Determinants of delayed or incomplete diphtheria-tetanus-pertussis vaccination in parallel urban and rural birth cohorts of 30,956 infants in Tanzania

Routine childhood immunizations prevents an estimated 2 to 3 million deaths each year due to diphtheria, tetanus, pertussis, and measles [1]. Nevertheless, the maximal potential benefits of vaccines have not been realized in low- and middle-income countries (LMICs) due to incomplete coverage and late vaccination. The World Health Organization estimated that in 2015, over 19 million infants did not receive routine immunizations with the overwhelming majority coming from LMICs [1]. While suboptimal vaccine coverage in LMICs has been reasonably well documented, delayed vaccination has received significantly less attention. Delayed vaccination prolongs the at risk period and leaves infants vulnerable during early infancy, when the risk of severe morbidity or death is the greatest for most vaccine-preventable diseases [2].

Incomplete or delayed vaccination in LMICs may result from numerous factors including poor immunization supply chains, suboptimal access to health services, or social and cultural factors [3]. Several studies suggest that higher birth order, low maternal education, and low socioeconomic status are associated with poor adherence to vaccination schedules [3]. Studies in the Netherlands [4], Ghana [5] and Guinea-Bissau [6], have found that low birth weight was associated with delayed vaccination. In rural Ghana, greater distance to the health facility was also associated with delayed vaccination [5]. Nevertheless, no studies to the best of our knowledge have examined how the magnitude of the predictors of delayed vaccination vary by setting within the same country.

In this study, we analyzed parallel prospective cohorts of newborns in urban Dar es Salaam and rural Morogoro Region, Tanzania to assess potential differences in the contribution of social, economic, and biological factors to delayed or not received Diphtheria-Tetanus-Pertussis (DTP) vaccination. Through these comparisons, we intend to inform interventions to improve complete and timely vaccination in urban and rural immunization programs.

In the parent NEOVITA trial, 32,483 (96.3%) newborns were screened for study eligibility and 31,999 (97.43%) were enrolled. The analysis of DTP1 and DTP3 included the 30,956 infants and 30,503 infants, respectively, who did not die and were not lost to follow-up before the start of the vaccination window. A participant flowchart is included in Additional file 1: Figure S1. Table 1 presents baseline characteristics of the urban Dar es Salaam and rural Morogoro study cohorts. The Dar es Salaam cohort had a lower proportion of low birth weight infants (8.7%) as compared to rural Morogoro (13.9%). In Dar es Salam 9.3% of infants were born SGA, while 13.6% were SGA in Morogoro. Infants from Morogoro were also of higher parity, with 17.7% being the 5th or higher child, as compared to 8.1% in Dar es Salaam. Mothers in Dar es Salaam were slightly more likely to attend their first antenatal care visit in the first trimester (13.3%) than mothers in Morogoro (9.6%). In addition, 13.4% of births occurred in the home in rural Morogoro, while there was a negligible amount (< 1%) of home births in the Dar es Salaam cohort. Mothers and fathers in Dar es Salaam were more likely to have completed school beyond the primary level and less likely to have had no formal schooling than mothers and fathers in Morogoro.

In Dar es Salaam, of 11,189 infants eligible for DTP1 analysis, 9906 (88.5%) were vaccinated on time, 336 (3.0%) were vaccinated late (> 90 days), and 947 (8.5%) were not observed to be vaccinated during follow-up to 1 year of age. Of 10,932 infants eligible for DTP3 analysis in Dar, 9181 (84.0%) were vaccinated on time, 200 (1.8%) were vaccinated late (> 7 months), and 1551 (14.2%) were not observed to be vaccinated to 1 year of age. In Morogoro Region, of 19,767 infants eligible for DTP1 analysis, 17,959 (90.9%) were vaccinated on time, 370 (1.9%) were vaccinated late, and 1439 (7.3%) were never vaccinated. Of 19,571 infants eligible for DTP3 analysis in Morogoro Region, 16,190 (82.7%) were vaccinated on time, 516 (2.6%) were vaccinated late, and 2865 (14.6%) were never vaccinated. The distribution of infant age (days) at vaccination with DTP1 and DTP3 by site are presented in Fig. 1. The bimodal shape of the DTP1 vaccination age in Fig. 1a is due to an increase in the Tanzanian recommended age at the first DTP dose from 4 weeks to 6 weeks with the introduction of the pentavalent vaccine. The shape of the DTP3 age histogram is monomodal since the recommended age at DTP3 vaccination was 14 weeks both before and after introduction of the pentavalent vaccine.

Risk factors associated with delayed or incomplete DTP1 and DTP3 vaccination are presented in Tables 2 and 3. We determined that there was no difference in risk by infant sex of delayed or no receipt of DTP1 or DTP3 vaccination in both urban Dar es Salaam and rural Morogoro cohorts. The association between low birth weight and delayed or incomplete DTP1 vaccination varied by study cohort (p-value for interaction 0.009). In Morogoro region, low birthweight infants had 31% (95% CI: 20–40%) and 29% (95% CI: 21–36%) reduced risk of delayed or not received DTP1 and DTP3 vaccination, respectively. There was no significant association of low birth weight with DTP vaccination in Dar es Salaam. There was also no relationship between prematurity or small-for-gestational-age births and vaccination timing in either setting.

Infants in the highest wealth quintile had a reduced risk of delayed or incomplete vaccination with DTP1 and DTP3 in both Dar es Salaam and Morogoro. This association between wealth quintile and delayed or incomplete vaccination was stronger in Morogoro than in Dar es Salaam for both DTP1 and DTP3 (p-values for interaction 0.02, < 0.001). The magnitude of the association between paternal education and delayed or incomplete vaccination was stronger in Morogoro for both DTP1 and DTP3 (p-value for interaction 0.04, 0.004). In rural Morogoro, infants of fathers with no formal education had a 59% (95% CI: 36, 87%) and 36% (95% CI: 22, 52%) increased risk of delayed or not received DTP1 and DTP3 vaccination, respectively as compared to those whose fathers completed primary school. In contrast, there was no indication that infants of non-educated fathers had an increased risk of delayed or not received vaccination in Dar es Salaam. Greater maternal education was significantly associated with reduced risk of delayed or incomplete vaccination with DTP3 in Morogoro (p-value < 0.001), but there was no significant association of maternal education with DTP3 vaccination in Dar es Salaam.

In terms of indicators of access to care, mothers in Dar es Salaam who attended their first ANC visit later in pregnancy had an increased risk of delayed or no receipt of both DTP1 (p-value < 0.001) and DTP3 (p-value < 0.001) vaccination. There was no relationship of ANC timing and vaccination status in rural Morogoro. Nevertheless, home births in the Morogoro Region had 1.20 (95% CI: 1.06, 1.35) and 1.17 (95% CI: 1.08, 1.27) times the risk of delayed or incomplete DTP1 and DTP3 vaccination as compared to facility births, respectively.

As expected, the percentage of infants with delayed vaccination increased when the earlier cut-offs of > 72 days for DTP1 and > 128 days for DTP3 were used in sensitivity analyses (Additional file 1: Tables S1 and S2). For DTP1, 15.3 and 11.8% of infants had delayed or incomplete vaccination in Dar es Salaam and Morogoro, respectively. For DTP3 vaccination, these percentages were 31.5 and 33.6%, respectively. In terms of risk factors, we found similar effect sizes in the sensitivity analysis (Additional file 1: Tables S1 and S2) and that the magnitude of the associations for wealth, paternal education, low birthweight, and ANC timing differed between Dar es Salaam and Morogoro Region.

In this prospective birth cohort study of 30,956 infants from urban Dar es Salaam and rural Morogoro Region, Tanzania, we determined that the proportion of infants with incomplete or late vaccination was similar between the contexts at about 10% for DTP1 and 15% for DTP3. We found some shared determinants of delayed or incomplete vaccination between settings; however, several factors differed in magnitude between urban and rural settings. In urban Dar es Salaam, infants born to young mothers, infants from households in the lowest wealth quintile, and mothers whose first ANC visit was in the third trimester were more likely to be vaccinated late or never. In rural Morogoro region, infants born normal birth weight, at home, to uneducated fathers, to poorer households, and to young mothers had an elevated risk of delayed or incomplete DTP vaccination.

Infants born to wealthier households were less likely to experience delayed or incomplete vaccination in both urban and rural settings in our study. The association between greater household wealth and reduced risk of delayed or incomplete vaccination has been previously noted [3]. Access to greater financial resources may address the costs associated with vaccination, including transport to the healthcare facility and parents foregoing daily wage to accompany their child to health clinics. Furthermore, individuals in the lowest wealth quintiles are more likely to live further from healthcare facilities [17], which is a commonly reported driver of poor vaccination uptake [18]. We did not have access to data on the distance to the nearest health center in our study. Financial and non-financial incentives to improve vaccination coverage have shown mixed effectiveness through extensive trials in LMICs [19, 20]. As a result, further implementation studies and trials are needed to determine effective interventions to overcome health care access barriers to vaccination among urban and rural poor families.

Younger maternal age was associated with delayed or incomplete vaccination with DTP3 in both Dar es Salaam and Morogoro region. However, lower maternal education was associated with an increased risk of delayed or incomplete vaccination with DTP3 in rural Morogoro, but there was no evidence of an association in Dar es Salaam. Strong relationships between maternal education, maternal age, and vaccination status have been observed in both high-income and LMIC settings, including Tanzania [3, 21]. We also found that lower paternal education, independent of maternal education, was associated with an increased risk of delayed or no receipt of DTP vaccination in both Dar es Salaam and Morogoro; however, the magnitude of the association was significantly stronger in rural Morogoro. Research has shown that lack of support by men can inhibit access of women to family planning and healthcare, particularly when men hold greater economic power within the households [22, 23]. Our data suggest that paternal sensitization and involvement interventions may be effective in improving vaccine uptake in rural Tanzania and similar settings, but these interventions may not as effective in improving vaccination timeliness in urban centers.

Low birthweight was associated with reduced risk of delayed or not received DTP1 and DTP3 vaccination in rural Morogoro Region, while there was no relationship in urban Dar es Salaam. This result is the opposite of findings from both high-income countries, like the Netherlands [4], and low-income countries, like Ghana [5], Guinea-Bissau [6], and Kenya [24], where low birthweight was associated with increased risk of delayed or incomplete vaccination. Preterm birth has also been shown to be associated with delayed vaccination in high-income settings [18]. In some LMIC settings, caregivers may be reluctant to bring ill or undernourished children to a clinic for vaccination due to social stigma and providers may delay vaccination of ill children [18]. Nevertheless, in rural Tanzania we found low birth weight infants were significantly less likely to have delayed or incomplete vaccinations. While some LMICs recommend waiting until infants have reached desired weights before administering certain vaccines [24], Tanzania’s Expanded Program for Immunization (EPI) does not do so [9]. Targeting low birth weight or preterm infants with additional counselling on vaccination does not appear to be a promising intervention to improve vaccination status in the study context.

Home birth was strongly associated with increased risk of delayed or no receipt of DTP vaccination in Morogoro region while third-trimester antenatal care visit was associated with an increased risk in Dar es Salaam. Distance to a healthcare facility is known to be a primary barrier to antenatal care, facility births, and timely childhood vaccination [18, 25, 26]. Furthermore, decreased health seeking behavior caused by negative perceptions of formal healthcare has also been shown to contribute to home births, poor antenatal care use, and delayed vaccination [26‐28]. However, there has not been widespread documentation of negative perceptions of routine child health services and vaccination in Tanzania. A qualitative study in rural Morogoro region has documented strong positive perception and high value of vaccination in the community [29]. Accordingly, interventions that target homebirths and improve healthcare system access in rural Tanzania and counselling interventions for women who book ANC late may significantly improve vaccination rates in urban Tanzania.

This study has several limitations. First, while the rural Morogoro region sample was a population-representative sample of all facility and home births within a demographic surveillance system, the Dar es Salaam site only recruited newborns born in a facility which may somewhat limit its representativeness. Nevertheless, the 2016 Tanzanian Demographic and Health Survey (DHS) found that 94% of women in Dar es Salaam delivered at a health facility and as a result the relatively small number of home births in this context would likely have minimal impact on our risk estimates [30]. The parent trial excluded infants in the first 3 days of life who were not able to feed orally. As a result, our study is not able to examine vaccination status among these high-risk infants; however, only 38 newborns were excluded for this reason in the parent trial and therefore would have minimal to no effect on our risk factor estimates. In addition, we did not collect information on the distance between home and the nearest health facility, which is known to influence the risk of delayed vaccination, particularly in rural areas [31]. Furthermore, in this study, fieldworkers conducted home visits primarily to assess infant vital status which may have had some impact on care-seeking behaviors; however, fieldworkers were not trained to counsel on vaccination timeliness and therefore the effect on delayed and no receipt of DTP vaccination is likely minimal. Nevertheless, there are several major strengths to this study. It is the largest birth cohort study of vaccination timing containing both an urban and rural sample. Furthermore, our prospective cohort design allowed us to accurately ascertain vaccination status and timing among infants who died during follow-up. Coverage and risk estimates from cross-sectional vaccination surveys, like the DHS, can be prone to survival bias since vaccination status is difficult to obtain post mortem as the vaccination card may be destroyed, mothers may travel after a child dies, or fieldworkers may be reluctant to ask for vaccination cards [32]. In addition, we prospectively followed infants from birth, which unlike cross-sectional studies, provides complete and accurate data on vaccination timing and birth risk factors.

We found that both urban and rural infants were at risk for not being vaccinated on time. While some of the risk factors for delayed or incomplete vaccination appear to be  important in both urban and rural settings, others varied significantly by context. Our data strongly suggests that national immunization programs should tailor interventions to the local context. In both urban Dar es Salaam and rural Morogoro Region, interventions targeting poor households, young mothers, and mothers with low education may decrease delayed or incomplete vaccination. However, in Morogoro Region, additional interventions focused on home births and fathers with low education appear to be particularly promising to improve timely vaccination. Future randomized trials and rigorous implementation evaluations should determine the effectiveness of intervention packages and strategies to address barriers to timely immunization in diverse urban and rural African settings.