Volume 26, Number 3—March 2020
Improving Quality of Patient Data for Treatment of Multidrug- or Rifampin-Resistant Tuberculosis
|Suggested steps||Additional notes|
|Persons responsible for capture and entry of data into electronic databases should be appropriately trained.
||• This includes obtaining a certificate in good clinical practice and training around the importance of confidentiality.
• This also includes training on the basics of MDR/RR-TB, relevant national guidelines, what to collect, how to collect it, and the importance of accuracy in the capture of data.
• These principles can be reinforced with detailed guidance for data capture and the definitions of the variables collected at the point of capture (e.g., within the electronic system or within a document kept where data are captured).
|Quality control measures (e.g., data safeguards) should be implemented to prevent implausible or “out-of-range” entries.
||• A warning can be implemented for continuous variables falling outside plausible ranges (e.g., age outside 0–99 y).
• Drop-down lists can be created to reduce/remove need for free form data entry (e.g., including the most common extrapulmonary TB sites within the dropdown or limiting responses for HIV-coinfection status to positive, negative, or not tested).
• Safeguards can be logical, which prevent certain data from being entered without a specific response in another section (e.g., CD4 and viral load cannot be filled in unless HIV-coinfection status is positive).
|Supervisors should have a standard quality assurance routine (e.g., perform routine follow-up for data accuracy of collected information).
||• Supervisors should have simple algorithms developed to detect implausible information that defy inbuilt measures (e.g., patients reported to be receiving a medicine to which their DST shows resistance).
• Complete checks should be run on at least 10% of records independently via dual extraction. These checks should be performed regularly and assessed by a supervisor with the goal of 95% accuracy.
• Corrective steps should be taken (e.g., further training, more comprehensive or routine checks of variables) when accuracy of data collection is an issue.
|Concurrent checks for data completeness should be performed with assessments of accuracy.||• Reminders can be developed that automatically signal that certain variables are not completed each time a patient record is updated. • In addition, preventing the “finalization” of a patient file until all variables are entered can be implemented—however, files should still be permitted to be saved, and other files opened and populated while patient files await finalization. • Completeness of data is of utmost importance—high frequency of absence of certain information may necessitate exclusion of entire datasets from particular analyses for which these data are required.|
Page created: January 09, 2020
Page updated: January 09, 2020
Page reviewed: January 09, 2020
The conclusions, findings, and opinions expressed by authors contributing to this journal do not necessarily reflect the official position of the U.S. Department of Health and Human Services, the Public Health Service, the Centers for Disease Control and Prevention, or the authors' affiliated institutions. Use of trade names is for identification only and does not imply endorsement by any of the groups named above.