View source for Diagnostic test (medical)

===Recognize futility of testing when disease prevalence is extremely low===
<!-- under development -->
Using [[Bayes Theorem]] may allow recognition that there are some settings where testing can be considered futile. Two conditions are necessary to establish futility:
# Being able to estimate the post-test probability of disease by having all the necessary information to do this: [[sensitivity  and specificity]] and prevalence of disease.
# Evidence-based analysis of what post-test probability of disease is considered futile. For example, in the screening of [[HIV]], decision analysis calculates that screening should occur whenever prevalence is approximately 0.2%.<ref name="pmid17146064">{{cite journal |author=Paltiel AD, Walensky RP, Schackman BR, ''et al'' |title=Expanded HIV screening in the United States: effect on clinical outcomes, HIV transmission, and costs |journal=Ann. Intern. Med. |volume=145 |issue=11 |pages=797–806 |year=2006 |pmid=17146064 |doi=}}</ref> However, this type of analysis is not available for many diseases and, when is available, usually includes value judgments about futility and cost that may not be universally accepted judgments.

Examples where thresholds are established or implied to justify testing or treatment include:
* [[HIV]] screening - the threshold is very low.<ref name="pmid17146064" /> Screening is recommended even if the prevalence is as low as 0.2%.<ref name="pmid17146064"/>
* [[Influenza]] treatment - the threshold is higher as the stakes are lower. For elderly patients, treatment should be initiated if probability of disease is 13% or more.<ref name="pimd12965940">{{cite journal | author = Rothberg M, Bellantonio S, Rose D | title = Management of influenza in adults older than 65 years of age: cost-effectiveness of rapid testing and antiviral therapy. | journal = Ann Intern Med | volume = 139 | issue = 5 Pt 1 | pages = 321-9 | year = 2003 | url = http://www.annals.org/cgi/content/abstract/139/5_Part_1/321 | id = PMID 12965940}}</ref> while for younger patients the threshold is 30%.<ref name="pmid12361816">{{cite journal | author = Smith K, Roberts M | title = Cost-effectiveness of newer treatment strategies for influenza. | journal = Am J Med | volume = 113 | issue = 4 | pages = 300-7 | year = 2002 | doi = 10.1016/S0002-9343(02)01222-6 | id = PMID 12361816}}</ref><ref name="pimd12965940">{{cite journal | author = Rothberg M, Bellantonio S, Rose D | title = Management of influenza in adults older than 65 years of age: cost-effectiveness of rapid testing and antiviral therapy. | journal = Ann Intern Med | volume = 139 | issue = 5 Pt 1 | pages = 321-9 | year = 2003 | url = http://www.annals.org/cgi/content/abstract/139/5_Part_1/321 | id = PMID 12965940}}</ref>

In the absence of specific analysis, another approach to determining the appropriate threshold is to use precedent. For example, in potentially lethal diseases such as [[pulmonary embolism]]<ref name="pmid16738268">{{cite journal |author=Stein PD, Fowler SE, Goodman LR, ''et al'' |title=Multidetector computed tomography for acute pulmonary embolism |journal=N. Engl. J. Med. |volume=354 |issue=22 |pages=2317–27 |year=2006 |pmid=16738268 |doi=10.1056/NEJMoa052367}}</ref>, [[acute coronary syndrome]]<ref name="pmid9867725">{{cite journal |author=Selker HP, Beshansky JR, Griffith JL, ''et al'' |title=Use of the acute cardiac ischemia time-insensitive predictive instrument (ACI-TIPI) to assist with triage of patients with chest pain or other symptoms suggestive of acute cardiac ischemia. A multicenter, controlled clinical trial |journal=Ann. Intern. Med. |volume=129 |issue=11 |pages=845–55 |year=1998 |pmid=9867725 |doi=}}</ref>, and [[pneumonia]]<ref name="pmid8995086">{{cite journal |author=Fine MJ, Auble TE, Yealy DM, ''et al'' |title=A prediction rule to identify low-risk patients with community-acquired pneumonia |journal=N. Engl. J. Med. |volume=336 |issue=4 |pages=243–50 |year=1997 |pmid=8995086 |doi=}}</ref><ref name="pmid15684204">{{cite journal |author=Carratalà J, Fernández-Sabé N, Ortega L, ''et al'' |title=Outpatient care compared with hospitalization for community-acquired pneumonia: a randomized trial in low-risk patients |journal=Ann. Intern. Med. |volume=142 |issue=3 |pages=165–72 |year=2005 |pmid=15684204 |doi=}}</ref>, in the best of health care settings 2-4% of patients have their diagnosis missed.Therefore, the precedent would be that whenever a serious disease is estimated to have more than a 2%-4% prevalence, the disease should be sought.

A [[randomized controlled trial]] showed a small reduction in test ordering when a computer displayed very low probabilities that a test would be abnormal.<ref name="pmid3339821">{{cite journal |author=Tierney WM, McDonald CJ, Hui SL, Martin DK |title=Computer predictions of abnormal test results. Effects on outpatient testing |journal=JAMA |volume=259 |issue=8 |pages=1194–8 |year=1988 |pmid=3339821 |doi=}}</ref>
<!-- PMID: 9391938 might help here -->