Service Delivery Indicators Health Survey 2013 - Harmonized Public Use Data
The SDI provides a set of metrics to benchmark the performance of schools and health facilities in Africa. The Indicators can be used to track progress within and across countries over time, and aim to enhance active monitoring of service delivery to increase public accountability and good governance. Ultimately, the goal of this effort is to help policymakers, citizens, service providers, donors, and other stakeholders enhance the quality of services and improve development outcomes. The perspective adopted by the Indicators is that of citizens accessing a service. The Indicators assemble objective and quantitative information from a survey of frontline service delivery units, using modules from the Public Expenditure Tracking Survey (PETS), Quantitative Service Delivery Survey (QSDS), and Staff Absence Survey (SAS). The SDI initiative is a partnership of the World Bank, the African Economic Research Consortium (AERC), and the African Development Bank. More information on the SDI survey instruments and data, and more generally on the SDI initiative can be found at: www.SDIndicators.org and www.worldbank.org/SDI, or by contacting SDI@worldbank.org.
Kind of data
Sample survey data [ssd]
Version 01: Harmonized, anonymous dataset for public distribution.
National coverage, representative of Uganda, public/private and rural/urban.
Central Region Eastern Region Kampala Northern Region Western Region
Unit of analysis
There are two different units of analysis: facilities and individuals. Individuals are assessed both for absence and for knowledge.
All primary health care facilities up to district hospitals.
Producers and sponsors
World Bank Group
Economic Policy Research Center
Data collection and processing
William and Flora Hewlett Foundation
A multi-stage clustered sampling strategy was adopted in the Uganda SDI. The first stage cluster selection was carried out independently within each explicit stratum. The primary cluster considered was at the county level, which was, therefore, the primary sampling unit (PSU). At the second stage, schools were selected and, at the third stage, health facility's staff for absence and clinicians for knowledge assessment. It was decided than within each stratum, except Kampala, 10 counties would be chosen with probability proportional to size (number of households
Sampling Frame for the 2013 Uganda SDI:
The sampling frame for the 2013 Uganda SDI was based on the Health Facility Inventory of 2012 provided by the Ministry of Health. The original sample frame contained 4,999 health facilities (HC2, HC3, HC4, and hospitals) with identifier variables from the region to the parish in which each facility was located. The inventory also included the ownership of each facility as well as whether it was functional. The final sample frame was purged of all non-functional facilities as well as large hospitals which are not part of the SDI study.
Stratification of the Sampling Frame for the 2013 Uganda SDI:
Although the SDI was usually representative of the national and urban and rural areas, in Uganda it was requested that the survey be also representative at the regional level. Because of its special status, Kampala was extracted from the Central region and considered a stratum in and of itself.
Unfortunately, the sample frame did not contain an urban/rural variable necessary for proper (implicit) stratification of the health facilities. With the help of the Uganda Bureau of Statistics (UBoS), further work was done to create an urban/rural variable and match each health facility according to its location.
Sample Size and Sample Allocation for the 2013 Uganda SDI:
To approximate the precision of the estimate, a previous similar survey or a survey measuring the same indicator could be very useful. Uganda National Panel Survey (UNPS) 2010/11 visited health facilities to measure absence rate - a key variable for the SDI. The UNPS absence variable was used for simulation of the appropriate sample size
Sampling Health Facilities and Health Workers:
After the total sample size and its allocation across regions were decided, the next step was to sample the actual health facilities that would be included in the final sample and the health workers and clinicians that would be selected for absence and competence assessment within each health facility. This was done using a two-stage sampling method. First, in each stratum, health facilities were chosen within the selected counties, and then, staff were selected in a second stage within each selected health facility during the visit.
The health facilities were chosen using probability proportional to size (PPS), where size was the number of households as provided by the UBoS community database. As for the selection of the cluster, the use of PPS implied that each household within a stratum had an equal probability for its health facility to be selected.
Finally, within each health facility, up to 10 staff were selected for estimating absence and for the administration of the vignettes. There were two different sampling procedures for measuring absence rate and assessing knowledge. For absence rate, 10 staff were randomly selected in the facility roster and the whereabouts of those staff was ascertained in a return surprise visit. For the knowledge assessment, however, up to 10 clinicians among those who regularly see patients and provide prescriptions were included in the sample. These procedures implied that health workers across strata and within health facility (for assessment), did not all have the same probability of selection. It was, therefore, warranted to compute weights for reporting the survey results.
To be representative of the population of interest, sample estimates from the 2013 Uganda SDI had to be properly weighted using a sampling weight, or expansion factor. Note that different weights needed to be applied depending on the relevant level for the variable (the health facility, staff or clinician). The basic weight for each entity was equal to the inverse of its probability of selection, which was computed by multiplying the probabilities of selection at each sampling stage. All the weights were computed and included in the dataset.
Dates of collection
Mode of data collection
Four modules were used, with multiple sections. All but module 4 (ressources) are in the archive.
Module 1 covers: (Section A) metadata; (B) general information; (C) Infrastructure; (D) Equipment, materials, and supplies; and (E) Drugs and consumables.
Module 2 covers: (Section A) basic demographic information on the facility; and (B) additional demographic information on a subset of staff.
Module 3 covers knowledge, as measured through seven patient case simulations.
Module 4 (not in the archive) covers resources and governance.
Data has been anonymized, but users commit to not seeking to re-identify statistical units in the data set.
The harmonized, anonymized datasets are available as public use files.
Researchers who feel that they need non-anonymized data should contact email@example.com with a statement of research objectives and a rationale for why they require such data. That will start the Research Use File discussion.
Use of the dataset must be acknowledged using a citation which would include:
- the Identification of the Primary Investigator
- the title of the survey (including country, acronym and year of implementation)
- the survey reference number
- the source and date of download
Disclaimer and copyrights
The user of the data acknowledges that the original collector of the data, the authorized distributor of the data, and the relevant funding agency bear no responsibility for use of the data or for interpretations or inferences based upon such uses.