Mild Cognitive Dysfunction and DM Control in Elderly Adults
Participants in these analyses were a subsample recruited for a cognition ancillary study from the New York City sample of the Informatics for Diabetes Education and Telemedicine (IDEATel) Project. IDEATel was designed to assess the feasibility and effectiveness of telemedicine for disease management in a sample of minority older adults with type 2 diabetes mellitus living in New York State. Their primary care providers recruited participants from New York City (Columbia University) and upstate New York (State University of New York Upstate Medical University at Syracuse). They were randomized between 2000 and 2002 to the intervention (a home-based interactive telemedicine unit used for televisits with a nurse educator, transmission of self-measured glucose and blood pressure data electronically, and web access in addition to usual care) or usual care alone. The primary endpoints were changes in HbA1c, blood pressure, and LDL-C. Participants (N = 1,665) were individuals aged 55 and older with a physician or medication-defined diagnosis of diabetes mellitus who were Medicare beneficiaries and resided in a federally designated medically underserved area. Participants were excluded if they were moderately to severely cognitively impaired; had a severe visual, mobility, or motor impairment; had a severe comorbid condition, a communication impairment, or no electrical outlet for the telemedicine unit; or planned to reside in another location for longer than 3 months. Details regarding inclusion and exclusion, the randomization scheme, the intervention, and evaluation of primary endpoints are described elsewhere.
An ancillary cognition study was initiated in the second phase (2004–08) of IDEATel at the Columbia University site. Five hundred participants who completed the first phase at this site (n = 775) continued on to the second phase. An additional 150 subjects were recruited to participate in the second phase. The only exclusion criteria for this ancillary study were unwillingness or inability to begin or complete the cognitive assessments. In addition to the global cognition measures administered in the first phase, an extended neuropsychological assessment with measures of memory, executive function, language, attention, and construction were administered in this ancillary study at 1-year intervals. Six hundred thirteen participants were included in this 5-year prospective analysis of the ancillary study data, 538 (87.7%) of whom had one follow-up visit, 437 (71.2%) two follow-up visits, 350 (57.1%) three follow-up visits, 231 (37.7%) four follow-up visits, and 90 (14.7%) five follow-up visits. Written informed consent was obtained from participants. The Columbia University institutional review board approved all protocols for this study.
Cognitive dysfunction was classified as executive and memory dysfunction because these types seem to be most relevant to the management of diabetes mellitus. Executive dysfunction is broadly defined as the ability to plan, initiate, and execute complex tasks (e.g., planning and completing diabetes mellitus treatment), whereas memory is defined as the ability to recall (e.g., remembering to take diabetes mellitus medications). Executive dysfunction was assessed based on the participant's performance on the Color Trails Test Part 2, a test of executive functioning and mental flexibility. The Color Trails Test is similar in form to the Trail-Making Test Part B, but it removes the cultural bias associated with unfamiliarity with the English alphabet. The majority of the cohort was Hispanic, and this test is well suited to individuals whose first language is not English. Instead of connecting numbers to Arabic letters, the Color Trails Test requires participants to connect numbers and colors. Each number is overlaid on a colored circle, pink or yellow. Participants are asked to connect the numbers 1–25 in ascending order but to alternate between colors (pink-1, yellow-2, pink-3, yellow-4, etc.). A cutoff for an abnormal score was defined as performance in less than the 16th percentile (roughly 1 standard deviation (SD) below the standardized test score mean). This cutoff was used because test performance 1 SD below the mean is typically used in clinical settings to categorize individuals as having abnormal cognitive functioning. Studies of cognition, in general, are increasingly studying this range of cognitive dysfunction as compared to the more frequently used classification of mild cognitive impairment (MCI), which uses 1.5 SDs as a cutoff for normal cognitive performance.
Memory dysfunction was assessed based on the participant's performance on the total immediate memory task of the Selective Reminding Test (SRT), a 12-item list-learning verbal memory test. SRT total score was converted to a T-score based on the participant's age, sex, education, and race and ethnicity. An abnormal T-score was defined as performance in under the 16th percentile (1 SD below the standardized test score mean) of SRT immediate recall T-scores in the sample. Based on this criterion, a participant with a T-score <28 on the SRT immediate recall task was classified with memory dysfunction.
Cognitive dysfunction was examined in terms of executive function and memory as an exposure in two ways: relating cognitive dysfunction at baseline to the outcomes and using all available follow-up data on cognitive dysfunction and classifying participants as ever or never having cognitive dysfunction. There was little difference in results between using the baseline cognitive dysfunction status definition and using the ever–never definition.
Changes in measures of metabolic control (HbA1c, SBP, LDL-C) were evaluated. Twelve-hour fasting blood samples were collected to assess HbA1c and lipid levels. HbA1c was analyzed using boronate affinity chromatography using high-performance liquid chromatography (Primus CLC 385; Primus, Kansas City, MO). Lipid levels were analyzed using enzymatic colorimetric methods (Vitros; Johnson & Johnson, New Brunswick, NJ). LDL-C was calculated using the Friedwald equation. SBP was determined as the average of the last two of three readings taken 1 minute apart (Dinamap PRO 100; Critikon, Tampa, FL).
Descriptive analyses were conducted using chi-square tests for categorical variables and the Kruskal–Wallis test for continuous variables to test for significant differences in participant characteristics between individuals with and without cognitive dysfunction, executive and memory type separately.
Linear mixed models were used to examine the longitudinal relationship between ever versus never presence of executive or memory dysfunction and changes in parameters of metabolic control (HbA1c, SBP, LDL-C). Models included random effects for intercepts (individuals) and clustering within primary care provider, which was the unit of recruitment in IDEATel. To estimate the association between executive or memory dysfunction and changes in measures of metabolic control, an interaction term between cognitive dysfunction status (executive or memory type) and time was included. Nonlinearity was assessed and goodness-of-fit statistics were used (Akaike Information Criterion and Schwarz's Bayesian Information Criterion) to assess model fit. Model fit was not improved with inclusion of a quadratic (group by time) or exponential term (group by e) for time. Demographic characteristics that are known to be associated with cognitive dysfunction and the metabolic measures (age, sex, years of education, race and ethnicity) were adjusted for. To account for residual effects of the IDEATel treatment, the randomization group assignment was included in the final model. Symptomatology of depression and cognitive dysfunction are similar and often mistaken for one another, so the presence or absence of depressive symptoms, as measured using the SHORT Comprehensive Assessment and Referral Evaluation Depression questionnaire, was included in the final model. Insulin, metformin, sulfonylurea, or thiazolidinedione medication use was further adjusted for in models of HbA1c to account for possible confounding by indication. Analyses were performed using Stata 13.0 (Stata Corp, College Station, TX).
Methods
Study Population and Design
Participants in these analyses were a subsample recruited for a cognition ancillary study from the New York City sample of the Informatics for Diabetes Education and Telemedicine (IDEATel) Project. IDEATel was designed to assess the feasibility and effectiveness of telemedicine for disease management in a sample of minority older adults with type 2 diabetes mellitus living in New York State. Their primary care providers recruited participants from New York City (Columbia University) and upstate New York (State University of New York Upstate Medical University at Syracuse). They were randomized between 2000 and 2002 to the intervention (a home-based interactive telemedicine unit used for televisits with a nurse educator, transmission of self-measured glucose and blood pressure data electronically, and web access in addition to usual care) or usual care alone. The primary endpoints were changes in HbA1c, blood pressure, and LDL-C. Participants (N = 1,665) were individuals aged 55 and older with a physician or medication-defined diagnosis of diabetes mellitus who were Medicare beneficiaries and resided in a federally designated medically underserved area. Participants were excluded if they were moderately to severely cognitively impaired; had a severe visual, mobility, or motor impairment; had a severe comorbid condition, a communication impairment, or no electrical outlet for the telemedicine unit; or planned to reside in another location for longer than 3 months. Details regarding inclusion and exclusion, the randomization scheme, the intervention, and evaluation of primary endpoints are described elsewhere.
An ancillary cognition study was initiated in the second phase (2004–08) of IDEATel at the Columbia University site. Five hundred participants who completed the first phase at this site (n = 775) continued on to the second phase. An additional 150 subjects were recruited to participate in the second phase. The only exclusion criteria for this ancillary study were unwillingness or inability to begin or complete the cognitive assessments. In addition to the global cognition measures administered in the first phase, an extended neuropsychological assessment with measures of memory, executive function, language, attention, and construction were administered in this ancillary study at 1-year intervals. Six hundred thirteen participants were included in this 5-year prospective analysis of the ancillary study data, 538 (87.7%) of whom had one follow-up visit, 437 (71.2%) two follow-up visits, 350 (57.1%) three follow-up visits, 231 (37.7%) four follow-up visits, and 90 (14.7%) five follow-up visits. Written informed consent was obtained from participants. The Columbia University institutional review board approved all protocols for this study.
Assessment of Cognitive Dysfunction
Cognitive dysfunction was classified as executive and memory dysfunction because these types seem to be most relevant to the management of diabetes mellitus. Executive dysfunction is broadly defined as the ability to plan, initiate, and execute complex tasks (e.g., planning and completing diabetes mellitus treatment), whereas memory is defined as the ability to recall (e.g., remembering to take diabetes mellitus medications). Executive dysfunction was assessed based on the participant's performance on the Color Trails Test Part 2, a test of executive functioning and mental flexibility. The Color Trails Test is similar in form to the Trail-Making Test Part B, but it removes the cultural bias associated with unfamiliarity with the English alphabet. The majority of the cohort was Hispanic, and this test is well suited to individuals whose first language is not English. Instead of connecting numbers to Arabic letters, the Color Trails Test requires participants to connect numbers and colors. Each number is overlaid on a colored circle, pink or yellow. Participants are asked to connect the numbers 1–25 in ascending order but to alternate between colors (pink-1, yellow-2, pink-3, yellow-4, etc.). A cutoff for an abnormal score was defined as performance in less than the 16th percentile (roughly 1 standard deviation (SD) below the standardized test score mean). This cutoff was used because test performance 1 SD below the mean is typically used in clinical settings to categorize individuals as having abnormal cognitive functioning. Studies of cognition, in general, are increasingly studying this range of cognitive dysfunction as compared to the more frequently used classification of mild cognitive impairment (MCI), which uses 1.5 SDs as a cutoff for normal cognitive performance.
Memory dysfunction was assessed based on the participant's performance on the total immediate memory task of the Selective Reminding Test (SRT), a 12-item list-learning verbal memory test. SRT total score was converted to a T-score based on the participant's age, sex, education, and race and ethnicity. An abnormal T-score was defined as performance in under the 16th percentile (1 SD below the standardized test score mean) of SRT immediate recall T-scores in the sample. Based on this criterion, a participant with a T-score <28 on the SRT immediate recall task was classified with memory dysfunction.
Cognitive dysfunction was examined in terms of executive function and memory as an exposure in two ways: relating cognitive dysfunction at baseline to the outcomes and using all available follow-up data on cognitive dysfunction and classifying participants as ever or never having cognitive dysfunction. There was little difference in results between using the baseline cognitive dysfunction status definition and using the ever–never definition.
Outcomes
Changes in measures of metabolic control (HbA1c, SBP, LDL-C) were evaluated. Twelve-hour fasting blood samples were collected to assess HbA1c and lipid levels. HbA1c was analyzed using boronate affinity chromatography using high-performance liquid chromatography (Primus CLC 385; Primus, Kansas City, MO). Lipid levels were analyzed using enzymatic colorimetric methods (Vitros; Johnson & Johnson, New Brunswick, NJ). LDL-C was calculated using the Friedwald equation. SBP was determined as the average of the last two of three readings taken 1 minute apart (Dinamap PRO 100; Critikon, Tampa, FL).
Statistical Analysis
Descriptive analyses were conducted using chi-square tests for categorical variables and the Kruskal–Wallis test for continuous variables to test for significant differences in participant characteristics between individuals with and without cognitive dysfunction, executive and memory type separately.
Linear mixed models were used to examine the longitudinal relationship between ever versus never presence of executive or memory dysfunction and changes in parameters of metabolic control (HbA1c, SBP, LDL-C). Models included random effects for intercepts (individuals) and clustering within primary care provider, which was the unit of recruitment in IDEATel. To estimate the association between executive or memory dysfunction and changes in measures of metabolic control, an interaction term between cognitive dysfunction status (executive or memory type) and time was included. Nonlinearity was assessed and goodness-of-fit statistics were used (Akaike Information Criterion and Schwarz's Bayesian Information Criterion) to assess model fit. Model fit was not improved with inclusion of a quadratic (group by time) or exponential term (group by e) for time. Demographic characteristics that are known to be associated with cognitive dysfunction and the metabolic measures (age, sex, years of education, race and ethnicity) were adjusted for. To account for residual effects of the IDEATel treatment, the randomization group assignment was included in the final model. Symptomatology of depression and cognitive dysfunction are similar and often mistaken for one another, so the presence or absence of depressive symptoms, as measured using the SHORT Comprehensive Assessment and Referral Evaluation Depression questionnaire, was included in the final model. Insulin, metformin, sulfonylurea, or thiazolidinedione medication use was further adjusted for in models of HbA1c to account for possible confounding by indication. Analyses were performed using Stata 13.0 (Stata Corp, College Station, TX).
Source...