A total of 14,372 studies were identified from database searches, and 11 studies were identified from review of reference lists and clinical trial registries. After removing duplicates, the abstracts and titles of 9331 studies were screened for inclusion, of which 68 studies met inclusion criteria. After reviewing 68 full-text articles, 36 unique studies met inclusion criteria and 32 were suitable for meta-analysis (Fig. 1). A summary of full-text articles excluded from further reviews is provided in Table S1. Eighteen (50%) studies were randomized trials, including 15 randomized controlled trials and three cluster randomized controlled trials. We identified one non-randomized controlled trial13. Seventeen (47%) studies were quasi-experimental, including ten studies using a historical control, six studies using a geographic control, and one study using both a historical and geographic control. Fourteen studies aimed to increase uptake of influenza vaccine, four to increase uptake of pertussis-containing vaccines, and six to increase uptake of influenza and pertussis-containing vaccines. Eleven studies aimed to increase uptake of the tetanus toxoid vaccine, and one study aimed to increase COVID-19 vaccine uptake.
PRISMA diagram for selection of studies evaluating the effect of interventions to increase the uptake of vaccines recommended during pregnancy.
Nineteen (53%) of the 36 included studies were conducted in the United States (Table 1). The remaining studies were conducted in one of 14 other countries; 12 (34%) in a low or middle-income country (LMIC) and 24 (67%) in a high-income country (HIC; Fig. S1). We identified 24 studies evaluating patient-level (demand-side) interventions and 17 studies that evaluated systems or provider-level (supply-side) interventions; 19 studies implemented patient-level interventions only13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,38, 11 studies implemented systems or provider-level interventions exclusively31,32,33,34,35,36,37,38,39,40,41, and six were bundled with interventions that targeting patients and providers or health systems42,43,44,45,46,47.
Risk of bias
Among the 17 randomized clinical trials, eight were deemed at low risk of bias, eight had some concerns, and one was deemed at high risk of bias (Table 2). Among the 15 non-randomized studies, one was deemed to be at low risk of bias, five were at moderate risk of bias, eight were at serious risk of bias, and one was at critical risk of bias (Table 3). The study deemed to be at critical risk of bias was due to self-selection into the intervention and limited comparability between groups13. This study was excluded from meta-analyses.
Patient-level interventions
Many of the interventions aligned well with evidence-based strategies recommended by the Community Preventive Services Task Force12. We identified 18 patient-level, demand-side only studies; three studies investigated the use of patient reminder and recall systems16,20,21, nine investigated the use of clinic-based patient education15,16,17,18,22,23,25,26,46, two studies investigated the use of patient or family incentive rewards through conditional cash transfers27,28, two studies investigated the use of persuasive messaging with patient education14,24, and one study investigated the use of patient-held immunization records38.
Handheld immunization records appeared to increase vaccination rates the most, albeit with low certainty (Table 4), with effect estimates for increasing tetanus toxoid vaccine uptake ranging from 4.85 (95% CI: 3.79, 7.42) to 6.41 (95% CI: 5.36, 7.67). The pooled RR was 5.65 (95% CI: 4.30, 7.42), with possibly substantial heterogeneity (I2 = 69%, tau = 0.03, P = 0.07). These estimates were also derived from multiple sites within a single study, making it difficult to draw firm conclusions. Two studies examined the use of patient incentives via conditional cash transfer programs, where participants received cash stipends in exchange for receiving antenatal care services27,28. In Nigeria, Sato and Fintan (2019) showed that payments in the amount of C300 or C800 (equivalent to $2.00 and $5.33 US dollars, respectively) increased tetanus toxoid vaccination modestly (RR 1.38; 95% CI: 1.28, 1.49 and RR 1.56; 95% CI: 1.46, 1.67, respectively)28. However, a similar conditional cash transfer program in rural India did not observe a significant effect on tetanus toxoid vaccination (RR 1.04; 95% CI: 0.97, 1.12)27. The pooled RR indicated a possible modest effect of conditional cash transfer programs (pooled RR 1.31; 95% CI: 1.03, 1.66) but with possibly considerable heterogeneity (I2 = 97%; tau=0.04; P < 0.001) and low certainty (Table 4).
Other patient-level interventions showed more modest effects on vaccine uptake during pregnancy. Although most studies evaluated the use of patient education strategies, these showed little to moderate improvements in vaccination rates and were entirely implemented in HIC settings. Patient education studies used printed materials (i.e., posters, statements, or pamphlets) or electronic materials (i.e., interactive website, videos, or text messages). One study used brief one-on-one education with a healthcare provider22. Effect estimates ranged from 0.98 (95% CI: 0.84, 1.15)16 to 2.11 (95% CI: 1.22, 3.67)22, with three studies showing a significant improvement in influenza and/or pertussis vaccine uptake17,22,26. The pooled RR indicated a modest effect of patient education on influenza or pertussis vaccine uptake (pooled RR 1.18; 95% CI: 1.04, 1.33), with possibly substantial heterogeneity (I2 = 63%; tau = 0.03; P < 0.01) (Fig. 2) and low certainty (Table 4). Results were similar when we considered any study implementing the effect of patient education, either alone or in combination with other interventions, and pooled RRs by vaccine suggested a possible modest increase in influenza vaccine uptake (pooled RR 1.21; 95% CI: 1.06, 1.37) but no increase in pertussis vaccine uptake (pooled RR 1.00; 95% CI: 0.96, 1.04) (Fig. S2).
Individual and pooled effect estimates for interventions that target patient-level demand-side interventions to increase the uptake of recommended vaccines during pregnancy. Pooled effect estimates were estimated by random effects meta-analysis overall and by intervention. CI indicates 95% confidence intervals, and RR indicates relative risk estimates.
Three studies evaluated the use of text message-based immunization reminders. Effect estimates for the use of text message reminders ranged from 0.90 (95% CI: 0.53, 1.54)20 to 1.06 (95% CI: 0.94, 1.19)21. The pooled RR indicated no effect of SMS-based immunization reminder systems on influenza vaccine uptake (pooled RR 1.01; 95% CI: 0.93, 1.09), with possibly low heterogeneity (I2 < 0.1; tau < 0.01; P = 0.51) and very low certainty (Table 4). Two small US studies (n < 36 per group) evaluated the use of different persuasive messaging using an affective messaging video or iBook14,24. Neither study showed a significant effect on influenza or pertussis vaccine uptake (pooled RR 0.94; 95% CI: 0.42, 2.11; I2 < 0.1; tau < 0.01; P = 0.92), and the evidence in support of a persuasive message was deemed to have very low certainty (Table 4).
Across all studies evaluating patient-level interventions exclusively, the evidence suggested a possible modest effect of patient-level interventions (pooled RR 1.37; 95% CI: 1.11, 1.69) with possibly considerable heterogeneity (I2 = 96%; tau = 0.24; P < 0.001) (Fig. 2). Among the 12 randomized controlled trials, six were deemed to be at low risk of bias, and six had some concerns (Table 2). Among the three quasi-experimental studies, one was at moderate risk, one at serious risk, and one at critical risk of bias (Table 3). Results from the six randomized controlled trials at low risk of bias showed no effect of patient education (n = 5/5)15,18,20,23,25 but increases in tetanus vaccination after cash incentives (n = 1)28.
We identified little evidence for publication bias (Egger statistic = 1.46; P = 0.33). However, examination of funnel plots showed more than expected larger studies with positive findings identified in the literature for patient-level interventions (Fig. S3), including patient education interventions (Fig. S4). Patient-level evaluations were deemed to have low certainty evidence in increasing vaccine uptake during pregnancy (Table 4).
Systems or provider-level interventions
Among the 11 studies of supply-side only interventions, three studies trialed some form of an Assessment-Feedback-Incentives-eXchange (AFIX) quality improvement program to increase influenza (n = 1) or both influenza and pertussis (n = 2) vaccine uptake32,34,46, three trialed the use of performance-based financing to improve antenatal care services (including tetanus toxoid vaccination)36,37,40, two trialed vaccine delivery systems for influenza or pertussis vaccination33,48, one trialed the use of a provider reminder system31, one trialed the use of a mobile health platform for improving the continuum of antenatal care35, and one trialed the use of socially-accountable medical education39.
Results did not strongly support supply-side interventions that were evaluated. Among AFIX interventions, effect estimates ranged from 0.94 (95% CI: 0.88, 1.00)46 to 1.54 (95% CI: 1.39, 1.71)32, with a pooled RR estimate of 1.13 (95% CI: 0.96, 1.33) with possibly considerable heterogeneity (I2 = 94%, tau = 0.03; P < 0.001) (Fig. 3) and very low certainty (Table 4). Effect estimates from the two LMIC-based studies of performance-based financing to increase tetanus toxoid vaccination ranged from 0.85 (95% CI: 0.78, 0.92)36 to 1.08 (95% CI: 0.98, 1.20)36. The pooled RR was 0.95 (95% CI: 0.83, 1.08), with possibly considerable heterogeneity (I2 = 85%; P < 0.001) and very low certainty (Table 4). The single study in India that evaluated the use of a mobile health (mHealth) platform for supporting case management and continuum of care for maternal and child health services showed a modest improvement in tetanus toxoid vaccination (RR 1.07; 95% CI: 1.06, 1.08)35 with very low certainty. The US study that evaluated the use of a healthcare provider reminder via an electronic health record alert, showing a 46% improvement in influenza vaccination rates (RR 1.46; 95% CI: 1.31, 1.63)31 with very low certainty. A study in Canada trialed different vaccine delivery systems and showed that administration of pertussis vaccine in obstetrics clinics or family medicine group practices resulted in higher uptake compared to the pre-intervention standard of vaccine administration in community clinics (RR 1.30; 95% CI: 1.17, 1.44 and RR 1.31; 95% CI: 1.17, 1.45, respectively)33 with very low certainty. Finally, one study in the Philippines investigated the use of a socially accountable medical education program to improve community health services, which admitted students who more accurately reflected the geographical, ethnic, and socio-economic diversity of the school’s reference population and applied an extended community-engaged service learning model with locally relevant curriculum. The socially accountable medical education program showed no effect on tetanus toxoid vaccination (RR 0.97; 95% CI: 0.93, 1.00)39 with very low certainty.
Individual and pooled effect estimates for interventions that target health systems and/or healthcare providers to increase the uptake of recommended vaccines during pregnancy. Pooled effect estimates were estimated by random effects meta-analysis overall and by intervention. AFIX indicates Assessment, Feedback, Incentive, eXchange programs, CI indicates 95% confidence intervals, and RR indicates relative risk estimates.
Results were similar when we considered provider or systems-level interventions when administered alone or in combination with other interventions. Seven studies evaluated provider education, either alone or in combination with patient-level interventions (Fig. S5). Pooled RR showed a modest improvement in influenza vaccine uptake (pooled RR 1.20; 95% CI: 1.02; 1.42) with possibly considerable heterogeneity (I2 = 89%; P < 0.001). A modest effect was similarly observed for pertussis vaccine (pooled RR 1.63; 95% CI: 0.74, 3.57), but with wider confidence intervals and possibly considerable heterogeneity (I2 = 92%; P < 0.001). Three studies examined provider reminders, either alone or in combination with other interventions (Fig. S6), and these showed modest improvement in either tetanus toxoid or influenza vaccine uptake (pooled RR 1.18; 95% CI: 0.97, 1.44) with possibly considerable heterogeneity (I2 = 94%; P < 0.001). Six US studies evaluated the nomination of an immunization champion (i.e., a dedicated member of staff who was compensated for championing immunization services and implementing quality improvement interventions) and enhanced electronic immunization documentation within practices (Fig. S7), showing possible modest improvement in influenza vaccine uptake (pooled RR 1.22; 95% CI: 1.04, 1.43) and pertussis vaccine uptake (pooled RR 1.39; 95% CI: 1.03, 1.57). However, meta-analyses indicated possibly considerable heterogeneity for both (I2 = 87% and I2 = 91%, respectively). Five studies evaluated the use of standing orders and AFIX programs and showed possible modest improvements in influenza vaccination (pooled RR 1.21; 95% CI: 1.03, 1.43) and pertussis vaccination (pooled RR 1.37; 95% CI: 0.99, 2.47), both with possibly considerable heterogeneity (I2 = 90% and I2 = 92%, respectively) (Fig. S8).
Across all studies evaluating provider or systems-level interventions exclusively, the evidence suggested a possible small effect of provider or systems-level interventions (pooled RR 1.09; 95% CI: 1.00, 1.19) with possibly considerable heterogeneity (I2 = 93%; tau=0.03; P < 0.001) (Fig. 3). Among the two randomized controlled trials, one was deemed to be at low risk of bias, and one had some concerns (Table 2). Among the seven quasi-experimental studies, two were at moderate risk, and five were at serious risk of bias (Table 3). Results from the one randomized controlled trial at low risk of bias found no increase in influenza vaccination during pregnancy after an opt-out approach to immunization48.
We observed no statistical evidence of publication bias in systems or provider-level interventions (Egger statistic = 0.28; P = 0.83); however, examination of funnel plots showed more studies with positive findings than would be expected appear in the published literature for provider or systems-level interventions (Fig. S9), including provider education (Fig. S10), provider reminders (Fig. S11), immunization champions and enhanced vaccine documentation (Fig. S12), and standing orders and AFIX programs (Fig. S13).
Patient and provider or systems-level interventions
Six studies were identified that evaluated ‘bundled’ multi-level interventions that addressed both demand and supply-side factors (Fig. 4)42,43,44,45,46,47. Five of these studies were conducted in the US and one in Pakistan. Bundled interventions often incorporated patient-level components like patient education and provider and systems-level components such as AFIX, enhanced vaccine documentation, and standing orders for vaccination. Effect estimates ranged from 1.00 (95% CI: 0.94, 1.06)36 to 5.42 (95% CI: 3.10, 9.48)34. Among the two randomized trials in the US, neither showed a significant effect of multi-level interventions on influenza or pertussis vaccine uptake45,46. Three US studies using a historical control reported a significant increase in influenza vaccine uptake42, pertussis vaccine uptake43, or influenza and pertussis vaccine uptake44. The pooled RR for multi-level interventions was 1.62 (95% CI: 1.09, 2.42) with possible high heterogeneity (I2 = 97%; P < 0.001) (Fig. 4) and very low certainty (Table 4). Among the three randomized controlled trials, one was deemed to be at low risk of bias, one had some concerns, and one was at high risk of bias (Table 2). Among the four quasi-experimental studies, two were at moderate risk of bias, and two were at serious risk of bias (Table 3). Results from the one randomized controlled trial at low risk of bias found no increase in influenza or pertussis vaccination during pregnancy after a multi-level intervention46. We observed no statistical evidence of publication bias (Egger statistic = 3.69; P-value = 0.14); however, the funnel plot indicated some asymmetry in studies with more positive findings published (Fig. S14).
Individual and pooled effect estimates for demand and supply-side interventions to increase the uptake of recommended vaccines during pregnancy. Pooled effect estimates were estimated by random effects meta-analysis. CI indicates 95% confidence intervals, and RR indicates relative risk estimates.
Ongoing studies
We identified seven ongoing registered clinical trials describing interventions to increase vaccination rates in pregnant people (Table S2). Four studies are being conducted in the US, two in Italy, and one in the Netherlands. Two target COVID-19 vaccine uptake, two target both influenza and pertussis vaccine uptake, one targets pertussis vaccine uptake exclusively, and another influenza vaccine uptake exclusively, and one targets respiratory syncytial virus vaccine uptake. Interventions focused on patient education, the AFIX framework, motivational interviewing, or tailored decision aids.
