THE FRAMINGHAM STUDY An Epidemiological Investigation of Cardiovascular Disease Rditors: William B. Kennel, M.D. and Tavia Gordon Section 24: The Framingham diet study: Diet and the regulation of serum cholestercl.We Franingsam Study 190 Fee Section 2h 1906 pu Page Editorial Note Introduction ake 1 Measurenent of atet ake 3 Usual. inteke ake Relationship of dietary iron and blood hemoglobin aha12 Relationship of diet to CHD incidence 2-13 Conclusions 2k-15, ‘Tables Table 1. Exclusions from Framingham dietery sample because of preexisting disease. Table 2. Summary description of the Framingham dietary sample. Table 3. Mean and standard deviction of cone characteristics of the average daily intake of food, by sex: Fremingham Diet Study. Table lt. Mean of sone characteristics of the average daily intake of food by age for men: Framingham Diet stuay. Mean of sone characteristics of the average daily intake of food by age for women: Framingham Diet study. ; Table 5. Correlation of calories with some other character- istics of the everage daily intake of food by sex and age: Framinghan Diet Study. Table 6. Correlation of serum cholesterol with some char- acteristics of the average daily intake of food by age end sex: Framingham Diet Study. : Gable 7. Neen serum cholesterol level according to daily caloric intake: Framingham Diet Study. Table 8. Mean serum cholesterol level. according to daily caloric intake per pound of body weight: Framingham Diet Study. Table 9. Daily caloric intake by Physical Activity Index and Table 10. body weight: Framingham Diet Study. Mean cholesterol Level according to Physical Activity Index and daily caloric intake by sex and age: Framingham Diet Study.Framinghan Study Section alt ables (continued) Table 11. Regression of serum cholesterol level on Physical Activity Index (PAI), daily caloric intake, and Framingham relative weight (FRW): Framingham Diet Study. Mean serum cholesterol level according to daily fat intake: Framingham Diet Study. Mean serum cholesterol level according to daily animal fat intake: Framingham Diet Study. Mean serum cholesterol level according to percent fat from animal sources: Framingham Diet Study. Mean serum cholesterol level according to ratio of intake of plant to animal fat: Framingham Diet study. Mean serum cholesterol level according to ratio of complex to simple carbohydrates (C/S ratio) in food intake: Framingham Diet Study. Mean serum cholesterol level accor@ing to daily dietary cholesterol intake: Framingham Diet Study. Mean serum cholesterol level according to daily intake of animal fat and dietary cholesterol: Framingham Diet Study. Regression of serum cholesterol on daily intake of animal fat and dietary cholesterol: Framingham Diet Study. Mean of some characteristics of the average daily intake of food by serum cholesterol level: Framingham Diet Study. Regression of specified diet characteristics on serum cholesterol: Framingham Diet Study. Regression of mean weight and weight change on serum cholesterol: Framingham Study. Mean daily iron intake at specified levels of blood hemoglobin: Framingham Diet Study.‘Framingham Study | section 24 Gables (continued) Table 24. Mean values for specified nutrient measurements | for persons who developed CHD subsequent to their dict interviews, by sex: Framingham Diet Study. Figures Figure 1. Mean value of average daily caloric intake for each three-month period: Framingham Diet Study, men, January 1957 to September 1960. Figure 2. Mean value of average daily caloric intake for each three-month period: Framingham Diet Study, wonen, January 1957 to September 1960. Figure 3. Mean volue of average daily fat intake for each three-month period: Framingham Diet Study, men, January 1957 to September 1960. Figure 4, Mean value of average daily fat intake for each ‘three-month period: Framingham Diet Study, women, January 1957 to September 1960. Figure 5.° Mean value of body weight for each three-month period: Framingham Diet Study, men, January 1957 ‘to September 1960. Figure 6. Mean value of body eight for each three-month period: Framingham Diet Study, women, January 1957 to September 1960. Figure 7. Calories in daily intake: Framingham Diet Study. Figure 8. Celories in daily intake per pound per body weigh Franinghan Diet Study. Figure 9. Protein in daily intake: Framingham Diet Study. Figure 10, Animal protein in daily intake: Framingham Diet Study. Figure 11. Plant protein in daily intake: Framingham Diet Study.Framinghan Study ' Section 2h Figures (continued) Figure Figure Figure Figure Figure Figure Figure R. 13. abe b. 1. 1. 18. Fat in daily intake: Fremingham Diet Study. Fat in daily intake per pound body weight: Framingham Diet Study. Animal fat in daily intake: Framingham Diet Stuay. Plant fat in daily intake: Framingham Diet Stuay. Percent calories from fat: Framingham Diet Stuay. Percent fat from animal sources: Framingham Diet Stuay. Ratio polyunsaturated fatty acid to saturated fatty acid in daily intake: Framingham Diet tudy. Iron in daily intake: Framingham Diet Study. Cholesterol in daily intake: Framingham Diet Study.Framingham Study ; Editorial wote: This report is based on a selection from a large body of tabular material prepared since the completion of data collection for the Fra- mingham Diet Study. Unfortunately, these data were never incorporated into a definitive report by the original investigators and a large emount of very careful and thoughtful vork has lain unused in the Framingham files. I have organized and edited this material, and to a limited extent supplemented it, so that it can be made part of the record of the Framing- ham Study. While sone effort was made to organize this material in a uni- form manner certain minor discrepancies, for example in age groupings, remain. Weedless to say, I alone am responsible for the selection and organization of the material end the conclusions are mine and do not necessarily reflect the judgements of the original investigators. Such @ disinterment and recastruction leaves a greet deal to be desired, including a proper credit to the people who did the original work and the large number of persons who had input into the planning, collection and analysis of the data, Two essential contributions mst be mentioned, however. George V. Mann, M.D. had primary responsibility for planning and supervising the study and Georgiana Pearson was the dietitian responsible for the data collection. Tavia Gordonhel Fram Study ‘The Framingham Diet Study: Diet and the Regulation of Serum Cholesterol In’ 1957-1960' @iet assessment vas undertaken fora sample of the Framingham Study cohort on their fifth biennial examination.* The method used was a modification of the aietary interview developed by Burke (1) and was directed to evaluating the "usual" long-term intake of each person interviewed. A sample of 1049 persons was drawn according to an elaborate samp- ling scheme. A 25 percent random sample was drawn from # listing of the original sample respondents. This sample was then modified as follows: (a) One half of the women sampled were removed. (>) Al persons who were mown dead vere removed (25 men, 7 wonen). (c) All persons who had missed the last two consecutive examinations were removed (50 men, 25 women). (4) All persons with the diagnosis of heart disease or definite hyper- tension were removed (Table 1). (Persons who vere diagnosed as having these conditions subsequent to the sampling but prior to the interview were, however, retained). The diagnoses comprised any form of definite heart @isease, including myocardial infarction, angina pectoris, and rheumatic heart disease. A total of 173 persons, 124 men and 49 women, were removed for these reasons. However, these were the only medical exclusions and other persons with known @isease, some of them involving dietary therapy, are included in the dietary sample. * A few interviews vere obteined on the fourth and sixth examinations. 1 Burke, B.S. The dietary history as a tool in research. J. Amer. Diet. Assce, 23: 1041-1046, December 1947.Framingham Study 2he2 The wives of all the men selected for the dietary sample vere then added in order to have these women available to supplement their musbanats | stories. The husbands of women in the random sample were not includea: | Since the degree of lipidemia is involved in the dietary hypotheses of atherogenesis and since this attribute hed been measured in past examina tions it was used as a criterion for augmenting the dietary sample. For ‘this purpose, all male subjects in the otherwise qualified population with cholesterol levels* of 300 ma% or more or below 170 mg were included in the dietary sample as "high cholesterol" and "low cholesterol" males, respectively. These groups promised to be unusually potent in the eval- vation of dietary hypotheses. The wives of these men were also included. These methods yielded a total sample of 1049 subjects, 503 men and 546 women. A summary of the dictary sample is shown in Table 2. These per- sons were to be studied in a period of 36 months. Of the 1049 persons in the sample only 912 yielded successful inter- views. The other 137 were accounted for as follows: Not interviewed 6 Refused 33 Dead 5 Moved 23 Unsuccessful interview 6 Failed 10 No pattern an Dieting sry Poor communication 4 Uncooperative aL Other 2 Tt was not intended by this procedure to provide an estimate of the usual diets in the Framingham Study group as a whole but rather to collect information to test the hypotheses that "fhe reyulation of the level of * On the initiel study determination.<3 \ SET Framingham Study 2lke3 serum cholesterol end the development of coronary heart disease are related to 1. he caloric balance 2. Level of animal fat intake 3. Level of vegetable fat intake 4. Level of protein intake 5. Level of cholesterol intake... ‘This psper presents data relating to the hypotheses respecting the regulation of serum cholesterol. To these has been added the hypothesis ‘thet serum cholesterol level is related to 6. Level of the ratio of complex to simple carbohydrates (G/s ratio). ‘THE MEASUREMENT OF DIET ‘Two reports have appeared describing the techniques used and the reliability of the measurement of dietary intake (2,3), While the neasurement of nutrient intake is exceedingly difficult the general con- clusion of these methodological studies was that the procedure provided a relatively valid and reliable measure of usual daily intake. However, diet interviews were obtained over a l-year period and it may vell be asked whether the technique remained stable over that period of tine. Figures 1 and 2 indicate that the total caloric intake reported for the study group decreased measurably over the 4-year period. In particular, levels during the last year of interviewing were sub- stantially less than during the first three years. If one specific nutrient is considered--total grams of fat--a parallel trend is noted, as 2 Dawber, T.R., Pearson, G., Anderson, P., Mann, C.V., Kannel, W.B., Shurtleff, D., McNamara, P.M. Dietary assessment in the epidemiologic study of coronary heart disease: the Framingham Study. Amer. J. Clinical Nutrition 11: 226-234, September 1962. 3 Mann, G.V., Pearson, G., Gordon, T., Dawber, T.R., Lyell, L.P., Shurtleff, D. Diet and cardiovascular disease in the Framingham Study. Amer. J. Clinical Nutrition 11: 200-226, September 1962.Franinghan Study aha’ might be anticipated. That this is not easily explained by a change with time in the persons interviewed may be Judged fron Figures 5 and 6 i which show trends in body weight. While the average weight was lover i for men interviewed at the end of the study than men interviewed at the | i i hegining of the stuty (e finding tert might bo conftowell as coasitbtint : with trends for total caloric intake) the opposite was true for vonen i: interviewed. Figures 1 and 2 might also be examined for evidence of seasonal | variation in reported intake. The interview was intended to elicit i “usual dntake; that 1s, something nore general than the recent diet of i the individual. If this effort was successful there should be no evidence | of seasonality in reported intake. Despite substantial variation from quarter to quarter in total calories reported, there is no evidence that this variation followed a seasonal cycle. USUAL INPAKE While it is difficult to specify exactly how the diet for this study group compares with that for the Framingham cohort as a whole or how the | Framingham dict compares with that for United States adults generally, he findings in a carefully evaluated group there may be some interest of American adults (Tables 3 end 4). on the average, men consumed 3156 calories and women consumed 2142 calories. Some 40 percent of these calories came from fat and 70 percent | of the fat was from animal sources. (There was little difference between nen and voren in these proportions). Calorie intake decreased with age in both men and women but the percent of calories from fat was about the | same across the age range. A slightly larger proportion of the fat cameFramingham Study abe5 from animal sources in older persons. Distributions of various nutrient measures are given in Figures 7-20. There is obviously considerable variation in ell these measures from one person to another. This is also evident in the standard deviatiom given in Table 3. While caloric intake had a considerable range in the study group, ‘there was a heavy concentration of reported intakes for men and women at the modal values (Figure 7). This appears visually in the marked “peakedness" of these curves. When caloric intake is expressed as cal- ories per pound of body weight this peakedness becomes less marked (Figure 8). The curves for the major nutrients also tend to have a large nunber of values at the mode combined with considerable dispersion (Figures 9-20). Probably the most important feature of these curves is the evidence they show that diet as reported for this Study Group is by no means uniform. There is a wide range of intake. All such assessments are, of course, dependent on one's perspective. In terms of diets in which only 15 or 20 percent of the calories come fron fat almost all values for the Study Group are "high," with roughly 90 percent of the group receiving 30 percent or more of their calories from fat. Clearly, if the effects being assessed appear only above or bellow some threshhold value, a population which was truncated at that value would be unsuitable for studying the effect. An additional piece of information of some interest is the corre- lation between the major diet components and totel caloric intake (Table 5).Framinghom Study 24-6 Tt will be noted that the major nutrients correlate highly with total caloric intake. Thus, it vould be anticipated that any findings for ‘total caloric intake would generally prefigure the findings for the major nutrients. On the other hand the various ratios and percentages tend to 4 have a much lover correlation with totel caloric intake. RELATION OF SPECIFIED DIET CHARACTERISTICS [0 SERUM CHOLESTEROL LEVELS The relationship of specified diet characteristics to serum choles- terol levels is summarized briefly in Table 6, which gives correlation coefficients with serum cholesterol for a number of diet characteristics. | The serum cholesterol value used is that determined at the time of the interview. Caloric balance The following measures related to caloric balance were looked at for @ possible relation to serum cholesterol levels. a) Calories per day b) Daily caloric intake per pound of body weight In addition, each measure (a or b) was examined separately at spec- ified levels of physical activity. Calories per day showed a slight nega- ‘tive association with serum cholesterol level (over all age groups) in men but no association in women. This is evident from the correlation coefficients. When the population is divided into persons above and below median caloric intake, (or median caloric intake per pound of body weight) those nen with higher caloric intake tend to have lover serum cholesterol levels; but there is no similar difference found in vomen (Tables 7 and 8).2S... Framingham Study eke ‘This finding is somewhat puzzling and it is reasonable to inquire if this ‘Is releted in sone way to the level of physical activity. Ina given weight group persons who report higher activity levels also report higher caloric intakes (Table 9). (This may be taken as a validation of both the diet interviews and the physical activity histories). Whst is nore, relative weight tended to be greater. in men who were more sedentary (caloric intake fixed) or who had a higher level of caloric intake (physical activity fixed). Again this tends to validate the measuring techniques used. These associations appear to be considerably stronger for men than women, although this may well be a function of the paucity of women reporting higher levels of physical activity. If, then, the population is divided into high and low classes on the basis of both physical activity and daily caloric intake, men in the same physical activity class tend to have higher serum cholesterol levels at lover caloric intake (Table 10). This finding 4s contrary to expectation. Constructing @ regression equation in the fom Serum cholesterol = a + b (physical activity index) + ¢ (calories per day) + d (Framingham relative weight ) we find thet for men aged 37-49 the regression is statistically eignifi- cant and that it is near statistical significance at 50-59. These regressions are primarily accounted for by the negative coefficient assigned to the daily caloric intake. For men 60-69 and for wonen the regressions are not statistically significant (Table 12). These findings are generally consistent with the tabular material already discussed.Framingham Study 2h-8 Level.of fat intake The following measures related to fat intake were used: a) Grams of fat per day i b)--Gramsof. animal fat per day ‘ ¢) Percent calories from fat a) Ratio of daily intake of plant fat to animal fat. Paralleling the findings for total calories there is a slight nega- i ‘tive association between daily intake of total fat (and also of animal fat) with serum cholesterol level, in men but not in women. This parallel is not surprising given the high correlation between fat intake end total caloric intake. No association between percent of calories from fat and i serum cholesterol level was shown; nor between ratio of plant fat intake to animel fat intake end seran cholesterol level (Tables 12-15). Protein inteke There was a trivial negative correlation between daily protein intake (in grams) and serum cholesterol level (Table 6).* G/ ratio ‘The evidence in Table 16 vhen taken in the context of Table 6 can be interpreted as showing no relationship between C/S ratio and serum cholesterol level. Dietary cholesterol There is no indication of 2 relationship between dietary cholesterol and serum cholesterol level (Table 17)* If the intake level of animal fat is held constant there is still no relation of cholesterol intake to serum cholesterol level (Table 18)* If (further) a multiple regression * The relationships indicated are unchanged if the specified nutrients are expressed in grams per pound of body weight.sas SSSR aaah Framingham Study akg is calculated in the form serun cholesterol = a + b (animal fat) + ¢ (dietary cholesterol) there is sso Little suggestion of an association between thie pair of variables and serum cholesterol level (able 19). The age group 60-69 in men is the only exception and it may be due to sampling variability. In that ege-sex group @ significant positive correlation between dietary cholesterol and serum cholesterol is translated into a significant bivar- jate regression. (Since the correlation between cholesterol intake and fat intake ranges between .606 and .717 in these various age-sex groups, it is unlikely that relationships not evident for either singly vould be evident for both jointly.) Other analyses ‘The failure to turn up uny positive association between food intake and serum cholesterol ievel in the Framingham Diet Study 1ea to the exploration of @ large number of variant analyses. ‘These vere uniformly unsuccessful in finding expected relationships. Only a fraction of these explorations are included in this report. Two other tables might be considered beyond those already presented. One is a look at mean dietary constituents for persons with "high" (300 mgs or greater) and "Low" (1ess than 180 ngf)) cholesterol Levels (Table 20). These groups are defined not by the serum cholesterol meas- urements used for drawing the diet sample but by the blood measurements made at the time of the diet intery: Another is a miltiple regression using some key diet variables (able 21). Neither alters the generally negutive conclusions already arrivedFramingham Study 2-10 | at, The regression equation for men 50-59 is statistically significant i (this may be interpreted as a reflection of the significant negative cor- relation between total caloric intake and serum cholesterol level in this | age-sex group) but the regression equations for the other age-sex groups are not. It is evident that the coefficients assigned to the independent variables follow no clear pattern from one age group to another. (The univariate correlation coefficients. This is a reflection of the use of | 7 andependent regressors rather than one but only for 50-59 year old men | does the miltiple correlation coefficient achieve the 5f significance Level. ) | Energy balance ; Food intake is part of an equation which includes reletive weight / and physical activity. This is 2 dynamic reletionship in which time is an important dimension. The cross-sectional data obtained by this study may reflect this relationship poorly, if at all. The following correla~ tions from the Framingham Diet Study are not very impressive. Correlation of Framinghan relative weight with Physical Activity Index Galories/day Both Men 31-49 +075 (.030) -121 (.167) 2125 50-59 ~.008 (~.032) +159 (.185) “275 ' 60-69 ~.083 (-.089) :120 (.110) +188 Women 31-49 +048 (024) +007 (.033) O48, 50-59 206 ¢.026) 7006 if oi.) 10K6 60-69 ~.196 (-.157) oho (.013) +199 | | ‘The parenthetical entries are correlations with actual weight. These aiffer trivially fron the correlations with relative weight.Framingham Stu eke. ‘The correlation of the physical activity index and daily intake of calories as expressed in the study data is moderate for men and poor for vonen: Men Women 37-49 years 39h O75 50-59 years “372 +135, 60-69 years sa +03 Considering the presumptive reliability of these two measurements the correlations for men are particularly interesting. Taking the data at face value (that is, ignoring sampling varietion) one of these measure- ments, at least, should have a better than 0.6 correlation with the "true" values it purports to describe. While this is far from ideal, it is better than 4s sonetines suggested. In all Likelihood the better measurenent in men is the measurement of diet so that the estimate of 0.6 is really on the low side. (Given certain simplifying assumptions respecting measurement error, x = Rez, where r is the sample correlation between two variables, R is the true correlation, », is the correlation between the sample values and true velues for one variable and x, 4s the corresponding correlation for the other variable. If ve make the worst assumption (that R = 1) ana ry = ty, then r, 2 /¥™ Reference (1) gives an estimte of the coefficient of reliability for calories based on 33 re-interviews of une 2, % 7 0.8h. % VE x, ‘that ancther estimate of r, is 0.92. hus, the estinate of 1, given in the text (x, 2 .6) vould eppear to be 8 conservative one). ae lov correlation vetween physical activity and caloric intake, on the one hand, and relative veight on the other, may be accounted for by the fact that for most people caloric intake and energy ekpenditureFraninghan Study alate are closely balanced and that relative weight is the end product of minor @ifferences between them accumilated over an extended period of tine. mus, if the balance between energy intake and expenditure is criti- cal in the control of serum cholesterol levels it might appear in the relationship of serum cholesterol level to weight change. An equation can be made in the form serun cholesterol (st Exan 4) = a +b (FRW) + ¢ (change in FEW) where FRW is the average relative weight over the first four examinations ‘and the change in FRW is measured by the Exam 4 FRW minus the Exam 1 FRW (table 22). In this analysis we use data from the entire Franinghan cohort. Generally speaking, both veight and weight change appear to have a weak association with serum cholesterol level. Thus, energy balance, as contrasted with diet as such, does appear to have sone relation to serum cholesterol level in this population, albeit not an impressive rela- tion. The data on energy balance, however, deserve a more extended analy- sis than would be appropriate in this report. RELATIONSHIP OF DIETARY IRON AND BLOOD HEMOGLOBIN A secondary hypothesis of the Franingham Diet Study, quite unrelated to the atherogenesis study, was thet there vas a direct relationship votween the intake of iron and blood hemoglobin levels. To the contrary, the study data suggest a veak negative association, not significantly aif- ferent from zero in most specific age-sex groupe but always negative, always about the same order of magnitude, and (for all age-sex groups combined) statistically significant.Framingham Study aberg Men Susber: Correlation 31-49 209 -0.075 50-59 az -0.091. 60-69 83 ~0.055, Women 31-49 2a 0.117 50-59 ka -0. 084 60-69 5 -0.08L Whether this arises from a tendency of persons to take more iron in their diet vhen informed they have "narginal anemia" is difficult to say for sure but it seems unlikely. If mean intekes of dietary iron ore examined for persons with low and high blood hemoglobin levels there is no indication of a dramatically high intake of iron in persons having Low blood hemoglobin levels. Rather, there is a trivial difference between the intakes for persons with high end low blood hemoglobin levels (Table 23). RELATIONSHIP OF DIS? 20 CHD INCIDENCE In undertaking the diet study at Framingham the primary interest was, of course, in the relation of dict to the development of coronary heart disease (CHD), It was felt, however, thet any such relationship would be an indirect one, diet influencing serum cholesterol level and serum cholesterol level influencing the risk of CHD. However, no rela-~ tionship could be discerned within the study cohort between food intake and serum cholesterol level. It does not necessarily follow, however, that food intake is unrelated to CHD incidence by sone other unspecified mechanism. In the period between the taking of the diet interviews and the ena of the 16-year follow-up, ‘7 cases of de novo CHD developed in the DietFramingham Study aheuh | Study group. The means for all the dist variables measured vere practi- cally the same for these cases as for the original cohort at risk i (Table 2t:). There is, in short, no suggestion of any relation between diet and the subsequent development of CHD in the study group, despite a | @istinct elevation of serum cholesterol level in men developing CHD. (fhe 1 large mean difference in serum cholesterol levels for men (Table 2) is | exaggerated by the soepling scheme vtich overseapled en with very high | and very tov serun cholesterol levels.) CONCLUSIONS | ee acm / reported diet intake and serum cholesterol level in the Framingham Diet Study Group. The one exception vas a weak negative association between | Wate negative findings on a sample can be argued to be consistent with positive findings in the universe, the data strongly suggest that if there really is any association between diet intake and serum chol- esterol level in the Framingham Study population it is probably a veak one. It 4s important to be perfectly clear what these findings mean and what they do not mean. They mean that in the Framingham Study Group, diet (as measured here) is not associated with concurrent differentials in serum cholesterol level. It does not mean that the difference between Framingham and some other population (say the people of Japan) in serum cholesterol level is unrelated to aifferences between these populations in diet. Still less does it mean that serum cholesterol levels cannot be changed by changes in dietary intake.r he amingham Study On the other hand, these findings suggest a cautionary note with respect to hypotheses relating diet to serum cholesterol levels. There is @ considerable range of serum cholesterol levels within the Framingham Study Group. Something explains this inter-individual variation, but it is not diet (as measured here). Clearly if there is to be an attumpt to manipulate serum cholesterol level in a general population it would be desirable to mow what these powerful but unspecified forces are. We have repeatedly stated that the findings relate to dict as measured ‘here. This implies that using sone other technique for meas- uring @iet intake or measuring some other dietary parameters than this | study did might yield different conclusions. It is important to recog- i nize this possibility. It is also important in considering some other i parameter included in this study. Furthermore, it should be remembered that, for men at least, interrial evidence indicates that the measurement of food intake was relatively good in this study. Two parameters other than serum cholesterol vere also considered. One was blood hemoglobin--was it directly velsted to iron intake? Tt was found there x 5 in fact, a very weak inverse relationship between these two variables, whereas (of course) the postulated relationship was in the other direction. ‘The other vas coronary heart disease incidence-- vas it related prospectively to diet? No relationship was found.Table 1. Exclusions from Framingham dietary sample because of preexisting disease. f } otal Men Women | Totel with preexisting disease 13 12h 49 | All heart disease, definite* TO 5. ig | Myocaraiel infarction uh uy - | Angina pectoris 1 13 & I Rheumatic heart disease 22 1 10 Hypertensive heart disease ly 2 5 1 Hypertension, definite 103 3 30 NOTE: No man remained in the diet sample if he had any of the specified conditions on his last exem. Some women in the sample do have some of these conditions, since there was no clinical selection for those drawn as spouses of men in the "aiet sample” or with i high or low cholesterol. * When more than one of these conditions were present the subject e was assigned to the first category on the list. i iTable 2. Summary description of the ngham dietary sample Age at time of interview Sex and Sample Category PotelL «37-32 o=d Ons 60-6: Total (both sexes) Lows dig 455, 328 Men Total Rendom sample Low cholesterol. High cholesterol Women Total Random sample Wives of men in random sample Wives of men with high or low cholesterolrN 3. Mean and standard deviation of some characteristics of the average daily intake of food, by sex: Framingham Diet Study ‘Tab’ Men Characteristic Mean Bede Mean Mumber of persons 437 4T5 Calories 3156 TOL. 7 elke 56L.1 Protein (grams ) Total 12h aba 88 19.6 Animal oh 20.1 6 16.6 | Plant 30 9.2 20 6.6 Fat (grams) otal 136 36.8 96 31.3 Animal 98 30.7 65 2h.3 Plant 38 18.0 30 16.4 Carbohydrate (grams ) Total 327 91.9 228 70.2 Simple 170 58.0 12k 4h. Complex 7 49.2 lok 37.5 Cholesterol (ng) To 220.9 ga 170.0 Tron (me) 19 3.8 a 3.2 Percent calories from Fat 39 5.7 40 5.7 { Simple carbohydrate aL 5.0 23 5.9 Complex carbohydrate 20 he ag 4.6 Percent fat from animal sources Te 1.0 6 wa. i Ratio of Fatty acids: polyunsaturated/saturated ond 0.13 one 0.33 Carbonyérates: complex/simple 1.000.390.9037Table 4. Mesn of some cher by age for men ; Characteristic Number of persons Calories Protein (grams) Total Animal Plant Fat (grams) Total Animal Plant Carbohydrate (grans ) Total Simple Complex Cholesterol (ng) Tron (mg) Percent calories fron Fat Simple carbohydrate Complex carbohyarate Percent fat from animal sources Ratio of Carbohydrates: complex/sinple eristi 3T- ho Rotel 39 uh Wah 33" 97 3156 3330 3260 lek 129 «12h ou or 66 303233 1360 Uk 98 103 98 3 keg 3T 357 3h3, 170 189 172 157 168 171 Toh 668666 19 ig 1g a9 ee aOeee39) Bless eh 20 «20k T2 TL 69 0.99 0.93. 1.08 45 4 87 3257 126 95 aL 139 100 39 33h ars 161 TOL ig aL 20 Te of the average daily inteke Framingham Diet Study 50- oh 8 31a. 125 96 30 135 oT 38 318 162 156 62 1g 39 aL 20 na of fooa 55- 60- 32, oh 52 3046 301K 12h ee 9 95 2B oar 133-139 8 oT 5 33 316 30h ye 160 13143 13t Tre 1B 3 » 22 aL i 19 73 15 0.94 0.99 1.05 0.92 2853 3 87 26} 135, 93 B 301 161 140 70 18 39 2at 20 0.92) Chara Nunbe Calor Prote! Carboh Tot Sim Com Choles: Iron (1 Percent Fat Siimy Come Percent Ratio c carb| 1 geble 4. Mean of some characteristica of the average daily intake of food ty age for women: Framingham Diet Study a 3T- Wo- 45- 50- 55- 60- B- 69) characteristic fotet 39 Wh 49 Sh 5D Gh 2h F number of persons ATS 2 12 100 909 63 55 2853 | calories elle 2508 2171 2181 2132 2068 1951 2098 Protein (grams) 3 Total 8 95 86 90 90 8 Bk 89 =) Animal 6 70 6 70 70 6 6 69 26 | Plant 20 2h al el 20 19° «19 20 Fat (grams) 25 total. 96 Ws 97 100 9% 90 85 9b 93 Animal, 65 3 & 6 6 63 60 67 32 Plant . 30 a J 31 29 aT a aT Carbohydrate (grams ) j 30L Total, {228 ark 230 225 aah 205 aah 230 161 Simple wk 151 19 120 126 lt 1a 15 yf 140 Complex lok lal 12 106 98 98 93 105 : 710 cholesterol (mg) Wge2 SWB 462 509 512468 Be Soh 18 | tron (mg) woop wow mw ow mw om F 39} Feat ho AL ho 4o 39 39 39 ho 23 Simple carbohydrate 230 2h 22 ba ahhh 20, Complex carbohydrate i = 19 20 9 Bap ago tsi Th) Pereent fat from animal sources 6 6 & § 8 on n i | Ratio of +92) Carbohydrates: complex/simple 0.89 0.83. 1.0L 0.95 0.81 0.81 0.82 0.91 iCorrelation of calo: average daily intake of food by sex and age: Characteristic Number of persons Protein Total Animal, Plant Fat Total Animal Plant Carbohydrate Total Simple Complex Cholesterol Iron Percent calories from Fat Simple carbohydr ate Complex carbohydrate Percent fat from animal sources Ratio of Fatty acids: polyunsaturated/saturated Carbonyarates: complex/simple 7 +56 “19 03 +o Men 50- 2 133 1T OT “TT ~ 002 -.02 ~15 60- 8 “15 -6L “19 “78 -66 +39 on 01, Fr s with some other characteristics of the Framingham Diet study Women a OS ig sates 239-153 +80 72 TA 162 55ST 8200679) wTL +92 +90 88 “172 .Th 80 TL 62.36 187 SB 91 -80 78.73 55 62 80.73. T2 28.35 095d ae 08 a.1L =.003. +205 06 1g -.08[ Table 6. Correlation of serum cholesterol with some characteristics of the average daily intake of food by age and sex: Framingham Diet Study Men Women Bf 50- 60- = Sf- «50> «B= Characteristic »®» 3% © Ww 59 6 Number of persons 21313286 ze. 4378 Calories 109 -.20 -.08 +06 +05 +05 Calories per body weight a7 +21 -08 -30 oh ‘As P Protein Total Oh -.07 -.0L -.03 -.03 -.10 Animal. +003 OL OL -.02 -.08 | Plant -.10 -.20 12 -.03 -.10 : Fat Total 06 -.10 -.02 a +15 Ok Animal 7.05.06 -.002 -.0k 610.0 j Plant 10h =.08 0h wh 2a ; Carbohydrate i Total --12 -.32 0-222 --06 -.04 07 ¥ Simple 12 -.30 0 =.14 OL =. 1k Complex --10 -.230 -.23 --12 -.02 -.06 Cholesterol 105-0523 =k 605 | Iron +06 -.07 =.04 — =.09 | Percent calories from | Fat 102 IL 02 =k 26 ~.003 Simple carbohydrate -206 -.290 -.14 +10 -.13 15 Complex carbohydrate -106 -613 0-623 2-07 =123 \ { Percent fat from animal sources -.0l .04 .03 «10 -.06 -.008 | Ratio of | | Fatty acids: 4 polyunsaturated/saturated 09-03. -.05 | -.0k 206-08 Carbohydrates: complex/simple .02 .09 .ol -.08 .a2 -.19 qTable 7. Mean serum chole: erol level according to daily caloric m Diet Study t intake: Framing! Nunber Mean Serum Cholesterol Level Men Above Median Below Median* © Above Nedian* Below Median* 37-bh Te 56 229 235 45-54 8h 82 235 253 55-69 ST 80 230 238 otal 213 218 231 aug, Women { 3T=bb 8 52 226 216 45-54, 89 86 243 246 55-69 53 8h 260 263 ToteL 220 2028 el 2hs * Median daily caloric intake: Men = 3124.79; women = 2081.08 NOTE: There are 6 men with unknown serum cholesterol. 5 are in the above median daily caloric intake and 1 in the below group. There are 33 women with unknown serum cholesterol. 17 ere in the above daily caloric intake end 16 in the below group. Serum cholesterol is measured in mgif Table 8. Mean serum cholesterol level accor@ing to deily caloric " intake per pound of body weight: Framinghan Diet Stuay id Number Mean Serum Cholesterol Level Men Above Median* Below Median Above Median* Below Median* 31-4 6 a 231. 232 45-54 19 8r 205 261 1 55-69 66 m 209 239 Total 212 219 208 ed ; Nomen 37-44 3 oT 225 217 45-54 96 19 aha, 2uy | 55-69 52 85 257 264, 3 | otal zal zal 2ho 2h6 i i * Median daily caloric intake per pound of body weight: Men = 18.66; women = 15.26# a eee Table 9. Daily caloric intake by P and body weight: Fra: Physical Activity Number fean Da: Ca’ c Intake Index * Above Median**Below Median** Total Above Mediar*Below Median™| Men/ 37-49 <30 al uk 2878 2965 2729 30-34 52 38 3209 3335 3036 Ble 38 36 3557 3637 3413 Unimown 5 10 3208 3551 3037 Total ug 98 3270 3366 3153 Men/50-69 <30 ho 3 2855 2905 279k 30-3 3h 48 2927 3171 2755 ake 23 ho 3399 3u5h 3368 Unimown 2 1 3e72 2875 oer Total 99 aL 30k3 3123 2979 Women/37-49 <30 25 25 2088 ig92 2185 30-34 62 B 225; 2254 2051 ae 16 23 2258 2350 219k ‘Unknown lo 6 2187 2180 2195 Total nie et 2215 2202 2225 Wonen/50-69 <30 52 39 2016 2023 2007 30-34 4g 55 2068 2030 aloe ake 16 a 2162 20K7 2329 Unknown, 8 6 2227 2156 2322 Total aes a 2068 2037 2103 * Physical Activity Index is based on the average number of hours per day spent at various levels of activity and equels 1.0 (hours sleep or rest) + 1.1 (hours sedentary) + 1.5 (hours slight activity) + 2.4 (hours moderate activity) + 5.0 (hous heavy activity). ** Median weight: Men = 167 pounds; women = 136 pounds r | nal 4 M| Table 10. Mean cholesterol level according to Physical Activity Index and daily caloric intake by sex and age: Framingham Diet Study : Physical { = Activity Number Mean Serum Cholesterol i Index Tove Median Below Median © Above Median* Below Median Men/37-b4 j 32+ 45 20 229 252 < 32 aL 32 229 227 Men/li5-5i 324 51 30 231 2g | < 32 30 4g 23h 257 { Men/55-59, | 32+ 3h 31 205 23h | < 32 23, 48 2 aba i | Women/37-4h 4 | jue 5h 26 220 202 | 18.8 19.3 18.8 18.4 ‘aman ‘37-49 222 6 us acy a1 2 Us.0 15.2 14.0 U.0 13.2 12.7 50-59 143, ° 6 SL 23 oO 13.8 wwe U1 13.6 13.6 60-69 6 1 AL 18 uw 2 13-7 12.3 14k 13.0 18.8, 1h.0 Total AL T 225 Lay 58 4 13.9 14.8 14.2 13.8 13.3 13.3 NOW: Hemoglobin 1s expressed as gx/100 al Nu cal Pre car SepTable 2h. Mean values for specified nutrient measurements for persons who developed CHD s ent to their dict interviews, by sex: Framingham Diet Study Men Women, i a eS Ge Actual Expected Actual © Expected Number of cases 2 360 6 hag Calories BUM. 3127.9 2162.7 2087.8 Protein (grems ) A Total 127.0 aeh.2 90.0 86.8 Animal 96.3 94.7 10.7 66.9 Plant 30.7 29.5 19.3 19.8 Fat (grans) Total Ho. 135.2 98.5 Animal 102.1. 98.0 10.3 Plant 38.2 31.2 28.2 Carbohydrate (grans ) Total 321.8 320.2 224.0 224.3 Simple 162.5 154.0 97-0 2o1.2 Complex 159.3 166.3 1eT.0 123.1 Cholesterol (grams) 708.3 6.2 519.9 MqT.5 Iron (milligrams ) 19.4 18.8 1.2 13.7 Percent Calories fron fat 39.8 38.9 40.6 39.3 Fat from animal sources 72.5 72.4 TLD 69.2 Calories from simple CHO 20.3 all 23.4 23.8 Calories from complex CHO 20.7 19.7 11-6 19.2 CHO from complex sources 50.9 48.3, 3h 45.0 c/s ratio La 1.0 0.9 0.9 Framingham relative weight (#) 104.5 102.0 99.8 101.3 Systolic blood pressure (mm Hg) 135.1 ‘131.2 139.3 134.0 é Serun cholesterol (mag) 270.8 2ho.T 2h9.6 2k8.2 Physical Activity Index 33.8 33-8 30.2 31.2 1 Age-specific means for total group adjusted to age-distribution of CHD cases. No CHD cases appeared outside the age range 40-6lI years.1 aunsiyuvaA 096T 6S61 8S6T LS61 dag unr 4eW 99q dag unr sew 99q dag unr sew rg dag unr sew inp dy -uer 399 -Inf dy -uer -}9Q -Inf -udy -uer 399 Inf -idy -uer fie freee eee ee alc ce ere) OR 000€ SaldyO1v9 00zE e dea,/sauojeg £:79-=q e @ OOve € € 92 OF I2 Lb LE BE v2 YE ZE OE BI FE ZE Polag ul Ua JO Jaquiny 0961 YIGW31d3S OL LS61 AMWANVE ‘NAW ‘AGALS 1310 WWHONINWY4 *C01¥ad HLNOW-33NHL HOWS X04 IMVLNI DIYOIVS ATIV JOVYIAW 40 INTVA NVIW | aundiyCALORIES Figure 2 MEAN VALUE OF AVERAGE DAILY CALORIC INTAKE FOR EACH THREE-MONTH PERIOD: FRAMINGHAM DIET STUDY, WOMEN, JANUARY 1957 TO SEPTEMBER 1960 Number of Women in Period 3735 25 36 42 37 25 44 29 46 35 41 22 19 2 2800 e 2600 b = 42.9 Calories/Year 2400 2200 2000 woot titty yyy yyy yy yy Jan- Apr- July Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jul Oct. Jan- ‘Apr-Jul. Mar Jun Sep Dec Mar Jun Sep Dec Mar Jun Sep Dec Mar Jun Sep 1957 1958 1959 1960 YEAR —- rn Figure 3.096T 6S6T 8961 LS61 dag unr sew 29g dag unp sew 99q dag unr sew oaq dag unr sew sin -ady -uep -0Q -In¢ -dy -uer 399 -INf -ady -uer -399 -Inf -Ady -uer osCOSSSTSCT CT Tees acre eres eee eee lea) ore e O€T e e OvT e sea,/Sweiy [g'9- = q e 8 6 ost € €¢ 96 OF Te dw dE BE be 96 26.96 B1..vE ZE Poled U! Ua JO JaquiNN 0961 YIGWILd]S OL LS61 ANVANVS ‘NAN ‘AGNLS 131d WVHONINWYS :01dad HLNOW-33YHL HOW] YOd SNVINI L¥4 ATIVG JOVYIAV JO INTWA NVAW € ansiy (swei) Lydg analy aV3A O96T 6S61 8S6T L961 dag unt se 28q dag unr sey vag das un Jew 2aq dag unr sew Inf _udy wer 490 “Inr ~ady -uer 499 inp -idy -uer 499 Int dy -uer aaa Tees lessee T Tess) 22 oot ott 4PA/SWRID OE" = g ozt z 6t @ Wy Ge Oy G2 wy Sz Le ze of cz se Le Polued Ul UaWOM Jo Jaquinyy 0961 UIGW31d3S O1 L661 AYVNNVE ‘NIWOM ‘AGNLS 1310 WVHONIWWUS *G01Ndd HLNOW-33YH1 HOW] YO INVINI LVd ATIVG JOVUaAY 40 aN1VA NVIW y analy SSSR meas eee ee (swesp) v4BODY WEIGHT (Pounds) Figure 5 MEAN VALUE OF BODY WEIGHT FOR EACH THREE-MONTH PERIOD: FRAMINGHAM DIET STUDY, MEN, JANUARY 1957 TO SEPTEMBER 1960 Number of Men in Period 32 34 18 36 32 36 24 38 37 47 21 30 2 23 3 e 180 175 b = -1.95 Pounds/Year 170 e 165 See eae] ttf eee tf ae (ae [case] Jan- Apr- Jul- Oct- Jan- Apr-Jul Oct- Jan- Apr- Jul- Oct- Jan- Apr- Jul Mar Jun Sep Dec Mar Jun Sep Dec Mar Jun Sep Dec Mar Jun Sep 1957 1958 1959 1960Da eet WINE ATIVG NI S3INO1WS L aun3ig + UVaA 096T 6S6T 8S6T LS6T dag unr 4eW 2G dag unr sew 99g dag unr Je 9aq deg Une sem Inp Ady -uer 990 nr -ady -uer 399 -Inp -idy -uer 499 np dy -uer rr leeea lee 429,/Spunod Z0'T = 4 z 61 2 Ib GE ob 62 bb gz 4€ ty 9€ G2 GE LE Pola ul uawom jo saquiny, OET GET ObT SbT OST 0961 U3GW31daS O1 LG61 ANVNNVI ‘NAWOM ‘AGNIS 1310 WYHONIWVES Old HNOW-33YHL HOV] Od LHOIIM AGO 40 INVA NVIW (spunog) LHOIaM AOStt — ll Figure 7 CALORIES IN DAILY INTAKE: FRAMINGHAM DIET STUDY Number Mean Std. dev. 25 Men 437 3156 = 701.7 475 2142 561.1 20 a 3 4 15 & © 5 = a & 10 2 e 5 0 0 1000 2000 3000 4000 5000 6000 ' CALORIES.— eee VLE Aw ree 6 aunsiy —_— AH9I3M AGOg GNNOd Yad Sado) ov GE 0€ GZ 02 ST OL S I T T T T A 8 SST Gly UBIO «. ev O6t Lev va —— “A@P "PIS Uea| JaquinN -AQALS 1310 WWHONIWWUd +LHIM AQO@ Y3d ONNOd Y3d 3MVLNI ATIVO NI S31NOTW9 8 oundi4 OL ST 02 dNowd JO LN3043d(swes5) NIBLOUd oot Ort O21 O01 08 96 Tv “nap “PIS 88 vel ueew Slv Ley Jequinn T T UaWOM ----= uw —— AQNLS 1310 WVHONIWWH4 MVINI ATIVG NI NIFLOUd 6 analy ol ST 02. dnNOud 40 LNIOY3dAL eunziy Di (swesp) SNIFLOUd TWWINY O8T O9T Oz 00T os T T 2 ty um TO 6 egy vay —— N@P ‘PIS ueaW —saquiny AQNLS 1310 WVHONINWHS “DNVINI ATIVG NI NIZLOYd TWWINY Ol amnaiy Ov or Ge dNOYD 40 LNFONad(swes9) NIZLOUd LNVId on 09 os ov oe oz ot r SRL T T i ! i t t t 1 ! t ! i ' ! 1 99 02 = Sky UAWOM -——~ H ze oe Lev ue] — St “nop "PIS ea JeqUINN W AGNLS 1310 WWHONINVYS SAVIN ATIVG NI NIZLOUd LNWId Heung ol st 4 GZ dNOUd JO LNIOUad| el analy (swesp) jy4 Gein Oe EEE Orel popes 0 7 0 a g or st oz .. | ETE 96 = Sze uawioyy -. nee 62 898 et rey vow — “OP “PIS UeAW Jaquiny AQNLS 1310 WVHONINWYS “UNVINI ATIVG NI Lyd 21 eung}y dNOYD 40 LN3OYadFigure 13 FAT IN DAILY INTAKE PER POUND BODY WEIGHT: FRAMINGHAM DIET STUDY Number Mean Std. dev. 20 —— Men 437 0.822 0.233 ---= Women 475 0.708 0.262 PERCENT OF GROUP O1 0.3 05 0.7 0.9 lt 13 15 17 FAT PER POUND BODY WEIGHT (Grams)“DIVINE ATIVG NI LW4 LNWId GL aingiy (sweip) L¥4 IWWINY osz sez 002 SLT ost set oot se oe : v T 0 $ ot st fz go Lp uaWom --=-- Loe 86 “ey = vay —— 02 ep PIS UeBW sJaquiny ANS 1310 WWHONIWWUd “AYVAN ATIVO NI LV TYWINY i aunty dNOYD JO IN30uadFigure 15 PLANT FAT IN DAILY INTAKE: FRAMINGHAM DIET STUDY Number Mean Std. dev. Fass — Men 437 38 18.0 2 Women 475 30 164 a PERCENT OF GROUP SLU aan 09 SS a “g ov “s 6€ nap “PIS URE Slp Ley Jaquiny, v4 WOYd S3INOTW) LN3Ou3d Sb ov SE 0€ T T uawom ———. uy — AQNLS 1310 WYHONIWYYd “LV4 WOUd S3U01¥9 INZONId 91 aundiy 02 dNOYD 40 IN30Y3dPERCENT OF GROUP 20 Figure 17 PERCENT FAT FROM ANIMAL SOURCES: FRAMINGHAM DIET STUDY Number Mean Std. dev. ee Men 437 2 110 ween Women 475 69 12.1 rN /_ ae ee ee ee 61 eundiy GIOV ALl¥4 G3LWYNLWS OL aloy ALLW4 G3LVYNLVSNNATOd OLY 60 80 £0 90 gO v0 £0 20 10 .. T T T T T T : ’ S or ST er ey Sly uawom ——— cL iv Ley vn — “ABP “PIS: ueay daquinn, 02 AGALS 1310 WVHONINWUS dNOud 40 LNI0YNad “2NVENI ATIVG NI GIOY ALLW4 OILVUNLVS O1 Gloy ALLW4 G3LVUNLYSNNATOd OlLVY SI aunaiyFigure 19 IRON IN DAILY INTAKE: FRAMINGHAM DIET STUDY i Number Mean Std. dev. 25 — Men 437 19 38 —-- Women 475 14 32 PERCENT OF GROUP(stead) ‘1OY4SLSITOHD oost O0ET OOIT 006 002 00s 00E 001 ae T T T : g i f H ot t 1 i I ST ! 1 1 1 i oz OOLT 26 Sly UaWom ----. STN ! —L,r—~—~C~C—C. Cs NY “A@P/PIS UeBW JequinN sz AGALS 1310 WWHONIWWUd “SVANE ATIVG NI 10N31S310H9 02 amndy dNOUD 40 LNIONadtha tho ain ant not spe ag rel oyFramingham Study Appendix, Throughout this paper the dste have been treated as if they derived from a simple random sample. However, the sampling scheme involved sev~ eral constraints which have to be considered, since they are bound to intro- duce some bias into the estimates of regression (or correlation) made for this paper (unpublished notes by Max Halperin). In all likelihood the extent of the induced bias is small. However, it As worth restricting at least one key table to data from the random sample to reassure the reader on this point. The table that is reproduced is Table 6. If the correlations between various diet characteristics and serum cholesterol are examined separately for the rendon sample and the supple- mentary groups, the following appears to be the case: Zn men the correlations appear to be smaller for the random semple than the supplementary group. Since the supplementary group of men vere those with either high or low cholesterol values at their initial deter- mination the effect of this type of selection does appear to be (as anticipated) an upward bias. The supplementary group of wouen were chosen not on the basis of their om characteristics but on the basis of their spouses' characteristics. Hence it is difficult to guess how they might differ from the random semple. Whst differences are evident in the cor- relations appear to be of a random nature. The generally negative conclusions in the report are only reinforced by these facts.Appendix Table. Correlation of serun cholesterol with some ] App characteristics of the average daily inteke of food vy age and sex: Framingham Diet Stuay (RANDOM SAMPLE ONLY). Men Women Bi 50-80 Be = Characteristic 2 2 8 i oS & Cha! Mumber of persons 139° ok 66 ag 5948 Tag Calories “OL 15-03 .00l 0h coy Cat, Calories per body weight tlk +612 0k =.0L | 0s ae cal) Protein Pre Total A = og, 7-06.25 t Animal 206-20 015-07 2.05 005 i Plant TOL =106 -.09 05-08-09 1 Fet Fat Total 104 005 OL =. ak Ly : Animal =.0L -.06 -2 a -.10 1 Plant 06 100 an i Carbohydrate Car) Total 08 eer ot 06) og ay ‘ Simple 2D 15 ior 108 hos, cop 1 Complex 70h -.18 oa OL 7-09.02 t Cholesterol Oh 08 sags, -12 -.18 cha Iron 103-06 -.00L -.10 207-26 Ind Percent calories from Per Fat ph mk soy +30 =. Simple carbohydrate wi 22 07 ak 5 gh Complex carbohydrate “Of =U 10 0h 120. -.05 Percent fat from animal sources +0 =.02 -lo “a --06 -.16 Per Ratio of Ret Fatty acta polyunsaturated/saturatea +03 =.06 -.18 +06 10 Carbohydrates: complex/simle 05 4 gp -06 =.17Appendix Table. Correlation of serum cholesterol with sone characteri of the ge daily of food by age and sex: Framingham D: (SUPPLEMENTARY GROUPS ). 3T- 50-60-37 50 0 Characteristic ig Bi so a8 6 oe umber of persons 7 3 20 132 8k 30 | Calories =120 027 -519 =-10 06.08 Calories per body veight 2h 232 0h 015 03 Protein | Total -.15 -.09 -.18 -.003. =.002-1T. i Animal -.0T 605 ML SOT =.00 «25 | Plant 120-033-130 ---16 02-2 Fat otal <7 -.07 =:16 -15 20 Animal =.09 -=.10 -.10 609 SL | Plant -.220 002 TB | Carboayarate otal 218-238 46-213, -.0T Simple -15 7.38 -.28 +02 Complex =16 =.30-.46 05-17 Cholesterol -16 -.06 «34 20 he | Iron -18 -.08 -.22 07 2h Percent calories from Fat 10816 R20 283 Simple carvohyarate el 40-126 OT =e Complex carbohyérate -.0% =.17 =.50 -.23) OL | Percent fat from animal sourees -.09 12 -.11 08 -.07— 23. | Ratio of Fatty acts: | polyunsaturated/saturated ag -0L es he OT --27 Carbohydrates: complex/simple -.0l .13. --03— - 1. +08 -.19 |