NIH-AARP Diet and Health Study Reaches Milestone

Breaking Ground for a Large Prospective Study

Nearly 25 years ago, investigators from the National Cancer Institute (NCI) mailed over three million members of the AARP (formerly known as the American Association of Retired Persons) a food frequency questionnaire to examine the relationship between diet, lifestyle, and cancer risk among older Americans. With the response of over half a million people, the NCI launched the NIH-AARP Diet and Health Study, a cohort that was, and remains, the largest prospective in-depth study on diet and health.

AARP members residing in six states (California, Florida, Louisiana, New Jersey, North Carolina, and Pennsylvania) and two metropolitan areas (Atlanta, Georgia, and Detroit, Michigan) received a 16-page questionnaire designed to take a complete inventory of their diet and to gather information on other important lifestyle factors. Analyses resulting from these data have greatly improved public knowledge of lifestyle risk factors and cancer, such as the health benefits of eating a Mediterranean diet and the adverse effects of sedentary behavior.

The study—conceived and led by the late Arthur Schatzkin, M.D., Dr.P.H.—was an enormous undertaking intended to address the methodological problems faced by prior epidemiologic investigations of diet and cancer. In the 1980s and 1990s there was a growing appreciation for the role of diet and lifestyle in health and disease risk, as well as concern over conflicting findings reported from mostly case-control studies. Early studies were limited by small population size resulting in insufficient power to detect true associations, recall bias common to case-control studies, measurement error in the ascertainment of dietary intake, and homogenous diets among study participants from one region. When writing about the study in 2005, Dr. Schatzkin noted, “I concluded, as did other nutritional epidemiologists around the world, that the field would be advanced only by ‘biting the bullet’ and carrying out such prospective studies, complex and expensive as they might be.” 

In 1987, he approached the AARP. The collaboration gave Dr. Schatzkin and his colleagues access to an immense mailing list of 30 million older American men and women. The study was designed and executed by biostatisticians and dietary experts in what was then called the NCI Division of Cancer Prevention and Control (now the Division of Cancer Control and Population Sciences—DCCPS and the Division of Cancer Prevention) in collaboration with DCEG investigators. Together, the research team selected states and metropolitan areas with active cancer registries that could also maximize the potential to recruit racial and ethnic minorities.

Study Characteristics

Over 567,000 people responded; 60% male and 40% female, with an average age of 62 years. The cohort provided a wide distribution of dietary intake. Body size was comparable to the national average, and despite efforts to collect a diverse group of subjects, participants were predominantly white and more educated compared to the general population. 

One year after the food frequency questionnaire, participants received a risk factor questionnaire, which asked detailed questions on physical activity, medications, cancer screening and family medical history, methods of cooking meat/meat preparation, and early life diet and body size. Investigators linked participant data with information from state cancer registries, the national death index, the Census (1990, 2000, and 2010), and Medicare claims; most recently they added geographical information systems data, providing a new opportunity to track the health outcomes of the large cohort.

Redefining Diet and Health Research

Over the past 25 years, the NIH-AARP Study has produced more than 400 original publications examining critical questions on the relationship between lifestyle factors—including diet, obesity, and physical activity, among others—and cancer outcomes. By the end of 2012, the most recent follow-up, approximately 72% of the participants were alive with an average age of 78 years, and the cohort had recorded approximately 99,000 cancer cases and 160,000 deaths. Research from this study was groundbreaking in the field of cancer epidemiology due to the cohort’s immense size and depth of data. Many of its findings have changed the way the public thinks about cancer and diet.

One of the most prominent topics addressed by the cohort is the effect of coffee-drinking on health outcomes. In 2012, Neal D. Freedman, Ph.D., senior investigator in Metabolic Epidemiology Branch (MEB), and colleagues determined that participants who drank coffee were less likely to die from heart disease, respiratory disease, stroke, injuries and accidents, diabetes, and infections, compared to those who did not drink coffee (Freedman ND et al, N Engl J Med 2012). Subsequent investigations examined the relationship of coffee and cancer risk, overturning older case-control studies that suggested coffee-drinking was associated with increased risk of pancreatic cancer (Guertin KA et al, Br J Cancer 2015). Further studies into the health benefits of coffee found that participants who drank the highest amounts of caffeinated coffee (four cups or more a day) reduced their risk of developing malignant melanoma, an aggressive type of skin cancer (Loftfield E et al, J Natl Cancer Inst 2015). 

The NIH-AARP study also made strides in research on the health effects associated with sedentary behavior and physical activity. Charles E. Matthews, Ph.D., senior investigator in MEB, and colleagues found that longer periods of sitting and television-viewing were associated with increased mortality (Matthews CE et al, Am J Clin Nutr 2012). Further investigations explored the trade-offs between sedentary behavior and physical activity: adults who were more active had lower overall mortality by replacing an hour of sitting with exercise (Matthews CE et al, Med Sci Sports Exerc 2015). With over a decade of research conducted through the cohort, investigators found physical activity was associated with a reduced risk of total cancer mortality and multiple cancer types, including esophageal, lung, kidney, colorectal, breast, stomach, bladder, and endometrial cancers (Patel AV et al, Med Sci Sports Exerc 2019). 

A Resource for Intramural and Extramural Scientists

While the NIH-AARP Diet and Health Study has been a valuable resource for DCEG, the data have also contributed to national and international efforts. Investigators from over 200 extramural institutions and NIH institutes, such as the National Institute of Aging, National Institute of Environmental Health Sciences, National Institute of Diabetes and Digestive and Kidney Diseases, and National Heart, Lung, and Blood Institute, have used its data. Several consortial projects have also benefited from this resource, such as the NCI Cohort Consortium, an extramural-intramural partnership that pools large datasets and biospecimens from over 50 cohorts to study health outcomes.

Evolution and Growth with the Advancement of Research and Technology

One of the remarkable aspects of the NIH-AARP cohort was its ability to evolve and grow as research and technologies advanced. From 2004 to 2006, another questionnaire was sent to obtain updated information on body weight, smoking, medications, and use of dietary supplements, menopausal hormonal therapy, and non-steroidal anti-inflammatory drugs (NSAIDs); to ask more detailed questions about physical activity and to gather self-reported non-cancer health outcomes. Blood or other biological specimens from the participants were not collected at the start of the study, however in 2004-2005, approximately 27,000 healthy participants provided mouthwash samples that were used initially for genetics research and now are being repurposed to study the microbiome.

The dataset’s far-reaching utility is further exemplified by its applicability to environmental studies. Mary H. Ward, Ph.D., senior investigator in the Occupational and Environmental Epidemiology Branch (OEEB), led the effort to identify the geographic coordinates of participants’ residential addresses at enrollment and follow-up, linking them to various environmental monitoring datasets and neighborhood contextual factors. Using these data, DCEG investigators reported that ambient ultraviolet radiation (UVR) exposure was inversely associated with total cancer risk, with differential patterns by cancer type, including increased risk for melanoma (Lin SW et al, Int J Cancer 2012). Currently, Tenure-Track Investigator Rena R. Jones, Ph.D., M.S., Dr. Ward, and Branch Chief Debra Silverman, Sc.D., in OEEB, are reconstructing residence histories for participants prior to study enrollment, which will allow linkages to historical databases for the study of long-term exposure to environmental risk factors, such as air pollution and risk of cancer and chronic diseases (Jones RR et al, Sci Tot Env, in press). 

Validation of Health Measurement Tools

The NIH-AARP Study has spurred the creation of more accurate and sensitive diet and health measurement tools for epidemiological research, many predating the proliferation of applications developed for smart phones. CHARRED: Computerized Heterocyclic Amines Resource for Research in Epidemiology of Disease, is an application designed by Rashmi Sinha, Ph.D., senior investigator in MEB, and colleagues to estimate the intake of mutagenic compounds in cooked meats. The CHARRED tool aids investigations of various meat-related hypotheses, specifically, the potential risk associated with mutagens like heterocyclic amines (HCAs), polycyclic aromatic hydrocarbons (PAHs), nitrates, nitrites, and heme iron.

Dr. Matthews and collaborators created the web-based Activities Completed over Time in 24 hours (ACT-24) to estimate total time spent and associated energy expenditure from sleeping, sitting, and engaging in physical activity during the previous day. This tool overcomes the limitations of traditional physical activity questionnaires, which relied on long-term recall to assess habitual behaviors and did not assess the full spectrum of sedentary behaviors. 

Amy F. Subar, Ph.D., M.P.H., R.D., of DCCPS, and colleagues developed the Automated Self-Administered 24-Hour Dietary Recall (ASA24) to capture high-quality 24-hour dietary recalls at a lower cost, eliminating the need for highly-trained interviewers, which are expensive and often impractical for large-scale nutrition research. 

Mitchell H. Gail, M.D., Ph.D. and Ruth Pfeiffer, Ph.D., senior investigators in the Biostatistics Branch used the NIH-AARP dataset to validate the cancer risk assessment models such as the Colorectal Cancer Risk Assessment Tool (CCRAT) and the Breast Cancer Risk Assessment Tool (BCRAT). Originally, BCRAT was developed with data from the Breast Cancer Detection Demonstration Project and the NCI Surveillance, Epidemiology, and End Results (SEER) from 1983 to 1987. Incidence rates of invasive breast cancer increased during the 1990s and decreased in the early 2000s. In recognizing that rate changes would affect model calibration, in 2010, Dr. Pfeiffer and colleagues evaluated the accuracy of BCRAT with data from the NIH-AARP and Prostate, Lung, Colorectal, Ovarian (PLCO) studies. They found the model underestimated the number of invasive breast cancers among women over the age of 50. Adjustment using recent breast cancer incidence rates from SEER improved its accuracy (Schonfeld SJ et al, J Clin Oncol 2010). Furthermore, the NIH-AARP dataset in combination with PLCO was used to develop a breast, endometrial, and ovarian cancer risk model that included potentially modifiable risk factors, such as body mass index (BMI), alcohol consumption and use of hormone replacement therapy (Pfeiffer RM et al, Plos Medicine 2013).

Looking to the Future

Now that the average NIH-AARP study participant is in their eighties, Christian C. Abnet, Ph.D., M.P.H., Branch Chief and senior investigator of MEB, anticipates that the most recent ascertainment of cancer incidence and mortality, which will provide data through 2018, may be the final linkage update for the study. Dr. Abnet reflected, “Given the study’s tremendous contributions to epidemiology, it is no doubt it will remain a valuable resource for future studies, and analyses of data and biospecimens will continue for many years.”