Comparison of information provided on diabetes distress when using PAID-20, PAID-11 or PAID-5. A registry-based study applying IRT analyses among 10,190 adults with type 1 diabetes in Norway

Aims

To 1) evaluate information provided on the latent diabetes distress trait when using the Problem Areas In Diabetes, 20 item (PAID-20), PAID-11 or PAID-5, and 2) evaluate the information provided by each scale item.

Methods

Using nationwide registry data from 10,190 individuals with type 1 diabetes in Norway, we applied Item Response Theory (IRT) analyses to compare test information curves for PAID-20, PAID-11, PAID-5 and their scale items.

Results

The test information curve for PAID-20 captured an overall broader range of the latent distress trait compared to the other versions. At test information level ≥ 10, the ranges (SD under/above mean) were: PAID-20: −1.30 to 3.32; PAID-11; −1.17 to 2.76; and PAID-5: −0.59 to 2.27. Six PAID-20 items had flat information curves with limited contribution to the latent trait, whereas the PAID-11 included 11 of the 12 items most frequently reported as serious problem areas.

Conclusions

The PAID-20 captures most information on the latent diabetes distress trait; however, the PAID-11 provides comparable information and captures the most serious problem areas. The PAID-5 provided limited information and covers limited aspects of distress. Overall, the PAID-11 may be a good choice for assessing diabetes distress in both research and clinical practice.