Over the past 20 years, prevention has become a major activity in primary care. During a typical day, primary care clinicians spend much of their time managing asymptomatic conditions in which the main goal is to prevent death or complications (e.g., hypertension, hyperlipidemia, osteoporosis). Many topics in ACP Medicine include information on screening or prevention of specific disorders in asymptomatic patients or those at increased risk [see Table 1]. This topic focuses primarily on preventive screening recommendations from the United States Preventive Services Task Force (USPSTF).
Rationale and Evolution of Preventive Care Guidelines
The rationale for delivering preventive care during an office visit is strong. In 2002, life expectancy in the United States was 77.4 years, an all-time high.1 Behavioral risk factors, including tobacco use, diet, and alcohol use, as well as factors such as hyper-lipidemia and hypertension, contributed to the most frequent causes of death [see Table 2]. From the viewpoint of clinical preventive services, modifiable risk factors such as these, rather than the diseases they affect, are the true causes of death.2
Primary care visits provide an opportunity to assess risk, discuss options, and recommend behaviors and treatments that have been proved to reduce the risks of diseases and death. During 2002, an estimated 558 million visits were made to primary care physicians in the United States, an overall rate of about two visits per person per year.3 On average, these physicians spent 20 minutes with the patient at each visit.
In 1975, Frame and Carlson published a series of articles that examined the quality of evidence for periodic screening conducted in the routine physical examination.4 These authors argued that any preventive strategy should meet certain criteria of accuracy and usefulness [see Table 3]. The criteria are helpful in understanding the controversy about screening proposals. Several scholars have pointed out that clinical intuition about screening is often wrong, leading to errors in inference about the effects of screening. Some of these logical fallacies and hidden assumptions are now well recognized and even find their way into board examinations [see Table 4].
The work of Frame and Carlson gave rise to evidence-based decision making in prevention. The Canadian Task Force on the Periodic Health Exam used independent reviews of the scientific literature and a set of rules to grade the strength of evidence supporting a clinical service.
The USPSTF, founded in 1984, was modeled on the Canadian Task Force. It published its first set of guidelines for clinical preventive services in 1989.5 The current USPSTF has experts from the specialties of family medicine, pediatrics, internal medicine, obstetrics and gynecology, geriatrics, preventive medicine, public health, behavioral medicine, and nursing.
Other expert panels also make recommendations about prevention [see Table5]. Despite general agreement that recommendations should be evidence based, opinions about the effectiveness of specific preventive services differ. These differences arise because interpretation of the evidence is ultimately a subjective process, especially regarding the balancing of benefits and risks—an equation that includes such disparate factors as mortality reduction, costs or burden of illness, and patient discomfort.
To avoid errors in judging the evidence and weighing benefits and harms, expert panels, as well as individual clinicians, should do the following: (1) use an independent systematic review to distinguish assertions based on evidence from those based on other grounds, (2) make the rationale for a recommendation explicit, and (3) be free from financial and political conflicts of interest. Although the use of these measures does not guarantee a correct decision, they represent the best safeguards against bias.
USPSTF Evidence Ratings
The USPSTF assigns an overall grade of A, B, C, D, or I to each prevention service. The grades reflect the overall strength of evidence and the magnitude of benefit, defined as benefits minus harms [see Table 6].
A grade of A indicates services that have solid supporting evidence and at least a moderate net benefit. A grade of B suggests that there are information gaps (so-called fair evidence) or that the benefits are only moderately greater than the harms for all patients. A grade of C denotes a toss-up, whereas a D grade indicates a service that is either proven ineffective or unlikely to provide benefits that outweigh the harms.
When there is too little evidence to determine whether or not a service works, the USPSTF assigns a grade of I for insufficient evidence. Some of the services with an I grade make good clinical sense and some are very promising, but without better research, it is not possible to say with confidence that they improve outcomes. Other grade I services have uncertain benefits but definite harms.
Noncancer Prevention Imperatives
Several preventive measures have earned an A grade on the strength of their good supportive evidence, substantially greater benefits than harms, and broadest applicability to primary care practice [see Table 7].
Immunization
The USPSTF has not issued recommendations about immunization since 1996, and those recommendations are now out of date. The Advisory Committee on Immunization Practices (ACIP), which consists of 15 experts in fields associated with immunization, is currently the only entity in the United States federal government that makes recommendations about immunizations. In contrast to the USPSTF, the ACIP does not use systematic reviews and does not usually describe the quality of evidence supporting a recommendation.
The ACIP publishes schedules for vaccination against certain infectious diseases in adults, depending on age and risk factors; these recommendations are discussed in individual ACP Medicine topics and are available on the Internet (http://www. cdc.gov/nip/recs/adult-schedule.pdf). For example, general recommendations include a tetanus-diphtheria booster every 10 years in all adults, influenza vaccination every year in adults 50 years of age and older, and pneumococcal vaccination once in adults 65 years and older.
Cancer Prevention
Only two cancer screening tests meet the USPSTF criteria for a strong recommendation: (1) Papanicolaou (Pap) smears for cervical cancer and (2) fecal occult blood testing or endoscopic procedures for colorectal cancer [see Table 9]. With both of these conditions, the aim of screening is to remove precancerous lesions, which prevents invasive cancer, saves the involved organ, and reduces disease-specific mortality. By contrast, the more controversial cancer screening tests, such as prostate-specific antigen (PSA) and mammography, detect invasive cancers and lead to aggressive treatments (prostatectomy and mastectomy) that often destroy the involved organ and that have more substantial morbidity than cone biopsy for cervical cancer and polypectomy for colorectal cancer.
Cervical cancer
Although no data from randomized, controlled trials support the value of the Pap smear in reducing mortality from cervical cancer, indirect evidence suggests that it is among the most effective cancer screening techniques.8 By current standards, the sensitivity of traditional Pap testing is low (51%).9,10 Cervical dysplasia is slow to progress to invasive carcinoma, however, so periodic screening can make up for the low sensitivity of a single exam.
The specificity of Pap smears for detection of dysplasia and cancer is 98%. False positive results occur infrequently, but Pap smears may correctly detect a large number of low-grade lesions that, without treatment, would remain stable or regress.11 As a consequence, many women who would never develop invasive cervical cancer are subjected to anxiety and to colposcopy and biopsy.
In a systematic review, the effectiveness of liquid-based cytology, computerized rescreening, and algorithm-based screening have been compared with that of conventional Pap smear screening in reducing the incidence and mortality of invasive cervical cancer. The review concluded that the liquid-based monolayer preparation (ThinPrep) appears to offer higher sensitivity but lower specificity than conventional Pap smears.10 However, the USPSTF could not determine whether the potential benefits of the three new screening approaches relative to conventional Pap smears are sufficient to justify a possible increase in potential harm or cost. They also found insufficient evidence to recommend for or against the routine use of human papillomavirus testing as a primary screening test for cervical cancer.
Colorectal cancer
Screening modalities for colorectal cancer include fecal occult blood testing (FOBT), sigmoidoscopy, double-contrast barium enema, colonoscopy, and computed tomographic colonogra-phy [see 12:V Colorectal Cancer].
Table 2 Major Causes of Death in the United States1*
|
Cause of Death |
Number of Deaths |
Age-Adjusted Death Rate (per 100,000 population) |
|
Diseases of the heart |
695,754 |
204.4 |
|
Malignant neoplasms |
558,847 |
194 |
|
Cerebrovascular diseases |
163,010 |
56.3 |
|
Chronic lower respiratory diseases |
125,500 |
43.7 |
|
Accidents (unintentional injuries) |
102,303 |
35.3 |
|
Diabetes mellitus |
73,119 |
25.4 |
|
Influenza and pneumonia |
65,984 |
22.7 |
|
Alzheimer disease |
58,785 |
20.2 |
*Preliminary data for 2002; these causes account for three quarters of all deaths.
FOBT is the only screening modality that has been shown in randomized controlled trials to reduce colorectal cancer mortality. In the Minnesota Colon Cancer Control Study, 33 volunteers 50 to 80 years of age were randomized to annual FOBT, biennial FOBT, or a control group. After 18 years of follow-up, colorectal cancer mortality was 33% lower in the annual FOBT group and 21% lower in the biennial group than in the control group.12 In this study, the slides were rehydrated, a technique that increases sensitivity but reduces specificity; during the trial, 38% of patients in the annual FOBT group underwent colonoscopy because of a positive test result. Two randomized, controlled trials from Europe have demonstrated 16% and 18% reductions in colorectal cancer mortality using FOBT.13,14 In the European trials, unlike in the Minnesota study, patients were drawn from the general population, the slides were not rehydrated, and all testing was biennial.
In the screening trials, FOBT reduced mortality from colon cancer but did not reduce all-cause mortality. For example, the Minnesota trial findings indicate that 10 years of screening would result in 12 (95% confidence interval, 1 to 24) fewer colon cancer deaths per 10,000 persons screened. In that trial, however, the 95% confidence interval for all-cause mortality was 334 to 350 with annual screening, 333 to 348 with biennial screening, and 336 to 351 in control subjects.12
Evidence for the efficacy of sigmoidoscopy comes from case-control studies, which suggest that the protective effect of a single sigmoidoscopy lasts at least 6 years. The results of a large United Kingdom trial of screening with flexible sigmoidoscopy are not yet complete. Preliminary results suggest that flexible sigmoidoscopy is safe and that about 5% of persons 55 to 64 years of age have high-risk polyps (three or more adenomas; size 1 cm or greater; villous, severely dysplastic, or malignant).15
Because of the imperfect sensitivity of FOBT and sigmoid-oscopy and because many patients who undergo these procedures end up requiring colonoscopy anyway, many clinicians are advising their average-risk patients to undergo colonoscopy as a screening test, either as a one-time procedure or periodically (e.g., every 10 years) beginning at age 50. Colonoscopy is the most sensitive test for detecting polyps; however, as for other slow-growing lesions, such as cervical dysplasia, it is not clear whether improved sensitivity for polyps at a single point in time will translate into fewer invasive cancers in the long run.
In 2002, for the first time, the USPSTF included screening colonoscopy as an option, but with the qualification that the potential added benefits of colonoscopy may not always be great enough to justify the increased risks and inconvenience.16 All colon cancer screening tests have a low yield—over 500 patients must be screened to prevent one invasive cancer17—so even a slightly increased rate of serious complications with colonos-copy might negate the benefit. Several gaps in the evidence base for colonoscopy can also be mentioned. First, the frequency of one procedure every 10 years was arrived at by means of mathematical models; in fact, no one knows how many patients will develop invasive cancer less than 10 years after a negative colonoscopy. Second, surveys suggest that gastroenterologists overuse colonoscopy for surveillance in patients who have clinically insignificant hyperplastic polyps or low-risk lesions, such as small adenomas. As a result, colonoscopic screening may lead to the use of a scarce, expensive resource, primarily in patients who have little chance of benefit. Third, the accuracy of colonoscopy when performed by the so-called average colonos-copist is not known. The primary advantage of colonoscopy, visualization of the entire colon, is negated if the operator cannot reach the cecum consistently or does not view the entire circumference of the lumen during the procedure.
No direct evidence supports the use of double-contrast barium enema for screening, and patients find it more uncomfortable than other alternatives. CT colonography may prove to be more sensitive and better tolerated than double-contrast barium enema and safer than colonoscopy; as of yet, however, there are insufficient data to determine whether it would result in better outcomes.18
Breast cancer
It was predicted that in the United States in 2004, invasive breast cancer would be diagnosed in an estimated 215,990 women; in situ disease would be diagnosed in 55,700 women; and 40,110 women would die of the disease.19 A 40-year-old woman has a 13.2% (approximately one in eight) chance of developing invasive breast cancer during her life, but her risk of developing breast cancer within 10 years is only 1.47% (approximately one in 68). Modalities for breast cancer screening include mammography, clinical breast examination, and breast self-examination.
|
Table 3 Criteria for Evaluating a Screening Program |
|
1. Does the program target a disease that causes serious morbidity and mortality that might be prevented by the service? |
|
2. Can the screening test accurately identify healthy people who are at high risk for developing advanced disease? |
|
3. Is the screening test feasible to use in primary care? |
|
4. Does treatment given before symptoms occur result in better outcomes than treatment given later? |
|
5. Do the overall benefits outweigh the harms of screening and treatment? |
Table 4 Sample Board Examination Questions About Screening
|
Question |
Answer and Explanation |
|
A screening test correctly identifies 95% of patients who have prexero-stosis and 95% of patients who are well. If 1 of every 500 patients has prexerostosis, what is the likelihood that a patient who has a positive test has the disease? |
The correct answer is 3%; the positive predictive value is commonly overestimated because of neglecting Bayes theorem. |
|
The 5-year survival of stage 0 lung cancer is 95%, versus 10% for more advanced stages. In usual care, 70% of patients present in advanced stages. When screening with a CT scan, 90% of patients have stage 0 disease. By how much will screening reduce mortality? |
The correct answer is that the effect of screening on survival cannot be determined; increasing detection of disease in a "curable stage" may improve 5-yr survival but does not necessarily reduce mortality because of overdiagnosis bias, length bias, and lead-time bias—for example, screening may detect slower-growing cancers that would never have become lethal. |
|
With improvements in treatment, mortality from advanced HIV infection has dropped by 63%. Because effective treatment is now available, screening and early treatment should result in even greater mortality reductions. True or false? |
The conclusion may be, but is not necessarily, true. If treatment of advanced disease is very effective, screening may not confer any additional advantage. Screening is most likely to improve outcomes when advanced disease is untreatable but treatment of earlier, asymptomatic disease can result in cure. |
Mammography
In 2000, a Danish meta-analysis of the major randomized trials of mammography concluded that there was no evidence that mammography reduced mortality from breast cancer.20 However, another analysis of the same trials conducted for the USPSTF concluded that mammography reduced breast cancer mortality in women 40 to 70 years of age.21 The controversy centered on disagreement about the quality of the randomized trials of mammography: the Danish investigators excluded five of the eight trials that showed mammography to be beneficial, whereas the United States investigators excluded only two of those eight trials on grounds of quality.
The USPSTF demoted mammography from grade A to grade B to reflect their view that the quality of the evidence was fair and that the net benefit (benefits minus harms) was moderate. Coming after the widely publicized Danish study, the USPSTF recommendation of grade B for mammography received a mixed reception. One independent review, published in 2003, confirmed the USPSTF view that although the trials were flawed, the balance of the evidence still favored screening mam-mography in women 40 years of age and older at least every 2 years.22 Conversely, the National Cancer Institute’s Physician Data Query program largely endorsed the idea that most of the mammography trials were seriously flawed.
The USPSTF’s most controversial decision regarding mam-mography was to promote screening in women 40 to 50 years of age from a grade C to a grade B. This was done because with several additional years of follow-up since the previous recommendations, in 1996, the pooled risk reduction for women who began screening at this age had become statistically significant. Nevertheless, the number needed to screen is higher, and the balance of benefits and harms narrower, in women 40 to 50 years of age than in older women.
For clinicians, the most difficult question is how to present information about the risks and benefits clearly and fairly to patients. At the time of an earlier controversy over the effectiveness of mammography in women 40 to 49 years of age, a survey of 509 women in the United States found that most believed the controversy was really about cost.
Table 5 Government-Sponsored Preventive Guidelines Programs
|
Organization |
Sponsorship |
Focus |
Web Site |
|
U.S. Preventive Services Task Force |
Agency for Healthcare Research and Quality |
Clinical preventive services |
http://www.ahrq.gov/clinic/uspstfix.htm |
|
Canadian Task Force on Preventive Health Care |
Health Canada (Canadian Federal Government) |
Clinical prevention, periodic health examination |
http://www.ctfphc.org |
|
Physician Data Query Program |
National Cancer Institute |
Cancer prevention |
http://cancernet.nci.nih.gov/cancertopics/pdq/screening |
|
Task Force on Community Preventive Services |
CDC |
Community, population, and health care system strategies |
http://www.thecommunityguide.org |
|
Advisory Committee on Immunization Practices |
DHHS and CDC |
Immunizations, bioterrorism response |
http://www.cdc.gov/nip/recs/adult-schedule.pdf |
|
National Heart, Lung, and Blood Institute |
National Institutes of Health |
Asthma, cholesterol, hypertension, obesity |
http://www.nhlbi.nih.gov/guidelines |
|
Board on Health Promotion and Disease Prevention |
Institute of Medicine |
Population-based public health measures and the public health infrastructure |
http://www.iom.edu/board.asp?id=3793 |
CDC—Centers for Disease Control and Prevention
DHHS—Department of Health and Human Services
Table 6 United States Preventive Services Task Force Grading System6
|
Grade* |
Strength of Evidence |
Magnitude of Benefit |
|
A |
Good |
Large |
|
B |
Good Fair |
Moderate Moderate to large |
|
C |
Fair to good |
Small |
|
D |
Fair to good |
None |
|
I |
Poor |
None to large |
*A—Service strongly recommended
B—Service recommended
C—No recommendation for or against
D—Service not recommended
I—Insufficient evidence
Women may interpret the lifetime risk of one in eight to be their immediate risk of developing breast cancer if they defer or miss their next mammo-gram.24 In deciding how to inform patients, clinicians should carefully consider the major criticisms of the USPSTF recommendation. These criticisms represent differences in values rather than disagreements over the facts. There are four principal issues:
1. Is reducing breast cancer mortality important? In the trials, which involved nearly half a million women, mammography clearly had no effect on all-cause mortality. The USPSTF, although fully aware of this fact, chose to base their assessment of the benefits on the narrower grounds of breast cancer mortality. They chose to let women decide for themselves whether reducing the risk of dying of breast cancer was important to them.
2. How large is the reduction in breast cancer mortality? Judging from the trials, about 1,200 women 40 to 70 years of age must be invited to be screened four to five times over 10 years to prevent one death from breast cancer. Of women 40 to 49 years of age, 1,792 (95% CI, 764 to 10,540) must be invited to be screened to prevent one death from breast cancer, a death that would not have occurred until about 20 years after screening began. The specification "invited to" is important: it is likely that the trials underestimated the true benefit because they are diluted by a large number of subjects who were assigned to have mammography but did not.25 Nevertheless, to benefit even one woman, a large number of women must have a large number of mammograms over many years.
3. Are these estimates from the randomized trials still valid? Evidence from the trials may be out of date. The first trial began in 1963, and the others began between 1976 and 1982. Improvements in mammography since then might translate into better outcomes than were seen in the trials. On the other hand, improved systemic treatment for clinically detected breast cancer may have eliminated the advantage that earlier detection conferred in the era of the trials.
4. How large are the harms? The USPSTF was criticized for ignoring or underestimating harms. In fact, the USPSTF considered the harms, but it was also influenced by evidence that many healthy women stated that they would be willing to take on these risks, as well as the morbidity associated with treatments, to avoid a breast cancer death.26
What are the harms? Women who get 10 annual mammo-grams have about a 50% chance that at least one of them is a false positive result; many of these false positive results necessitate a biopsy. Of women who are found to have invasive cancer, about 30 must undergo major surgery, or surgery plus radiation or tamoxifen, to prevent one death from breast cancer. In addition, screening identifies many women with ductal carcinoma in situ, and many of these women also undergo surgery, with uncertain benefit. In sum, many women experience immediate morbidity from treatment; without screening, most of them would not have had consequences of their breast cancer (and no morbidity from mastectomy) for many years, if ever.
Table 7 Strongly Recommended Noncancer Preventive Services in Adults*
|
Service |
Candidates |
Established Benefits |
|
Aspirin for primary prevention of cardiovascular events |
Adults at high cardiovascular risk |
Reduces the risk of stroke |
|
Blood pressure screening |
All adults |
Reduces the risk of stroke |
|
Screening for lipid disorders |
Men 35 yr of age and older; women 45 yr of age and older; and younger adults at increased risk for coronary artery disease |
Reduces overall mortality, as well as mortality from cardiovascular disease |
|
Chlamydial infection screening |
Sexually active women 25 yr of age and younger; other asymptomatic women at increased risk for infection |
Reduced the risk of pelvic inflammatory disease in one randomized trial36 |
|
Hepatitis B virus (HBV) infection screening |
Pregnant women |
Reduces prenatal transmission of HBV |
|
Syphilis screening |
Persons at increased risk for infection; all pregnant women |
Penicillin treatment during pregnancy reduces the risk to the fetus of acquiring congenital syphilis |
|
HIV screening^ |
Pregnant women High-risk men and women |
Reduces prenatal transmission of HIV Delays mortality from HIV disease and permits counseling to reduce transmission |
|
Screening for asymptomatic bacteriuria |
Pregnant women (urine culture at 12-16 weeks’ gestation) |
Prevents symptomatic urinary tract infections, low birth weight, and preterm delivery |
*As per the United States Preventive Services Task Force.
As per the CDC.
Table 8 Recommended Vaccination Schedule for Health Care Workers7*
|
Vaccine |
Schedule |
|
Tetanus-diphtheria |
Every 10 yr after complete primary series or for persons lacking documentation of vaccination |
|
Influenza |
Annual |
|
Pneumococcal (polysaccharide) |
For persons with medical indications^ or at risk for exposure |
|
Hepatitis B |
For persons lacking documentation of vaccination or evidence of disease |
|
Hepatitis A |
No data to support a recommendation |
|
Measles, mumps, rubella (MMR) |
For persons lacking documentation of vaccination or history of disease} |
|
Varicella |
For persons lacking documentation of vaccination or history of disease |
*As per the Advisory Committee on Immunization Practices.
tMedical indications: Chronic pulmonary disorders, excluding asthma; cardiovascular disease; diabetes mellitus; chronic liver disease, including liver disease as a result of alcohol abuse (i.e., cirrhosis); chronic renal failure; chronic renal failure or nephrotic syndrome; functional or anatomic asplenia (e.g., sickle cell disease or splenectomy); lymphoma, multiple myeloma, generalized malignancy, or organ or bone marrow transplantation; chemotherapy with alkylating agents, antimetabolites, or long-term systemic corticosteroids; or cochlear implants.
tMeasles component: Adults born before 1957 can be considered immune to measles. Health care workers born during or after 1957 should receive two doses of MMR vaccine unless they have a medical contraindication, documentation of one or more dose, or other acceptable evidence of immunity. Mumps component: One dose of MMR vaccine should be adequate for protection. Rubella component: Administer one dose of MMR vaccine to women whose rubella vaccination history is unreliable, and counsel women to avoid becoming pregnant for 4 wk after vaccination. For women of childbearing age, routinely determine rubella immunity and counsel regarding congenital rubella syndrome. Do not vaccinate pregnant women or those planning to become pregnant during the next 4 wk. For women who are pregnant and susceptible, vaccinate as early as possible in the postpartum period.
