Study setting
The small nation of Malawi is part of the sub-Sahara of Africa and is a landlocked country located in the southeastern part of the continent. It is bordered to the north and northeast by the United Republic of Tanzania; to the east, south, and southwest by the People’s Republic of Mozambique; and to the west and northwest by the Republic of Zambia [18][19][20]. Malawi is one of the poorest countries in the world with GNI per capita of US $320 and about 85% of the population live in rural areas [21]. The economy of Malawi is based primarily on agriculture, which accounts for 30 percent of the gross domestic product (GDP) [22]. Health care services in Malawi are provided through both the public and the private sectors [23]. The public sector includes all facilities under the Ministry of Health (MoH), Ministry of Local Government and Rural Development, the Ministry of Forestry, the Police, the Prisons and the Army. The private sector consists of private for-profit and private not for profit providers mainly Christian Association of Malawi (CHAM) [11][23]. The public sector provides services free of charge while the private sector charges user fees for its services. There are currently 977 health facilities in Malawi comprising 113 hospitals, 466 health Centres, 48 dispensaries, 327 clinics, and 23 health posts. These health facilities are managed by the government (472), CHAM (163), Private (214) and Non-government Organizations (NGOs) (58) and company (69) [5]. All these forms of institutions provide immunization services in Malawi.
Data sources
The current study analyzed the data obtained from the Malawi Demographic and Health Survey (MDHS) 2004, 2010, and 2015-16. The methodology used in these surveys can be obtained in detail elsewhere [10][24][25]. In brief, the surveys employed a two-stage sampling designed to produce nationally representative samples. The surveys utilized sampling frames from the Malawi Population and Housing Census (MPHC) conducted in 1998 and 2008. The first stage selected 850, 849, and 522 clusters also known as standard enumeration areas (SEAs) proportional to population in 2015-16, 2010, and 2004 respectively. The second stage involved selection of 27,516, 27,307, and 15,041 households from the SEAs with an equal probability systematic selection in 2015-16, 2010, and 2004 respectively.
Data collection
Using face-to-face interviews, data were collected on the demographic, social, and economic characteristic of the respondents and their households. Respondents were asked to show a health passport or any other document where (NAME)’s of vaccines were written down. If the respondents could not show a health or immunization card, they were then asked to recall any vaccinations ever received to prevent (NAME) from getting diseases, including vaccinations received in campaigns or immunization days or child health days. Particularly, respondents were asked to report whether the (NAME) received BCG, polio, pentavalent, rotavirus, pneumococcal and measles vaccines. As regards polio, pentavalent, rotavirus, and pneumococcal vaccines, respondents were further asked to report a number of times the (NAME) received each specific vaccine [10][24][25]. Of the selected households, 11,698 in 2004, 23,020 in 2010 and 24,562 in 2015, women were interviewed representing 97.7%, 96.9%, and 95.7% respectively. As recommended by the WHO [10], the present study included children aged 12–23 months. Thus, the final samples analyzed were 2,211, 3,741, and 3,225 children in 2004, 2010, and 2015-16 respectively.
Inclusion criteria
All live children aged 12–23 months prior to each survey, living with their guardians and had information on immunization were included in this study.
Bottle neck analysis
Bottleneck analysis (BNA) is an approach based on Monitoring of Results for Equity System (MoRES) for planning equity-focused interventions and identifying bottlenecks in their uptake [26]. The MoRES was developed in 2010 as part of UNICEF’s refocus on equity to ensure that UNICEF is as effective as possible in the protection and promotion of children’s rights [27]. The BNA framework is premised on the notion that effective coverage of services is influenced by four main domains namely: supply, demand, quality, and environment. (1) Supply determinants of services are predominantly controlled by the health care delivery system and have three important components: commodities, human resources, and geographic access. (2) Demand determinants of services are predominantly controlled by the community, and have two important components known as initial utilization and continuous utilization of services (3) Quality determinants of services are predominantly controlled by the health care delivery system and relate to the services being able to meet the quality standards set within national guidelines [13]. However, in this study, we focused on the demand and quality determinants of services since the DHS does not have data on supply demands.
Variables and operational definitions
Immunization coverage
Immunization coverage is the proportion of children aged 12–23 month who received the recommended EPI vaccine antigens compared to the total number of infants who survived in the given target population. The immunization coverage rate is measured by comparing the number of antigens actually administered versus the total number of infants who survived in the given target population.
Partially immunized
Partially immunized children is defined as any child aged 12–23 months who missed some of the scheduled prescribed vaccines antigens considered to protect them against vaccine-preventable diseases.
Unimmunized
Unimmunized children are children who have not defaulted any of the scheduled EPI vaccinations. Usually unimmunized children is considered to be those aged 12–23 months and have not received DTP3.
Dropout rate
The dropout rate is calculated by comparing the number of infants who initiate the vaccination schedule against those that complete it and usually two domains are habitually used to calculate the dropout rate. These measures are the Penta vaccine and MCV1. Specifically, the dropout rate can be calculated subtracting children who received Penta1 from those who received Penta3 then divide by those who received Penta1 (Penta1–Penta3) ÷ Penta1 x 100%). Also it can be can be calculated between the children who received Penta1 and MCV1 divide by those who received Penta1 (Penta1–measles) ÷ Penta1 x 100%). The WHO recommends that the coverage of both the Penta1 to Penta3 and that of Penta1to MCV1dropout rates should be less than 10% so as to have the better-quality of immunization coverage as well as to have the reduced rates under-five morbidity and mortality [28][29]. It should be taken into consideration that the dropout rate of more than 10% reflect underutilization of immunization services.
Initial Utilization
An initial utilization of an immunization program is define as the proportion of children who received Penta1 vaccine during the past year in region/district. The numerator for this indicator is the number of children aged 12–23 months who received either BCG or Penta1 vaccine while the denominator is the number of children under 12–23 months eligible for Pentavalent 1 vaccination.
Continuous Utilization
The continuous utilization of an immunization program is defined as the proportion of children who received Penta3 vaccine during the past year in region/district. This indicator uses the number of children received Penta 3 vaccine as it numerator and the number of children under 12–23 months eligible for Penta 3 vaccination the denominator.
Adequate coverage
Adequate immunization coverage is defined as the percentage of children aged 12–23 months who were immunized with MR1 during the past year in region/district. The indicator is calculated by diving the number of children received MR1 vaccine by the number of children under 12–23 months eligible for MR1 vaccination.
Fully immunized coverage (FIC)
The FIC is defined children aged 12–23 months who BCG, OPV3, Penta3, PCV3, Rota2 and MCV1 vaccines. The number of children fully vaccinated by 12–23 months according to the vaccination calendar timeline is considered as the numerator and the number of children under 12 eligible for full vaccination as the denominator.
Control variables
The control variables used in the present study included the year in which the survey was carried out and geographical region. In order to establish a trend in immunization coverage a 15 years period was sampled, the years included 2004, 2010, and 2015-16. The geographical region included northern, central, and southern and was used a proxy for administrative divisions in Malawi. The geographical region was chosen so as to establish the most underperforming area, thus for better policy implications.
Statistical analysis
Data were analyzed using SAS software version 9.4 (SAS Institute Inc., Cary, NC, USA) and Stata version 15 (Stata Corp, College Station, TX, USA). All analyses were performed separately for 2004, 2010, and 2015-16. Data were presented as frequency and percentages, and where necessary data was presented in the form of charts. Using Pearson’s χ2, the bivariate analysis was performed to test the differences in distribution between groups (Penta1 vs Penta3, Penta1 vs MCV1, initial utilization [yes/no,], continued utilization [yes/no], adequate coverage [yes/no) and FVC [yes/no]). The univariate analyses were conducted using binary logistic regression to examine the magnitude of unimmunized with Penta3 and MCV1. The results of the univariate logistic regression were presented as the odds ratio (OR) with their corresponding 95% CI. The statistical significance was considered when p-values were < 0.05.
Ethics statement
The protocols for 2004, 2010, and 2015-16 MDHS were reviewed and approved by the Malawi National Health Sciences Research Committee (NHSRC), the Institutional Review Board of International Classification Function (ICF) Macro, and the Centers for Disease Control (CDC) in Atlanta. Data collection was implemented by the National Statistics Office (NSO). At the beginning of each interview informed consent was obtained from the participants. The authors sought permission from the DHS program for the use of the data. The data obtained from respondents were anonymous as names not written down thus ethics approval for this study was not required.