International survey of people living with chronic conditions (PaRIS survey): effects of general practitioner non-participation on the representativeness of the Norwegian patient data.

Background: The International Survey of People living with Chronic Conditions (OECD-PaRIS survey), aims to systematically gather patient-reported experiences (PREMs) and - outcomes (PROMs) and potential predictors for these outcomes for persons with chronic conditions as well as information from professionals about health care provided. In such patient surveys, the advantages of a multilevel (nested) approach in which patients are sampled 'within providers' need to be balanced against the potential for bias if patient populations from participating GPs significantly differ from those of non-participating GPs. The objective was to assess the effects of general practitioner (GP) non-participation on the representativeness of the Norwegian patient data of the International Survey of People living with Chronic Conditions (OECD-PaRIS survey).

Methods: To test all aspects of the first main PaRIS survey, it was preceded by a field trial which this paper reports on the Norwegian part of. For the Norwegian part of the field trial in 2022, we randomly sampled and surveyed 75 GPs and 125 patients 45 years and older for each GP, regardless of whether their GP were also participating in the study. GPs were sampled from a national register that included all GPs. The surveys were primarily digital, but we sent postal questionnaires to non-digital patients and non-responding digital patients. We compared GP and patient characteristics as well as patient-reported experiences and outcomes according to GP participation status in bivariate analysis, supplemented with multiple linear regressions with PREMs/PROMs as dependent variables and participation status as independent adjusting for significant patient factors.

Results: 17 of 75 sampled GPs participated (22.7%), of which 993 of 2,015 patients responded (49.3%). 3,347 of 7,080 patients of non-responding GPs answered (47.3%). Persons with chronic conditions from participating GPs reported significantly better patient-centred coordinated care (p = 0.017), overall experiences with the GP office the last 12 months (p = 0.004), mental well-being (p = 0.039) and mental health (p = 0.013) than patients from non-participating GPs. The raw differences between participating and non-participating GPs on patient-reported experiences and - outcomes varied from 1.5 to 2.9 points on a 0-100 scale, and from 2.2 to 3.0 after adjustment for case-mix.

Conclusions: The Norwegian field trial indicates that estimates based on participants in the PaRIS survey may modestly overestimate patient-reported experiences and -outcomes at the aggregated level and the need for more research within and across countries to identify and address this potential bias.