Table 2 Characteristics of included studies.

Loevinsohn 1987 [19]

Santa Rosa del Penon in the pacific northwest of Nicaragua

Mass vaccination campaigns; stationary clinics and mobile clinics with or without food supplementation as an incentive.

1) Measurement of attendance rates at mobile and stationary well child clinics.

2) Repeated surveys of clinic attendance; little detail of data collection or quality control; study used attendance rather than vaccination status as the outcome; lack of contemporaneous control group makes order effects possible.

Attendance at clinics; personnel time.

Regular mobile N = 425, 63.3%

Mass vaccination campaigns N= 889, 77,1%

Stationary with food N = 764, 94.1%

Mobile with food N = 547, 99.2%

Stationary clinics took up half as much healthy worker time as mobile clinics Person hours per village served MVC 8.5; Mobile clinics 38;Stationary clinics 19.

Attendance declined in a linear fashion with distance from stationary clinics.

Cutts 1990 [20]

Mothers from Mozambique Maputo 210 children aged 12-23 months

Comprehensive and integrated intervention: Outreach teams visited in 3 consecutive monthly 'pulses'; communications system to inform villages about arrival of mobile teams. Training of representatives from grass roots organisations (ten-family leaders); development of community-based volunteers from grass roots organizations; Door-to-door canvassing.

Health departments and Executive Councils formed intersect oral communities to organize activities with emphasis on mobilization

1) Cumulative BCG vaccination rates in the pulse project districts in 1984 (pre) and 1987 (post).

2) Difficult to apply rigorous study methods in a war zone; Post survey used EPI cluster sampling method; vaccination rates obtained from health cards; lack of contemporaneous control group makes order effects possible.

Comparison of coverage before and after program acceleration.

Measles vaccination 1985-1987

Beira (1% increase)

Inhambane (13% increase)

Tete (22% increase)

Quelimane (31% increase).

Zimicki 1994 [16]

Philippines Pilot in Manila (1988) and nationally (1990) Mothers or permanent carers with children under 2 years

The mass-media element of the campaign was March-Sept 1990.

It focussed on measles as a way of bringing mothers to health centres, mainly in urban areas.

Four television and four radio advertisements were broadcast, and advertisements were printed in newspapers reminding people of vaccination day. Concentrated on the dangers of measles Other promotional materials included posters and bunting.

1) Two surveys of the carers of children aged < 2 years to measure a change in knowledge) and vaccination rates. A pre-post study of 60 health centres in the same areas.

2) National campaign, so parallel control group not possible; used a rigorous multi-stage cluster sampling with weighted analysis adjusted for clustering to standardise the 1990 sample to 1989 sample; lack of contemporaneous control

group makes order effects possible.

12-23 month vaccine coverage (all 8 vaccines). 2-8 month vaccine coverage (at least 4 vaccines).

Vaccination started on time and finished on time.

Mean number of vaccinations:

Rate difference

53.6% (1989) to 64.5% (1990) (RD 10.9 (2.8-19.0)

-starting on time (12.3 (1.5-23.1)

-finished on time (24.0 (12.2-35.8)

-appropriate early (8.5 (0-17.1)

Exposure to the mass-media campaign and knowledge level

RD2.32 (2.19-2.46)

Increase knowledge increase vaccination (1989-1990)

Absolute diff 0.77 (P < 0.0001).

Brugha 1996 [22]

Three towns in Eastern region of Ghana

Program of home visits during which

parents or carers were advised to take the children to the next under five's clinic of their choice, and were

given a referral note for the clinic.

The intervention targeted parents of unvaccinated children. Up to 3 additional visits by a nurse over the next 6 months if the child did not complete vaccination.

1)Cluster randomised trial was conducted in the largest of the three towns.

2) Quality rating in Table 3.

Completed vaccination rates before and after the intervention using

Road to Health cards, clinic records supplemented by maternal history.

Vaccination coverage rose from 59.5% to 86% in the intervention group compared with 60.7% to 66.7% in the control group. The difference in the increases in the intervention and control groups was statistically significant (P < 0.005).

Vaccine coverage also rose in the other two towns that were also subject to the intervention but did not participate in the randomised trial.

Tulchinsky 1997 [21]

Communities in Hebron, the West Bank. 69 villages in Hebron and 20 in other areas

Village Health Rooms (VHR) implemented by village leaders.

Each village health room is staffed by a female village guide selected by the village leaders and health office. The guide arranges mothers' visits. They also organize national vaccination days and have a teaching as well as a service role.

1) Data from individual patient records; household surveys to determine community basic demographic information and immunisation status.

2) Data collection based on secure records but lack of contemporaneous control group makes order effects possible.

Coverage utilization and improved health status; costs and program longevity

Coverage compared VHR with baseline data from the village household survey. 90% of children up to the age of 2 years had received measles or MMR and 96% had 3 or 4 doses of DPT.

There was little changed from previous estimates but in the past children generally had vaccination delayed into the third year of life.

Amin 1997 [18]

Villages from 5 NGO regions in Bangladesh. 3,564 married women under the age of 50

NGO provision of small collateral-free area focused credit.

1) Cross sectional survey Multistage cluster sampling strategy; recruitment from villages where NGOs maintained rural credit programs and control areas where NGOs had no presence.

2) Parallel study of intervention and control areas; able to study credit recipients and non-recipients in program areas. Multivariate analyses to adjust for potential confounders.

Vaccination status amongst loanees and non-loanees from credit and non-credit areas

Mean age 29, av 3.1 children per household

Last born vaccinated

Total area 62.4%

Credit program 67.8%

non-member 58.8%

Comparison area 49.4%

Last born less than one

Total area 66.7%

Credit area 71.8%

non-member 63.3%

Comparison area 50.7%

Infant mortality rates were lower in members of credit schemes in program areas.

Hutchinson 2006 [17]

People living in rural areas of Bangladesh

'Smiling Sun' communication program included a variety of important health-related messages. The delivery media included signboards, television drama series, television advertisements, radio spots, press ads in newspapers and local publicity.

Messages related to maternal and child health, family planning and communicable disease control (incl vaccination).

1) Cross sectional survey using two-stage cluster sampling; correlation between exposure to the campaign and reported vaccination status was calculated. Extensive costing data collected.

2) National campaign, so parallel control group not possible. Used bivariate probit likelihood method to estimate program effectiveness.

Self-reported exposure

to the 'Shining Sun' media campaigns and simultaneous self-reporting of key health-related behaviours (use of ante-natal care and use of childhood vaccinations).

Cost-effectiveness analyses.

Mothers who recalled seeing Smiling Sun promotional material were more likely than those who did not to complete DPT vaccination (64% vs 48%).

Marginal effectiveness remained positive after adjustment.

Cost-effectiveness:

National level data:

$0.30/additional child vaccinated for measles and $0.36/additional child with DPT3.

Local promotional activities:

$32/additional child vaccinated for measles and $37/additional child vaccinated with DPT3.

Andersson 2009 [6]

Lasbela Pakistan parents of 12-23 months old

Three structured discussions with one in every ten thousand respondents.

1) Discussion showed findings about vaccine uptake from baseline survey.

2) Focused on the costs and benefits of childhood vaccination.

3) Focused on local action plans, including options for sharing transport and childhood costs.

1) Cluster randomised controlled trial.

2) Quality rating in Table 3.

Uptake of measles and full DPT vaccination.

Measles

OR 2.20 (95% CI 1.2-3.88)

DPT

OR 3.36 (95% CI 2.03-5.56).