Baseline data showed that the study population comprised patients with a wide range of age and duration of disease. As patients could be either untreated or in need of a treatment switch, this study possibly included patients who received antipsychotic medications for years, but eventually had to be switched due to treatment-emergent adverse events or insufficient efficacy.
The percentages of patients with known concomitant hypertension (16.7%), lipid metabolism disorder (6.7%) and diabetes (5.6%) appeared moderate compared to numbers from German primary care patients (hypertension 31.6%, lipid metabolism disorder 23.4%, diabetes 9.4%) [29
]. However, the vital signs and laboratory data collected at baseline revealed high blood pressure in 54.8%, increased triglycerides in 52.5% and increased blood glucose in 14.1% of the patients. This remarkable discrepancy emphasizes how important the actual monitoring of vital signs and blood values is in patients with schizophrenia, as seemingly, a large proportion of these patients were neither aware of their somatic health status nor adequately diagnosed and treated for cardiovascular risk factors.
Regarding baseline differences between the treatment groups (Prev-AP and New-AP), only two cohorts contrasted perceptibly from the others: One was the small (N = 16) group of New-Typ. These patients had clinically noticeable high mean values for BMI (32.3 kg/cm²), waist circumference (111.3 cm) and blood pressure (SBP/DBP 134.6/84.6 mmHG), and 12 of them (75%) actually met the criteria of MetS (AHA/NHLB). Though this cohort was too small for reliable statistical evidence, a possible explanation might be that these patients were switched/newly initiated on typical antipsychotics, because their metabolic and cardiovascular risk was already evident and these substances were assumed to have a lower risk of treatment-emergent metabolic adverse events. Though, in our study, the perception of lower risk of metabolic adverse events through typical antipsychotics was not supported by the baseline values found in the Prev-Typ cohort.
The other treatment cohort with noteworthy baseline values was Prev-None. These previously untreated patients showed numerically lower mean values for BMI, blood pressure, prevalence of somatic concomitant disease and practically all laboratory parameters than any other Prev-AP cohort, but had a comparatively higher symptom severity at baseline (mean CGI-S 4.2).
Apart from Prev-None, the Prev-AP cohorts did not contrast clearly with respect to baseline values; the highest percentages of patients with laboratory values out of normal range dispersed in different treatment groups for different parameters (see Table ). This possibly reflects that changes in metabolic parameters may occur in patients treated with any antipsychotic medication, though these may differ in grade and type according to the properties of the respective substance and the patients' individual risk factors.
The prevalence of MetS in the FAS of 42.8% (AHA/NHLB definition) at baseline was comparable to the findings from the CATIE study, which reported a baseline MetS prevalence of 42.7% in an US-American sample of patients with schizophrenia [28
The Prev-AP cohorts who had received some previous antipsychotic treatment showed no statistically significant differences in MetS-rates (AHA/NHBL). However, patients who entered our study untreated (Prev-None) had a baseline MetS prevalence of 24.7%, which was significantly lower than in any other cohort but Prev-Risp (42.4%, but overlapping CI). For comparison, Moebus et al. [30
] reported a MetS prevalence rate of 28.6 ± 0.45% (AHA/NHLB criteria) in a cross-sectional sample of 33,502 primary care patients in Germany. Considering that Moebus' patients had a higher mean age than our study sample (53.0 ± 15.8 years in men and 50.9 ± 16.2 years in women versus 43.1 ± 13.1 and 47.3 ± 13.1 years, respectively, in our study), the prevalence of MetS in the Prev-None cohort appears to resemble the rates seen in primary care patients.
Considering the changes in MetS prevalence, the differences between baseline and month-3 lacked significance for all New-AP groups. Though, looking at the mean change of the particular MetS-components, a trend to increase was apparent in lipids, which could be a possible early predictor.
The results from logistic regression models at visit 2 indicate that the factors "increased CRP
", "concomitant somatic diseases
", and "concomitant non-psychiatric medication
" increased the odds to develop MetS, while "female sex
" and "smoking
" decreased them. The factors "concomitant somatic disease
" and "concomitant non-psychiatric medication
" are in part comprised in the MetS definitons, and CRP is an established indicator of cardiovascular risk [31
]. We did not expect, however, to find that smoking decreased the odds for MetS; this might possibly be an effect of the appetite reducing properties of nicotine [33
Regarding the lower MetS-odds for women, data from the German general population [34
] show women to have a lower incidence of cardiovascular and cerebrovascular events than men up to the age of 64, after which the respective rates converge (cardiovascular) or even become inverted (cerebrovascular). The review of cardiovascular risk factors in women by Evangelista and MacLaughlin [35
], comprising international data published between 1990 and 2008, provided similar results. Considering the age structure of our study sample (FAS: mean age 45.2 years, Q1 36 years, Q3 54 years) our results fit well into the general picture.
They do, however, contradict the results from the CATIE study: McEvoy et al. [28
] reports MetS-prevalences of 36.0% in men and 51.6% in women (fasting cohort, N = 689); the higher risk for MetS in women was a universal finding in all age groups, races and ethnicities. However, CATIE was a controlled clinical trial, so apart from country specific confounders as behavioral and dietary habits; possible selection bias might have impacted the results.
Several limitations of this study should be considered: As the study did not reach the required sample size, the analyses were underpowered, and therefore logistic regression models might have failed to detect all effects associated with MetS. Furthermore, the observational period of three months might have been too short to observe certain changes in metabolic status as e.g. development of insulin resistance or the processes leading eventually to increased CRP. Due to the observational design, treatment cohorts were defined post-hoc, depending on the actual case numbers treated with each antipsychotic, and compounds which were less frequently prescribed had to be grouped.