1. Describe data cleaning checks.
Organizing data; physically (properly labeling and putting in folders of completed forms ordered by stratum and cluster and numbering of the completed forms) similarly, proper filing of vaccine card images and keeping the file names as variable in the survey data facilitates ease of cleaning
Creating a unique identifier by combining different variables helps to uniquely identify respondents. Which in turn helps to locate the records and correct errors during cleaning, it help also identify duplicate records.
Running frequency of variables is important to see for missing value, and outliers.
Conducting logical cleaning:
If card is not available then there should not be vaccine receipt dates,
If age is collected (age in months, age in years) then age calculated at time of the survey and age in months plus age in years should be similar
The age of the child should be within the eligible age bracket of the survey dates,
Vaccine receipt date should be after child birth date, if a new vaccine was introduced like IPV, IPV receipt date should be after the national or state IPV vaccine launch date, age validity of vaccines should be checked (Penta1 at 6 weeks, measles at 9 months etc...), intervals between vaccines should be at least 28 days.
Once all cleaning codes are written, and run all records with errors will be corrected, duplicates will be deleted and those children who are not eligible for the survey will be excluded and the cleaned file will be saved with a different file name leaving the original file as it is. This will help to track and document all the changes made by reading the cleaning codes.
2. Data weighting plan
The three stages for weighting are: 1) calculating the design weight, 2) adjusting for non-response and 3) and post stratifying to match for population total.
The first and second steps are almost always necessary. To implement all the steps, proper documentation of the process of selection of the clusters, the total and selected EAs in each stratum, if there was segmentation in any of the EAs, total and selected house holds in each EAs, total respondents and total eligible in house holds in the EAs, list of missed households or absent, and refusal are required. For the third step (post stratification) to implement it is important to know if coverage will be estimated across strata and nationally, and if there was over/under sampling. Post stratification requires also the availability of accurate strata level total population.
i. In the Nigeria data the first two will be used to calculate the final weight. The design weight will be calculated 1st by calculating the probability of respondents selection. Hence the number of EAs selected will be divided by the total EAs in each stratum and if there was segmentation of EA it will be multiplied by that probability of selection of the segment (if EA is divided in to 5 equal parts, then the probability of selecting one segment will be 1/5), then the probability of respondent selection will be calculated by dividing the respondents in house holds by total eligible in House-holds.
The design weight will be inverse of the probability of selection
Weight_design=1/p_selection
ii. Adjusting for non-response
This require proper documentation of house holds listed and house holds completed, not present or refused during survey. The non-response rate is the inverse of the response rates for the different stages.
Response_ ratei=Stage1reponse rate x stage2 response rate (…)
iii. Post-stratified weights are adjusted to make the sum of weights in each stratum proportional to the known eligible population. This is important if national or across strata estimates of coverage are going to be prepared and if there is under or over sampling. Accurate population by stratum is required to apply post stratified weights.
3. Prepare table shells for five indicators
i . Table 1. Crude coverage
| Zone Name | State name | Penta3 coverage | 95%CI | DEFF | N(Unweighted) | N(Weighted) | Penta3 coverage boys | Penta3 coverage girls |
| North East | Zonal | |||||||
| North East | Adamawa | |||||||
| North East | Bauchi | |||||||
| North East | Borno | |||||||
| North East | Gombe | |||||||
| North East | Taraba | |||||||
| North East | Yobe | |||||||
| South South | Zonal | |||||||
| South South | Akwa Ibom | |||||||
| South South | Bayelsa | |||||||
| South South | Cross River | |||||||
| South South | Delta | |||||||
| South South | Edo | |||||||
| South South | River |
note: The Penta3 estimate are weighted coverages. And the source of infomration is card plus history. the 95%CI is two side CI.
MICS represent Multiple Indicator Cluster Survey (MICS) and national immunization coverage survey.
Note: point estimates wrapped in parentheses () indicated that the estimate is based on based on 25-49 respondents and asterisk (*)] indicate that the estimate was based on fewer than 25 respondents.
ii. Table 2. Dropout
| Zone name | state name | PENTA1-PENTA3 Dropout % | N(Unweighted) | Penta1-Penta3 drpop out boys | Penta1-Penta3 drpop out gilrs |
| North East | Zonal | ||||
| North East | Adamawa | ||||
| North East | Bauchi | ||||
| North East | Borno | ||||
| North East | Gombe | ||||
| North East | Taraba | ||||
| North East | Yobe | ||||
| South South | Zonal | ||||
| South South | Akwa Ibom | ||||
| South South | Bayelsa | ||||
| South South | Cross River | ||||
| South South | Delta | ||||
| South South | Edo | ||||
| South South | River |
note:The dropout rate is between penta to penta3. The source of information is card or maternal recall.
The drop out rate is calculated as per the VCQI indicator definition and the reulst are unweighted.
MICS represent Multiple Indicator Cluster Survey (MICS) and national immunization coverage survey.
Note: point estimates wrapped in parentheses () indicated that the estimate is based on based on 25-49 respondents and asterisk (*)] indicate that the estimate was based on fewer than 25 respondents.
iii. Table 3. Valid coverage
| Zone name | State name | Valid measles coverage | N(Unweighted) | valid measles coverage boys | valid measles coverage girls |
| North East | Zonal | ||||
| North East | Adamawa | ||||
| North East | Bauchi | ||||
| North East | Borno | ||||
| North East | Gombe | ||||
| North East | Taraba | ||||
| North East | Yobe | ||||
| South South | Zonal | ||||
| South South | Akwa Ibom | ||||
| South South | Bayelsa | ||||
| South South | Cross River | ||||
| South South | Delta | ||||
| South South | Edo | ||||
| South South | River |
note: this is unweighted valid measles coverage received before tha age of 1 year. The coverage is calculated among children with recorded birth date and who had documneted measles dose. Measles was considered valid when received at or after 270 days.
MICS represent Multiple Indicator Cluster Survey (MICS) and national immunization coverage survey.
Note: point estimates wrapped in parentheses () indicated that the estimate is based on based on 25-49 respondents and asterisk (*)] indicate that the estimate was based on fewer than 25 respondents.
iv. Table 4. Card availability
| Zone name | State name | card availability% | 95%CI | N(Weighted) | N(Unweighted) | card avilabiliy% boys | card avilabiliy% girls |
| North East | Zonal | ||||||
| North East | Admawa | ||||||
| North East | Bauchi | ||||||
| North East | Borno | ||||||
| North East | Gombe | ||||||
| North East | Taraba | ||||||
| North East | Yobe | ||||||
| South South | Akwa Ibom | ||||||
| South South | Bayelsa | ||||||
| South South | Cross River | ||||||
| South South | Delta | ||||||
| South South | Edo | ||||||
| South South | River |
Note:The card avilability is weighted %. Card avilable is defined when the card was avilable and there was at leaset one vaccine receipt date written in the card.
Note: point estimates wrapped in parentheses () indicated that the estimate is based on based on 25-49 respondents and asterisk (*)] indicate that the estimate was based on fewer than 25 respondents.
MICS represent Multiple Indicator Cluster Survey (MICS) and national immunization coverage survey.
v. Table 5. Fully vaccianted crude coverage
| Zone | State | % fully vaccinated | 95%CI | N(Unweighted) | N(Weighted) | Fully vaccinated% boys | Fully vaccinated% girls |
| North East | Zonal coverage | ||||||
| North East | Adamawa | ||||||
| North East | Bauchi | ||||||
| North East | Borno | ||||||
| North East | Gombe | ||||||
| North East | Taraba | ||||||
| North East | Yobe | ||||||
| South South | Zonal coverage | ||||||
| South South | Akwa Imbo | ||||||
| South South | Bayelsa | ||||||
| South South | Cross River | ||||||
| South South | Delta | ||||||
| South South | Edo | ||||||
| South South | River |
Note:Fully vaccinated is defined as a child who received at least BCG,OPV1,OPV2,OPV3,DPT1,DPT2,DPT3, and MCV1.
The reported coverage is weighted. Source of immunization is card plus maternal recall or hsitory by the time of the survey. Two side 95%CI is reported.
MICS means Multiple Indicator Cluster Survey (MICS).
Note: point estimates wrapped in parentheses () indicated that the estimate is based on based on 25-49 respondents and asterisk (*)] indicate that the estimate was based on fewer than 25 respondents.
4. Completed table (Calculate the results for one of your tables)
The selected indicator (table 1) is crude Penta3 coverage for 12-23 months old children by card or history. The results were generated for the 12 states in North East and South South zones and agregated zonal coverages for the two zones as well.
-Crude weighted Penta3 coverage by card or history by the time of the survey
-Lower Bound and Upper Bound 95% CI
-Un Weighted and weighted number of survyed children by state and zone
- Crude weighted Penta3 coverage by card or history for boys
-Crude weighted Penta3 coverage by card or history for girls
5. Table filling Syntax
Stata syntax was prepared to generate the results for the tables and the tables were copied to excel.
Syntax
********* Analysis to populate Crude penta3 coverage for children 12-23 months old surveyed in the MICS in 2016-2017*****
********* The source of data is both card or maternal recall***************
**The results are for 12 zones in the North East and South South Zones in Nigeria**
**The results included: Crude Penta3 coverage,95% CI, weighted and unweighted numbers, coverage by sex and DEFF***
use "C:\Users\Teklay\Desktop\DATA\0Nigeria\Nigeria_data_for_Scholar_course.dta", clear
***** generate the total weighted number of 12-23months old children by state: denominator which is the same as the sum of psweight*****
egen eligible_12_23months_wt =total(psweight), by (RI02)
***** generate the weighted number of 12-23months old children Penta3 vaccinated by for each vaccinated child*************************************
gen penta3_c_h_wt_vac_child = got_crude_penta3_c_or_h * psweight
*******generate weighted number of children vaccinated by state for penta3 from card or history source*******************************
egen penta3_c_h_wt_vac_by_state =total(penta3_c_h_wt_vac_child), by (RI02)
******* generate weighted penta3 coverage******
gen penta3_by_c_h_wt_cov_state = (penta3_c_h_wt_vac_by_state/eligible_12_23months_wt)*100
tabstat penta3_by_c_h_wt_cov_state, by(RI02) stat(mean)
*** setting the weight and weighted coverage******
svyset clusterid, strata(stratumid) weight(psweight)
svy: proportion got_crude_penta3_c_or_h if RI02=="Akwa Ibom"
***** for wilson confidence interval
svy: proportion got_crude_penta3_c_or_h if RI02=="Akwa Ibom",citype(wilson)
**How to calculate the confidence interval for states at the same time****
svy: proportion got_crude_penta3_c_or_h, citype(wilson) over(MICS_5_hh7)
** calculating the zonal weighted coverage for the two zones
svy: proportion got_crude_penta3_c_or_h, citype(wilson) over(level2id)
*** clculating the design effect for each of the 12 starta and two zones*****
svy: proportion got_crude_penta3_c_or_h if RI02=="Adamawa"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Akwa Ibom"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Bauchi"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Bayelsa"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Borno"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Cross River"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Delta"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Edo"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Gombe"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Rivers"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Taraba"
estat effects
svy: proportion got_crude_penta3_c_or_h if RI02=="Yobe"
estat effects
***** calculating design effects for zonal level*******
svy: proportion got_crude_penta3_c_or_h if level2name=="NORTH EAST"
estat effects
svy: proportion got_crude_penta3_c_or_h if level2name=="SOUTH SOUTH"
estat effects
****** weighted number of observations per state*******
tab got_crude_mcv_c_or_h [iweight=psweight]
****unweighted observation****
tab got_crude_mcv_c_or_h [iweight=psweight]
******weighted crude penta3 coverage card or history for boys, by state*****
svy: proportion got_crude_penta3_c_or_h if MICS_5_hl4==1, over(MICS_5_hh7)
******weighted crude penta3 coverage card or histiry for girls, by state*****
svy: proportion got_crude_penta3_c_or_h if MICS_5_hl4==2, over(MICS_5_hh7)
******weighted crude penta3 coverage card or history for boys, by zone*****
svy: proportion got_crude_penta3_c_or_h if MICS_5_hl4==1, over(level2id)
******weighted crude penta3 coverage card or history for girls, by zone*****
svy: proportion got_crude_penta3_c_or_h if MICS_5_hl4==2, over(level2id)
**** calculating unweighted number of observations per strata boys******
svyset, clear
**** calculating coverage by sex, education, wealth quantile, ethnicity, and state for Crude penta1, penta3 and measles by card plus history****
******************** Crude Penta1 card plus history************************
*** calculating penta1 card plus history crud coverage by state****
svy: proportion got_crude_penta1_c_or_h, citype(wilson) over(MICS_5_hh7)
*** calculating penta1 card plus history crud coverage by sex****
svy: proportion got_crude_penta1_c_or_h, citype(wilson) over(MICS_5_hl4)
*****calculating penta1 card plus history crud coverage by ethncity****
svy: proportion got_crude_penta1_c_or_h, citype(wilson) over(MICS_5_ethnicity)
*****calculating penta1 card plus history crud coverage by education****
svy: proportion got_crude_penta1_c_or_h, citype(wilson) over(MICS_5_ed4a)
*****calculating penta1 card plus history crud coverage by wealth quantile****
svy: proportion got_crude_penta1_c_or_h, citype(wilson) over(MICS_5_windex5)
***** calculating crude penta3 card plus history coverage **************************
*** calculating penta3 card plus history crud coverage by state****
svy: proportion got_crude_penta3_c_or_h, citype(wilson) over(MICS_5_hh7)
*** calcualting penta3 card plus history crud coverage by sex****
svy: proportion got_crude_penta3_c_or_h, citype(wilson) over(MICS_5_hl4)
*****calculating penta3 card plus history crud coverage by ethncity****
svy: proportion got_crude_penta3_c_or_h, citype(wilson) over(MICS_5_ethnicity)
*****calculating penta3 card plus history crud coverage by education****
svy: proportion got_crude_penta3_c_or_h, citype(wilson) over(MICS_5_ed4a)
*****calculating penta3 card plus history crud coverage by wealth quantile****
svy: proportion got_crude_penta3_c_or_h, citype(wilson) over(MICS_5_windex5)
***** calculating crude MCV coverage by card plus history*****
**************************************************************
***** calculating measles card plus history crud coverage by state****
svy: proportion got_crude_mcv_c_or_h, citype(wilson) over(MICS_5_hh7)
*** calculating mcv1 card plus history crud coverage by sex****
svy: proportion got_crude_mcv_c_or_h, citype(wilson) over(MICS_5_hl4)
****calculating mcv card plus history crud coverage by ethncity***
svy: proportion got_crude_mcv_c_or_h, citype(wilson) over(MICS_5_ethnicity)
*****calculating mcv card plus history crud coverage by education****
svy: proportion got_crude_mcv_c_or_h, citype(wilson) over(MICS_5_ed4a)
*****calculating mcv card plus history crud coverage by wealth quantile****
svy: proportion got_crude_mcv_c_or_h, citype(wilson) over(MICS_5_windex5)
6. Graphical summary of Crude Penta3 vaccination coverage by state
The crude Penta3 coverage by the time of the survey by card plus maternal recall with errors bars is given in fig 1.
7.Summarize methods, results, strengths and weakness
i. summary of methods
Stata version 15 was used for analysis. The number of clusters, households and children by sex and state, and the characteristics of caretakers in terms education, ethnicity and wealth quantile, and card receipt and retention was described. Source of immunization and reasons for not being vaccinated were grouped into 5 categories and tabulated aginest states to see dominant reasons by state(Table-4 in annex section). The The weighted coverage and confidence interval were calculated for crude penta1, penta3 and MCV1 with 95% confidence interval.
ii. Summary of results
There were 655 clusters, 1649 households and 1728 children surveyed in the 12 states in North East and South South zones. Most 1573 (95.4%) of the households had on 12-23 months old child surveyed,73(4.4%) had two and the remaining 3 households had 3 children. In terms of education, most (53.14) of the care takers had secondary/ secondary-technical education , followed by primary education (24.16%),higher education (11.92% ) and no formal education (10.78%). In terms of ethnicity, 35.88% were Hausa, followed by Igbo (4.51%),Yoruba(.75%) and the 58.85% of the caretakers had were from other ethnic groups. In terms of wealth index quantile, Poorest (21.35%),Second (19.21%), Middle (22.34%), Fourth(21.24%) and Richest were (15.86%), Table-2.
Of all the surveyed children 1025(59%) had ever received immunization card and out of those received 644(63%) was available during the survey. Most of the children received immunization service in governmental health centers(61%) followed by governmental hospitals (28%), governmental mobile or outreach services (8%). Private facilities including NGO accounted for about 4% of immunization service delivery. There was variation in place of immunization service receipt by states. In Edo private facilities and NGO contribution was relatively higher 15%, on the other-hand in Borno governmental hospital were the dominate source of immunization and the contribution of governmental mobile or outreach services was significantly high (24%), table 2.
Coverage of Penta1, Penta3 and measles by card plus history by the time of the survey by ethnicity, sex, education, wealth index quantile, state are summarized in (Table-3). Generally coverage are higher in South South compared to the North East.
Access to immunization
Access to immunization (as measured by crude penta1 coverage) is generally low in the North East zone 46.7%(95%CI:39.0-54.7%) compared to 79.9%(95%CI:75.9-83.4%) in South South zone. Among North East States, all states have poor access (below 80%), Borno state has relatively better access 72.9%(95%CI:57.7-84.1%) and Yobe state has very low access 17.5%(95%CI:10.5-27.7%). On the other hand access was good(above 80%) in 4 of the six states in South South zone, and below 80% in Bayelsa and Delta states.
Immunization utilization
The Penta1 to Penta3 dropout was generally high or above 30% in all six states in North East zones in particular the Penta dropout was highest in Taraba state (54%) followed by Yobe state (50%).
However, Penta dropout was below 30% in states in South South zone except for Bayelsa state where the dropout is 30.5%.
Most states in the North East have both access and utilization problem. On the other hand the problem in the South South seems more of utilization problem.
There is no significant difference in immunization coverage by sex between boys and girls. Immunization coverage as measured by crude penta3 is highest among women with higher education, 75.4%(95%C:62.5-85.0%) and very low among those with non-formal education, 17.1%(10.1-27.4%). Similarly, crude penta3 coverage was highest among those children from richest families, 63.9%(95%CI:55.5-71.6%) compared to those children from poorest families (10.5%(95%CI:7.3-15.0%).
Reasons for not being vaccinaed is described in annex (Table 4): service delivery obstacles are dominant reasons for not being vaccinated in Gombe(32%),Taraba(32%) and Yobe (29%), while lack of faith in immunization or concern on side effects of immunization are dominant cause for not being vaccinated children in Bauchi state(41%).
iii. sumarry strengthes and weaknesses
Strengths
The sampling method employed probability sampling and all the variables as per the VQCI are captured in the data collection. Employing tablets for data collection can be considered a strength as it will reduce the time that would have taken for data entry. The survey followed the variables and definitions of (standards) that are described in the WHO document entitled “VCQI Indicator List with Specifications - v1.9”. This eases the analysis and helps to calculate standard indicators which make it understandable and comparable with other survyes. The data were collected by a Multiple Indicator Cluster Survey (MICS) team, so the quality of household listing is expected to be high. The outcome for every household that was selected was know which is very useful for weighting and adjusting for non response.
Weakness
As population for groups such as sex in each state is not given, post stratification can not be done. No visits were made to health facilities which could have increased the vaccine receipt dates availability rate. Additional survey on service delivery frequency by facility type, cold chain functionality, EPI service interruption and underlying causes could have been added to the survey which could help explain to explain the differences of coverage in the different states.
Annexes
Table 2. The profile summary of the 12 states in terms of clusters, households and children 12-23 months survyed, card avilability and source of immunization among vaccinated children.
Tabl-3, describes estimate crude coverage with 95% confidence interval of Pent1, Penta3 and measles by state, sex, caretakers education and wealth quantile.
Table 4.describes, reasons for not being vaccinated in a children were grouped into five groups as follows:
i. EPI schedule awareness (22a: thought the child was fully vaccinated, 22b: unawareness of need for immunization 22c: unaware of need to return for 2nd or 3d dose)
ii. Family problems related (22e:post pond for other time, 22f:care taker/mother busy or 22g:family problem)
iii. Faith/concern ( 22h: no faith in immunization, 22i:fear of side effects, 22j:myths/rumours,22k: believe there were contraindication)
iv. Obstacles: (22l:place of immunization too far,22m: time of immunization inconvenient, 22n: vaccinator absent, 22o: vaccine not available, 22p:long waiting time)
v. The remining reasons were grouped as others
Table 4. Reasons for not being vaccinated
