A review

Obesity and screening for breast, cervical, and colorectal cancer in women

Authors

  • Sarah S. Cohen MS,

    Corresponding author
    1. Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina
    2. International Epidemiology Institute, Rockville, Maryland
    • Department of Epidemiology, School of Public Health, CB 7435, University of North Carolina, Chapel Hill, NC 27599
    Search for more papers by this author
    • Fax: (301) 424-1054

  • Rachel T. Palmieri MSPH,

    1. Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina
    Search for more papers by this author
  • Sarah J. Nyante MSPH,

    1. Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina
    Search for more papers by this author
  • Daniel O. Koralek MA, MS,

    1. Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina
    2. Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, North Carolina
    Search for more papers by this author
  • Sangmi Kim PhD,

    1. Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina
    Search for more papers by this author
  • Patrick Bradshaw MS,

    1. Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina
    Search for more papers by this author
  • Andrew F. Olshan PhD

    1. Department of Epidemiology, University of North Carolina, Chapel Hill, North Carolina
    2. Lineberger Comprehensive Cancer Center, University of North Carolina, Chapel Hill, North Carolina
    Search for more papers by this author

Abstract

The literature examining obesity as a barrier to screening for breast, cervical, and colorectal cancer has not been evaluated systematically. With the increasing prevalence of obesity and its impact on cancer incidence and mortality, it is important to determine whether obesity is a barrier to screening so that cancers among women at increased risk because of their body size can be detected early or prevented entirely. On the basis of 32 relevant published studies (10 breast cancer studies, 14 cervical cancer studies, and 8 colorectal cancer studies), the authors reviewed the literature regarding associations between obesity and recommended screening tests for these cancer sites among women in the U.S. The most consistent associations between obesity and screening behavior were observed for cervical cancer. Most studies reported an inverse relation between decreased cervical cancer screening and increasing body size, and several studies reported that the association was more consistent among white women than among black women. For breast cancer, obesity was associated with decreased screening behavior among white women but not among black women. The literature regarding obesity and colorectal cancer screening adherence was mixed, with some studies reporting an inverse effect of body size on screening behavior and others reporting no effect. Overall, the results indicated that obesity most likely is a barrier to screening for breast and cervical cancers, particularly among white women; the evidence for colorectal cancer screening was inconclusive. Thus, efforts to identify barriers and increase screening for breast and cervical cancers may be targeted toward obese women, whereas outreach to all women should remain the objective for colorectal cancer screening programs. Cancer 2008. © 2008 American Cancer Society.

The objective of a cancer screening program is to detect cancerous and precancerous lesions in asymptomatic individuals, which, with effective treatment, will decrease cancer-related morbidity and mortality. In the U.S., screening programs exist for cancers of the breast, cervix, and colon and rectum, which will account for an estimated 326,290 new cancer cases and 69,850 cancer deaths in 2007 among women in the U.S.1

For breast cancer, the U.S. Preventive Services Task Force (USPSTF) and the American Cancer Society (ACS) both recommend screening mammography for women aged ≥40 years who are at average risk of breast cancer, either annually (ACS) or every 1 to 2 years (USPSTF).2, 3 For cervical cancer, the ACS recommends Papanicolaou (Pap) tests every 1 or 2 years beginning 3 years after the onset of sexual activity or age 21 years, whichever comes first. After age 30 years, the screening interval may be relaxed to 2 or 3 years based on previous Pap test results and other risk factors.4 The USPSTF recommends that screening end at age 65 years, whereas the ACS recommends ceasing at age 70 years, and both the ACS and the USPSTF recommend that women who have had a total hysterectomy for a benign condition not be screened for cervical cancer.5–7 For colorectal cancer, the ACS recommends either colonoscopy every 10 years, sigmoidoscopy every 5 years, or yearly fecal occult blood test (FOBT), beginning at age 50 years.4, 8 The USPSTF has not established evidence that colonoscopy is effective in reducing mortality from colorectal cancer and, thus, recommends screening women aged ≥50 years with FOBT, sigmoidoscopy, or both.9

National screening rates for breast and cervical cancer are relatively high. On the basis of 2004 Behavioral Risk Factor Surveillance System (BRFSS) data, 74.9% of women aged >40 years received a mammogram within the past 2 years, and 86% of women aged >18 years received a Pap test within the past 3 years.10 In contrast, colorectal cancer screening rates are much lower. According to data from the 2003 National Health Interview Survey (NHIS), only 33% of adults reported receiving endoscopy (ie, sigmoidoscopy or colonoscopy) in the previous 5 years, and 15% reported an FOBT in the previous year.11

The percentage of American women who are overweight or obese has been increasing steadily in recent decades.12, 13 Between 1976 and 2004, the percentage of overweight women (ie, body mass index [BMI] ≥25, calculated as weight in kilograms divided by the square of height in meters [kg/m2]) rose markedly, from 38.7% to 57.1% in white women and from 62.6% to 79.5% in black women.13 According to National Health and Nutrition Examination Survey data for 2000 through 2004, 31.5% of white women and 51.6% of black women were obese (BMI ≥30 kg/m2).13

The rapid increase in the prevalence of obesity, coupled with the suggestion that 20% of cancer deaths among women in the U.S. in 2000 were attributable to obesity14 and the evidence indicating that obesity is a modifiable risk factor for both postmenopausal breast cancer and colorectal cancer,15 highlights the importance of cancer screening among overweight and obese women. To our knowledge, the literature on obesity as a barrier to cancer screening among women has not been evaluated systematically. The purpose of this review was to evaluate the evidence regarding the effect of obesity on the receipt of recommended screening tests for breast, cervical, and colorectal cancer among women in the U.S. With the increasing prevalence of obesity and its impact on cancer incidence and mortality, it is imperative to determine whether obesity is a barrier to screening so that cancers among women who already are at increased risk can be detected as early as possible or prevented altogether.

MATERIALS AND METHODS

We conducted a PubMed search between January and February 2007 for each of 3 cancer sites (breast, cervical, and colorectal). We used a standard set of terms for body size for all 3 searches that included the following terms: obesity, body mass index, BMI, obese, overweight, body weight, and body size. For breast cancer, the screening search terms included breast cancer screening, mammography, and mammogram. For the cervical cancer search, we used the terms cervical cancer screening, Papanicolaou test, Pap test, Pap smear, pelvic examination, and gynecologic examination. The colorectal cancer search terms included colorectal cancer screening; colonoscopy; sigmoidoscopy; and fecal occult blood test. We limited our search to articles written in English. For our search of references related to colorectal cancer screening, we included articles that reported study results for women only or stated that sex-stratified results were equivalent. We screened references from each search first on the basis of title and abstract and then by reviewing the full article. Articles that were considered relevant were those with any data that either estimated the prevalence of screening behaviors or characteristics by body size or that estimated the relative risk or relative prevalence of screening by body size. Two investigators independently reviewed all of the references and abstracted the selected articles; discrepancies were resolved by consensus.

Our search related to mammography initially produced 743 articles. After review, we identified 16 articles that specifically addressed the relation between screening mammography and obesity. Of those 16 articles, 7 were excluded because they dealt with populations outside of the U.S. This exclusion criterion was used because healthcare systems and access to cancer screening services vary widely between countries. The cited references of the remaining 9 articles suggested 1 additional reference, yielding a total of 10 articles for this review (Table 1).16–25 Our search of cervical cancer screening and obesity yielded 192 references. We identified 11 relevant articles on cervical cancer screening and body size. Three additional articles were identified from the cited references, for a total of 14 studies to be included in this review (Table 2).16, 19–23, 26–32 Our initial search of colorectal cancer screening and obesity resulted in 275 references. After review, 6 articles were identified for inclusion. The cited references from the initial articles identified 2 additional studies, resulting in 8 studies for inclusion in this review (Table 3).33–40

Table 1. Association Between Obesity and Mammography Screening Behavior for Breast Cancer
Cross-sectional studyStudy descriptionResults
  • BMI indicates body mass index; OR, odds ratio; 95% CI, 95% confidence interval; RD, risk difference; NR, not reported; SE, standard error; HRS, Health and Retirement Study; AHEAD, Asset and Health Dynamics Among the Oldest Study.

  • Results are the most adjusted as reported by the individual studies.

  • *

    BMI was treated as a continuous variable in analyses.

  • Women rated their body size on a scale from 1 (very thin) to 9 (obese) by using a graphic and grouped into thin (1–3), medium (4–6), and obese (7–9) categories.

Fontaine 199820 Mammogram within past 3 y
 Location and dateUS, 1992BMI, kg/m2*OR [95% CI]
 Population characteristicsAged ≥18 y; US residents who self-reported sociodemographic information and use of healthcare services; 80% white25.11.00
 Analysis sample size3105350.81 [0.59–1.12]
  400.73 [0.45–1.19]
Wee 200023 Mammogram within past 2 y
 Location and dateUS, 1994BMI, kg/m2RD [95% CI]
 Population characteristicsAges 50–75 y; 81% white18.5–24.90
 Analysis sample size350225–29.9−2.8[−6.7–0.9]
  30–34.9−5.3 [−11.1–0.5]
  35–39.9−4.5 [−12.5–3.4]
  ≥40−8.8 [−22.9–5.3]
Fontaine 200121 Mammogram within past 2 y
 Location and dateUS, 1998BMI, kg/m2OR [95% CI]
 Population characteristicsAged ≥40 y; 84% white<18.51.32 [1.13–1.54]
 Analysis sample size38,68218.5–24.91.00
  25–29.91.00 [0.94–1.07]
  30–34.91.12 [1.02–1.23]
  35–39.91.13 [0.98–1.30]
  ≥401.32 [1.09–1.59]
Carney 200217 Mammogram within past 2 y
 Location and dateNH, US, 1996Mean±SE BMI, kg/m2Screening behavior
 Population characteristicsAged ≥50 y; race NR26.1 ± 0.3Adhering
 Analysis sample size62527.6 ± 0.4Nonadhering
Wee 200424 Mammogram within past 2 y: White women only
 Location and dateUS, 1998BMI, kg/m2OR [95% CI]
 Population characteristicsAges 50–75 y; 74% white18.5–24.91.00
 Analysis sample size527725–29.90.98 [0.92–1.04]
  30–34.90.94 [0.86–1.01]
  35–39.90.83 [0.68–0.96]
  ≥400.90 [0.67–1.07]
  Mammogram within past 2 y: Black women only
  BMI, kg/m2OR [95% CI]
  18.5–24.91.00
  25–29.91.19 [1.01–1.32]
  30–34.91.22 [0.98–1.39]
  35–39.91.37 [1.13–1.50]
  ≥400.95 [0.60–1.25]
Colbert 200418 Age at first mammogram
 Location and dateUS, 2000–2002Body typeMean age [95% CI], y
 Population characteristicsWomen receiving first mammogram; excluded women on Medicare; 90% whiteThin40.5 [39.5–41.6]
 Analysis sample size786Medium41.7 [41–42.4]
  Obese44.7 [42.5–46.9]
Ostbye 200522 Mammogram within past 2 y, White women only
 Location and dateUS, 1995, 1996, 2000BMI, kg/m2OR [95% CI]
 Population characteristicsHRS: ages 51–61 y, 82% white; AHEAD: ages ≥70 y, 86% white<18.50.91 [0.66–1.26]
 Analysis sample sizeHRS, 4439; AHEAD, 401018.5–24.91.00
  25–29.90.90 [0.78–1.05]
  30–34.90.73 [0.60–0.88]
  35–39.90.69 [0.51–0.93]
  ≥400.59 [0.40–0.88]
  Mammogram within past 2 y: Black women only
  BMI, kg/m2OR [95% CI]
  <18.50.71 [0.33–1.53]
  18.5–24.91.00
  25–29.91.13 [0.79–1.62]
  30–34.90.97 [0.65–1.45]
  35–39.91.03 [0.61–1.76]
  ≥401.07 [0.60–1.92]
Ferrante 200619 Mammogram within past 2 y
 Location and dateNJ, 2000–2003BMI, kg/m2OR [95% CI]
 Population characteristicsAges 40–74 y; 10% white, non-Hispanic; 50% Hispanic18–29.91.00
 Analysis sample size180930–34.90.99 [0.71–1.39]
  35–39.91.40 [0.90–2.18]
  >401.13 [0.72–1.78]
Amy 200616 Mammogram within past 2 y
 Location and dateUS, date NRBMI, kg/m2Prevalence
 Population characteristicsAges 40–80 y; BMI >25 kg/m2; 68% white25–3586%
 Analysis sample size33835–4581.2%
  45–5580.3%
  >5567.7%
Zhu 200625 No mammogram in past 2 y: White women only
 Location and dateUS, 2000BMI, kg/m2OR [95% CI]
 Population characteristicsAges 40–80 y; 84% white<18.51.8 [1.3–2.7]
 Analysis sample size850118.5–24.91.00
  25–29.90.9 [0.8–1.1]
  30–34.91.0 [0.8–1.2]
  35–39.91.1 [0.8–1.4]
  ≥401.4 [1.0–1.9]
  No mammogram in past 2 y: Black women only
  BMI, kg/m2OR [95% CI]
  <18.51.0 [0.2–4.1]
  18.5–24.91.00
  25–29.90.8 [0.6–1.2]
  30–34.90.6 [0.4–1]
  35–39.90.9 [0.5–1.7]
  ≥401.3 [0.7–2.4]
Table 2. Association Between Obesity and Papanicolaou Smear Screening Behavior for Cervical Cancer
Cross-sectional studyStudy descriptionResults
  • BMI indicates body mass index; POR, prevalence odds ratio; 95% CI, 95% confidence interval; Pap, Papanicolaou; r2, correlation coefficient; NR, not reported; IRD, incidence rate difference; RR, relative risk.

  • Results are the most adjusted as reported by the individual studies.

  • Ponderal Index = height (inches)/(cube root of weight [pounds]).

  • Ideal bodyweight = 3.83×height (inches)–105.7.

  • §

    Results are from a reanalysis of the 2000 National Health Interview Survey data originally analyzed and reported by Wee 2005.30

Williams 197231 Correlation between increasing Ponderal Index and receipt of yearly Pap smear test
 Location and dateBoston, date NRr2=−0.24 
 Population characteristicsAges 35–54 y; mothers of 9th graders; race NR  
 Analysis sample size161  
Lubitz 199529 No Pap smear at recommended visit
 Location and dateInd, 1989Bodyweight categoryPOR [95% CI]
 Population characteristicsNo hysterectomy; excluded if weight was <60% of ideal weight; % white NR; 62% black≤130% Of ideal bodyweight1.00
 Analysis sample size970130%–200% Of ideal bodyweight1.20 [0.86–1.67]
  >200% Of ideal bodyweight1.20 [0.58–2.47]
Fontaine 199820 No Pap smear within past 3 y
 Location and dateUS, 1992BMI, kg/m2POR [95% CI]
 Population characteristicsAged ≥18 y; 80% white25.11.00
 Analysis sample size6314351.29 [1.04–1.58]
  401.46 [1.07–1.98]
Simoes 199942 No Pap smear within past 1 y
 Location and dateMo, 1994BMI, kg/m2POR [95% CI]
 Population characteristicsAged ≥18 y; included women with hysterectomy; 85% white≤27.31.0
 Analysis sample size1609>27.31.2 [0.9–1.5]
Wee 200023 Pap smear within past 3 y: White women only
 Location and dateUS, 1994BMI, kg/m2IRD [95% CI]
 Population characteristicsAges 18–75 y; no hysterectomy; 73% white18.5–<250
 Analysis sample size740525–<30−3.4 [−6.4, −0.5]
  30–35−9.4 [−13.5, −5.2]
  35–<40−8.3 [−14.2, −2.3]
  ≥40−8.8 [−16.9, −0.7]
  Pap smear within past 3 y: Black women only
  18.5–<250
  25–<30−2.5 [−8.0, 3.1]
  30–35−0.2 [−4.5, 4.1]
  35–<40−0.5 [−7.8, 6.8]
  ≥401.7 [−5.0, 8.4]
Fontaine 200121 No Pap smear within past 2 y
 Location and dateUS, 1998BMI, kg/m2POR [95% CI]
 Population characteristicsAged ≥18 y; 85% white<18.51.21 [1.09–1.34]
 Analysis sample size72,88918.5–24.91.00
  25–29.91.13 [1.07–1.18]
  30–34.91.22 [1.14–1.30]
  35–39.91.43 [1.30–1.57]
  ≥401.69 [1.49–1.92]
Amonkar 200226 Pap smear within past 1 y
 Location and dateAppalachia, 1997BMI, kg/m2POR [95% CI]
 Population characteristicsAged ≥18 y; 79% white<301.00
 Analysis sample size12,949>300.80 [0.66–0.96]
Coughlin 200427 Pap smear within past 3 y
 Location and dateUS, 1999BMI, kg/m2Proportion [95% CI]
 Population characteristicsAged ≥18 y; no hysterectomy; 75% white<18.583.7% [81.8–85.7%]
 Analysis sample size49,564>18.5–<2586.7% [86.1–87.3%]
  25–2986.5% [85.7–87.2%]
  >3085.9% [84.9–86.8%]
Datta 200528 No Pap smear within past 2 y
 Location and dateUS, 1995BMI, kg/m2POR [95% CI]
 Population characteristicsAges 21–65 y; black only<201.3 [1.1–1.6]
 Analysis sample size40,00920–24.91.00
  25–29.91 [0.9–1.1]
  ≥301.6 [1.4–1.7]
Ostbye 200522 Pap smear within past 2 y: White women only
 Location and dateUS, 1996 and 2000BMI, kg/m2POR [95% CI]
 Population characteristicsAges 50–64 y; 82% white<18.50.74 [0.55–0.98]
 Analysis sample size1996, 4434; 2000, 400918.5–24.91.00
  25–29.90.78 [0.68–0.89]
  30–34.90.68 [0.57–0.80]
  35–39.90.59 [0.45–0.78]
  ≥400.50 [0.35–0.71]
  Pap smear within past 2 y: Black women only
  <18.50.98 [0.47–2.05]
  18.5–24.91;00
  25–29.91.50 [1.07–2.12]
  30–34.91.22 [0.84–1.77]
  35–39.91.13 [0.70–1.85]
  ≥400.75 [0.45–1.26]
Wee 200530 Pap smear within past 3 y: All women
 Location and dateUS, 2000BMI, kg/m2RR [95% CI]
 Population characteristicsAges 18–75 y; no hysterectomy; 71% white18.5–<251.00
 Analysis sample size12,170≥400.94 [0.87–0.99]
  Pap smear within past 3 y: White women only
  BMI, kg/m2Prevalence
  18.5–<2587%
  25–<3087%
  30–<3583%
  35–<4085%
  ≥4078%
  Pap smear within past 3 y: Black women only
  18.5–<2588%
  25–<3089%
  30–<3589%
  35–<4083%
  ≥4087%
Amy 200616 Pap smear within past 2 y
Location and dateUS, date NRBMI, kg/m2Prevalence
 Population characteristicsAged ≥21 y; BMI >25 kg/m2; 68% white25–3586.3%
 Analysis sample size48135–4585.8%
  45–5584.3%
  >5568.3%
Ferrante 200619 Pap smear within past 3 y
 Location and dateNJ, 2000–2003BMI, kg/m2POR [95% CI]
 Population characteristicsAges 40–74 y; no hysterectomy; 10% white18–29.91.00
 Analysis sample size155430–34.90.74 [0.51–1.09]
  35–39.91.31 [0.77–2.24]
  >400.68 0.42–1.13]
  <301.00
  ≥300.75 [0.58–0.99]
Wu 200632§ No Pap smear within past 3 y: White women only
 Location and dateUS, 2000BMI, kg/m2RR [95% CI]
 Population characteristicsAges 18–75 y; no hysterectomy; screening Pap smear tests only; 79% white<18.50.87 [0.59–1.27]
 Analysis sample size991018.5–24.91.00
  25–29.90.94 [0.79–1.13]
  30–34.91.13 [0.90–1.39]
  35–39.91.01 [0.72–1.37]
  ≥401.27 [0.89–1.77]
  No Pap smear within past 3 y: Black women only
  <18.50.26 [0.04–1.40]
  18.5–24.91.00
  25–29.91.07 [0.73–1.50]
  30–34.91.18 [0.76–1.75]
  35–39.91.93 [1.03–3.01]
  ≥401.29 [0.70–2.11]
Table 3. Association Between Obesity and Screening Behavior for Colorectal Cancer
StudyStudy descriptionResults
  • BMI indicates body mass index; POR, prevalence odds ratio; 95% CI, 95% confidence interval; OR, odds ratio; NR, not reported; CRC, colorectal cancer; RD, risk difference; FOBT, fecal occult blood test; KPMCP, Kaiser Permanente Medical Care Program.

  • Results are the most adjusted as reported by the individual studies.

  • Results are for the total sample; results for women only were not reported.

  • Endoscopic screen refers to flexible sigmoidoscopy within the past 5 years or colonoscopy within the past 10 years.

  • §

    Only controls from the population-based case–control studies were included in the screening analyses.

Cross-sectional
 Chao 200433 Ever received an endoscopic screen
  Location and dateUS, 1997BMI, kg/m2POR [95% CI]
  Population characteristicsAged ≥50 y; 98% white women<18.50.99 [0.87–1.13]
  Analysis sample sizeWomen, 71,04218.5–24.91.00
  25–29.90.89 [0.85–0.93]
  30–39.90.86 [0.81–0.91]
  ≥400.71 [0.59–0.85]
 Heo 200435 Sigmoidoscopy within past 5 y
  Location and dateUS, 2001BMI, kg/m2OR [95% CI]
  Population characteristicsAged ≥50 y; % white women NR; 82.3% white among total sample18.5–<251.00
  Analysis sample sizeWomen, 52,10625–<30NR
  30–<350.86 [0.78–0.94]
  35–<400.88 [0.79–0.99]
 Rosen 200437 Met CRC screening guidelines
  Location and dateUS, 1999BMI, kg/m2RD [95% CI]
  Population characteristicsAges 51–80 y; BMI ≥18.5 kg/m2; % white women NR; 81% white among total sample18.5–24.90
  Analysis sample sizeWomen, 28,120≥35−5.6 [−–8.5, −2.6]
  Endoscopic screen within past 5 y
  18.5–24.90
  ≥35−4.9 [−7.7, −1.9]
  FOBT within past 1 y
  18.5–24.90
  ≥35−3.7 [−6.2, −1.1]
 Seeff 200438 FOBT within past 1 y
  Location and dateUS, 2000BMI, kg/m2OR [95% CI]
  Population characteristicsAged ≥50 y; no prior history of CRC; % white women NR; 83% white among total sample<251.00
  Analysis sample sizeFOBT, 11,480 (6689 women); endoscopic screen, 11,588 (6774 women)25–290.99 [0.87–1.14]
  ≥301.07 [0.92–1.25]
  Endoscopic screening within past 10 y
  <251.00
  25–291.10 [0.98–1.23]
  ≥301.09 [0.92–1.25]
  Met CRC screening guidelines
  <251.00
  25–291.07 [0.96–1.20]
  ≥301.11 [0.98–1.27]
 Wee 200540 FOBT within past 1 y
  Location and dateUS, 2000BMI, kg/m2OR [95% CI]
  Population characteristicsAges 50–75 y; % white women NR; % white among total sample NR<18.50.7 [0.4–1.1]
  Analysis sample sizeTotal sample, 11,42718.5–<251.0
  25–<301 [0.9–1.1]
  30–<351.0 [0.8–1.2]
  35–<401.2 [0.9–1.5]
  ≥401.1 [0.8–1.5]
  Endoscopic screen within past 5 or 10 y
  <18.50.8 [0.5–1.2]
  18.5–<201.0
  25–<301.2 [1.0–1.3]
  30–<351.1 [1.0–1.3]
  35–<401.0 [0.7–1.2]
  ≥401.1 [0.8–1.5]
  Met CRC screening guidelines
  <18.50.7 [0.5–1.0]
  18.5–<251.0
  25–<301.1 [1.0–1.2]
  30–<351.0 [0.9–1.2]
  35–<401.1 [0.9–1.4]
  ≥401.1 [0.8–1.5]
 Menis 200636 Met CRC screening guidelines
  Location and dateMd, 2002BMI, kg/m2OR [95% CI]
  Population characteristicsAged ≥50 y; % white women NR; 79% white among total sample<251.00
  Analysis sample sizeTotal sample, 3436; women, 214325–29.91.05 [0.83–1.33]
  ≥300.84 [0.65–1.09]
Observational
 Slattery 200439 Sigmoidoscopy within past 10 y
  Location and dateNorthern Calif and Utah, 1991–1995 and 1997–2002BMI, kg/m2OR [95% CI]
  Population characteristicsAges 30–79 y; no prior history of CRC, familial polyposis, ulcerative colitis, or Crohn disease; % white women NR; 82% and 88% white for total sample among the rectal and Utah/KPMCP studies, respectively<251.00
  Analysis sample sizeWomen, 1231§25–292.3 [1.5–3.5]
  ≥301.8 [1.2–2.8]
 Ferrante 200634 Met CRC screening guidelines (including barium enema within past 5 y)
  Location and dateNJ and Pa, 2003–2004BMI, kg/m2OR [95% CI]
  Population characteristicsAged ≥50 y; alive and still a primary care patient at time of chart audit; % white women, NR; 80% white among total sample<301.00
  Analysis sample sizeTotal sample, 1297; women, 657≥300.75 [0.62–0.91]

RESULTS

Obesity Classifications

Obesity was measured most frequently by BMI, which was usually grouped according to standard cutoff points defined by the World Health Organization (WHO) as underweight (BMI <18.5 kg/m2), healthy weight (BMI 18.5–24.9 kg/m2), overweight (BMI 25–29.9 kg/m2), obesity Class I (BMI 30–34.5 kg/m2), obesity Class II (BMI 35–39.9 kg/m2), and obesity Class III or extreme obesity (BMI ≥40 kg/m2).41 Alternative categorizations used by some authors included combining underweight and healthy weight women,36, 38, 39 combining healthy and overweight women,19 and using a dichotomous measure of BMI (with obesity most often defined as ≥30 kg/m2).19, 26, 34, 42 Other measures of obesity included self-rated body type,18 ideal body weight (calculated as 3.83 × height in inches − 105.7; women whose body weight was >130% of their ideal body weight were considered obese)29 and the Ponderal Index (calculated as height in inches/cube root of weight in pounds).31

Breast Cancer Screening

Outcome definitions

‘Recent mammography’ (also referred to as ‘compliance with breast cancer screening’) was defined most often as having had at least 1 mammogram within the past 2 years,16, 17, 19, 21–25 although a mammogram within the past 3 years was used in 1 study.20 Age at first mammogram was examined in 1 study.18

Study designs

The majority of the mammography studies were cross-sectional and included data from single hospitals,18 regional mammography networks, health plans, and convenience samples16, 17, 19 and from national population surveys, including the BRFSS,21 the NHIS,20, 23–25 and the Health and Retirement Study (HRS).22 In most studies, mammography information, height, and weight were obtained from self-report; a few studies obtained these data through medical chart abstraction.17–19 In general, women aged >40 years16, 19, 21, 25 or aged >50 years17, 22–24 were included in the analyses.

Study findings

Three studies examined the relation between increasing categories of BMI and failure to have a recent mammogram separately among white women and black women.22, 24, 25 Among white women, these studies consistently demonstrated that women in the highest BMI categories were less likely to have had a recent mammogram than healthy weight women with odds ratios (ORs) for the most extreme body size categories ranging from 0.59 to 0.90.22, 24, 25 The results were less consistent among black women, with either no association22 or associations in the opposite direction of those observed among white women.24, 25 In studies that did not stratify by race, the observed associations between recent mammography and obesity were less clear. Two unstratified studies demonstrated a positive association between increasing BMI category and an increased likelihood of not having had a recent mammogram.21, 23 However, in 2 other unstratified studies, the authors reported no significant association between BMI and recent mammography,19, 20 although the trend was suggestive in the report by Fontaine et al.20

In the few studies that examined underweight women (BMI <18.5 kg/m2), several indicated that underweight women were less likely to be compliant with mammography screening than healthy weight women.21, 22, 25 In a race-stratified analysis, Ostbye et al. reported that underweight black women were less likely to have had a recent mammogram than healthy weight black women (OR, 0.71; 95% confidence interval [95% CI], 0.33–1.53 for black women) but observed a null association for underweight white women.22 In contrast, Zhu et al. observed a null effect among black women and a deterring effect of underweight on mammography compliance in white women.25

A few studies reported only screening rates or characteristics by body size. Amy et al. observed similar rates of screening in the past 2 years for women with a BMI from 25 kg/m2 to 35 kg/m2, from 35 kg/m2 to 45 kg/m2, and from 45 kg/m2 to 55 kg/m2 (80%–86%) but observed that screening rates were much lower among women with a BMI >55 kg/m2 (68%).16 Carney et al. reported that women who adhered to a 2-year screening protocol had a lower mean BMI compared with women who did not adhere to screening recommendations,17 and Colbert et al. reported that women with a larger body size were older when they initiated mammography use.18

Cervical Cancer Screening

Outcome definitions

Definitions of screening adherence for cervical cancer varied across studies. Six studies measured Pap test screening within the past 3 years,19, 20, 23, 27, 30, 32 and 4 studies measured Pap test screening within the past 2 years.16, 21, 22, 28 Two groups presented results for compliance with annual Pap screening.26, 42

Study designs

Most of the studies that examined cervical cancer screening and obesity were cross-sectional, and many used data from national surveys, including the NHIS,20, 23, 30, 32 BRFSS,21, 26, 27, 42 and HRS.22 Except for 2 studies that used medical chart review,19, 29 all data regarding timing and receipt of Pap tests, as well as body size measures, were self-reported. The majority of studies included women aged ≥18 years20, 21, 23, 26, 27, 30, 32, 42 or aged ≥21 years,16, 28 whereas a few studies were restricted to middle-aged or older women.19, 22, 31

Study findings

Several studies reported an inverse association between increasing categories of BMI and compliance with cervical cancer screening.19–23, 26, 28, 30, 32, 42 Three of the 5 studies that presented race-stratified results demonstrated that overweight and obese white women had decreased odds of screening compliance compared with healthy weight women, whereas the association for black women was weaker or was inconsistent across increasing categories of BMI.22, 23, 30 In contrast, a fourth study was a reanalysis of the Wee et al. analysis of 2000 NHIS data and demonstrated no associations for white women but an inverse association between increasing obesity and Pap test compliance for black women.32 A fifth race-stratified study came from the Black Women's Health Study, which includes women of higher educational attainment and socioeconomic status than the general U.S. population of black women. The results were in contrast to the majority of the race-stratified analyses in that, among blacks only, obese women were more likely to be noncompliant with cervical cancer screening than healthy weight women.28

Two studies by Fontaine et al. indicated an inverse association between increasing categories of BMI and cervical cancer screening among women of all races20, 21; 1 study reported no significant interaction between race and BMI category (although the P value for the interaction term was .07),21 whereas the other study made no comment on the possibility of effect measure modification by race.20 Among the studies that examined only a dichotomous measure of body size, obesity (measured as either BMI ≥30 kg/m2 or as BMI ≥27.3 kg/m2)19, 26, 42 was associated with decreased odds of having a recent Pap smear among all women.

Several studies also demonstrated a positive association between extreme body weight values and delayed cervical screening.19–21, 23, 30, 32 In a study that measured ideal weight, Lubitz et al. observed that women with a weight >200% of ideal body weight were 1.20 times as likely (95% CI, 0.58–2.47) to be noncompliant with cervical cancer screening than nonobese women.29 Amy et al. were the only authors to report results for cervical cancer screening among women with a BMI >55 kg/m2. Screening compliance was 18% lower compared with women who had a BMI from 25 kg/m2 to 35 kg/m2 in that study.16 Of the 3 studies that reported results for underweight women, 2 studies indicated that being underweight was associated with decreased cervical screening compliance.21, 28 In the third study, no association was observed for white women, whereas being underweight was associated with a large decrease in noncompliance among black women (OR for not having had a Pap smear, 0.26; 95% CI, 0.04–1.40).32

Colorectal Cancer Screening

Outcome definitions

Studies of colorectal cancer screening examined 1 or more of the colorectal screening modalities in relation to body size, including FOBT, sigmoidoscopy, and colonoscopy. Some studies measured whether study participants had “met screening guidelines” by completing at least 1 of the recommended screening tests during the recommended time frame. Adequate screening usually was defined as FOBT in the past year, sigmoidoscopy within the past 5 years, or colonoscopy within the past 10 years.

Study findings

We identified 6 cross-sectional studies,33, 35–38, 40 1 cohort study,34 and 1 case–control study39 that examined BMI in association with colorectal cancer screening compliance. Screening behavior, height, and weight data were ascertained by self-report in 7 studies33, 35–40 and by medical chart review in the remaining study.34

Five studies reported lower rates of adherence to colorectal cancer screening guidelines in women with higher BMI.33–37 In an analysis conducted in the Cancer Prevention Study II Nutrition Cohort, women with a BMI ≥25 kg/m2 were less likely to have undergone screening endoscopy in the past 5 years than women of healthy weight with ORs ranging from 0.71 to 0.89.33 Similarly, obese patients were less likely (OR, 0.75; 95% CI, 0.62–0.91) to have met colorectal cancer screening guidelines than nonobese patients in a study of primary care practice patients in New Jersey and Pennsylvania.34 Heo et al. reported that women with a BMI between 30 kg/m2 and 40 kg/m2 were less likely to have undergone screening sigmoidoscopy in the past 5 years compared with women with a BMI <25 kg/m2.35 An analysis from the 1999 BRFSS also indicated that there were lower rates of colorectal cancer screening among extremely obese women (BMI ≥35 kg/m2) compared with women who had a BMI <25 kg/m2. However, there were no substantial differences in screening rates in overweight women (BMI 25–29.9 kg/m2) or obese women (BMI 30–34.9 kg/m2; rates not reported).37

Three studies did not report lower compliance with colorectal cancer screening in overweight or obese women. Slattery et al. reported that obese women (BMI ≥30 kg/m2), in fact, were more likely to have undergone a sigmoidoscopy in the past 10 years than women with a BMI <25 kg/m2.39 Two analyses of the 2000 NHIS identified no clear relation between BMI and screening for colorectal cancer with the use of either FOBT or endoscopy.38, 40

DISCUSSION

We reviewed 32 published studies on the association between obesity and screening for breast, cervical and colorectal cancers. The studies of breast and cervical cancer screening generally supported the hypothesis that being overweight or obese is associated with decreased compliance with cancer screening recommendations, and the associations for breast cancer were slightly less consistent compared with the associations for cervical cancer screening. The literature for colorectal cancer screening was more variable, although approximately 50% of the reviewed studies indicated that there may be a similar inverse association between body size and screening behavior. Associations between body size and colorectal screening may be more difficult to elucidate because of the multiple modalities used in screening for colorectal cancer. Specific modalities may be recommended based in part on body size, and individual studies have not always investigated each modality individually.

Several analyses in the breast and cervical literature reported different results across racial groups, indicating that studies that did not stratify by race may have masked important findings. In the studies we reviewed, obesity appeared to be associated negatively with screening among white women, but the effect was less pronounced or was not observed among black women. These results are consistent with research suggesting that, compared with white women, black women report less body dissatisfaction, prefer a larger body size, and define a larger ideal body size.43, 44 In studies that did not stratify by race, the negative effects of obesity on screening behaviors that were observed among white women in race-stratified analyses may have been attenuated by including black women, among whom the effects generally were close to or at the null.

The role of obesity as a barrier to screening is a fairly recent research topic; and, perhaps because of this, we observed that there was a lack of consistency in study methodology that limited our ability to compare results across studies. For example, we observed that not all studies considered whether study participants were at risk for incident disease. In the cervical cancer literature, only 5 analyses excluded women who had undergone hysterectomy,23, 27, 28, 30, 32 although Pap smears are not recommended for women without a cervix.5–7 Simoes et al. stated that their results did not differ when women who reported undergoing a hysterectomy were excluded from the analysis, and they included these women in their final results.42 However, because of the high rate of hysterectomy among women of screening age,45 this may be an important exclusion criteria that was ignored in many studies. In contrast, for breast and colon cancer, the vast majority of women studied were at risk for incident disease.

Another variation between studies was that no single definition of overweight or obese was used consistently. Except for 2 early studies,29, 31 most studies used BMI as a measure of body size. BMI generally was used in combinations of standard categories set forth by the WHO.41 The effect of obesity on decreased screening compliance often was more pronounced in the highest categories of BMI, making it likely that analyses using dichotomized BMI or less precise categories at the upper range of BMI may have masked associations that would have been observed by examining women in the more extreme categories of obesity. Similarly, grouping underweight and healthy weight women into a single reference group erroneously may show no association between screening noncompliance and body size if underweight women also are less likely to be screened compared with healthy weight women, as demonstrated in some studies.21, 22, 25, 28, 40

Authors also adjusted their results for very different sets of confounders, making it difficult to compare across studies. The most extreme example involved the analyses by Wu et al. and Wee et al., who analyzed the 2000 NHIS data using different exclusion criteria and confounders.30, 32 Wee et al. reported that BMI was not associated with Pap testing among black women,30 whereas Wu et al. reported that black women with a BMI between 35 kg/m2 and 39.9 kg/m2 were almost twice as likely not to have had a Pap smear as healthy weight women.32 Wu et al. used the more appropriate exclusion criteria by excluding women with a history of cervical cancer and women who had had a Pap smear for a reason other than a routine examination,32 but Wee et al. reported a more systematic approach to the selection of confounders.30 This extreme example involving the direct comparison of 2 analyses from the same data with differing results indicates that readers should be cautious in comparing results across studies without careful attention to differences in study methodology.

A final methodological concern was that the vast majority of the studies reviewed here relied on self-reported height and weight data to define BMI. Although several of the authors cited older literature indicating high concordance for measured and self-reported height and weight values,46 a more recent review indicated that self-reported height tends to be overestimated and that weight tends to be underestimated.47 In particular, heavier women reportedly are more likely to under-report their weight and over-report their height, leading to underestimates of their true BMI.48 If this height and weight misclassification was present among heavier women in the studies we reviewed, then the reported ORs may have been overestimates of the true effects. Future studies using measured heights and weights would eliminate this particular methodological concern.

Overall, we found few studies that were designed to address the reasons why overweight and obese women are less likely to receive recommended cancer screening examinations and why there may be differences by race. In a study examining the reasons why women do not receive cancer screening examinations, Wee et al. reported many reasons that were not necessarily weight-related, including embarrassment, discomfort, not having thought about screening, being unaware that screening was necessary, and not having any problems.30 Few studies have focused specifically on the reasons why overweight or obese women may be less likely to receive screening services at the recommended time intervals. Fear of embarrassment in the examination room, negative reactions from healthcare providers, and lectures about weight were cited as reasons why overweight and obese women were more likely to cancel a physician appointment49 or were more reluctant to obtain Pap tests.50 Amy et al. reported that women commonly reported advice to lose weight and embarrassment about being weighed as “barriers to healthcare” and that the proportion of women reporting these barriers increased with increasing BMI.16 Those authors also observed that other common factors cited as “barriers to healthcare” were small gowns, tables, and equipment.16 Their survey of gynecologic healthcare providers corroborates the patient fears that there will be inadequate equipment for larger women; most providers reported access to longer speculums for pelvic examinations (80%) and larger examination tables (60%), but only 21% reported that larger gowns were available for large patients.16

A review by Zayat et al. suggested that obese patients were receiving less preventive care, because preventive care was not a top priority for patients or providers.51 In the report by Amy et al., 85% of gynecologic healthcare providers surveyed reported that caring for obese patients was more difficult than caring for others, although there was no direct evidence to suggest that healthcare providers would be less likely actually to provide preventive care to overweight or obese patients.16 In fact, Wee et al. observed that, among women who had seen a general practitioner or gynecologist in the past 12 months, a greater proportion of obese women reported having received a physician recommendation for Pap testing.30

Likely hypotheses presented to date regarding the reasons that overweight and obese women may be less likely to receive preventative screening include emotional barriers in patients,49 provider bias,49 competing healthcare needs of overweight and obese patients,51 and inadequate equipment for larger body sizes.16 However, there are few quantitative studies on this subject. Future studies that examine specific questions related to these hypotheses will be valuable in pin-pointing the factors that prevent all eligible women from participating in cancer screening programs. Identifying the extent to which these and other factors influence patient- and physician-related screening behaviors will provide direction for the development of public health interventions focused on weight-related health concerns and positive communication about weight between patients and providers.

Collectively, the literature reviewed here indicates that screening behavior may be affected by body size, although the extent of this problem varies by cancer site and by race. For breast and cervical cancers, screening rates are relatively high, and it most likely will prove difficult to reach the women who are not currently receiving screening examinations. However, because of the observed racial differences in obesity rates by race and the evidence linking differential screening behaviors to both race and body size, screening programs targeted to overweight and obese women may be useful in reaching those in greatest need of regular breast and cervical cancer screening. For colorectal cancer, however, screening rates remain low overall, and it is not clear that obese women are less likely to receive colorectal screening than healthy weight women. Thus, targeted outreach campaigns based on body size are not warranted when more general outreach programs still are advisable.

Acknowledgements

We thank Dr. Etta D. Pisano and Dr. Robert S. Sandler for helpful discussion during the writing of this article

Ancillary