Although there has been much research both on self-care and on cognitive function in diabetes, the relationship between the two remains under-researched.
In one of the few studies in this area, Sinclair et al (2000) examined whether or not cognitive impairment is associated with changes in self-care behaviour and use of health and social services in a community-based case control study of older people with diabetes. Cognitive function was assessed using two global cognition measures, the Mini-Mental State Examination (MMSE; Folstein et al, 1975) and the Clock-Drawing Test (CDT; Shulman, 2000). Self-care was measured by recording the number of people who were solely responsible for self-medication and blood glucose (BG) monitoring and their attendance at a specialist diabetes clinic. Use of the CDT demonstrated that 65% and 72% of people with diabetes, respectively, placed the clock numbers and hands correctly, compared with 76% and 84% of controls. Elderly people with diabetes displayed a significant cognitive dysfunction that was associated with poorer abilities in diabetes self-management and greater dependency.
Asimakopoulou and Hampson (2002) argued that the lack of consensus over the cognitive functions and instruments that should be assessed and used in people with diabetes makes this area of research problematic. In their study, 51 people with type 2 diabetes completed a battery of cognitive tests and the Summary of Diabetes Self-Care Activities questionnaire (SDSCA; Toobert and Glasgow, 1994), but only a few associations between cognitive functioning and self-management were observed. This lack of association may be due to limited statistical power or the absence of a significant practical association between self-reported self-care and specific cognitive skills. One of the few significant associations that was found was the inverse relationship between self-reported memory problems and number of diabetes problem-solving strategies, although self-reported memory complaints were not a reliable indicator of objective cognitive function in the study.
Better dietary self-management was predicted by better general and diabetes-specific abstract reasoning as assessed by the modified Wisconsin Card Sorting Test (Hart et al, 1998) and the Diabetes Problem-Solving Interview (Toobert and Glasgow, 1991), respectively. Better exercise self-management was predicted by better scores on a test of mental flexibility, the Serial Subtractions of 7s (Lezak, 1995), and generating more problem-solving strategies in the Diabetes Problem-Solving Interview was predicted by fewer subjective memory problems. This study assessed self-reported self-care through the SDSCA. In later work, however, Asimakopoulou and Hampson (2005) showed that the SDSCA can be prone to recall biases in people with diabetes.
Other studies have examined self-care on the basis of medication adherence and glycaemic control. Rosen et al (2003) examined the association between cognitive performance and adherence to prescribed medication, HbA1c and missed appointments. Cognitive function was assessed using a variety of global and specific neuropsychological tests. Adherence to metformin was measured using pill bottle caps, which contained a microprocessor that recorded the date and times of bottle openings; the caps were placed on the patients’ prescribed antihyperglycaemic medication.
Medication adherence was independently associated with scores on the Stroop word test (a measure of attention and flexibility, Lezak, 1995) and with Trails B completion time (a measure of motor speed, visual scanning, attention and flexibility; Reitan and Wolfson, 1993) but, interestingly, neuropsychological performance was not associated with HbA1c levels. Missed appointments were associated with impaired performance on the MMSE (Mini Mental State Exmination). The authors concluded that ‘cognitive abilities should be considered when counselling patients concerning their adherence’.
More recently, Trimble et al (2005) assessed the ability of the CDT to predict problematic insulin administration skills in older adults with diabetes. Thirty individuals who had not used insulin before were taught to self-administer a sham insulin injection with an insulin pen using a standardised protocol. Injections were performed for 7 days, after which self-administration was re-tested. An abnormal CDT was significantly associated with more problems in learning to perform the sham injections (measured as those who were unable to correctly complete all steps of the protocol, or those who omitted all or part of a step), although a small number of people with a normal CDT also demonstrated major problems. The results were in line with other studies that noted the frequency of abnormal CDTs in older people (Sinclair et al, 2000) and the frequency of errors in older people self-administering insulin (Coscelli et al, 1992), and suggested that the CDT is a valuable predictor of potential problems with insulin administration skills in elderly individuals.
Finally, Munshi et al (2006) assessed the relationship between global cognitive function as measured by the MMSE, CDT and Clock in Box (CIB) tests, as well as glycaemic control, measured by HbA1c, in older adults with diabetes. In total, 34% of people had low scores on the CIB and 38% had low scores on the CDT. Both CIB and CDT were superior at identifying those with cognitive dysfunction, compared with MMSE. CIB was more sensitive in predicting poor glycaemic control than CDT. Both clock tests were inversely correlated with HbA1c levels, suggesting that cognitive function may play a role in the control of diabetes.
Table 1 summarises the tests the reviewed studies have used, the domains they measure and the reported main findings.
It thus appears that the few studies that have assessed diabetes self-management and cognition together tend to argue for a relationship between cognitive dysfunction and impaired self-care in people with diabetes. In order to help clinicians to identify and assist those who are less likely to be able to self-manage their diabetes, it is important to be able to assess whether or not cognitive impairment is associated with ability to self-care.
Assessing cognition and self-care in primary care settings: A practical guide
Variability in the way that cognitive function and self-management have been assessed is evidenced from the literature. Measures of cognition have included global (MMSE or CDT) or specific (Stroop or SS7) function tests, while measures of self-care have included self-report (SDSCA) and HbA1c levels. Although self-reports of any health behaviour can be unreliable (Abraham and Hampson, 1996), HbA1c as a measure of self-care (rather than glycaemic control) is also problematic as it can be affected by a myriad of factors that don’t relate to self-care, including how aggressive a treatment regime is, sickness and stress. Despite these difficulties, we suggest some straightforward ways of assessing cognition and self-care in primary care settings.
It has been argued that ‘a strong background in neuropathology, neuroanatomy, basic neuropsychological principles, clinical and cognitive psychology’ is needed before diagnostic cognitive testing is undertaken (Lezak, 1995). In light of this, we propose that some testing can take place by non-experts in neuropsychology, as long as the test results are used only as signs for further referral, rather than as diagnostic instruments.
Two global function tests that have been used successfully before with older people with diabetes are the MMSE and the Clock-Drawing Test. The former is the most widely used dementia screening tool, takes approximately 10 minutes to administer and consists of questions relating to attention, orientation, memory, calculation and language. It relies heavily on language and, as such, would not be suitable for non-English speakers (although it is available in different languages). For example, individuals are, among other tasks, asked to tell the examiner the year, month, date, day and time and to spell the word ‘world’ backwards. They are also asked to name three objects that are in the examination room and, a few minutes later, unexpectedly recall them (Figure 1). Although a reliable indicator of moderate-to-severe cognitive impairment, the MMSE is not sensitive enough to detect mild cognitive impairment. This may not necessarily be a problem as mild cognitive impairment is unlikely to be related to diabetes self-care activities in any significant way (Asimakopoulou and Hampson, 2002).
The Clock-Drawing Test is another popular measure that is quick and easy to administer. Participants are given a circle (4–10cm in diameter), told that it represents a clock face and are instructed to ‘put in the numbers so that it looks like a clock and then set the time to 10 minutes past 11’. The test assesses executive function and, in particular, abilities such as planning, visuo-spatial ability, abstract reasoning and concentration. The test can be scored in several ways (Shulman, 2000) although 4- (Death et al, 1993) and 5-point systems (Shulman et al, 1993) are probably the quickest and easiest. Using the latter, the individual’s drawing is assessed from being perfect (scored 5) to showing inaccurate representation of 10 past 11 when the overall visuo-spatial organisation is good (scored 3), down to 0 for inability to make any reasonable representation of a clock (Shulman et al, 1993). Completed examples of the CDT using this severity scale are shown in Figure 2.
In their study of the validity of the CDT compared with the MMSE, Nishiwaki et al (2004) found that the CDT was better at detecting moderate/severe cognitive impairment than mild impairment. The test sensitivity was better for females and increased with age. Higher CDT scores were associated with higher mortality from cerebrovascular disease. The authors noted that in isolation, the MMSE might not detect mild impairment, while the CDT might produce a large number of false positives; used together however, it has been argued that these tests can be reliable predictors of moderate-to-severe cognitive dysfunction.
In terms of self-care, a revised version of the Toobert and Glasgow (1991) Summary of Diabetes Self-Care Activities measure is a reliable and valid way to assess self-care via self-report (Toobert et al, 2000). This brief questionnaire assesses, among other aspects of self-care, dietary behaviour, exercise, glucose monitoring and medication taking, in separate sections. It can be completed while the individual waits to be seen by their diabetes healthcare professional and can be scored very quickly (Toobert et al, 2000). Each section then provides a clear quantitative view of the individual’s self-care efforts over the past week. It is suggested that alongside HbA1c readings, this measure may be used as a preliminary indicator of diabetes self-care areas that the person is struggling with and, as such, be instrumental in helping an informed discussion between clinicians and patients.
We have reviewed several studies examining whether or not there is a relationship between cognitive function and self-care in type 2 diabetes. In doing so, we have highlighted the variability in measures used to assess cognition and self-care and have proposed some straightforward tools that can be easily obtained and used in primary care to assess cognition and self-care. We have also noted that these tests are not meant to replace clinical judgement or offer a diagnosis. As they are fairly insensitive in detecting mild cognitive impairment, their usefulness with people who clinicians suspect might fall into this category is questionable. Assuming they are used ethically and alongside clinical opinion, research has shown that they can be useful indicators of people’s cognitive functioning.
We conclude that there is evidence supporting a relationship between cognitive dysfunction and self-care in diabetes and, as such, clinicians may find it helpful to assess both of these in people with type 2 diabetes. Further work in determining the clinical relevance to diabetes self-care and overall medical management of both minor and moderate cognitive changes is needed.