## Abstract

Treatment-resistant depression (TRD) refers to a condition where individuals with major depressive disorder have inadequate or no response to treatment. Although functional disability is a prominent and costly feature of treatment resistance, very little is known about the factors that contribute to and maintain functional impairment in TRD. This is the first study to report the neurocognitive profile of TRD and the relationships among neurocognition, symptoms, and functioning in this syndrome. Results indicated that patients with TRD (N = 33) exhibit mildly reduced performance across all neurocognitive domains with a superimposed moderate impairment in verbal working memory. Neurocognition was associated with functional competence (what one can do), whereas depressive symptoms were associated with functional performance (what one actually does). Understanding the psychological mechanisms related to functioning may help us move toward recovery in this chronically ill group.

## Introduction

Many individuals with major depressive disorder (MDD) have inadequate or no response to treatment. Operationally, this has been defined as treatment-resistant depression (TRD). Individuals with TRD are usually more severely ill in terms of chronicity and prognosis (Kornstein & Schneider, 2001). TRD is characterized by a failure to achieve the remission of clinical symptoms following at least one trial of an antidepressant medication with established efficacy at an appropriate dose and duration that is sufficient to produce a robust therapeutic effect (e.g., 12 weeks; Fava, 2003; Thase & Rush, 1995). Varying degrees of resistance have been proposed, ranging from the aforementioned criterion (Stage I), to response failures in three other classes of pharmacotherapy, plus bilateral electroconvulsive therapy (Stage V; Thase & Rush, 1995). Although TRD has been defined with several different sets of criteria, all definitions center on the fact that these patients experience chronic depression, with recalcitrant symptoms and continuous episodes, and have a poor treatment response.

The functional impairments related to these poor outcomes are not well understood. Individuals with TRD experience impairment in many areas of psychosocial functioning, including academic and occupational achievement, interpersonal relationships, independent living, and community participation (Petersen et al., 2004). Interestingly, the clinical symptoms associated with depression account for surprisingly little variance and do not fully account for the breadth and persistence of functional impairment experienced by individuals with this disorder (Harvey, 2011; Judd, Paulus, Wells, & Rapaport, 1996). In 2000, the World Health Organization ascertained that depression was the fourth leading cause of disease burden worldwide and projected that by 2020, depression would be the leading cause of disease burden in developed countries. To our knowledge, no epidemiological studies have specifically examined TRD, although reports suggest that only 20% of patients with MDD achieve full functional recovery. Of the 80% that do not achieve functional recovery, 20% remain chronically disabled (Andrews, 2001; Judd et al., 1998). Despite the evidence for chronic and profound disability, very little information exists on the factors that contribute to the functional impairments observed in TRD. It is likely that multiple factors unique to treatment resistance, such as the magnitude of neurocognitive impairment, the severity of depressive symptoms, and the chronicity of the illness, are related to functioning, but the exact nature of these relationships remains unclear.

One of the most prominent predictors of functional outcomes in other chronic psychiatric disorders, such as schizophrenia and bipolar disorder, is neurocognitive functioning (Bowie and Harvey, 2008, 2010). The neurocognitive profile of TRD has not been reported. However, in MDD, neurocognitive deficits are evident across multiple domains, including impairments in attention (Landro, Stiles, & Sletvold, 2001; Weiland-Fiedler et al., 2004), processing speed (Tsourtos, Thompson, & Stough, 2002), memory (Purcell, Maruff, Kyrios, & Pantelis, 1997), visuospatial processing (Bulbena & Berrios, 1993; Porter, Gallagher, Thompson, & Young, 2003), verbal and non-verbal learning (Basso & Bornstein, 1999), motor functioning (Borkowska & Rybakowski, 2001; Landro et al., 2001), and executive functioning (Merriam, Thase, Haas, Keshavan, & Sweeny, 1999; Paradiso, Lamberty, Garvey, & Robinson, 1997; Trichard et al., 1995). Neurocognitive performance is associated with mood state, such that neurocognitive impairments are greater when the individual is depressed; however, deficits persist even during states of euthymia (Harvey, 2011).

Although there is evidence to suggest that neurocognition is associated with functional outcomes in depression (Jaeger, Berns, Uzelac, & Davis-Conway, 2006), this literature base is scant, has not focused on treatment-resistant samples, and has not examined functioning at multiple levels and dimensions. Since mood symptoms do not fully account for the breadth of impairment, the association between neurocognition and functional impairment in MDD is in need of investigation.

The purpose of this study is to examine the relationships among symptoms, course of illness, neurocognition, and functional impairment in TRD. Specifically, we examined concurrent predictors of functional competence (performance-based assessment in the laboratory) and interview-based ratings of functional performance (real-world functioning) in a sample of individuals with TRD.

## Method

### Participants

Outpatients (N = 33) with a diagnosis of MDD were recruited from the Mood Disorders Research and Treatment Service at Providence Care Mental Health Services in Kingston, Ontario. The overall purpose of the study was to examine changes in cognition and functioning following cognitive remediation; data for this paper were obtained from baseline assessments. Participants recruited for this study had evidence of at least Stage I treatment resistance according to the Thase and Rush (1995) definition. One participant had a history of electroconvulsive therapy prior to baseline, with the last treatment occurring more than 2 years before the baseline assessment. Exclusion criteria included a medical diagnosis associated with neurocognitive impairment (e.g., dementia, cerebrovascular accident, traumatic brain injury), a reading level below Grade 6, as assessed by the Wide Range Achievement Test-Reading Recognition subtest (Wilkinson, 1993), or uncorrectable sensory/perceptual conditions. The presence of these exclusion criteria indicates a high probability of problems of understanding instructions and uncorrectable hearing or visual impairments that would violate the standardized assessment procedures.

The participant sample was composed of 23 women and 10 men between the ages of 18 and 74 (M = 45.8, SD = 13.0); 79% of the participants were White, 12% were more than one race, and 9% chose to not report their race. Participants had a range of 10–20 years of education (M = 14.1, SD = 2.5) and approximately 70% of the sample was unemployed at the time of the assessment. Approximately 58% of the participants had subsidized housing where they received governmental (n = 14) or family/friend (n = 5) financial support for their residence. Age of first psychiatric hospitalization ranged from 12 to 56 years (M = 21.9, SD = 20.2). Participants were hospitalized 1–50 times (M = 4.6, SD = 12.2) and spent a total of 0.5–54 months in the hospital (M = 5.1, SD = 12.7) over their lifetime. The amount of time elapsed since their last hospitalization ranged from 1.5 to 240 months (M = 33.8, SD = 57.6). Approximately 6% of the sample experienced psychotic symptoms, 6% had a history of alcohol dependence or abuse, 6% had a history of substance dependence or abuse, 33% had at least one comorbid anxiety disorder, and 6% had comorbid borderline personality disorder. See Table 1 for means and standard deviations of all measures.

Table 1.

Means and standard deviations for all neurocognitive, clinical, and functioning measures

Observed range M SD Impaired observed N (expected N = 2.3) Not impaired observed N (expected N = 30.7) χ2 (p-value)
Neurocognition
Neurocognition compositea −3.0 to 0.51 −0.92 0.85 25 15.07 (<.001)
Sustained attentiona −3.0 to 1.1 −0.86 1.28 13 20 53.19 (<.001)
Verbal memorya −2.2 to 1.8 −0.76 1.00 24 20.83 (<.001)
Verbal working memorya −3.0 to 1.1 −1.65 1.00 22 11 180.47 (<.001)
Information processing speeda −2.9 to 1.0 −0.81 0.98 25 15.07 (<.001)
Verbal fluencya −2.4 to 2.0 −0.59 1.30 27 6.34 (.012)
Response inhibitiona −2.4 to 1.5 −0.15 0.84 32 0.80 (.371)
Set shiftinga −6.4 to 1.0 −1.00 1.87 24 20.83 (<.001)
Symptoms
Montgomery–Åsberg Depression Rating Scaleb
possible range (0–60)
5–39 25.06 8.10
Beck Anxiety Inventoryb
possible range (0–63)
3–51 23.33 11.57
Competence
Social Skills Performance Assessmentb
possible range (0–95)
37–80 66.18 10.35
possible range (0–17)
0–17 8.91 8.67
Performance
LIFE-RIFT 1 (work)c
possible range (1–5)
1–5 4.18 1.18
LIFE-RIFT 2 (interpersonal)c
possible range (1–5)
2–5 3.94 1.14
LIFE-RIFT 3 (satisfaction)c
possible range (1–5)
2–5 3.48 0.71
LIFE-RIFT 4 (recreation)c
possible range (1–5)
1–5 3.24 1.32
Observed range M SD Impaired observed N (expected N = 2.3) Not impaired observed N (expected N = 30.7) χ2 (p-value)
Neurocognition
Neurocognition compositea −3.0 to 0.51 −0.92 0.85 25 15.07 (<.001)
Sustained attentiona −3.0 to 1.1 −0.86 1.28 13 20 53.19 (<.001)
Verbal memorya −2.2 to 1.8 −0.76 1.00 24 20.83 (<.001)
Verbal working memorya −3.0 to 1.1 −1.65 1.00 22 11 180.47 (<.001)
Information processing speeda −2.9 to 1.0 −0.81 0.98 25 15.07 (<.001)
Verbal fluencya −2.4 to 2.0 −0.59 1.30 27 6.34 (.012)
Response inhibitiona −2.4 to 1.5 −0.15 0.84 32 0.80 (.371)
Set shiftinga −6.4 to 1.0 −1.00 1.87 24 20.83 (<.001)
Symptoms
Montgomery–Åsberg Depression Rating Scaleb
possible range (0–60)
5–39 25.06 8.10
Beck Anxiety Inventoryb
possible range (0–63)
3–51 23.33 11.57
Competence
Social Skills Performance Assessmentb
possible range (0–95)
37–80 66.18 10.35
possible range (0–17)
0–17 8.91 8.67
Performance
LIFE-RIFT 1 (work)c
possible range (1–5)
1–5 4.18 1.18
LIFE-RIFT 2 (interpersonal)c
possible range (1–5)
2–5 3.94 1.14
LIFE-RIFT 3 (satisfaction)c
possible range (1–5)
2–5 3.48 0.71
LIFE-RIFT 4 (recreation)c
possible range (1–5)
1–5 3.24 1.32

Note: LIFT-RIFT = Longitudinal Interval Follow-up Evaluation Range of Impaired Functioning Tool.

aRepresented as z-scores with respect to the normative group.

bRepresented as a total sum score.

cLongitudinal Interval Follow-up Evaluation Range of Impaired Functioning Tool; Represented as the maximum value among multiple items within that domain.

### Measures

#### Neurocognition

The neurocognitive battery consisted of a selection of standardized tests that are frequently used in research and clinical settings. Tests were selected based on neurocognitive domains previously found to be impaired in MDD. Raw data were converted to age-corrected standard scores based on normative data from healthy control subjects, found in each test's manual. A neurocognitive composite score was computed by calculating an equally weighted average from the following domains.

“Sustained attention” refers to maintaining focus on a set of stimuli, while ignoring irrelevant stimuli. The Continuous Performance Test-Identical Pairs Version (Cornblatt, Risch, Faris, Friedman, & Erlenmeyer-Kimling, 1988) is a computerized test of sustained attention. In different trials, two-, three-, or four-digit numbers were presented and participants responded by pressing a response key when the number presented was identical to the previous number. A measure of signal detection, d′, was calculated based on the proportion of correct detections and false alarms.

“Verbal learning” refers to learning new information, such as a list of words over a series of trials. The Hopkins Verbal Learning Test (Brandt & Benedict, 2001) is a list-learning task of verbal declarative memory in which participants listened to a 12-word list over three learning trials. The dependent variable is the number of words correctly recalled by the respondent over the three trials.

“Verbal working memory” refers to the capacity to simultaneously store and manipulate information. The Letter Number Sequencing Test (Gold, Carpenter, Randolph, Goldberg, & Weinberger, 1997) is an auditory working memory test that requires participants to listen to presentations of intermixed numbers and letters of increasing length and to mentally reorder (i.e., to organize numbers in ascending order, followed by letters in alphabetical order) the presented lists before repeating them back to the test administrator. This verbal working memory task engages both the maintenance and manipulation components of working memory. The dependent variable is the number of correct responses.

“Information processing speed” refers to the ability to quickly and accurately respond to stimuli in the environment. Information processing speed was assessed using the Symbol Coding Task (Keefe, 1999) and the Trail Making Test Part A (TMT-A; Partington & Leiter, 1949). The Symbol Coding Task is a test of processing speed that requires the participant to rapidly and accurately match numbers to nonsense syllables. The dependent variable is the total number of correct responses in 90 s. The TMT-A requires the subject to connect 25 consecutive numbers that are arranged in irregular locations on a sheet of paper as quickly as possible. The dependent variable is the total time to completion.

“Verbal fluency” refers to the ability to produce words efficiently under contextual demands. The Controlled Oral Word Association Test and Animal Naming Tests (Nuechterlein & Green, 2006; Spreen & Benton, 1977) were used to measure phonological and semantic fluency, respectively. These tests required the participant to orally produce as many words as possible, beginning with a specific letter or category, in a 60-s interval. The dependent variable is the number of correct words produced in 60 s.

“Executive functioning” refers to the ability to plan ahead, solve problems, engage in abstract thinking, and manage other cognitive skills. Executive functioning was assessed as two separate components. Response inhibition was assessed using the Stroop Color-Word Test (Golden, 1978), a measure of the color-word interference effect, wherein the participant has to inhibit their natural response in order to name the color of the ink in which the words are printed, ignoring the non-matching written color name that was printed for each item. Participants have to focus on a salient feature of stimuli while simultaneously inhibiting response to a distracting feature of the same stimuli. The dependent variable is calculated as a difference score that subtracts the Stroop predicted performance from the observed performance. Set-shifting was assessed using the TMT-B (Partington & Leiter, 1949), which required the subject to alternate their attention between two salient features of a stimulus. It involved connecting 25 encircled numbers and letters in alternating order. The dependent variable is the time to completion in seconds.

A “Neurocognitive Composite Score (NCS)” was an equally weighted average of all the neurocognitive domains. It was calculated by converting the aforementioned neurocognitive domains into z-scores. The information processing speed and verbal fluency domains consisted of more than one neurocognitive measure; therefore, the z-scores for the dependent variables within these two domains were averaged to create the domain scores. Finally, the z-scores from all of the neurocognitive domains were averaged to create the NCS. The computation for the NCS has consistently been used in other research (e.g., Keefe, 1999; Nuechterlein & Green, 2006).

#### Symptoms

Measures of depressive and anxiety symptoms were collected. The 10-item Montgomery–Åsberg Depression Rating Scale (MADRS; Montgomery & Asberg, 1979) was used to measure the severity of depressive symptoms via a 45-min semi-structured interview rated by Master's level graduate students enrolled in a clinical psychology program. The MADRS item scales range from 0 to 6 for a total score range of 0–60. Higher scores indicate more severe depressive symptoms.

The Beck Anxiety Inventory (BAI; Beck, Epstein, Brown, & Steer, 1988) is a 21-item self-report instrument that was used to assess somatic, subjective, and panic-related symptoms. The BAI employs a 4-point scale ranging from 0 to 3 with a total score that ranges from 0 to 63. Higher scores indicate more severe anxiety.

#### Performance-based measures of functional competence

Performance-based measures are objective, laboratory-based assessments in which participants engage in everyday situations through role play. These measures allow for the assessment of an individual's capabilities, rather than their performance in the real world. A major issue in the interpretation of functional disability in psychiatric disorders is that factors other than competence are related to real-world performance (Gupta, Bassett, Iftene, & Bowie, 2012). Therefore, we also used self-report and clinician-rated behavioral observations as measures of real-world behavior.

##### Interpersonal competence

The Social Skills Performance Assessment (SSPA; Patterson, Moscona, McKibbin, Davidson, & Jeste, 2001) is a measure of social competence and communication. After a brief practice session, patients initiate and maintain a conversation for 3min in each of two situations: greeting a new neighbor and calling a landlord to request a repair for an ongoing leak. The sessions are audio-taped and scored by a blind trained rater who is unaware of diagnosis and all other data. Dimensions of social skills scores include interest, fluency, clarity, focus, negotiation ability, persistence, and social appropriateness. Each item was scored on a scale from 0 to 5, for a total score range of 0–95. Higher scores indicate better interpersonal competence. Raters were trained to the gold standard ratings proposed by the instrument developers (ICC = 0.86).

### Data Analysis

Descriptive statistics and a repeated-measures analysis of variance were used to illustrate the neurocognitive profile of TRD. Stepwise regression analyses were used to determine the variables associated with functional skill impairment and real-world behavior. Concurrent demographic, course of illness, symptom, and neurocognitive variables were entered as predictors for each functional skill domain (interpersonal and adaptive competence) and clinician-rated real-world behavior (work, interpersonal, satisfaction, and recreation).

## Results

### Neurocognitive Profile of TRD

The repeated-measures analysis of variance revealed that the neurocognitive domains were significantly different, F(6,26) = 6.35, p < .001, partial η2 = 0.59. Follow-up pairwise comparisons between the neurocognitive domains were examined using paired-samples t-tests with a Bonferroni correction of 0.05/21 = 0.002. The follow-up pairwise comparisons revealed significantly reduced performance in verbal working memory compared with verbal learning, verbal fluency, information processing speed, and response inhibition (p < .001). All other neurocognitive domains were not statistically different from each other. Two-tailed Pearson's correlations were conducted among the neurocognitive domains to examine the potential for multicollinearity. Almost all correlation coefficients were below 0.50, with three correlation coefficients falling between 0.50 and 0.70, below standard cutoffs for multicollinearity (Crown, 1998). Since the current study did not employ a healthy comparison group, we also conducted a one-sample t-test on the z-scores for each of the neurocognitive domains to determine whether they were significantly different from zero. All neurocognitive domains were significantly different from zero (p < .005). Furthermore, a one-sample χ2 analysis was conducted for each neurocognitive domain using a z-score cutoff of ≤−1.50 as an index of impairment. The expected values of the χ2 were selected in order to reflect the proportion of individuals who would be impaired in a normal population, such that a significant χ2 suggests that the proportion of individuals in the current TRD population is significantly different than would be expected in a healthy control sample. The χ2 analyses were all significant (all ps < .02), with the exception of response inhibition (p = .37), suggesting that the current sample was truly impaired on measures of neurocognitive performance (Table 1).

### Stepwise Linear Regressions

#### Performance-based functional competence

Stepwise regressions were conducted to determine the variables associated with functional skill impairment (interpersonal and adaptive competence) and clinician-rated real-world behavior. Variables that were significantly correlated with the dependent variable were entered into one step and analyzed using a stepwise regression. The regression equation for interpersonal competence was significant with age at baseline entering first, F(1,29) = 8.98, p = .006, R2Δ = .24, followed by sustained attention in Step 2, F(2,28) = 8.56, p = .001, R2Δ = .14, and depressive symptoms in Step 3, F(3,27) = 9.87, p < .001, R2Δ = .14. The regression equation for adaptive competence was significant for set shifting only, F(1,31) = 13.11, p = .001, R2Δ = .30. Table 2 displays detailed results of the functional competence regression analyses.

Table 2.

Regression coefficients for interpersonal and adaptive competence

Interpersonal competence B SE β Partial correlations t p
Model 1 Age at baseline 0.02 0.01 0.49 .48 3.00 .006
Model 2 Age at baseline 0.02 0.01 0.40 .45 2.64 .014
Sustained attention 0.18 0.07 0.39 .43 2.54 .017
Model 3 Age at baseline 0.02 0.01 0.31 .38 2.63 .014
Sustained attention 0.20 0.07 0.43 .50 3.16 .004
Depressive symptoms 0.02 0.01 0.34 .41 2.85 .008
Set shifting 1.62 0.45 0.55 .52 3.62 .001
Interpersonal competence B SE β Partial correlations t p
Model 1 Age at baseline 0.02 0.01 0.49 .48 3.00 .006
Model 2 Age at baseline 0.02 0.01 0.40 .45 2.64 .014
Sustained attention 0.18 0.07 0.39 .43 2.54 .017
Model 3 Age at baseline 0.02 0.01 0.31 .38 2.63 .014
Sustained attention 0.20 0.07 0.43 .50 3.16 .004
Depressive symptoms 0.02 0.01 0.34 .41 2.85 .008
Set shifting 1.62 0.45 0.55 .52 3.62 .001

#### Clinician-rated real-world behavior

Poorer clinician-rated real-world work behavior was significantly associated with severity of depressive symptoms, F(1,30) = 5.35, p = .028, R2Δ = .15. Impairments in interpersonal relations, F(1,30) = 8.55, p = .007, R2Δ = .22, and general satisfaction, F(1,30) = 15.42, p < .001, R2Δ = .34, were significantly associated with the severity of depressive symptoms. Recreation was significantly associated with the severity of depressive symptoms in Step 1, F(1,27) = 14.41, p = .001, R2Δ = .35, and sustained attention in Step 2, F(2,26) = 14.82, p < .001, R2Δ = .19. Refer to Table 3 for detailed results of the regression analyses for real-world behavior.

Table 3.

Regression coefficients for real-world behavior

B SE β Partial correlations t p
Work skills
Model 1
Depressive symptoms 0.06 0.03 0.39 .39 2.31 .028
Interpersonal Behavior
Model 1
Depressive symptoms 0.07 0.02 0.47 .47 2.92 .007
General satisfaction
Model 1
Depressive symptoms 0.05 0.01 0.58 .58 3.93 .000
Recreation
Model 1
Depressive symptoms 0.10 0.03 0.59 .59 3.80 .001
Model 2
Depressive symptoms 0.09 0.02 0.55 .62 4.10 .000
Sustained attention −0.45 0.14 −0.43 −.52 −3.21 .004
B SE β Partial correlations t p
Work skills
Model 1
Depressive symptoms 0.06 0.03 0.39 .39 2.31 .028
Interpersonal Behavior
Model 1
Depressive symptoms 0.07 0.02 0.47 .47 2.92 .007
General satisfaction
Model 1
Depressive symptoms 0.05 0.01 0.58 .58 3.93 .000
Recreation
Model 1
Depressive symptoms 0.10 0.03 0.59 .59 3.80 .001
Model 2
Depressive symptoms 0.09 0.02 0.55 .62 4.10 .000
Sustained attention −0.45 0.14 −0.43 −.52 −3.21 .004

## Discussion

To our knowledge, this was the first study to examine the relationships among neurocognition, symptoms, and functional outcomes in TRD. We sought to determine which neurocognitive, course of illness, and clinical variables were associated with functioning in this unique sample.

Individuals with TRD display mildly reduced performance in global neurocognition and across all neurocognitive domains with a superimposed moderate impairment in verbal working memory. Neurocognitive performance in this sample was worse relative to other MDD samples that did not explicitly recruit treatment resistant patients (e.g., Paradiso et al., 1997; Weiland-Fiedler et al., 2004), although direct comparisons are necessary to further examine the relative effect of treatment resistance.

In the current study, participants' ability to initiate and maintain goal-directed conversations was associated with older age at baseline, greater sustained attention, and more severe symptoms of depression. Older individuals are more likely to have had the necessary opportunities for acquiring interpersonal skills. Additionally, sustained attention is a neurocognitive skill that is necessary for following conversations and is likely to be an important asset in successful interpersonal interactions. The relationship between greater interpersonal skill and greater severity of depressive symptoms is somewhat surprising. One might expect that more severe mood symptoms would be related to greater impairment in social competence (Kessler et al., 2003). A possible explanation, which requires further study, is that better interpersonal skills may be related to a heightened ability to attend to and critically appraise the negative social reactions of others and this may perpetuate depressive symptomatology.

Our measure of complex financial management was associated with better set-shifting. The ability to process information efficiently and switch between cognitive tasks provides an advantage in the timed Advanced Finances task and may be an important treatment target for programs aimed at improving functional competence in individuals with chronic depression.

Impairment in all four of the real-world behavior domains (work, interpersonal relations, satisfaction, and recreation) was associated with more severe depressive symptoms. Greater impairment in recreational activity was also associated with poor sustained attention.

### Limitations and Future Directions

As this study is the first to examine the relationships among neurocognition, symptoms, and functional outcomes in TRD, it has brought to light several important lessons and generated new ideas for future research. First, it is important to acknowledge that due to sparse literature in this area, the theory that informed our hypotheses stemmed from research on other psychiatric groups. In the current study, neurocognition was associated with interpersonal and adaptive competence, but was not significantly related to measures of real-world behavior in TRD. Symptoms appear to play a much larger role in real-world functioning in TRD compared with other disorders such as bipolar disorder and schizophrenia (Bowie et al., 2010). Given limited literature on neurocognition and functioning in MDD, let alone TRD, this study, although limited, can provide insight into future directions for research in this field. There are several distinctions between TRD and other psychiatric disorders that could account for some of the unexpected findings in this study. Symptoms of depression are cyclical in nature even when severe and persistent. In general, individuals with depression have better premorbid functioning and experience a substantial drop in functioning with the onset of symptoms. It is plausible that the neurocognitive deficits observed in depression are state rather than trait-like as it has been proposed that during euthymia the brain can repair itself and improve neurocognitive functioning (Hammar, Lund, & Hugdahl, 2003; Neu et al., 2005). The aforementioned distinctions are important candidates for future research aiming to understand the complex relationships among neurocognition, symptoms, and functional outcomes in TRD.

A formal structured assessment of diagnosis was not conducted in this study, although diagnoses were made by senior clinicians in a tertiary clinic. Furthermore, the assessment of functional competence in the laboratory and functional behaviors in the real world are challenging, and there is no gold standard measure. The instrument used to assess functional performance was the LIFE-RIFT. Although this is a clinician-rated tool, it is based on a semi-structured interview that elicits self-reports of functioning from the participant and is therefore, in essence, a self-report measure. As such, this measure is susceptible to biases in self-report, particularly as the negative attributional style of depressed thinking is known to affect recall of autobiographical memories (Kuyken & Dalgleish, 2011).

This study was limited by a small sample size. Post hoc power analyses were conducted using the program G*Power (Erdfelder, Faul, & Buchner, 1996). Power for all regressions was low (range 0.38–0.57). Given that this is the first study to report on these relationships, albeit with a small sample size, replication of these results in independent and larger samples is critical to developing a fuller understanding of the relationships among symptoms, neurocognition, and functioning in TRD. The current study is a first step toward a better understanding of this unique disorder.

It is important to keep in mind that the current sample was comprised of patients with severe TRD who were referred to a tertiary mood disorders clinic. As such, these findings may not generalize to all individuals with MDD but they do represent a particularly important group considering the economic burden and personal suffering in TRD. TRD is characterized by high prevalence of comorbid psychiatric conditions and multiple somatic complaints, which restrains the ability to sample cohorts without confounding clinical characteristics. Although comorbid conditions may be contributing to functional impairment in this group, the present sample is likely to be closer to the clinical reality of TRD than a sample without comorbidities.

Finally, this was a pilot study that did not incorporate a demographically matched healthy comparison group. Although the current sample displayed reduced performance across all neurocognitive domains, future research would benefit from including a control group in order to make comparisons about the severity of neurocognitive and functioning deficits in TRD.

### Implications

Individuals with TRD are a unique subset of those with MDD who suffer from chronic, persistent symptoms that are unresponsive to the standard pharmacological and psychological treatment. Better understanding of the factors associated with the functional outcome in this group is of great importance in developing new approaches to treatment.

The current study revealed several exciting new findings. Neurocognition and depressive symptoms are associated with functional competence (what one can do) but depressive symptoms are more strongly related to functional performance (what one actually does in the real world). As such, there may be additional intrinsic or extrinsic factors, such as motivation, dysfunctional attitudes, or stigma, that mediate the relationship between competence and performance and are worthy of investigation in future studies. Our findings suggest that improvement of neurocognitive and functional abilities might provide limited transfer to changes in real-world behavior if mood symptoms are constant. Finally, neurocognition emerged as significantly associated with interpersonal and adaptive functioning. Research in other psychiatric groups has demonstrated that neurocognitive remediation therapy is effective at improving neurocognition and that these effects generalize to improved functioning in schizophrenia, bipolar disorder, and MDD (Deckersbach et al., 2010; Naismith et al., 2011; Wykes et al., 2009). Moreover, these improvements are particularly effective within a broader psychosocial treatment (Bell, Zito, Greig, & Wexler, 2008; Hogarty et al., 2004; Hogarty, Greenwals, & Eack, 2006). Researchers and clinicians will need to take a unique approach to understanding functioning in order to develop novel and effective interventions and move forward in promoting both symptomatic and functional recovery in this unique group.

## Funding

This work was supported by funds to Dr Bowie by Queen's University, a Leader's Opportunity Award from the Canadian Foundation for Research and Innovation and an Early Researcher Award from the Ontario Ministry of Research and Innovation.

None declared.

## Acknowledgements

The authors are grateful to Theresa Bernard for coordinating the study and formatting and editing the manuscript; Allisha Patterson, Jeremy Stewart, and Talia Troister for conducting symptom assessments; and Emma Bassett, Kaelen Boyd, Lauren David, Lisa Gou, Meighen Roes, Laura Stefanik, Stephanie Taillefer, and Morgan Todd for data collection.

## References

Andrews
G.
Should depression be managed as a chronic disease?
British Medical Journal
,
2001
, vol.
322
(pg.
419
-
421
)
Basso
M. R.
Bornstein
R. A.
Relative memory deficits in recurrent vs. first-episode major depression on a word-list learning task
Neuropsychology
,
1999
, vol.
13
(pg.
557
-
563
)
Beck
A. T.
Epstein
N.
Brown
G.
Steer
R. A.
An inventory for measuring clinical anxiety: The Beck Anxiety Inventory
Journal of Consulting and Clinical Psychology
,
1988
, vol.
56
(pg.
893
-
897
)
Bell
M. D.
Zito
W.
Greig
T.
Wexler
B. E.
Neurocognitive enhancement therapy with vocational services: Work outcomes at two-year follow-up
Schizophrenia Research
,
2008
, vol.
105
(pg.
18
-
29
)
Borkowska
A.
Rybakowski
J. K.
Neuropsychological frontal lobe tests indicate that bipolar depressed patients are more impaired than unipolar
Bipolar Disorders
,
2001
, vol.
3
(pg.
88
-
94
)
Bowie
C. R.
Depp
C.
McGrath
J. A.
Wolyniec
P.
Mausbach
B. T.
Thornquist
M. H.
, et al.  .
Prediction of real-world functional disability in chronic mental disorders: A comparison of schizophrenia and bipolar disorder
American Journal of Psychiatry
,
2010
, vol.
AiA
(pg.
1
-
9
)
Bowie
C. R.
Harvey
P. D.
Communication abnormalities predict functional outcomes in chronic schizophrenia: Differential associations with social and adaptive functions
Schizophrenia Research
,
2008
, vol.
103
(pg.
240
-
247
)
Brandt
J.
Benedict
R. H. B.
The Hopkins Verbal Learning Test-Revised
,
2001
Odessa, FL
Psychological Assessment Resources
Bulbena
A.
Berrios
G. E.
Cognitive function in the affective disorders: a prospective study
Psychopathology
,
1993
, vol.
26
(pg.
6
-
12
)
Cornblatt
B. A.
Risch
N. J.
Faris
G.
Friedman
D.
Erlenmeyer-Kimling
L.
The Continuous Performance Test, Identical Pairs Version (CPT-IP): I. New findings about sustained attention in normal families
Psychiatric Research
,
1988
, vol.
26
(pg.
223
-
238
)
Crown
W. H.
Statistical models for the social and behavioral sciences: Multiple regression and limited-dependent variable models
,
1998
Westport, CT
Praeger Publishers
Deckersbach
T.
Nierenberg
A. A.
Kessler
R.
Lund
H. G.
Ametrano
R. M.
Sachs
G.
, et al.  .
Cognitive rehabilitation for bipolar disorder: An open trial for employed patients with residual depressive symptoms
CNS Neuroscience and Therapeutics
,
2010
, vol.
16

5
(pg.
298
-
307
)
Erdfelder
E.
Faul
F.
Buchner
A.
GPOWER: A general power analysis program
Behavior Research Methods Instruments and Computers
,
1996
, vol.
28
(pg.
1
-
11
)
Fava
M.
Diagnosis and definition of treatment-resistant depression
Biological Psychiatry
,
2003
, vol.
53
(pg.
649
-
659
)
Gold
J. M.
Carpenter
C.
Randolph
C.
Goldberg
T. E.
Weinberger
D. R.
Auditory working memory and Wisconsin Card Sorting Test performance in schizophrenia
Archives of General Psychiatry
,
1997
, vol.
54
(pg.
159
-
165
)
Golden
C. J.
The Stroop Color and Word Test
,
1978
Chicago: IL
Stoelting Company
Gupta
M.
Bassett
E.
Iftene
F.
Bowie
C. R.
Functional outcomes in schizophrenia: Understanding the competence-performance discrepancy
Journal of Psychiatric Research
,
2012
, vol.
46
(pg.
205
-
211
)
Hammar
A.
Lund
A.
Hugdahl
K.
Long-lasting cognitive impairment in unipolar major depression: A 6-month follow-up study
Psychiatric Research
,
2003
, vol.
118
(pg.
189
-
196
)
Harvey
P. D.
Mood symptoms, cognition, and everyday functioning in major depression, bipolar disorder, and schizophrenia
Innovations in Clinical Neuroscience
,
2011
, vol.
8

10
(pg.
14
-
18
)
Heaton
R. K.
Marcotte
T. D.
Mindt
M. R.
J.
Moore
D. J.
Bentley
H.
, et al.  .
The impact of HIV-associated neuropsychological impairment on everyday functioning
Journal of the International Neuropsychological Society
,
2004
, vol.
10
(pg.
317
-
331
)
Hogarty
G. E.
Flesher
S.
Ulrich
R.
Carter
M.
Greenwald
D.
Pogue-Geile
M.
, et al.  .
Cognitive enhancement therapy for schizophrenia: Effects of a 2-year randomized trial on cognition and behavior
Archives of General Psychiatry
,
2004
, vol.
61

9
(pg.
866
-
876
)
Hogarty
G. E.
Greenwals
D. P.
Eack
S. M.
Durability and mechanism of effects of cognitive enhancement therapy
Psychiatric Services
,
2006
, vol.
57

12
(pg.
1751
-
1757
)
Jaeger
J.
Berns
S.
Uzelac
S.
Davis-Conway
S.
Neurocognitive deficits and disability in major depressive disorder
Psychiatric Research
,
2006
, vol.
145
(pg.
39
-
48
)
Judd
L. L.
Akiskal
H. S.
Maser
J. D.
Zeller
P. J.
Endicott
J.
Coryell
W.
, et al.  .
A prospective 12-year study of subsyndromal and syndromal depressive symptoms in unipolar major depressive disorders
Archives of General Psychiatry
,
1998
, vol.
55
(pg.
694
-
700
)
Judd
L. L.
Paulus
M. P.
Wells
K. B.
Rapaport
M. H.
Socioeconomic burden of subsyndromal depressive symptoms and major depression in a sample of the general population
American Journal of Psychiatry
,
1996
, vol.
153
(pg.
1411
-
1417
)
Keefe
R. S. E.
Brief assessment of cognition in schizophrenia
,
1999
Durham, NC
Duke University Medical Center
Kessler
R. C.
Berglund
P.
Demler
O.
Jin
R.
Koretz
D.
Merikangas
K. R.
, et al.  .
The epidemiology of major depressive disorder: Results from the National Comorbidity Survey Replication (NCS-R)
Journal of the American Medical Association
,
2003
, vol.
289

23
(pg.
3095
-
3105
)
Kornstein
S. G.
Schneider
R. K.
Clinical features of treatment-resistant depression
Journal of Clinical Psychiatry
,
2001
, vol.
62

16
(pg.
18
-
25
)
Kuyken
W.
Dalgleish
T.
Overgeneral autobiographical memory in adolescents at risk for depression
Memory
,
2011
, vol.
19

3
(pg.
241
-
250
)
Landro
N. I.
Stiles
T. C.
Sletvold
H.
Neuropsychological function in nonpsychotic unipolar major depression
Neuropsychiatry, Neuropsychology, and Behavioral Neurology
,
2001
, vol.
14
(pg.
233
-
240
)
Leon
A. C.
Solomon
D. A.
Mueller
T. I.
Turvey
C. L.
Endicott
J.
Keller
M. B.
The Range of Impaired Functioning Tool (LIFE-RIFT): A brief measure of functional impairment
Psychological Medicine
,
1999
, vol.
29
(pg.
869
-
878
)
Merriam
E. P.
Thase
M. E.
Haas
G. L.
Keshavan
M. S.
Sweeny
J. A.
Prefrontal cortical dysfunction in depression determined by Wisconsin Card Sorting Test performance
American Journal of Psychiatry
,
1999
, vol.
156
(pg.
780
-
782
)
Montgomery
S. A.
Asberg
M.
A new depression scale designed to be sensitive to change
British Journal of Psychiatry
,
1979
, vol.
134
(pg.
382
-
389
)
Naismith
S. L.
Diamon
K.
Carter
P. E.
Norrie
L. M.
M. E.
Lewis
S. J. G.
, et al.  .
Enhancing memory in late-life depression: The effects of a combined psychoeducation and cognitive training program
American Journal of Geriatric Psychiatry
,
2011
, vol.
19

3
(pg.
240
-
248
)
Neu
P.
Bajbouj
M.
Schilling
A.
Godemann
F.
Berman
R. M.
Schlattmann
P.
Cognitive function over the treatment course of depression in middle-aged patients: Correlation with brain MRI signal hyperintensities
Journal of Psychiatric Research
,
2005
, vol.
39
(pg.
129
-
135
)
Nuechterlein
K. H.
Green
M. F.
MATRICS Consensus Cognitive Battery manual
,
2006
Los Angeles
University of California
S.
Lamberty
G. J.
Garvey
M. J.
Robinson
R. G.
Cognitive impairment in the euthymic phase of chronic unipolar depression
Journal of Nervous and Mental Disease
,
1997
, vol.
185
(pg.
748
-
754
)
Partington
J. E.
Leiter
R. G.
Partington's Pathway Test
Psychological Services Center Bulletin
,
1949
, vol.
1
(pg.
9
-
20
)
Patterson
T. L.
Moscona
S.
McKibbin
C. L.
Davidson
K.
Jeste
D. V.
Social skills performance assessment among older patients with schizophrenia
Schizophrenia Research
,
2001
, vol.
48

2–3
(pg.
351
-
60
)
Petersen
T.
Papakostas
G. I.
Mahal
Y.
Guyker
W. M.
Beaumont
E. C.
Alpert
J. E.
, et al.  .
Psychosocial functioning in patients with treatment resistant depression
European Psychiatry
,
2004
, vol.
19
(pg.
196
-
201
)
Porter
R. J.
Gallagher
P.
Thompson
J. M.
Young
A. H.
Neurocognitive impairment in drug-free patients with major depressive disorder
British Journal of Psychiatry
,
2003
, vol.
182
(pg.
214
-
220
)
Purcell
R.
Maruff
P.
Kyrios
M.
Pantelis
C.
Neuropsychological function in young patients with unipolar major depression
Psychological Medicine
,
1997
, vol.
27
(pg.
1277
-
1285
)
Spreen
O.
Benton
A. L.
Neurosensory Center Comprehensive Examination for Aphasia (NCCEA)
,
1977
Victoria
University of Victoria Neuropsychology Laboratory
Thase
M. E.
Rush
A. J.
Bloom
F. E.
Kupfer
D. J.
Treatment resistant depression
Psychopharmacology: The fourth generation of progress
,
1995
New York
Raven Press
(pg.
1081
-
1097
)
Trichard
C.
Martinot
J. L.
Alagille
M.
Masure
M. C.
Hardy
P.
Ginestet
D.
, et al.  .
Time course of prefrontal lobe dysfunction in severely depressed in-patients: A longitudinal neuropsychological study
Psychological Medicine
,
1995
, vol.
25
(pg.
79
-
85
)
Tsourtos
G.
Thompson
J. C.
Stough
C.
Evidence of an early information processing speed deficit in unipolar major depression
Psychological Medicine
,
2002
, vol.
32
(pg.
259
-
265
)
Weiland-Fiedler
P.
Erickson
K.
Waldeck
T.
Luckenbaugh
D. A.
Pike
D.
Bonne
O.
, et al.  .
Evidence for continuing neuropsychological impairments in depression
Journal of Affective Disorders
,
2004
, vol.
82
(pg.
253
-
258
)
Wilkinson
G. S.
Wide-Range Achievement Test 3: Administration manual
,
1993
Wilmington, Del
Wide Range
Wykes
T.
Reeder
C.
Landau
S.
Matthiasson
P.
Haworth
E.
Hutchinson
C.
Does age matter? Effects of cognitive rehabilitation across the age span
Schizophrenia Research
,
2009
, vol.
113
(pg.
252
-
258
)