Response to Letter Regarding Article, “Carotid Intima-Media Thickness Is Associated With the Progression of Cognitive Impairment in Older Adults”
We appreciate the interest of Dr Kawada in our recently published article with reference to an important issue about the statistical validity and causality of observational studies. We conducted a 5-year cohort study to evaluate the association between carotid atherosclerosis and progression of cognitive impairment in 348 older adults. We used multivariate logistic regression analyses with adjustment for age, education, mood, and general health status to predict the progress of cognitive impairment or mild cognitive impairment. In our analyses, 56 (16.1%) of the 348 included subjects showed progression of cognitive dysfunction, and 50 (18.2%) of 275 subjects who were cognitively normal at the baseline evaluation were newly diagnosed with mild cognitive impairment.
We agree that sex is an important cardiovascular disease risk factor. The objective of our study was to investigate the association of cardiovascular disease risk factors, including carotid intima-media thickness (CIMT) and other surrogate markers, with the future risk of clinically diagnosed mild cognitive impairment or dementia. When we compared various baseline cardiovascular disease risk factors, including sex, according to the progression of cognitive dysfunction, we found that the sex distribution did not differ between the progression and nonprogression groups. In this comparison, the CIMT differed between these 2 groups, and therefore, we performed multivariable logistic regression analyses to test the independent association between CIMT and the progression of cognitive dysfunction. We included well-established risk factors for cognitive impairment (age, education, mood, general health status) and the parameters that differed between the progression and nonprogression groups (CIMT and hypertension) as independent variables in the logistic regression analyses. We did not include sex as an independent variable because previous studies reported inconsistent results about the sex difference of dementia incidence,1,2 and our baseline comparison show no difference in the sex distribution. When we included sex as an independent variable, the association between CIMT and the progression of cognitive dysfunction was maintained (P=0.049, hazard ratio=1.245 in the fully adjusted model).
The rule of thumb that recommends using logistic and Cox models with a minimum of 10 outcome events per predictor variable is based on 2 simulation studies.3,4 In our study, we used 8 predictor variables, and the number of events was 56 (events per predictor variable=7). We agree that the results from a small number of the events should be interpreted with caution. However, situations violating this rule commonly arise to address confounding factors persuasively, especially in observation studies. A recent study suggested that the rule of thumb might be too conservative and that 5 to 9 events per predictor variable could be acceptable.5 In addition, we presented 3 statistical models that included 5, 7, or 8 predictor variables in the logistic regression analyses, and the events per predictor variable value was ≥10 in the model, including 5 predictor variables, and ≥5 in all statistical models. Moreover, a greater CIMT was associated with the development of mild cognitive impairment or dementia with similar hazard ratios and significance levels in all statistical models.
The mean IMT is measured as an index of arterial wall thickening, whereas the maximum IMT provides an index of arterial wall thickening and atheroma formation. We measured mean value of IMT in the carotid artery with a computerized method and determined CIMT as the average of all measurements from both the left and right arteries. Although we did not include the maximum IMT in the analyses, carotid artery plaque formation indicating atheroma formation was included in the analyses.
Jae Hoon Moon, MD, PhD*
Soo Lim, MD, PhD*
Hak Chul Jang, MD, PhD
Department of Internal Medicine
Seoul National University Bundang Hospital
Seoul National University College of Medicine
Seongnam-si, Kyunggi-do, Korea
Stroke welcomes Letters to the Editor and will publish them, if suitable, as space permits. Letters must reference a Stroke published-ahead-of-print article or an article printed within the past 3 weeks. The maximum length is 750 words including no more than 5 references and 3 authors. Please submit letters typed double-spaced. Letters may be shortened or edited.
- © 2015 American Heart Association, Inc.