OBJECTIVE: To examine the association of voluntary vs involuntary weight loss with incidence of cancer in older women.
DESIGN: Prospective cohort study from 1993 to 2000, with cancer incidence identified through record linkage to a cancer registry.
SUBJECTS: A total of 21 707 postmenopausal women initially free of cancer.
MESUREMENTS: Women completed a questionnaire about intentional and unintentional weight loss episodes of ≥20 pounds during adulthood.
RESULTS: Compared with women who never had any ≥20 pounds weight loss episode, women who ever experienced intentional weight loss ≥20 pounds but no unintentional weight loss had incidence rates lower by 11% for any cancer (RR=0.89, 95% CI 0.79–1.00), by 19% for breast cancer (RR=0.81, 95% CI 0.66–1.00), by 9% for colon cancer (RR=0.91, 95% CI 0.66–1.24), by 4% for endometrial cancer (RR=0.96, 95% CI 0.61–1.52), and by 14% for all obesity-related cancer (RR=0.86, 95% CI 0.74–1.01) after adjusting for age, body mass index, waist-to-hip ratio, physical activity, education, marital status, smoking status, pack-years of cigarettes, current estrogen use, alcohol use, parity, and multivitamin use. Furthermore, although overweight women were at increased risk of several cancers, women who experienced intentional weight loss episodes of 20 or more pounds and were not currently overweight were observed to have an incidence of cancer similar to nonoverweight women who never lost weight. Unintentional weight loss episodes were not associated with decreased cancer risk.
CONCLUSIONS: These findings suggest that intentional weight loss might reduce risk of obesity-related cancers.
The prevalence of overweight and obesity has risen sharply in the United States in the last decade.1,2,3 Currently, over 60% of US adults are overweight and over 30% are obese. This rise is especially alarming considering the number of adverse health outcomes associated with excess weight.4,5 Excess weight may be the second leading cause of preventable death in the United States.1,2 The medical risks of excess weight include increased risk of certain cancers, namely breast, colon, endometrial, and kidney cancers.4,6,7
Health concerns are common reasons people offer for attempting to lose weight.8 At any given time, approximately 40% of women in the United States are dieting, and dieting is even more frequent among overweight women.8,9 Thus, intentional weight loss is one of the most common health-related practices.8,10 Despite the high prevalence of dieting, whether intentional weight loss can reduce cancer risk remains unclear. Several epidemiological studies have observed that individuals who lose weight surprisingly have an increased risk of total mortality compared with those who maintained a stable body weight.11,12,13,14,15,16,17 However, a limitation of most epidemiologic studies of weight loss has been that they have not looked at intentional and unintentional weight loss separately. Failure to separate intentional and unintentional weight losses may lead to bias. Conventional wisdom argues that intentional weight loss will reduce risk factors for disease and perhaps obesity-related cancers. At the same time, unintentional weight loss often accompanies illness.15,18,19,20
Very few studies have examined weight loss, regardless of intentionality, and cancer incidence.21 Two studies have assessed the relationship of intentional weight loss and cancer mortality,14,22 but not cancer incidence. We therefore examined the associations of intentional and unintentional weight loss episodes with cancer incidence from 1993 to 2000 in a prospective study of postmenopausal women. It was hypothesized that women in the Iowa Women's Health Study who experienced at least one intentional weight loss episode of 20 or more pounds would have a reduced risk of incident breast, colon, endometrial, and grouped obesity-related cancers vs comparable weight women who never had a 20 or more pound weight loss episode.
Subject population and questionnaire
The Iowa Women's Health Study is a prospective cohort study of health risks in older women. The cohort was assembled in January 1986, when 41 836 of 98 030 randomly selected women between the ages 55 and 69 y, who had a valid driver's license in 1985, completed a mailed survey.6 Although driver's license information indicated that respondents were 3 months older than nonrespondents, had lower body mass index (BMI) by 0.4 kg/m2, and were more likely to live in rural counties of Iowa, the associations of body weight recorded on the driver's license and cancer incidence were similar in respondents and nonrespondents.23 The baseline questionnaire collected demographic information such as current age, marital status, and educational attainment. In addition, the baseline questionnaire included questions on smoking status and amount, multivitamin use, alcohol intake over the past year, hormone replacement therapy status, and reproductive history. Participants were asked three questions about whether they participated in any leisure time physical activity and, if so, the frequency of moderate- and heavy-intensity activities. These latter two questions were combined to create a three-level activity score (low, medium, and high). Prevalent cancers were ascertained by asking the women whether they had ever been told by a physician that they had any form of cancer, excluding skin cancer. Enclosed with the questionnaire were a paper tape measure and written instructions to have a friend measure waist (2.5 cm above the umbilicus) and hip (maximal protrusion) circumferences.24 Waist-to-hip circumference ratio was calculated. Participants reported their current weight and height, maximum adult weight, and weight at ages 18, 30, 40, and 50 y. BMI was computed. The self-measured or self-reported anthropometric variables obtained by this protocol were valid (interclass correlation coefficient with measures by a trained technician, ≥0.84) and reliable (interclass correlation of measures at two different periods, ≥0.85).24
In addition to the baseline questionnaire, subsequent questionnaires were mailed in 1988 (91% response by the cohort), 1990 (90%), 1992 (83%), and 1997 (79%). The 1992 questionnaire also included questions on current body weight, marital status, regular alcohol use, estrogen use, and the intentionality of weight loss episodes.
Assessment of intentionality of weight loss
Intentional weight loss was assessed with questions about the number of weight loss episodes of 5–9, 10–19, 20–49, and 50 or more pounds reported for each of three age periods (18–39, 40–54, and 55 y at the time of follow-up in 1992). For example, ‘Between the ages of 18–39, about how many different times did you lose each of the following amounts of weight on purpose, excluding pregnancy or illness?’ Unintentional weight loss was measured by asking questions about the number of unintentional weight loss episodes of ≥20 pounds reported for each of the same three age periods. For example, ‘Between the ages of 18 and 39, about how many times did you lose 20 or more pounds when you weren't trying to, for example, because of illness?’ These questions had adequate test–retest reliability and were correlated in expected directions with other health and demographic variables.25
Participants were classified into one of four categories of weight loss since age 18 y. The four categories were: (1) at least one intentional weight loss episode of 20 or more pounds, but no unintentional weight loss episodes; (2) at least one unintentional weight loss episode of 20 or more pounds, but no intentional weight loss episodes; (3) at least one intentional weight loss episode of 20 or more pounds, and at least one unintentional weight loss episode of 20 or more pounds; or (4) neither an intentional weight loss episode nor an unintentional weight loss episode of 20 or more pounds. A 20-pound weight loss criterion was selected to represent significant weight loss and to maintain comparability between intentional and unintentional weight loss episodes.
Ascertainment of cancer incidence
Cancer incidence was identified by computer linkage with State Health Registry of Iowa, a National Cancer Institute-supported Surveillance, Epidemiology, and End Results cancer registry.
The base sample for the present analyses consisted of women who responded to the third follow-up questionnaire in 1992 (n=33 017). Women with a history of cancer at baseline in 1986 or an incident cancer by 1992 were excluded from these analyses (n=4190). A total of 21 707 women were free of cancer and completed enough questions on weight loss episodes to be classified into one of the four weight loss categories. Perhaps due to the complicated format of the questionnaire, 22% of the women did not complete enough questions to be classified into a lifetime weight loss category. Women who did vs those who did not complete sufficient weight loss questions were generally similar. Those who completed the weight loss questions were slightly more likely to have reported prevalent diseases or health conditions on one of the mail surveys (1986–1992).
Analyses were conducted using Cox proportional hazards regression with the SAS program PHREG (SAS Institute, Inc., Cary, NC, USA). Incident end points of interest were: any cancer, breast cancer, colon cancer, endometrial cancer, and ‘obesity-related cancer’ (any of the preceding three sites or kidney cancer). Person-years of follow-up were calculated as the time elapsed from January 1993 questionnaire to either the incidence date of cancer, date of emigration from Iowa, death, or else the end of the year 2000. Multivariate-adjusted relative risks and their 95% confidence intervals (CI) were computed from the regression coefficients and their standard errors.
The main independent variable, category of weight loss, was entered using three dummy variables: ever lost 20 or more pounds intentionally only, ever lost 20 or more pounds unintentionally only, and ever lost 20 or more intentionally plus 20 or more pounds unintentionally. The reference group was women who had never lost 20 or more pounds. Two models were run to adjust for confounding variables: model A included age in 1992, BMI in 1992, and BMI-squared; model B included model A variables plus variables measured at baseline (1986)—waist-to-hip ratio, physical activity level (low vs high), educational attainment (high school or less, more than high school), smoking amount (pack-years), parity (first live birth >30 y or nulliparous vs first live birth ≤30 y) and multivitamin use (yes vs no)—and variables assessed in 1992—marital status (currently married or not), smoking status (yes or no), and estrogen use (current vs other).
The cohort included 21 707 postmenopausal women who were free of cancer through 1992. Table 1 shows demographic and health characteristics of women by weight loss category. Overall, 53% had never lost 20 or more pounds in one episode, 17% had lost 20 or more pounds intentionally, 18% had lost 20 or more pounds unintentionally, and 11% had lost 20 pounds intentionally plus 20 pounds unintentionally. Given the large sample size, statistically significant (P<0.05) global differences existed among weight loss categories for all baseline or 1992 variables except current estrogen use. The average age for women in the cohort in 1992 was 68 y. Women who never lost 20 or more pounds or lost 20 or more pounds unintentionally had a lower prevalence of current overweight than the two groups reporting intentional weight loss episodes of 20 pounds or more. Women who had never lost 20 or more pounds, or reported only intentional weight loss episodes, were more likely to be currently married and were less likely to be current smokers in 1992.
Cancer incidence and category of weight loss episodes
As shown in Table 2, after adjusting for age and BMI, a history of intentional weight loss episodes of 20 pounds or more by itself was associated with a 12% decreased risk of any incident cancer, 21% decreased risk of breast cancer, 18% reduced risk of colon cancer, 7% reduced risk of endometrial cancer, and 18% reduced risk of obesity-related cancers, compared with never having lost ≥20 pounds. Further adjustment for waist-to-hip ratio, physical activity, education, marital status, smoking status, pack-years of cigarette use, current estrogen use, alcohol use, age at first live birth, and multivitamin use attenuated these relative risk estimates only slightly, although some confidence intervals included one. In contrast, unintentional weight loss of ≥20 pounds by itself was associated with a 5–11% higher risk of any incident cancer compared with never losing 20 or more pounds, but this association was not statistically significant and was attenuated after multivariate adjustment. Otherwise, unintentional weight loss episodes of ≥20 pounds, with or without a history of intentional weight loss, were not associated with any of the incident cancers studied.
Weight loss category, overweight status, and cancer incidence
Table 3 shows the relative risks of any incident cancer, breast cancer, or obesity-related cancers by weight loss category and overweight status. Compared to women who were not overweight and who never had a weight loss episode of 20 or more pounds (the reference category), women who were currently overweight uniformly had greater cancer incidence. For example, compared with reference group women, women who were overweight and had never lost 20 or more pounds had 14% higher incidence of any cancer, 46% higher incidence of breast cancer, and 38% higher incidence of obesity-related cancers. However, intentional weight loss sufficient to no longer be overweight was associated with cancer risk equivalent to the reference group. That is, women who had lost 20 or more pounds at least once intentionally and were not currently overweight had a relative risk of any cancer (RR=1.01), breast cancer (RR=0.79), and obesity-related cancer (RR=0.91) equivalent to women who were not overweight and had never lost 20 pounds. In contrast, women with unintentional weight loss of 20 or more pounds and who were not overweight had cancer risks slightly higher (RR=1.06–1.13) than the reference group of women.
As our main analysis was based upon reports of large weight loss episodes in the past and not upon calculated weight change, we ran a supplemental analysis. We computed weight change from each participant's maximum adult weight prior to the IWHS baseline questionnaire in 1986 and her reported weight in 1992. Weight change was categorized for the cohort as lost 10 or more pounds from maximum (n=10 218), gained 10 or more pounds (n=1803), or a reference category of less than a 10-pound change from maximum (n=11 955). In a multivariate model comparable to model B in Table 2, women who lost at least 10 pounds from their maximums had a reduced risk of breast cancer (RR=0.85, 95% CI 0.73–0.98) and obesity-related cancer (RR=0.92, 95% CI 0.81–1.06) compared with the reference category.
This prospective investigation found that, compared with never having a weight loss episode of 20 or more pounds and adjusted for age and BMI, one or more intentional weight loss episodes of 20 or more pounds were associated with a 12% reduced incidence of any cancer, 21% reduced incidence of breast cancer, 18% reduced incidence of colon cancer, 7% reduced incidence of endometrial cancer, and 18% reduced incidence of obesity-related cancers. These associations were not always statistically significant after multivariate adjustment. Furthermore, although overweight women were at increased risk of several cancers, women who experienced intentional weight loss episodes of 20 or more pounds and were not currently overweight were observed to have an incidence of cancer similar to nonoverweight women who never lost weight. These findings suggest that intentional weight loss might reduce risk of obesity-related cancers. In a supplemental analysis, we also showed that 10 or more pounds weight loss from maximum adult weight (regardless of intentionality) also was associated with reduced risk of breast cancer and obesity-related cancer.
To our knowledge, there are no other studies of weight loss intentionality and incident cancer, but one study has looked at weight loss and cancer mortality. In the Cancer Prevention Study I, Williamson et al22 examined mortality from a group of obesity-related cancers (gallbladder, breast, cervix, uterine, and ovary). Intentional weight loss in overweight women with obesity-related health conditions was associated with a 30% reduction in overall cancer mortality, and an even greater reduction of 40–50% in obesity-related cancer mortality. The findings of Williamson et al were independent of the amount of weight loss,22 but that study could not separate whether the effect of intentional weight loss was greater for cancer incidence or for cancer survival. Nevertheless, our findings and those of Williamson et al suggest that intentional weight loss may reduce cancer incidence and mortality.
Strengths and limitations
Strengths of our study were that it was prospective and had detailed information about weight loss episodes including intentionality and amount of weight lost. At the same time, there are some limitations. First, the IWHS cohort is ethnically homogenous and therefore may not be generalizable to other populations. Second, measures of intentionalilty of weight loss episodes were not validated and potential reporting biases may exist. For instance, overweight women may be more likely to label unintentional weight loss episodes as intentional weight loss episodes since weight loss is in all likelihood sought-after. Also, French et al14 observed that more recent weight loss episodes in this older cohort were more easily recalled and more likely to be unintentional. However, the weight loss questionnaire was moderately reliable, and indirect validity for intentionality of the weight loss history questionnaire was provided by its correlation in expected direction with other measures such as BMI.14,26,27 Third, we did not ask for the starting and stopping weights for weight loss episodes or whether weight losses were maintained or for how long. Fourth, a fair number of women in the IWHS cohort were excluded from these analyses due to insufficient numbers of weight loss questions answered.
This study found that compared with never losing 20 or more pounds, an intentional weight loss episode of 20 more pounds was associated with a reduced risk of any incident cancer, breast, colon, endometrial, and obesity-related cancers. The results suggest that modest weight loss among overweight individuals might reduce the incidence of obesity-related cancers.
Flegal KM, Carroll MD, Ogden CL, Johnson CL . Prevalence and trends in obesity among US adults, 1999–2000. JAMA 2002; 288: 1723–1727.
Kuczmarski RJ, Carroll MD, Flegal KM, Troiano RP . Varying body mass index cutoff points to describe overweight prevalence among US adults: NHANES III (1988 to 1994). Obes Res 1997; 5: 542–548.
Flegal KM, Carroll MD, Kuczmarski RJ, Johnson CL . Overweight and obesity in the United States: prevalence and trends, 1960–1994. Int J Obes Relat Metab Disord 1998; 22: 39–47.
National Task Force on the Prevention and Treatment of Obesity. Overweight, obesity, and health risk. Arch Intern Med 2000; 160: 898–904.
National Institutes of Health. National Institutes of Health Consensus Development Conference Statement: health implications of obesity. Ann Intern Med 1985; 103: 147–151.
Folsom AR, Kaye SA, Prineas RJ, Potter JD, Gapstur SM, Wallace RB . Increased incidence of carcinoma of the breast associated with abdominal adiposity in postmenopausal women. Am J Epidemiol 1990; 131: 794–803.
Fund WCR. Food, nutrition, and the prevention of cancer: a global perspective. American Institute for Cancer Research: Washington, DC; 1997.
Brownell KD, Jeffery RW . Improving long-term weight loss: pushing the limits of treatment. Behav Ther 1987; 18: 353–374.
Williamson DF, Serdula MK, Anda RF, Levy A, Byers T . Weight loss attempts in adults: goals, duration, and rate of weight loss. Am J Public Health 1992; 82: 1251–1257.
Serdula MK, Mokdad AH, Williamson DF, Galuska DA, Mendlein JM, Heath GW . Prevalence of attempting weight loss and strategies for controlling weight. JAMA 1999; 282: 1353–1358.
Andres R, Muller DC, Sorkin JD . Long-term effects of change in body weight on all-cause mortality. A review. Ann Intern Med 1993; 119 (7 Part 2): 737–743.
Blair SN, Shaten J, Brownell K, Collins G, Lissner L . Body weight change, all-cause mortality, and cause-specific mortality in the Multiple Risk Factor Intervention Trial. Ann Intern Med 1993; 119 (7 Part 2): 749–757.
Folsom AR, French SA, Zheng W, Baxter JE, Jeffery RW . Weight variability and mortality: the Iowa Women's Health Study. Int J Obes Relat Metab Disord 1996; 20: 704–709.
French SA, Folsom AR, Jeffery RW, Williamson DF . Prospective study of intentionality of weight loss and mortality in older women: the Iowa Women's Health Study. Am J Epidemiol 1999; 149: 504–514.
Lee IM, Paffenbarger Jr RS . Change in body weight and longevity. JAMA 1992; 268: 2045–2049.
Lissner L, Odell PM, D'Agostino RB, Stokes III J, Kreger BE, Belanger AJ, Brownell KD . Variability of body weight and health outcomes in the Framingham population. N Engl J Med 1991; 324: 1839–1844.
Pamuk ER, Williamson DF, Madans J, Serdula MK, Kleinman JC, Byers T . Weight loss and mortality in a national cohort of adults, 1971–1987. Am J Epidemiol 1992; 136: 686–697.
Kassirer JP, Angell M . Losing weight—an ill-fated New Year's resolution. N Engl J Med 1998; 338: 52–54.
Pamuk ER, Williamson DF, Serdula MK, Madans J, Byers TE . Weight loss and subsequent death in a cohort of US adults. Ann Intern Med 1993; 119 (7 Part 2): 744–748.
Reynolds MW, Fredman L, Langenberg P, Magaziner J . Weight, weight change, mortality in a random sample of older community-dwelling women. J Am Geriatr Soc 1999; 47: 1409–1414.
French SA, Folsom AR, Jeffery RW, Zheng W, Mink PJ, Baxter JE . Weight variability and incident disease in older women: the Iowa Women's Health Study. Int J Obes Relat Metab Disord 1997; 21: 217–223.
Williamson DF, Pamuk E, Thun M, Flanders D, Byers T, Heath C . Prospective study of intentional weight loss and mortality in never-smoking overweight US white women aged 40–64 years. Am J Epidemiol 1995; 141: 1128–1141.
Bisgard KM, Folsom AR, Hong C-P, Sellers TA . Mortality and cancer rates in nonrespondents to a prospective study of older women: 5-year follow-up. Am J Epidemiol 1994; 139: 990–1000.
Kushi LH, Kaye SA, Folsom AR, Soler JT, Prineas RJ . Accuracy and reliability of self-measurement of body girths. Am J Epidemiol 1988; 128: 740–748.
French SA, Jeffery RW, Folsom AR, Williamson DF, Byers T . Weight variability in a population-based sample of older women: intercorrelation and reliability of measures. Int J Obes Relat Metab Disord 1995; 19: 22–29.
Meltzer AA, Everhart JE . Self-reported substantial 1-year weight change among men and women in the United States. Obes Res 1995; 3 (Suppl 2): 123s–134s.
Meltzer AA, Everhart JE . Unintentional weight loss in the United States. Am J Epidemiol 1995; 142: 1039–1046.
The Iowa Women's Health Study was funded by a grant (RO1 CA39742) from the National Cancer Institute. The contents of this paper are solely the responsibilities of the authors and do not necessarily represent the official view of the National Cancer Institute. We thank Dr Simone French for previous work that helped define the weight loss categories, Ms Ching Ping Hong for SAS programming advice, and Ms Laura Kemmis for manuscript preparation.
About this article
CA: A Cancer Journal for Clinicians (2018)
Cancer Epidemiology (2018)