Can health workers reliably assess their own work? A test-retest study of bias among data collectors conducting a Lot Quality Assurance Sampling survey in Uganda.

Beckworth, Colin, Davis, Rosemary H., Faragher, Brian and Valadez, Joseph ORCID: https://orcid.org/0000-0002-6575-6592 (2014) 'Can health workers reliably assess their own work? A test-retest study of bias among data collectors conducting a Lot Quality Assurance Sampling survey in Uganda.'. Health Policy and Planning, Vol 30, Issue 2, pp. 181-186.

Full text not available from this repository.

Abstract

BACKGROUND
Lot Quality Assurance Sampling (LQAS) is a classification method that enables local health staff to assess health programmes for which they are responsible. While LQAS has been favourably reviewed by the World Bank and World Health Organization (WHO), questions remain about whether using local health staff as data collectors can lead to biased data.

METHODS
In this test-retest research, Pallisa Health District in Uganda is subdivided into four administrative units called supervision areas (SA). Data collectors from each SA conducted an LQAS survey. A week later, the data collectors were swapped to a different SA, outside their area of responsibility, to repeat the LQAS survey with the same respondents. The two data sets were analysed for agreement using Cohens' kappa coefficient and disagreements were analysed.

RESULTS
Kappa values ranged from 0.19 to 0.97. On average, there was a moderate degree of agreement for knowledge indicators and a substantial level for practice indicators. Respondents were found to be systematically more knowledgeable on retest indicating bias favouring the retest, although no evidence of bias was found for practices indicators.

CONCLUSIONS
In this initial study, using local health care providers to collect data did not bias data collection. The bias observed in the knowledge indicators is most likely due to the 'practice effect', whereby respondents increased their knowledge as a result of completing the first survey, as no corresponding effect was seen in the practices indicators.

Item Type:	Article
Subjects:	QY Clinical Pathology > QY 4 General works W General Medicine. Health Professions > Professional practice > W88 Administrative work. Teaching. Research WA Public Health > WA 20.5 Research (General)
Faculty: Department:	Clinical Sciences & International Health > Clinical Sciences Department Clinical Sciences & International Health > International Public Health Department
Digital Object Identifer (DOI):	https://doi.org/10.1093/heapol/czt110
Depositing User:	Helen Fletcher
Date Deposited:	28 Aug 2014 08:37
Last Modified:	06 Sep 2019 11:29
URI:	https://archive.lstmed.ac.uk/id/eprint/3602

Statistics

View details

Actions (login required)

Edit Item