Cognitive impairment is common in patients with bipolar disorder (BD) and is associated with poorer functional outcomes [
1‐
5] and longer time to recovery after a first episode [
6]. Over time, increasing illness burden appears to be associated with increased neurocognitive dysfunction [
7]. Despite strong associations between cognitive impairment and functional outcomes, treatment for these symptoms is inadequate. Medications produce modest overall improvement in cognitive symptoms, and may even worsen some aspects of cognition [
8‐
11]. As few as one third of patients achieve functional recovery over time [
12]. At least partial disability is reported in approximately 80 % of patients with BD; as many as 65 % of patients report being unemployed even after clinical recovery, and patients experience significant disability in daily living and social functioning [
13]. As cognition is among the strongest predictors of functional outcomes (e.g., [
3,
4]), and consistent with a European College of Neuropsychopharmacology expert report [
13], it is critical to target this key symptom dimension in patients with BD [
1,
4,
12,
13].
Cognitive deficits in bipolar disorder
Patients with BD exhibit deficits in memory, executive function, and processing speed/efficiency that persist during euthymic phases and over time [
14‐
18]; patients with BD with a history of psychosis (BDP) may be especially cognitively impaired [
14,
19,
20]. Cognitive remediation (CR) is a promising treatment approach addressing neurocognitive dysfunction in the hopes that improved cognitive performance will translate to better community outcomes. A meta-analysis of 40 studies of CR for adults with schizophrenia (SZ) found that these programs produced moderate, durable effects on cognitive functioning [
21]. A recent meta-analysis of 16 studies reported moderate effects of CR on cognition in affective illness (primarily affective psychosis) similar to findings in SZ [
22]. Despite the substantial overlap in cognitive impairments between SZ and BD – qualitatively and in some studies quantitatively – only one study to date has reported outcomes after CR in patients with BD. In a 14-session cognitive training program focusing on neurocognitive deficits and residual symptoms of depression in patients with BD, CR was associated with improved executive functioning, which was related to improved vocational performance [
23].
Challenges in cognitive remediation
Typical CR paradigms require multiple training sessions weekly, which may be prohibitive for potential study participants. Many of the CR programs that have shown efficacy in patients with SZ have studied chronic patients who are able to attend sessions at community-based programs multiple times per week, due to high levels of disability and low instrumental role involvement (e.g., [
24]). While patients with BD exhibit considerable disability, especially compared to premorbid functioning, patients with SZ or schizoaffective disorder (SZA) exhibit greater morbidity and disability on average [
25], indicating that treatments requiring multiple weekly sessions at study sites may be unrealistic, especially for patients who are even partially engaged in work or school roles. With rapid advancements in online cognitive training programs, there is considerable potential for Internet-based treatments to improve access to training for patients. However, reports suggest that face-to-face “bridging” sessions, which create an opportunity for subjects to explicitly consider implementation of newly-developed cognitive skills in daily life, are important for generalizing the effect of CR treatments to daily functioning [
26]. Thus, a tension exists between reducing participant burden through remote access and conducting treatment delivery in the most effective manner.
Inclusion of adequate comparison conditions is challenging in CR paradigms given the active nature of the treatment. “Placebo” conditions are particularly challenging with these types of programs, as participants are keenly aware of the elements of participation throughout the program. The majority of projects that have included an active control group have selected individual or group rehabilitative or therapeutic activities such as Enriched Supportive Therapy [
27], supportive group therapy [
28], cognitive behavioral therapy (CBT) [
29], Integrated Psychological Therapy [
30], treatment as usual plus group activities [
31], vocational rehabilitation, occupational therapy, supported employment [
32‐
34], computer skills training plus day treatment [
35] or social skills training [
36]. These types of comparison conditions allow for the control of hours of study staff contact and other factors associated with involvement/adherence to a psychosocial intervention; however, they do not control for other aspects of the CR intervention such as computer interaction or game engagement, and they do not reasonably permit double-blindness – a particular challenge for interpretation of results as poor masking of treatment allocation in clinical trials has been shown to inflate treatment effects [
21].
Recently several game-format comparison conditions have been developed for CR paradigms attempting to match the computer/gaming elements of CR but with nonspecific content; these include computer games or action videogames [
24,
37‐
39], some paired with additional group or coaching activities to match for other elements of the active treatment condition [
37,
38]. No studies to date have published findings including Internet-based controls that can be overseen via remote access; however, as Internet-based CR programs are in active development and use, comparable Internet-based paradigms that will permit active comparison are needed.
CR programs vary considerably in terms of methodology, and it is not clear which if any aspect(s) of treatment design influence outcomes. Variables to consider include total duration of treatment, number of sessions, treatment intensity (i.e., “density” of sessions), drill and practice approaches versus strategy use, and individual versus group format, among many other characteristics. A meta-analysis of 40 CR studies did not find any effect of treatment characteristics on cognitive outcomes [
21], although programs that included strategy use showed a trend toward greater functional improvements [
21,
40].
Finally, it is possible that participant characteristics may influence treatment response. In their meta-analysis, Wykes and colleagues [
21] found no effect of most baseline characteristics including age, baseline cognitive functioning, or chlorpromazine equivalents on outcomes after CR; however, lower clinical symptoms at baseline were associated with greater effects of treatment on cognitive outcomes.
In this report we aimed to (1) describe our Internet-based neuroplasticity-informed CR protocol for BDP, (2) describe the development of our Internet-based, dose-matched active control, and (3) assess the ability of our study design to remain blinded.