Methods
A prospective naturalistic observational study was conducted in the Department of Psychiatry of San Matteo Hospital in Pavia. The study was approved by local Ethics Review Committee.
We chosed to recruit only women because of their higher risk of developing drug related arrythmias. Consecutive female inpatients admitted from August 2003 to April 2004, with schizophrenia, bipolar disorder or schizoaffective disorder, were considered eligible for the study. Diagnoses were made by two staff psychiatrists (one attending and one resident psychiatrist), after reaching a clinical consensus in accordance to the DSM IV.
Pharmacological and medical history were obtained.
Included patients had to be free from psychiatric medications for at least 48 hours. Patients who were taking fluoxetine untill three days before recruitment were excluded, because of the long half life of this drug (3–5 days). Also patients treated with depot preparations were excluded. Non psychoactive drugs, like cardiovascular drugs, were allowed only if they were not reported to alter QT interval.
Patients with disturbances of cardiac rate and rhythm, history of prolonged QTc, family history of sudden death, QTc interval greater than 470 ms in the ECG performed at admission, alterations of hepatic or renal function and substance abusers patients were also excluded.
For each subject, therapy was started according to the clinical evaluation of psychiatrist in charge of the patient. The first group (Group 1) included women who were treated with only an antipsychotic (concomitant benzotropine treatment was permitted, as also zolpidem for insomnia was), the second group (Group 2) was composed of female patients who started treatment with antipsychotics in association with either an antidepressant or lithium. The dosage equivalent of haloperidol was calculated [19].
Two ECGs were obtained for each patient, the first before the beginning of treatment and the second after four days of treatment with the patients on the full therapeutic daily dose of antipsychotic prescribed by the clinician, generally after one week from the beginning of the treatment, except for the patients treated with clozapine, who had the second ECG four days after the end of titration (generally after two weeks of therapy).
On the same day of the first ECG, blood samples were drawn for the evaluation of potassium serum levels from all the participants. When the second ECG was administered, blood was drawn to obtain potassium and additionally magnesium serum levels as well as serum levels of the antipsychotic agent.
Samples for plasmatic levels determination were drawn before the first drug dose in the morning. Serum antipsychotic levels were analyzed by high-performance liquid chromatography with ultraviolet detection.
The ECGs were obtained by standard 3-leads resting ECG procedure in the supine position and analyzed by a resident cardiologist (A. V.) who was blind to the patient's condition, study hypothesis, treatment status, serum levels of antipsychotics and was not involved in patient care. The QTc interval was calculated with the Bazzett formula. The QT interval was assessed in both DII and V2 leads. It was decided to focus on DII leads measurements due to the higher variability of measurements in precordial leads.

Statistical Analysis
A repeated measures analysis of variance was used to test the effect of treatments on QTc (within subject factor: time of ECG examination, between subject factor: therapy group). Fisher Exact Test was used to compared the number of patients who exceed the threshold of borderline QTc values in Group 1 and Group 2. Indipendent T-Sample Tests were used to test the differences between baseline QTc values, ages and duration of illness using the statistical package Stata 7.0 (Stata Coorporation, 2001).