SHM annual report: what is it that makes SHM’s data high quality?

Released two weeks ago, the SHM annual report for 2014 gives a comprehensive overview of SHM’s activities during the past calendar year. As well as detailing research activities, publications, and the financial statements, it also provides an extensive report on the activities carried out by the data management and quality control unit. Largely unseen by the outside world, these activities nonetheless contribute significantly to the high quality data generated by SHM and are vital to SHM’s research and scientific output.  Here we give a brief overview of what the data management and quality control unit did in 2014.

Highly-trained data collectors

SHM’s database contains anonymous data from more than 23,000 individuals infected with HIV. In 2014, data were collected from just over 18,000 of these individuals. The patient data and quality control unit is responsible, in part, for the collection and entry of these data into the SHM database. This work is carried out by highly-trained data collectors who receive regular training on specific items and issue-specific support from a helpdesk. Furthermore, as part of the personal coaching programme, an average of four patient files per data collector were selected and subjected to a quality control. The outcome of these checks was discussed with the data collector and, if necessary, used as the basis for item-specific training.

Ongoing protocol reviews

Ongoing efforts are undertaken to review and improve the data collection protocols in line with changes and developments in the field. In 2014, this review resulted in changes in the collection of data on HIV transmission and recent HIV infections, diabetes mellitus, cardiovascular disease, viral hepatitis and liver work-ups. These improvements are relayed back to the data collectors during specially-organised review days (see article on the latest data collectors’ review day elsewhere in this newsletter).

Automated and manual quality controls

SHM carries out both automated and manual quality checks on the collected data. For example, in 2014, automated checks based on 170 validation rules selected 17,520 records for subsequent verification by the data collectors. Despite an increase in the number of validation rules, the number of records selected for verification dropped by 19% in 2014 compared to 2013. This highlights how effective these automated checks are in improving data quality.

In addition, SHM data quality staff carried out a wide range of manual checks. These manual checks targeted data essential for SHM’s scientific output, consistency within the data, a random selection of new items for which data were collected in 2014 for the first time, and more complex data that could be used for training purposes. Data from 502 patients were randomly selected and checked, while all reported cases of cardiovascular disease and other endpoints (n=798) were also checked. Finally, cause of death was validated for all deceased patients (n=211). In total, during the course of 2014, the HIV treatment centres were visited regularly by the data quality staff and data from 1,800 patients underwent manual verification.

This is just a brief overview of the many activities carried out by SHM data management and quality control unit during 2014. For a more detailed overview, please take a look at our Annual Report, which you can now download from our website by clicking on this link.

For a short animated summary of the work that goes into collecting and preparing the data for analyses, please click here.