Table 3. SCZ case characteristics across cluster groups.
Table 3A. SCZ case characteristics across cluster groups in national training set
| Feature | Cluster 1 (medium EDU) |
Cluster 2 (low EDU) |
Cluster 3 (high EDU) |
Cluster 4 (ID) |
P | |
|---|---|---|---|---|---|---|
| N (%) | 3,865 (56.6%) | 1,739 (25.5%) | 798 (11.7%) | 421 (6.2%) | – | |
| Clustering input variables | Age at first SCZ diagnosis, mean (SD) | 0.09 (0.96) | −0.23 (0.99) | 0.35 (0.91) | −0.05 (1.10) | 5.55×10−48 ‡ |
| ID, N (%) | 0 (0%) | 0 (0%) | 0 (0%) | 421 (100%) | - | |
| EDU, mean (SD) | 0.12 (0.27) | −1.10 (0.27) | 2.17 (0.19) | −0.70 (0.71) | <1×10−300 ‡ | |
| Parental EDU, mean (SD) | 0.03 (0.99) | −0.27 (0.90) | 0.63 (1.06) | −0.43 (0.86) | 3.08×10−117 ‡ | |
| Number of BIP contacts, mean (SD) | 0.01 (1.00) | −0.06 (0.76) | 0.09 (1.28) | −0.03 (0.78) | 0.002 ‡ | |
| Birth year, mean (SD) | 1968 (8.61) | 1969 (9.03) | 1969 (7.61) | 1969 (8.41) | 0.001 ‡ | |
| Male sex, N (%) | 2,444 (63.2%) | 1,148 (66.0%) | 408 (51.1%) | 253 (60.1%) | 7.04×10−12 ‡ | |
| Premorbid cognitive ability (males), mean (SD) | −0.43 (0.99) | −1.01 (0.91) | 0.38 (0.90) | −1.75 (0.59) | 6.38×10−130 ‡ | |
| Attempt/completed suicide, N (%) | 497 (12.9%) | 284 (16.3%) | 66 (8.3%) | 60 (14.3%) | 4.40×10−7 ‡ | |
| Death, N (%) | 322 (8.3%) | 198 (11.4%) | 28 (3.5%) | 37 (8.8%) | 1.53×10−9 ‡ | |
| Ever hospitalized for more than 200 days, N (%) | 629 (16.3%) | 487 (28.0%) | 69 (8.6%) | 76 (18.1%) | 2.73×10−36 ‡ | |
| Ever use of clozapine, N (%) | 879 (22.7%) | 464 (26.7%) | 141 (17.7%) | 103 (24.5%) | 2.61×10−7 ‡ | |
Abbreviations: SCZ, schizophrenia; BIP, bipolar disorder; EDU, educational attainment; ID, intellectual disability. Parental EDU is either from mother or from father if only one among them is available; if both mother’s and father’s EDU are available, take the mean. Age at first SCZ diagnosis, EDU, parental EDU, and number of BIP contacts are regressed on birth year and sex and then take the standardized residuals within the population case cohort. Premorbid cognitive ability is Z-score standardized by birth year in the whole population cohort. The hospitalization >200 days is the median length of hospitalization for those in top decile of hospitalization. Statistical comparisons are one-way ANOVA for continuous variables and chi-square test for categorical variables.
Indicates results exceeding Bonferroni-corrected significance threshold (N=11, P < 0.004).