Finally, 1,320 record sets were derived from 1,245 patients (see the data supplement accompanying the online version of this letter). Risperidone users had more frequent incidence of hypocalcemia (≤2.1 mmol/l) (N=50/328) than comparison subjects (N=19/973) (odds ratio=9.21, 95% confidence interval [CI]=5.34—15.88, χ2=88.4, df=1, p<0.001). The mean calcemia level was significantly lower in the risperidone group (2.24 mmol/l [SD=0.13] vs. 2.33 mmol/l [SD=0.12], t=−11.7, df=1318, p<0.001). Additionally, there was significant correlation between the risperidone dose (mean daily dose: 2.5 mg [SD=1.2], 95% CI=2.4—2.8) and the serum calcium level (F=98.8, df=1, p<0.001; rs=−0.299, p<0.001), with the lowest concentrations at approximately 3 mg (Figure 1). Multivariable logistic regression, with confounders well-documented in the source data (age, gender, current drug treatment, data sets) (see the data supplement), confirmed independent association of risperidone and hypocalcemia (adjusted odds ratio=9.21, 95% CI=5.02—17.81, df=1, p<0.001). Both in the total number of hypocalcemic subjects (N=73) and in those with hypocalcemia within 99 percentiles (N=69), no significant association with any other study variable was found.