths occurring on the 15th were 91% (190/208), 83% (172/207) and 83% (136/163) in the 3 cohorts. The estimated mortality reduction indicated no difference as to whether the 15th of the month was considered an exact date of death or whether deaths registered to have occurred on the 15th were treated as interval censored using the 1st and the 30th as the interval in which the death could have occurred. These 2 approaches both estimated MRRs as 0.78 (0.65–0.93). Including deaths classified as being due to accidents did not change the reduction in mortality, with the aMMR being 0.80 (0.67–0.96).
The under-5 mortality level has declined from approximately 300/1000 in 2000 to 150/1000 in 2006 (Fig., Supplemental Digital Content 6, http://links.lww.com/INF/C271). In none of the years was the mortality decline as large as between 2005 and 2006.
Number Needed to Vaccinate
Mortality during the year following the index dates in the precampaign years was 2.5% (2.3%–2.7%) and 1.9% (1.7%–2.2%) during the year after the campaign. The risk difference was 0.6 percentage point (0.2–0.9) and consequently the number needed to vaccinate to avoid 1 death was 179 (106–526).
We found a 20% (4%–34%) reduction in mortality the year following the 2006 MV campaign compared with the 2 previous years. The lower mortality was not explained by protection against measles infection. Consistent with many other studies, the beneficial effect of MV was most marked in girls and it was also strongest among children who had received routine MV previously, although none of the interactions was statistically significant.
Strengths and Weaknesses
In a before-and-after study, observed differences could be caused by other factors varying at the same time. However, the observed tendencies are consistent with the hypotheses regarding the beneficial nonspecific effects of MV: the drop in mortality between 2004 to 2005 and 2006 appeared larger than declines observed in previous years (Fig., Supplemental Digital Content 6, http://links.lww.com/INF/C271), and it tended to be stronger during the first 6 months of follow-up, excluding the period after November 2006 when bed nets were distributed. The drop in mortality also tended to be stronger for girls than for boys and for children who had previously received MV. By stratifying the analysis for village cluster, children in the same village were compared, and the distribution of risk factors for death therefore should be similar for children compared in the post- and precampaign years. Hence, we would expect little effect of adjusting for background factors, as observed for maternal education and age.
Guinea-Bissau has higher mortality in the rainy season (June to November) than in dry season29; we therefore compared the mortality during the months after the campaign, with mortality during the same period the previous years. The duration of the nonspecific effect of MV is unknown. We limited postcampaign follow-up to 1 year since vaccination strategies changed in 2007 when the BHP started offering vaccines at the 6 monthly routine visits30 and the national program discontinued the DTP booster dose at 18 months of age, as we expected both changes to lower mortality.
The analysis had to be made as an intention-to-treat analysis as it otherwise would have been impossible to know with whom to compare in the years before the campaign. We confirmed reception of MV campaign for at least 90% of the children. Coverage figures provided by the Measles and Rubella Initiative indicate 92% coverage. Hence, the before-after comparison is likely to resemble a comparison of children who received MV in a campaign and children who did not. Still an intention-to-treat analysis will be conservative because some children did not receive the intervention. A before-after study rather than a comparison of participants and nonparticipants was chosen because the small group of nonparticipant is likely to differ from the participants and because the follow-up could only start after classifying participation status at a visit after the campaign.
In addition to MV, the children received VAS and mebendazole during the 2006 campaign and the effects of the 3 interventions are difficult to disentangle. VAS had also been distributed in campaigns in the previous years although only once per year, whereas the post-MV campaign cohort received VAS twice (Table, Supplemental Digital Content 1, http://links.lww.com/INF/C266). In a randomized trial of VAS administered with vaccines, we recently found that VAS compared with placebo had no overall effect on child survival but increased the mortality in boys and reduced it in girls.31 Similarly, a recent trial from India of one million children found no overall effect of biannual distribution of VAS.32 Hence, it seems unlikely that VAS should be the major driver in the mortality decline after the 2006 campaign although it may amplify a beneficial effect in girls and counteract it in boys.31 It also seems unlikely that deworming would have such a strong effect on child survival.33,34 In the years before the MV campaign (but not during the year after the campaign), children aged 0–59 months received OPV that may also have beneficial nonspecific effects35,36 and would therefore tend to mask a benefit of MV campaign. No single cause of death seemed to explain the lower mortality after the campaign (Table, Supplemental Digital Content 4, http://links.lww.com/INF/C269).
Consistency With Previous Studies
Despite the huge investment in delivering MV in campaigns, the effect on survival has not been assessed previously. From a disease-specific perspective, measles revaccination campaigns should have a limited effect on child mortality because most children are already protected against measles infection. Several studies have compared mortality of measles-vaccinated and measles-unvaccinated children. In 10 cohort studies from the 1970s to 1980s, the estimated MMRs ranged from 0.14 to 0.70.12 In a previous study from the same village clusters, children were followed between 1990 and 1996; measles-vaccinated children aged 6–17 months had an MRR of 0.51 (0.28–0.95) during 6 months of follow-up (excluding measles deaths).13 In the late 1990s, children in Guinea-Bissau were randomized to an early MV or a control group receiving inactivated polio vaccine at 6 months. Both groups should receive an MV at 9 months. When the war broke out in 1998, a group of children had not received their 9 months’ vaccination. Among these, the MRR for measles-vaccinated children compared with inactivated polio vaccine-vaccinated children was 0.30 (0.08–0.87).18 In these studies, the mortality in the unvaccinated group was considerably higher than that in the present study. With the decline in mortality since the 1990s, presumably because of fewer infectious disease deaths, we would expect a lower nonspecific effect on mortality.
We found a stronger beneficial effect for girls than for boys. As shown in Table (Supplemental Digital Content 2, http://links.lww.com/INF/C267), no sex-differential distribution of background factors could explain this difference. Stronger beneficial nonspecific effects of MV for girls have been observed consistently in both randomized trials and observational studies in Guinea-Bissau, which does not have sex-differential treatment,26,30 as well as in settings with sex-differential treatment.37
Interestingly, when assessing survival after inspection of a vaccination card after the index dates, the effect of MV in campaign on all-cause mortality was only seen in children who had already received routine MV and hence were protected against measles infection already. In a quasi-experimental study from the early 1980s when the first MV campaigns were conducted and before DTP was introduced in the rural areas in Guinea-Bissau, children who happened to be younger than 9 months at the time of the first MV campaign were also offered MV after 9 months of age. Children who received 2 doses of MV had significantly lower mortality between 9 and 59 months of age than children who received only 1 dose of MV between 9 and 11 months of age, the reduction in mortality being 59% (15%–81%).20 More recently, in a randomized trial of early measles vaccination, children who received MV at 4.5 and 9 months had 29% (−1% to 50%) lower mortality between 9 and 36 months of age than children who received 1 dose of MV at 9 months of age.17 Hence, the beneficial nonspecific effects of MV seem to be amplified by a booster MV as also supported by the present study, and the 20% reduction observed in the present study is comparable with the result from a randomized trial.
Interpretation and Implications
We observed 20% lower mortality during 12 months of follow-up. Like in previous studies,12,13,17 excluding measles deaths did not alter the estimated beneficial effect of MV. Hence, the reduction in child mortality cannot be explained by the prevention of measles infection. Six monthly visits took place throughout the study period, and we know of no other disease outbreaks during 2004 to 2006, which could explain the mortality pattern. The effect tended to be strongest for those who had received both routine and MV campaign, corroborating that beneficial nonspecific effects of MV were the main drivers of the mortality reduction. Noteworthy, if a booster MV is associated with strong additional survival benefits, the program for a second dose of MV in the second year of life, which is currently being rolled out with funding from GAVI, the Vaccine Alliance,38 may have a much larger effect on child survival than assumed. Future cluster randomized studies of MV campaigns should seek to measure the effect of MV campaign on all-cause mortality and whether the effect differs by number of prior MVs.
Few studies on the biological mechanisms underlying nonspecific effects of MV have been undertaken,39,40 and the mechanisms are largely unknown. Recent studies have shown that Calmette-Guérin bacillus vaccination induces epigenetic changes that affect the responses to unrelated pathogens.41 MV also may induce such changes. Within a randomized trial, we recently observed that children randomized to early MV versus no MV at 4–6 months of age had higher levels of monocyte chemoattractant protein-1, 6 weeks postrandomization.39 In the present study, the second dose of MV seemed to carry additional beneficial nonspecific effects. It has been shown previously that providing MV in the presence of maternal measles-specific antibodies results in the lower attained antibody levels but at the same time in stronger beneficial nonspecific effects.42 We have speculated that the high-affinity maternal measles-specific antibodies may (1) bind to the most dominant epitopes and thereby lead to a broader response to other epitopes on the MV virus, resulting in protective cross-reactivity memory responses or (2) result in rapid formation of antibody-antigen complexes, which may lead to enhanced T-cell responses.42 Such mechanisms also could explain the beneficial effect of the second dose of MV observed in the present study because the first MV induces high-affinity measles-specific antibodies, and the second vaccination occurs in the presence of these antibodies. If vaccination in the presence of measles antibodies is responsible for the beneficial effect, children would benefit from receiving the first vaccine early rather than later as is recommended when measles infection is better controlled.2 The second dose is provided to improve immunity against measles,2 but the benefit may even be stronger in children who have seroconverted already.
The number needed to vaccinate to prevent one death during the first year after vaccination was only 179 (106–526), and thus, the cheap MV has the potential to prevent many nonmeasles-related deaths. The effect was evident in our study in which three-quarters of the children had already received routine MV before the campaign and particularly strong among these children. Hence, cutting routine MV immunization services, the number of MV doses or MV campaigns in connection with approaching measles elimination and eventual eradication could have negative effects on child survival.
Mortality levels were stable during 2004 and 2005, but a mortality drop was observed after the 2006 measles campaign. Corroborating findings from previous observational studies and randomized trials, the beneficial effect of MV was significant in its own right for girls. Furthermore, contrary to assumptions about specific effects, the beneficial effect of MV campaign was also significant in its own right among children who had received routine MV but not among measles-unvaccinated children. Both findings support that the mortality decline was due to nonspecific effects of MV rather than protection against measles infection or uncontrolled confounding.
Author contributions: A.B.F., H.R., C.S.B. and P.A. designed the study; A.B.F., C.M., S.B., S.T., L.S., M.P., M.F. and P.A. supervised data collection, data entry and/or data cleaning; A.B.F., P.A. and H.R. analyzed the data and A.B.F. had full access to the data, wrote the first manuscript draft and had primary responsibility for its final content. All authors contributed to and approved the final manuscript.
2. . Measles vaccines: WHO position paper. Wkly Epidemiol Rec. 2009;84:349–360
3. Zuber PL, Conombo KS, Traoré AD, et al. Mass measles vaccination
in urban Burkina Faso, 1998. Bull World Health Organ. 2001;79:296–300
4. Chuang SK, Lau YL, Lim WL, et al. Mass measles immunization campaign: experience in the Hong Kong Special Administrative Region of China. Bull World Health Organ. 2002;80:585–591
5. Goodson JL, Kulkarni MA, Vanden Eng JL, et al. Improved equity in measles vaccination
from integrating insecticide-treated bednets in a vaccination campaign, Madagascar. Trop Med Int Health. 2012;17:430–437
6. Hu X, Xiao S, Chen B, et al. Gaps in the 2010 measles SIA coverage among migrant children in Beijing: evidence from a parental survey. Vaccine. 2012;30:5721–5725
7. Munyoro MN, Kufa E, Biellik R, Pazvakavambwa IE, Cairns KL. Impact of nationwide measles vaccination
campaign among children aged 9 months to 14 years, Zimbabwe, 1998–2001. J Infect Dis. 2003;187(suppl 1):S91–S96
8. Lowther SA, Curriero FC, Kalish BT, et al. Population immunity to measles virus and the effect of HIV-1 infection after a mass measles vaccination
campaign in Lusaka, Zambia: a cross-sectional survey. Lancet. 2009;373:1025–1032
9. Amouzou A, Habi O, Bensaïd KNiger Countdown Case Study Working Group. . Reduction in child mortality
in Niger: a Countdown to 2015 country case study. Lancet. 2012;380:1169–1178
10. Aaby P, Bukh J, Lisse IM, et al. Measles vaccination
and reduction in child mortality
: a community study from Guinea-Bissau. J Infect. 1984;8:13–21
11. Aaby P, Simondon F, Samb B, et al. Low mortality after mild measles infection compared to uninfected children in rural West Africa. Vaccine. 2002;21:120–126
12. Aaby P, Samb B, Simondon F, et al. Non-specific beneficial effect of measles immunisation: analysis of mortality studies from developing countries. BMJ. 1995;311:481–485
13. Kristensen I, Aaby P, Jensen H. Routine vaccinations and child survival: follow up study in Guinea-Bissau, West Africa. BMJ. 2000;321:1435–1438
14. Desgrees du Lou A, Pison G, Aaby P. Role of immunizations in the recent decline in childhood mortality and the changes in the female/male mortality ratio in rural Senegal. Am J Epidemiol. 1995;142:643–652
15. . Influence of measles vaccination
on survival pattern of 7-35-month-old children in Kasongo, Zaire. The Kasongo Project Team. Lancet. 1981;1:764–767
16. Koenig MA, Khan MA, Wojtyniak B, et al. Impact of measles vaccination
on childhood mortality in rural Bangladesh. Bull World Health Organ. 1990;68:441–447
17. Aaby P, Martins CL, Garly ML, et al. Non-specific effects of standard measles vaccine at 4.5 and 9 months of age on childhood mortality: randomised controlled trial. BMJ. 2010;341:c6495
18. Aaby P, Garly ML, Balé C, et al. Survival of previously measles-vaccinated and measles-unvaccinated children in an emergency situation: an unplanned study. Pediatr Infect Dis J. 2003;22:798–805
19. 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-pertussis and inactivated poliovirus: reanalysis of West African studies. Lancet. 2003;361:2183–2188
20. Aaby P, Andersen M, Sodemann M, et al. Reduced childhood mortality after standard measles vaccination
at 4-8 months compared with 9-11 months of age. BMJ. 1993;307:1308–1311
22. Høj L, da Silva D, Hedegaard K, et al. Factors associated with maternal mortality in rural Guinea-Bissau. A longitudinal population-based study. BJOG. 2002;109:792–799
23. Balé C, Garly ML, Martins C, et al. Risk factors for measles in young infants in an urban African area with high measles vaccination
coverage. Pediatr Infect Dis J. 2011;30:689–693
24. Aaby P, Jensen H, Gomes J, et al. The introduction of diphtheria-tetanus-pertussis vaccine and child mortality
in rural Guinea-Bissau: an observational study. Int J Epidemiol. 2004;33:374–380
25. Hornshoj L, Benn CS, Fernandes M, et al. Vaccination coverage and out-of-sequence vaccinations in rural Guinea-Bissau: an observational cohort study. BMJ open. 2012;2:e001509
26. Martins CL, Benn CS, Andersen A, et al. A randomized trial of a standard dose of Edmonston-Zagreb measles vaccine given at 4.5 months of age: effect on total hospital admissions. J Infect Dis. 2014;209:1731–1738
27. Veirum JE, Sodemann M, Biai S, et al. Routine vaccinations associated with divergent effects on female and male mortality at the paediatric ward in Bissau, Guinea-Bissau. Vaccine. 2005;23:1197–1204
28. Austin PC. Absolute risk reductions and numbers needed to treat can be obtained from adjusted survival models for time-to-event outcomes. J Clin Epidemiol. 2010;63:46–55
29. Veirum JE, Biai S, Jakobsen M, et al. Persisting high hospital and community childhood mortality in an urban setting in Guinea-Bissau. Acta Paediatr. 2007;96:1526–1530
30. Fisker AB, Hornshøj L, Rodrigues A, et al. Effects of the introduction of new vaccines in Guinea-Bissau on vaccine coverage, vaccine timeliness, and child survival: an observational study. Lancet Glob Health. 2014;2:e478–e487
31. Fisker AB, Bale C, Rodrigues A, et al. High-dose vitamin A with vaccination after 6 months of age: a randomized trial. Pediatrics. 2014;134:e739–e748
32. Awasthi S, Peto R, Read S, et al. Vitamin A supplementation every 6 months with retinol in 1 million pre-school children in north India: DEVTA, a cluster-randomised trial. Lancet. 2013;381:1469–1477
33. Awasthi S, Peto R, Read S, et al.DEVTA (Deworming and Enhanced Vitamin A) Team. Population deworming every 6 months with albendazole in 1 million pre-school children in North India: DEVTA, a cluster-randomised trial. Lancet. 2013;381:1478–1486
34. Donnen P, Brasseur D, Dramaix M, et al. Vitamin A supplementation but not deworming improves growth of malnourished preschool children in eastern Zaire. J Nutr. 1998;128:1320–1327
35. Aaby P, Hedegaard K, Sodemann M, et al. Childhood mortality after oral polio immunisation campaign in Guinea-Bissau. Vaccine. 2005;23:1746–1751
36. Aaby P, Rodrigues A, Biai S, et al. Oral polio vaccination and low case fatality at the paediatric ward in Bissau, Guinea-Bissau. Vaccine. 2004;22:3014–3017
37. Hirve S, Bavdekar A, Juvekar S, et al. Non-specific and sex-differential effects of vaccinations on child survival in rural western India. Vaccine. 2012;30:7300–7308
39. Jensen KJ, Søndergaard M, Andersen A, et al. A randomized trial of an early measles vaccine at 4½ months of age in Guinea-Bissau: sex-differential immunological effects. PLoS One. 2014;9:e97536
40. Bertley FM, Ibrahim SA, Libman M, et al. Measles vaccination
in the presence of maternal antibodies primes for a balanced humoral and cellular response to revaccination. Vaccine. 2004;23:444–449
41. Kleinnijenhuis J, Quintin J, Preijers F, et al. Bacille Calmette-Guerin induces NOD2-dependent nonspecific protection from reinfection via epigenetic reprogramming of monocytes. Proc Natl Acad Sci U S A. 2012;109:17537–17542
42. Aaby P, Martins CL, Garly ML, et al. Measles vaccination
in the presence or absence of maternal measles antibody: impact on child survival. Clin Infect Dis. 2014;59:484–492
child mortality; measles vaccination; vaccination campaigns; nonspecific/heterologous effects of vaccines
Supplemental Digital Content
Copyright © 2015 Wolters Kluwer Health, Inc. All rights reserved.