Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
Randomized Controlled Trial
. 2019 Jul 25;19(1):163.
doi: 10.1186/s12874-019-0791-z.

Using joint models to disentangle intervention effect types and baseline confounding: an application within an intervention study in prodromal Alzheimer's disease with Fortasyn Connect

Floor M van Oudenhoven  1   2 , Sophie H N Swinkels  2 , Tobias Hartmann  3   4 , Hilkka Soininen  5   6 , Anneke M J van Hees  2 , Dimitris Rizopoulos  1
Affiliations
Randomized Controlled Trial

Using joint models to disentangle intervention effect types and baseline confounding: an application within an intervention study in prodromal Alzheimer's disease with Fortasyn Connect

Floor M van Oudenhoven et al. BMC Med Res Methodol. .

Abstract

Background: Many prodromal Alzheimer's disease trials collect two types of data: the time until clinical diagnosis of dementia and longitudinal patient information. These data are often analysed separately, although they are strongly associated. By combining the longitudinal and survival data into a single statistical model, joint models can account for the dependencies between the two types of data.

Methods: We illustrate the major steps in a joint modelling approach, motivated by data from a prodromal Alzheimer's disease study: the LipiDiDiet trial.

Results: By using joint models we are able to disentangle baseline confounding from the intervention effect and moreover, to investigate the association between longitudinal patient information and the time until clinical dementia diagnosis.

Conclusions: Joint models provide a valuable tool in the statistical analysis of clinical studies with longitudinal and survival data, such as in prodromal Alzheimer's disease trials, and have several added values compared to separate analyses.

Keywords: Alzheimer’s disease; Baseline imbalance; Fortasyn; Intervention effect; Joint model.

PubMed Disclaimer

Conflict of interest statement

FO, SS and AH are employees of Danone Nutricia Research. HS and TH were supported by a grant from the European Commission for the LipiDiDiet study (FP7-211696 LipiDiDiet). HS has served as advisory board member for ACImmune and MERCK. Institution, UEF, has received funding from Nutricia for extension studies of LipiDiDiet Trial (no personal payment). Author DR declares to have no competing interests.

Figures

Fig. 1
Fig. 1
Histograms of Mini-Mental State Examination at baseline for the test and control group. The test group contains more values at the lower end of the histogram
Fig. 2
Fig. 2
Observed longitudinal profiles for CDR-SB for six randomly selected patients. A higher CDR-SB score indicates a worsening of a patient’s status
Fig. 3
Fig. 3
Observed longitudinal profiles for NTB memory domain for six randomly selected patients. A lower NTB memory domain score indicates a worsening of a patient’s status
Fig. 4
Fig. 4
Schematic representation of a joint model. β2 and β3 denote the constant respectively time-varying indirect intervention effect on the longitudinal marker, α is the effect of the longitudinal marker on the survival outcome and γ1 is the direct effect on the survival outcome
Fig. 5
Fig. 5
Graphical representation of different ways of modelling the association between the longitudinal and survival process. The different graphs respectively denote the current value (a), the current value plus the rate of change (b) and the cumulative effect (i.e., the AUC) of the longitudinal trajectory (c)
Fig. 6
Fig. 6
Separate effects as estimated by the joint model for CDR-SB. The separate components exp(γ1) (direct effect; dashed line), exp(α×β2) (indirect constant effect; solid line) and exp(α×β3×t) (indirect time-varying effect; dot dashed line), that together form the hazard ratio for the total intervention effect as estimated from the joint model for CDR-SB, plotted as separate effects in a without and in b with correction for baseline MMSE in the survival sub-model
Fig. 7
Fig. 7
Total intervention effect as estimated by the joint model for CDR-SB. The total intervention effect on the hazard of dementia diagnosis as estimated from the joint model (solid line) and Cox model (dashed line) for CDR-SB, in a without and in b with correction for baseline MMSE in the survival sub-model. Corresponding 95% percentile confidence bands (light grey corresponding to the joint model and dark grey corresponding to the Cox model) were based on 2500 bootstrap samples
See this image and copyright information in PMC

References

    1. Aisen PS, Cummings J, Jack CR, Morris JC, Sperling R, Frölich L, Jones RW, Dowsett SA, Matthews BR, Raskin J, et al. On the path to 2025: understanding the Alzheimer’s disease continuum. Alzheimer’s Res Ther. 2017;9(1):60. doi: 10.1186/s13195-017-0283-5. - DOI - PMC - PubMed
    1. Petersen RC. Mild cognitive impairment as a diagnostic entity. J Intern Med. 2004;256(3):183–94. doi: 10.1111/j.1365-2796.2004.01388.x. - DOI - PubMed
    1. Vaughan RM, Coen RF, Kenny R, Lawlor BA. Semantic and phonemic verbal fluency discrepancy in mild cognitive impairment: Potential predictor of progression to alzheimer’s disease. J Am Geriatr Soc. 2018;66(4):755–9. doi: 10.1111/jgs.15294. - DOI - PubMed
    1. Munro CE, Donovan NJ, Amariglio RE, Papp KV, Marshall GA, Rentz DM, Pascual-Leone A, Sperling RA, Locascio JJ, Vannini P. The impact of awareness of and concern about memory performance on the prediction of progression from mild cognitive impairment to alzheimer disease dementia. Am J Geriatr Psychiatr. 2018;26(8):896–904. doi: 10.1016/j.jagp.2018.04.008. - DOI - PMC - PubMed
    1. Baldeiras I, Santana I, Leitão MJ, Gens H, Pascoal R, Tábuas-Pereira M, Beato-Coelho J, Duro D, Almeida MR, Oliveira CR. Addition of the A β42/40 ratio to the cerebrospinal fluid biomarker profile increases the predictive value for underlying alzheimer’s disease dementia in mild cognitive impairment. Alzheimer’s Res Ther. 2018;10(1):33. doi: 10.1186/s13195-018-0362-2. - DOI - PMC - PubMed

Publication types