DOMINO study was a non-interventional, retrospective, observational, multicenter study involving patients diagnosed with schizophrenia and aimed at evaluating the rate of persistence with AOM over a 6-month period, along with the influence of baseline demographic and clinical variables on the likelihood to be persistent [3]. The present study (DOMINO II) reports on a longer term follow-up, over a mean period of 48.8 months from AOM initiation, involving the patients who were persistent for 6 months or longer in DOMINO study. Our primary objective was to evaluate the long-term persistence with AOM treatment, as defined as being still on AOM at the time of the last follow-up observation. The secondary objectives included the evaluation of efficacy and tolerability. The study was approved by the Ethical Committee/Institutional Review Board at each recruitment site. Written informed consent was obtained from all participants or their legal guardians. DOMINO study [3] was conducted between June 1st 2015 and July, 11th 2017. All patients presenting for a visit, who were at least 18 years of age, were diagnosed with schizophrenia, and had started ALAI (at least 1 injection) at least 6 months before the inclusion visit, were continuously recruited. Retrospective information from the start of A-LAI treatment (index date, baseline timepoint) until the follow-up visit (inclusion visit) was collected from A-LAI start to the follow-up visit (inclusion visit). The present DOMINO II study, extended the observation period to November 23, 2020.

The study was conducted between June 1, 2015 and November 23, 2020, in 16 clinical sites in Italy. All patients were at least 18 years and carried a diagnosis of schizophrenia, as established via a clinical interview conducted to verify the presence of the Diagnostic and Statistical Manual for Mental Disorders, 5th Edition (DSM 5) criteria.

Persistence was defined as receiving AOM at the last follow up visit and measured by time to all-cause AOM discontinuation and was assumed to reflect AOM’s efficacy, safety, and tolerability from both patient and treating health care provider (HCP)’s perspectives. Non-persistence was declared if the patient missed, for any reason (discontinuation, loss to follow up or skipped AOM doses), at least 2 consecutive or 3 non-consecutive injections of AOM. A missed dose of AOM was defined as a lapse of more than 45 days from the previous AOM injection. Electronic and paper clinical records were reviewed and treating HCP were interviewed as per the date of AOM interruption or discontinuation and reasons, when applicable.

The analyses were performed using Statistical Package for Social Science (SPSS-IBM, version 21.0 or later). Level of significance was set at a p value < 0.05 for two-tailed hypothesis. The “Full Analysis Set” (FAS) population was used to evaluate the primary objective. The FAS population included all patients persistent to AOM at the end of the DOMINO study observational period (6 months) for whom the information about the primary variables (long term persistence with AOM treatment) was available. The continuous variables were reported by sample statistics: n (number of observations), number of missing data, mean, standard deviation (SD), minimum, first quartile (Q1), median, third quartile (Q3), and maximum. The summary statistics mean, median, standard deviation, minimum, maximum, Q1 and Q3 were presented to one more decimal places as that used to collect the data.

Treatment persistence was defined as the proportion of patients who maintained the same medication (AOM) in a specific period. Treatment persistence might be considered as a proxy for treatment efficacy and safety, given that lack of efficacy or poor tolerability are among the main reasons for discontinuation. As in DOMINO study, non-persistence was declared if a patient missed 2 consecutive or 3 non-consecutive injections of AOM during the retrospective follow-up period. Treatment persistence was evaluated using Kaplan Meier survival method. Person-years at risk were computed from the beginning of AOM treatment to the point of treatment discontinuation, end of the study (last retrospective assessment, scheduled to be completed by the end of November 2020), or loss to follow-up, whichever occurred first. The event for “discontinuation” and the determination of RR was the interruption of AOM treatment. Patients lost to follow-up were considered as censored. The starting time for the Kaplan Meier analysis was the index date (date of AOM initiation).

Using a Cox Regression model demographic and clinical characteristics were evaluated for their effect on the long-term persistence to AOM. The variables included in the model were sourced from medical records or directly verified with the patients: age, gender, marital status, education, occupation, living situation and family support, number of previous acute episodes, schizophrenia, comorbidities, history of non-adherence in the 3 months prior to the index date, alcohol and/or drug abuse, reason to initiate AOM treatment, last antipsychotic treatment prior to AOM, CGI-S at index date and starting AOM dose.

Prior to the main analysis of the primary objective, univariate analyses (Chi Square tests, Student T tests or Mann–Whitney tests, as applicable) were performed to test the association between demographic and clinical characteristics of patients and long-term persistence with AOM (each baseline variable was compared between patients still in treatment at the end of the long-term follow-up period and patients that interrupted/discontinued the treatment before long term follow-up visit).

Univariate and multivariate Hazard Ratios (HR) with 95% Confidence Intervals (CIs) were calculated by regression model in which demographic and clinical characteristics were the independent variables and persistence was the dependent variable. The Cox Regression model produces a survival function that predict the probability of the discontinuation (event of interest) occurrence at a given time t for given predictor variable values. The shape of the survival function and the regression coefficients for the predictors were estimated from observed subjects. Cox’s semi-parametric model was used in the data analysis to explain the effect of explanatory variables on hazard rates.

The FAS population was used to evaluate all the secondary objectives. The analysis of the secondary objectives was essentially descriptive. Data were presented using summary statistics. All categorical variables were summarized in frequency and percentage. The continuous variables were reported by sample statistics: n (number of observations), number of missing data, mean, SD, minimum, Q1, median, Q3, and maximum. The summary statistics mean, median, standard deviation, minimum, maximum, Q1 and Q3 were presented to one more decimal places as that used to collect the data.

CGI-Severity score (index date, DOMINO inclusion visit date, long term follow-up visit date) was compared using the non-parametric Wilcoxon Signed-Rank test. The test is a non-parametric analogous of the one-sample t test. This test can be used to make inferences about a population mean or median, without requiring the assumption of normally distributed data. In addition, all CGI–severity score data collected for each patient after the index date, were analyzed descriptively.