Background: Second generation antipsychotics (SGA) induce substantial weight gain but the mechanisms responsible for this phenomenon remain speculative.
Objective: To explore eating behaviors among SGA-treated patients and compare them with nonschizophrenic healthy sedentary individuals (controls).
Methods and Procedures: Appetite sensations were recorded before and after a standardized breakfast using visual analog scales. Three hours after breakfast, a buffet-type meal was offered to participants to document spontaneous food intake and food preferences. Satiety quotients (SQs) were calculated to determine the satiation of both meals and the Three-Factor Eating Questionnaire (TFEQ) was used to document eating behaviors. Body composition and abdominal fat distribution were assessed.
Results: Compared with controls (n = 20), SGA-treated patients (n = 18) showed greater adiposity indices (P ≤ 0.04). Patients' degree of hunger was also higher following the standardized breakfast (P = 0.03). Moreover, patients had significantly higher cognitive dietary restraint, disinhibition, and susceptibility to hunger scores than the reference group (P ≤ 0.05). Disinhibition in the reference group was positively associated with hunger triggered by external cues (r = 0.48, P = 0.03) whereas internal cues seem to mainly regulate emotional susceptibility to disinhibition in patients (r = 0.56, P = 0.02). Higher strategic restraint behavior in patients was associated with decreased satiation right after the buffet-type meal (r = −0.56, P = 0.02).
Discussion: These exploratory findings suggest that patients under SGA seem to develop disordered eating behaviors in response to altered appetite sensations and increased susceptibility to hunger, a factor which may influence the extent of body weight gain triggered by these drugs.
Second generation antipsychotics (SGA) have become widely used as first-line pharmacotherapy for psychosis. One of the main advantages of the second generation of antipsychotics over the first generation is their broad spectrum of efficacy on positive, negative, and cognitive symptoms of schizophrenia as well as their reduced propensity for extrapyramidal side effects (1). Despite these undeniable advantages, the use of SGA has been associated with significant weight gain (2), dyslipidemia such as elevated plasma triglyceride concentrations (3), hyperglycemia favoring new onset of type 2 diabetes, exacerbation of existing type 1 or 2 diabetes, and even diabetic ketoacidosis (4). A recent study has also reported that young SGA-treated patients (n = 87, mean age = 28 years) were characterized by an atherogenic risk profile (5). Moreover, some authors have suggested that weight gain induced by SGA is a major determinant for patients' noncompliance, low self-esteem, and overall poorer quality of life (6) threatening the quality of care and therapeutic alliance. For instance, it has been reported that weight gain may lead to treatment discontinuation, exposing schizophrenic patients to an increased risk of relapse (7).
Weight gain associated with the use of SGA drugs has thus become a source of concern. It is generally believed that there are multiple mechanisms by which SGA may expose patients to positive energy balance although their precise nature remains speculative. From a clinical point of view, SGA-treated patients report modifications in their food habits since the introduction of their drug therapy in response to increased appetite and lack of satiation (8). However, despite these frequent complaints by patients, the effect of SGA on eating behaviors, as potential contributors to a positive energy balance, has not been extensively investigated and remains an issue that is at least partly misunderstood. Therefore, the main objectives of this study were to explore indicators of eating behaviors (appetite sensations, satiation, food preferences and spontaneous intake, cognitive dietary restraint, disinhibition, and susceptibility to hunger) using validated tools among patients treated with a SGA, and to compare them with values of healthy sedentary-untreated individuals (reference group).
Methods and Procedures
This cross-sectional study was conducted on eligible SGA-treated patients recruited from a clinic specializing in first psychoses in the Québec City suburbs, Canada. The Ethics Review Board of the Centre de Recherche Université Laval Robert-Giffard approved the study, and a signed informed consent form was obtained from all participating patients. To be involved in the study, patients had to be treated with a single SGA medication for at least 3 months to allow a washout period of any previous antipsychotic. Exposure to other previous first generation antipsychotics or SGA as well as current concomitant medication for depression, anxiety, and/or mood disorders were not considered as exclusion criteria whereas treatment with drugs altering appetite, satiety, or body weight was considered as an exclusion criterion. Patients had to be aged between 18 and 65 years (outpatients) and had to be sedentary (practicing <30 min of continuous physical activity per week). Eighteen male patients were recruited (clozapine: n = 2, olanzapine: n = 9, risperidone: n = 2, quetiapine: n = 3, and ziprasidone: n = 2). Patients with DSM-IV schizophrenia or other related psychoses were mainly diagnosed by their referring psychiatrist (schizophrenia: n = 15, schizoaffective disorders: n = 2, and delusional disorder: n = 1).
A sample of 20 nonschizophrenic healthy men taking no regular medication was recruited as a reference control group. Participants were selected on the basis of similar age and weekly physical activity level as patients. The research Ethic Review Board of Hôpital Laval approved the study and a signed informed consent form was obtained from each participating subject.
Psychiatric history, including diagnosis, dose of antipsychotic, age at the onset of the psychiatric illness, age at the beginning of antipsychotic treatment, duration of illness, and number of prior psychiatric hospitalizations were recorded for patients and not the reference group. Moreover, psychiatric assessments including Positive and Negative Syndrome Scale (PANSS) (9), Global Assessment of Functioning (GAF) (10), Social and Occupational Functioning Assessment Scale (SOFAS) (11) and Calgary Depression Scale for Schizophrenia (CDSS) (12) were performed.
Anthropometric and adiposity indices
Anthropometric measurements: Body weight, height, and BMI were determined following the procedures of the National Institutes of Health on the standardization of anthropometric measurements (13), whereas waist circumference was measured following the recommendations of van der Kooy and Seidell (14).
Body composition: Fat mass, fat-free mass, percent body fat mass, and percent body fat-free mass were assessed using dual-energy X-ray absorptiometry on a Lunar Prodigy equipment (Madison, WI) (15).
Abdominal adipose tissue distribution: Abdominal adipose tissue areas were performed by computed tomography with a General Electric scanner (Milwaukee, WI) as described previously (16). Subjects were examined in the supine position with both arms stretched above the head. Two computed tomography scans were performed at the abdominal level (lumbar vertebra L2–L3 and L4–L5) using a radiograph of the skeleton as a reference to establish the position of the scan to the nearest millimeter. Total abdominal adipose tissue area was calculated by delineating the surface with a graph pen and then computing the adipose tissue surface using an attenuation range of −190 to −30 Hounsfield units. Abdominal visceral adipose tissue area was measured using the same method. The abdominal subcutaneous adipose tissue area was calculated by subtracting the amount of abdominal visceral adipose tissue from the total abdominal adipose tissue area. Calculations of the partial volume of visceral and subcutaneous adipose tissue were carried out by multiplying the distance between the two slices by their average areas, as described previously by Paré et al. (17).
All participants were evaluated after a 12-h overnight fast. They were then asked to eat a standardized breakfast designed to provide 14% of dietary energy from proteins, 40% from carbohydrates, and 46% from lipids with an energy content of 835 kcal (food quotient: 0.83). Breakfast consisted of white bread (100 g), butter (6 g), peanut butter (32 g), cheddar cheese (40 g), and orange juice (250 g). Participants were asked to eat the whole breakfast, which was eaten within a 30-min period, under standardized conditions.
Food preference test and spontaneous intake
A cold buffet-type meal composed of a variety of foods was offered to participants ∼3 h after the standardized breakfast to measure macronutrient preferences and spontaneous intake under conditions reproducing free-living conditions. Portions of each food were larger than the expected intake. A large diversity in protein, lipid, and carbohydrate sources was available in order to facilitate the detection of macronutrient preferences: dairy products (gruyere, mozzarella and cottage cheeses, yogurt, partly skimmed milk), fruits and vegetables (lettuce, fresh tomatoes, carrot, orange, apple, unsweetened orange juice), grain products (white and wheat kaiser breads, soda crackers), meat and alternatives (cooked ham and turkey), fat food products (butter, mayonnaise, salad dressing, salmon mousse, liver pate, plain potato chips), refine sugar products (shortcake butter and chocolate cookies, soft drinks, ketchup), and food with no energy (mustard, water). They had ad libitum access to food and were instructed to eat a normal meal, within a 30-min period, until they had reached satiety. To determine macronutrient preferences and spontaneous intake from the buffet, food was weighed before and after the buffet to quantify the intake of each food. Energy, protein, lipid, and carbohydrate intakes were calculated using the Canadian Nutrient File (18) and/or information on food labels.
Before, immediately after, and every 10 min for a period of 1 h after the standardized breakfast test, as well as immediately before and after the buffet, participants were asked to record their appetite sensations such as “hunger” and “prospective food consumption” using visual analog scales (19). Subjects were asked to indicate by a vertical line, on a scale of 0–150 mm, how they felt at the moment of completing the following questions: how hungry do you feel? (not hungry at all-as hungry as I ever felt) and how much food do you think you could eat? (nothing at all-a large amount). Participants were also asked to evaluate their appreciation of the meal. The appetite sensation responses to the standardized meal test were evaluated by calculating the area under the curve for the 1-h postprandial period.
Satiating effect (satiety quotient)
Appetite sensations measured before and after a meal can also reveal information about the satiating capacity of food, which can be expressed as satiety quotient (SQ). Kissileff first introduced the satiating efficiency of food (20), which aimed at measuring the extent to which a meal could reduce subjective appetite sensations per unit of energy intake (e.g., kcal, kJ). Green et al. (21) extended this concept by taking into account the temporal effect of foods and suggested calculation of the SQ for each appetite sensation measured over time. Differences in SQ for hunger sensation have been observed between meals of fixed energy content (21) (like the standardized breakfast) or ad libitum meals (like the buffet) (21). SQ (mm/kcal) represents the effect of food intake on appetite sensations relative to energy intake. Calculated at various time points following consumption, the SQ gives a temporal profile of the satiating efficiency of the breakfast. The higher the quotient, the greater the satiating effect of a meal. SQ was calculated as follows:
SQ was calculated immediately after standardized breakfast (T: 0 min) as well as every 10 min over a 60-min period following the meal (T: 10, 20, 30, 40, 50, and 60 min). To evaluate the mean postprandial period satiating effect of standardized breakfast on satiety parameters (hunger and prospective food consumption), all time point quotients (0–60 min) for a given parameter were averaged. SQ was also determined immediately after the buffet-type meal.
Three-Factor Eating Questionnaire
The Three-Factor Eating Questionnaire (TFEQ) is a 51-item questionnaire developed by Stunkard and Messick (22) and is used to assess three behaviors associated with eating: cognitive dietary restraint, disinhibition, and susceptibility to hunger. Specifically, cognitive dietary restraint is the intent to restrict food intake to control body weight (21 items; score ranging from 0 to 21). Disinhibition is an overconsumption of food in response to a variety of stimuli, such as emotional stress, associated with a loss of control on food intake (16 items; score ranging from 0 to 16). Finally, susceptibility to hunger refers to food intake in response to feelings and perceptions of hunger (14 items; score ranging from 0 to 14). More specific subscores, suggested by Bond et al. (23), for these three general eating behaviors were determined. First, cognitive dietary restraint has been divided into strategic dieting behaviors, which are specific behavior that might be used to control weight (four items; score ranging from 0 to 4), attitude to self-regulation, which assesses subject's overarching perspective on eating and weight control (five items; score ranging from 0 to 5), and avoidance of fattening food, which is predominantly behavioral (four items; score ranging from 0 to 4). Disinhibition is divided into three subscores (habitual, emotional, and situational susceptibility to disinhibition) (23). Habitual susceptibility to disinhibition describes behaviors that may occur when circumstances could predispose to recurrent disinhibition (five items; score ranging from 0 to 5). Emotional susceptibility to disinhibition defines a behavior that is associated with negative affective states (three items; score ranging from 0 to 3), and situational susceptibility to disinhibition refers to overeating initiated by specific environmental cues (five items; score ranging from 0 to 5). Finally, susceptibility to hunger is divided into two specific subscores (internal and external locus for hunger) (six items for each subscore; ranging from 0 to 6) (ref. 23). Internal hunger refers to the type of hunger that is interpreted and regulated internally, whereas external hunger is triggered by environmental cues.
Differences between patients and the reference group for continuous variables were analyzed using Student's unpaired t-test (adiposity indices, food preference test, appetite sensations, SQ, and eating behaviors according to the TFEQ). A regression model was used to adjust differences observed between the two groups for appetite sensations (24) and TFEQ scores (25) for the confounding effect of obesity (adjustment for BMI). Spearman correlation coefficients were computed to quantify relationships between adiposity indices, food intake from the buffet, and SQ with eating behaviors scores. Results were considered significant with P values ≤0.05. All analyses were performed using the SAS statistical system, version 8.0 (SAS Institute, Cary, NC).
Patients were not on their first antipsychotic treatment. Overall, they were treated since 35.3 ± 22.0 months with other previous first generation antipsychotics or SGA. However, study patients were on their current SGA for at least 3 months (mean treatment duration of current SGA: 24.6 ± 19.7 months). Moreover, patients were generally considered clinically stable according to scores obtained with the PANSS (61 ± 13), GAF (58 ± 16), SOFAS (61 ± 15), and the CDSS (2 ± 2) psychiatric scales.
Table 1 shows that for a similar age, SGA-treated patients were overweight (BMI ≥ 25.0 kg/m2), compared with the reference group, and were characterized by abdominal obesity, as assessed by waist circumference ≥100 cm and confirmed by computed tomography scan. Indeed, study patients showed significantly more total abdominal tissue volume as well as an increased visceral adipose tissue volume when compared with the reference group.
Table 1. Adiposity indices of second generation antipsychotic-treated patients and reference group
As demonstrated in Table 2, patients had markedly higher cognitive dietary restraint, disinhibition and susceptibility to hunger scores compared with the reference group. Patients scores for strategic dieting behaviors, habitual and emotional susceptibility to disinhibition, and susceptibility to hunger triggered by either internal or external cues were at least twofold higher as compared with the reference group. Of note, all differences detected between the two groups regarding TFEQ scores remained significant after adjustment for the confounding effect of obesity (adjustment for BMI).
Table 2. Eating behaviors according to the TFEQ of second generation antipsychotic-treated patients and reference group
It is shown in Table 3 that, among patients, susceptibility to hunger (r = 0.49; P ≤ 0.05) as well as susceptibility to hunger triggered by internal locus scores (r = 0.56; P ≤ 0.05) were positively associated with emotional susceptibility to disinhibition, and similar trends were observed for disinhibition total score as well as for habitual susceptibility to disinhibition, whereas no relationship was noted for susceptibility to hunger triggered by external locus. Among the reference group, susceptibility to hunger total score and its subscores for either internal or external locus were positively correlated with total disinhibition and its subscores.
Table 3. Intercorrelations between susceptibility to hunger and its subscores and the other eating behaviors according to the TFEQ of second generation antipsychotic-treated patients and reference group
Subsequently, intercorrelations between TFEQ's subscores and adiposity indices were performed (data not shown). Among patients, higher cognitive dietary restraint was associated with lower body weight (r = −0.42; P = 0.08), but did not reach statistical significance. No intercorrelation between TFEQ's restraint subscores and adiposity indices was observed among healthy control subjects.
Appetite sensations after standardized breakfast
As illustrated in Figure 1a, patients' degree of hunger differed from the reference group after consumption of the standardized breakfast. Indeed, after adjusting for baseline values, the degree of hunger between the two groups was statistically higher for SGA-treated patients compared with reference control subjects at time 10, 20, and 60 min after the meal as well as when measured as postprandial area under the curve (P = 0.03). Thus, to better evaluate the satiating efficiency of the standardized breakfast on the degree of hunger, SQ was calculated every 10 min over a 60-min period after the meal. As shown in Figure 1b, SGA-treated patients had a significantly lower satiating effect every 10 min for a 60-min period after the standardized breakfast compared with the reference group (P ≤ 0.02). Moreover, the mean postprandial satiating efficiency of standardized breakfast on the degree of hunger was also statistically different between the two groups (P = 0.002), suggesting that patients were overall less physiologically satisfied by the standardized breakfast. These differences between the two groups remained significant even after adjustment for BMI (degree of hunger: (P = 0.03), satiating postprandial effect for degree of hunger: (P = 0.002)).
Appetite sensations, macronutrient preferences, and spontaneous intake at the buffet-type meal
Table 4 shows that no significant difference for macronutrient preferences and spontaneous energy intake was observed between the two groups during the ad libitum buffet-type meal. Moreover, as opposed to the results obtained after the standardized breakfast with respect to the satiating efficiency of a meal, no difference was found immediately after the ad libitum buffet between the two groups. Indeed, SQ for prospective food consumption tended to be lower among the reference group. Intercorrelations between TFEQ's subscores and macronutrient preferences and spontaneous intake from the buffet were also performed. Results revealed that, among patients, higher avoidance of fattening food was associated with lower energy, absolute lipid, and carbohydrate intakes from the buffet (−0.55 ≤ r ≤ −0.57; P ≤ 0.02). In the same order of idea, it is illustrated in Figure 2 that, among patients treated with SGA, lower satiation from the buffet (lower SQ for prospective food consumption) was associated with higher strategic dieting behavior (r = −0.56; P = 0.02), whereas a positive correlation was observed in the reference group (r = 0.48; P = 0.03). However, a trend was noted with disinhibition score and SQ for prospective food consumption among SGA-treated patients (r = −0.40; P = 0.1) while no relationship was observed for the reference group.
Table 4. Macronutrient preferences and satiety quotient (SQ) from buffet of second generation antipsychotic-treated patients and reference group
SGA-treated patients report altered eating behaviors since the introduction of their pharmacological therapy when they are seen in a clinical setting. The main purpose of this study was to investigate this issue by documenting variations in appetite sensations, satiation, macronutrient food preferences and spontaneous intake, daily dietary profile, cognitive dietary restraint, disinhibition and susceptibility to hunger of 18 SGA-treated patients, who were compared with 20 sedentary healthy untreated-individuals. Using validated tools commonly used in the scientific community, our results showed that patients under SGA were characterized by a cluster of impaired eating behaviors as compared to the reference group, a clinical factor which obviously may lead to a positive energy balance.
Results demonstrate that SGA-treated patients presented impaired appetite sensations under standardized conditions. Indeed, the degree of hunger, as recorded with visual analog scales, was higher over the hour following the consumption of a standardized breakfast. Furthermore, the assessment of satiating efficiency using the SQ developed by Green et al. (21) confirmed that SGA-treated patients were less physiologically satiated over time after a given meal. Therefore, results obtained during the postprandial period after standardized breakfast suggest that a given energy-fixed meal provides a less powerful satiating efficiency among patients treated with SGA when compared with control subjects.
Therefore, it was also important to demonstrate that SGA-treated patients were characterized by a cluster of altered eating behaviors, in order to explore some behavioral mechanisms as potential contributors to a positive energy balance under free-living conditions. Results obtained using the TFEQ (23) confirmed that patients' susceptibility to hunger, triggered by either internal or external cues, was two times higher than that of the reference group. Patients were also characterized by higher disinhibition total score, defined as overconsumption of food in response to a variety of stimuli, associated with a loss of control of food intake. In both groups, the presence of susceptibility to hunger total score was associated with increased total and subscores of dietary disinhibition but these relationships seemed to be driven differently i.e., in response to different locus. Indeed, among the reference group, total and subscores of disinhibition tended to be positively associated with hunger triggered by either internal or external cues whereas among patients, internal cues seem to mainly control episodes of disinhibition. These results suggest that, among patients, episodes of overconsumption of food might be driven in response to impaired and/or lack of control of appetite sensations.
The presence of such higher scores for susceptibility to hunger and tendencies for recurrent episodes of dietary disinhibition among SGA-treated patients, when compared with reference subjects, are concordant with the SGA-induced weight gain hypothesis, based on classical neurotransmitters serotonin and histamine because both have a well-confirmed role in the control of appetite. The affinities of antipsychotic agents for a large series of receptor types were evaluated in humans for correlation with antipsychotic-associated weight gain. Results of this trial revealed that H1 receptor blockade represented the strongest correlation with body weight gain (26). In that sense, it is possible to imply that the effects of SGA on appetite sensations, as documented in this study for the first time with validated indicators, might be triggered by their interaction with brain receptors and may influence energy intake as well as the excessive body weight gain observed in this population.
Therefore, it was surprising to detect no difference between the two groups in food preferences and spontaneous intake from the ad libitum buffet-type meal. However, the presence of higher disinhibition scores may explain, at least in part, this phenomenon. Indeed, it has been suggested by Kissileff that highly disinhibited subjects tend to behave differently in response to instructions when evaluated in laboratory. Indeed, if instructed to eat a normal meal, subjects tend to restrict their intake whereas they eat significantly more when asked to eat without limit or “binge” (27). In this study, the presence of higher susceptibility to hunger and dietary disinhibition scores observed among patients was also accompanied by higher levels of cognitive dietary restraint. In patients, the intensity of dietary restraint was associated with lower adiposity indices and overall energy consumption, as assessed by the ad libitum buffet-type meal test in which patients were instructed to eat a normal meal. Therefore, it seemed that patients who were able to restrain their food intake ate less and were leaner. Thus, we may speculate that patients recently treated with SGA may have developed restricted eating behaviors as a mechanism of defense in response to altered appetite sensations, increased susceptibility to hunger and disinhibition, thus limiting body weight gain. Results of this study observed among SGA treated-patients are concordant with previously reported cross-sectional findings performed in the general population, demonstrating that intentional suppression of food intake (restraint) was associated with lower total energy intake (28,29,30,31) and adiposity indices (25,26,27,28,29,30,31,32,33,34).
However, conversely to the reference group, higher strategic dieting behavior, i.e., a more flexible and gradual behavioral approach to eating, dieting, and weight (23), was associated with lower satiating efficiency of the ad libitum buffet-type meal (decreased SQ calculated immediately after the consumption of the meal) in SGA-treated patients. These results suggest that patients presenting higher levels of cognitive food intake control are those who eat less, but are also those who are less satiated by a meal offered under free-living conditions, a factor which may obviously lead to a positive energy balance. Indeed, the achievement of constant cognitive suppression of food intake over time is hardly feasible and this is concordant with some aspects of the dieting counter-regulation phenomenon described by Herman and Polivy (as reviewed by Polivy (35)). In brief, according to this paradigm, food deprivation that occurs among dieters (achieved with cognitive dietary restriction) appears to produce a tendency to overeat, which constitutes a potential explanation of unsuccessful long-term dieting interventions among restrained eaters. These eaters display a tendency toward excessive eating or even bingeing when restrictions are lifted.
In this study, the duration of overall antipsychotic treatment (including all previous first generation antipsychotics or SGA therapy) was recorded and was associated with increased weight (r = 0.46; P = 0.05), BMI (r = 0.54; P = 0.02), and waist circumference (r = 0.59; P = 0.01), as well as with decreased internal susceptibility to hunger (r = −0.49; P = 0.04). Moreover, the duration of current SGA therapy was associated with decreased cognitive dietary restraint (r = −0.50; P = 0.03) and strategic dieting behavior (r = −0.52; P = 0.03) as well as with decreased emotional susceptibility to disinhibition (r = −0.62; P = 0.006). Therefore, it is realistic to speculate that the second generation therapy has an effect on adiposity indices over time which may be attributable to an increased food intake as a result of a decreased intentional suppression of food intake. In this regard, early nutritional interventions should be done among SGA-treated patients to limit compensatory eating behaviors, such as high level of cognitive dietary restraint, and potential subsequent body weight gain.
The limitations of this study suggest that future research should also consider the possible effect of psychotic disorders in general rather than SGA treatment. Indeed, a rather scarce literature on schizophrenic eating habits describes exaggerated preference for unhealthy palatable nutrition, poor in fiber, in fruits and in vegetables but excessive in saturated fat and carbohydrates (36,37,38,39,40). These patterns may be worsened further by treatment with SGA. Therefore, a different study design would, of course, be needed to examine this aspect such as a between-subjects design comparing psychotic patients on SGA medications to first generation antipsychotics. Moreover, although some analyses from this study were adjusted for the confounding effect of obesity, a larger sample would have permitted to perform BMI-matched analyses to eliminated the possible confounding effect of obesity.
In conclusion, weight maintenance is hardly feasible in a context of altered eating behaviors. Indeed, those eating behaviors contribute to the greater adiposity indices observed among SGA-treated patients and to obesity-related health risk. This study is a first step in the understanding of the clinical mechanisms underlying the SGA-induced altered eating behaviors which may lead to a positive energy balance and contribute to body weight gain. Such findings are essentials to investigate how these molecules influence energy balance in order to properly intervene among mentally ill patients whose weight gain is linked to a life-saving drug therapy.
A.T. is partly funded by the Canada Research Chair on Physical Activity, Nutrition, and Energy Balance. M.-A.R. is Director of the Clinical Research Unit of the Centre de Recherche Université Laval Robert-Giffard in Québec City. R.-H.B. is Director of an early psychosis clinic, the Clinique Notre-Dame-des-Victoires, in Québec City. M-E.J. is recipient of a scholarship from the Social Sciences and Humanities Research Council of Canada, and M.B. is recipient of a scholarship from the Fonds de la recherche en santé du Québec. We thank the staff of the clinic specializing in first psychoses, the Clinique Notre-Dame-des-Victoires, as well as the staff from the Hôpital Laval Research Centre for their dedicated work during data collection.