An Evaluation of the Accuracy of Patient Data in NEDSS 10/23/2010 Andrea E. Palmer Graduate Research Assistant Center for TB Control and Prevention Maryland Department of Health and Mental Hygiene Infectious Disease and Environmental Health Administration MISSION To improve the health of Marylanders by reducing the transmission of infectious diseases, helping impacted persons live longer, healthier lives, and protecting individuals and communities from environmental health hazards We work in partnership with local health departments, providers, community based organizations, and public and private sector agencies to provide public health leadership in the prevention, control, monitoring, and treatment of infectious diseases and environmental health hazards. Background Transition of case entry from state-level (TIMS) to local-level (NEDSS) Maryland, to date, has no formal, established system of data quality assurance Initial overview of the data in Maryland made to assess QA needs 1
Consequences of Bad Data Inaccurate follow-up of services to patients Inadequate resources (i.e. funding, staff, facilities, drugs and supplies) Inaccurate evaluation and policy Misrepresentation of the public health burden of TB Inability to measure TB program indicators that are based on surveillance data *Special thanks to Elvin Magee at the CDC for use of this slide. Maryland s Counties Highest Morbidity Counties in 2009 Montgomery 70 cases (32%) Prince George s 65 cases (30%) Baltimore County 25 cases (11%) Baltimore City 17 cases (8%) These four counties account for 81% of Maryland s cases Methods Scope of project All counties with at least 1 case in 2009 Chart Selection Low case counts High case counts Review 2
Data Entry and Analysis Entry Excel Analysis SPSS v. 18 Results Brief overview of Incorrect fields Missing fields Fields without Documentation Possible explanations Possible Areas of funding Errors within funding variables Examples of errors Incorrect fields 491 out of 8060 (6%) applicable responses incorrect overall 1. Date reported 41% incorrect (39/96) 20d. Culture of tissue/other fluids: date result reported 33% incorrect (15/45) 18c. Sputum culture: date result reported 33% incorrect (28/86) 41b. Culture conversion: date specimen collected 31% incorrect (16/52) 18b. Sputum culture 20% incorrect (17/86) 3
Missing Fields 625 out of 8060 (8 %) applicable responses were missing 22B3. CT Scan Evidence of miliary TB 42% missing g( (25/60) 48a. Was follow-up susceptibility testing done? 33% missing (19/58) 22B2. CT Scan Evidence of cavity 33% missing (20/61) 47a. Directly Observed Therapy (DOT) 25% missing (15/57) 22A3. X-Ray Evidence of miliary TB 21% missing (18/85) Fields with No Documentation 305 out of 8060 (4%) applicable responses with no chart documentation 35. Immigration status at first entry 28% not documented (28/99) 31. Injecting drug use 21% not documented (21/99) 32. Non-injecting drug use 21% not documented (21/99) 33. Excess alcohol 21% not documented (21/99) Possible Explanations Clerical/Human Error Misunderstanding of variable definition Lack of updates to NEDSS 4
Example Areas of Funding Foreign-born HIV Coinfection Injecting Drug Use Non-Injecting Drug Use Excess Alcohol Use Homelessness Errors within Example Funding Variables Variable Proportion Percent Incorrect Incorrect Foreign born 0/100 0% HIV Coinfection 9/100 9% Injecting Drug Use 4/100 4% Non Injecting Drug Use 5/100 5% Excess Alcohol Use 6/100 6% Homelessness 4/100 4% Examples of Errors within Funding Variables Homelessness Misunderstanding of variable definition Excess Alcohol Use Lack of updates to NEDSS 5
Outcomes and What s Next This assessment will help to: Identify where and to what extent the errors are occurring Raise awareness of the importance of data quality measures Next steps Address training needs Quick reference guides Quality assurance measures Regularly scheduled conference calls Acknowledgments Questions? 6
Maryland Infectious Disease and Environmental Health Administration http://eh.dhmh.md.gov/ideah 7