THE AIM OF THE PROJECT
The overall aim of the project was to investigate protective factors for cognitive aging. The literature has so far strongly focused on risk factors whereas factors which may be protective in cognitive aging have been less investigated. The project is based on a longitudinal study, within the Swedish Twin Registry; including participants aged 80 years and older, covering five measurement points across 8 years. The longitudinal data was linked to information collected 25 years earlier, from the time when the participants were in their midlife. Further aims of the project were to explore differences between men and women regarding protective factors, to explore the genes/heredity and environmental aspects of the protective factors, and to test an alternative time-scale to structure the longitudinal data. Given the fact that chronological age becomes less informative in very old age there have been attempts to use time to death instead of time from birth (chronological age) as time structure (all participants are deceased). Although the goal of the project was to shift focus from risk factors to protective factors, risk factors have also been studied.

THE THREE MOST IMPORTANT RESULTS OF THE PROJECT
1. Protective factors for cognitive health should be examined already in midlife: Given the fact that the cognitive symptoms that precede Alzheimer's disease are measurable up to 20 years before a clinical diagnosis is made, it is crucial that the potential protective/risk factors are examined at least 20 years before the outcome, to avoid confounding of cause and effect. Despite this, studies with relatively short follow-up dominate the research field. One of the project's most important results comes from the study where psychosocial conditions and health status in midlife were examined in relation to cognitive aging. Apart from the genetic marker APOE Ɛ4, which proved to be the main explanatory factor for cognitive aging, it was found that the strongest contribution to cognitive stability in old age was education as well as social group position, two strongly correlated factors. Moreover, it was found that absence of family-conflicts in midlife, as well as normal BMI and good health (no cardio- or cerebrovascular diseases) in midlife were related to cognitive stability during aging. These findings were further examined among monozygotic twins, where one twin showed cognitive stability but not the other. The twin pairs were therefore genetically identical and did not differ in terms of BMI, health status, or level of education. What differed between the twin pairs was that the cognitively stable twin smoked less and had lower alcohol intake in midlife, and was involved to a greater extent in leisure activities in midlife that reflected intellectual, cultural and self-improving activities.
2. Light alcohol intake, often seen as being protective of cognitive health, is more likely to be a risk factor: There are many studies which show a positive association between light alcohol intake and cognitive health, and therefore many studies conclude that light alcohol consumption can protect against cognitive aging. This interpretation is likely to be incorrect because many of those who abstain completely from alcohol often do it for health reasons; the group of abstainers is therefore biased towards poorer health and thus not suitable as a control group. One of the important results from the project show that identical questions about alcohol intake, addressed to the same individuals with a 25 years interval (midlife, old age), strengthened this misleading conclusion such that in old age the seemingly beneficial effect of alcohol appeared to be considerably stronger than when the questions were asked in midlife. In an attempt to control for the abstainer-bias, the dose-response relationship between alcohol intake and cognitive performance was examined, including only alcohol consumers. Of note is that the participants were born around the turn of the latest century and were in midlife in the sixty's, which means that alcohol intake was very moderate and significantly lower than what is the norm today (the average intake being less than 1 unit per week). Moreover, there were no heavy drinkers in this sample. Despite very low alcohol consumption reported by the participants, clear linear dose-response correlation was found between alcohol consumption and cognitive test performance, especially the memory tests. Thus, higher alcohol consumption in midlife was clearly linked to worse memory performance measured 25 years later.
3. Differences between men and women: The project showed no differences in cognitive function or cognitive decline between the sexes. However, there were gender differences in the factors which are of importance for cognitive aging. For example, there were differences in types of leisure activities reported by men and women in midlife, and these activities were differentially related to cognitive aging. Women reported mainly activities related to domestic activities, and it was found that higher domestic activities were related to faster cognitive decline. Men, on the other hand, reported activities related to self-realization to a greater extent, which was associated with higher level of cognitive performance. All these relationships were corrected for differences in level of education and age. Another result, which shows the importance of taking notice of gender differences, showed that women are more likely than men to develop vascular dementia as a result of diabetes.

NEW RESEARCH QUESTIONS GENERATED FROM THE PROJECT
During the project it has become very clear that search for significant influencing factors of importance for cognitive aging must take a life course perspective, which means that examination of factors at earlier stages of life should be the main focus. An example of a new research interest, generated by the project, is to further examine how changes in alcohol consumption across the life course relate to cognitive aging.

INTERNATIONAL REPRESENTATION OF THE PROJECT
The international representation of the project is reflected by publications in international scientific journals and by regular participation at international conferences and meetings. For example, findings related to the project have been presented at the American Gerontological Society (GSA) annual meetings, at the Nordic Gerontological Society biannual meetings, and at the World Congress of Gerontology (IAGG). Another type of international representation is the international research collaborations. One example is the collaboration within the IALSA research network (Integrative Analysis of Longitudinal Studies of Aging and Dementia). Within this collaboration, the same hypothesis was tested in three similar longitudinal studies in the US, UK, and Sweden (Kelly et al., 2016). Another example is the collaboration between fourteen studies around the world, resulting in a meta-analysis including over 2 million participants (Chatterjee et al., 2016).

RESEARCH COMMUNICATION OUTSIDE THE SCIENTIFIC COMMUNITY
There have been several occasions at which research findings have been presented outside the scientific community. Examples of these include; open lectures for Folkuniversitetet in Gothenburg; open lecture for the general public at the annual International Science Festival in Gothenburg; and open lectures for SPF Seniorerna - The Swedish Association for Senior Citizens.

THE TWO MOST IMPORTANT PUBLICATIONS
One of the most important publications, I believe, will be the one showing adverse effect of alcohol on memory functions in aging (Hassing, under review). This finding, which shows that even very low alcohol consumption is negatively related to memory performance, is important because it contradicts the common interpretation of this relationship, that is that light alcohol consumption could have a protective effect on cognitive aging.
The other important publication is the one that is a result of collaboration between researchers working within 14 studies around the world (Chatterjee et al., 2016). By pooling a number of studies, a basis for advanced analysis was created, which led to interesting results, among others, how diabetes differently affects dementia risk in men and women. The study has received considerable attention and is already highly cited.

PUBLICATION STRATEGY
The publication strategy of the project is to publish in highly ranked international peer-reviewed journals, with high impact within the field, and as far as it goes with an open access alternative. The open access aspect has been ensured by paying for direct open access (Björk Praetorius et al., 2016), and by parallel publications via GUP (Gothenburg University Publications) as far as possible (Dahl et al., 2013; Kelly et al., 2016; Praetorius et al., 2013; Praetorius et al., 2014).