Melodem (Methods in longitudinal dementia research) is an international initiative aiming at harmonizing analytic approaches across longitudinal studies of dementia risk or cognitive decline.

Several methodological challenges arise in studies of the determinants of dementia risk and cognitive decline. Some challenges, such as unmeasured confounding or missing data, are common in many research areas; others, such as outcome measurement error and lack of a “gold standard” outcome assessment, are more pervasive or more severe in dementia research.

Currently, different researchers handle these challenges differently, making it difficult to directly compare studies and combine evidence. Although some methodological differences across studies arise because analytic methods are explicitly tailored to the study design and realities of the data at hand, other differences arise for less substantive reasons. Modifiable sources of inconsistency include the absence of consensus and definitive standards for best analytic approaches; divergent disciplinary traditions in epidemiology, clinical research, biostatistics, neuropsychology, psychiatry, geriatrics and neurology; and software and technical barriers.

Melodem was launched in 2012, under the coordination of Carole Dufouil (INSERM, University of Bordeaux, France) and Maria Glymour (UCSF, USA) with the support of the Fondation Plan Alzheimer. Two meetings have been organized in Paris, in 2013 and 2014, to discuss progress and future directions, with several ancillary meetings in the intervening period.

Melodem was originally organized around five working groups that represent major challenges:

  • (1) Selection, i.e., handling selection stemming from study participation, attrition, and mortality;
  • (2) Measurement, i.e., dealing with the quality of measurements of exposure and outcomes and how imperfect measurement quality affects analysis and interpretation of results;
  • (3) Alternative time scale, i.e., specifying the time-scale and the shape of trajectories in longitudinal models
  • (4) Time-varying exposures and confounding, i.e., accounting for changes in explanatory variables;
  • (5) High-dimensional data, i.e., analyzing complex and multi-dimensional data such as neuroimaging, genomic information, or database linkages.

Each working group has one to two key coordinators who organize regular phone/web meeting to discuss specific challenges within the topic area and develop manuscripts that detail these challenges and potential methods for addressing them.


For each working group and challenges the ultimate goals are:

  • To compare methods conceptually and empirically (across data sets)
  • To reach consensus on preferred methods, considering current and future data availability
  • To identify barriers to adoption (e.g., software availability) and facilitate dissemination

Additional working groups such as analyses of secular trends and biomarkers validation are planned to launch in 2015.

The Melodem website will provide updates on the working groups’ progress and publications as well as make available statistical programs/methods/guidelines whenever possible.

As of January 2015, Melodem counts more than a hundred researchers as members. Melodem activities are supervised and guided by its international scientific advisory board (David Bennett (Rush University, Chicago, USA) Carol Brayne (Cambridge University, UK), Monique Breteler (DZNE, Bonn, Germany), Geneviève Chêne (Institute (ITMO) of Public Health, Paris, France), Charli De Carli (University of California, Davis, USA), Ronald Petersen (Rochester University, USA)