Background: There is no symptom-based patient-reported outcomes (PRO) measurement available in IBD. Disease scores contain a mixture of PRO and physician's observations and have shown serious limitations in clinical trials. Comparison between healthcare professionals (HCP) and patient (P) reports on scores' items is a first step toward disease scores refinement. In our IBD cohort study, we were able to collect P and HCP-reported symptoms independently. We assessed the agreement between both measures, and tested the correlation between the general well-being item (GWB) and two health-related quality of life (HRQoL) measures.
Methods: Between 2012 and 2015, we collected CDAI and MTWAI items 1) during follow-up medical visits, 2) through P self-reported follow-up questionnaire, except lab values. We compared items independently reported by HCP and P, stratified by diagnostic and Δt HCP-P reports. We calculated the Cohen's kappa (κ) statistic for agreement. A quadratic weight was applied for more severely serious disagreements. For EIM & complications, we computed a pooled κ based on the average between observed and expected probability of agreement over sub-items. A pooled κ was computed to summarize agreement over all examined variables. We also collected SF-36 and IBDQ scores. Pearson correlation coefficients r were calculated between both scores and GWB reports of HCP and P.
Results: 2427 reports could be evaluated (Δt: 537<1 month, 390 1–2, 1500 2–6), referring to 1385 patients (52% females, 58% CD).
The best overall κ was found at Δt 1–2 months, moderate for number of stools/wk and antidiarrheal treatment (AT) in CD, moderate to good for nocturnal diarrhea and bloody stools in UC. Agreement on GWB was low to very low. P-reported GWB were well correlated with IBDQ (CD: r=0.65, UC: r=0.67), SF-36 physical (PCS) (CD: r=0.52, UC: r=0.58) an SF-36 mental (MCS) component scores (CD: r=0.47, UC: r=0.46). Correlation of PCS resp. IBQD with HCP-reported CD-GBW was moderate at Δt <1 and 2–3 months (r=−0.45 and −0.53, resp. −0.43 and −0.48), but correlation with MCS remained low (r<0.40) whatever Δt. For UC, HCP-reported GBW moderately correlated with IBDQ at Δt<1 and 1–2 months (r=−0.48 and −0.47), but was low when Δt >2. Correlation with PCS and MCS remained low whatever Δt.
Conclusions: The agreement was low for many scores' items, except two per disease. Among scores' items with high weight, eg CDAI AT or GWB, agreement was surprisingly low. P-GWB correlated with HRQoL scores better than HCP, especially for scores related to mental or emotional aspects.