Table 2.
Phenotypic domains and statistical methods used in the included studies
Hypothesis-driven studies | Data-driven studies | |
n = 8 | n = 25 | |
Number of phenotypic domains used for subtyping - N (%) | ||
Single-domain | 7 (87.5) | 9 (36.0) |
Two domains | 1 (12.5) | - |
> = 3 domains | 16 (64.0) | |
Phenotypic domain(s) - N (%) | ||
Demographic | 2 (25.0) | 8 (32.0) |
Motor | 3 (37.5) | 18 (72.0) |
Cognitive | 1 (12.5) | 17 (68.0) |
Emotional | 1 (12.5) | 17 (68.0) |
Autonomic | 1 (12.5) | 11 (44.0) |
Treatment | - | 3 (12.0) |
Non-clinical Biomarkers* | - | 5 (20.0) |
*Imaging (n = 4), biochemical (n = 1) | ||
Statistical approaches | ||
Pre-determined groups | 8 (100) | - |
Hierarchical cluster analysis | - | 7 (28.0) |
Non-hierarchical cluster analysis | - | 16 (64.0) |
Hierarchical and Non-hierarchical cluster analysis | 1 (4.0) | |
Other | 1 (4.0) |
Five studies were not included in this analysis for the following reasons: combined hypothesis- and data-driven approach (n = 2) and other methods (regression analyses, n = 2; subtype criteria defined post-hoc, n = 1).