Step 2. Analyse reporting timeliness and completeness

§ Are the data complete?

The data is not complete. A reducing trend is observed in the completeness of reporting from 100% in January 2014 to 93% in December 2014.

§ If not, which regions have completeness issues?

The regions of Center and North West are responsible for the poor completeness of reporting observed. In Center region, the reporting was 100% in January 2014 but this reduced to 80% in December 2014. Similarly in North West, reporting reduced from 100% in January to 80.8% in December 2014. Thou South West had 100% completeness through out the year, a drop-in reporting was observed in December 2014.

§ What is the implication of that for interpretation of the data?

The implication is that in making decision based on the data, caution is required in the regions where data is incomplete.

§ Are the data sent in a timely way?

The data was are not sent timely in most regions. As seen in figure below across the three years, timeliness varied between 95%, 93% and 90% in January of each year and ended at 82%, 90% and 82% for year 2014, 2013,2012 respectively.

§ If not, which regions have timeliness issues?

In the table below, average timeliness of reporting is shown. Highlighted in yellow are the regions with decreasing trend in timeliness across the 3 years which included East, North and West. All regions have timeliness issues with no region making 100% in 2014.

§ Is timeliness improving or getting worse over time?

Timeliness of reporting is getting worse over time across all regions with 2014 having the worse performance across all regions.

§ What kind of indicators or visualization did you use to analyse completeness and timeliness?

Trend line was used to analyse the completeness and timeliness of reporting. Also average annual reporting was presented in a table.

Step 3. Scan for outliers and other inconsistences

§ Can you find any obvious outliers (monthly values that seem too high or too low compared to the average)

Yes

§ For which region(s)?

Center, East, Far North, North West

§ Can you spot any major differences in numerator data for doses that are normally given at the same time, and that should be somewhat consistent?

OPV3 and PENTA3 are antigens given together. In subtracting the recorded number of children for OPV3 from PENTA3, wide discrepancy was observed which varied from month to month and region to region.

§ Can you spot any other possible mistakes?

No

§ What could be some other reasons for differences between different vaccines, apart from data quality issues?

§ What kind of indicators or visualization did you use to help you spot outliers and consistency between doses?

For Outliers, I used an excel function to determines a boundary for each year. The boundary was used to identify outliers with TRUE specified when reported number of children are out of the range. Using Conditional formatting, all the values that were true were highlighted.

For Consistency, I subtracted the number of children given PENTA 3 from number of children given OPV3. The resulting difference is the discrepancy between the two antigens administered at the same time.

Step 4. Analyse coverage trends.

Refer to tab 5, coverage.

§ Are the coverage estimates by region consistent, or are there obvious problems with the data?

The coverage figures are consistent overtime in most provinces however Center and North West reported coverages above 100%. It would be essential to compare the figures with other sources in order to validate the results.

§ There was a big drop in coverage last year: what caused this drop?

The observed drop occurred due to sudden change in the denominator used for calculating the coverage. In 2011 denominator used for calculating coverage was 455,681 which suddenly dropped to 434,507 in 2012. Sudden changes like this should be properly documented for decision makers in order to guide them aright.

Step 5. Compare your denominator to the UN Population estimate.

Refer to Tab 6

§ Which denominator estimate seems more plausible and why?

The UNDP Surviving Infant figure looks more plausible due to consistency overtime rising from 453987 to 473749. Surviving infant figures based on country figures dropped from 468,989 to 442,584 and back to 475,534. Also the increment in population figures doesn’t seem consistent overtime as well.