Introduction Due to increasingly restrictive privacy regulations, it has become difficult to access representative general population samples. In the UK, commercially available online datasets are available that pool data from several sources, but are generally untested in terms of completeness and accuracy.
Aim To evaluate the performance of a commercially available online dataset (192.com) as a population sampling frame.
Methods Firstly, 192.com was employed to identify age- and sex-matched population controls for a case-control study. Selected persons were sent a health questionnaire. Secondly, 192.com was challenged with a sample of consecutive hospital attendees whose NHS data allowed testing for coverage and accuracy. Population representativeness was assessed by comparing demographic characteristics of persons who were/were not identifiable via 192.com, using simple descriptive statistics.
Results From 814 age- and sex-matched controls identified, a 25% questionnaire response rate was achieved. 3% were returned with address inaccuracy, and no complaints were received from any recipient. Of the 163 hospital attendees, 135 (83%) were identified on the website. Comparing those who were/were not identified revealed no differences in mean age (identified=59 yrs, vs 60 yrs; PMann–Whitney=0.45), sex (identified=50% Female vs 57%; Pχ2=0.51), deprivation (PMann–Whitney=0.71) or rurality (Pχ2=0.11). Of those identified, an age estimate was quoted for 131 (97%) of which 113(86%) were accurate.
Conclusions This is the first evaluation of an online population sampling frame of which we are aware. The website, 192.com, acted as a feasible sampling frame, demonstrating good participant acceptability, coverage, accuracy and representation. In the future, such resources could provide researchers with alternative sampling opportunities.