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

Background: Delayed vaccination increases the time infants are at risk for acquiring vaccine-preventable diseases. Factors associated with incomplete vaccination are relatively well characterized in resource-limited settings; however, few studies have assessed immunization timeliness. Methods: We conducted a prospective cohort study examining Diphtheria-Tetanus-Pertussis (DTP) vaccination timing among newborns enrolled in a Neonatal Vitamin A supplementation trial (NEOVITA) conducted in urban Dar es Salaam (n = 11,189) and rural Morogoro Region (n = 19,767), Tanzania. We used log-binomial models to assess the relationship of demographic, socioeconomic, healthcare access, and birth characteristics with late or incomplete DTP1 and DTP3 immunization. Results: The proportion of infants with either delayed or incomplete vaccination was similar in Dar es Salaam (DTP1 11.5% and DTP3 16.0%) and Morogoro (DTP1 9.2% and DTP3 17.3%); however, the determinants of delayed or incomplete vaccination as well as their magnitude of association differed by setting. Both maternal and paternal education were more strongly associated with vaccination status in rural Morogoro region as compared to Dar es Salaam (p-values for heterogeneity < 0.05). Infants in Morogoro who had fathers and mothers with no education had 36% (95% CI: 22-52%) and 22% (95% CI: 10-34%) increased risk of delayed or incomplete DTP3 vaccination as compared to those with primary school education, respectively. In Dar es Salaam, mothers who attended their first antenatal care (ANC) visit in the 3rd trimester had 1.55 (95% CI: 1.36-1.78) times the risk of delayed or not received vaccination as compared to those with a 2nd trimester booking, while there was no relationship in Morogoro. In rural Morogoro, infants born at home had 17% (95% CI: 8-27%) increased risk for delayed or no receipt of DTP3 vaccination. In both settings, younger maternal age and poorer households were at increased risk for delayed or incomplete vaccination. Conclusion: We found some risk factors for delayed and incomplete vaccination were shared between urban and rural Tanzania; however, we found several context-specific risk factors as well as determinants that differed in their magnitude of risk between contexts. Immunization programs should be tailored to address context-specific barriers and enablers to improve timely and complete vaccination. This study utilized data from a randomized, double-blind controlled trial of Neonatal Vitamin A supplementation (NEOVITA). This trial enrolled 31,999 children in urban Dar es Salaam and rural Morogoro, Tanzania from August 2010 to March 2014 (ACTRN12610000636055). Briefly, newborns born at home or at a health facility were randomized to receive a mega-dose of vitamin A (50,000 IU) or placebo on the day of birth or within 3 days; trial recruitment and detailed data collection procedures have been presented elsewhere [7]. Newborns whose mothers were younger than 15 years old or older than 49 years were excluded from participation in the trial. In urban Dar es Salaam, 11,895 participants were enrolled at antenatal clinics and in labor wards of public health facilities in Kinondoni, Ilala, and Temeke districts. In Morogoro Region, 20,104 participants were recruited within a rural Health and Demographic Surveillance System (HDSS) that covers approximately 2400 km2 and allowed for enrollment of both health facility and home births in Kilombero, Ulanga and Kilosa districts. Newborns were eligible for randomization if they were able to feed orally, were born within the past three days, were not previously enrolled in other clinical trials, intended to stay in the study area for at least six months post-delivery, and the parents provided written informed consent to participate. At the baseline visit, trained study staff administered a baseline questionnaire to mothers in order to collect information on demographic, socioeconomic, and health factors. Trained fieldworkers visited the infants 1 and 3 days after enrollment, as well as 1, 3, 6, and 12 months after birth. During these visits, fieldworkers assessed the morbidity and mortality of the infants and ascertained information about the infants’ recent nutrition intake and any immunizations the infant received since the last visit [7, 8]. All newborns enrolled in the trial were given a child health card at the time of randomization. DTP vaccination status was determined by field workers who reviewed the child health card recorded dates of vaccination at each study visit (1 day, 3 days, and then 1, 3, 6, and 12 months). Child health cards were not available at less than 0.1% of all study visits and in these few cases the date of vaccination was reported by the mother or infant caregiver. In this study, we analyzed DTP1 and DTP3 vaccination timing. Tanzania Mainland recommended DTP1 vaccination at 4 weeks of age; however, during the trial in 2012, the pentavalent DTP + Haemophilus influenza type B (Hib) + Hepatitis B (HepB) vaccine was introduced and the recommended age at vaccination was increased to 6 weeks [9]. Tanzania recommended DTP3 at 14 weeks of age throughout the study. There is no universal definition of delayed vaccination for DTP vaccine, although most, including the Pan American Health Organization, define late vaccination as one month or more after the recommended age of vaccination [10]. At the time of the study, LMICs were advised to set their recommended age at DTP1 vaccination at 6 weeks (4 weeks–2 months) [11, 12]. As a result, in the primary analysis we define delayed DTP1 vaccination as receipt > 90 days of age based on the upper recommended age for DTP1 (2 months or 60 days) plus 30 days late. In a sensitivity analysis, we define delayed vaccination as > 72 days based on the Tanzanian recommended age for the first pentavalent vaccine dose (6 weeks or 42 days) plus 30 days late. Due to the staggered roll-out of the pentavalent vaccine program during the study, we were not able to determine if infants were due for their first DTP vaccination at 4 weeks or 6 weeks and therefore we only present the conservative definition of 72 days late in the sensitivity analysis. Infants who were lost to follow up, died or were vaccinated before 15 days were excluded from the DTP1 analysis. The standard recommended age range for DTP3 vaccination in LMICs was 14 weeks [12 weeks–6 months]) [11, 12]. As a result, we defined delayed DTP3 vaccination as receipt > 210 days (7 months) of age in the main analysis. In a sensitivity analysis, we defined delayed DTP3 vaccination as > 128 days of age based on the Tanzanian recommended age of 14 weeks (98 days) plus 30 days late. Infants who were lost to follow up, died or were vaccinated before 60 days were excluded from the DTP3 analysis. Univariate and multivariate relative risks of delayed or incomplete DTP1 and DTP3 vaccination (> 90 days and > 210 days, respectively) were calculated using log-binomial models stratified by Dar es Salaam and Morogoro region [13]. We also present sensitivity analyses using delayed vaccination definitions of > 72 days and > 128 days for DTP1 and DTP3, respectively. Log-binomial models did not converge in a few instances and in these cases log-Poisson models, which provide consistent but not fully efficient estimates of the relative risk and its confidence intervals, were used [14]. Preterm birth was defined as delivery at < 37 weeks gestation as assessed by maternal report of last menstrual period. Small-for-gestational-age (SGA) was defined, with the use of Oken standards, as birth weight < 10th percentile for gestational age and sex [15]. A wealth index was generated based on household ownership of assets, and households were categorized into wealth quintiles stratified by Dar es Salaam and Morogoro residence [16]. Due to collinearity, low birth weight was modeled separately from preterm birth and small-for-gestational age. P-values for trend in categorical analyses were calculated by treating the median value of each category as a continuous variable. The log-rank test was used to assess the statistical significance of potential effect modification of predictors of interest by study site (Dar es Salaam versus Morogoro). Missing data were retained with use of the missing indicator method. All P values were 2-sided with a P <  0.05 considered statistically significant. All of the analyses for this study were conducted using R version 3.3.1.