INTRODUCTION — Reproduced and adapted with permission from: Fletcher GS. Clinical Epidemiology: The Essentials, 6th Edition, Lippincott Williams & Wilkins 2020. For more information, please visit www.lww.com.
“If a patient asks a medical practitioner for help, the doctor does the best he can. He is not responsible for defects in medical knowledge. If, however, the practitioner initiates screening procedures, he is in a very different situation. He should have conclusive evidence that screening can alter the natural history of disease in a significant proportion of those screened.”
Archie Cochrane and Walter Holland, 1971
Most doctors are attracted to medicine because they look forward to curing disease. But all things considered, most people would prefer never to contract a disease in the first place; or, if they cannot avoid an illness, they prefer that it be caught early and stamped out before it causes them any harm. To accomplish this, people without specific complaints undergo interventions to identify and modify risk factors to avoid the onset of disease or to find disease early in its course so that early treatment prevents illness. When these interventions take place in clinical practice, the activity is referred to as preventive care.
Preventive care constitutes a large portion of clinical practice [1]. Clinicians should understand its conceptual basis and content. They should be prepared to answer questions from patients such as, “How much exercise do I need, Doctor?” or “I heard that a study showed antioxidants were not helpful in preventing heart disease. What do you think?” or “There was a newspaper ad for a calcium scan. Do you think I should get one?”
This topic will discuss the principles that support the development of recommendations for preventive care. Specific recommendations for preventive care in adults and children are discussed in numerous other topics in UpToDate. A selection of these topics includes:
●(See "Overview of preventive care in adults".)
●(See "Geriatric health maintenance".)
●(See "Overview of cancer prevention".)
●(See "Screening tests in children and adolescents".)
PREVENTIVE ACTIVITIES IN CLINICAL SETTINGS — In the clinical setting, preventive care activities often can be incorporated into the ongoing care of patients, such as when a doctor checks the blood pressure of a patient complaining of a sore throat or orders pneumococcal vaccination in an older person after dealing with a skin rash. At other times, a special visit just for preventive care is scheduled; thus the terms annual physical, periodic checkup, or preventive health examination.
Types of clinical prevention — There are four major types of clinical preventive care: immunizations, screening, behavioral counseling (sometimes referred to as lifestyle changes), and chemoprevention. All four apply throughout the lifespan.
Immunization — Childhood immunizations to prevent 15 different diseases largely determine visit schedules to the pediatrician in the early months of life. Human papillomavirus (HPV) and meningococcal vaccinations are recommended for adolescents. Adult immunizations include diphtheria, pertussis, and tetanus (DPT) boosters as well as vaccinations to prevent influenza, pneumococcal pneumonia, and hepatitis A and B. COVID-19 vaccinations are recommended for children, adolescents, and adults. (See "Standard immunizations for children and adolescents: Overview", section on 'Routine schedule' and "Overview of preventive care in adults", section on 'Immunization' and "COVID-19: Vaccines".)
Screening — Screening is the identification of asymptomatic diseases or risk factors. Screening tests start in the prenatal period (such as testing for Down syndrome in the fetuses of older pregnant women) and continue throughout life (eg, when inquiring about hearing in older adults). Scientific principles of screening are discussed later in this topic.
Behavioral counseling (lifestyle changes) — Clinicians can give effective behavioral counseling to motivate lifestyle changes. Clinicians counsel patients to stop smoking, eat a prudent diet, drink alcohol moderately, exercise, and engage in safe sexual practices. It is important to have evidence that (i) behavior change decreases the risk for the condition of interest, and (ii) counseling leads to behavior change before spending time and effort on this approach to prevention. (See 'Levels of prevention' below.)
Chemoprevention — Chemoprevention is the use of drugs to prevent disease. It is used to prevent disease early in life (eg, folate during pregnancy to prevent neural tube defects and ocular antibiotic prophylaxis in all newborns to prevent gonococcal ophthalmia neonatorum) but is also common in adults (eg, statin treatment for hypercholesterolemia).
LEVELS OF PREVENTION — Merriam-Webster’s dictionary defines prevention as “the act of preventing or hindering” and “the act or practice of keeping something from happening” [2]. With these definitions in mind, almost all activities in medicine could be defined as prevention. After all, clinicians’ efforts are aimed at preventing the untimely occurrences of the 5 Ds: death, disease, disability, discomfort, and dissatisfaction. However, in clinical medicine, the definition of prevention has traditionally been restricted to interventions in people who are not known to have the particular condition of interest. Three levels of prevention have been defined: primary, secondary, and tertiary prevention (figure 1).
Primary prevention — Primary prevention keeps disease from occurring at all by removing its causes. The most common clinical primary care preventive activities involve immunizations to prevent communicable diseases; drugs; and behavioral counseling. Prophylactic surgery has become more common, with bariatric surgery to prevent complications of obesity and ovariectomy and mastectomy to prevent ovarian and breast cancer in women with certain genetic mutations.
Primary prevention has eliminated many infectious diseases from childhood. In American men, primary prevention has prevented many deaths from two major killers: lung cancer and cardiovascular disease. Lung cancer mortality in men decreased by one-third over three decades in the United States [3]. This decrease followed smoking cessation trends among adults, without organized screening and without much improvement in survival after treatment for lung cancer. Heart disease mortality rates in men have decreased by half over the past several decades [4] not only because medical care has improved, but also because of primary prevention efforts such as smoking cessation and use of antihypertensive and statin medications. Primary prevention is now possible for cervical, hepatocellular, skin, and breast cancer; bone fractures; and alcoholism.
A special attribute of primary prevention involving efforts to help patients adopt healthy lifestyles is that a single intervention may prevent multiple diseases. Smoking cessation decreases not only lung cancer but also many other pulmonary diseases, other cancers, and, most of all, cardiovascular disease. Maintaining an appropriate weight prevents diabetes and osteoarthritis as well as cardiovascular disease and some cancers.
Primary prevention at the community level can also be effective. Examples include immunization requirements for students, no-smoking regulations in public buildings, chlorination and fluoridation of the water supply, and laws mandating seatbelt use in automobiles and helmet use on motorcycles and bicycles. Certain primary prevention activities occur in specific occupational settings (use of earplugs or dust masks), in schools (immunizations), or in specialized health care settings (use of tests to detect hepatitis B and C or HIV in blood banks).
For some problems, such as injuries from automobile accidents, community prevention works best. For others, such as prophylaxis in newborns to prevent gonococcal ophthalmia neonatorum, clinical settings work best. For still others, clinical efforts can complement community-wide activities. In smoking prevention efforts, clinicians help individual patients stop smoking, and public education, regulations, and taxes prevent teenagers from starting to smoke.
Secondary prevention — Secondary prevention detects early disease when it is asymptomatic and when treatment can stop it from progressing. Secondary prevention is a two-step process, involving a screening test and follow-up diagnosis and treatment for those with the condition of interest. Testing asymptomatic patients for HIV and routine Pap smears are examples. Most secondary prevention is done in clinical settings.
As indicated earlier, screening is the identification of an unrecognized disease or risk factor by history taking (eg, asking if the patient smokes), physical examination (eg, a blood pressure measurement), laboratory test (eg, checking for proteinuria in a diabetic), or other procedure (eg, a bone mineral density examination) that can be applied reasonably rapidly to asymptomatic people. Screening tests sort out apparently well persons (for the condition of interest) who have an increased likelihood of disease or a risk factor for a disease from people who have a low likelihood. Screening tests are part of all secondary and some primary and tertiary preventive activities.
A screening test is usually not intended to be diagnostic. If the clinician and/or patient are not committed to further investigation of abnormal results and treatment, if necessary, the screening test should not be performed at all.
Tertiary prevention — Tertiary prevention describes clinical activities that prevent deterioration or reduce complications after a disease has declared itself. An example is the use of beta-blocking drugs to decrease the risk of death in patients who have recovered from myocardial infarction. Tertiary prevention is really just another term for treatment, but treatment focused on health effects occurring not so much in hours and days but months and years. For example, in diabetic patients, good treatment requires not just control of blood glucose. Searches for and successful treatment of other cardiovascular risk factors (eg, hypertension, hypercholesterolemia, obesity, and smoking) help prevent cardiovascular disease in diabetic patients as much as, and even more than, good control of blood glucose. In addition, diabetic patients need regular ophthalmologic examinations for detecting early diabetic retinopathy, routine foot care, and monitoring for urinary protein to guide use of angiotensin-converting enzyme inhibitors to prevent renal failure. All these preventive activities are tertiary in the sense that they prevent and reduce complications of a disease that is already present.
Confusion about primary, secondary, and tertiary prevention — Over the years, as more and more of clinical practice has involved prevention, the distinctions among primary, secondary, and tertiary prevention have become blurred. Historically, primary prevention was thought of as primarily vaccinations for infectious disease and counseling for healthy lifestyle behaviors, but primary prevention now includes prescribing antihypertensive medication and statins to prevent cardiovascular diseases and performing prophylactic surgery to prevent ovarian cancer in women with certain genetic abnormalities. Increasingly, risk factors are treated as if they are diseases, even at a time when they have not caused any of the 5 Ds. This is true for a growing number of health risks, for example, low bone mineral density, hypertension, hyperlipidemia, obesity, and certain genetic abnormalities. Treating risk factors as disease broadens the definition of secondary prevention into the domain of traditional primary prevention.
In some disciplines, such as cardiology, the term secondary prevention is used when discussing tertiary prevention. “A new era of secondary prevention” was declared when treating patients with acute coronary syndrome (myocardial infarction or unstable angina) with a combination of antiplatelet and anticoagulant therapies to prevent cardiovascular death [5]. Similarly, “secondary prevention” of stroke is used to describe interventions to prevent stroke in patients with transient ischemia attacks.
Tests used for primary, secondary, and tertiary prevention, as well as for diagnosis, are often identical, another reason for confusing the levels of prevention (and confusing prevention with diagnosis). Colonoscopy may be used to find a cancer in a patient with blood in their stool (diagnosis); to find an early asymptomatic colon cancer (secondary prevention); remove an adenomatous polyp, which is a risk factor for colon cancer (primary prevention); or to check for cancer recurrence in a patient treated for colon cancer (a tertiary preventive activity referred to as surveillance).
Regardless of the terms used, an underlying reason to differentiate levels among preventive activities is that there is a spectrum of probabilities of disease and adverse health effects from the condition(s) being sought and treated during preventive activities, as well as different probabilities of adverse health effects from interventions that are used for prevention at the various levels. The underlying risk of certain health problems is usually much higher in diseased than healthy people. For example, the risk of cardiovascular disease in diabetic patients is much greater than in asymptomatic non-diabetic patients. Identical tests perform differently depending on the level of prevention. Furthermore, the tradeoffs between effectiveness and harms can be quite different for patients in different parts of the spectrum. False-positive test results and overdiagnosis (both discussed later in this chapter) among people without the disease being sought are important issues in secondary prevention, but they are less important in treatment of patients already known to have the disease in question. The terms primary, secondary, and tertiary prevention are ways to consider these differences conceptually.
SCIENTIFIC APPROACH TO CLINICAL PREVENTION — When considering what preventive activities to perform, the clinician must first decide with the patient which medical problems or diseases they should try to prevent. This statement is so clear and obvious that it would seem unnecessary to mention, but the fact is that many preventive procedures, especially screening tests, are performed without a clear understanding of what is being sought or prevented. For instance, clinicians performing routine checkups on their patients may order a urinalysis. However, a urinalysis might be used to search for any number of medical problems, including diabetes, asymptomatic urinary tract infections, renal cancer, or renal failure. It is necessary to decide which, if any, of these conditions is worth screening for before undertaking the test. One of the most important scientific advances in clinical prevention has been the development of methods for deciding whether a proposed preventive activity should be undertaken [6]. The remainder of this topic describes these methods and concepts.
Three criteria are important when judging whether a condition should be included in preventive care (table 1):
●The burden of suffering caused by the condition
●The effectiveness, safety, and cost of the preventive intervention or treatment
●The performance of the screening test
BURDEN OF SUFFERING — Only conditions posing threats to life or health (the 5 Ds) should be included in preventive care. The burden of suffering of a medical condition is determined primarily by (i) how much suffering (in terms of the 5 Ds) it causes those afflicted with the condition, and (ii) its frequency.
How does one measure suffering? Most often, it is measured by mortality rates and frequency of hospitalizations and amount of health care utilization caused by the condition. Information about how much disability, pain, nausea, or dissatisfaction a given disease causes is much less available.
The frequency of a condition is also important in deciding about prevention. A disease may cause great suffering for individuals who are unfortunate enough to get it, but it may occur too rarely, especially in the individual’s particular age group, for screening to be considered. Breast cancer is an example. Although it can occur in much younger women, most breast cancers occur in women older than 50 years of age. For 20-year-old women, annual breast cancer incidence is 1.6 in 100,000 (about one-fifth the rate for men in their later 70s) [7]. Although breast cancer should be sought in preventive care for older women, it is too uncommon in average 20-year-old women and 70-year-old men for screening. Screening for very rare diseases means not only that, at most, very few people will benefit, but screening also results in false-positive tests in some people who are subject to complications from further diagnostic evaluation.
The incidence of what is to be prevented is especially important in primary and secondary prevention because, regardless of the disease, the risk is low for most individuals. Stratifying populations according to risk and targeting the higher-risk groups can help overcome this problem. This practice is frequently done by concentrating specific preventive activities on certain age groups or medical conditions, or by using predictive modeling to incorporate many risk factors.
EFFECTIVENESS OF TREATMENT — Randomized controlled trials are the strongest scientific evidence for establishing the effectiveness of treatments. It is usual practice to meet this standard for tertiary prevention (treatments). On the other hand, to conduct randomized trials when evaluating primary or secondary prevention requires very large studies on thousands, often tens of thousands, of patients, carried out over many years, sometimes decades, because the outcome of interest is rare and often takes years to occur. The task is daunting, and all the difficulties of randomized controlled trials are magnified manyfold.
Other challenges in evaluating treatments in prevention are outlined in the following text for each level of prevention.
Treatment in primary prevention — Whatever the primary intervention (immunizations, drugs, behavioral counseling, or prophylactic surgery), it should be efficacious (able to produce a beneficial result in ideal situations) and effective (able to produce a beneficial net result under usual conditions, taking into account patient compliance). Because interventions for primary prevention are usually given to large numbers of healthy people, they also must be very safe.
Randomized trials — Virtually all recommended immunizations are backed by evidence from randomized trials, sometimes accomplished relatively quickly when the outcomes occur within weeks or months, as in childhood infections. Because pharmaceuticals are regulated, primary and secondary preventive activities involving drugs (eg, treatment of hypertension and hyperlipidemia in adults) also usually have been evaluated by randomized trials. Randomized trials are less common when the proposed prevention is not regulated, as is true with vitamins, minerals, and food supplements, or when the intervention is behavioral counseling.
Observational studies — Observational studies can help clarify the effectiveness of primary prevention when randomization is not possible.
●Example: What is the risk of transmission of HIV infection in gay men with undetectable viral levels who engage in condomless sex? Viral suppression would theoretically reduce, and potentially eliminate, transmission. However, even with undetectable levels, HIV-positive individuals continue to harbor the virus. It would be ethically questionable and logistically difficult to do a randomized trial to determine the risks of condomless sex in gay men. Investigators conducted a cohort study of 782 gay male couples, one partner with virally suppressed HIV and the other HIV-negative, who were followed for new HIV infection [8]. At a median of two years of follow-up and 76,000 episodes of condomless sex, no new cases of intra-partner HIV occurred.
However, observational studies are vulnerable to bias. The conclusion that undetectable HIV virus reduces and potentially eliminates transmission is reasonable from a biologic perspective and from the dramatic result. It is on even firmer ground since the finding is replicated in different populations (such as among heterosexuals in different ethnic groups) and in different clinical contexts (such as mother-to-child transmission in pregnancy).
Safety — With immunizations, the occurrence of adverse effects may be so rare that randomized trials would be unlikely to uncover them. One way to study this question is to track illnesses in large datasets of millions of patients and to compare the frequency of an adverse effect linked temporally to the vaccination.
●Example: The safety of COVID-19 vaccines has been an important concern, particularly given the widespread use of the vaccines and the hesitancy of some individuals to consent to vaccination. Initial randomized controlled trials demonstrating the efficacy and safety of the vaccines were not large enough to detect rare adverse events. Cases of thrombosis and thrombocytopenia syndrome (TTS) were reported after the introduction of adenoviral-vectored COVID-19 vaccines [9,10]. Subsequent studies used surveillance data to estimated the incidence of TTS after COVID-19 vaccination. For example, a study calculated the incidence of TTS in the United States from cases submitted to the Vaccine Adverse Event Reporting System, as well administrative data on vaccine administration [11]. There were 57 confirmed cases: 54 after an adenovirus-based vaccine (Janssen/Johnson & Johnson) and three after mRNA-based vaccines. The incidence rates (cases per million vaccine doses) were 3.83 and 0.000855, respectively. This and similar studies provide more evidence that the initial case reports of TTS did pick up a real risk associated with adenoviral-vectored COVID-19 vaccination; the very low incidence of TTS after the administration of mRNA-based vaccines may be no different than the baseline rate in the population.
Population-based surveillance systems can provide better estimates of risk than case reports alone. Even so, associations found in surveillance systems are relatively weak evidence for a causal relationship because they are observational in nature, may have incomplete reporting of events, and often do not have information on all important possible confounding variables.
Counseling — United States laws do not require rigorous evidence of effectiveness of behavioral counseling methods. Nevertheless, clinicians should require scientific evidence before incorporating routine counseling into preventive care; counseling that does not work wastes time, costs money, and may harm patients. Research has demonstrated that certain counseling methods can help patients change some health behaviors. Smoking cessation efforts have led the way with many randomized trials evaluating different approaches.
●Example: Smoking kills approximately 450,000 Americans each year. But what is the evidence that advising patients to quit smoking gets them to stop? Are some approaches better than others? These questions were addressed in a review of all studies done on smoking cessation, focusing on randomized trials [12,13]. There were 49 trials identified that assessed counseling with or without medication. Counseling increased quit rates by more than half, and adding medication to counseling further improved smoking cessation (figure 2). Pharmacotherapy with bupropion (a centrally acting drug that decreases craving), varenicline (a nicotine receptor agonist), nicotine gum, nasal spray, or patches were effective. More intensive counseling increased abstinence rates more than less intensive counseling.
Treatment in secondary prevention — Treatments in secondary prevention are generally the same as treatments for curative medicine. Like interventions for symptomatic disease, they should be both efficacious and effective. Unlike usual interventions for disease, however, it typically takes years to establish that a secondary preventive intervention is effective, and it requires large numbers of people to be studied. For example, one of the major randomized trials of prostate cancer screening recruited 162,000 men followed for 13 years to detect a decreased in death from prostate cancer of 1 per 10,000 person-years of follow-up [14].
A unique requirement for treatment in secondary prevention is that treatment of early, asymptomatic disease must be superior to treatment of the disease when it would have been diagnosed in the usual course of events, when a patient seeks medical care for symptoms. If outcome in the two situations is the same, screening does not add value.
●Example: Lung cancer is the leading cause of cancer-related death in the United States. Why, then, did no major professional medical group recommend screening for lung cancer throughout the first decade of the 21st century? Several randomized trials begun in the 1970s and 1980s found screening was not protective. In one study of the use of chest radiographs and sputum cytology to screen for lung cancer, male cigarette smokers who were screened every four months and treated promptly when cancer was found did no better than those not offered screening and treated only when they presented with symptoms. Twenty years later, death rates from lung cancer were similar in the two groups: 4.4 per 1000 person-years in the screened men versus 3.9 per 1000 person-years in men not offered screening [15]. However, in 2011, a randomized controlled trial of low-dose computed tomography (CT) screening reported a 20 percent reduction in lung cancer mortality after a median of 6.5 years [16]. This trial convincingly demonstrated for the first time that treatment of asymptomatic lung cancer found on screening decreased mortality.
Treatment in tertiary prevention — All new pharmaceutical treatments in the United States are regulated by the US Food and Drug Administration (FDA), which almost always requires evidence of efficacy from randomized clinical trials. It is easy to assume, therefore, that tertiary preventive treatments have been carefully evaluated. However, after a drug has been approved, it may be used for new, unevaluated indications. Patients with some diseases are at increased risk for other diseases; thus, some drugs are used not only to treat the condition for which they are approved but also to prevent other diseases for which patients are at increased risk. The distinction between proven therapeutic effects of a medicine for a given disease and its effect in preventing other diseases is a subtle challenge facing clinicians when considering tertiary preventive interventions that have not been evaluated for that purpose. Sometimes, careful evaluations have led to surprising results.
●Example: Three randomized trials have shown that tight control of blood sugar (bringing the level down to normal range) in patients with type 2 diabetes did not prevent cardiovascular disease and mortality any better than looser control (see "Glycemic control and vascular complications in type 2 diabetes mellitus", section on 'Macrovascular disease'). The diabetic medications used in these studies were approved after randomized controlled studies. However, the outcomes leading to approval of the drugs was not prevention of long-term cardiovascular disease (tertiary prevention) among diabetic patients, but rather the effect of the medications on blood sugar levels. The assumption was made that these medications would also decrease cardiovascular disease because observational studies had shown blood sugar levels correlated with cardiovascular disease risk. Putting this assumption to the test in rigorous randomized trials produced surprising and important results. As a result, tertiary prevention in diabetes has shifted toward including aggressive treatment of other risk factors for cardiovascular disease. With such an approach, a randomized trial showed that cardiovascular disease and death in diabetics were reduced about 50 percent over a 13-year period compared with a group receiving conventional therapy aimed at controlling blood sugar levels; at 21 years of follow-up, the intervention group had lived a median of 7.9 years longer [17,18].
METHODOLOGIC ISSUES IN EVALUATING SCREENING PROGRAMS — Several problems arise in the evaluation of screening programs, some of which can make it appear that early treatment is effective after screening when it is not. These issues include the difference between prevalence and incidence screens and three biases that can occur in screening studies: lead-time, length-time, and compliance biases.
Prevalence and incidence screens — The yield of screening decreases as screening is repeated over time, as demonstrated in a figure (figure 3). The first time that screening is carried out (the prevalence screen), cases of the medical condition will have been present for varying lengths of time. During the second round of screening, most cases found will have had their onset between the first and second screening. (A few will have been missed by the first screen.) Therefore, the second (and subsequent) screenings are called incidence screens. Thus, when a group of people is periodically rescreened, the number of cases of disease in the group drops after the prevalence screen. This means that the positive predictive value for test results will decrease after the first round of screening.
Special biases — The following biases are most likely to be a problem in observational studies of screening.
Lead-time bias — Lead time is the period of time between the detection of a medical condition by screening and when it ordinarily would have been diagnosed because a patient experienced symptoms and sought medical care (figure 4). The amount of lead time for a given disease depends on the biologic rate of progression of the disease and how early the screening test can detect the disease. When lead time is very short, as is true with lung cancer, it is difficult to demonstrate that treatment of medical conditions picked up on screening is more effective than treatment after symptoms appear. On the other hand, when lead time is long, as is true for cervical cancer (on average, it takes 20 to 30 years for carcinoma in situ to progress to clinically invasive disease), treatment of the medical condition found on screening can be very effective.
How can lead time cause biased results in a study of the efficacy of early treatment? As shown in a figure (figure 4), because of screening, a disease is found earlier than it would have been after the patient developed symptoms. As a result, people who are diagnosed by screening for a deadly disease will, on average, survive longer from the time of diagnosis than people who are diagnosed after they develop symptoms, even if early treatment is no more effective than treatment at the time of clinical presentation. In such a situation, screening would appear to help people live longer, spuriously improving survival rates when, in reality, they would have been given not more “survival time” but more “disease time.”
An appropriate method of analysis to avoid lead-time bias is to compare age-specific mortality rates rather than survival rates in a screened group of people and a control group of similar people who do not get screened, as in a randomized trial (table 2). Screening for breast, lung, and colorectal cancers are known to be effective because randomized controlled trials have shown that mortality rates of screened persons are lower than those of a comparable group of unscreened people.
Length-time bias — Length-time bias occurs because the proportion of slow-growing lesions diagnosed during screening is greater than the proportion of those diagnosed during usual medical care. As a result, length-time bias makes it seem that screening and early treatment are more effective than usual care.
Length-time bias occurs in the following way. Screening works best when a medical condition develops slowly. A given type of cancer, however, typically demonstrates tumors with a wide range of growth rates. Some of them grow slowly, some very fast. Screening tests are likely to find mostly slow-growing tumors because they are present for a longer period of time before they cause symptoms. Fast-growing tumors are more likely to cause symptoms that lead to diagnosis in the interval between screening examinations (figure 5 and figure 6). Screening, therefore, tends to find tumors with inherently better prognoses. As a result, the mortality rates of cancers found through screening may be better than those not found through screening, but screening is not protective in this situation.
Compliance bias — The third major bias that can occur in prevention studies is compliance bias. Compliant patients tend to have better prognoses regardless of preventive activities. The reasons for this are not completely clear, but on average, compliant patients are more interested in their health and are generally healthier than noncompliant ones. For example, a randomized trial that invited people for screening found that volunteers from the control group who were not invited but requested screening had better mortality rates than the invited group, which contained both compliant people who wanted screening and those who refused [19]. The effect of patient compliance, as distinct from treatment effect, has primarily involved medication adherence in the placebo group and has been termed placebo adherence.
●Example: An analysis was done to determine if health outcomes differed in the placebo arm of a randomized trial among patients who demonstrated different degrees of placebo adherence [20]. The original study was a tertiary prevention trial that randomized asymptomatic patients with left ventricular dysfunction (cardiac ejection fraction <35 percent) to active treatment (enalapril) or placebo. The analysis showed that after three years, patients randomized to placebo who took at least 75 percent of the placebo medication (“highly adherent” patients) had half the mortality rate of “lower adherent” patients randomized to placebo. Explanations for this strange result were sought. The difference in mortality was not changed after adjusting for several risk factors, including serious illness. The placebo medication itself was chemically inert. Placebo adherence appears to be a marker of healthier people even though the underlying mechanism remains to be determined.
Biases from length time and patient compliance can be avoided by relying on studies that have concurrent screened and control groups that are comparable. In each group, all people experiencing the outcomes of interest must be counted, regardless of the method of diagnosis or degree of participation (table 2). Randomized trials are the strongest design because patients who are randomly allocated will have comparable numbers of slow- and fast-growing tumors and, on average, comparable levels of compliance. These groups then can be followed over time with mortality rates, rather than survival rates, to avoid lead-time bias. If a randomized trial is not possible, results of population-based observational studies can be valid. In such cases, bias can be minimized if the screened and control groups are made up of similar populations, the control population does not have access to screening, and both populations have careful follow-up to document all cases of the outcome being studied.
Because randomized controlled trials and prospective population-based studies are difficult to conduct, take a long time, and are expensive, investigators sometimes try to use other kinds of studies, such as historical cohort studies or case-control studies, to investigate preventive maneuvers.
●Example: To test whether periodic screening with colonoscopy reduces mortality from colorectal cancer, researchers conducted a case-control study in a large health care system [21]. They compared the frequency of colonoscopy screening over the previous 10 years in patients with and without colorectal cancer, matched for age, sex, health plan enrollment duration, and geographical region. To deal with lead- and length-time biases, they investigated screening only in people who were known to have died (case group) or not to have died (control group) from colorectal cancer. To deal with compliance bias, they adjusted their results for the number of general periodic health examinations and the use of other colorectal cancer screen tests. They excluded patients with the presence of medical conditions that could have led to both increased screening and increased likelihood of colorectal cancer. The investigators found that colonoscopy followed by early therapy prevented two-thirds of deaths related to colorectal cancer. The results did not change when the analysis was stratified by whether patients had received other colorectal cancer screening. The study thus employed methods to reduce bias for observational studies in general (restriction, matching, stratification, and adjustment) as well as to address those specific to screening (lead-time, length-time, and compliance biases).
PERFORMANCE OF SCREENING TESTS — The following criteria for a good screening test apply to all types of screening tests, whether they are history, physical examination, or laboratory tests.
High sensitivity and specificity — The very nature of searching for a disease in people without symptoms means that prevalence is usually very low, even among high-risk groups who were selected because of age, sex, and other risk characteristics. A good screening test must, therefore, have a high sensitivity so that it does not miss the few cases of disease present. It must also be sensitive early in the disease, when the subsequent course can still be altered. If a screening test is sensitive only for late-stage disease, which has progressed too far for effective treatment, the test would be useless. A screening test should also have a high specificity to reduce the number of people with false-positive results who require diagnostic evaluation.
Sensitivity and specificity are determined for screening tests much as they are for diagnostic tests, with one major difference. The sensitivity and specificity of a diagnostic test are determined by comparing the results with another test (the gold standard). In screening, the gold standard for the presence of disease often is not only another, more accurate test but also a period of time for follow-up. The gold standard test is routinely applied only to people with positive screening test results to differentiate between true and false-positive results. A period of follow-up should also be applied to all people who have a negative screening test result in order to differentiate between true- and false-negative test results.
Follow-up is particularly important in cancer screening, where interval cancers, cancers not detected during screening but subsequently discovered over the follow-up period, occur. When interval cancers occur, the calculated test sensitivity is lowered.
●Example: In one study, prostate-specific antigen (PSA) levels were measured in stored blood samples collected from a cohort of healthy men (physicians) who later were or were not diagnosed with prostate cancer [22]. Thirteen of 18 men who were diagnosed with prostate cancer within a year after the blood sample had elevated PSA levels (>4.0 ng/mL) and would have been diagnosed after an abnormal PSA result; the other five had normal PSA results and developed interval cancers during the first year after a normal PSA test. Thus, sensitivity of PSA was calculated as 13 divided by (13 + 5), or 72 percent.
A key challenge is to choose a correct period of follow-up. If the follow-up period is too short, disease missed by the screening test might not have a chance to make itself obvious, so the test’s sensitivity may be overestimated. On the other hand, if the follow-up period is too long, disease not present at the time of screening might be found, resulting in a falsely low estimation of the test’s sensitivity.
Detection and incidence methods for calculating sensitivity — Calculating sensitivity by counting cancers detected during screening as true positives and interval cancers as false negatives is sometimes referred to as the detection method (figure 7). The method works well for many screening tests, but there are two difficulties with it for some cancer screening tests. First, as already pointed out, it requires that the appropriate amount of follow-up time for interval cancers be known; often, it is not known and must be guessed. The detection method also assumes that the abnormalities detected by the screening test would go on to cause trouble if left alone. This is not necessarily so for several cancers, such as prostate cancer.
●Example: Histologic prostate cancer is common in men, especially older men. A review of autopsy studies showed that the prevalence of prostate cancer among men were 5 percent among in their 20s, 30 percent in their 50s, and 83 percent among men in their 70s [23]. Screening tests can find such cancers in many men, but for most the cancer will never become invasive. Thus, when the sensitivity of prostate cancer tests such as PSA is determined by the detection method, the test may look quite good because the numerator includes all cancers found, not just those with malignant potential.
The incidence method calculates sensitivity by using the incidence in persons not undergoing screening and the interval cancer rate in persons who are screened (figure 7). The rationale for this approach is that the sensitivity of a test should affect interval cancer rates, not disease incidence. For prostate cancer, the incidence method defines sensitivity of the test as one minus the ratio of the interval prostate cancer rate in a group of men undergoing periodic screening to the incidence of prostate cancer in a group of men not undergoing screening (control group). The incidence method of calculating sensitivity gets around the problem of counting “benign” prostate cancers, but it may underestimate sensitivity because it excludes cancers with long lead times. True sensitivity of a test is, therefore, probably between the estimates of the two methods.
Low positive predictive value — Because of the low prevalence of most diseases in asymptomatic people, the positive predictive value of most screening tests is low, even for tests with high specificity. (The reverse is true for negative predictive value, because when prevalence is low, the negative predictive value is likely to be high.) Clinicians who perform screening tests on their patients must accept the fact that they will have to work up many patients who have positive screening test results but do not have disease. However, they can minimize the problem by concentrating their screening efforts on people with a higher prevalence for disease.
●Example: The incidence of breast cancer increases with age, from approximately 1 in 100,000/year at age 20 to 1 in 200/year over age 70. Also, sensitivity and specificity of mammography are better in older women. Therefore, a lump found during screening in a young woman’s breast is more likely to be non-malignant than a lump in an older woman. In a large study of breast cancer screening, finding cancer after an abnormal mammogram varied markedly according to the age of women; in women in their 40s, about 32 women without cancer experienced further workup for every woman who was found to have a malignancy (figure 8) with a positive predictive value of 3.0 percent [24]. However, for women in their 80s, the number dropped to about 7, with a positive predictive value of 12.5 percent.
Simplicity and low cost — An ideal screening test should take only a few minutes to perform, require minimum preparation by the patient, depend on no special appointments, and be inexpensive.
Simple, quick examinations such as blood pressure determinations are ideal screening tests. Conversely, tests such as colonoscopy, which are expensive and require an appointment and bowel preparation, are best suited for diagnostic testing in patients with symptoms and clinical indications. Nevertheless, screening colonoscopy has been found to be highly effective in decreasing colorectal cancer mortality, and a negative test does not have to be repeated for 10 years. Other tests, such as visual field testing for the detection of glaucoma and audiograms for the detection of hearing loss, fall between these two extremes.
The financial “cost” of the test depends not only on the cost of (or charge for) the procedure itself but also on the cost of subsequent evaluations performed on patients with positive test results. Thus sensitivity, specificity, and predictive value affect cost. Cost is also affected by whether the test requires a special visit to the clinician. Screening tests performed while the patient is seeing their clinician for other reasons (as is frequently the case with blood pressure measurements) are much cheaper for patients than tests requiring special visits, extra time off work, and additional transportation. Cost also is determined by how often a screening test must be repeated.
Taking all these issues into account sometimes leads to surprising conclusions.
●Example: Several different tests can be used to screen for colorectal cancer. They include annual stool-based tests, computed tomography (CT) colonography every five years, sigmoidoscopy every five years, and colonoscopy every 10 years. The tests vary greatly in their upfront costs, ranging from USD $20 for guaiac fecal occult blood tests to more than $1000 for screening colonoscopy. However, the cost per year of life saved by screening is not very different for these tests; all were in an acceptable range by United States standards [25]. This is true for several reasons. First, the simpler, cheaper tests have to be done more often. Fecal occult blood tests are recommended yearly, whereas colonoscopy is recommended every 10 years. The cheaper tests also produce more false-positive results that lead to more testing (and, therefore, more costs). Finally, they produce more false-negative results and miss patients who actually have cancer. This leads to increased costs for care of patients with more advanced cancer.
Safety — It is reasonable and ethical to accept a certain risk for diagnostic tests applied to sick patients seeking help for specific complaints. The clinician cannot avoid action when the patient is severely ill, and does their best. It is quite another matter to subject presumably well people to risks. In such circumstances, the procedure should be especially safe. This is partly because the chances of finding disease in healthy people are so low. Thus, although colonoscopy is hardly thought of as a dangerous procedure when used on patients with gastrointestinal complaints, it can cause bowel perforation. In fact, if the perforation rate of colonoscopy is 1 per 1000, when colonoscopy is used to screen people in their 50s, colon perforations occur more often than the number of cancers found.
Concerns have been raised about possible long-term risks with the increasing use of CT scans to screen for coronary artery disease or, in the case of whole-body scans, a variety of abnormalities. The radiation dose of CT scans varies by type, with a CT scan for coronary calcium on average being the equivalent of about 30, and a whole-body scan about 120, chest radiographs. One estimate of risk projected 29,000 excess cancers as a result of 70 million CT scans performed in the United States in a single year [26]. If these concerns are correct, CT scans used to screen for early cancer could themselves cause cancer over subsequent decades.
Acceptable to patients and clinicians — If a screening test is associated with discomfort, it usually takes several years to convince large percentages of patients to obtain the test. This has been true for Pap smears, mammograms, sigmoidoscopies, and colonoscopies. By and large, however, the American public supports screening.
The acceptability of the test to clinicians can determine which tests patients receive. For example, as already noted, several alternatives are available for colorectal cancer screening that differ in convenience. In trials of alternative screening tests, patients were more likely to complete the stool tests. Nevertheless, colonoscopy is the predominant screening test in the United States, despite requiring more time and expertise to complete. Physicians in the United States generally consider colonoscopy superior to the alternatives and this may explain much of preferential use of this test [27].
UNINTENDED CONSEQUENCES OF SCREENING — Adverse effects of screening tests include discomfort during the test procedure (the majority of women undergoing mammography say that the procedure is painful, although usually not so severe that patients refuse the test), long-term radiation effects after exposure to radiographic procedures, false-positive test results (with resulting needless workups and negative labeling effects), overdiagnosis, and incidentalomas. The last three will be discussed in this section.
Risk of false-positive result — A false-positive screening test result is an abnormal result in a person without disease. As already mentioned, tests with low predictive values (resulting from low prevalence of disease, poor specificity of the test, or both) are likely to lead to a higher frequency of false positives. False-positive results, in turn, can lead to negative labeling effects, inconvenience, and expense in obtaining follow-up procedures. In certain situations, false-positive results can lead to major surgery. In a study of ovarian cancer screening, 8.4 percent (3285) of 39,000 women had a false-positive result and one-third of those underwent surgery as part of the diagnostic evaluation of the test result. Because of false-positive screening tests, five times more women without ovarian cancer had surgery than those with ovarian cancer [28].
False-positive results account for only a minority of screening test results (only about 10 percent of screening mammograms are false positives). Even so, they can affect large percentages of people who get screened. This happens in two ways. Most clinicians do not perform only one or two tests on patients presenting for routine checkups. Modern technology, and perhaps the threat of lawsuits, has fueled the propensity to “cover all the bases.” Automated tests allow clinicians to order several dozen tests with a few checks in the appropriate boxes.
When the measurements of screening tests are expressed on interval scales (as most blood tests are), and when normal is defined by the range covered by 95 percent of the results (as is usual), the more tests the clinician orders, the greater the risk of a false-positive result. In fact, as shown in a table (table 3), if the clinician orders enough tests, “abnormalities” will be discovered in virtually all healthy patients. A spoof entitled “The Last Well Person” commented on this phenomenon [29].
Another reason that many people may experience a false-positive screening test result is that most screening tests are repeated at regular intervals. With each repeat screen, the patient is at risk for a false-positive result.
●Example: In a clinical trial of lung cancer screening with low-dose spiral computed tomography (CT) or chest radiograph, the first round of screening produced false-positive results in 21 percent of people screened with CT and 9 percent of people screened with chest radiograph; after the second round, the total proportion of people experiencing a false positive had increased to 33 percent in the CT group and 15 percent in the chest radiograph group [30]. In a related study, participants received screening tests for prostate, ovarian, and colorectal cancer as well as chest radiographs for lung cancer, for a total of 14 tests over three years. The cumulative risk of having at least one false-positive screening test result was 60.4 percent for men and 48.8 percent for women. The risk for undergoing an invasive diagnostic procedure prompted by a false-positive test was 28.5 percent for men and 22.1 percent for women [31].
Risk of negative labeling effect — Test results can sometimes have important psychological effects on patients, called a labeling effect. A good screening test result produces either no labeling effect or a positive labeling effect.
A positive labeling effect may occur when a patient is told that all the screening test results were normal. Most clinicians have heard such responses as, “Great, that means I can keep working for another year.” On the other hand, being told that the screening test result is abnormal and more testing is necessary may have an adverse psychological effect, particularly in cancer screening. Some people with false-positive tests continue to worry even after being told everything was normal on follow-up tests. Because this is a group of people without disease, negative labeling effects are particularly worrisome ethically. In such situations, screening efforts might promote a sense of vulnerability instead of health and might do more harm than good.
●Example: A study of men with abnormal prostate-specific antigen (PSA) screening test results who subsequently were declared to be free of cancer after workup found that, one year after screening, 26 percent reported worry about prostate cancer, compared with 6 percent among men with normal PSA results, and 46 percent reported their wives or significant others were concerned about their partner’s potential for prostate cancer versus 14 percent of those with normal results [32]. Anxiety has also been reported after false-positive results tests for other cancers as well as in blood pressure screening. In summary, people do not respond well when told, “Your screening test result was not quite normal, and we need to do more tests.”
Labeling effects are sometimes unpredictable, especially among people who know they are at high risk for a genetic disease because of a family history. Studies of relatives of patients with Huntington disease (a neurologic condition with onset in middle age causing mental deterioration leading to dementia, movement disorders, and death) have found that those with positive genetic tests had no worsening in psychological health, perhaps because they were no longer dealing with uncertainty. Studies of women being tested for genetic mutations that increase their risk for breast and ovarian cancer have also found that women testing positive for the mutation experience little or no psychological deterioration.
Risk of overdiagnosis (pseudodisease) in cancer screening — The rationale for cancer screening is that the earlier a cancer is found, the better the chance of cure. Therefore, the thinking goes, it is always better to find cancer as early as possible. This thesis has been challenged by the observation that incidence often increases after the introduction of widespread screening for a particular cancer. A temporary increase in incidence is to be expected because screening moves the time of diagnosis forward, adding early cases to the usual number of prevalent cancers being diagnosed without screening, but the temporary bump in incidence should come down to the baseline level after a few years. With several cancers, however, incidence has remained at a higher level, as illustrated for prostate cancer in a figure (figure 9). It is as if screening caused more cancers. How could this be?
Some cancers are so slow growing (some even regress) that they do not cause any trouble for the patient. If such cancers are found through screening, they are called pseudodisease; the process leading to their detection is called overdiagnosis because finding them does not help the patient. Overdiagnosis is an extreme example of length-time bias. The cancers found have such a good prognosis that they would never become evident without screening technology. Some estimates are that as many as 50 percent of prostate cancers diagnosed by screening are due to overdiagnosis.
As research unravels the development of cancer, it appears that a sequence of genetic and other changes accompany pathogenesis from normal tissue to malignant disease. At each step, only some lesions go on to the next stage of carcinogenesis. It is likely that overdiagnosis occurs because cancers early in the chain are being picked up by screening tests. The challenge is to differentiate those early cancers that will go on to cause morbidity and mortality from those that will lie dormant throughout life, even though pathologically they appear the same. Generally, screening technology is not able to do this.
To determine if and to what degree overdiagnosis occurs, it is necessary to compare a screened group with a similar unscreened group and determine incidence and disease-specific mortality rates (not survival rates). This can be done by long-term randomized trials or by careful population-based observational studies.
●Example: Neuroblastoma, a tumor of neurologic tissue near the kidney, is the second most common tumor occurring in children. Prognosis depends on the stage of the disease and is better when the tumor is diagnosed during infancy. Treatment involves surgery and chemotherapy. A simple urine test for catecholamine metabolites can be used to screen for the tumor. Japanese studies showed improved survival rates after screening, but lead time and a historical control group could have biased the results. Also, there was no population-based registry of childhood cancers in Japan to ensure ascertainment of all neuroblastoma cases. Finally, at least some cases of neuroblastomas regress without treatment, which raised the possibility of overdiagnosis.
Two population-based studies were, therefore, undertaken in Germany [33] and Quebec [34], in which screening was offered for all infants in certain areas, while infants in other areas were not screened and acted as concurrent controls. Existing tumor registries were used to track all cases of and deaths due to neuroblastoma. In both studies, the incidence of neuroblastoma doubled in the screened group, but mortality rates from neuroblastoma over the subsequent five years were equivalent in the screened groups and the unscreened groups. It appeared that the screening test primarily detected tumors with a favorable prognosis, many of which would have regressed if left undetected. Meanwhile, highly invasive disease was often missed. Investigators of both studies concluded that screening infants for neuroblastoma leads to overdiagnosis but does not reduce mortality from neuroblastoma.
Overdiagnosis has been shown in randomized trials of screening for lung and breast cancer. It also can occur when detecting precancerous abnormalities in cervical, colorectal, and breast cancer screening, that is, with cervical dysplasia, adenomatous polyps, and ductal carcinoma in situ (abnormalities sometimes termed predisease). It is important to understand that overdiagnosis can coexist with effective screening and that although randomized trials and population studies can help determine the amount of overdiagnosis, it is impossible to identify it in an individual patient.
Incidentalomas — Over the past couple of decades, using CT as a screening test has become more common. CT has been evaluated rigorously as “virtual colonoscopy” for colorectal cancer screening and also for lung cancer screening. It has been advocated as a screening test for coronary heart disease (with calcium scores) and for screening in general with full-body CT scans. Unlike most screening tests, CT often visualizes much more than the targeted area of interest. For example, CT colonography visualizes the abdomen and lower thorax. In the process, abnormalities are sometimes detected outside the colon. Masses or lesions detected incidentally by an imaging examination are called incidentalomas.
●Example: A systematic review of 17 studies found that incidentalomas were common in CT colonography; 40 percent of 3488 patients had at least one incidental finding, and 14 percent had further evaluation [35]. Workups uncovered early-stage non-colonic cancers in about 1 percent of patients and abdominal aortic aneurysms >5.5 cm in size in about 0.1 percent. Therefore, incidentalomas were very common but included some potentially important conditions. There was no information about how many patients who did not receive further evaluation went on to suffer serious consequences. Also, most of the detected cancers were ones for which screening is not recommended. With inadequate evidence about benefits and likelihood of increased costs in pursuit of extracolonic findings, the Center for Medicare and Medicaid Services (CMS) declined to cover screening CT colonography in the Medicare population.
CHANGES IN SCREENING TESTS AND TREATMENTS OVER TIME — Careful evaluation of screening has been particularly difficult when tests approved for diagnostic purposes are then used as screening tests before evaluation in a screening study, a problem analogous to using therapeutic interventions for prevention without evaluating them in prevention. For example, computed tomography (CT) scans and magnetic resonance imaging (MRI) were developed for diagnostic purposes in patients with serious complaints or known disease, and prostate-specific antigen (PSA) was developed to determine whether treatment for prostate cancer was successful. All of these tests are now commonly used as screening tests, but most became common in practice without careful evaluation. Only low-dose CT scans for lung cancer screening underwent careful evaluation prior to widespread use. PSA screening became so common in the United States that when it was subjected to a careful randomized trial, more than half the men assigned to the control arm had a PSA test during the course of a trial. When tests are so commonly used, it is difficult to determine rigorously whether they are effective.
Over time, improvements in screening tests, treatments, and vaccinations may change the need for screening. As indicated earlier, effective secondary prevention is a two-step process: a good screening test followed by a good treatment for those found to have disease. Changes in either one may affect how well screening works in preventing disease. At one extreme, a highly accurate screening test will not help prevent adverse outcomes of a disease if there is no effective therapy. Screening tests for HIV preceded the development of effective HIV therapy; therefore, early in the history of HIV, screening could not prevent disease progression in people with HIV. With the development of increasingly effective treatments, screening for HIV increased. At the other extreme, a highly effective treatment may make screening unnecessary. With modern therapy, the 10-year survival rate of testicular cancer is over 95 percent, so high that it would be difficult to show improvement with screening for this rare cancer. As human papillomavirus (HPV) vaccination for prevention of cervical cancer becomes more widespread and able to cover more carcinogenic types of HPV, the need for cervical cancer screening should decrease over time. Some studies of mammography screening have not found the anticipated mortality benefits seen in earlier studies, partly because breast cancer mortality among women not screened was lower than in the past, probably due to improved treatments. Thus, with the introduction of new therapies and screening tests, effectiveness of screening will change and ongoing reevaluation is necessary.
WEIGHING BENEFITS AGAINST HARMS OF PREVENTION — How can the many aspects of prevention discussed above be combined to make a decision whether to include a preventive intervention in clinical practice? Conceptually, the decision should be based on weighing the magnitude of benefits against the magnitude of harms that will occur as a result of the action. This approach has become common when making treatment decisions; reports of randomized trials routinely include harms as well as benefits.
A straightforward approach is to present the benefits and harms for a particular preventive activity in some orderly and understandable way. Whenever possible, these should be presented using absolute, not relative risks. The estimated key benefits and harms of annual mammography for women in their 40s, 50s, and 60s are summarized in a figure (figure 10) [7,36,37]. Such an approach can help clinicians and patients understand what is involved when making the decision to screen. It can also help clarify why different individuals and expert groups come to different decisions about a preventive activity, even when looking at the same set of information. Different people put different values on benefits and harms [38].
Another approach to weighing benefits and harms is a modeling process that expresses both benefits and harms in a single metric and then subtracts harms from benefits. (The most common metric used is the quality-adjusted life year [QALY].) The advantage of this approach is that different types of prevention (eg, vaccinations, colorectal cancer screening, and tertiary treatment of diabetes) can all be compared with each other, which is important for policymakers with limited resources. The disadvantage is that for most clinicians and policymakers, it is difficult to understand the process by which benefits and harms are handled. Most models require numerous assumptions and interacting components that are not readily apparent.
Regardless of the method used in weighing the benefits and harms of preventive activities, the quality of the evidence for each benefit and harm must be evaluated to prevent the problem of “garbage in, garbage out.” Several groups making recommendations for clinical prevention have developed explicit methods to evaluate the evidence and take into account the strength of evidence when making their recommendations. (See "Overview of clinical practice guidelines", section on 'Evidence based'.)
If the benefits of a preventive activity outweigh the harms, the final step is to determine the economic effect of using it. Some commentators like to claim that “prevention saves money,” but it does so only rarely. (Exceptions are arising due to increased cancer costs, which make prevention relatively more attractive. For example, chemotherapy for colorectal cancer has become so expensive that some analyses now find screening for this cancer saves money.) Even so, most preventive services recommended by groups who have carefully evaluated the data are as cost-effective as other clinical activities.
Cost-effectiveness analysis is a method for assessing the costs and health benefits of an intervention. All costs related to disease occurrence and treatment should be counted, both with and without the preventive activity, as well as all costs related to the preventive activity itself. The health benefits of the activity are then calculated, often in terms of QALYs. The change in costs and benefits are compared with an existing standard, such as no screening or the current screening regimen. The cost-effectiveness is expressed as the incremental cost-effectiveness ratio: the incremental cost (the cost of the intervention minus the costs of the current standard) for each unit of incremental benefit (the benefit of the intervention minus the benefit of current standard).
●Example: Herpes zoster, a common condition that arises in older and immunocompromised adults, is characterized by a painful rash. The rash usually resolves within weeks, but about 10 percent of individuals experience post-herpetic neuralgia, pain at the rash site lasting months or longer; in rare cases, severe complications can occur such as loss of hearing or vision. Herpes zoster is caused by reactivation of the varicella zoster virus that persists in latent form after the initial infection that causes chickenpox. It can be prevented with vaccination in older adults.
A study was performed to assess the cost-effectiveness of a new vaccine for herpes zoster [39]. The effectiveness of the vaccine was based on randomized trial data showing rates of more than 90 percent reduction in the incidence of herpes zoster and post-herpetic neuralgia, and 85 percent after four years. In absolute terms, per 1000 patients the vaccine prevented nine cases of zoster and one case of post-herpetic neuralgia each year. These outcomes were translated into QALYs based on prior published estimates. For example, an average case of herpes zoster was equivalent to the loss of about 10 days of healthy life. Costs were from a societal perspective in the United States, considering the full range of costs no matter who is paying them. The costs included those that were direct (such as the costs of the vaccine, office visits, and treatment of complications) and indirect (such as loss of productivity and income from missed work). The study found that for adults aged 50 or older, vaccination cost USD $47,000 or less per QALY gained compared with no vaccination. In sensitivity analyses, the cost-effectiveness model was most affected by the duration of the vaccine effectiveness, though even if the vaccine effect lasted only 10 years rather than the assumed 19 years, the cost-effectiveness ratio remained less than $100,000 per QALY gained. In the United States, the usual threshold that indicates cost-effective care is below $50,000 to $100,000.
The effort to gather all the information needed to make a decision whether to conduct a preventive activity in clinical practice is not something a single clinician can accomplish, but when reviewing recommendations about prevention, individual clinicians can determine if the benefits and harms of the activity are presented in an understandable way and if the recommendation has taken into account the strength of the evidence. They can also look for estimates of cost-effectiveness. With these facts, they should be able to share with their patients the information they need. Patients can then make an informed decision about preventive activities that takes into account the scientific information and their individual values.
SUMMARY
●There are four major types of clinical preventive care: immunizations, screening, behavioral counseling (sometimes referred to as lifestyle changes), and chemoprevention. Screening is the identification of an asymptomatic disease, unhealthy condition, or risk factor. (See 'Types of clinical prevention' above.)
●This topic identifies primary prevention as interventions to keep disease from occurring (eg, immunization for communicable disease); secondary prevention as detection of early asymptomatic disease (eg, screening); and tertiary prevention as reducing complications of disease (eg, eye examinations in patients with diabetes). This nomenclature is applied differently by some other disciplines. (See 'Levels of prevention' above.)
●Three criteria are important when deciding what conditions to screen for (see 'Scientific approach to clinical prevention' above):
•The burden of suffering caused by the condition (including issues related to disease prevalence and severity)
•The effectiveness, safety, and cost of the preventive intervention or treatment
•The performance of the screening test
●There are several methodologic issues to consider in evaluating studies of screening effectiveness: whether data include both prevalence and incidence screens or prevalence only; and the potential for bias, including lead-time, length-time, and compliance bias. (See 'Methodologic issues in evaluating screening programs' above.)
●Adverse effects of screening include physical injury from some screening tests, negative labeling, false-positive tests resulting in unnecessary follow-up and overdiagnosis, and detection of incidentalomas. (See 'Unintended consequences of screening' above.)
●When reviewing recommendations about prevention, individual clinicians can determine if the benefits and harms of the activity are presented in an understandable way, using absolute rather than relative risks, and if the recommendation has taken into account the strength of the evidence. They can also look for estimates of cost-effectiveness. They should share this information with their patients to help them make an informed decision about preventive activities that takes into account the scientific information and their individual values. (See 'Weighing benefits against harms of prevention' above.)
ACKNOWLEDGMENT — The UpToDate editorial staff acknowledges Suzanne Fletcher, MD, and Robert Fletcher, MD, MSc, who contributed to an earlier version of this topic review.
