Bulletin of the World Health Organization

Vaccination timing of low-birth-weight infants in rural Ghana: a population-based, prospective cohort study

Maureen O’Leary a, Sara Thomas a, Lisa Hurt b, Sian Floyd a, Caitlin Shannon c, Sam Newton d, Gyan Thomas e, Seeba Amenga-Etego e, Charlotte Tawiah-Agyemang e, Lu Gram a, Chris Hurt f, Rajiv Bahl g, Seth Owusu-Agyei e, Betty Kirkwood a & Karen Edmond h

a. Department of Infectious Disease Epidemiology, Faculty of Epidemiology and Population Health, London School of Hygiene & Tropical Medicine, Keppel St, London, WC1E 7HT, England.
b. Institute of Primary Care and Public Health, University of Cardiff, Cardiff, Wales.
c. Engender Health, New York, United States of America (USA).
d. Department of Community Health, Kwame Nkrumah University of Science and Technology, Kumasi, Ghana.
e. Kintampo Health Research Centre, Kintampo, Ghana.
f. Institute of Cancer and Genetics, University of Cardiff, Cardiff, Wales.
g. Department of Maternal, Newborn, Child and Adolescent Health, World Health Organization, Geneva, Switzerland.
h. School of Paediatrics and Child Health, University of Western Australia, Crawley, Australia.

Correspondence to Maureen O’Leary (email: maureen.oleary@lshtm.ac.uk).

(Submitted: 28 July 2015 – Revised version received: 11 February 2016 – Accepted: 11 February 2016.)

Bulletin of the World Health Organization 2016;94:442-451D. doi: http://dx.doi.org/10.2471/BLT.15.159699


Approximately 14% of infants born in low- and middle-income countries have a low birth weight (weighing < 2.50 kg at="">1 It has been reported that in high-income settings, low-birth-weight infants have an increased risk of vaccine-preventable diseases, such as pertussis,2 invasive pneumococcal disease35 and Haemophilus influenzae type b (Hib).6 However, it is not known whether such risk exists in low-income settings. Timely vaccination of low-birth-weight infants, including booster doses, is important because these infants have lower passive immunity before vaccination7 and may respond sub-optimally to primary vaccination.8 Vaccination has similar efficacy and safety in low-birth-weight infants compared with non-low-birth-weight infants,8 and therefore vaccination is recommended at the same chronological age as other infants.9

Studies from high-income settings indicate that low-birth-weight infants are vaccinated later than non-low-birth-weight infants.10,11 Regardless of whether they are at increased risk, delayed vaccination of low-birth-weight infants prolongs their risk period for contracting vaccine-preventable diseases, especially Hib and Streptococcus pneumoniae,3,12 which are most prevalent in the first few months of life. Studies of the effect of low birth weight on timely vaccination in low-income settings, however, are lacking.

We aimed to measure the timing of vaccination of low-birth-weight infants compared with non-low-birth-weight infants by analysing data from a population-based, prospective cohort study in Ghana. Our primary objectives were to assess whether low birth weight is a determinant of delayed first and third dose diphtheria–tetanus–pertussis (DTP1 and DTP3) vaccination; and whether maternal education or socioeconomic status modified the association between birth weight and vaccination with DTP1 and DTP3. As a secondary objective, we aimed to quantify other determinants of delayed DTP1 and DTP3 vaccination.


Study design and setting

We studied a cohort of infants nested within a large randomized, double-blind, placebo-controlled trial of neonatal vitamin A supplementation conducted in Ghana between August 2010 and February 2013.13 The trial was conducted at the Kintampo Health Research Centre in Kintampo, Ghana. The trial procedures and study area have been described elsewhere.14

Ethics approval for the study was granted by the ethics committees of the World Health Organization (WHO), the London School of Hygiene & Tropical Medicine and the Kintampo Health Research Centre.

DTP vaccination in Ghana is recommended at 6, 10 and 14 weeks of age. Children are vaccinated at health facilities, community health planning system compounds or mobile outreach clinics. For each administered vaccine, the date and place of administration and vaccine batch number are usually documented in the child health record book. These may also be documented on a vaccination card or in the mother’s antenatal card. Infants who have never attended a child health clinic may not have a written record.

Enrolment and data collection

Trained fieldworkers enrolled all consenting women aged 15–49 years residing in the study area into a reproductive surveillance system to document pregnancies and deliveries. All infants born in the study area were assessed for eligibility (eligible infants were aged ≤ 3 days at screening, could suck or feed and were staying in the study area for 6 months after enrolment) and mothers were asked for informed written consent for enrolment in the trial. Infants were weighed using calibrated electronic (38%; 8723 of enrolled infants) or spring (62%; 14 232) scales, to record birth weights to the nearest 0.1 kg (electronic scales) or 0.2 kg (spring scales). Only five (0.2%) infants were weighed later than 72 hours after delivery. The fieldworkers collected data on infant (sex and multiple delivery), maternal (age, education, occupation, and illness before delivery) and household characteristics (ethnicity, religion, socioeconomic status, distance to health facility and number of children in household).

The enrolled infants were visited monthly for the first year of life to collect data on the types and dates of vaccines given. We looked for written documentation of vaccines from all possible sources, including the child health record book, the mother’s antenatal card and vaccination cards. The infant’s caregiver (usually the mother) was also asked to recall what vaccines had been given. We also collected data on the infant’s vital status and on illnesses since the previous visit.

Follow-up started at birth. It ended at the vaccination date for vaccinated infants, and the end of the risk period for unvaccinated infants not lost to follow-up. For those lost to follow-up before the end of the risk period, follow-up ended on the last date the written record was viewed, for unvaccinated infants whose record was viewed; or on the last date the infant was seen, for unvaccinated infants whose record was never viewed; or on the date of death, for unvaccinated infants who died before the end of the risk period and whose record was viewed after their death.

For the analyses we included all infants from the trial with known vaccination status and dates and with complete data on covariates. We excluded infants who were lost to follow-up or died before the vaccination due date.


We classified infants’ vaccination status as follows: (i) vaccinated, date known (written record had a plausible vaccination date); (ii) vaccinated, date unknown (record had clearly documented vaccination but with the date missing, illegible or implausible); (iii) unvaccinated (record was seen but had no documented vaccination date or any evidence of vaccination; or record was never seen and mother consistently reported infant had never been vaccinated); or (iv) vaccination status unknown (mother reported that infant had been vaccinated but did not specify the vaccine; or infant was never seen in follow-up).

We categorized birth weight into four standard categories: ≥ 2.5 kg (i.e. non-low birth weight); 2.0–2.4 kg; 1.5–1.9 kg; <>1,15

Outcome measures

The study outcomes were delayed receipt of DTP1 and DTP3. There is no standard approach to the assessment of delayed vaccination and several definitions based on predefined cut-offs have been described.1618 To assess how the effect of birth weight may vary over time, we defined risk periods for delayed vaccination up to 4, 8 and 12 weeks after the vaccination due date. For DTP1 we therefore analysed vaccination rates from birth up to 10, 14 and 18 weeks of age. For DTP3 we analysed vaccination from birth up to 18, 22 and 26 weeks of age.

Data analysis

The data were double-entered and processed at the Kintampo Health Research Centre using the SQL Server 2008 data management system (Microsoft Corp., Redmond, USA). Inconsistencies and errors in the vaccination dates were corrected, with senior fieldworkers visiting mothers to review the written record and verify the dates if necessary.

All analyses were conducted using the Stata package version 13.1 (StataCorp, College Station, USA). We generated Kaplan–Meier curves of time to vaccination in low-birth-weight compared with non-low-birth-weight infants in the first year of life for DTP1 and DTP3. Vaccination rate ratios, adjusted for a priori selected factors, were obtained for each risk period using multivariable Poisson regression, informed by a hierarchical framework of the recognized determinants of vaccination (Fig. 1).12,1618 The initial model included distal determinants of vaccination, then intermediate determinants were added, followed by birth weight and, finally, infant illness at the time the vaccine was due (as this was considered to be a possible mediator of the association between birth weight and vaccination).19 We assessed the statistical association between vaccination and each explanatory variable using likelihood ratio tests and 95% confidence intervals (CI). We also investigated whether the association between birth weight and vaccination varied by maternal education or socioeconomic status by testing the interaction of birth weight with these variables.

Fig. 1. Hierarchical framework of determinants of infant vaccination in the prospective cohort study in rural Ghana, 2010–2013
Fig. 1. Hierarchical framework of determinants of infant vaccination in the prospective cohort study in rural Ghana, 2010–2013
Note: The effects of distal variables were hypothesized to be mediated by intermediate and proximal variables, and intermediate variables by the proximal variables. Among the proximal variables, illness was hypothesized to mediate the effect of birth weight. Mediation is indicated by arrows linking the variables across different levels.

Two sets of sensitivity analyses were undertaken. First, to assess whether delayed DTP3 vaccination simply reflected delayed DTP1 vaccination, we repeated the DTP3 analyses, starting follow-up at receipt of DTP1 vaccination and ending 12 weeks after receipt of DTP1. Second, to examine the effect of possible misclassification of vaccine status for infants categorized as never vaccinated but whose written record was never viewed, we excluded these infants and repeated the analyses of DTP1 vaccination up to 18 weeks from birth and DTP3 vaccination up to 26 weeks.


Of 27 330 live births identified in the study area, 26 414 infants were screened for eligibility for the trial and 22 955 were enrolled (Fig. 2); 22 361 (97.4%) and 22 192 (96.7%) infants were included in the analysis of DTP1 and DTP3 respectively. Low-birth-weight infants were more likely to be excluded from our analysis, as were those with illness reported around the vaccination due date, those from multiple births and those born to mothers of lower socioeconomic status, of non-Akan ethnicity, with lower education, with lower employment grades or living more than 5.0 km from a health facility (Table 1). Of the infants included in the DTP1 analysis, 18 979 (84.9%) were normal birth weight and 3382 (15.1%) were low birth weight: 2916 (13.0%) weighed 2.0–2.4 kg, 386 (1.7%) 1.5–1.9 kg and 80 (0.4%) < 1.5 kg. the="" birth="" weight="" distribution="" was="" the="" same="" for="" infants="" in="" the="" dtp3="" analysis:="" 18 850="" (84.9%)="" weighed="" ≥ 2.5 kg,="" 2886="" (13.0%)="" 2.0–2.4 kg,="" 378="" (1.7%)="" 1.5–1.9 kg="" and="" 78="" (0.4%)="">< 1.5 kg>

Fig. 2. Identification, recruitment and inclusion of participants in the prospective cohort study on infant vaccination in rural Ghana, 2010–2013
Fig. 2. Identification, recruitment and inclusion of participants in the prospective cohort study on infant vaccination in rural Ghana, 2010–2013
DTP1: first dose of diphtheria–tetanus–pertussis vaccine; DTP3: third dose of diphtheria–tetanus–pertussis vaccine.
Note: Overall and disaggregated percentages may not agree due to rounding.

Delayed vaccination

Birth weight

Although uptake of vaccination was high (> 95%) for all infants by 1 year of age, low birth weight was associated with later vaccination for both DTP1 and DTP3. Median ages at DTP1 vaccination were 8 weeks (interquartile range, IQR: 6.7–9.6 weeks) for infants weighing ≥ 2.5 kg at birth; 8.3 weeks (IQR: 6.9–9.9) for those 2.0–2.4 kg; 8.4 weeks (IQR: 6.9–10.7) for those 1.5–1.9 kg and 9 weeks (IQR: 7.4–11.9) for those < 1.5 kg. for="" dtp3,="" the="" corresponding="" median="" ages="" at="" vaccination="" were="" 18.4="" weeks="" (iqr:="" 16.3–22.1),="" 18.6="" weeks="" (iqr:="" 16.6–22.3),="" 19.6 weeks="" (iqr:="" 16.6–23.3)="" and="" 20.4 weeks="" (iqr:="" 17.7–25.1),="">

The Kaplan–Meier curves showed that DTP1 vaccination rates over the days since birth were also lower for infants weighing < 1.5 kg and="" those="" weighing="" 1.5–1.9 kg="" compared="" with="" those="" weighing="" ≥ 2.5 kg="">). After adjustment for other variables, there was evidence of progressively delayed vaccination with decreasing birth weight (P-value for trend < 0.0001). infants="" weighing="">< 1.5 kg at="" birth="" had="" a="" dtp1="" vaccination="" rate="" approximately="" 40%="" lower="" than="" non-low-birth-weight="" infants="" by="" the="" age="" of="" 10 weeks="" (adjusted="" rate="" ratio,="" arr: 0.58;="" 95%="" ci:="" 0.43–0.77)="" and="" age="" 18 weeks="" (arr:="" 0.63;="" 95%="" ci:="" 0.50–0.80).="" infants="" weighing="" 1.5–1.9 kg="" had="" vaccination="" rates="" approximately="" 25%="" lower="" than="" non-low-birth-weight="" infants="" at="" these="" time="" points="" (arr:="" 0.71;="" 95%="" ci:="" 0.62–0.81="" and="" arr:="" 0.76;="" 95%="" ci:="" 0.69–0.85,="" respectively;="" table 2,="" available="" at:="">

Fig. 3. Time to vaccination with first dose and third dose of diphtheria–tetanus–pertussis vaccine, by birth weight in the prospective cohort study in rural Ghana, 2010–2013
Fig. 3. Time to vaccination with first dose and third dose of diphtheria–tetanus–pertussis vaccine, by birth weight in the prospective cohort study in rural Ghana, 2010–2013
DTP1: first dose of diphtheria–tetanus–pertussis vaccine; DTP3: third dose of diphtheria–tetanus–pertussis vaccine.

Similar results were observed for DTP3 (Fig. 3). The findings were also similar for DTP1 and DTP3 vaccination at 8 weeks after the due date (Table 2).

Adjusting for illness had little effect on the magnitude of the association between birth weight and vaccination for both DTP1 and DTP3 (Table 2).

Other variables

Younger maternal age, lower educational attainment, and longer distance to the nearest health facility were associated with moderate reductions in the DTP1 and DTP3 vaccination rates of approximately 10–20% at ages 10 and 18 weeks, whereas higher employment grade was associated with moderate increased vaccination rates at these ages (Table 3, available at: http://www.who.int/bulletin/volumes/94/5/15-159699). In the final model (after adjusting for potential mediating variables) low socioeconomic status of mothers was associated with a 15% increased DTP3 vaccination rate at 18 weeks, whereas no association with DTP1 vaccination was observed. Muslim religion and larger family size were associated with > 10% reduction in DTP3 vaccination rates but had no, or only a small, association with DTP1 vaccination rates. None of the other variables measured had notable associations with DTP1 or DTP3 vaccination rates at any ages.

Sensitivity analyses

Adjusting for late vaccination with DTP1 decreased the effect size for the association between birth weight and the rate of DTP3 vaccination for infants weighing 1.5–1.9 kg (12 weeks after DTP1 aRR: 0.98; 95% CI: 0.85–1.13) compared with an aRR of 0.82 (95% CI: 0.73–0.92) at 12 weeks after the DTP3 due date, but the effect size for infants weighing < 1.5 kg was="" largely="" unchanged="">

Excluding unvaccinated infants whose written record was never seen had little impact on the effect size of the explanatory variables for DTP1 or DTP3.

Modifying factors

When we looked at other factors that might modify the association between birth weight and delayed vaccination there was no evidence that the effect of birth weight on vaccination with DTP1 or DTP3 varied by socioeconomic status (P-values for interaction all > 0.4), or that the rate of vaccination with DTP1 varied by maternal education, when measured at age 10 weeks (P = 0.3338) or age 18 weeks (P = 0.2675). However, for DTP3 vaccination there was some evidence that the effect of birth weight on the vaccination rate at age 18 weeks (P = 0.0219) and age 26 weeks (P = 0.0813) varied with maternal education, with a more pronounced reduction in vaccination rate among smaller infants born to mothers with higher educational attainment (aRR for infants weighing < 1.50 kg="" at="" age="" 18="" weeks:="" 0.37;="" 95%="" ci:="" 0.19–0.72;="" arr="" at="" 26="" weeks:="" 0.63;="" 95%="" ci:="" 0.50–0.80).="" when="" infants="" with="" delayed="" receipt="" of="" dtp1="" were="" excluded="" from="" the="" analysis,="" this="" effect="" was="" no="" longer="">


The results of this study provide evidence that low-birth-weight infants in Ghana are vaccinated later than non-low-birth-weight infants. The effect persisted up to 12 weeks after the vaccination due date and was evident for both DTP1 and DTP3, even after controlling for other determinants of delayed vaccination.

The results are consistent with previous reports from high-income countries of delayed vaccination in low-birth-weight infants.10,11,2022 In addition, a study of low-birth-weight infants in Guinea-Bissau, which did not look at timeliness, reported lower uptake of DTP1 at 8 weeks of age among smaller low-birth-weight infants compared with larger low-birth-weight infants.23 A North American study reported that both parents and vaccine-providers had erroneous beliefs that initiation of vaccination depended on the degree of prematurity and the infant’s weight.22 In addition, a review of 47 studies in the grey literature from low-income settings reported parental reluctance to bring sick, weak or malnourished children for vaccination for reasons of social stigma and fatalism; these have also been cited as reasons for non-vaccination by vaccine providers.19

Low-birth-weight infants in low-income settings are known to have higher rates of illness and death in the first year of life than non-low-birth-weight infants.15,2427 Data from high-income settings indicate that they also have higher rates of illness from vaccine-preventable diseases.24,6 The risk and consequences of illness related to vaccine-preventable diseases in low-birth-weight infants in low-income settings is not known and may differ from those in high-income settings. Without this information it is difficult to fully understand the implications of delayed vaccination on clinical outcomes for these infants. However, we do know that delayed vaccination of these infants will prolong their risk period for contracting these diseases and may also reflect an underuse of health services by the caregivers of these infants. Given this increased risk of illness and death, it is essential that all opportunities for vaccination and health care for low-birth-weight infants be exploited.

We also identified several additional determinants of delayed vaccination – low maternal age and educational attainment and longer distance to the nearest health facility – that reflect persisting inequities in access to and uptake of vaccination in our study population. This is consistent with previous findings from the study area16 and the issue of inequities in coverage of vaccination have featured in global vaccine policy.28

The strengths of our study include the high quality population-based surveillance system and low loss to follow-up. Almost all infants were weighed within 72 hours of delivery by trained fieldworkers using calibrated scales, thus minimizing the likelihood of misclassification of infants by birth weight. Similarly, we collected high quality data on vaccination – from both written records and maternal recall – and we employed a rigorous approach to resolving inconsistencies in these data. Although recall data is used in the generation of routine vaccine uptake estimates,29 their validity may vary.30,31 The validity of our recall data was maximized by the continuous nature of the data collection in our study. Infants with recall data accounted for less than 0.6% of all infants included in the analyses and had little impact on our estimates. Furthermore, the inclusion of over 22 000 infants ensured that the study had sufficient power to show effects in small subgroups.

Aspects of this study that may have affected the generalizability of our findings are that our study sample may have experienced more timely vaccination compared with the general population. A higher proportion of low-birth-weight infants than non-low-birth-weight infants were excluded from our analyses, either because they did not meet the inclusion criteria for enrolment in the trial or because more of them were lost to follow-up or had missing data (including missing vaccination data) than non-low-birth-weight infants. Those excluded could have experienced greater delay in receiving their vaccines compared with the included low-birth-weight infants, possibly causing some underestimation of the association between low birth weight and timely vaccination in our population. As less than 5% of enrolled infants were excluded, this was unlikely to have changed the results appreciably and important delays in vaccination were still observed among low-birth-weight infants. Mothers of enrolled infants were asked about their infant’s vaccination status at monthly visits, possibly increasing their awareness of the need to vaccinate their infants. This increased awareness, however, would not have been differentially affected by birth weight and would lead to an overall underestimation of delayed vaccination.

Other limitations are that we did not have reliable data on gestational age and therefore we were not able to assess whether delayed vaccination was associated with prematurity or whether all low-birth-weight infants were affected regardless of gestational age. This study was also not designed to assess the association between delayed vaccination and clinical outcomes such as vaccine-preventable diseases or hospitalizations. Consequently, we do not know whether those infants who had delayed vaccination were more likely to contract vaccine-preventable diseases or to report elevated rates of illness or hospitalization. Eleven explanatory variables were included in our secondary analysis, thus increasing the potential for type 1 errors (finding statistically significant results by chance alone). Finally, we did not collect any qualitative data on the reasons for delayed vaccination of low-birth-weight infants in our study sample. This limits our interpretation of the findings. It may be that vaccination was delayed for reasons beyond the control of both the caregivers and the vaccine providers, such as lack of vaccines or staff, although it is reasonable to assume that these would not be distributed differently among low-birth-weight compared with non-low-birth-weight infants.


Current global policy on vaccination advocates the identification of groups that are underserved by vaccination,28 yet data on uptake and timeliness of vaccination in low-birth-weight infants are not currently included in routine evaluations of vaccination programmes. These data are feasible to collect in low-income settings; doing this would contribute to a more comprehensive evaluation of the performance of vaccination programmes and would inform the development of strategies to improve uptake and timing of vaccination in all countries.28 Even though several organizations in high-income countries have made specific recommendations about vaccination of low-birth-weight infants,9,32 international guidelines are lacking.

Efforts to improve the vaccination of low-birth-weight infants, for example by education of caregivers and vaccine-providers, are warranted. Further research is needed in low-income countries to understand the reasons for delayed vaccination of low-birth-weight infants and to inform strategies to improve the timeliness of vaccination.


We would like to thank the mothers and infants who participated in this study, the staff of the Kintampo Health Research Centre, WHO and Robin Nesbitt. Seth Owusu-Agyei is also affiliated with the London School of Hygiene & Tropical Medicine.


The work was funded by the Bill & Melinda Gates Foundation.

Competing interests:

None declared.