Ten‐year course of treated bipolar I disorder: The role of polarity at onset

Abstract Introduction Early‐stage predictors of illness course are needed in bipolar disorder (BD). Differences among patients with a first depressive versus maniac/hypomanic episode have been stated, although in most studies, memory bias and time from onset to start of specialized treatment might interfere. The aim was to compare the first 10 years of illness course according to polarity at onset. Methods 49 type I BD patients admitted for treatment for a first‐time affective episode and a following 10‐year attendance to the institution were included. A retrospective year by year comparison according to polarity at onset (depressive (DPO) or maniac (MPO)) was performed. Cramer's V and Cohen d were computed to determine effect size. Results 59.2% (n = 29) started with MPO. Both groups were similar in demographic and social outcome characteristics, clinical features, and treatment variables. Patients with DPO reported more depressive episodes than MPO patients (U = 149.0 p < .001, Cohen's d = 0.87); both groups had a similar number of manic episodes. Only during the first year of follow‐up, suicide attempts (SA) were more frequent in patients with DPO while the presence of a psychotic episode and psychiatric hospitalizations were more frequent in the MPO group. Conclusion According to these findings, it can be concluded that illness onset is only indicative of depressive predominant polarity but is not related to other poor prognostic variables after the first year of illness onset, in treated BD. SA in the first year of an affective disorder could represent a marker of BD.


INTRODUCTION
Bipolar disorder (BD) is a chronic and debilitating mental illness, present in approximately 2.4% of the world population (Merikangas et al., 2011). BD represents the sixth cause of disability worldwide and conveys a poor prognosis due to functional impairment and the presence of residual symptoms (Judd, 2002;Tohen et al., 2003).
Polarity at illness onset is defined as the pole (depressive or manic/hypomanic) at which a bipolar patient presents his or her first affective episode. Due to the high clinical heterogeneity of BD and its poor prognosis, polarity at illness onset has been studied as an early predictor of illness course in BD (Cremaschi et al., 2017;Daban et al., 2006;Etain et al., 2012;Forty et al., 2009;Perlis et al., 2005;Perugi et al., 2000;Tundo et al., 2015). The importance of studying polarity at illness onset is grounded on the possibility of a "glimpse" into the future course of illness, which may in turn guide clinicians in developing treatment and secondary prevention strategies at early stages of disease.
Polarity at illness onset was shown to be associated with predominant polarity, defined as the polarity of two thirds of total episodes (Carvalho et al., 2014;Daban et al., 2006;Perugi et al., 2000;Tundo et al., 2015). Patients with depressive first-episode polarity (depressive polarity at onset, DPO) have shown a more chronic course of illness with greater number and longer duration of depressive episodes; whereas patients with maniac polarity at onset (MPO) tend to have more manic or hypomanic episodes Etain et al., 2012;Forty et al., 2009;Perlis et al., 2005). The importance of the latter relies in the functional impairment that accompanies a greater number of episodes.
In terms of poor prognostic variables, DPO has been associated with more suicide attempts. On the other hand, MPO has been associated with a higher prevalence of psychotic symptoms and a greater number of hospitalizations (Cha et al., 2009;Chaudhury et al., 2007;Daban et al., 2006;Neves et al., 2009;Perlis et al., 2005;Perugi et al., 2000).
Previous studies have been of great value at identifying associations between polarity at illness onset and poor prognostic variables.
Nonetheless, they lack control over important variables: reported polarity at onset was based on recollection from the patient; disease duration at study entry was different between studied patients; disease follow-up also differed from one patient to another; report of number, polarity and hospitalizations of subsequent episodes was mostly obtained through interviews; patients were not treatment naïve at study entry (Table 1). The importance of controlling these variables rests on the fact that the number of episodes and poor prognostic variables may be influenced by the duration and management of disease and not by the polarity of onset itself. Also, data collection from clinical interviews, instead of the objective procurement of information from clinical records, entrails recall memory bias regarding the number and severity of episodes (Martino et al., 2016).
Given these limitations in our current knowledge on the possible predictive capability of polarity at onset, we aimed to compare long-term clinical variables in BD according to illness polarity in the first episode of disease (manic or depressive), in patients with an institutional follow-up of their first 10 years of illness. We chose to assess variables from three major areas: (1) current social outcomes such as employment, years of education and marital status; (2)

METHOD
This study was approved by the Ethics Committee of the National Institute of Psychiatry at Mexico City with number CEI/C/018/2016. Participants gave their written consent for inclusion.

Subjects
Patients were identified from the Affective Disorders' Clinic at the National Institute of Psychiatry Ramón de la Fuente Muñíz (INPRFM) in Mexico City, a highly specialized psychiatric facility dedicated to research, training, and inpatient and outpatient treatment for psychiatric conditions. The Affective Disorders' Clinic is composed of treating psychiatrists specialized in bipolar disorder who follow international guidelines for the treatment of BD (Bandelow et al., 2012;Yatham et al., 2018) with an individualized evidence-based treatment.
The present study included patients that had been admitted to the National Institute of Psychiatry for a first-time mood episode, defined as mood symptoms sufficient to fulfill DSM criteria for either a depressive, a hypomanic or a manic episode (onset of first mood episode had occurred at most 2 years before admittance and had remained untreated during that period) and had from then-on continued to attend this institution for at least 10 years, at a rate of at least one consultation a year in those 10 years, so memory bias could be significantly reduced (Martino et al., 2016). Additionally, patients who had received an initial diagnosis different from BD, during follow-up were changed to bipolar disorder I, and from then-on this remained the principal diagnosis according to clinical records. Only the first 10 years of institutional attendance were analyzed. Clinical records from a total of 116 cases with a first mood episode and current BD I diagnosis were reviewed. Sixty-seven cases failed to have at least one consultation a year during the 10-year period and were excluded from the study, therefore, 49 patients were included in the analysis. BD remained the principal diagnosis).

Retrospective measurement
The following data from each individual's first 10 years of institutional attendance was gathered from medical records: -Current demographics (age, gender) and social outcomes (

Statistical analysis
All data were analyzed using SPSS version 21. Data is presented in frequencies and percentages for categorical variables and means and standard deviations (SD) for continuous variables.   The presence of a suicide attempt was more frequent in patients with DPO only during the first year (Cramer's V = 0.32) and a tendency to significant differences was observed in the third year, with similar percentages reported in both groups in the following years ( Figure 2). The presence of a psychotic episode (Figure 3  However, we believed avoiding memory bias in the recollection of the course of BD and having precise information on the first 10 years of illness evolution in every patient was necessary to establish the association between polarity at onset and illness course.

Demographic characteristics, clinical features, and current social outcomes
As found in previous studies, polarity at onset was associated with the polarity of subsequent depressive episodes (Carvalho et al., 2014;Colom et al., 2006;Perugi et al., 2000;Tundo et al., 2015): DPO patients had more depressive episodes than MPO patients, while there were no differences regarding the total number of manic episodes. This finding may serve as an indicator that DPO patients need more intensive prevention and management of depressive episodes to diminish the burden and impairment that come alongside. Indeed, an outstanding third of DPO patients had a suicide attempt in the first year of illness onset. The finding that a lifetime suicide attempt is higher in DPO patients than in MPO patients has been reported in several studies (Schaffer et al., 2015), and this study confirmed that the percentage of suicide attempts is higher year by year in the DPO group, although both groups have an important decrease after the third year. Although suicide attempts are complex phenomena, we believe that a suicide attempt in the first year of an affective disorder could be a marker of BD as a rate over 30% is much higher than the suicide-attempt rate reported in the onset of other clinical entities: González-Pinto et al.
(2007) reported a suicide attempt in 8% of their sample at enrollment in their study of first-episode affective and non-affective psychotic inpatients; Shen et al. (2019) reported that 20.1% of their sample had a history of suicide attempts in their study of drug naïve patients with major depressive disorder, although these were not first-episode patients.
Furthermore, this rate is certainly much higher than that reported in general population (Centers for Disease Control and Prevention (CDC), 2005 ). However, conclusions on this matter will only be drawn with further studies using the proper methodology.
The difference in psychotic symptoms between groups was notable in the first year, with a much higher prevalence among MPO, present in over half the patients, but from that point on, differences tended to disappear. One should consider that it is very likely that only the most severe presentations of BD reach institutional management in their first year of illness evolution (Dagani et al., 2017) and this may account for such severe illness onset (suicide attempts in DPO and psychosis in MPO).
Hospital admittance is more prevalent in patients with a manic initial episode, despite the high prevalence of suicide attempts in the DPO group; we assume that this is due to the fact that manic episodes tend to be more disruptive and is congruent with other findings (Atigari et al., 2015), however, it could represent a red flag for treating physicians, as the severity of depression could be being underestimated.
After the first year, there were no differences in terms of poor prognostic variables such as number of hospitalizations and psychotic symptoms between groups. This finding is the opposite from what has been described by other studies (Azorin et al., 2011;Baldessarini et al., 2014;Cha et al., 2009;Chaudhury et al., 2007;Cremaschi et al., 2017;Daban et al., 2006;Etain et al., 2012;Forty et al., 2009;Garcia-Lopez et al., 2009;Kassem et al., 2006;Neves et al., 2009;Perlis et al., 2005;Perugi et al., 2000;Tundo et al., 2015). We believe this could be due to differences in study methods, mainly the lack of memory bias in this study, but could also be attributed to illness course modification due to treatment: in this study, all patients were treatment naïve when institutional attendance started, but from then on, psychiatric treatment was ongoing. Treatment in psychiatry includes, but is not restricted to pharmacological therapy (Akiskal and Tohen (2011) There are several limitations in this study. Most limitations derive from the retrospective methodology used in the study: we had to exclude from analysis unreliable variables, many of which would have given a better understanding of the phenomenon (e.g., response to lithium; evolution of pharmacological treatment; certainty in the presence or absence of comorbid disorders; information regarding the patient's agreement for hospitalizations, among others). This was mainly due to substantial differences in clinical records as our study covers a time span of 30 years (the first 10 years of evolution for each patient, the first being admitted in 1991 and the last in 2009): policies, treating physicians, guidelines and available therapeutic options changed considerably during this time. However, many interesting variables were reliable, notably the number and polarity of episodes during these years, which we believe adds important information to our current knowledge on polarity in type I BD.
Also, it is difficult to generalize these findings as this sample centered on type I BD. The inclusion of patients with type II BD was originally intended. However, the sample reached for BD II was too small (n = 9) to allow for comparisons between BD I and BD II. We decided to sacrifice the representativeness in order to gain methodological strength.
The fact that this is a population of treated BD and one with excellent adherence, also affects generalizability. These characteristics make this population less vulnerable to severe clinical outcomes and relapses due to their strict medical supervision and probably due to patient's insight (de Barros Pellegrinelli et al., 2013). It is also very probable that these patients represent a population with a severe form of disease onset, and probably a very effective support network as all reached a highly specialized facility in the first year, which is very uncommon: a several years delay in specialized treatment for psychiatric disorders is sadly the rule rather than the exception, and seems to be a worldwide problematic (del Valle et al., 2017;Fikretoglu et al., 2010;Goldberg et al., 2019;Green et al., 2012;Ki et al., 2014;Stagnaro et al., 2019).
The small sample size must also be mentioned, although one must consider the difficulty in fulfilling inclusion criteria, having the prior statement in consideration.
Yet another limitation that must be stated is related to differences in diagnostic criteria between DSM-IV and DSM-5 (Kessing et al., 2021): as previous criteria of DSM were followed for the detection of polarity at onset, we cannot rule-out the possibility that some patients might have had a first hypomanic episode of short duration as their first affective episode. Little is known on the subset of patients with short duration hypomanias (Miller et al., 2016), and the specific question on the evolution from a first affective episode of these characteristics must be addressed in future studies.
To our knowledge, this is the first study that evaluates the illness course of treated bipolar disorder year by year in the first 10 years of illness evolution and compares it according to polarity at illness onset.
The main strength of the present study relies on the fact that included patients started the follow-up at disease onset, therefore reducing the possibility of recall bias and the possible confounding effect of untreated BD. Another important aspect of the studied population is the control over disease duration since all patients were followed during their first 10 years of illness course (with at most 2 years variation).
In conclusion, our study shows that in treated BD, illness onset is only indicative of depressive predominant polarity but is not related to other poor prognostic variables; adds evidence of the effectiveness of psychiatric treatment in this disorder; and highlights suicide attempts in the first year of an affective disorder as a possible marker of BD. Further longitudinal studies in different populations are needed to allow for generalization of these findings as well as comparisons between suicide attempts in the first year of other affective disorders.

DATA SHARING
The data that support the findings of this study are available from the corresponding author upon reasonable request.

TRANSPARENT PEER REVIEW
The peer review history for this article is available at https://publons. com/publon/10.1002/brb3.2279