A 4 year follow-up study of cognitive functioning in patients with type 2 diabetes mellitus

The baseline examination (2002–2004) included 122 patients with type 2 diabetes and 56 control participants aged between 56 and 80 years, matched on age, sex and estimated IQ [8, 11]. Patients were recruited through general practitioners in the region. Control participants were recruited among the spouses or acquaintances of the patients. For inclusion the patients had to have had type 2 diabetes for at least 1 year, be functionally independent and Dutch speaking. Exclusion criteria for all participants were a psychiatric or neurological disorder (unrelated to diabetes) that could influence cognitive functioning and a history of alcohol or substance abuse or dementia. Control participants with a fasting blood glucose ≥7.0 mmol/l were also excluded. At follow-up 4 years later, seven participants had died, four could not be contacted and 59 were not willing or able to participate. Reasons for not participating were: lack of interest (n = 28); comorbidity (n = 22; three reported dementia [two patients, one control]); and other reasons (n = 9). The remaining 108 participants were re-examined between 2006 and 2008 (mean follow-up time 4.1 ± 0.4 years). One patient with type 2 diabetes was excluded because of severe comorbid disease and one control participant fulfilled the criteria for type 2 diabetes and was therefore excluded from the control group, leaving 106 participants (68 patients and 38 control participants) in the present analysis.

The non-participants (n = 70) did not differ from the participants (n = 106) with regard to baseline age, sex or estimated premorbid IQ (all p > 0.05). To control for possible selective loss at follow-up we examined the cognitive status of both non-participants and participants (±1 week after their participation in the follow-up examination) with the Dutch version of the Telephone Interview for Cognitive Status [12], a 12 item screening instrument designed to identify persons with dementia [13]. The Interview was slightly modified (TICS-m) by including a delayed word-list recall which resulted in a maximum score of 50 [14]. The TICS-m was obtained from 43 of the 70 non-participants (i.e. 73% of the 59 non-participants who were still alive and could be contacted) and 99 of the 106 participants (93%). TICS-m scores were normally distributed across the whole study sample. The TICS-m performance for the non-participants was similar to the participants (non-participants mean score 35.4 ± 5.2, participants 36.5 ± 4.6, F[1,140] = 1.67, p = 0.20). Only three patients with type 2 diabetes (two non-participants, one participant) and two control participants (one non-participant, one participant) performed below the cut-off score of 28 (χ ²[1] = 0.24, p = 0.62), which is indicative for cognitive impairment [15]. Hence, among the non-participants four patients with type 2 diabetes (3% of baseline sample) and two controls (4% of baseline sample) had cognitive impairment based on self-reported dementia or a low TICS-m score (χ ²[1] = 0.10, p = 0.76). The study was approved by the medical ethics committee of the University Medical Center Utrecht, Utrecht, the Netherlands. Written informed consent was obtained from all participants.

At follow-up all participants performed an extensive neuropsychological assessment, identical to the baseline examination [8]. Parallel versions were used for memory tests to control for possible material-specific learning effects [16]. The neuropsychological assessment consisted of 11 verbal and non-verbal tasks, administered in a fixed order that took about 90 min to complete. The tasks were divided into five cognitive domains to reduce the amount of neuropsychological variables in the analysis and for clinical clarity [8]. This division was made a priori, according to standard neuropsychological practice and cognitive theory, as described in detail in Lezak et al. [16]. The domain ‘abstract reasoning’ was assessed by Raven Advanced Progressive Matrices (12 item short form). The domain ‘memory’ included four subdomains: ‘working memory’ assessed by the forward and backward digit span of the Wechsler Adult Intelligence Scale, 3rd edition (WAIS-III) and the Corsi Block-tapping Task (the product scores of the span length × the number of correctly recalled sequences were recorded [17]); ‘immediate memory and learning rate’, including verbal memory assessed by the Rey Auditory Verbal Learning Test and visual memory assessed by the Location Learning Test; ‘forgetting rate’ assessed by the delayed task of the Rey Auditory Verbal Learning Test and the Location Learning Test; and ‘incidental memory’ assessed by the delayed trial of the modified Taylor Complex Figure. The domain ‘information-processing speed’ was assessed by the Trail-making Test—Part A, the Stroop Color–Word Test (Parts I and II), and the subtest Digit Symbol of the WAIS-III. The domain ‘attention and executive function’ was assessed by the Trail-making Test—Part B (ratio score), the Stroop Color–Word Test (Part III; ratio score), the Brixton Spatial Anticipation Test, a letter fluency test using the ‘N’ and ‘A’, and category fluency (animal naming). The domain ‘visuoconstruction’ was assessed by the copy trial of the modified Taylor Complex Figure.

To compare the five different cognitive domains between the two groups the raw test scores were standardised into z scores per cognitive domain. These z scores were calculated by using the pooled mean of baseline scores of the whole study sample. The z score for each domain was derived by calculating the mean of the z scores for tests comprising that domain. Depressive symptoms were assessed with the Dutch version of the Beck Depression Inventory, 2nd edition (BDI-II [18]). The total score on this self-rated depressive symptoms inventory and the proportion of persons scoring >13 were recorded [19].

Procedures at baseline and follow-up were identical [8]. Medical history was assessed with a standardised questionnaire addressing diabetes duration, medication use, history of cardiovascular disease (including stroke), smoking and alcohol consumption. Fasting glucose, HbA_1c and cholesterol levels were measured with standard laboratory testing. Blood pressure was measured in a seated position at three time-points during the half-day visit (Omron MX3; Omron, Mannheim, Germany). Hypertension was defined as a systolic blood pressure >160 mmHg, a diastolic blood pressure >95 mmHg or self-reported use of blood-pressure-lowering medication. ‘Any macrovascular event’ was defined as a history of myocardial infarction, stroke, or surgery or endovascular treatment for coronary, carotid or peripheral (legs, abdominal aorta) artery disease. Retinopathy was defined as a score of ≥1.5 on the Wisconsin Epidemiologic Study of Diabetic Retinopathy scale [20]. Neuropathy was defined as a score ≥6 on a modified version of the Toronto Clinical Neuropathy Scoring System [21, 22].

Between-group differences in characteristics were analysed with analysis of variance for continuous variables, Mann–Whitney U tests for non-parametric data and χ ² tests for proportions. The primary outcome measures were the z scores of the five cognitive domains, which were analysed with repeated-measures ANOVA, including the effect of time, group and the time × group interaction. The effect of time reflects the mean change in cognitive performance over time for the whole study sample; the effect of group reflects the mean difference between the patients with type 2 diabetes and the control group; the time × group interaction reflects the additional change over time attributable to diabetes status. p values <0.05 were considered statistically significant.

In a secondary analysis additional adjustment for BDI-II depressive symptoms was performed to examine a possible confounding effect of depression. Moreover, within the diabetes group cognitive functioning (domains information processing speed and attention and executive functioning) was compared between patients with high or low baseline HbA_1c (dichotomised at the median level of 6.6 mmol/l), with or without baseline hypertension and with or without ‘any macrovascular event’ at baseline.