Door – to – door immunization strategy for improving access and utilization of immunization Services in Hard-to-Reach Areas: a case of Migori County, Kenya

Migori County is found to the southwestern part of Kenya and borders Homa Bay County (North), Kisii County (North East), Narok (South East), Tanzania (West and South) and Lake Victoria to the West. The county also borders Uganda via Migingo islands in Lake Victoria. It is divided into eight subcounties. It has an area of 2586.4 km² with an estimated population of 1,119,184 in 2018 [26]. It has a population density of 353 per square kilometer and 43% of the population live below the poverty line [27]. The main economic activities include agriculture, fishing, manufacturing and small – scale mining. The county has both rural and urban populations. Poor road network within the county limits access & utilization of health services especially during rainy seasons. The population around the Tanzania and Narok borders, fishing zones on the shores of Lake Victoria and the goldmines is migratory in nature depending on the existing cash flow and economic gains for daily living thereby playing disrupting continued uptake and utilization of health services including immunization. The county also has the Roho and Legio Maria religious sects that discourage use of health services (Fig. 1).

This was a cross – sectional review of District Health Information System 2 (DHIS2) immunization data for July and August 2018 for Migori County. During the presidential immunization RRI in July 2018, poorly performing wards/facilities were mapped using the Quantum Geographical Information System (QGIS) application that supports viewing, editing and analysis of geospatial data for the two study periods (Fig. 2). The unreached rural-urban and socio – economic hard –to – reach populations were identified. A total of 64 health facilities with over 100 unimmunized children each between January 2017 and June 2018 from all the eight subcounties were identified. In July, routine fixed-point immunization services were provided whereas in August, fixed – point plus a 3 – day enhanced door – to – door defaulter tracing by community health volunteers (CHVs) and household immunization by nurses was conducted in the 64 health facilities (out of 185 immunizing facilities) with high burden of unimmunized children. Each health facility was allocated three CHVs under the supervision of the community health assistant (CHA) and one frontline healthcare worker who was the vaccinator. The exercise was supervised by the subcounties’ health management teams (subcounty EPI coordinator and subcounty community strategy focal person). Overall supervision and coordination of logistics was provided by the county EPI logistics team. Importantly, the county, subcounties and the health facilities provide daily immunization services and had adequate stocks of routine vaccines both before and during the mobile immunization period. Afya Halisi, a 5-year USAID funded project facilitated the ground logistics – transportation, lunches and coordination airtime for the county and subcounty teams and lunches and transportation for the healthcare workers and the CHVs during the mobile exercise.

The select facilities generated the immunization defaulters lists from the Immunization Permanent Registers (MOH 510) and the immunization diaries. These lists were assigned to the CHVs from the catchment areas who mapped out the community based on the unvaccinated children’s burden. Community mobilization for the immunization services was conducted by the respective CHVs who provided households with information on benefits of starting and completing all the immunization services as per the prescribed MOH schedule. Together with the vaccinator, the team moved from house to house, scrutinized Mother Child Health (MCH) records (MOH 216) or the immunization card, BCG scar mark as well as mother’s or guardian’s verbal verification to identify the unimmunized children. The CHVs had earlier received basic orientation on the MCH records. Importantly, in areas considered vast, the team of CHVs mobilized all the eligible unvaccinated children to central points where the vaccinator would visit together with the CHA and vaccinate the children. To maintain the cold – chain of the antigens, the vaccinators carried the vaccines in cold boxes and vaccine carriers each day. After vaccination, the vaccinator and the CHA would tally the immunization tally sheet (MOH 702), update the MCH records, counsel the client on the importance of scheduled immunizations and book for the next appointment for the subsequent follow – ups/immunizations. The CHVs and the vaccinators updated their immunization defaulters’ records for reconciling/updating of the immunization permanent registers daily during the exercise. At the end of the exercise, the vaccinators updated the daily immunization data in the Immunization Services Uptake Summary (MOH 710) both the static/fixed point immunization services at the facility and the mobile immunization services. Data from 64 health facilities in Migori County were summarized and routinely entered in the national District Health Information System (DHIS2) by the subcounty health records officers. Technical assistance was provided to the health records team to ensure timely, complete, and accurate reporting of immunization indicators.

Immunization coverage as per the immunization schedule at birth – 2 weeks, 6, 10 & 14 weeks, 9 months and at 18 months were computed as follows: birth – 2 weeks, BCG and OPV 0; 6 weeks – Pentavalent 1; 10 weeks – Pentavalent 2; 14 weeks – Pentavalent 3; 9 months – Measles/Rubella 1; proportion of fully immunized children at 1 year and at 18 months – Measles/Rubella 2. Drop – out rates for Pentavalent 3 and Measles 1 were also computed for utilization of immunization services. It is assumed that with adequate stocks for all antigens, each child receives all the due antigens at a given immunization visit as per the national immunization schedule.

Immunization data for July and August 2018 were extracted from the DHIS2 and entered in Microsoft Office Excel 2013 for data cleaning. Cleaned data was then exported to STATA version 12 for analysis. Differences in proportions of the coverages and drop – outs for the two time periods were compared through z – tests (for the 64 facilities)/paired t – tests (for the 8 subcounties) and MOH summary data and program reports were also reviewed to triangulate the information obtained from the DHIS2. Drop-out rates were calculated as follows: (i) Penta1 to Penta3 dropout rate = (Penta1 – Penta3) ÷ Penta1 × 100%, where: Penta1 is the number (or percentage) receiving the first pentavalent vaccine dose and Penta 3 is the number (or percentage) receiving the third dose; (ii) Penta1 to Measles dropout rate = (Penta1 – Measles) ÷ Penta1 × 100%, where: Penta1 is the number (or percentage) receiving the first pentavalent vaccine dose and Measles is the number (or percentage) receiving the Measles dose [28]. Our study was not only interested in the significance test alone. The use of the parametric tests allowed us to acknowledge the special population from which the samples came, and this is best done with estimates of parameters and confidence intervals [29]. Confidence intervals were calculated at the 95% level and P – values ≤0.05 were considered statistically significant.

Overall, there was a marked improvement in the coverage by facilities in the number of unimmunized children compared to the previous routine fixed immunization service delivery (Fig. 3).

The findings indicate an increase in the total number of children immunized per antigen and by age from July to August (facilities: 10744 vs 14,809; subcounties: 20543 vs 25,550) as indicated in Table 2 below. At the facility level, the increase was more marked in children reached over the age of 1 year for BCG (19 times), Pentavalent 1 (22 times), Pentavalent 3 (19 times), MR1 (5 times) and MR2 (9 times) antigens as per the recommended immunization schedule. About 7 % (296 of 4396) of the children who received MR1 were above the age of 1 year. Similarly, about 30% (1211 of 4079) of the children received MR2 beyond 2 years of age. At the subcounty level, similar trends were observed as follows: Pentavalent 1 > 1 year (22 times), Pentavalent 3 > 1 year (11 times) and MR2 > 2 years (6 times) (Table 1).

Vaccination coverage by antigen for the facilities indicated that OPV 0 and BCG had lower coverages (less than 90%) of the expected per antigen for the two reference periods (OPV 0: 75.67 vs 84.76 and BCG: 74.44 vs 89.94). Vaccine coverages significantly increased from July to August for OPV 0 (75.6–84.8%, P = 0.0008, CI = 3.57–14.43), BCG (74.4–89.9%, P = 0.0001, CI = 8.01–14.98), Pentavalent 3 (92.4–112.0%, P = 0.0001, CI = 9.96–29.52) and Measles/Rubella 1 (81.7–111.5%, P < 0.0001, CI = 19.97–39.61). The proportion of children fully immunized by 1 year of age increased significantly from 78.56 to 103.86 (P < 0.0001, CI = 16.09–34.5). Drop – out rates for Penta 3 and Measles/Rubella 1 decreased significantly (P < 0.05) (Table 2).

The 64 facilities with the high numbers of unimmunized children had a net effect on the overall vaccination coverage and drop – outs per antigen at the subcounties’ level. All the vaccine coverages for OPV 0, BCG, Pentavalent 1, Pentavalent 3 and Measles/Rubella 1 were above 90% in August with the highest gains observed in Pentavalent 3 and Measles/Rubella 1. Drop – outs in Pentavalent 3 and Measles/Rubella 1 were all less than 10% (Pentavalent 3 drop – out: 3.97 vs − 5.28, P = 0.0084, CI = − 17.00 - -2.3) and MR1 drop - out: 8.02 vs − 7.82, P = 0.0006, CI = − 25.52 - -4.75) (Table 3).

Overall, the month of August registered the highest coverage with at least 90% coverage for all antigens and above 95% for MR1 as recommended (Table 4).