Residency · Residency · Preventive Medicine
Cancer Screening Principles: Benefits, Harms, and Overdiagnosis
Overview
Cancer screening aims to detect cancer or precancerous lesions in asymptomatic individuals to reduce mortality and morbidity. Effective screening requires: detectable preclinical phase, accurate test, effective treatment of early-stage disease, and net benefit exceeding harms. Not all cancers benefit from screening -- the target condition must have a natural history amenable to early intervention. Key harms include false positives, overdiagnosis, overtreatment, psychological distress, and procedural complications. Wilson and Jungner criteria (1968) remain foundational, supplemented by modern frameworks emphasizing net benefit, equity, and shared decision-making.
Foundational Principles
Wilson and Jungner Criteria (1968)
Important health problem. Accepted treatment available. Facilities for diagnosis and treatment available. Recognizable latent or early symptomatic stage. Suitable test available. Test acceptable to the population. Natural history of the condition adequately understood. Agreed-upon policy on whom to treat. Cost-effective (cost of case-finding balanced against total medical expenditure) Ongoing process, not a "once and for all" project.
Modern Additions
Net benefit must outweigh net harm at the population level. Health equity considerations: screening must not exacerbate disparities. Shared decision-making: patients must be informed of both benefits and harms. Consideration of overdiagnosis and its downstream effects. System capacity: follow-up pathways must be in place before screening is implemented.
Test Performance Characteristics
Sensitivity and Specificity
Sensitivity (true positive rate): proportion of those with disease who test positive; high sensitivity minimizes missed cases (false negatives) Specificity (true negative rate): proportion of those without disease who test negative; high specificity minimizes false positives. Trade-off: increasing sensitivity typically decreases specificity and vice versa (receiver operating characteristic curve) Screening tests generally prioritize high sensitivity (rule out disease)
Predictive Values
Positive predictive value (PPV): probability that a positive test result is a true positive; heavily influenced by disease prevalence. Negative predictive value (NPV): probability that a negative test result is a true negative. In low-prevalence populations (most screening scenarios), even a test with high sensitivity and specificity will have a low PPV -- most positive results will be false positives. Example: a test with 95% sensitivity and 95% specificity in a population with 1% prevalence has a PPV of only ~16%.
Likelihood Ratios
Positive likelihood ratio (LR+): sensitivity / (1 - specificity); how much a positive test increases the odds of disease. Negative likelihood ratio (LR-): (1 - sensitivity) / specificity; how much a negative test decreases the odds of disease. LR+ >10 or LR- <0.1 are considered strong; independent of prevalence.
Number Needed to Screen (NNS)
Number of individuals who must be screened to prevent one death or one case of advanced disease. Example: mammography NNS to prevent one breast cancer death over 10 years: ~1,000-2,000 for women 50-59. Communicates the yield of screening at the population level.
| Test Characteristic | Definition | Clinical Significance |
|---|---|---|
| Sensitivity | Proportion of diseased who test positive | High sensitivity minimizes missed cases (rule out) |
| Specificity | Proportion of non-diseased who test negative | High specificity minimizes false positives (rule in) |
| PPV | Probability that positive test is true positive | Heavily dependent on prevalence; low in screening |
| NPV | Probability that negative test is true negative | High in low-prevalence populations |
| LR+ | Sensitivity / (1 - specificity) | >10 is strong; shifts post-test probability up |
| LR- | (1 - sensitivity) / specificity | <0.1 is strong; shifts post-test probability down |
| NNS | Number needed to screen to prevent one death/case | Communicates population-level yield |
| Bias | Mechanism | Effect on Apparent Benefit | How to Avoid |
|---|---|---|---|
| Lead-time bias | Earlier detection without delayed death | Inflates survival statistics | Use mortality endpoints in RCTs |
| Length-time bias | Preferential detection of indolent cancers | Overestimates screening benefit | Randomized trial design |
| Overdiagnosis bias | Detection of cancers that would never cause harm | Inflates incidence and survival | Compare mortality rates, not survival |
| Healthy screenee bias | Volunteers are healthier than non-volunteers | Screened group appears healthier | Randomized trial design |
Biases in Screening Evaluation
Lead-Time Bias
Screening detects disease earlier in its natural history, making survival from diagnosis appear longer even if the time of death is unchanged. Survival statistics (5-year survival) are inflated by lead time; only mortality rate reduction in RCTs is valid evidence of screening benefit.
Length-Time Bias
Screening preferentially detects slowly progressive cancers (long preclinical detectable phase) Aggressive, fast-growing cancers are more likely to present as interval cancers (between screening rounds) Results in overrepresentation of indolent cancers among screen-detected cases.
Overdiagnosis Bias (Pseudodisease)
Detection of cancers that would never have become clinically apparent during the person's lifetime. Contributes to apparent increases in incidence and survival without corresponding mortality reduction. The screened population appears to have better outcomes simply because their cancers include overdiagnosed cases.
Selection Bias (Healthy Screenee Bias / Volunteer Bias)
Individuals who participate in screening tend to be healthier and more health-conscious. Observational studies of screening benefit are confounded by this effect. RCTs minimize this bias through randomization.
Overdiagnosis
Definition and Mechanism
A cancer is overdiagnosed when it is detected by screening but would not have caused symptoms, morbidity, or death during the patient's remaining lifetime. Occurs because some cancers are inherently indolent, non-progressive, or regress; or because the patient would have died of another cause before the cancer became symptomatic. Cannot be identified at the individual level -- can only be estimated at the population level.
Magnitude by Cancer Type
Prostate cancer (PSA screening): estimated overdiagnosis rate ~20-50% of screen-detected cancers. Breast cancer (mammography): estimated ~10-25% of screen-detected invasive cancers; higher for DCIS. Thyroid cancer: massive overdiagnosis in countries with widespread ultrasound screening (South Korea: 15-fold incidence increase with no mortality change) Lung cancer (LDCT): estimated ~3-25% overdiagnosis. Melanoma: rising incidence with stable mortality suggests significant overdiagnosis of thin melanomas.
Consequences of Overdiagnosis
Unnecessary surgery, radiation, chemotherapy with associated morbidity. Psychological burden: anxiety, depression, cancer identity. Financial toxicity: treatment costs, insurance implications. Diversion of healthcare resources from patients with clinically significant disease.
Shared Decision-Making in Cancer Screening
Components
Informing patients of potential benefits (mortality reduction, catching cancer early) Informing patients of potential harms (false positives, overdiagnosis, complications of diagnostic workup) Using absolute rather than relative risk reduction for transparent communication. Incorporating patient values and preferences into the screening decision. Decision aids: visual tools (icon arrays, pictographs) improve patient understanding.
Where SDM Is Most Important
USPSTF C-grade recommendations (selective offering based on individual factors) PSA screening for prostate cancer (USPSTF C for ages 55-69) Lung cancer screening (USPSTF B, but SDM explicitly required) Breast cancer screening (start age and interval vary by guideline)
<image>A 2x2 contingency table illustrating sensitivity, specificity, PPV, and NPV with a worked example using a hypothetical cancer screening test (sensitivity 90%, specificity 95%, prevalence 1%). The table shows the number of true positives, false positives, false negatives, and true negatives in a population of 10,000. The PPV is calculated showing that even with a "good" test, most positives are false positives in low-prevalence screening. Cancer screening test characteristics education illustration.</image>
<image>A diagram illustrating lead-time bias in cancer screening. Two parallel timelines compare a screened and an unscreened patient with the same cancer. The screened patient is diagnosed earlier (at the preclinical detectable phase), while the unscreened patient is diagnosed at symptom onset. Both die at the same time. The "survival" from diagnosis appears longer in the screened patient (lead time) without any actual change in outcome. Lead-time bias education illustration.</image>
<image>An infographic showing estimated overdiagnosis rates for major cancers detected through screening: prostate cancer via PSA (20-50%), breast cancer via mammography (10-25%), thyroid cancer via ultrasound (up to 80-90% in high-screening populations), lung cancer via LDCT (3-25%), and melanoma (indeterminate but suspected). Each cancer is paired with an icon and a brief explanation of the clinical significance of overdiagnosis in that context. Cancer screening overdiagnosis education illustration.</image>
<image>An icon array (pictograph) for communicating breast cancer screening outcomes to patients: 1,000 women screened with mammography over 10 years. Icons show: 1-2 lives saved from breast cancer death, 5-15 overdiagnosed, 100-200 false positive results leading to additional testing, 20-30 undergoing unnecessary biopsy, and 800+ receiving reassurance. This visual tool is an example of a decision aid for shared decision-making. Cancer screening communication education illustration.</image>
Clinical Pearls
In low-prevalence populations, even highly accurate screening tests produce more false positives than true positives -- understanding PPV and prevalence is essential for counseling patients. Five-year survival rates cannot be used to evaluate screening effectiveness because of lead-time and length-time biases -- only RCTs showing mortality reduction provide valid evidence. Overdiagnosis is the most important harm of cancer screening that patients are least likely to understand -- thyroid cancer screening in South Korea (15x incidence increase, flat mortality) is the clearest cautionary example. Shared decision-making is mandatory for PSA screening and lung cancer screening -- not optional. For boards: be able to calculate PPV from sensitivity, specificity, and prevalence; distinguish lead-time from length-time bias; define overdiagnosis; and apply Wilson and Jungner criteria.
References
- Wilson JMG, Jungner G. Principles and Practice of Screening for Disease. WHO; 1968.
- Welch HG, Black WC. Overdiagnosis in cancer. J Natl Cancer Inst. 2010;102(9):605-613.
- Ahn HS, et al. Korea's thyroid cancer "epidemic" -- screening and overdiagnosis. N Engl J Med. 2014;371(19):1765-1767.
- Hoffman RM, et al. Screening for prostate cancer. Cochrane Database Syst Rev. 2014;(1):CD004720.
- USPSTF. Screening for breast cancer. JAMA. 2024;331(22):1918-1930.
