Assessment of data quality of an EMR system, and the effect of CD4 guideline change and patient escort on time to linkage to and retention in ART: a study at Kiambu County Referral Hospital
Muthee, Winnie G.
MetadataShow full item record
Kiambu County Referral Hospital adopted electronic medical records in the HIV clinic for longitudinal follow up of patients enrolled in care. The hospital also implemented CD4 guideline changes on antiretroviral therapy eligibility and a patient escort system to facilitate timely linkage of patients to treatment after HIV diagnosis. This study first evaluated data quality of the electronic medical records database, which was then utilized to assess the effect of CD4 guideline changes and patient escort on time to linkage to and retention in antiretroviral therapy. This study is a cross-sectional study that involved evaluation of medical records of eligible patient populations. For the data quality study, the sample size was based on random sampling on Microsoft Excel using a line list that was extracted from the electronic medical record database. This sample of electronic records was then compared against a similar sample of physical records. For the assessment of effect of CD4 guideline change and patient escort on time to linkage to and retention in antiretroviral therapy, a complete census of the eligible population was done using the electronic medical record database. There was over 70% completion rate for most clinical indicators evaluated in both electronic and physical records. Both ART start date (96 % vs 77%) and baseline CD4 (73% vs 56%) were significantly more complete in the physical records compared to electronic records. Almost all evaluated indicators had a mismatch rate of less than 10% between physical and electronic records, except for “date started ART” that had the highest mismatch rate of 20%. There was a significant decline in median time to linkage to ART from 51 days in the 350 CD4 cut-off group to 16 days in the “test and treat” group. There was a significantly higher retention in ART, and lower overall attrition and lost to follow up for the “test and treat” group. However, there was no significant difference in retention, overall attrition, lost to follow up and mortality between the pre and post patient escort groups. The study findings have important implications for the use and interpretation of data derived from EMR databases for ongoing patient follow up and retention in treatment programmes as well as operational research. The “test and treat” strategy showed improvements in both median time to linkage to and retention in ART. More research on patient escort as a behavioral intervention for linkage and retention in ART, as well as impact of “test and treat” strategy on mortality is needed.