Patterns of cancer screening in primary care from 2005 to 2010
Kathryn J. Martires MD,
Department of Graduate Medical Education, Scripps Mercy Hospital, San Diego, California
Department of Dermatology, Kaiser Permanente Los Angeles Medical Center, Los Angeles, California
Corresponding author: Kathryn Martires, MD, Department of Dermatology, Kaiser Permanente Los Angeles Medical Center, 1515 North Vermont Avenue, 5th Floor, Los Angeles, CA 90027; Fax: (323) 783-1629; firstname.lastname@example.org
We thank Andrew Lee, BS, for contributing to the statistical analysis and graphic representations in this study.
Cancer screening recommendations vary widely, especially for breast, prostate, and skin cancer screening. Guidelines are provided by the American Cancer Society, the US Preventive Services Task Force, and various professional organizations. The recommendations often differ with regard to age and frequency of screening. The objective of this study was to determine actual rates of screening in the primary care setting.
Data from the National Ambulatory Medical Care Survey were used. Only adult visits to non–federally employed, office-based physicians for preventive care from 2005 through 2010 were examined. Prevalence rates for breast, pelvic, and rectal examinations were calculated, along with the rates for mammograms, Papanicolaou smears, and prostate-specific antigen tests. Factors associated with screening, including age, race, smoking status, and insurance type, were examined using t tests and chi-square tests.
In total, 8521 visits were examined. The rates of most screening examinations and tests were stable over time. Clinical breast examinations took place significantly more than mammography was ordered (54.8% vs 34.6%; P < .001). White patients received more mammography (P = .031), skin examinations (P < .010), digital rectal examinations (P < .010), and prostate-specific antigen tests (P = .003) than patients of other races. Patients who paid with Medicare or private insurance received more screening than patients who had Medicaid or no insurance (P < .010).
Guidelines for cancer screening in the primary care setting vary, especially with regard to breast, skin, and prostate examinations. Recommendations for cervical screening are less controversial, although they are constantly evolving. Several organizations, including the American Cancer Society (ACS), the US Preventive Services Task Force (USPSTF), and major organizations within each field, offer recommendations guiding clinical care. The opinions of each professional society's recommendations often differ with regard to age of initiation and discontinuation of screening, frequency of screening, and screening tool.
Clinical breast examinations are recommended by the ACS and the American College of Obstetricians and Gynecologists (ACOG)[1-3] but are not recommended by the USPSTF or the World Health Organization because of the risks of anxiety and unnecessary biopsies.[4, 5] In contrast, all major organizations recommend mammography; however, controversy exists with regard to the age of initiation. The USPSTF recommended against routine screening with mammography in women aged <50 years in 2009, which resulted in significant controversy.
Screening for melanoma with routine skin examinations is recommended by the ACS. However, the USPSTF cites insufficient evidence to support screening. The Institute of Medicine and the American Academy of Dermatology recommend screening for those at high risk, including older, white men.[8, 9] Professional dermatology organizations are constantly examining this issue, attempting to provide better evidence of the benefit of routine screening.
Prostate cancer screening with the prostate-specific antigen (PSA) test and digital rectal examination (DRE) is no longer recommended by the American Urologic Association (AUA). Instead, the AUA recommends shared decision-making for PSA screening in men ages 55 to 69 years based on possible harms of over-diagnosis and overtreatment. Before this recommendation, the AUA had recommended routine annual screening with the PSA test. The USPSTF recommends against either the PSA test or DRE. In contrast, the ACS continues to recommend annual screening with the PSA test, but does not require the DRE.
Although strong evidence supports the ACOG, ACS, and USPSTF guidelines outlining the pathways of screening for cervical cancer using cytology,[2, 12, 13] the evidence does not support routine use of the pelvic examination for screening purposes. Its annual use is recommended by the ACOG, although the ACOG acknowledges the lack of evidence for its use alone in the asymptomatic patient. It is recommended against by the USPSTF for ovarian cancer screening.
In this study, we examine the most recent data representing the rates of screening examinations currently conducted in a nationally representative population of primary care practices. We compare the rates of examinations that are and are not recommended, and we also examine the factors that influence screening rates.
MATERIALS AND METHODS
Data from the National Ambulatory Medical Care Survey, a publicly available survey conducted by the Centers for Disease Control and Prevention, was used. We examined only visits to non–federally employed, office-based physicians (general, family practice, and internal medicine groups) for nonillness care and preventive care from 2005 through 2010 in adult patients aged ≥18 years. Visits were limited to routine physical examinations and general medical examinations. We excluded pregnant patients who were undergoing prenatal follow-up, and we also excluded any visits related to injuries. The rates of breast examinations and mammography were examined only among women aged >40 years, pelvic examinations and Papanicolaou (Pap) smears were examined among women ages 18 to 70 years, and DRE and PSA tests were examined among men ages 50 to 75 years. The rate of skin examinations was examined in patients aged ≥18 years and again in patients aged ≥50 years, because there are no guidelines for the age at which skin examinations should begin. The ranges were based on current screening recommendations.
The National Ambulatory Medical Care Survey data provide information from the nation's largest study of ambulatory care provided at physicians' offices. Participants are composed of office-based physicians randomly selected from information obtained from the American Medical Association and the American Osteopathic Association. The US Census Bureau acts as the field data-collection agent. Participants provide data through the use of survey instruments on 30 randomly selected patient visits during a randomly assigned, 1-week reporting period. Thirty visits are used from each physician to equalize reporting levels among the sample regardless of practice size. The data are collected either by the physician or by the physician's staff; although, in 51.5% of visits during 2009, for example, data were collected by Census field representatives trained in data collection. The field staff also makes completeness checks. In addition, manual clerical edits, during which ambiguous entries are reclassified or recoded, are performed upon the receipt of data for central processing. Computer edits for code ranges and inconsistencies also are performed. Error rates for coding range between 0% and 1%, depending on the survey item. Item nonresponse rates are ≤5%. Imputations are performed for missing data.
The survey instrument varies slightly from year to year but includes the following data: patient date of visit; zip code; birth date; sex; ethnicity; race; source of payment; tobacco use; medical history; vital signs, including height and weight; whether the visit is related to an injury or poisoning; whether the provider is the primary care physician or whether the provider has seen the patient before; the reason for the visit; the primary diagnosis or diagnoses; diagnostic tests, including physical examination, imaging, and blood tests; the procedures, health education, nonmedication treatment, medications, or immunizations provided; type of provider; time spent with the provider; and follow-up. All items are multiple-choice except for medications and immunizations, which may be written in by the provider. The questionnaire and related documentation are available online .
We calculated the prevalence rate for each of the examinations and tests among eligible visits per year. We used the Fisher exact test to compare the prevalence across years. We compared the prevalence of examinations from 2008 to 2010 with those from 2005 to 2007. In addition, we examined the factors associated with whether or not each of the examinations took place. These included age, ethnicity, race, sex, smoking status, source of payment (Medicaid and self-pay vs Medicare and private insurance), practitioner type (MD or DO vs nurse practitioner or physician assistant), and the number of chronic medical conditions. Race was divided into 4 groups: white, black, Asian, and other, which included American Indian, native Alaskan, native Hawaiian, or other Pacific Islander. Ethnicity was defined as Hispanic or Latino and non-Hispanic or non-Latino. Age and the number of chronic medical conditions were examined continuously. We used t tests for continuous variables and chi-square tests for categorical variables to examine the influence of each variable on each screening test or examination. The significance level was set at P < .05. Statistical analyses were conducted using the Statistical Analysis System (SAS) statistical software package (SAS Institute Inc., Cary, NC).
The prevalence rate of clinical breast examinations among eligible visits was significantly higher than that of mammography (54.8% vs 34.6%; P < .001). The rate of pelvic examinations did not significantly differ from that of Pap smears (P = .087). PSA tests were ordered significantly more often than rectal examinations were conducted (33.1% vs 23%; P < .001). The rate of breast examinations and pelvic examinations did not differ significantly (P = .125). Mammography was ordered significantly less often than Pap smears were conducted (P < .001). The rates of all screening tests or examinations were significantly higher than skin examinations among individuals aged ≥18 years (26.2%; P < .001), skin examinations among those aged >50 years (27.8%; P < .001), and rectal examinations (23%; P < .001), which did not differ in prevalence from one another (P = .106) (Fig. 1).
The rate of pelvic examinations decreased over time, with 59.9% prevalence at visits during 2005 to 2007 versus 56.7% prevalence during 2008 to 2010 (P = .016). The rate of skin examinations in adults aged ≥18 years increased over time from 25.1% to 27.1% (P = .032). It is noteworthy that this increase was most pronounced among patients aged ≥50 years, for whom the prevalence increased from 26.2% to 29.3% (P = .013). The prevalence of clinical breast examinations, mammography, Pap smears, rectal examinations, and PSA tests did not differ significantly across years (Fig. 2).
When examining age, younger patients (mean age, 46 years) received more clinical breast examinations than older patients (mean age, 50.6 years; P < .001), whereas older patients had mammography ordered more than younger patients (mean age, 53.7 years vs 47.6 years, respectively; P < .001). Older patients had more DREs done (mean age, 52.1 years vs 47.8 years; P < .001) and PSAs drawn (mean age, 59.9 years vs 47.9 years; P < .001) than younger patients. Younger patients had more pelvic examinations (mean age, 44.8 years vs 51.3 years; P < .001) and Pap smears done (mean age, 44.8 years vs 51.4 years; P < .001) than older patients. Patients with a higher number of chronic medical conditions (mean ± standard deviation: 1.94 ± 1.08 conditions) received Pap smears less often than those with a lower number of conditions (mean ± standard deviation: 1.63 ± 0.91 conditions; P < .001). There were no differences with regard to the number of chronic medical conditions for clinical breast examinations, mammograms, DREs, PSA tests, pelvic examinations, Pap smears, or skin examinations (Table 1).
Table 1. The Relation of Age and Number of Chronic Medical Conditions to Screeninga
Screening Examination or Tool
Abbreviations: DRE, digital rectal examination; Pap, Papanicolaou; PSA, prostate-specific antigen; SD, standard deviation.
Older patients received fewer breast exams, but more mammography. Older patients had significantly more DREs and PSA tests than younger patients. Younger patients had significantly more pelvic examinations and Pap smears than older patients. Patients with a higher number of chronic medical conditions received fewer Pap smears than those with a lower number of conditions.
White patients received more skin examinations or DREs than black, Asian, or other patients (all P < .001). Non-Hispanic, non-Latino patients received more skin examinations or DREs than Hispanic or Latino patients (P = .012 and P = .001, respectively). White and Asian patients also had mammography ordered (P = .031) and PSA drawn (P = .003) more than black and other patients. Non-Hispanic, non-Latino patients also had mammography ordered (P = .014) and PSA drawn (P < .001) more than Hispanic or Latino patients. There were no differences with regard to race or ethnicity for clinical breast examinations (Table 2).
Table 2. The Relation of Race and Ethnicity to Screeninga
Prevalence Rate, %
Prevalence Rate, %
Screening Examination or Tool
Hispanic or Latino
Abbreviations: DRE, digital rectal examination; Pap, Papanicolaou; PSA, prostate-specific antigen.
White patients received more DREs, PSAs, and skin examinations than black, Asian, or other patients. Non-Hispanic, non-Latino patients received more skin examinations, mammograms, DREs, and PSAs than Hispanic or Latino patients. Non-Hispanic, non-Latino patients had more mammograms, DREs, PSAs, and skin examinations than Hispanic or Latino patients.
Nonsmokers received more skin examinations (P = .003), breast examinations (P < .001), pelvic examinations (P = .022), and rectal examinations (P < .001) than smokers. They also had mammograms and (P = .002) and Pap smears (P = .022). There was no difference with regard to smoking status for patients who had PSA drawn (Fig. 3). Patients received more skin examinations (P = .017), rectal examinations (P = .031), Pap smears (P = .002), or mammograms (P = .008) by physician assistants or nurse practitioners than by physicians (Table 3). There was no difference with regard to practitioner type for clinical breast or pelvic examinations. Patients who paid with Medicare or privately received more screening than patients who had Medicaid or no insurance (self-pay) (P < .01). It is noteworthy that, when patients without insurance were excluded, the trend remained the same (P < .01) (Fig. 4).
Table 3. Patients Received More Skin Examinations, Digital Rectal Examinations, Papanicolaou Smears, and Mammograms by Nonphysicians (Physician Assistants or Nurse Practitioners) Than by Physicians (MD or DO)
Our study validates the disparity that exists in cancer screening with regard to race and socioeconomic status. Numerous studies have addressed this issue,[17-22] especially in recent years as the debate regarding universal health care has made its way to the national foreground. Studies have demonstrated that this disparity is minimized or does not exist in systems with universal access to care.[23, 24] We observed that white patients received all screening examinations and tests significantly more frequently than black, Asian, Hispanic, and other patients. These findings are in line with numerous studies demonstrating a lack of access to care among nonwhites, stigma preventing certain cultures from seeking routine medical care like gynecologic screening,[26, 27] and distrust and misunderstanding of the US medical system among particular ethnic groups. Many studies have demonstrated increased mortality for black patients with skin, breast, prostate, and cervical cancers compared with their white counterparts.[17, 18, 24, 29] This is especially worrisome for black men with prostate cancer, for whom studies indicate that mortality is 2 to 3 times higher than for white men.[30, 31] Newer studies demonstrate this disparity not only with black patients but also with southeast Asian and Hispanic patients. Those studies demonstrate why finding solutions to closing these gaps is especially imperative.
Not only did we observe a large disparity with regard to race and ethnicity, but our findings also confirmed the role of insurance in cancer screening. Patients who were paying with private insurance or Medicare were far more likely to receive all examinations and tests than patients who were paying with Medicaid or who were without insurance. When we excluded patients without insurance, the results were similar. Medicare coverage is considered by many to be superior to Medicaid programs because of accessibility, selection of providers, and reimbursement rates. Although both programs theoretically cover recommended cancer screening examinations, our data suggest that the screening of patients with Medicaid falls short. This is line with studies documenting the importance of insurance on cancer screening.
Many of our findings are consistent with current recommendations. This is especially true with regard to age. The mean age for starting clinical breast examinations was 46 years, whereas that for starting mammography was 51 years. Recommendations from the ACS suggest that yearly breast examinations and mammograms should begin at age >40 years.[1, 2] Although the USPSTF does not recommend breast examinations, it does suggest that mammograms begin at age 50 years. The mean ages for starting DREs and PSA tests were 52 years and 60 years, respectively, which fell well within the commonly recommended age range of 50 to 75 years.[35, 36] In the era of cost-effective medicine, however, it may become increasingly important to act based on strict evidence based-recommendations. A recent National Ambulatory Health Care (NAMCS) study demonstrated that many services, including breast and prostate screening, continue to be over utilized.
One such screening measure may include clinical breast examinations, the rate of which was higher than that for mammography, despite research suggesting no additional mortality benefit with clinical breast examinations. The sensitivity of detecting breast cancers with mammogram versus mammogram plus clinical breast examination is similar; however clinical breast examinations lead to higher rates of false-positive results. Because of such evidence, the USPSTF does not recommend routine clinical breast examinations. They are still recommended, however, by the ACS and the ACOG,[1, 3] likely based on slight gains in sensitivity of detection.
In our study, the rate of pelvic examinations did not differ significantly from that of Pap smears. Although the definition of a pelvic examination is not entirely clear from NAMCS data, we interpreted this as distinct from a Pap smear, entailing an external visualization, speculum examination, and/or bimanual pelvic examination for use in cervical and ovarian cancer screening. Whereas cytology with a Pap smear is routinely recommended by most professional organizations every 3 years,[12, 13, 41] evidence supporting the role of pelvic examinations is lacking. Despite this, our study is similar to a recent study of US gynecologists demonstrating that screening with bimanual pelvic examinations is performed routinely. Forty-seven percent of survey participants in that study cited detection of ovarian cancer as the reason for performing the examination. The evidence indicates that the likelihood of detecting ovarian cancer with screening is so low that it is recommended against by the USPSTF. In addition, pelvic examinations often are associated with significant embarrassment, discomfort, and anxiety. Our data indicate perhaps some concordance with such evidence, as the rate of pelvic examinations is slowly declining.
Despite the rising incidence of melanoma,[45, 46] skin cancer screening is currently not recommended by the USPSTF because of a lack of evidence of mortality benefit.[6, 47] However, it is recommended by the ACS. Skin cancer screening by primary care physicians is not routinely provided. More recent evidence suggests, however, that physician detection is associated with thinner melanomas and a significant survival advantage.[7, 49] Although our data demonstrate that skin cancer screening has significantly increased over a 10-year period, the rates of other examinations that similarly lack evidence of mortality benefit, including breast and pelvic examinations,[39, 42] continue to be significantly higher than the rate for skin cancer screening. This is consistent with findings from a NAMCS study that examined skin cancer screening in 1997.
Lack of time is commonly cited as a reason why screening is not performed, particularly for skin, breast, and prostate cancers.[52-55] To our surprise, the number of chronic medical conditions did not seem to have a significant influence on the rate of examinations or tests. This was true in all cases except Pap smears, for which a higher number of chronic medical conditions meant a lower rate of examinations. This may be explained by our inclusion of only routine physical examinations in our study, which may not have addressed the management of chronic issues. However, this result is still promising. According to our study, a patient's complex medical history does not preclude conducting recommended routine cancer screening in the setting of scheduled preventive visits. This is in contrast to other studies, which have indicated that patients with chronic conditions may experience neglect of unrelated disorders during primary care visits.
Also unexpected in our study was the finding that smokers were less likely to be screened than nonsmokers. It is well established that smoking is a risk factor for and plays a significant role in the pathogenesis of most cancers. It is possible that smokers were screened less often because of the presence of additional comorbidities associated with smoking[58, 59] as well as the extra time it may have taken to counsel smoking cessation. Smokers may also be less likely to consult physicians for preventive health maintenance visits. Regardless of the reason, this finding highlights a need for the risk stratification of patients who require screening. For example, although routine skin cancer screening is not recommended, evidence suggests that screening in at-risk populations, such as elderly white men, is likely to yield a significant mortality benefit compared with screening all patients. Risk stratification should be based on other factors as well, such as family history.
Our study also demonstrates the changing role of the physician in the practice of primary care. Significantly more screening examinations and tests took place when the visit was conducted by a nurse practitioner or a physician assistant rather than by a physician. Nurse practitioners and physician assistants conducted routine screening examinations more often than physicians, whereas physicians addressed more acute complaints and chronic illness visits. This speaks to the expanding role of nurse practitioners and physician assistants in primary care. More and more practices, not only in the community but at academic centers as well, are integrating nonphysician practitioners as part of an effort to address the primary care shortage. The role of the nurse practitioner is not clearly defined.[63, 64] Our study documents the increasing importance of nonphysician practitioners in providing preventive care and demonstrates an area in which the efforts of physician assistants and nurse practitioners may be especially realized.
Limitations of this study are related to our use of data from a publicly available survey instrument. The survey form is completed by the provider, provider staff, or NAMCS field staff. Variables in the survey form were largely objective, because they were based on data from the visit. Although this helps to minimize some bias traditionally associated with questionnaires administered to patients, survey of the providers may be associated with variability in how the survey form was completed. For example, some may have completed forms during or immediately after the visit, whereas others may have completed them at a later time, making the data more prone to recall bias. Providers also knew they were being observed during the week of visits and may have altered their clinical behavior, whether intentional or not. In addition, we were limited by the data available from the survey form. We would have liked to include data regarding colon cancer screening; however, the survey form did not explicitly collect data for fecal occult blood tests, sigmoidoscopy, or gastrointestinal referral. Although the survey form did collect data for colonoscopy, in our preliminary analysis, we observed that it was seldom ordered. We hypothesized that this was because of the recommendation for routine colonoscopy only every 10 years and the availability of flexible sigmoidoscopy as an acceptable alternative to colonoscopy for cancer screening. It is also a procedure that is not consistently ordered by primary care providers themselves but, rather, upon referral to gastroenterology. We included only screening examinations and tests in our study that were likely done or ordered by the provider at the time of the primary care visit.
Cancer screening is a key component of primary care. The goal of cancer screening is to balance the risk of diagnostic workup with mortality benefit. In our population-based study representing primary care practices across the United States, we identified trends in cancer screening practices that have important implications for the practice and delivery of health care. It is important to document patterns of cancer screening to identify areas of strength and areas that need significant improvement.
This study was supported by an American Society for Dermatologic Surgery Cutting Edge Research Grant.