View Report Details

Cognitive function, emotional well-being and dietary factors in the Caerphilly cohort of older men

Project Code: N05039

31/08/2004

Department of Epidemiology, Statistics and Public Health, Cardiff University
Elwood, P ;
University of Bristol
Appleton, K;
Department of Epidemiology, Statistics and Public Health, Cardiff University
Pickering, J; Hughes, J; Fehily, A; Gallacher, J

The work described in this report is based on data collected over 25 years from a representative sample of older men in Caerphilly, South Wales(2,500 men aged 45-59 years at baseline). All the data had been collected prior to the application for the present grant. However a considerable amount of work had to be done extracting appropriate data, deriving indices and organising these for analyses. 

The aims set for work under the present grant were to examine associations between diet and 1) cognitive function, and 2) emotional well-being.   A summary of the main findings follows. 

Method of presentation of results in the tables.As an example, part of table C.1.1, the first of the main tables in the report, is reproduced below. The mean CRT (choice reaction time) in the fifth of men with the lowest dietary protein intake (224 men) is shown, namely 902msec SD 252msec. The differences in the mean levels of subsequent fifths of men are then shown – simply as differences. Thus, the fifths of men with increasing levels of protein intakes had reaction times that were 16msec shorter, 26msec shorter, 2msec longer and 5msec shorter. The significance of this sequence of figures was judged from the continuous original reaction times (that is, ungrouped data) giving a probability level of 0.664.

Column 2 shows the same CRT results for the various fifths of men, adjusted for confounding by the factors shown in a footnote (age, BMI, smoking, social class, education and energy).   Again, the significance of the trend in these adjusted differences has been assessed using the original data and not grouped data, and the probability level stated in this case as P < 0.405.

1. Unadjusted

2. Adjusted for age, BMI, smoking, social class, education, energy

Associations with having been breast fed.

This was undoubtedly the most important outcome of the whole project. In men whose birth weight had been below the median, breast feeding was associated with significantly higher test results in reasoning and in word power. Two SDs below the median birth weight, breast feeding was associated with an extra six points, or 70% of a SD. We judge this to be a most important result, indicating that substantial benefits of breast feeding are detectable 60 and more years later. As we admit in the full report, the men in our cohort who had not been breast fed will have received diluted, sweetened cow’s milk, whereas today’s formulae infant milks are highly sophisticated and contain various fatty acid and other additives. On the other hand, these new milks have never been rigorously tested in a long term follow-up study such as ours and so the message of our work still stands – namely, there is no adequate substitute for breast milk.

A paper on breast feeding and cognitive function has been accepted by the JECH and should be published very shortly. The paper acknowledges the support of FSA, and a copy was sent to FSA before submission.

C.1. Associations with the intakes of the major nutrients

Early analyses of fat intake suggested relationships with cognitive function and with mood, and in view of the concerns expressed by Muldoon and others about a possible increase in deaths from suicide, violence and accidents in the cholesterol lowering trials, we found these suggested relationships of considerable interest. Further analyses however (Table C1.2 C1.5) showed that adjustments for confounding removed all significance.

Data on alcohol intake show a beneficial association with cognitive function (Table C.1.7.). If further analyses confirm that this is real, it is surprising and requires detailed investigation.

C.2. Associations with fruit & vegetable intakes

The intakes of fruit and vegetables in Caerphilly is very low indeed – less than 5% of the men consume what is recommended by WHO and other bodies Table C.2.1.). Nevertheless, the stability of intakes of these foods over the ten years covered by the three semi-quantitative food frequency questionnaires, is high.

The associations between the fruit, vegetables and the combined fruit & vegetable consumption, and cognitive function are shown in table C.2.2The results give little evidence of any important associations.

Furthermore, what associations there are with the three cognitive tests, are fairly severely reduced by adjustment for the effects of social class etc. We were not encouraged by this.

The effects of fruit and vegetable consumption appear on the whole to be beneficial on emotional well being – that is, the men with the lowest intakes have negative scores, while the men with higher intakes have positive scores. The effects are however small, though the results do suggest that there might be a true association between fruit and vegetable intake and ‘contentment’. At the same time, the general absence of significance in the table give reason for caution.

C.3. Associations with blood antioxidants

We did extensive analyses of these data. The data are very extensive, comprising 17 variables in all. In order to handle these we conducted a principal components analysis and this yielded two ‘Factors’. In summary, our antioxidant ‘Factor 1’, which is truly dependent upon antioxidants, shows many significant relationships with cognitive function, while ‘Factor 2’, which is mainly dependent upon inflammatory indices, shows nothing suggestive. Detailed analyses of the individual antioxidants which contributed to ‘Factor 1’ confirmed significance.

Again, we did a large number of analyses of relationships with the indices of emotional well-being. None of these gave promise of anything of value

Clearly further work will be justified and we have had preliminary discussions with experts on antioxidant mechanisms.

C.4. Associations with having been breast fed.

This part of the work has already been mentioned above.

C.5. Associations with blood homocysteine.

No meaningful associations were seen between either homocysteine level (C.5.1. and 2)or dietary folate intake (C.5.3. and 4).

C.6. Associations with individual food items.

We recognise the dangers of ‘trawling’ through an almost endless list of food items and so we only include here data for those foods for which we had some expectation – however slight - of a meaningful relationship.

We examined relationships with tomatoes (Table C.6.1.), a major source of lycopene, a powerful antioxidant. There is no significant relationship after adjusting for confounding. The AH4 alone suggested a significant and beneficial relationship. We got no encouragement from apples (table C.6.2.) which, in earlier work, had shown significant relationships with lung function. Nor did the data on carrots or tea give any promise (tables C.6.3 and C.6.4.).

C.7. Associations with food patterns.

Through the work of Baker, Kromhout and others, we became interested in food ‘patterns’. Published work appears to have focussed on two approaches to food patterns: 1) associations with an ‘ideal’ dietary; and 2) relationships with whatever distinct patterns emerge from a statistical analysis of the diets recorded by a representative sample of subjects.

We did not feel competent to define an ‘ideal’ dietary pattern, but we did use this approach in relation to two definable diets. In one we grouped the men by the amount of dairy foods they had recorded in their dietary assessments. The other was to use the protein/carbohydrate ratio in each man's diet in order to derive an ‘Atkins’ kind of index. Neither of these approaches appeared to yield anything of value. (see tables C.7.1. C.7.2 and C.7.3.).

The main work we did however was using principal components analyses to derive patterns within the food items recorded by 665 men in the cohort who had recorded 7-day weighed intakes. A small number of statistically distinct patterns emerged. For convenience we termed these: Healthy (high fruit and vegetables and low fat consumption); Traditional (high meat, moderate vegetables); and Sandwich (high refined grains and high fat). 

Nothing of promise emerged (C.7.4 and C.7.5). Further analyses are however in progress with a colleague who has published on this approach and we are discussing analyses of the 2,500 questionnaire data in the same way.   In this we will examine both relationships with incident disease, and also with cognitive function and emotional wellbeing.

•  Main File
•  N05039 - Annex

Some of the files on this site may be in a format that your computer can't read. However, you can download Readers and Viewers for the following document types below:

• PDF - Download Acrobat Reader
• DOC - Download MS Word Viewer
• XLS - Download MS Excel Viewer
• PPT - Download MS Powerpoint Viewer