Number of patients needed to discriminate between subgroups in patient reported outcome measures

Background: Patient reported outcome-measures (PROs) are increasingly used in orthopedics. Information on number of patients needed in different settings is warranted.

Aim: To assess the number of patients needed for different PROs to discriminate between subgroups of age, gender, and diagnosis.

Methods: 5777 primary THA patients, operated 1‐2, 5‐6, and 10‐11 years ago. SF‐12 Health Survey (SF-12), EQ-5D, Oxford 12‐item Hip Score (OHS), and Hip dysfunction and Osteoarthritis Outcome Score (HOOS) were included. The different PRO subscales abilities to discriminate between groups were studied using analysis of variance. The hypothetical number of subjects needed to find the significant difference in PRO mean value between groups (assuming a significance level of 5 % and a power of 85 % to detect differences between the actual groups in our current study) was estimated for each PRO subscales with sample size calculations or by power calculations and simulated ANOVA F tests, depending on the number of groups.

Results: To discriminate between gender, the least number needed to find a statistically significant difference in mean sum score in each group was 298 (OHS) while HOOS QoL required the most number of subjects (760 in each group). PCS had the least number needed in relation to diagnoses (51 patients per group needed), while HOOS Pain required the most (116 patients per group needed). Concerning age, the least number needed was 270 (EQ-VAS), and OHS required the most (1566 in each group).

Discussion/ Conclusion: Subgroup analyses in a hip registry setting require group-sizes from 51 to 1566 dependent on descriptive factor and choice of PRO which seems feasible in registry context.