Population-Based Versus Hospital-Based Data in Amyotrophic Lateral Sclerosis-A Factor to Consider?

Background: Over the past years, some studies in amyotrophic lateral sclerosis (ALS) have provided heterogeneous findings regarding demographic and clinical data as well as the impact of various prognostic factors. It is well known that these inconsistencies might be caused by a selection bias in hospital-based data sets. In this study, we sought to further characterize this selection bias.

Methods: We compared hospital-based data from the ALS center at Ulm University (UC; n = 3833; 1997-2021) with the population-based ALS registry Swabia (SR; n = 852; 2010-2020).

Results: Patients from UC were younger (age of onset 60.9 [IQR 52.4-68.9] vs. 65.0 [57.0-72.7]), had a higher share of males (60.5% vs. 56.3%), a longer diagnostic delay (10.5 [IQR 6.4-18.4] months vs. 6.9 [IQR 3.4-12.1] months), a higher prevalence of the "definite" category according to El Escorial diagnostic criteria (60.9% vs. 11.2%), a higher share of familial cases (12.9% vs. 6.3%), a slower progression rate (points of ALS functional rating scale revised lost per month -0.54 [IQR -1.02 to -0.28] vs. -0.79 [IQR -1.47 to -0.43]), and (among all deceased patients) a higher share of percutaneous endoscopic gastrostomy (26.7% vs. 17.7%) and non-invasive ventilation (34.3% vs. 25.3%).

Conclusions: The observed differences likely indicate a selection bias in hospital-based data, which may be attributed, among others, to the willingness to travel large distances to a specialized center, the desire to participate in clinical studies, and the attitude toward life-prolonging measures. These differences must be considered when interpreting and generalizing study results from hospital-based populations.