Diagnosis & Screening
Assessing the benefits and harms of diagnosis or screening is challenging. This handout provides an overview of different sources of bias that needs to be considered when evaluating a diagnostic test.
Lead time bias, length bias, and over-diagnosis
Lead time bias
Earlier diagnosis will lead to an apparent longer survival even when diagnosis and treatment does not affect or worsens the natural disease history. Lead time bias needs to be considered when a claim is made that earlier diagnosis improves survival.
Length time bias
Diseases that progress fast (e.g., the lead to mortality quickly) are less likely to be detected with diagnosis or screening than diseases that progress slowly or remain latent for long periods.
Overdiagnosis(Bleyer and Welch, 2012; Welch and Black, 2010)
The diagnosis of a screen-detected disease which would not go on to cause morbidity or mortality.Large natural variability in disease progression/regression causes difficulty in determining whose disease will progress to the stage of clinical symptoms, morbidity, and mortality. This is turns determines who needs and who does not need treatment. Chronic diseases including cancers may spontaneously regress. If a disease can regress, early diagnosis can become even more problematic.
Positive consequences of diagnostic testing
Can be expressed with different metrics. The gold standard for evaluating diagnostic testing is improved patient outcomes. A minimum requirement for a diagnostic test is that it has an impact on treatment decisions.
Negative consequences of diagnostic testing
-Psychological harm to patient of being labeled as “diseased”
-Physical harm from the diagnostic test itself (e.g. biopsy and surgery induced angiogenenis or cancer seeding(Retsky et al., 2007), ionizing radiation and cancer)
-Cost of diagnostic test which can reduce available resources for other diagnostic tests which may be more costly or treatment.
-Diagnostic tests can lead to a sequence of evidence of unnecessary or unhelpful treatments that may result in more harm than good.
-High false positive rates are a common challenge when screening for diseases with a low prevalence. Even a perfect sensitivity and a high specificity can result in a high rate of false positive when the prevalence of disease is low.
-Incidentalomas (e.g., (Fainstein Day et al., 2004))
Morediagnostic information is not necessarily better
-Additional information from ultrasound and computed tomography and misdiagnosis rate of appendicitis (Flum et al., 2005).
-Additional information can be “toxic” (Taleb, 2007) – More diagnostic test information can lead to more hypotheses that need to be explored and may provide more chances to get stuck on the wrong hypothesis.
-Additional diagnostic testing and rate of incorrect cancer diagnoses (Burton et al., 1998)
-Additional diagnostic information and orthodontic diagnoses (Han et al., 1991)
-The influence of cephalometrics on orthodontic treatmentplanning (European Journal of Orthodontics 30 (2008) 630–635).
-The use of cross-sectional CT on implant planning (
Most Reliable evidence comes from randomized controlled trials and systematic reviews of randomized controlled trials
-The most reliable evidence in favor of a diagnostic test is the conduct of a randomized controlled trial where access to the diagnostic test is randomized. If those individuals whom were randomly assigned to the diagnostic test have a higher probability of having a beneficial outcome than those without access to the diagnostic test, the diagnostic test is beneficial
-The largest and most expensive trials in the United States is the prostate, lung, colon, and ovarian (PLCO) cancer trial. The results demonstrated that screening for ovarian cancer(Buys et al., 2011)and prostate cancer (Andriole et al., 2009)failed to provide a benefit, screening with a CT for lung cancer improved survival outcomes(Reduced Lung-Cancer Mortality with Low-Dose Computed Tomographic Screening, 2011).
-Currently, there is only one randomized controlled trial the evaluated the impact or oral cancer screening on mortality. Theresults of this trial shows that there was no significant impact of screening on mortality but the trend was promising and in favor of screening(Sankaranarayanan et al., 2005).
-Many dental diagnostic tests have not been submitted to an evidence-based assessment.
Evidence-Based Assessment of diagnostic tests
-Aframework for assessing the clinical value of diagnostic tests was published by Tatsioni(Tatsioni et al., 2005) (freely available at
References
Andriole GL, Crawford ED, Grubb RL, 3rd, Buys SS, Chia D, Church TR et al. (2009). Mortality results from a randomized prostate-cancer screening trial. N Engl J Med 360(13):1310-1319.
Bleyer A, Welch HG (2012). Effect of three decades of screening mammography on breast-cancer incidence. The New England journal of medicine 367(21):1998-2005.
Burton EC, Troxclair DA, Newman WP, 3rd (1998). Autopsy diagnoses of malignant neoplasms: how often are clinical diagnoses incorrect? JAMA 280(14):1245-1248.
Buys SS, Partridge E, Black A, Johnson CC, Lamerato L, Isaacs C et al. (2011). Effect of screening on ovarian cancer mortality: the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Randomized Controlled Trial. JAMA 305(22):2295-2303.
Fainstein Day P, Guitelman M, Artese R, Fiszledjer L, Chervin A, Vitale NM et al. (2004). Retrospective multicentric study of pituitary incidentalomas. Pituitary 7(3):145-148.
Flum DR, McClure TD, Morris A, Koepsell T (2005). Misdiagnosis of appendicitis and the use of diagnostic imaging. J Am Coll Surg 201(6):933-939.
Han UK, Vig KW, Weintraub JA, Vig PS, Kowalski CJ (1991). Consistency of orthodontic treatment decisions relative to diagnostic records. Am J Orthod Dentofacial Orthop 100(3):212-219.
Reduced Lung-Cancer Mortality with Low-Dose Computed Tomographic Screening. (2011). N Engl J Med.
Retsky MW, Demicheli R, Gukas ID, Hrushesky WJ (2007). Enhanced surgery-induced angiogenesis among premenopausal women might partially explain excess breast cancer mortality of blacks compared to whites: an hypothesis. Int J Surg 5(5):300-304.
Sankaranarayanan R, Ramadas K, Thomas G, Muwonge R, Thara S, Mathew B et al. (2005). Effect of screening on oral cancer mortality in Kerala, India: a cluster-randomised controlled trial. Lancet 365(9475):1927-1933.
Taleb N (2007). The black swan : the impact of the highly improbable. 1st ed. New York: Random House.
Tatsioni A, Zarin DA, Aronson N, Samson DJ, Flamm CR, Schmid C et al. (2005). Challenges in systematic reviews of diagnostic technologies. Ann Intern Med 142(12 Pt 2):1048-1055.
Welch HG, Black WC (2010). Overdiagnosis in cancer. Journal of the National Cancer Institute 102(9):605-613.
Contact Information : Philippe Hujoel - email:
1