Dietary fat intake and early mortality patterns – data from The Malmö Diet and Cancer Study

Authors


Margret Leosdottir MD, Department of Medicine, University Hospital (UMAS), S-205 02 Malmö, Sweden.
(fax: +46 40 92 32 72; e-mail: margret.leosdottir@med.lu.se).

Abstract.

Objectives.  Most current dietary guidelines encourage limiting relative fat intake to <30% of total daily energy, with saturated and trans fatty acids contributing no more than 10%. We examined whether total fat intake, saturated fat, monounsaturated, or polyunsaturated fat intake are independent risk factors for prospective all-cause, cardiovascular and cancer mortality.

Design.  Population-based, prospective cohort study.

Setting and subjects.  The Malmö Diet and Cancer Study was set in the city of Malmö, southern Sweden. A total of 28 098 middle-aged individuals participated in the study 1991–1996.

Main outcome measures.  Subjects were categorized by quartiles of relative fat intake, with the first quartile used as a reference point in estimating multivariate relative risks (RR; 95% CI, Cox's regression model). Adjustments were made for confounding by age and various lifestyle factors.

Results.  Women in the fourth quartile of total fat intake had a significantly higher RR of cancer mortality (RR 1.46; CI 1.04–2.04). A significant downwards trend was observed for cardiovascular mortality amongst men from the first to the fourth quartile (P = 0.028). No deteriorating effects of high saturated fat intake were observed for either sex for any cause of death. Beneficial effects of a relatively high intake of unsaturated fats were not uniform.

Conclusions.  With the exception of cancer mortality for women, individuals receiving more than 30% of their total daily energy from fat and more than 10% from saturated fat, did not have increased mortality. Current dietary guidelines concerning fat intake are thus generally not supported by our observational results.

Introduction

The effect of dietary fat on disease development, longevity and mortality has long been a popular focus of interest, dating from the early 20th century when a Russian scientist produced atherosclerosis in animals by feeding them with high fat, high cholesterol diets [1]. From the 1950s through to the 1970s evidence accumulated from various epidemiological, ecological and clinical trials supporting the theory that diets rich in animal fats and poor in unsaturated vegetable fats produced unfavourable risk profiles for cardiovascular disease and cancer [2–4]. Results from these studies founded most of the dietary guidelines by amongst others the World Health Organization and American Heart Association, the basis of which is still used in reviewed editions today [5, 6].

Over the last decades evidence from large-scale epidemiological studies has been emerging, partly defying the previously believed hypotheses, with voices rejecting the fat-disease hypotheses becoming more prominent [7–9]. Most researchers today agree on total fat intake not being a risk factor for cardiovascular disease or cancer [3, 9, 10]. The role of dietary fat in the development and treatment of obesity has also been under question [11, 12]. The largest bulk of evidence still pointing towards dietary fats influencing disease development, involves increased risk of various cancers and cardiovascular disease in relation to a high intake of trans fatty acids and saturated fats, as well as lower risks with diets rich in unsaturated fats, especially polyunsaturated fats [2, 3, 9, 13]. All the same, many studies have showed negative findings or even opposite results [8].

In spite of conflicting evidence, most institutions and governmental authorities issuing dietary guidelines still encourage limiting the percentage of calories from total fat in the diet, usually to <30%, with special focus on limiting saturated and trans fatty acids [14–16].

The aim of this study was to examine whether total fat intake, saturated, monounsaturated or polyunsaturated fat intake, or the ratio between unsaturated and saturated fat intake, are independent risk factors for prospective all-cause mortality, cardiovascular mortality and cancer mortality, within a large population-based cohort.

Material and methods

The Malmö Diet and Cancer Study

The Malmö Diet and Cancer Study (MDC), a large-scale, population-based, prospective cohort study, was designed in the early 1990s to identify dietary risk factors in relation to cancer mortality [17]. A total of 74 138 individuals identified through national population registries, including all men born between 1923 and 1945 and all women born between 1923 and 1950 living at the time in Malmö, the third largest city in Sweden, were invited to participate. Baseline examinations took place at the MDC Centre at the University Hospital in Malmö between 1991 and 1996. A 141-item questionnaire addressing various lifestyle factors and medical history was administered. Trained nurses conducted a medical examination to evaluate body composition and to measure blood pressure. Blood samples were drawn from the participants and stored in a bio-bank. A 168-item dietary questionnaire and a 7-day menu-diary were administered to evaluate diet composition [18]. The recruitment stopped when 28 098 individuals had completed both questionnaires and attended the physical examination. Information on mortality has been gathered from local and national registries for a follow-up period of approximately 6.6 years. Lund University Ethics Committee approved the MDC study.

Study population

All 28 098 individuals who completed both questionnaires and attended the physical examination, 11 063 males (39.4%) and 17 035 females (60.6%), were included in the primary analysis.

Dietary, lifestyle and anthropometric variables

The dietary questionnaire recorded consumption of regularly eaten foods other than cooked meals during the past year. The menu-diary recorded descriptions of cooked meals, nutrient supplements and cold beverages for seven consecutive days. Studies on the reproducibility and validity of these methods have previously been published [19–21].

Intake of independent energy sources (fat, protein, carbohydrates and alcohol) and fibre was recorded in grams consumed per day. Relative fat intake, calculated as percentage of nonalcohol energy coming from fat, was used in this study, as this measurement of dietary fat intake is used in most nutrient recommendations around the world, including those followed in the Nordic countries [22].

Information on smoking habits, socio-economic status, marital status, and physical activity was retrieved from the lifestyle questionnaire. A detailed description of these variables has recently been published in this journal [23]. In short, subjects were defined as current smokers, former smokers or never-smokers. Socio-economic status was recorded according to the Swedish population census as blue collar workers, white collar workers and employers/self-employed [24]. Marital status was categorized as single, divorced or widowed versus married or co-habiting. Physical activity at work was self-rated as very light, light, medium heavy or very heavy. For leisure time physical activity categories of low, moderate and high levels of leisure time physical activity were adapted from the Minnesota Leisure Time Physical Activity Questionnaire [25]. Information on prior medical history was acquired from the questionnaire (for diabetes) and from local or national registries (for diagnosis of cancer, myocardial infarction, or stroke).

Weight (kg), height (m) and blood pressure (mmHg) were measured by trained nurses at one of the visits to the MDC centre. From the weight and height, body mass index (kg m−2) was calculated.

Statistical analysis

Cox's proportional hazards regression was used to estimate multivariate relative risks (RR) and 95% confidence intervals (CI). All analyses were stratified by gender. Adjustments were made for confounding by age, smoking habits, alcohol consumption, socio-economic status, marital status, leisure time- and work-related physical activity, body mass index, fibre intake, and blood pressure for cardiovascular mortality. All three types of fat (saturated, monounsaturated and polyunsaturated) were simultaneously included in the multivariate analyses. Adjustment was made for total fat in the regression analysis of the unsaturated (mono- and poly-)/saturated fat ratio. To evaluate whether individuals likely to have poorer health, leading to both higher mortality rates and altered dietary habits, would bias the results, calculations were repeated after excluding subjects with <1 year of follow-up and subjects with prior history of myocardial infarction, stroke, diabetes, or cancer (n = 4611).

Univariate General Linear Model procedure was used to calculate age-adjusted mean values. One-way anova and chi-square tests were used to compare mean values between quartiles. Statistical analyses were performed using SPSS version 11.5 for Windows. A traditional significance level of P < 0.05 was used.

Results

Tables 1 and 2 show baseline characteristics for subjects, categorized by survival status at the end of follow-up. Till the end of year 2000, over a mean follow-up period of 6.6 years (range 0.1–9.8 years) a total of 1250 deaths were registered. Mean age at death was 66.2 years (47.3–77.9) for women and 66.7 years (48.2–77.8) for men. Cancer was the leading cause of death, responsible for 59% of deaths amongst women (310 of 522) and 43% amongst men (313 of 728). Breast cancer was the most common cancer causing death amongst the women (16%), followed by lung cancer (12%). Amongst men, lung (19%) and prostate (15%) cancers were most common. Ninety-seven of 522 women (19%) and 242 (33%) men died from cardiovascular disease. Of the 28 098 subjects participating in the study, 139 individuals (81 women and 59 men) were lost to follow-up, the main reason being permanent emigration.

Table 1.  Baseline characteristics for women categorized by survival status at the end of follow-up. Subjects lost to follow-up (n = 81) are not included. Figures are presented as percentages or age-adjusted mean values with range in parentheses
 Alive (N = 16 432)Dead (all causes) (N = 522)Dead from CVD (N = 97)Dead from cancer (N = 310)
Mean age in years (range)57.3 (45–73)61.8 (45–73)63.8 (45–73)60.9 (45–73)
Social status
 Blue collar worker (%)38.043.858.342.2
 White collar worker (%)54.549.033.352.0
 Employer/self-employed (%)7.67.28.35.9
Smoking habits
 Never smoked (%)44.535.427.836.5
 Former smoker (%)27.823.824.722.6
 Current smoker (%)27.640.847.441.0
Marital status
 Married/co-habiting (%)60.850.354.648.9
 Single/divorced/widowed (%)39.249.745.451.1
Alcohol consumption (g pure ethanol day−1)7.7 (0.0–109.5)6.8 (0.0–73.7)6.3 (0.0–45.3)6.8 (0.0–73.7)
Physical activity score7935 (0–316 120)7490 (0–39 825)7184 (0–33 600)8011 (0–39 825)
Blood pressure (mmHg)
 Systolic139.3 (61–230)140.4 (90–210)147.2 (110–210)138.9 (98–200)
 Diastolic84.0 (40–136)84.6 (56–126)86.9 (65–126)84.2 (60–110)
Body mass index (kg m−2)25.4 (14.0–50.9)25.3 (15.1–44.8)25.6 (15.1–39.7)25.4 (16.7–44.8)
Total energy intake (kcal day−1)1999 (504–5556)2015 (742–4104)1923 (776–4072)2063 (742–4085)
Relative fat intake (% of total energy)38.4 (9.2–71.5)39.0 (13.3–60.6)37.9 (9.2–71.5)39.5 (13.3–60.6)
Table 2.  Baseline characteristics for men categorized by survival status at the end of follow-up. Subjects lost to follow-up (n = 58) are not included. Figures are presented as percentages or age-adjusted mean values with range in parentheses
 Alive (N = 10277)Dead (all causes) (N = 728)Dead from CVD (N = 242)Dead from cancer (N = 313)
Mean age in years (range)59.1 (46–73)62.7 (46–73)63.3 (48–73)62.8 (46–73)
Social status
 Blue collar worker (%)35.342.346.736.7
 White collar worker (%)47.143.637.647.9
 Employer/self-employed (%)17.614.115.715.4
Smoking habits
 Never smoked (%)29.017.215.317.9
 Former smoker (%)43.540.040.539.3
 Current smoker (%)27.542.944.242.8
Marital status
 Married/co-habiting (%)73.261.858.568.1
 Single/divorced/widowed (%)26.838.241.531.9
Alcohol consumption (g pure ethanol day−1)15.5 (0.0–194.0)14.3 (0.0–161.6)12.9 (0.0–92.1)14.8 (0.0–95.4)
Physical activity score8386 (0–209 556)7969 (0–95 200)7606 (0–95 200)8270 (0–58 790)
Blood pressure (mmHg)
 Systolic143.9 (90–230)147.3 (90–220)149.2 (92–220)146.3 (108–210)
 Diastolic87.9 (52–136)89.8 (60–128)91.1 (60–128)88.9 (60–115)
Body mass index (kg m−2)26.3 (13.9–50.7)26.2 (15.2–43.4)26.7 (17.4–43.4)25.9 (15.6–38.4)
Total energy intake (kcal day−1)2603 (570–6789)2598 (999–8304)2484 (999–5229)2675 (1296–8304)
Relative fat intake (% of total energy)39.7 (4.7–68.1)39.7 (13.0–60.3)39.3 (19.1–58.4)39.9 (19.2–60.1)

Diet composition

Tables 3 and 4 show diet composition and baseline characteristics for women and men, respectively, in the four quartiles of total fat intake. Fat intake in the whole cohort varied from 31% (10th percentile) to 46% (90th percentile) of total energy intake for women and 32% (10th percentile) to 48% (90th percentile) for men. For both sexes, the increase in total fat intake from the first to the fourth quartiles was attributable to increased saturated fat intake (P < 0.001, Tables 3 and 4).

Table 3.  Diet composition and baseline characteristics within different quartiles of relative fat intake for women. Figures are presented as age-adjusted mean values or percentages
 1st2nd3rd4th
  1. N = number of individuals within each quartile; n = number of deaths from all causes. BMI, body mass index; SBP, systolic blood pressure.

N (n)4258 (121)4259 (138)4259 (112)4259 (151)
% Energy from fat30.836.540.346.1
 % From saturated fats41.242.543.445.3
 % From monounsaturated fats34.834.934.734.1
 % From polyunsaturated fats16.716.015.614.9
Monounsaturated/saturated fat ratio0.860.840.810.77
% Energy from carbohydrates52.047.043.738.7
% Energy from protein16.616.215.815.2
Fibre intake (g day−1)21.319.618.316.8
Vegetable and fruit intake (g day−1)465412369317
Energy intake (kcal day−1)1832196420392163
Age57.957.757.357.0
Socio-economic status
 % Blue collar worker41.338.337.335.7
 % White collar worker51.854.155.555.5
 % Self-employed6.97.67.38.5
% Current smoker20.924.429.337.6
Prior history of cancer, diabetes, myocardial infarction or stroke (%)18.618.017.318.3
Use of antihypertensive medication (%)20.517.016.414.2
BMI (kg m−2)25.525.625.424.9
SBP (mmHg)141140139138
Alcohol intake (g pure ethanol day−1)5.97.38.29.3
Physical activity score8407807576827537
Table 4.  Diet composition and baseline characteristics within different quartiles of relative fat intake for men. Figures are presented as age-adjusted mean values or percentages
 1st2nd3rd4th
  1. N = number of individuals within each quartile; n = number of deaths from all causes. BMI, body mass index; SBP, systolic blood pressure.

N (n)2765 (186)2766 (183)2766 (155)2766 (204)
% Energy from fat31.737.841.747.7
 % From saturated fats40.141.642.444.6
 % From monounsaturated fats35.435.435.434.7
 % From polyunsaturated fats17.416.516.015.0
Monounsaturated/saturated fat ratio0.900.870.850.80
% Energy from carbohydrates51.746.242.937.6
% Energy from protein16.015.815.414.8
Fibre intake (g day−1)24.822.120.718.7
Vegetable and fruit intake (g day−1)416357326284
Energy intake (kcal day−1)2408254326582798
Age59.459.559.260.0
Socio-economic status
 % Blue collar worker37.036.534.934.8
 % White collar worker48.146.547.245.8
 % Self-employed15.017.117.919.5
% Current smoker20.924.728.140.6
Prior history of cancer, diabetes, myocardial infarction or stroke (%)16.713.711.611.1
Use of antihypertensive medication (%)24.721.320.518.6
BMI (kg m−2)26.326.426.326.1
SBP (mmHg)144144144144
Alcohol intake (g pure ethanol day−1)12.614.116.718.5
Physical activity score9028848981117798

The average intake of each type of fat in different quartiles can be seen in Tables 5–7. Intakes of unsaturated fats were more or less in line with Nordic dietary guidelines, which recommend 10–15% of total daily energy intake coming from cis monounsaturated fats and 5–10% from polyunsaturated fats [22]. Note that trans fatty acids were not recorded as a separate variable in our study and are thus incorporated into the monounsaturated, and to some extent, polyunsaturated variables.

Table 5.  Relative risks (95% CI) for total mortality by quartiles of relative fat intake for women and men. Adjusted for age, alcohol consumption, smoking, social class, marital status, physical activity, BMI and fibre intake. Saturated, monounsaturated and polyunsaturated fats were included simultaneously in the multivariate analysis. Adjustments were made for total fat intake for the ratio between unsaturated and saturated fats. Also shown is the percentage of daily energy intake (EI) that the relevant fat contributes
 QuartilesP for trend
1234
  1. aP = 0.049. bP = 0.017. cP = 0.046.

Women
Total fat
 RR (95% CI)1.00 (ref)1.08 (0.84–1.40)0.93 (0.71–1.22)1.22 (0.94–1.58)0.26
 % of EI30.836.540.346.1
Saturated fat
 RR (95% CI)1.00 (ref)0.96 (0.73–1.26)0.82 (0.60–1.10)0.89 (0.64–1.23)0.39
 % of EI12.215.217.521.8
Monounsaturated fat
 RR (95% CI)1.00 (ref)1.17 (0.88–1.56)1.28 (0.92–1.76)1.44 (1.00–2.08)a0.06
 % of EI10.612.614.016.0
Polyunsaturated fat
 RR (95% CI)1.00 (ref)0.80 (0.62–1.03)0.77 (0.59–1.01)0.85 (0.64–1.13)0.22
 % of EI4.35.46.38.1
Monounsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.94 (0.74–1.21)0.82 (0.63–1.07)0.99 (0.77–1.27)0.39
 Ratio1.000.860.770.65
Polyunsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.93 (0.72–1.20)0.93 (0.72–1.20)0.93 (0.72–1.20)0.35
 Ratio0.560.410.330.23
Men
Total fat
 RR (95% CI)1.00 (ref)0.92 (0.75–1.13)0.77 (0.62–0.95)b0.89 (0.72–1.10)0.14
 % of EI31.737.841.747.7
Saturated fat
 RR (95% CI)1.00 (ref)0.85 (0.67–1.07)1.04 (0.81–1.32)0.91 (0.69–1.19)0.72
 % of EI12.315.317.622.3
Monounsaturated fat
 RR (95% CI)1.00 (ref)0.82 (0.65–1.04)0.76 (0.59–0.99)c0.88 (0.65–1.19)0.37
 % of EI11.013.314.716.9
Polyunsaturated fat
 RR (95% CI)1.00 (ref)0.88 (0.71–1.09)1.02 (0.82–1.27)0.85 (0.66–1.09)0.44
 % of EI4.55.76.78.6
Monounsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.95 (0.77–1.17)1.00 (0.81–1.24)1.11 (0.90–1.37)0.51
 Ratio1.000.900.800.67
Polyunsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.94 (0.76–1.16)0.93 (0.75–1.15)1.11 (0.90–1.38)0.58
 Ratio0.590.430.350.24
Table 6.  Relative risks (95% CI) for cardiovascular mortality by quartiles of relative fat intake for women and men. Adjusted for age, alcohol consumption, smoking, social class, marital status, physical activity, BMI and fibre intake. Saturated, monounsaturated and polyunsaturated fats were included simultaneously in the multivariate analysis. Adjustments were made for total fat intake for the ratio between unsaturated and saturated fats. Also shown is the percentage of daily energy intake (EI) that the relevant fat contributes
 QuartilesP for trend
1234
  1. aP = 0.021.

  2. bP = 0.023.

  3. cP = 0.027.

Women
Total fat
 RR (95% CI)1.00 (ref)0.99 (0.57–1.72)0.80 (0.45–1.43)0.74 (0.40–1.36)0.25
 % of EI30.836.540.346.1
Saturated fat
 RR (95% CI)1.00 (ref)0.89 (0.49–1.62)0.76 (0.39–1.45)0.55 (0.26–1.17)0.16
 % of EI12.215.217.521.8
Monounsaturated fat
 RR (95% CI)1.00 (ref)1.66 (0.88–3.15)1.91 (0.92–3.94)1.53 (0.65–3.64)0.34
 % of EI10.612.614.016.0
Polyunsaturated fat
 RR (95% CI)1.00 (ref)0.82 (0.47–1.42) 0.46 (0.24–0.89) a0.63 (0.33–1.22)0.10
 % of EI4.35.46.38.1
Monounsaturated/saturated fat
 RR (95% CI)1.00 (ref)1.04 (0.61–1.78)0.81 (0.45–1.46)0.84 (0.46–1.53)0.39
 Ratio1.000.860.770.65
Polyunsaturated/saturated fat
 RR (95% CI)1.00 (ref)1.03 (0.60–1.76)0.81 (0.45–1.46)0.84 (0.46–1.54)0.82
 Ratio0.560.410.330.23
Men
Total fat
 RR (95% CI)1.00 (ref)0.76 (0.53–1.09)0.74 (0.52–1.07)0.65 (0.45–0.94)b0.03
 % of EI31.737.841.747.7
Saturated fat
 RR (95% CI)1.00 (ref)1.03 (0.69–1.53)1.24 (0.82–1.89)0.94 (0.58–1.53)0.98
 % of EI12.315.317.622.3
Monounsaturated fat
 RR (95% CI)1.00 (ref) 0.63 (0.42–0.95) c0.72 (0.47–1.12)0.61 (0.36–1.03)0.11
 % of EI11.013.314.716.9
Polyunsaturated fat
 RR (95% CI)1.00 (ref)0.81 (0.56–1.19)1.08 (0.74–1.58)0.99 (0.65–1.53)0.64
 % of EI4.55.76.78.6
Monounsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.93 (0.65–1.34)0.89 (0.62–1.30)1.11 (0.77–1.59)0.59
 Ratio1.000.900.800.67
Polyunsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.80 (0.55–1.16)0.82 (0.57–1.19)1.03 (0.72–1.48)0.77
 Ratio0.590.430.350.24
Table 7.  Relative risks (95% CI) for cancer mortality by quartiles of relative fat intake for women and men. Adjusted for age, alcohol consumption, smoking, social class, marital status, physical activity, BMI, and fibre intake. Saturated, monounsaturated and polyunsaturated fats were included simultaneously in the multivariate analysis. Adjustments were made for total fat intake for the ratio between unsaturated and saturated fats. Also shown is the percentage of daily energy intake (EI) that the relevant fat contributes
 Quartilesp for trend
1234
  1. aP = 0.029.

  2. bP = 0.038.

Women
Total fat
 RR (95% CI)1.00 (ref)1.08 (0.77–1.53)1.06 (0.75–1.50)1.46 (1.04–2.04)a0.03
 % of EI30.836.540.346.1
Saturated fat
 RR (95% CI)1.00 (ref)1.21 (0.84–1.75)1.04 (0.70–1.57)1.15 (0.92–1.79)0.78
 % of EI12.215.217.521.8
Monounsaturated fat
 RR (95% CI)1.00 (ref)1.00 (0.68–1.46)1.12 (0.73–1.70)1.47 (0.92–2.34)0.08
 % of EI10.612.614.016.0
Polyunsaturated fat
 RR (95% CI)1.00 (ref)0.70 (0.50–0.98)b0.80 (0.57–1.13)0.77 (0.53–1.12)0.28
 % of EI4.35.46.38.1
Monounsaturated/saturated fat
 RR (95% CI)1.00 (ref)1.07 (0.77–1.48)0.91 (0.65–1.28)1.10 (0.79–1.53)0.86
 Ratio1.000.860.770.65
Polyunsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.81 (0.58–1.13)0.99 (0.72–1.37)0.90 (0.65–1.25)0.51
 Ratio0.560.410.330.23
Men
Total fat
 RR (95% CI)1.00 (ref)1.08 (0.79–1.47)0.74 (0.53–1.05)1.00 (0.72–1.39)0.56
 % of EI31.737.841.747.7
Saturated fat
 RR (95% CI)1.00 (ref)0.89 (0.63–1.25)0.92 (0.63–1.35)0.98 (0.64–1.48)0.82
 % of EI12.3%15.3%17.6%22.3%
Monounsaturated fat
 RR (95% CI)1.00 (ref)0.86 (0.61–1.21)0.67 (0.44–1.01)0.92 (0.58–1.44)0.55
 % of EI11.013.314.716.9
Polyunsaturated fat
 RR (95% CI)1.00 (ref)0.97 (0.70–1.33)0.99 (0.71–1.39)0.84 (0.58–1.23)0.40
 % of EI4.55.76.78.6
Monounsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.93 (0.67–1.28)0.90 (0.65–1.25)1.13 (0.82–1.55)0.84
 Ratio1.000.900.800.67
Polyunsaturated/saturated fat
 RR (95% CI)1.00 (ref)0.98 (0.71–1.35)0.93 (0.67–1.28)1.03 (0.74–1.42)0.70
 Ratio0.590.430.350.24

Fibre intake as well as fruit and vegetable intake decreased significantly from the first to the fourth quartiles of total fat, saturated fat, monounsaturated- and polyunsaturated fat for both sexes (P < 0.001 for both sexes, all types of fat). By including fibre intake as a continuous variable in our multivariate analyses, it was significantly negatively correlated to all-cause mortality and cardiovascular mortality for all types of fat for men (P < 0.05) but not for women. No significant effects of fibre intake were observed on cancer mortality for either sex in the multivariate analyses.

Total mortality

Relative risks for total mortality for different quartiles of fat intake for men and women can be seen in Table 5. Women in the fourth quartile of total fat intake had somewhat higher RR of total mortality (RR 1.22; CI 0.94–1.58) even though not statistically significant. This risk increase was mostly applicable to a high RR for the same quartile for cancer mortality (RR 1.46; CI 1.04–2.04, P = 0.029). Women in the fourth quartile of monounsaturated fat intake had higher total mortality (RR 1.44; CI 1.00–2.08, P = 0.049). The second and third quartiles also had RR >1.00, even though not significant (Table 5). No significant difference in RR was observed between quartiles of saturated fat intake, polyunsaturated fat intake, or the ratio between mono- or polyunsaturated fat and saturated fat intake for women.

For men, a significantly lower RR of total mortality was observed in the third quartile of total fat intake (RR 0.77; CI 0.62–0.95, P = 0.017), with RR for the second and fourth quartiles also <1.00 (Table 5). No trend towards higher mortality was observed for saturated fat intake. For unsaturated fats, men in the third quartile for monounsaturated fats had significantly lower total mortality (RR 0.76; CI 0.59–0.99, P = 0.046). No significant difference or trend was observed between quartiles of polyunsaturated fat or the unsaturated/saturated fat ratios.

After excluding individuals with <1 year follow-up, prior history of myocardial infarction, stroke, cancer or diabetes (3101 women and 1510 men) significance did not hold for any of the results. These exclusions involved 16.5% of the cohort, thus decreasing the power of the calculations considerably. However, the observed trends in mortality did not change.

Cardiovascular mortality

Relative risks of cardiovascular mortality for quartiles of all types of fat can be seen in Table 6. For women, no significant difference was observed between quartiles of total fat or saturated fat intake. For men, a significant trend towards lower cardiovascular mortality in upper quartiles of total fat intake was observed (P = 0.028) with the RR for men in the fourth quartile being 0.65 (CI 0.45–0.94, P = 0.023) (Fig. 1). No difference was observed between quartiles of saturated fat intake for men. When looking at unsaturated fats, women in the third quartile of polyunsaturated fat had lower RR (0.46; CI 0.24–0.89, P = 0.021), as did men in the second quartile of monounsaturated fat intake (0.63; CI 0.42–0.95, P = 0.027). Having relatively high intakes of monounsaturated or polyunsaturated fats compared with saturated fats did not show benefit for either sex.

Figure 1.

Relative risk (RR; 95% CI) of cardiovascular mortality by quartiles of relative fat intake (% of daily energy intake coming from fat) for men. Adjusted for age, smoking, alcohol intake, socio-economic status, marital status, leisure time and work-related physical activity, fibre intake, body mass index and systolic blood pressure. Men in the first quartile (reference group) received on average 31.7% of their daily calorie load from fat, compared with 47.7% for the men in the fourth quartile. P-value for downward trend from the first to the fourth quartile = 0.028. *P = 0.023.

As with total mortality, significance did not hold after excluding subjects with short follow-up, prior history of myocardial infarction, stroke, cancer or diabetes.

Cancer mortality

Relative risks of cancer mortality for quartiles of all types of fat can be seen in Table 7. Women in the fourth quartile of total fat intake (46.1% of daily energy from fat) had a significantly higher RR of dying from cancer (RR 1.46; CI 1.04–2.04, P = 0.029) (Fig. 2). This was mostly attributable to a high RR for women in the fourth quartile of monounsaturated fat intake (RR 1.47; CI 0.92–2.34, P = NS). RR for women in the second quartile of polyunsaturated fat intake was significantly lower, with an RR of 0.70 (CI 0.50–0.98, P = 0.038). Saturated fat intake, and the ratio between monounsaturated or polyunsaturated fat and saturated fat, did not show any significant effect on cancer mortality for women. For men, significant differences in cancer mortality between quartiles were not revealed for any type of fat.

Figure 2.

Relative risk (RR; 95% CI) of cancer mortality by quartiles of relative fat intake (% of daily energy intake coming from fat) for women. Adjusted for age, smoking, alcohol intake, socio-economic status, marital status, leisure time and work-related physical activity, fibre intake, and body mass index. Women in the first quartile (reference group) received on average 30.8% of their daily calorie load from fat, compared with 46.1% for the women in the fourth quartile. P-value for upward trend from the first to the fourth quartile = 0.034. *P = 0.029.

The observed trends in mortality were unchanged after excluding individuals with <1 year follow-up, prior history of myocardial infarction, stroke, cancer or diabetes, although significance did not hold.

Discussion

With the exception of cancer mortality for women in the highest quartile of relative fat intake, individuals receiving more than 30% of their total daily energy from fat did not have increased mortality. Men in the fourth quartile of total fat intake, receiving almost 50% of their total energy intake from fat, had the lowest cardiovascular mortality. Receiving more than 10% of total energy intake from saturated fat did not have a significant effect on all-cause, cardiovascular or cancer mortality for men or women. Beneficial effects of relatively high intakes of unsaturated fats were not uniform, and having a high index of unsaturated fat compared with saturated fat intake did not have any detectable effect on mortality.

When looking at cardiovascular mortality, the first large-scale prospective observational study on the dietary fat–cardiovascular disease relationship was the ecological Seven Countries study [2]. In the 15-year follow-up publication, a strong correlation was observed between all-cause mortality, cancer mortality and cardiovascular mortality, and the ratio between monounsaturated and saturated fat in the diet [26]. A somewhat weaker association was observed between mortality and saturated fat intake, but total fat intake did not have any effect. Prospective cohort studies, published at a similar time and over the next decade, showed weak or no correlation between total fat and saturated fat and cardiovascular mortality and/or disease [27–31]. However, dietary assessment methods used in these studies and confounding for which adjustments were made varied considerably. Adjustments for other types of fat, especially trans fatty acids, were not uniform. Also, adjustments for fibre intake were generally not made in these studies. It has been shown that high-fat Western diets tend to be poor in fibre [32, 33]. Fibre intake has been shown to protect against cardiovascular disease and to a less extent some types of cancer [34, 35]. Thus, not adjusting for fibre intake when evaluating the effect of fat on disease incidence and mortality, especially cardiovascular disease, produces an important confounding effect. In more recent epidemiological studies, where adjustments for fibre intake were made, total fat intake and cardiovascular disease and/or mortality were as in our study not positively correlated, and saturated fat intake showed no or borderline significant correlation [36, 37]. One of those, the Health Professionals Follow-up Study (HPFS), underlined the important effect fibre intake has on the fat–cardiovascular disease relationship, with RR for the upper quintiles of fat intake dropping by approximately 20% when additional adjustments were made for fibre intake [37]. In the Nurses Health Study (NHS), a prospective cohort study of over 80 000 women, no statistically significant effects of high intakes of total or saturated fats were noticed, even though fibre intake was not adjusted for in their multivariate analysis [38].

Studies on the effect of dietary fats on cerebrovascular disease are much sparser. The two most recent large-scale cohort studies have shown no effect or even beneficial effects of increased total and saturated fat intake on stroke incidence [39, 40].

Several reviews on prospective cohort studies examining the effects of unsaturated fats on cardiovascular disease have been published [4, 8, 40]. Most conclude that high intakes of polyunsaturated fats protect against cardiovascular disease. However, in the more than 20 studies reviewed, results are not unanimous, with some studies showing null or negative findings [4, 8, 36–38, 41]. Monounsaturated fats seem to provide little if any protective effects [36, 38, 41]. In the two before-mentioned studies on cerebrovascular disease, polyunsaturated fats did not give beneficial effects, with monounsaturated fats providing no or slight benefit [39, 40]. In our study, women in the third quartile of polyunsaturated fat intake had the lowest cardiovascular mortality, but high polyunsaturated fat intake did not show any benefit for men. Monounsaturated fat intake however showed some protecting effects for men, with the second quartile having the lowest RR and the third and fourth quartiles having RR <1.00, even though not statistically significant. The opposite was observed for women, with RR in the second to fourth quartiles over 1.00. However, the number of cardiovascular cases amongst women was few, thus giving poor statistical power to the calculations with wide CI. A separate analysis within the same cohort, evaluating the effect of dietary fat on fatal and nonfatal cardiovascular events, with more than 2000 registered endpoints, awaits publication.

The fact that trans fatty acids are included in the unsaturated fat variables in our study, could have confounded the results considerably. Also, the limited variation in unsaturated fat intake in our study population, with mean intakes for the first as well as the fourth quartiles lying approximately within recommended intakes, diminishes the likelihood of revealing significant effects [22].

Unlike the ecological Seven Countries Study's results, the ratio between mono- or polyunsaturated fats and saturated fats did not show any beneficial effects for either sex in our study. In prospective cohort studies, the NHS and the HPFS, a similar ratio, the Keys score (incorporating polyunsaturated fat, saturated fat and cholesterol intake, with higher scores indicating relatively lower intakes of polyunsaturated fats as opposed to saturated fats and cholesterol), was evaluated [37, 38]. The Keys score was positively correlated to coronary heart disease in the NHS, but not in the HPFS. When evaluating the effect on fatal coronary disease in HPFS, significance was lost after adjustment for fibre intake [37].

When looking at cancer mortality, our results showed significantly higher RR for women in the fourth quartile of total fat intake. Monounsaturated fat intake seemed to be the largest contributor to this effect. For men, total fat, subtypes of fat, and the unsaturated/saturated fat ratio had no effect on cancer mortality. This gender difference might thus be attributable to female cancers such as breast, ovarian, and endometrial tumours, which jointly accounted for 38% of cancer deaths amongst women. Evidence from multiple large prospective cohort studies has confirmed that total fat intake probably has no effect on breast cancer risk [9, 42, 43]. In non-Mediterranean Western countries, where monounsaturated fats in the diet mostly come from meat and dairy products and not from olive oil, no significant effects of diets rich in monounsaturated fat on breast cancer incidence have been observed [9, 42, 43]. Previous publications from the EPIC study (European Prospective Investigation into Cancer and Nutrition), to which the MDC cohort contributes, have indeed shown that the main source of monounsaturated fats in this population was margarines and animal products [44].

Some facts have to be taken into account when interpreting our findings. First, it has been shown in this cohort that fat-rich diets tended to be poor in vegetables and fruit [33]. Inverse associations between vegetable and fruit intake and breast cancer have been reported repeatedly [43]. Secondly, the fact that trans fatty acids were not recorded separately in our study might have confounded our observed effect of monounsaturated fat intake. Some studies on trans fatty acids and breast cancer have shown increased breast cancer risk with higher levels of trans fatty acids in adipose tissue [45]. However, studies on the effect of trans fatty acids on breast cancer are sparse, and results have not been unanimous [45, 46].

Limitations

The participation rate in the MDC Study was quite low (approximately 40%). Hence, the risk of selection bias must be considered when interpreting our results. The diet in MDC was only evaluated at one point in time, and the number of people changing their diet during the follow-up period is unknown. Confounding effects from dietary factors not adjusted for in this study, for example diet composition and nutritional value, cannot be ruled out.

The follow-up time in our study was relatively short, with mean age at death being 66.5 years and thus only early mortality being observed. Other factors might be of greater importance for death later in life.

Pooling all cancer deaths together might confound our results, as dietary fats have been shown to effect different types of cancers in different ways [9]. The same might be applicable to cardiovascular mortality, as dietary fats do not seem to effect cerebrovascular disease incidence in the same way as cardiovascular disease and mortality [4, 8, 36–41].

Factors such as vitality, biological ageing and subclinical disease can affect appetite, energy intake, palatability and dietary habits, as well as disease development and mortality. As such, they can produce confounding effects in large-scale epidemiological studies like the current study. These limitations have to be taken into account when interpreting the results.

Underreporting of total energy and fat intake when using self-reported dietary assessment methods is considerable [47, 48]. Mattisson et al. [49] showed in a subsample of the MDC cohort that 41.6% of the subjects would be classified as under-reporters. Various adjustment models, which take biomarker measurements such as doubly labelled water and urinary nitrogen into account, have been created to try and further elucidate the effect of misreporting [47, 50]. Even though such models were not directly applied in the MDC Study, the dietary assessment instruments used in MDC were validated against weighed food records and urinary nitrogen measurements, showing correlation with r-values in the range of 0.50–0.60 for total energy intake and energy-adjusted fat and protein intake, which is comparable with the highest observations in similar validation studies [20]. Also, by stratifying analysis by sex and adjusting for confounding by body mass index, smoking, physical activity, and socio-economic factors, we partly compensate for differences between individuals in factors with impact on diet reporting, thus further limiting the biasing effect on our results.

Future implications

For the last 50 years, internationally used dietary guidelines recommending low-fat diets, have had great impact on both the lives of ordinary healthy free-living people and the food industry as a whole [6–8]. Whilst total energy intake over the last decades has increased to some extent in the Western world, the percentage of total daily energy coming from total fat and saturated fats especially has decreased, with intake of carbohydrates, mostly from refined sugars, increasing [4, 51]. Life expectancy has increased steadily during the same time, with incidence and death rates from heart disease and stroke diminishing dramatically [6, 52]. The changing nutritional- and lifestyle habits have, however, bred new problems with an overwhelming rise in the incidence of obesity and diabetes [53]. In the light of these changing trends, recent dietary and lifestyle advice from health authorities and professionals have, on the individual and population-based level, increasingly stressed a balanced energy intake and regular physical activity to avoid obesity, and to eat a diet rich in fruits, vegetables and fibre, including calories from all energy sources [54]. Our results support this trend, and not the change towards low-fat diets. However, studies with long-term follow-up and hard endpoints are still needed for further understanding the effect of diet on disease development and prevention. We also need a better understanding of the biological mechanisms behind our food choices, with the subjects’ vitality possibly being an important determinant of high energy and high fat intake [23].

Conclusions

With the exception of cancer mortality for women in the highest quartile of relative fat intake, individuals receiving more than 30% of their total daily energy from fat and more than 10% from saturated fat, did not have increased mortality. Beneficial effects of relatively high intakes of unsaturated fats were not uniform. With our results added to the pool of evidence from large-scale prospective cohort studies on dietary fat, disease and mortality, traditional dietary guidelines concerning fat intake are thus generally not strongly supported.

Conflict of interest statement

No conflict of interest was declared.

Acknowledgements

This project was supported by grants from The Swedish Scientific Council, The Swedish Cancer Foundation, Anna Jönssons Memorial Fund, the Swedish Heart and Lung Foundation, The European Commission, and The Region of Skåne, Sweden.

Ancillary