RE: Resolving the Denominator Dilemma within the Ghana Health Service: Response from Zambia
Puta, Chilunga
Comments by MR. Kaliki, Ministry of Health, Zambia
Well noted.
As you may be aware, the main source of health statistics the health sector uses to monitor and evaluate the implementation of various health sector programs such as the National Health Strategic Plan (NHSP) and Millennium Development Goals (MDGs) health targets, is the Health Management Information System (HMIS).
However, there have been concerns like you rightly pointed out on the high coverage of service statistics (e.g. immunization and antenatal coverage), for some districts which happen to go beyond 100%. Anecdotal evidence on the performance of these indicators has reviewed that this situation may be attributed to the following factors:
I. The ratio for target population (i.e. under ones, under-fives, expected deliveries and pregnancies), has not been reviewed and has remained constant since 1998;
II. The estimated catchment population of some health facilities is under estimated partly due to item (I) highlighted above;
III. Fragmented criteria and/or standard applied in the establishment of catchment population for health facilities;
IV. An overlap of some health facility catchment population with the Census Supervisory Areas (CSA) which distort the actual facility population.
Therefore, in order to validate the above noted observations and effectively address the challenges the Ministry faces in using these service statistics, for informed decision making, there is need to conduct a methodical assessment that would streamline and harmonise head count catchment populations being used by most health facilities with the official Census Population figures, particularly in districts with coverage of service indicators of over 100%.
The assessment was conducted in four sampled districts of the Western and Copperbelt based on the following criteria:
(i) A border district with high population dynamics due to immigration and emigration influences;
(ii) A district with rural population aspect which is static or stable;
(iii) A district with very low or less external influence in terms of movement of the people;
(iv) Performance of some indicators particularly immunization and antenatal coverage( i.e. consistently high coverage and consistently low coverage).
The results of this small assessment showed that Chingola, Kalulushi, Luashya, Lufwanyama, Masaiti and Mufulira districts in Copperbelt Province, had negative variances implying that these districts have the highest proportion of head counts compared to the census figures. Similarly, Kalabo, Senanga and Sesheke districts in Western Province had negative variances implying that these districts have higher proportion of head counts than census figures
To address this long standing problem the following steps need to be considered:
(i) There is need to revise the coding system in the data collection tools particularly the registers so that the actual distance of each client is reflected;
(ii) There is need to revise the estimated proportions for target population (i.e. under ones, under- fives, expected deliveries and pregnancies), which has not been reviewed and has remained constant since 1998;
(iii) There is need to harmonise the current overlap of some facilities with the Census Supervisory Areas (CSAs) which happen to distort the actual facility population;
(iv) There is need to come up with standard criteria in the establishment of catchment population for health facilities.
CSO may not entirely be blamed for the over 100% coverage figures being noted in the system but we need to address the reasons outlined above (refer to attached report)
I thank you and look forward to your comments and possible discussion to address this long standing issue…
From: bidini...@googlegroups.com [mailto:bidini...@googlegroups.com]
On Behalf Of Puta, Chilunga
Sent: 24 February 2015 14:35
To: bidini...@googlegroups.com
Cc: dominic k atweam
Subject: Resolving the Denominator Dilemma within the Ghana Health Service
Resolving the Denominator Dilemma within the Ghana Health Service
The generation of some of the service indicators depends on the use of population data. This population data is derived from the census published by the Ghana Statistical Service (GSS). The age groups population data from the population census is often used as the denominator in generating some of the service coverage indicators.
From the 2000 census and now the latest 2010 census for Ghana, the use of the age group segments as provided by the GSS results in generation of coverage often above 100% especially for the EPI.
To address this anomaly the EPI programme from data gathered from the various NIDs started using 4% as the proportion for the children less than one year and by extension expected pregnancy for all the Regions to generate the coverage indicators. This results in acceptable coverage figures below 100%. It however also results in some very low figures for some districts whose under one population was far below the 4% that was used.
From the recent census of Ghana, the average proportion of children less than one year out of the total population was given as 3.1%. It however ranges from 2.7 to 3.2. Using these proportions for children less than one year give Penta 3 coverage ranging from 121% to 96%, although in the various districts you will from coverage surveys find children who have not been vaccinated.
From the recent NID, (2010 and 2014) the proportions for the regions are higher than from the 2010 census, using these proportions also give more realistic coverage.
A decision was made by GHS to use the 4% and going forward the Ghana Health Service-Policy Planning Monitoring and Evaluation Division is implementing the DHIS2 e-Tracker as a child (immunization and growth promotion) and mother (ANC delivery, PNC and FP) registry for Child and Maternal Health Services. The e-Tracker does the following and is currently deployed in some districts in Ghana:
· Collect transactional data - set up automated aggregation queries - populate the aggregated data warehouse directly - all in one system!
· Enrol individuals into longitudinal and chronic programs - schedule visits - set up automated SMS reminders - track missed appointments - improve retention.
· Define your own programs with stages - decide what to collect at each stage - all through the user interface.
· Generate daily or weekly visit schedules (work plans) for your facility or community health workers.
· Tools for tracking and following up children and mothers who do not come to scheduled visits.
· Set up detailed maternal or neonatal death audits - analyse your data using the tabular reports with both case-based data and ad-hoc aggregation.
· Collect data using mobile phones - online in web browser or offline with java clients
The results we are getting are interesting as children registered so far in the phase one districts are more than what the Ghana Statistical Service have published , we are documenting all this and the service can then make a strong case to local and political authorities to resolve the denominator dilemma.
Dominic K Atweam
--
You received this message because you are subscribed to the Google Groups "BID Initiative Discussions" group.
To unsubscribe from this group and stop receiving emails from it, send an email to
bidinitiativ...@googlegroups.com.
To post to this group, send email to
bidini...@googlegroups.com.
To view this discussion on the web visit
https://groups.google.com/d/msgid/bidinitiative/BLUPR0701MB1025BFC1970702172C00EBE0D0160%40BLUPR0701MB1025.namprd07.prod.outlook.com.
For more options, visit https://groups.google.com/d/optout.
Vincent Seaman
Hi All,
Sorry to be weighing in late on this discussion, but I would like to let you know about a project we are currently developing that might provide some answers.
We have been working with Oak Ridge National Labs and Andy Tatem (University of Southampton) over the past year to develop an population model that uses feature-extracted (structure level) high resolution satellite imagery and ground-based microcensus data to provide population counts/demographics at a resolution of 90 meters. These numbers should be fairly accurate (>85%) and will not depend on any administrative/census data.
We are just now evaluating the initial output from northern Nigeria, and expect to move on to other countries in Q3-4 this year. The population modeling will also be part of our core data layers project (admin boundaries, settlement names/locations, population, transport) that will begin in June in 3-4 target countries (TBD).
If anyone is interested in this, please let me know. We can likely get access to the imagery and will support the FE, so all that is needed is some ground-trothing data that should be fairly easy to collect if you already have people in-country or are working with the government.
Vince
Sent: Tuesday, February 24, 2015 5:40 AM
To: Puta, Chilunga; bidini...@googlegroups.com
Cc: dominic k atweam
Subject: RE: Resolving the Denominator Dilemma within the Ghana Health Service: Response from Zambia
Eduardo Jezierski
9 Analyzing Telco CDRs to Calibrate Infant Census
Challenge | Unknown baseline population makes it hard to estimate coverage and can lead to stockouts or false immunization statistics. Also population varies at different rates throughout the year. |
Impact |  Medium |
Cost | Medium |
Maturity | Research → Prototype → Pilot → Scale |
Description | Mobile network operators save historical information about subscribers and their cell phone movements (in many formats, but here referred collectively as call detail records (CDRs). It is possible to analyze anonymized and aggregated CDR records for target areas, to see how mobile activity can help inform census of infant population. This approach has been demonstrated useful for modeling the spread of infectious disease in Eastern Africa. |
Implementation Plan |
|
Background Info | Validated by Harvard School of Public Health in Kenya in Dengue and Malaria surveillance and census modeling |
Tags |
|
10 Analyzing Satellite Maps to Calibrate Population & Infant Census
Challenge | Unknown baseline population makes it hard to estimate coverage and can lead to stockouts or false immunization statistics. |
Impact | Medium |
Cost | Medium |
Maturity | Research → Prototype → Pilot → Scale |
Description | Establish a program to analyze maps to see growth of urbanization, road and deforestation / agriculture / land use pattern (via visible and multispectral imaging) to create adjustment factors for population counts. |
Implementation Plan |
|
Background Info | Used by the polio program from the Bill & Melinda Gates Foundation, albeit likely academically. There are increasing freely available high-resolution, high-frequency imagery available, as well as growing libraries of open source algorithms for settlement identification and sizing. Due to changes in regulations and space investments, by 2015, weekly 10cm-class imagery may be freely available for use for humanitarian programs. |
Tags |
|
To view this discussion on the web visit https://groups.google.com/d/msgid/bidinitiative/6FC67AAB6CE093438FBACBA60159E4A92DAD015E%40EMSMBXUM02.seattle.main.gatesfoundation.org.