Skip Navigation


American Journal of Epidemiology Advance Access originally published online on November 14, 2006
American Journal of Epidemiology 2007 165(3):344-350; doi:10.1093/aje/kwk017
This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
165/3/344    most recent
kwk017v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (4)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Proust-Lima, C.
Right arrow Articles by Jacqmin-Gadda, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Proust-Lima, C.
Right arrow Articles by Jacqmin-Gadda, H.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?

American Journal of Epidemiology Copyright © 2006 by the Johns Hopkins Bloomberg School of Public Health All rights reserved; printed in U.S.A.

PRACTICE OF EPIDEMIOLOGY

Sensitivity of Four Psychometric Tests to Measure Cognitive Changes in Brain Aging-Population–based Studies

Cécile Proust-Lima1,2, Hélène Amieva2,3, Jean-François Dartigues2,3 and Hélène Jacqmin-Gadda1,2

1 INSERM E0338, Bordeaux, France
2 Université Victor Segalen Bordeaux 2, Bordeaux, France
3 INSERM U593, Bordeaux, France

Correspondence to Cécile Proust-Lima, INSERM E0338, Institut de Santé Publique, d'Epidémiologie et de Développement, Université de Bordeaux 2, 146 rue Léo Saignat, 33076 Bordeaux Cedex, France (e-mail: cecile.proust{at}isped.u-bordeaux2.fr).

Received for publication April 11, 2006. Accepted for publication July 5, 2006.


   ABSTRACT
TOP
 ABSTRACT
INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX
 References
 
Choosing the measure of cognition in an epidemiologic study investigating cognitive changes over time is a challenging question. A powerful measure must be able to detect small cognitive changes in all the range of cognition observed in the target population. This work aims at comparing the sensitivity to detect cognitive changes in the observed range of cognition of four widely used psychometric tests in an aging-population–based study through a nonlinear latent process model, assuming that the psychometric tests are nonlinear noisy transformations of their common factor. With data from the French prospective cohort study PAQUID (1989–2001), the authors found that the Mini-Mental State Examination and the Benton Visual Retention Test exhibited a better sensitivity to cognitive changes in low levels of cognition, while the Digit Symbol Substitution Test was more sensitive to changes in high levels of cognition. In contrast, the Isaacs Set Test shortened at 15 seconds appeared to be sensitive to small changes in all the range of cognition and, thus, represents an appropriate measure of cognition in population-based studies including both highly normal and severely impaired subjects.

aging; cognition; dementia; epidemiologic methods; neuropsychological tests

Abbreviations: BVRT, Benton Visual Retention Test; DSST, Digit Symbol Substitution Test; IQR, interquartile range; IST15, Isaacs Set Test shortened at 15 seconds; MMSE, Mini-Mental State Examination


   INTRODUCTION
TOP
 ABSTRACT
 INTRODUCTION
MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX
 References
 
With the increasing number of people suffering from neurodegenerative diseases, especially Alzheimer's disease, investigating cognitive changes over time has received growing attention in population-based cohort studies for understanding the natural history of the neurodegenerative diseases and in intervention trials designed to assess the effects of drugs on neuropsychological functioning (1, 2). In these studies, cognition is generally assessed through a battery of psychometric tests repeatedly administered to the subjects. Collecting several cognitive tests may be useful, because this allows exploration of the various cognitive domains (memory functioning, attention, or executive functions) and because the tests often have different metrologic properties. In particular, some tests are very sensitive to small changes at high levels of cognition, while others are more sensitive to changes at the lower levels. As a consequence, when studying the effects of drugs or the aging process on cognition, results may differ considerably according to the properties of the test being used (3). A way to deal with this problem is to study the change over time of the common latent cognitive level underlying the battery of cognitive tests used (4).

However, administering an extensive battery of neuropsychological tests can turn out to be difficult, because of the duration (and thus cost) of the evaluation and also because subjects presenting cognitive impairment are more prone to refuse long testing sessions than cognitively intact elderly subjects (5). In this way, it would be of substantial interest to compare the properties of commonly used cognitive tests to highlight arguments for selecting a restricted number of tests and even only one according to the aim of the study.

Brevity and ease of use are obviously relevant criteria by which to select a test for population-based studies. Nonetheless, it is more crucial to select a test able to detect small changes in cognition in all the range of cognitive levels observed in the target population (1, 2), particularly in cases of prolonged follow-up or long-term prevention studies. Indeed, the range of general cognitive level targeted will be different according to whether the study sample consists of subjects from population-based cohorts, patients consulting memory clinics for memory troubles, or patients with Alzheimer's disease enrolled in pharmacologic trials. Until now, when a single psychometric test is collected, the Mini-Mental State Examination (MMSE) (6) is usually preferred, because it gives a brief measure of global cognitive functioning (7). However, as the MMSE suffers from a strong ceiling effect, it is not suitable to identify slight declines in high levels of cognition (7, 8) and, thus, is not appropriate to study normal cognitive aging in prospective studies, particularly among highly educated people.

The aim of this work is to compare the sensitivity to cognitive change of four tests widely used in clinical practice: the MMSE, the Isaacs Set Test, the Benton Visual Retention Test, and the Digit Symbol Substitution Test. More specifically, we would like to identify the most appropriate test to measure cognitive changes over time in heterogeneous populations, including both highly normal and severely impaired subjects, as encountered in population-based studies. To answer this purpose, we use a nonlinear latent variable model for longitudinal multivariate data in which psychometric tests are assumed to be nonlinear parameterized transformations of a common factor (4). The common factor is a latent process representing the latent cognitive process underlying the psychometric tests. It is related to age through a linear mixed model for describing the latent cognitive evolution according to age. By estimating flexible transformations between the psychometric tests and the common factor, we are able to compare the metrologic properties of the psychometric tests.

This approach may be viewed as an extension of Item Response Theory (9) to handle repeated measurements of quantitative outcomes (the summary scores of each test) instead of binary outcomes (the individual binary items of one test). The link functions estimate the mean score of each test given the values of the latent process similarly to the Item Characteristic Curve in Item Response Theory models that estimates the probability of correct response given the latent ability. In addition, we are able to estimate evolution with time of the latent process and the test scores.


   MATERIALS AND METHODS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
RESULTS
 DISCUSSION
 APPENDIX
 References
 
Population
PAQUID is a prospective cohort study initiated in 1988 in southwestern France (Dordogne and Gironde) to explore functional and cerebral aging. In brief, 3,777 subjects who were 65 years or older and were living at home at enrollment were included in the cohort and were followed up for six times with a visit at 1 year, 3 years, 5 years, 8 years, 10 years, and 13 years after the initial visit, except at 1 year in Dordogne. At each visit, a neuropsychological evaluation and a diagnosis of dementia were carried out at home. Letenneur et al. (10) offer a detailed description of the PAQUID program.

Neuropsychological evaluation
In PAQUID, a battery of psychometric tests was used to evaluate cognition. In this paper, we focus on the following four tests.

  • The Mini-Mental State Examination (6) is a summed score evaluating various dimensions of cognition (memory, calculation, orientation in space and time, language, and word recognition). It is used as an index of global cognitive performance and ranges from 0 to 30.
  • The Isaacs Set Test shortened at 15 seconds (11) evaluates verbal fluency abilities and speed of verbal production. Subjects have to give a list of words (with a maximum of 10) belonging to a specific semantic category in 15 seconds. Four semantic categories were successively used (cities, fruits, animals, and colors). The score ranges from 0 to 40.
  • The Benton Visual Retention Test (12) evaluates immediate visual memory. After the presentation for 10 seconds of a stimulus card displaying a geometric figure, subjects are asked to choose the initial figure among four possibilities; 15 figures are successively presented. The score ranges from 0 to 15.
  • The Digit Symbol Substitution Test (13) explores attention and psychomotor speed. Given a code table displaying the correspondence between pairs of digits (from 1 to 9) and symbols, the subjects have to fill in blank squares with the symbol that is paired to the digit displayed above the square. The subjects have to fill in as many squares as possible in 90 seconds. In PAQUID, the score ranges from 0 to 76 even if the theoretical maximum is 90.

Statistical model
The statistical model assumes that the correlation between the tests is induced by a latent common cognitive factor. Thus, each test is a noisy measure of a test-specific nonlinear transformation of the common factor. The evolution of the common factor was modeled using a linear mixed model (14) that aims at evaluating changes over time of a repeated outcome (here, the latent common factor), accounting for correlation between the repeated measures on each subject. The linear mixed model included random intercept, age, and age squared in accordance with other longitudinal aging studies (15, 16) that showed quadratic cognitive evolutions. We added a Brownian motion to account for individual deviations from this quadratic evolution and thus relaxed the parametric form of the model. Test-specific random intercepts were added to evaluate if, for a same level of the latent common factor, two subjects scored differently at the tests.

We used flexible nonlinear transformations to link each psychometric test with the latent common factor. The chosen flexible transformations were beta cumulative distribution functions that offer a large variety of shapes (concave, convex, or sigmoid) using only two estimated parameters per test. The estimated beta cumulative distribution functions represent the shapes of the transformations between the latent common factor and each of the four tests and provide interesting metrologic properties for each test. The complete methodology was previously detailed in the report by Proust et al. (4) and is described in the Appendix.

Sample
Measurements at the initial visit were excluded from the analysis because of a first passing effect previously described (17). Indeed a test-retest improvement was observed between the initial visit and the visit at 1 year, possibly due to the stress induced by the first evaluation at the initial visit or a practice effect observed after the first evaluation. Of the 3,777 subjects in PAQUID, we retained 3,043 subjects who were followed up after the initial visit, and we studied the evolution of the common factor between the visits at 1 and 13 years. The Digit Symbol Substitution Test (DSST) was not completed at the visit at 3 years. Moreover, a subsample of PAQUID completed a nutritional questionnaire at the visit at 3 years that could impact on the Isaacs Set Test shortened at 15 seconds (IST15) score for the fruit and animal categories. So, we excluded the measurements of IST15 at the visit at 3 years. From the sample of 3,043 subjects, we retained a sample of 2,252 subjects with at least one measure at each neuropsychological test between visits at 1 and 13 years to ensure that every subject brought information about each neuropsychological test. The median number of measures was four for the MMSE (interquartile range (IQR): 2–5) and the Benton Visual Retention Test (BVRT) (IQR: 2–5) and three and two for the IST15 (IQR: 1–4) and the DSST (IQR: 1–4), respectively.


   RESULTS
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
DISCUSSION
 APPENDIX
 References
 
Description of the sample
Characteristics of the subjects in the sample are described in table 1: 56.7 percent of the subjects were women, and 73 percent had at least graduated from primary school. The mean age at the beginning of the follow-up (visit at 1 year) was 75.16 (standard error: 6.24) years. Among the 2,252 subjects, 14 had a positive diagnosis of dementia at the visit at 1 year, and 368 subjects were diagnosed as demented during the follow-up. At the 1-year visit, 1,623 subjects completed the MMSE with a median score of 28 (IQR: 26–29), 1,622 subjects completed the IST15 with a median score of 28 (IQR: 24–33), 1,619 subjects completed the BVRT with a median score of 11 (IQR: 9–13), and 1,614 completed the DSST with a median score of 28 (IQR: 20–37) (table 2).


View this table:
[in this window]
[in a new window]

  		TABLE 1. Demographic and health characteristics in the sample (n = 2,252), the PAQUID Study, France, 1989–2001

 

View this table:
[in this window]
[in a new window]

  		TABLE 2. Description of the cognitive performances for the four psychometric tests at the visit at 1 year, with number of measures, range, median with interquartile range, and mean with standard error, the PAQUID Sudy, France, 1989–2001

 
Cognitive evolution with age
The latent common cognitive factor ranges from 0 to 1, with 0 corresponding to a minimal cognitive performance and 1 corresponding to a maximal cognitive performance. The predicted mean evolution of the latent common cognitive factor and its 95 percent confidence interval are displayed in figure 1, part A. It shows a decline of cognitive performances in the population according to age, with an acceleration of the decline at older ages. This evolution can be interpreted as the mean cognitive decline with age in a population of subjects aged 65 years or older, including both demented and nondemented subjects.


Figure 1
View larger version (9K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 1. Predicted mean evolution (solid line) with 95% confidence interval (dashed lines) for the common factor (A) and for the four tests (Mini-Mental State Examination (MMSE), Benton Visual Retention Test (BVRT), Digital Symbol Substitution Test (DSST), and Isaacs Set Test shortened at 15 seconds (IST15)) (B), the PAQUID Study, France, 1989–2001.

 
The predicted mean evolutions of the four tests and their 95 percent confidence intervals are displayed in figure 1, part B. The decline of the DSST is quite linear, whereas the decline observed in the other tests is nonlinear, with an acceleration at older ages.

Comparison of the sensitivity of the four tests
The estimated transformation shapes between each test and the common factor are displayed in figure 2. They are very different for the four tests: The transformation is concave for the MMSE and the BVRT, convex for the DSST, and close to linear for the IST15. Moreover, the MMSE and the BVRT cover, respectively, only 78 percent and 85 percent of the common factor range, while the DSST and the IST15 cover almost all of the entire range with, respectively, 94 percent and 95 percent of the common factor range.


Figure 2
View larger version (9K):
[in this window]
[in a new window]
[Download PowerPoint slide]
  		FIGURE 2. Estimated link function (solid line) between the common factor and each of the four tests (Mini-Mental State Examination (MMSE), Benton Visual Retention Test (BVRT), Digital Symbol Substitution Test (DSST), and Isaacs Set Test shortened at 15 seconds (IST15)) with the 95% pointwise confidence interval (dashed lines), the PAQUID Study, France, 1989–2001.

 
Ceiling effect.
For the MMSE and the BVRT, the maximum value of the test is reached for a value of the common factor lower than its maximum (respectively, 0.78 and 0.87), which shows that these tests are not appropriate to assess cognitive changes in subjects with a very high general cognitive level of functioning. This underlines the ceiling effect of these two tests. In contrast, the maximum values of the IST15 and the DSST are reached for a value of the common factor close to 1, underlining that these tests do not suffer from a ceiling effect and can identify changes at the highest levels of cognition.

The MMSE transformation.
The nonlinear shape of the MMSE transformation shows that a decline of the MMSE does not correspond to the same intensity of decline of the common factor in all the range of the test. Indeed, a loss of 0.1 in the common factor scale between 0.7 and 0.6 represents a loss of 1 point for the MMSE, whereas the same loss between 0.3 and 0.2 represents a loss of more than 4 points for the MMSE. This shows that a decline in the MMSE cannot be interpreted without accounting for the initial value and that the MMSE is more appropriate to detect a decline at low cognitive levels than at high cognitive levels.

The BVRT transformation.
The shape of the BVRT transformation is quite linear at the low cognitive levels (corresponding to a value under 10 at the BVRT or under 0.5 on the common factor scale). It reveals that, under a value of 0.5 for the common factor, a decline in the BVRT represents the same intensity of decline for the common factor whatever the initial cognitive level. However, at high levels of cognition, the slope of the curve decreases. Thus, at the higher levels of cognition, the BVRT is less sensitive to small cognitive changes.

The DSST transformation.
The shape of the DSST transformation is quite linear at the high levels of cognition, corresponding to a score higher than 25 or a common factor level higher than 0.54. This means that, above this value, whatever the initial cognitive level, a decline of the DSST corresponds to the same decline in the latent cognitive level. In contrast, at the low levels of cognition, the shape is nonlinear, revealing that 1 point lost at the DSST corresponds to a greater decline of the common factor level under a score of 25. In this way, subjects with a very low level of general cognitive functioning (lower than 0.1) tend to score 0 on the DSST. This could be explained by the fact that subjects with a low level of cognition can probably not even understand the instructions of the DSST. Thus, DSST is appropriate to identify small changes in cognition among subjects with a high to medium general cognitive level but not among impaired subjects.

The IST15 transformation.
In contrast with the other tests, the IST15 transformation is close to linear for the entire range of cognition. This shows that the IST15 proves to be sensitive to change in all the ranges of cognition; 1 point lost at the IST15 score in higher values of the score has the same interpretation in terms of cognitive decline as 1 point lost in the lower values of the score. Thus, the IST15 can be useful for evaluating cognition in a heterogeneous population, including both subjects with a high general cognitive functioning and subjects with a severe cognitive impairment.


   DISCUSSION
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
APPENDIX
 References
 
By modeling jointly four psychometric tests and their latent common factor, we have been able to compare their distributions in all the range of cognition. In this way, we showed that MMSE and BVRT were not sensitive to cognitive changes at high levels of cognition and thus were not appropriate to study cognitive aging in prospective studies including highly educated people. On the contrary, we showed that the DSST was very sensitive to cognitive changes at high levels of cognition. However, as it was less sensitive to cognitive changes at low levels of cognition, it was not suitable for measuring cognitive changes in heterogeneous populations consisting of both highly normal and severely impaired subjects. In contrast, the IST15 appeared to be a satisfactory cognitive measure in all the range of cognition, which is of substantial interest when studying cognitive aging in population-based cohort studies.

The IST15 has several assets when compared with the three other tests. First, it does not suffer from a floor effect or a ceiling effect. Indeed, using cognitive measures with border effects can lead to misleading results (especially underestimated declines) when investigating cognitive changes, since initial scores are often differentially distributed among exposure groups and the sensitivity of the tests to identify cognitive changes is thus different among these groups (1). Second, the Isaacs Set Test when shortened at 15 seconds, as well as the DSST, includes a speed component that may explain its high sensitivity to changes at upper levels of cognition. Indeed, the speed component plays a key role in cognitive aging, and it has been shown, for example, that most age-related differences in cognition were due to the decrease in the processing speed (18). Finally, the IST15 is a very brief test, and its instructions are easily understandable. It can therefore be performed in large population-based studies even with severely impaired subjects.

The methodology we proposed in this paper has several advantages that should be discussed. First, the estimated link functions between the test scores and the latent process make it possible to compare properties of the tests and, especially, their sensitivity to detect cognitive changes within the entire range of cognition. This is done by modeling jointly various psychometric tests for which the hypothesis of a common factor is sensible. By the way, it is worth noting that the latent common factor in this model is actually defined according to the pool of psychometric tests used in the analysis. Computing the model with other tests involving different cognitive components could have an impact on the common factor evolution. In this analysis, we used tests that both are frequently used and explore different domains of cognition, because we wanted to select one test for exploring general cognitive decline in heterogeneous populations. The methodology could also be used for selecting sensitive measures in a specific domain of cognition. In this case, based on his/her knowledge or on other analyses such as principal component analyses, the researcher must choose the tests that are assumed to measure the same latent cognitive ability in this specific domain and then apply the methodology to the selected tests.

A second asset of the methodology is that, thanks to the estimated transformations of tests, the tests are no longer constrained to follow a Gaussian distribution as in a standard linear mixed model. In this way, even if longitudinal evolutions of the four tests, as presented in figure 1, part B, could have actually been estimated using linear mixed models, they would have been obtained under the wrong Gaussian assumption.

Finally, as parameters are estimated using the maximum likelihood estimators, results are robust to data missing at random (i.e., when the probability that data are missing does not depend on unobserved values given the past observed values). Simpler analyses that aim at comparing empirical means of the tests for different age groups are often biased by the missing data process, especially when the cognitive level and the dropout are linked, as was previously shown in the PAQUID cohort (17). In this previous work, it was also shown that the missing at random assumption was probably not strictly true, but the impact on the estimated evolution was slight (17, 19). Moreover, even if missing data may blur the comparison of evolution of the tests' scores, it is very unlikely that they biased the comparison of test sensitivity, which is the main objective of this paper. This was checked by comparing transformations estimated on four subsamples defined by the time of dropout (dropout after the visits at 3, 5, 8, and 10 years or complete follow-up) in the spirit of pattern mixture analysis (20). Whatever the pattern of dropout, the estimated transformations were very similar (results not shown).

Some methodological issues of this analysis should, however, be discussed. First, as the results rely on a parametric model, adequate fit of the model to the data has been carefully checked using postfit methods based on the residuals and the predictions developed in Proust et al. (4) (results not shown). An essential part of the model is the link function between the tests and the common factor. The beta cumulative distribution function was chosen because this transformation was flexible enough to exhibit very different shapes and depended on only two parameters per test. However, complementary analyses have been performed estimating the link functions on a basis of splines instead of the beta cumulative distribution functions; they have led to very similar results while raising more numerical problems due to the large number of parameters.

Second, in the PAQUID Study, MMSE was the first test fulfilled during each testing session. Consequently, it was more frequently completed than the three other tests, particularly among impaired subjects. To ensure that test-specific parameters were estimated on the same sample and to maintain comparability between the tests, we required that every subject had at least one measure at each test. The 791 subjects excluded from the sample were older (median age: 78.6 vs. 73.1 years) and less educated (51.5 percent did not graduate from primary school vs. 27 percent in the sample) than the subjects included in the selected sample, but the range of the observed scores was the same. Note also that using longitudinal data and keeping incident cases of dementia in the sample increased the observed range of cognition and allowed us to compare evolution of each test over time.

In conclusion, our results show that the Isaacs Set Test shortened at 15 seconds could be a good candidate to measure cognitive changes in a general population. More generally, the methodology used in this study provides some clues to thoughtfully select the appropriate measures of cognition collected in a study according to the nature of the target population and the objective of the study.


   APPENDIX
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX
References
 
Model specification
We consider K neuropsychological tests. For each test k, k = 1, ..., K, each subject i, i = 1, ..., N, and each occasion j, j = 1, ..., nik, the measure of the neuropsychological test yijk is collected at time tijk, tijk being different for each test and each subject. The latent process that represents the common factor of the K neuropsychological tests is modeled by use of the following linear mixed model, including a quadratic function of time and a Brownian motion (wi(t))t≥0 with variance term {sigma}Formula x t:

Formula
The vector of random effects ui = (u0i,u1i,u2i)T follows a multivariate normal distribution with mean vector 0 and variance covariance matrix D. The mean evolution of the common factor is represented by the fixed effects µ0, µ1, and µ2.

The observed score value yijk is linked to the value of the common factor at the time of measurement {Lambda}i(tijk) through a nonlinear link function hk that is a beta cumulative distribution function depending on two test-specific parameters {eta}k = ({eta}1k, {eta}2k). This leads to the following measurement model:

Formula
where the test-specific random intercept {alpha}ik follows a Gaussian distribution with mean 0 and variance {sigma}Formula. It takes into account the residual individual variability between tests after adjustment on the latent common factor, that is, the fact that two subjects with the same latent cognition can score differently at the psychometric tests. At last, {varepsilon}ijk are independent Gaussian errors with mean 0 and variance {sigma}Formula.


   ACKNOWLEDGMENTS
 
This study was funded by Novartis AG, SCOR insurance Agrica, Conseil Général de la Gironde, and Conseil général de la Dordogne.

Conflict of interest: none declared.


   References
TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 APPENDIX
 References
 

  1. Morris MC, Evans DA, Hebert LE, et al. (1999) Methodological issues in the study of cognitive decline. Am J Epidemiol 149:789–93.[Abstract/Free Full Text]
  2. Yesavage JA and Brooks JO 3rd. (1991) On the importance of longitudinal research in Alzheimer's disease. J Am Geriatr Soc 39:942–4.[Web of Science][Medline]
  3. Galasko DR, Gould RL, Abramson IS, et al. (2000) Measuring cognitive change in a cohort of patients with Alzheimer's disease. Stat Med 19:1421–32.[CrossRef][Web of Science][Medline]
  4. Proust C, Jacqmin-Gadda H, Taylor JM, et al. A nonlinear model with latent process for cognitive evolution using multivariate longitudinal data. Biometrics (Advance Access: doi: 10.1111/j.1541-0420. 2006. 00573.x.).
  5. Dartigues JF, Commenges D, Letenneur D, et al. (1997) Cognitive predictors of dementia in elderly community residents. Neuroepidemiology 16:29–39.[Medline]
  6. Folstein MF, Folstein SE, McHugh PR. (1975) "Mini-Mental State." A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res 12:189–98.[CrossRef][Web of Science][Medline]
  7. Tombaugh TN and McIntyre NJ. (1992) The Mini-Mental State Examination: a comprehensive review. J Am Geriatr Soc 40:922–35.[Web of Science][Medline]
  8. Commenges D, Gagnon M, Letenneur L, et al. (1992) Statistical description of the Mini-Mental State Examination for French elderly community residents. Paquid Study Group. J Nerv Ment Dis 180:28–32.[CrossRef][Web of Science][Medline]
  9. Baker F. (2001) The basics of Item Response Theory. (ERIC Clearinghouse on Assessment and Evaluation, College Park, MD).
  10. Letenneur L, Commenges D, Dartigues JF, et al. (1994) Incidence of dementia and Alzheimer's disease in elderly community residents of south-western France. Int J Epidemiol 23:1256–61.[Abstract/Free Full Text]
  11. Isaacs B and Kennie AT. (1973) The Set test as an aid to the detection of dementia in old people. Br J Psychiatry 123:467–70.[Abstract/Free Full Text]
  12. Benton A. (1965) Manuel pour l'application du Test de Rétention Visuelle. Applications cliniques et expérimentales. (In French). (Centre de Psychologie appliquée, Paris, France).
  13. Wechsler D. (1981) WAIS-R manual. (Psychological Corporation, New York, NY).
  14. Laird NM and Ware JH. (1982) Random-effects models for longitudinal data. Biometrics 38:963–74.[CrossRef][Web of Science][Medline]
  15. Hall CB, Lipton RB, Sliwinski M, et al. (2000) A change point model for estimating the onset of cognitive decline in preclinical Alzheimer's disease. Stat Med 19:1555–66.[CrossRef][Web of Science][Medline]
  16. Amieva H, Jacqmin-Gadda H, Orgogozo JM, et al. (2005) The 9 year cognitive decline before dementia of the Alzheimer type: a prospective population-based study. Brain 128:1093–101.[Abstract/Free Full Text]
  17. Jacqmin-Gadda H, Fabrigoule C, Commenges D, et al. (1997) A 5-year longitudinal study of the Mini-Mental State Examination in normal aging. Am J Epidemiol 145:498–506.[Abstract/Free Full Text]
  18. Salthouse TA. (1996) The processing-speed theory of adult age differences in cognition. Psychol Rev 103:403–28.[CrossRef][Web of Science][Medline]
  19. Jacqmin-Gadda H, Commenges D, Dartigues J. (1999) Analysis of longitudinal Gaussian data with missing data on the response variable. (In French). Rev Epidemiol Sante Publique 47:525–34.[Web of Science][Medline]
  20. Little R. (1993) Pattern-mixture models for multivariate incomplete data. J Am Stat Assoc 88:125–34.[CrossRef][Web of Science]

Add to CiteULike CiteULike   Add to Connotea Connotea   Add to Del.icio.us Del.icio.us    What's this?


This article has been cited by other articles:


Home page
 J Gerontol A Biol Sci Med SciHome page
A. Soumare, B. Tavernier, A. Alperovitch, C. Tzourio, and A. Elbaz
A Cross-Sectional and Longitudinal Study of the Relationship Between Walking Speed and Cognitive Function in Community-Dwelling Elderly People
J Gerontol A Biol Sci Med Sci, October 1, 2009; 64A(10): 1058 - 1065.
[Abstract] [Full Text] [PDF]

This Article
Right arrow Abstract Freely available
Right arrow FREE Full Text (PDF) Freely available
Right arrow All Versions of this Article:
165/3/344    most recent
kwk017v1
Right arrow Alert me when this article is cited
Right arrow Alert me if a correction is posted
Services
Right arrow Email this article to a friend
Right arrow Similar articles in this journal
Right arrow Similar articles in ISI Web of Science
Right arrow Similar articles in PubMed
Right arrow Alert me to new issues of the journal
Right arrow Add to My Personal Archive
Right arrow Download to citation manager
Right arrow Search for citing articles in:
ISI Web of Science (4)
Right arrowRequest Permissions
Right arrow Disclaimer
Google Scholar
Right arrow Articles by Proust-Lima, C.
Right arrow Articles by Jacqmin-Gadda, H.
Right arrow Search for Related Content
PubMed
Right arrow PubMed Citation
Right arrow Articles by Proust-Lima, C.
Right arrow Articles by Jacqmin-Gadda, H.
Social Bookmarking
Add to CiteULike   Add to Connotea   Add to Del.icio.us  
What's this?