Cognitive deficits may contribute to the characteristic symptoms of obsessive–compulsive disorder (OCD), such as chronic doubt, repetitive checking, behavioural inflexibility, and ritualistic behaviours. In contemporary neurobiological models of OCD, a specific neural circuit is implicated, comprising the orbitofrontal cortex, the cingulate cortex and the caudate nucleus. ^{Reference Insel1} It has been difficult, however, to link findings regarding brain circuit dysfunction and structural differences, cognitive impairment and symptomatology to create a comprehensive, coherent model of OCD. Many studies have attempted to use neuropsychological tests to evaluate deficits in cognitive functioning that may reflect brain functioning impairment associated with OCD. The majority of findings suggest impairment in visuospatial abilities, memory performance and various executive functions, although general intellectual abilities and attention span appear preserved. ^{Reference Greisberg and McKay2} This pattern of deficits, however, has not been consistently observed across studies and conclusions regarding specific deficits associated with OCD have not been possible. ^{Reference Greisberg and McKay2–Reference Kuelz, Hohagen and Voderholzer4} In addition, almost all of the existing neuropsychological studies of OCD are cross-sectional and do not provide evidence of whether the deficits preceded the onset of the disorder. Discriminating between deficits that play an aetiologically significant role and those that may result from symptomatology has far-reaching theoretical and clinical implications.

The current study used longitudinal data to extend the literature on the relationship between OCD and neuropsychological functioning. We tested the hypothesis that deficit neuropsychological test performance at age 13 would predict OCD diagnosed at age 32. On the basis of biological models and cross-sectional studies, we hypothesised that the strongest associations between childhood neuropsychological test performance and adult OCD would be within the visuospatial/visuomotor and executive domains, whereas general IQ and verbal ability would be unimpaired.

Method

Results

Neuropsychological performance

Independent samples t-tests were conducted to examine group differences in overall intellectual functioning at age 13. The small difference (as measured by prorated WISC–R full-scale IQ) between those with (mean 103.31, s.d. = 18.04) and without (mean 108.09, s.d. = 14.36) OCD at age 32 was not significant (t(685) = 1.18, P = 0.24).

Table 1 contains neuropsychological test results for cohort members with and without a diagnosis of OCD at age 32. Odds ratios show the effect of neuropsychological performance at age 13 on risk for OCD at age 32, controlling for socioeconomic status and gender. Although there were no group differences on the WISC–R freedom from distractibility or verbal comprehension factors, performance of the OCD group on the WISC–R perceptual organisation factor was 0.75 s.d. lower than that of the control group. Performance on the ROCF was also significantly worse in the OCD group (0.82 s.d. and 0.91 s.d. lower for the copy and delayed recall conditions, respectively), as was performance on WISC–R Mazes (0.55 s.d. lower for the OCD group than for the control group). Finally, the OCD group was significantly slower than the control group on the Grooved Pegboard test. The Pegboard effect persisted following adjustment for hand preference. No significant differences were found between groups on the COWAT, the TMT, the RAVLT or the WCST. However, on WCST total errors the OCD group scored 0.5 s.d. below controls. This difference represents a moderate effect size, but was not significant because of the small size of the OCD group.

Table 1 Univariate logistic regression of neuropsychological tests scores at age 13 on diagnosis of obsessive-compulsive disorder (OCD) at age 32 ^a

Variable	No OCD, mean (s.d.) (n = 687)	OCD, mean (s.d.) (n = 13)	No OCD v. OCD, OR (95% CI) b
WISC–R factors, z-score
Perceptual organisation	0.02 (0.96)	–0.73 (1.12)	0.4 (0.25–0.76) **
Freedom from distractibility	–0.03 (0.98)	0.07 (1.03)	1.1 (0.62–2.03)
Verbal comprehension	–0.01 (0.96)	–0.15 (1.31)	0.8 (0.45–1.49)
Rey–Osterrieth Complex Figure Test, z-score
Copy,	0.05 (0.90)	–0.77 (1.26)	0.6 (0.40–0.82) **
Delayed recall	0.03 (1.00)	–0.88 (0.68)	0.4 (0.24–0.71) **
Rey Auditory –Verbal Learning Test, z-score
Trials 1–4	0.00 (1.00)	–0.22 (1.18)	0.8 (0.47–1.37)
Post interference test	0.01 (1.00)	–0.44 (1.20)	0.7 (0.39–1.09)
Delayed recall	0.02 (0.99)	–0.32 (1.23)	0.7 (0.41–1.22)
Wisconsin Card Sorting Test, z-score
Total errors	–0.01 (1.00)	0.49 (1.02)	1.6 (0.94–2.63)
Perseverative errors	0.00 (1.00)	0.05 (1.21)	1.1 (0.62–1.80)
Trail Making Test (TMT) A and B, z-score
TMT B, s	0.00 (1.02)	0.15 (0.86)	1.1 (0.77–1.58)
TMT B–A, s	0.00 (1.02)	0.15 (1.01)	1.1 (0.61–1.11)
Grooved Pegboard Test, z-score
Right hand	–0.02 (1.00)	0.93 (2.00)	1.5 (1.11–1.94) **
Left hand	–0.02 (0.95)	0.59 (1.15)	1.6 (1.08–2.45) *
WISC–R Mazes, z-score	0.04 (0.95)	–0.51 (1.14)	0.6 (0.38–0.97) *
Verbal Fluency Controlled Oral Word Association Test, z-score	0.00 (1.00)	–0.20 (1.00)	0.8 (0.45–1.44)

WISC–R, Wechsler Intelligence Scale for Children–Revised

a. n varies slightly for some risk factors owing to missing data

b. Adjusted for gender and socioeconomic status

* P<0.05

** P<0.01

Discussion

We found a relationship between premorbid impairment in specific cognitive domains at age 13 and OCD at age 32. Compared with the rest of the cohort, the OCD group demonstrated poor performance on tests that assess visuospatial and visuomotor ability but performance on tests of verbal comprehension, verbal fluency and verbal memory was comparable with that of controls. Performance of the OCD group on tests of executive functioning, such as planning, cognitive flexibility and set-shifting, was mixed. Given evidence of gender dimorphism in OCD, ^{Reference Zohar, Gross-Isseroff, Hermesh and Weizman13} it is significant that these effects persisted despite controlling for gender.

Although there are no comparable prospective studies, on the basis of cross-sectional studies of neuropsychological functioning and OCD, we expected impairment on tasks that reflect visuospatial, visuoconstructive and visuomotor skills and non-verbal memory. ^{Reference Kuelz, Hohagen and Voderholzer4} The poor performance observed across the ROCF, Grooved Pegboard, WISC–R Mazes and the WISC perceptual organisation factor supported our predictions. We also expected to observe some relation between executive skills and OCD; however, the findings were divergent. Performance of the OCD group on tests that evaluated both visuospatial/visuomotor skills and executive functioning (ROCF, WISC–R Mazes) was impaired, whereas performance on tests that evaluated other executive functions such as set-shifting and divided attention (WCST, TMT) was comparable between the groups. The equivalent performance of the OCD group on the WISC–R full-scale IQ, as well as on the WISC–R verbal tests, the TMT and the WCST, suggests that there is a pattern of specific deficits associated with OCD, rather than overall cognitive or neuropsychological impairment.

A previous study of OCD with this same birth cohort at age 18 found that the OCD group did not differ significantly on any of the neuropsychological tests at age 13. ^{Reference Douglass, Moffitt, Dar, McGee and Silva14} Changes in the OCD criteria from DSM–III ¹⁵ to DSM–IV have reduced the reported prevalence of the disorder in the general population. ^{Reference Crino, Slade and Andrews16} The application of the slightly more stringent DSM–IV diagnostic criteria may have resulted in the identification of a more valid OCD group; only 1.9% of age 32 study members met criteria for OCD compared with the 4% prevalence rate in the earlier study. ^{Reference Douglass, Moffitt, Dar, McGee and Silva14} In addition, from a developmental perspective, study members who were diagnosed with OCD at age 32 may represent a more severe and persistent subgroup.

A limitation to the current study was the small number of study members who had OCD outcomes at age 32. In addition, although the findings suggest impairment in visuospatial and visuomotor performance and non-verbal memory, neuropsychological tests assess performance in more than one domain. For example, memory dysfunction on the ROCF may be secondary to impaired executive functioning, in particular the inability to apply efficiently elaborated strategies. ^{Reference Savage, Keuthen, Jenike, Brown, Rauch and Jenike17} In addition, consistent with high rates of comorbidity generally found among mood and anxiety disorders, ^{Reference Brown, Campbell, Lehman, Grisham and Mancill18} the majority of the OCD group met criteria for an additional Axis I disorder. This introduces the possibility that differences between the OCD and the comparison group were partially accounted for by the presence of comorbid psychopathology. Finally, although neuropsychological measures provide clues about neuroanatomical substrates, we cannot directly conclude neurobiological dysfunction from poor test performance.

Despite these limitations, the current findings are noteworthy because of the unique longitudinal nature of the study. We have identified associations between childhood neuropsychological test performance and adult OCD that span two decades. We found selective deficits in visuospatial and visuoconstructive abilities and non-verbal memory, with intact verbal functioning and ambiguous frontal lobe functioning. Theories arguing that developmental disruption to brain maturation processes (such as abnormal postnatal pruning in the ventromedial prefrontal cortex) contributes to OCD ^{Reference Rosenberg and Keshavan19} may view these premorbid childhood deficits as support in that individuals were exhibiting neuropsychological deficits before they had developed the full OCD syndrome. It is important to note that OCD is likely to be multiply determined ^{Reference Grisham, Anderson and Sachdev20} and not all participants with these deficits developed OCD. Consistent with a diathesis–stress model of OCD, whether particular neuropsychological deficits result in the onset of OCD may depend on psychosocial factors, such as parenting styles or critical stressful life events.

Funding

The Dunedin Multidisciplinary Health and Development Research Unit is supported by the New Zealand Health Research Council. This research was supported by grants from the US National Institute of Mental Health (MH 45070, MH49414), the UK Medical Research Council grant G0100527, and a Postdoctoral Fellowship with the School of Psychiatry, University of New South Wales.