AJCN. First published ahead of print January 7, 2015 as doi: 10.3945/ajcn.114.094854.

Adherence to cancer prevention guidelines and cancer incidence, cancer mortality, and total mortality: a prospective cohort study1–4 Geoffrey C Kabat, Charles E Matthews, Victor Kamensky, Albert R Hollenbeck, and Thomas E Rohan ABSTRACT Background: Several health agencies have issued guidelines promoting behaviors to reduce chronic disease risk; however, little is known about the impact of such guidelines, particularly on cancer incidence. Objective: The objective was to determine whether greater adherence to the American Cancer Society (ACS) cancer prevention guidelines is associated with a reduction in cancer incidence, cancer mortality, and total mortality. Design: The NIH-AARP Diet and Health Study, a prospective cohort study of 566,401 adults aged 50–71 y at recruitment in 1995–1996, was followed for a median of 10.5 y for cancer incidence, 12.6 y for cancer mortality, and 13.6 y for total mortality. Participants who reported a history of cancer or who had missing data were excluded, yielding 476,396 subjects for analysis. We constructed a 5-level score measuring adherence to ACS guidelines, which included baseline body mass index, physical activity, alcohol intake, and several aspects of diet. Cox proportional hazards models were used to compute HRs and 95% CIs for the association of the adherence score with cancer incidence, cancer mortality, and total mortality. All analyses included fine adjustment for cigarette smoking. Results: Among 476,396 participants, 73,784 incident first cancers, 16,193 cancer deaths, and 81,433 deaths from all causes were identified in the cohort. Adherence to ACS guidelines was associated with reduced risk of all cancers combined: HRs (95% CIs) for the highest compared with the lowest level of adherence were 0.90 (0.87, 0.93) in men and 0.81 (0.77, 0.84) in women. Fourteen of 25 specific cancer sites showed a reduction in risk associated with increased adherence. Adherence was also associated with reduced cancer mortality: HRs (95% CIs) were 0.75 (0.70, 0.80) in men and 0.76 (0.70, 0.83) in women, and reduced all-cause mortality: HRs (95% CIs) were 0.74 (0.72, 0.76) in men and 0.67 (0.65, 0.70) in women. Conclusions: In both men and women, adherence to the ACS guidelines was associated with reductions in all-cancer incidence and the incidence of cancer at specific sites, as well as with reductions in cancer mortality and total mortality. These data suggest that, after accounting for cigarette smoking, adherence to a set of healthy behaviors may have considerable health benefits. This trial was registered at www.clinicaltrials.gov as NCT00340015. Am J Clin Nutr doi: 10.3945/ajcn.114.094854. Keywords body mass index, cancer incidence, cancer prevention guidelines, diet, physical activity

INTRODUCTION

In 1993, Breslow and Breslow (1) reported that participants in the Alameda County study who adhered to 7 health practices had

reduced mortality and disability. The 7 practices included avoiding the following risk factors: excessive alcohol intake, smoking, being obese, sleeping fewer or more than 7–8 h, having little physical activity, eating between meals, and not eating breakfast. After adjustment for the effects of age, sex, health status, and social support, rates of disability and death for those with good health practices were approximately half those with poor practices, and rates for the intermediate category were about two-thirds as great. Since that publication, several organizations, including the American Cancer Society (ACS),5 the World Cancer Research Fund/ American Institute for Cancer Research (WCRF/AICR), and the American Heart Association, have published guidelines designed to reduce the risk of cardiovascular disease, cancer, and other chronic diseases (2–6). A number of studies have examined the relation of adherence to the various guidelines to total mortality and cancer incidence (7–17). Only 2 studies (16, 17) have examined the relation between adherence to the guidelines and incidence of specific cancers. Although differing in many respects, these studies suggest that adherence to health guidelines may have a substantial impact on reducing the risk of chronic disease and mortality. Given that most previous studies were small and, with one exception (17), were not able to comprehensively examine sitespecific cancer incidence, we used information on 476,396 participants in the NIH-AARP Diet and Health Study to examine the association of adherence to the ACS guidelines, which primarily focus on cancer prevention, with all-cancer incidence and incidence of cancer at specific anatomic sites, cancer mortality, and total mortality. (Throughout, we use the term adherence to 1 From the Department of Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, NY (GCK, VK, and TER); the Division of Cancer Epidemiology and Genetics, National Cancer Institute, Bethesda, MD (CEM); and AARP, Washington, DC (ARH). 2 Supported by institutional funds from the Albert Einstein College of Medicine. 3 Supplemental Tables 1 and 2 are available from the “Supplemental data” link in the online posting of the article and from the same link in the online table of contents at http://ajcn.nutrition.org. 4 Address correspondence to G Kabat, Department of Epidemiology and Population Health, Albert Einstein College of Medicine, 1300 Morris Park Avenue, Bronx, NY 10461. E-mail: [email protected]. 5 Abbreviations used: ACS, American Cancer Society; BRFSS, Behavioral Risk Factor Surveillance System; cpd, cigarettes per day; EPIC, European Prospective Investigation into Cancer and Nutrition; MET, metabolic equivalent task; WCRF/AICR, World Cancer Research Fund/American Institute for Cancer Research. Received June 27, 2014. Accepted for publication November 26, 2014. doi: 10.3945/ajcn.114.094854.

Am J Clin Nutr doi: 10.3945/ajcn.114.094854. Printed in USA. Ó 2015 American Society for Nutrition

Copyright (C) 2015 by the American Society for Nutrition

1 of 12

2 of 12

KABAT ET AL.

mean the number of guidelines met by each participant.) The ACS guidelines suggest that limiting alcohol intake, making healthy dietary choices, maintaining a healthy body weight, and engaging in regular moderate to vigorous physical activity can reduce the risk of developing cancer, cancer mortality, and total mortality. Of the various recommendations regarding cancer prevention, the ACS guidelines were first issued .10 y ago and are the most influential in the United States. We hypothesized that increased adherence would be associated with improved outcomes.

SUBJECTS AND METHODS

Study population The NIH-AARP Diet and Health Study (www.clinicaltrials. gov; NCT00340015) is a large prospective cohort study of AARP members initiated in 1995–1996. The rationale for and design of the study have been described in detail previously (18). In brief, a baseline questionnaire was mailed to 3.5 million AARP (formerly, the American Association of Retired Persons) members between the ages of 50 and 71 y who resided in 6 states (California, Florida, Louisiana, New Jersey, North Carolina, or Pennsylvania) or in 2 metropolitan areas (Atlanta, GA, or Detroit, MI) with existing population-based cancer registries. In

total, 617,119 questionnaires were returned, and 566,398 were satisfactorily completed. The study was approved by the National Cancer Institute Special Studies Institutional Review Board, and return of the questionnaire signified consent. Information on lifestyle behaviors and covariates At baseline, participants completed a questionnaire that elicited information on demographics, including race-ethnicity and education, lifestyle factors such as current weight and height, strenuous physical activity at work and outside of work, smoking history, and diet (18). Women were asked about their use of postmenopausal hormones. Participants were asked whether they had ever smoked as many as 100 cigarettes during their lifetime; those who responded yes were asked whether they smoked currently or if they had quit and, if so, how long ago. Those currently smoking or who had quit smoking within the past year were considered current smokers. All smokers were asked how many cigarettes they usually smoked per day. Assessment of physical activity was based on the average frequency (never, rarely, 1–3 times/mo, 1–2 times/wk, 3–4 times/ wk, and $5 times/wk) during the past year that participants had engaged in activities of any type that lasted $20 min and caused either increases in breathing or heart rate or working up a sweat. To assess dietary intake, participants were asked about their usual

TABLE 1 Distribution of selected characteristics by health score in the NIH-AARP Diet and Health Study, by sex1 ACS score

Men n Age, y BMI, kg/m2 Alcoholic drinks/d Red meat intake, g/d Fruit and vegetable intake, cup equivalents/d Energy intake, kcal/d Vigorous physical activity ($3 times/wk), % Never smoked, % College graduate/postcollege education, % White race, % Self-reported health status fair/poor, % Women n Age, y BMI, kg/m2 Alcoholic drinks/d Red meat intake, g/d Fruit and vegetable intake, cup equivalents/d Energy intake, kcal/d Vigorous physical activity ($3 times/wk), % Never smoked, % College graduate/postcollege education, % White race, % Hormone therapy, % Nulliparous, % Age at menarche #12 y, % Self-reported health status fair/poor, %

0–3 (least adherent)

4–5

6

7

8–11 (most adherent)

33,062 61.2 6 5.42 31.4 6 5.5 3.1 6 5.4 112.3 6 71.0 3.1 6 1.9 2293 6 1035 3.8 20.5 34.2 93.4 21.8

75,174 61.7 6 5.4 28.3 6 4.3 1.5 6 3.4 96.2 6 65.1 3.4 6 2.0 2066 6 909 22.2 24.6 38.8 92.7 14.5

44,412 62.1 6 5.4 27.3 6 3.6 1.1 6 2.6 83.9 6 60.9 3.8 6 2.3 2011 6 858 47.9 28.3 42.7 92.2 11.1

43,318 62.4 6 5.3 26.7 6 3.3 0.9 6 1.9 73.4 6 58.0 4.2 6 2.4 1975 6 822 59.6 30.4 45.7 92.2 9.5

90,855 62.7 6 5.3 25.4 6 2.9 0.7 6 1.4 51.4 6 46.2 5.0 6 2.6 1913 6 745 84.2 35.8 53.9 92.8 6.7

20,407 61.1 6 5.4 33.5 6 7.7 0.8 6 2.3 76.2 6 54.3 2.9 6 1.9 1718 6 814 2.0 39.6 21.4 89.7 43.1 16.2 53.7 25.3

44,132 61.6 6 5.4 28.9 6 6.4 0.5 6 1.4 60.2 6 45.9 3.3 6 2.1 1609 6 739 13.0 40.9 24.8 89.1 49.3 14.8 50.3 17.2

27,607 61.8 6 5.4 27.2 6 5.0 0.5 6 1.3 49.6 6 41.1 3.8 6 2.4 1582 6 715 28.1 43.4 27.8 88.7 53.0 14.9 48.2 13.0

29,298 61.9 6 5.4 26.1 6 4.6 0.4 6 1.0 43.6 6 37.9 4.1 6 2.4 1566 6 688 41.5 45.4 30.3 88.9 55.3 15.0 47.8 10.7

68,131 62.1 6 5.4 24.2 6 3.2 0.3 6 0.7 32.9 6 29.8 4.8 6 2.6 1557 6 625 76.5 47.2 36.8 90.7 59.2 14.6 46.7 5.9

P-trend values were obtained from linear regression for continuous variables and from global x2 test for categorical variables. The P value for all trends over health score was ,0.0001. ACS, American Cancer Society. 2 Mean 6 SD (all such values). 1

3 of 12

IMPACT OF CANCER PREVENTION GUIDELINES

frequency of intake and portion size of 124 food items over the past 12 mo, according to 3 predefined categories of portion size and 10 predefined frequency categories ranging from “never” to “$6 times/d” for beverages and from “never” to “$2 times/d” for foods. The nutrient database for this food-frequency questionnaire was constructed by using the US Department of Agriculture 1994–96 Continuing Survey of Food Intake by Individuals (19). Alcohol intake was estimated as the number of drinks per day from beer, wine, and hard liquor combined (1 drink being a 12-oz beer, a 5-oz glass of wine, or a 1.5-oz shot of liquor; 1 oz = 28.35 g). Creation of the guidelines adherence score The ACS guidelines focus on 4 preventive factors: maintaining a healthy body weight, engaging in moderate to vigorous physical activity, healthy dietary choices, and limiting alcohol intake. Rather than scoring adherence to each guideline as either yes (meets guideline) or no (does not meet guideline), we used multiple levels of each variable to exploit the heterogeneity of exposure and better capture the degree of adherence. Initially, we created scores for each guideline component, with higher scores denoting greater adherence. The score for healthy body weight was based on 4 levels of BMI [in kg/m2; 18.5–24.9, 25.0–29.9, 30.0–34.9, and $35.0 (score = 3, 2, 1, and 0, respectively)]. Information on height and weight earlier in life was not available for the cohort; therefore, we were unable to assess maintenance of a healthy body weight over time, as defined in the ACS guidelines. The physical activity score was based on 4 levels of strenuous physical activity [never/rarely/1–3 times per month, 1–2 times per week, 3–4 times per week, and $5 times per week (score = 0, 1, 2, and 3)]. For the diet score, we created quartiles of fruit and vegetable intake (score = 0, 1, 2, and 3), quartiles of the ratio of whole grains to total grains (whole plus refined) (score = 0, 1, 2, and 3), and quartiles of intake of red plus processed meats (score = 3, 2, 1, and 0). The 3 dietary scores were combined for a possible total of 10 points overall (i.e., 0–9) and then collapsed into 4 categories (0–3, 4, 5, and 6–9) and assigned an overall dietary adherence value (score = 0, 1, 2, and 3). The alcohol score included

3 levels: score = 0 (men, $3 drinks/d; women, $2 drinks/d), score = 1 (nondrinkers), and score = 2 (men, 1–2 drinks/d; women, 1 drink/d). Finally, we created a single overall adherence score by combining the body weight (0–3), physical activity (0–3), overall diet (0–3), and alcohol (0–2) scores, yielding a total score ranging from 0 to 11 points. Because the numbers of subjects at the extremes were small, we created 5 categories of the total adherence score for analysis (0–3, 4–5, 6, 7, and 8–11). Follow-up and ascertainment of cases In the NIH-AARP study, vital status was determined by linkage of the cohort to the Social Security Administration Death Master File, the National Death Index Plus (for participants who could also be matched to the Death Master File), and cancer registry records. Participants’ responses to questionnaires and other mailings were also used to confirm vital status. Incident cases were identified from cancer registries in the original 6 states and 2 metropolitan areas, plus Texas and Arizona, states to which participants most commonly moved during follow-up. A validation study indicated that study procedures identified w90% of all incident cancers within the 8 registries (20). For cancer incidence, follow-up time extended from the date of receipt of the completed questionnaire (between 1995 and 1996) to the date of death, date of diagnosis of a first incident primary cancer (except nonmelanoma skin cancer), participant relocation out of the registry ascertainment area, or 31 December 2006, whichever came first. For cancer mortality, follow-up was complete through 2008. For total mortality, follow-up was complete through 2009. Mortality ascertainment had greater than 95% accuracy (21). Exclusions Among those with completed questionnaires, we excluded subjects who had questionnaires completed by proxy respondents (n = 15,760), had a history of cancer (n = 49,318), died or moved out of the study area before study entry (n = 1916), were identified as having cancer through death reports only (n = 2152), had zero

TABLE 2 Association of adherence to ACS guidelines and all-cancer incidence, cancer mortality, and total mortality, by sex, in the NIHAARP Diet and Health Study1 ACS score

All-cancer incidence Men Women Cancer mortality Men Women Total mortality Men Women

n

0–3

4–5

6

7

8–11

P-trend

50,762 23,022

1.002 1.00

0.96 (0.94, 0.99)3 0.92 (0.88, 0.96)

0.95 (0.92, 0.98) 0.88 (0.84, 0.93)

0.93 (0.90, 0.96) 0.82 (0.78, 0.86)

0.90 (0.87, 0.93) 0.81 (0.77, 0.84)

,0.0001 ,0.0001

10,599 5594

1.00 1.00

0.89 (0.84, 0.95) 0.89 (0.81, 0.97)

0.88 (0.82, 0.94) 0.83 (0.75, 0.92)

0.84 (0.78, 0.91) 0.83 (0.75, 0.92)

0.75 (0.70, 0.80) 0.76 (0.70, 0.83)

,0.0001 ,0.0001

55,814 25,619

1.00 1.00

0.89 (0.88, 0.92) 0.89 (0.85, 0.92)

0.86 (0.83, 0.89) 0.79 (0.75, 0.83)

0.81 (0.80, 0.85) 0.74 (0.71, 0.78)

0.74 (0.72, 0.76) 0.67 (0.65, 0.70)

,0.0001 ,0.0001

1 HRs (95% CIs) were obtained from Cox proportional hazards models. All HRs were adjusted for age (continuous), educational level (less than high school graduate, high school graduate, some college, college graduate/postgraduate), ethnicity (white, black, other), 14-level smoking variable (never; quit $10 y, 1–20 cpd; quit $10 y, 21–40 cpd; quit $10 y, .40 cpd; quit 5–9 y, 1–20 cpd; quit 5–9 y, 21–40 cpd; quit 5–9 y, .40 cpd; quit 1–4 y, 1–20 cpd; quit 1–4 y, 21–40 cpd; quit 1–4 y, .40 cpd; current, 1–20 cpd; current 21–40 cpd; current .40 cpd; missing), marital status (married, single, separated/divorced, widowed), and energy intake (continuous). ACS, American Cancer Society; cpd, cigarettes per day. 2 Reference category. 3 HR; 95% CI in parentheses (all such values).

4 of 12

KABAT ET AL.

follow-up time (n = 65), had a daily log-transformed calorie intake of .3 SDs beyond the mean (i.e., ,436 or .6575 calories/d) (n = 4019), or were missing information (mainly BMI) needed to construct the score (n = 16,792). Our final analytic cohort consisted of 476,396 subjects (286,821 men, 189,575 women). Statistical analysis Cox proportional hazards models, with attained age as the time scale, were used to estimate HRs and 95% CIs for the association of the extent of adherence to the prevention guidelines with cancer incidence, cancer mortality, and total mortality. Covariates included in the basic multivariable models were age (continuous), education (less than high school graduate, high school graduate, some college, college graduate/postgraduate), race (white, black, other), 14-level smoking variable [never; quit $10 y, 1–20 cigarettes per day (cpd); quit $10 y, 21–40 cpd; quit $10 y, .40 cpd; quit 5–9 y, 1–20 cpd; quit 5–9 y, 21–40 cpd; quit

5–9 y, .40 cpd; quit 1–4 y, 1–20 cpd; quit 1–4 y, 21–40 cpd; quit 1–4 y, .40 cpd; current, 1–20 cpd; current 21–40 cpd; current .40 cpd; missing], marital status (married, single, separated/divorced, widowed), and energy intake (continuous). In addition, we examined the association of adherence with risk of cancer at individual sites in men and women separately and, for sites with smaller numbers, in both sexes combined. In this analysis, for non–smoking-related sites, smoking was categorized as never smoked, quit $5 y, quit 1–4 y, and current smoker; for smoking-related sites, intensity of smoking (1–10, 11–20, 21–30, 31-40, .40 cpd) was also included. We carried out analyses stratified by smoking status (never, former, current smoker); for current and former smokers, intensity of smoking was included in models as a covariate (1–10, 11–20, 21–30, 31–40, .40 cpd). We tested for interactions of the ACS score with smoking status by including product terms (never, former, current smoker 3 score, categorized as 0–1, 2–3, and 4) and referring 23 the

FIGURE 1 HRs and 95% CIs for the association of level of adherence to ACS guidelines and cancer incidence by smoking status and by sex. HRs were adjusted for age, education, race, detailed smoking exposure, marital status, and energy intake. ACS, American Cancer Society.

IMPACT OF CANCER PREVENTION GUIDELINES

absolute difference in the log-likelihoods of models with and without the interaction terms to the x2 distribution with the appropriate df. The relative importance of the 4 component scores and smoking was assessed by computing adjusted HRs for each score level, relative to the lowest, with all 3 overall outcomes. The HR for each component score was adjusted for the 3 other component scores, smoking, and other covariates. In addition, to determine how alternative categorizations might affect the results, we carried out 2 alternative analyses. First, because recommendations can be viewed as dichotomous (meets the guideline or does not), we dichotomized each of the 4 components and created a score from 0 to 4, representing the sum of the number of recommendations met by each participant. This 5-level categorical variable was entered into models with covariates. Second, we examined the association of the original 12-point score (before it was collapsed into a 5-point score).

5 of 12

We conducted 2 sensitivity analyses: 1) excluding the first 3 y of follow-up and 2) excluding those in poor health at baseline (i.e., those whose self-reported general health was “fair,” “poor,” or unknown) and retaining those whose reported health was “excellent,” “very good,” or “good.” We computed population attributable risks to estimate the number of cancers and deaths that would have been prevented if the entire study population had adhered to the highest level of the ACS guidelines, under the assumption that the associations observed in this study are causal (22). Tests for trend across categorical variables were conducted by assigning the median value of each category and modeling this variable as a continuous variable. All analyses were done with SAS version 9.1 (SAS Institute). All statistical tests were 2-sided, and P values were assessed at a level of 0.05 for statistical significance for main effects and ,0.1 for tests of interaction.

FIGURE 2 HRs and 95% CIs for the association of level of adherence to ACS guidelines and cancer mortality by smoking status and by sex. HRs were adjusted for age, education, race, detailed smoking exposure, marital status, and energy intake. ACS, American Cancer Society.

6 of 12

KABAT ET AL.

We tested the proportional hazards assumption by using PROC LIFETEST (SAS Institute). RESULTS

A total of 73,784 incident first cancers were identified (50,762 men, 23,022 women) over a median follow-up of 10.5 y. In addition, 16,193 cancer deaths (10,599 men, 5,594 women) and 81,433 total deaths (55,814 men, 25,619 women) were ascertained over a median follow-up of 12.6 y and 13.6 y for cancer mortality and total mortality, respectively. In both men and women, extent of adherence to the ACS guidelines was inversely associated with BMI and meat intake and positively associated with physical activity, fruit and vegetable intake, proportion of never smokers, level of education, and self-described health at baseline (Table 1). Among women, age at menarche and ever use of hormone therapy

were positively associated with ACS score, whereas mean parity did not vary by score. Extent of adherence was associated with a modest reduction in all-cancer incidence, with HRs (95% CIs) for the highest compared with the lowest score of 0.90 (0.87, 0.93) in men and 0.81 (0.77, 0.84) in women (Table 2). Significant reductions were seen for the highest score category in all groups defined by smoking status, except female current smokers; however, there was little evidence of a declining trend over increasing score levels (Figure 1). There was a significant interaction between score and smoking status in men, which was driven by a greater reduction in incidence among men, with the highest level of adherence among current smokers. Adherence was associated with reduced total cancer mortality in a graded fashion (Table 2): HRs (95% CIs) for the highest level of adherence relative to the lowest were 0.75 (0.70, 0.80) in men and 0.76 (0.70, 0.83) in women. However, the pattern of the

FIGURE 3 HRs and 95% CIs for the association of level of adherence to ACS guidelines and total mortality by smoking status and by sex. HRs were adjusted for age, education, race, detailed smoking exposure, marital status, and energy intake. ACS, American Cancer Society.

7 of 12

IMPACT OF CANCER PREVENTION GUIDELINES TABLE 3 Association of ACS guidelines score with risk of incident cancer at specific sites, by sex, in the NIH-AARP Diet and Health Study1 ACS score Cancer site/type Oral Men Women Esophagus Men Women Stomach Men Women Small intestine Both sexes combined Colon Men Women Rectum Men Women Liver Men Women Gallbladder Both sexes combined Pancreas Men Women Breast Women Endometrium Women Ovary Women Cervix Women Bladder Men Women Kidney Men Women Prostate Men Larynx Both sexes combined Lung Men Women Melanoma Men Women Thyroid Men Women Brain Both sexes combined Hodgkin lymphoma Both sexes combined NHL Men Women

No. of cases2

0–3

4–5

6

7

8–11

P-trend

862 292

1.00 (ref) 1.00 (ref)

0.80 (0.66, 0.96)3 1.06 (0.73, 1.54)

0.82 (0.65, 1.03) 0.92 (0.60, 1.40)

0.77 (0.60, 0.98) 0.93 (0.61, 1.42)

0.79 (0.64, 0.97) 0.71 (0.48, 1.06)

0.06 0.03

685 95

1.00 (ref) 1.00 (ref)

0.78 (0.63, 0.96) 1.11 (0.60, 2.08)

0.80 (0.62, 1.02) 0.75 (0.36, 1.58)

0.73 (0.56, 0.95) 0.59 (0.26, 1.30)

0.59 (0.46, 0.76) 0.59 (0.29, 1.17)

0.001 0.02

700 200

1.00 (ref) 1.00 (ref)

0.81 (0.65, 1.01) 1.34 (0.79, 2.29)

0.68 (0.52, 0.88) 1.22 (0.68, 2.18)

0.84 (0.45, 0.79) 1.20 (0.67, 2.15)

0.62 (0.49, 0.79) 1.21 (0.71, 2.05)

0.0007 0.93

228

1.00 (ref)

0.62 (0.42, 0.91)

0.55 (0.35, 0.87)

0.55 (0.38, 0.86)

0.53 (0.36, 0.79)

0.008

2844 1287

1.00 (ref) 1.00 (ref)

0.80 (0.72, 0.89) 0.81 (0.67, 0.97)

0.67 (0.59, 0.76) 0.70 (0.57, 0.86)

0.65 (0.57, 0.74) 0.75 (0.61, 0.91)

0.52 (0.47, 0.59) 0.65 (0.54, 0.78)

,0.0001 ,0.0001

1417 582

1.00 (ref) 1.00 (ref)

0.79 (0.68, 0.92) 0.80 (0.61, 1.04)

0.82 (0.69, 0.98) 0.78 (0.58, 1.05)

0.77 (0.64, 0.93) 0.59 (0.43, 0.81)

0.60 (0.51, 0.72) 0.64 (0.49, 0.83)

,0.0001 0.0003

442 121

1.00 (ref) 1.00 (ref)

0.74 (0.58, 0.96) 0.83 (0.45, 1.56)

0.57 (0.41, 0.78) 1.00 (0.52, 1.95)

0.53 (0.37, 0.74) 1.06 (0.55, 2.04)

0.52 (0.39, 0.70) 0.77 (0.42, 1.43)

,0.0001 0.61

115

1.00 (ref)

0.53 (0.31, 0.89)

0.47 (0.25, 0.87)

0.52 (0.29, 0.96)

0.35 (0.20, 0.62)

0.002

1094 587

1.00 (ref) 1.00 (ref)

0.99 (0.83, 1.18) 1.01 (0.75, 1.37)

0.89 (0.72, 1.10) 0.99 (0.72, 1.38)

0.79 (0.63, 0.99) 1.22 (0.89, 1.67)

0.80 (0.65, 0.97) 1.04 (0.78, 1.40)

0.002 0.50

9072

1.00 (ref)

0.94 (0.88, 1.01)

0.91 (0.84, 0.99)

0.82 (0.76, 0.89)

0.81 (0.76, 0.87)

,0.0001

1518

1.00 (ref)

0.71 (0.61, 0.83)

0.62 (0.52, 0.73)

0.48 (0.40, 0.57)

0.40 (0.34, 0.46)

,0.0001

702

1.00 (ref)

1.02 (0.78, 1.33)

0.93 (0.69, 1.25)

0.86 (0.64, 1.16)

0.95 (0.73, 1.23)

0.40

143

1.00 (ref)

1.21 (0.69, 2.11)

0.89 (0.46, 1.71)

1.07 (0.57, 2.01)

0.98 (0.55, 1.75)

0.62

3607 605

1.00 (ref) 1.00 (ref)

0.86 (0.78, 0.94) 1.02 (0.77, 1.36)

0.85 (0.76, 0.95) 1.17 (0.87, 1.59)

0.88 (0.78, 0.99) 0.98 (0.72, 1.34)

0.81 (0.73, 0.90) 0.97 (0.73, 1.28)

0.002 0.57

1519 523

1.00 (ref) 1.00 (ref)

0.84 (0.73, 0.98) 0.76 (0.58, 0.99)

0.88 (0.75, 1.05) 0.69 (0.51, 0.94)

0.78 (0.65, 0.94) 0.53 (0.38, 0.73)

0.62 (0.52, 0.73) 0.54 (0.41, 0.71)

,0.0001 ,0.0001

22,931

1.00 (ref)

1.04 (1.00, 1.09)

1.07 (1.02, 1.12)

1.06 (1.01, 1.12)

1.04 (1.00, 1.09)

0.13

620

1.00 (ref)

0.84 (0.68, 1.05)

0.69 (0.52, 0.91)

0.70 (0.52, 0.94)

0.82 (0.64, 1.05)

0.06

5960 3387

1.00 (ref) 1.00 (ref)

0.99 (0.92, 1.07) 0.98 (0.88, 1.09)

0.96 (0.88, 1.05) 0.89 (0.78, 1.00)

0.91 (0.83, 1.00) 0.87 (0.77, 0.99)

0.85 (0.78, 0.93) 0.94 (0.84, 1.05)

,0.0001 0.12

3538 1210

1.00 (ref) 1.00 (ref)

1.06 (0.94, 1.18) 1.10 (0.88, 1.36)

1.09 (0.96, 1.24) 1.02 (0.80, 1.30)

1.03 (0.90, 1.17) 1.15 (0.92, 1.45)

1.19 (1.07, 1.33) 1.21 (0.98, 1.49)

0.002 0.04

248 317

1.00 (ref) 1.00 (ref)

0.91 (0.60, 1.39) 0.82 (0.55, 1.22)

1.15 (0.73, 1.81) 1.14 (0.76, 1.70)

1.27 (0.81, 2.00) 0.97 (0.64, 1.47)

1.15 (0.75, 1.75) 0.82 (0.57, 1.20)

0.18 0.53

732

1.00 (ref)

1.21 (0.94, 1.56)

1.23 (0.93, 1.63)

1.07 (0.79, 1.43)

1.32 (1.02, 1.71)

0.12

113

1.00 (ref)

0.45 (0.26, 0.78)

0.55 (0.30, 1.03)

0.67 (0.37, 1.22)

0.57 (0.33, 0.98)

0.32

1837 980

1.00 (ref) 1.00 (ref)

1.13 (0.97, 1.31) 0.91 (0.73, 1.14)

1.08 (0.91, 1.28) 0.98 (0.77, 1.24)

0.99 (0.83, 1.17) 0.95 (0.75, 1.21)

1.03 (0.88, 1.20) 0.91 (0.73, 1.12)

0.41 0.56 (Continued)

8 of 12

KABAT ET AL.

TABLE 3 (Continued ) ACS score Cancer site/type Multiple myeloma Men Women Leukemia Men Women

No. of cases2

0–3

4–5

6

7

8–11

P-trend

602 275

1.00 (ref) 1.00 (ref)

0.86 (0.67, 1.11) 0.88 (0.58, 1.33)

0.93 (0.70, 1.24) 0.99 (0.64, 1.53)

0.97 (0.73, 1.29) 0.74 (0.47, 1.18)

0.78 (0.60, 1.02) 0.74 (0.49, 1.10)

0.21 0.08

1026 359

1.00 (ref) 1.00 (ref)

1.03 (0.85, 1.24) 0.78 (0.55, 1.10)

0.97 (0.78, 1.20) 0.87 (0.60, 1.26)

0.84 (0.67, 1.06) 0.76 (0.52, 1.11)

0.80 (0.65, 0.98) 0.69 (0.49, 0.98)

0.002 0.07

1 HRs (95% CIs) were obtained from Cox proportional hazards models. All sites adjusted for age (continuous), educational level (less than high school graduate, high school graduate, some college, college graduate/postgraduate), ethnicity (white, black, other), smoking status (never, former, current smoker), marital status (married, single, separated/divorced, widowed), and energy intake (continuous). Breast, ovarian, and endometrial cancers also were adjusted for menopausal status, age at menarche, age at first birth, parity, and hormone therapy use (ever/never). Breast cancer also was adjusted for family history of breast cancer in a first-degree relative and mammographic screening (never, in past 5 y, .5 y ago). Colon and rectal cancer also were adjusted for family history and colonoscopy screening. Prostate cancer also was adjusted for family history and rectal examination and prostate-specific antigen test. Melanoma also was adjusted for ultraviolet exposure by using estimated ground-level erythemal dose for the period 1978–1993 by linking Total Ozone Mapping Spectrometer data (from the National Aeronautics and Space Administration; http://toms.gsfc.nasa.gov) to the latitude and longitude of the census tract of residence at baseline for all cohort members. For cancers of the oropharynx, esophagus, colon, rectum, liver, pancreas, bladder, kidney, larynx, and lung, as well as all-cancer incidence, smoking intensity (1–10, 11–20, 21–30, 31–40, .40 cigarettes per day) was included as a covariate. ACS, American Cancer Society; NHL, nonHodgkin lymphoma; ref, reference. 2 The number of noncases in each analysis was obtained by subtracting the number of male cases at each site from 286,821 men and subtracting the number of female cases at each site from 189,575 women. 3 HR; 95% CI in parentheses (all such values).

associations by strata of smoking status was inconsistent (Figure 2). There was no association in never smokers of either sex or among female current smokers. Only among female former smokers was there a clear and significant decreasing trend with increasing score. There was a statistically significant interaction between score and smoking status in men but not in women. Risk of all-cause mortality decreased monotonically with increasing score in both men and women: the HRs (95% CIs) for the highest compared with the lowest category were 0.74 (0.72, 0.76) in men and 0.67 (0.65, 0.70) in women (Table 2). A graded inverse association of score with total mortality was seen in never smokers, former smokers, and current smokers of both sexes (Figure 3). Reductions in the HR were largest in never smokers, intermediate in former smokers, and smallest in current smokers (Figure 3). In both men and women, there was a significant interaction between smoking status and score. High compared with low score was associated with a significantly reduced risk of cancer in 16 of 25 anatomic sites examined in at least one sex (Table 3). (We considered the association with a cancer site statistically significant if both the highest and the lowest quintiles of the score and the P value for the linear trend over quintiles were statistically significant.) Significant inverse associations were seen for esophagus (men), stomach (men), small intestine (both sexes combined), colon (both sexes), rectum (both sexes), liver (men), gallbladder (both sexes combined), pancreas (men), breast (women), endometrium (women), bladder (men), kidney (both sexes), lung (men), and leukemia (men). The largest reductions (associated with the highest adherence category) were seen for gallbladder (65%), endometrium (60%, women), liver (48%, men), and colon (48%, men). Six of the sites for which we computed sex-specific HRs (oral cavity, esophagus, colon, rectum, kidney, and leukemia) showed a significant inverse association with score in both men and women (Table 3; Figure 4). The only statistically significant

positive association was that for the highest score level with melanoma in men; in women, the HR for the highest score level was elevated but not statistically significant (1.21; 95% CI: 0.98, 1.49). The linear trend was statistically significant in both sexes. The relative magnitude of the associations for the 4 components of the adherence score varied by outcome and by sex (Table 4). For cancer incidence, BMI followed by diet were strongest in women, whereas in men, alcohol followed by diet were the strongest factors. For cancer mortality, diet followed by BMI, followed by physical activity were strongest in both sexes. For total mortality, BMI followed by physical activity were the strongest factors in both sexes. Smoking showed the strongest association with cancer mortality, followed by total mortality, followed by all-cancer incidence. For all 3 outcomes in both sexes, smoking was a stronger risk factor than any of the component scores. In analyses based on alternative approaches to the construction of adherence scores, entailing the use of the sum of binary adherence scores and the 12-point adherence score, the patterns of the results were very similar to those of the main analysis and showed monotonic decreasing trends for the 3 outcomes in both sexes (Supplemental Tables 1 and 2). In the 2 sensitivity analyses, excluding the first 3 y of follow-up and those in poorer health (“fair,” “poor,” or “missing,” representing 14% of participants), the results for overall cancer incidence, cancer mortality, and total mortality were not materially altered. For this latter analysis, health status was not included as a covariate (data not shown). The formal test for nonproportional hazards was significant due to the large sample size, but the log-log survival plot did not indicate any marked deviations from normality. Population attributable risks for cancer incidence, cancer mortality, and total mortality were 2.6% and 6.6%, 10.3% and 11.2%, and 14.5% and 21.0% in men and women, respectively.

IMPACT OF CANCER PREVENTION GUIDELINES

9 of 12

FIGURE 4 Forest plot showing HRs and 95% CIs for the association of adherence to ACS guidelines (highest compared with lowest of 5 categories) and risk of cancer at 25 anatomic sites obtained from Cox proportional hazards models. In addition, the P value for the linear trend over levels of the score is presented. Different sites were adjusted for different covariates (see below). Each cancer site is represented by a square and a horizontal line denoting the point estimate and the 95% CI, respectively. The size of the square represents the statistical weight of the outcome, reflecting the number of cases. The diamonds represent the HRs for all cancers combined in males and females. All sites were adjusted for age (continuous), education, race, smoking status and intensity, marital status, and energy intake. Breast, ovarian, and endometrial cancers were also adjusted for menopausal status, age at menarche, age at first birth, parity, and hormone therapy use (ever/never). Breast cancer was also adjusted for family history of breast cancer in a first-degree relative and mammographic

10 of 12

KABAT ET AL.

DISCUSSION

To our knowledge, this is the largest study to date to examine the relation of adherence to a set of health-related guidelines to cancer risk and mortality. Compliance with the ACS guidelines was associated with a modest, albeit statistically significant, reduction in overall cancer incidence but with a substantially reduced risk of cancer in 14 of 25 sites examined. Reductions in risk ranged from 15% (lung cancer in men) to 65% (gallbladder, men and women combined), and many of these associations were consistent in men and women for sites common to both sexes. Reductions in risk were seen for cancer mortality. Strong, monotonic inverse associations were observed with total mortality in both men and women, and this pattern was present in never, former, and current smokers. Secondary analyses indicated that the pattern of results was similar when alternative categorizations of the score were used. As in our study, previous studies have reported stronger associations of adherence to healthy behaviors with total mortality compared with cancer mortality and cancer incidence (12, 13, 16, 17). For example, McCullough et al. (12) observed an RR of 0.58 for the highest compared with the lowest level of adherence to the ACS guidelines for all-cause mortality in both sexes, whereas the RRs for cancer mortality were 0.70 and 0.76 in men and women, respectively. In analyses of the European Prospective Investigation into Cancer and Nutrition (EPIC) cohort, HRs for the highest compared with the lowest level of adherence to the WCRF/AICR guidelines with total mortality, cancer mortality, and all-cancer incidence in men and women, respectively, were 0.71 and 0.62, 0.86 and 0.76, and 0.84 and 0.81 (13, 17). In agreement with our results, 2 previous studies have reported significant inverse associations of adherence score with risk of cancer at a number of major anatomic sites [colon, lung, breast, and endometrium (16); colorectal, stomach, breast, endometrium, lung, kidney, upper aerodigestive tract, liver, and esophagus (17)]. Our results concerning pancreatic cancer are consistent with those of Jiao et al. (23), who assessed the association of a different healthy lifestyle score with risk of pancreatic cancer in the NIH-AARP study; after adjusting for smoking, they too found a significant reduction in pancreatic cancer incidence in men but not in women. Only melanoma incidence was positively associated with a higher score; however, the trend over levels of the score in both men and women was weak. This association could be due to chance or to confounding. We were not able to adjust for risk factors for melanoma (skin color, eye color, hair color) because these data were not available. To our knowledge, no previous study has examined the association between adherence to health guidelines and risk of melanoma. Assessment of the contribution of components of the overall score showed that their relative importance varied by outcome and by sex. Alcohol intake and diet score were inversely associated with all-cancer incidence in men, whereas BMI and physical activity were not significant. In contrast, BMI was the

strongest predictor in women, followed by alcohol, diet, and physical activity. Similar male/female differences in predictors of cancer incidence were reported in EPIC (17). In our study, diet score and then BMI were the strongest predictors of reduced cancer mortality in both men and women. In EPIC (13), the plant foods component was the strongest predictor of cancer death in both sexes. BMI and physical activity were the strongest predictors of total mortality in both men and women in the present study. In EPIC (13), BMI, followed by intake of plant foods, followed by physical activity were the strongest predictors of total mortality. In the ACS analysis (12), maintenance of a healthy BMI was the component most strongly associated with lower mortality. Studies examining the association of compliance with health guidelines and major health outcomes have differed in the population studied, the age range of the population, the specific items included in the questionnaires, the specifics of the set of guidelines, and the construction of a score. In view of this diversity, it is noteworthy that these studies show consistent associations and that the strongest associations tend to be seen for all-cause mortality. The value of these studies is that they attempt to assess the overall pattern of health-related behaviors rather than focusing on individual factors, such as fiber or saturated fat. In addition to including specific items that have shown consistent associations with health outcomes, a high score is likely to reflect greater “health consciousness” in general; thus, use of a score may classify people according to other risk factors that were not measured (e.g., abdominal obesity, job-related or leisure-time physical activity, sleep hours) and that may be related to morbidity/ mortality. It is known that many health-related behaviors and characteristics tend to be correlated (24). The stronger inverse associations observed with all-cause mortality and cancer mortality compared with cancer incidence suggest several possible explanations. It is possible that individuals with higher scores may have access to higher-quality health care. Also, adherence to healthy behaviors may be associated with a better prognosis following diagnosis of a chronic disease. One study found that among female cancer survivors, adherence to the WCRF/AICR guidelines was associated with lower all-cause mortality (25). The modest overall inverse association of score with cancer incidence resulted from its being a weighted average of many large inverse associations with specific cancer sites as well as null associations with others. The larger overall reduction in cancer incidence in women (HR: 0.81) compared with men (HR: 0.90) was driven by inverse associations for breast and endometrial cancer, which accounted for 46% of incident cancers in women. In conformity with previous studies (8–10), the association of smoking with the 3 outcomes was stronger than the associations for any of the score components individually. For this reason, careful adjustment for smoking and stratification by smoking status are crucial to assess the associations of other health behaviors. We found robust inverse associations of the score with total

screening (never, in past 5 y, or .5 y ago). Colon and rectal cancer were also adjusted for family history and colonoscopy screening. Prostate cancer also was adjusted for family history and rectal examination and prostate-specific antigen test. Melanoma was also adjusted for UV exposure by using estimated groundlevel erythemal dose for the period 1978–1993 by linking Total Ozone Mapping Spectrometer data (from the National Aeronautics and Space Administration; http://toms.gsfc.nasa.gov) to the latitude and longitude of the census tract of residence at baseline for all cohort members. For cancers of the oropharynx, esophagus, larynx, lung, bladder, kidney, pancreas, and liver, as well as all-cancer incidence, smoking intensity (1–10, 11–20, 21–30, 31–40, or .40 cigarettes/d) was included as a covariate. ACS, American Cancer Society; INTEST, intestine; LYMPH, lymphoma; NHL, non-Hodgkin lymphoma.

11 of 12

IMPACT OF CANCER PREVENTION GUIDELINES

TABLE 4 Association of component scores with cancer incidence, cancer mortality, and total mortality, by sex, in the NIH-AARP Diet and Health Study1 Cancer incidence, HR (95% CI) Score components2

Cancer mortality, HR (95% CI)

Total mortality, HR (95% CI)

Men

Women

Men

Women

Men

Women

1.00 (ref) 0.95 (0.93, 0.98) 0.96 (0.94, 0.99) 0.94 (0.92, 0.96) ,0.0001

1.00 (ref) 0.97 (0.93, 1.01) 0.98 (0.94, 1.02) 0.93 (0.90, 0.96) ,0.0001

1.00 (ref) 0.90 (0.86, 0.96) 0.86 (0.81, 0.91) 0.83 (0.78, 0.87) ,0.0001

1.00 (ref) 0.91 (0.83, 0.99) 0.90 (0.83, 0.98) 0.88 (0.82, 0.94) 0.0003

1.00 (ref) 0.97 (0.95, 0.99) 0.94 (0.91, 0.96) 0.91 (0.89, 0.93) ,0.0001

1.00 (ref) 0.92 (0.88, 0.96) 0.92 (0.88, 0.95) 0.86 (0.83, 0.89) ,0.0001

1.00 (ref) 1.06 (1.01, 1.11) 1.06 (1.01, 1.11) 1.03 (0.99, 1.08) ,0.0001

1.00 (ref) 0.92 (0.87, 0.97) 0.84 (0.80, 0.89) 0.81 (0.77, 0.85) ,0.0001

1.00 (ref) 0.97 (0.88, 1.07) 0.89 (0.82, 0.98) 0.88 (0.82, 0.97) 0.0005

1.00 (ref) 1.01 (0.90, 1.13) 0.90 (0.81, 1.00) 0.87 (0.79, 0.97) 0.0003

1.00 (ref) 0.74 (0.71, 0.77) 0.62 (0.60, 0.64) 0.64 (0.62, 0.67) ,0.0001

1.00 (ref) 0.73 (0.70, 0.77) 0.60 (0.60, 0.65) 0.61 (0.59, 0.64) ,0.0001

1.00 (ref) 0.99 (0.97, 1.02) 1.01 (0.98, 1.03) 0.98 (0.96, 1.01) ,0.0001

1.00 (ref) 0.98 (0.94, 1.01) 0.96 (0.92, 0.99) 0.96 (0.93, 1.00) ,0.0001

1.00 (ref) 0.98 (0.93, 1.03) 0.91 (0.86, 0.96) 0.90 (0.85, 0.95) ,0.0001

1.00 (ref) 0.94 (0.88, 1.02) 0.89 (0.83, 0.96) 0.94 (0.86, 1.02) 0.01

1.00 (ref) 0.82 (0.80, 0.84) 0.78 (0.76, 0.80) 0.75 (0.73, 0.77) ,0.0001

1.00 (ref) 0.81 (0.78, 0.84) 0.77 (0.74, 0.80) 0.77 (0.74, 0.80) 0.02

1.00 (ref) 0.90 (0.87, 0.92) 0.92 (0.90, 0.94) ,0.0001

1.00 (ref) 0.89 (0.85, 0.92) 0.89 (0.85, 0.92) 0.24

1.00 (ref) 1.02 (0.95, 1.08) 0.90 (0.85, 0.94) ,0.0001

1.00 (ref) 1.03 (0.94, 1.12) 0.90 (0.84, 0.98) 0.05

1.00 (ref) 1.20 (1.17, 1.23) 0.91 (0.89, 0.94) ,0.0001

1.00 (ref) 1.28 (1.23, 1.34) 0.92 (0.89, 0.96) ,0.0001

1.00 (ref) 1.07 (1.05, 1.09) 1.29 (1.23, 1.37) 1.46 (1.42, 1.51) 1.06 (1.01, 1.11)

1.00 (ref) 1.16 (1.12, 1.20) 1.33 (1.25, 1.42) 1.64 (1.58, 1.70) 1.23 (1.14, 1.32)

1.00 (ref) 1.57 (1.48, 1.66) 2.74 (2.50, 3.01) 3.62 (3.39, 3.86) 1.66 (1.48, 1.85)

1.00 (ref) 1.51 (1.41, 1.63) 2.26 (2.01, 2.56) 3.40 (3.16, 3.66) 1.63 (1.40, 1.89)

1.00 (ref) 1.38 (1.35, 1.41) 2.12 (2.03, 2.21) 2.46 (2.39, 2.53) 1.49 (1.42, 1.56)

1.00 (ref) 1.44 (1.39, 1.48) 2.13 (2.01, 2.25) 2.72 (2.62, 2.82) 1.59 (1.48, 1.70)

3

Diet 0 1 2 3 P-trend BMI, kg/m2 $35.0 30.0–34.9 25.0–29.9 18.5–24.9 P-trend Physical activity Never/rarely/1–33/mo 1–23/wk 3–43/wk $53/wk P-trend Alcohol4 Heavier Never Lighter P-trend Smoking Never Quit $5 y Quit 1–4 y Current Missing

1 HRs (95% CIs) were obtained from Cox proportional hazards models. HRs for each score component were adjusted for the other components and, in addition, were adjusted for age (continuous), educational level (less than high school graduate, high school graduate, some college, college graduate/ postgraduate), ethnicity (white, black, other), and marital status (married, single, separated/divorced, widowed). ref, reference. 2 Score components were diet, BMI, physical activity, and alcohol. Smoking was not a score component but was included as a covariate. 3 Four-level score based on sum of component scores for fruit and vegetable intake, ratio of whole-grain intake to total grain intake, and quartiles of red meat plus processed meat intake. 4 For men, #2 drinks/d; for women, #1 drink/d.

mortality after adjusting for smoking as well as in analyses stratified by smoking status. In addition, the reduction in total mortality among individuals with the highest compared with the lowest level of adherence was significantly greater among never smokers of both sexes (0.68 in men and 0.64 in women) compared with that in current smokers (0.81 in men and 0.75 in women). Although previous studies consistently found no difference in the association of score with outcome by smoking status (12–17), there were a much larger number of deaths in the current study. Among the strengths of the present study are its prospective design and its large size; inclusion of both sexes; the availability of information on a wide range of exposures, including detailed information on diet; and the completeness of follow-up. The large number of outcomes enabled us to examine associations by smoking status and to examine associations for a wide range of cancer sites, including the small intestine, gallbladder, and liver, which have not been examined previously. Our study also had limitations. Specifically, self-reports of physical activity, height and weight, and dietary and alcohol intake are susceptible to measurement error, and reported be-

havior at one point in time may not be representative of long-term behavior. Previous studies indicate that the stability of the various components of the score from adolescence into adulthood ranges from low to high, depending on the behavior (10); however, there may be greater stability at older ages. In any event, misclassification due to changes in behaviors, if random, would be expected to reduce the magnitude of the observed associations. Furthermore, these limitations also apply to previous studies. Another limitation is that we were not able to assess maintenance of a healthy BMI because data on height and weight in early adulthood were unavailable. Participants in the NIH-AARP cohort were aged 50–71 y at enrollment, and their habits may not be representative of those of the general population. Furthermore, the cohort contains only small percentages of minorities (4% black, 2% Hispanic). However, mean BMI for men and women in our study was in good agreement with data from the Behavioral Risk Factor Surveillance System (BRFSS), a cross-sectional telephone survey of a representative sample from each state’s noninstitutionalized civilian residents aged 18 y and older, which reported BMI based on self-reported height and

12 of 12

KABAT ET AL.

weight for the period 1999–2002 for ages 50–69 y (men: 27.4 in AARP and w27.7 in BRFSS; women: 27.0 in AARP and w27.3 in BRFSS) (26). In conclusion, the present study indicates that, after accounting for smoking, adherence to guidelines relating to body weight, physical activity, alcohol, and diet was associated with a modest reduction in cancer incidence overall but with significant reductions in a large number of cancer sites. Reductions in risk were also seen with cancer mortality and particularly with total mortality. Our findings suggest that adherence to healthy behaviors may reduce cancer incidence as well as mortality. The authors’ responsibilities were as follows—GCK and TER: designed the research project and wrote the manuscript; CEM and ARH: conducted the research; VK: provided essential materials; and GCK: performed the statistical analysis and had primary responsibility for the final content. All authors read and approved the final manuscript. None of the authors reported a conflict of interest relating to this study.

REFERENCES 1. Breslow L, Breslow N. Health practices and disability: some evidence from Alameda County. Prev Med 1993;22:86–95. 2. Byers T, Nestle M, McTiernan, Doyle C, Currie-Williams A, Gansler T, Thun M; American Cancer Society 2001 Nutrition and Physical Activity Guidelines Advisory Committee. American Cancer Society guidelines on nutrition and physical activity for cancer prevention: reducing the risk of cancer with healthy food choices and physical activity. CA Cancer J Clin 2002;52:92–119. 3. Kushi LH, Byers T, Doyle C, Bandera EV, McCullough M, McTiernan A, Gansler T, Andrews KS, Thun MJ; American Cancer Society 2006 Nutrition and Physical Activity Guidelines Advisory Committee. American Cancer Society guidelines on nutrition and physical activity for cancer prevention: reducing the risk of cancer with healthy food choices and nutritional activity. CA Cancer J Clin 2006;56:254–81. 4. Lichtenstein AH, Appel LJ, Brands M, Carnethon M, Daniels S, Franch HA, Franklin B, Kris-Etherton P, Harris WS, Howard B, et al. Diet and lifestyle recommendations revision 2006: a scientific statement from the American Heart Association Nutrition Committee. Circulation 2006;114:82–96. 5. World Cancer Research Fund/American Institute for Cancer Research. Food, nutrition, and the prevention of cancer: a global perspective. Washington, DC: AICR; 2007. 6. Kushi LH, Doyle C, McCullough M, Rock CL, Demark-Wahnefried W, Bandera EV, Gapstur S, Patel AV, Andrews K, Gansler T, et al. American Cancer Society guidelines on nutrition and physical activity for cancer prevention: reducing the risk of cancer with healthy food choices and physical activity. CA Cancer J Clin 2012;62:30–67. 7. Knoops KT, de Groot LC, Kromhout D, Perrin AE, Moreiras-Varela O, Menotti A, van Staveren WA. Mediterranean diet, lifestyle factors, and 10-year mortality in elderly European men and women. JAMA 2004; 292:1433–9. 8. van Dam RM, Li T, Spiegelman D, Franco OH, Hu FB. Combined impact of lifestyle factors on mortality: prospective cohort study of US women. BMJ 2008;337:a1440. 9. Khaw KT, Wareham N, Bingham S, Welch A, Luben R, Day N. Combined impact of health behaviours and mortality in men and women: the EPIC-Norfolk prospective population study. PLoS Med 2008;5:e12. 10. Kvaavik E, Batty GD, Ursin G, Huxley R, Gale CR. Influence of individual and combined health behaviors on total and cause-specific mortality in men and women: the United Kingdom Health and Lifestyle Survey. Arch Intern Med 2010;170:711–8.

11. Nechuta SJ, Shu XO, Li HL, Yang G, Xiang YB, Cai H, Chow WH, Ji B, Zhang X, Wen W, et al. Combined impact of lifestyle-related factors on total and cause-specific mortality among Chinese women: prospective cohort study. PLoS Med 2010;7:e1000339. 12. McCullough ML, Patel AV, Kushi LH, Patel R, Willett WC, Doyle C, Thun MJ, Gapstur SM. Following cancer prevention guidelines reduces risk of cancer, cardiovascular disease, and all-cause mortality. Cancer Epidemiol Biomarkers Prev 2011;20:1089–97. 13. Vergnaud AC, Romaguera D, Peeters PH, van Gils CH, Chan DS, Romieu I, Freisling H, Ferrari P, Clavel-Chapelon F, Fagherazzi G, et al. Adherence to the World Cancer Research Fund/American Institute for Cancer Research guidelines and risk of death in Europe: results from the European Prospective Investigation into Nutrition and Cancer cohort study. Am J Clin Nutr 2013;97:1107–20. 14. Cerhan JR, Potter JD, Gilmore JM, Janney CA, Kushi LH, Lazovich D, Anderson KE, Sellers TA, Folsom AR. Adherence to the AICR cancer prevention recommendations and subsequent morbidity and mortality in the Iowa Women’s Health Study cohort. Cancer Epidemiol Biomarkers Prev 2004;13:1114–20. 15. Ford ES, Bergmann MM, Kro¨ger J, Schienkiewitz A, Weikert C, Boeing H. Healthy living is the best revenge: findings from the European Prospective Investigation into Cancer and Nutrition–Potsdam Study. Arch Intern Med 2009;169:1355–62. 16. Harnack L, Nicodemus K, Jacobs DR, Folsom AR. An evaluation of the Dietary Guidelines for Americans in relation to cancer occurrence. Am J Clin Nutr 2002;76:889–96. 17. Romaguera D, Vergnaud AC, Peeters PH, van Gils CH, Chan DS, Ferrari P, Romieu I, Jenab M, Slimani N, Clavel-Chapelon F, et al. Is concordance with World Cancer Research Fund/American Institute for Cancer Research guidelines for cancer prevention related to subsequent risk of cancer? Results from the EPIC study. Am J Clin Nutr 2012;96: 150–63. 18. Schatzkin A, Subar AF, Thompson FE, Harlan LC, Tangrea J, Hollenbeck AR, Hurwitz PE, Coyle L, Schussler N, Michaud DS, et al. Design and serendipity in establishing a large cohort with wide dietary intake distributions. Am J Epidemiol 2001;154:1119–25. 19. Subar AF, Midthune D, Kulldorff M, Brown CC, Thompson FE, Kipnis V, Schatzkin A. Evaluation of alternative approaches to assign nutrient values to food groups in food frequency questionnaires. Am J Epidemiol 2000;152:279–86. 20. Michaud DS, Midthune D, Hermansen S, Leitzmann M, Harlan LC, Kipnis V, Schatzkin A. Comparison of cancer registry case ascertainment with SEER estimates and self-reporting in a subset of the NIHAARP diet and health study. J Registry Manage. 2005;32:70–5. 21. Hermansen SW, Leitzmann MF, Schatzkin A. The impact on National Death Index ascertainment of limiting submissions to Social Security Administration Death Master File matches in epidemiologic studies of mortality. Am J Epidemiol 2009;169:901–8. 22. Szklo M, Nieto FJ. Epidemiology: beyond the basics. Gaithersburg (MD): Aspen Publishers; 2000:104. 23. Jiao L, Mitrou PN, Reedy J, Graubard BI, Hollenbeck AR, Schatzkin A, Stolzenberg-Solomon R. A combined healthy lifestyle score and risk of pancreatic cancer—a large cohort study. Arch Intern Med 2009; 169:764–70. 24. Krick JP, Sobal J. Relationships between health protective behaviors. J Community Health 1990;15:19–34. 25. Inoue-Choi M, Robien K, Lazovich D. Adherence to the WCRF/AICR guidelines for cancer prevention is associated with lower mortality among older female cancer survivors. Cancer Epidemiol Biomarkers Prev 2013;22:792–802. 26. Ezzati M, Martin H, Skjold S, Vander Hoorn S, Murray CJ. Trends in national and state-level obesity in the USA after correction for self-report bias: analysis of health surveys. J R Soc Med 2006;99: 250–7.