Skip to content
PaperFren

Concept · medicine

Diagnostic Accuracy

Follow Diagnostic Accuracy — see important new research and changes in evidence.

Change log

What changed

Dated edits to this page's evidence: studies added or removed from a claim, claims added or withdrawn, and new explanations tagged here. Rewordings are not listed.

Show 6 earlier
  • Sep 14, 2026

    High sensitivity/specificity in a selected clinic sample does not guarantee useful predictive values in a low-prevalence screening population—the CA125 GP analysis is the cautionary case in this library.

  • Sep 14, 2026

    If sensitivity and specificity are high, positive predictive value will be high too.

    • Added a misconception
  • Sep 14, 2026

    Plasma NfL/GFAP dementia associations are the same kind of result as a smartphone AF diagnostic study.

    • Added a misconception
  • Sep 14, 2026

    Assay accuracy automatically means better patient outcomes.

    • Added a misconception
  • Sep 14, 2026

    Diagnostic Accuracy findings always generalise to every clinic.

    • Removed a misconception
  • Sep 3, 2026

    • Concept page published

Diagnostic accuracy studies estimate how well a test classifies disease—often via sensitivity, specificity, predictive values, or discrimination (AUC)—against a reference standard or outcome. Related work includes risk scores and prognostic biomarkers that forecast future events rather than confirm current disease.

A bedside assay with 95% sensitivity, a low-prevalence screening PPV, and a plasma protein that forecasts dementia years ahead are all “accuracy-ish,” but they answer different clinical questions and break in different ways.

Evidence

What the evidence shows

Drawn from 8 studies in this library. Each finding starts with a plain-language takeaway, then the denser detail. Supports means evidence for a finding; Challenges means evidence against a stated position; Qualifies marks scope with a short note on each study’s contribution. Challenged positions are labeled — they are not findings.

  • Phone pulse waveforms can detect atrial fibrillation accurately versus ECG in clinic samples.

    Smartphone photoplethysmography detected atrial fibrillation with sensitivity about 95.6% and specificity about 96.6% versus ECG in an analysed clinic sample (adequate signal required; pacing excluded).

    1 supporting · 2 qualifying

    Qualifies

    1. 1How well does CA125 find ovarian cancer in GP care?different setting — low-prevalence GP screening PPV, not clinic AF vs ECG
    2. 2Blood GFAP and NfL predict dementia years aheaddifferent question — long-horizon prognosis, not point-of-care arrhythmia detection

    Study comparison

    StudyRoleDesignNPopulationOutcome
    Smartphone PPG to detect atrial fibrillation2019SupportsCross-sectionalDiagnostic accuracy vs 12-lead ECG in primary careN=223 · 241 enrolled; 223 analysed after exclusionsPrimary-care patients screened for atrial fibrillation with FibriCheck PPGSensitivity and specificity of mobile PPG for AF vs 12-lead ECG
    How well does CA125 find ovarian cancer in GP care?2020Qualifiesdifferent setting — low-prevalence GP screening PPV, not clinic AF vs ECGCohortPopulation-based primary-care EHR cohortN=50780 · Women with CA125 tested in English primary careWomen having CA125 measured in English general practicePPV, sensitivity, specificity, and AUC of CA125 ≥35 U/ml for ovarian cancer
    Blood GFAP and NfL predict dementia years ahead2024Qualifiesdifferent question — long-horizon prognosis, not point-of-care arrhythmia detectionCohortUK Biobank prospective plasma GFAP/NfL with registry dementia follow-upN=48542 · 1,312 incident all-cause dementia over ~13 yearsUK Biobank adults with baseline Olink plasma GFAP and NfLIncident all-cause dementia / ADRD HRs and prediction AUC for GFAP and NfL
  • A multiplex assay can identify Staphylococcus from blood cultures with very high accuracy.

    A multiplex molecular assay on monomicrobial blood cultures identified Staphylococcus at genus level with sensitivity about 99.4% and specificity about 99.7% in a large evaluated set.

    1 supporting · 2 qualifying

    Qualifies

    1. 1Can a rapid test diagnose sickle cell at the bedside?different target — sickle-cell bedside detection, not blood-culture ID
    2. 2Smartphone PPG to detect atrial fibrillationdifferent modality — PPG AF screening, not microbiology ID
  • In GP care, CA125 looks accurate but a positive rarely means ovarian cancer when disease is rare.

    In GP care, CA125 ≥35 U/ml for ovarian cancer showed sensitivity about 77%, specificity about 93.8%, and AUC about 0.92, but PPV was only about 10.1% because incidence was about 0.9%—illustrating prevalence’s grip on predictive value.

    1 supporting · 2 qualifying

    Qualifies

    1. 1Smartphone PPG to detect atrial fibrillationhigher-prevalence clinic AF sample — PPV/NPV not transportable as raw sensitivities
    2. 2Diabetes prevalence and risk-score accuracyrelated lesson — risk-score AUCs with number-needed-to-screen trade-offs

    Study comparison

    StudyRoleDesignNPopulationOutcome
    How well does CA125 find ovarian cancer in GP care?2020SupportsCohortPopulation-based primary-care EHR cohortN=50780 · Women with CA125 tested in English primary careWomen having CA125 measured in English general practicePPV, sensitivity, specificity, and AUC of CA125 ≥35 U/ml for ovarian cancer
    Smartphone PPG to detect atrial fibrillation2019Qualifieshigher-prevalence clinic AF sample — PPV/NPV not transportable as raw sensitivitiesCross-sectionalDiagnostic accuracy vs 12-lead ECG in primary careN=223 · 241 enrolled; 223 analysed after exclusionsPrimary-care patients screened for atrial fibrillation with FibriCheck PPGSensitivity and specificity of mobile PPG for AF vs 12-lead ECG
    Diabetes prevalence and risk-score accuracy2008Qualifiesrelated lesson — risk-score AUCs with number-needed-to-screen trade-offsCross-sectionalSUNSET Amsterdam population sample; ethnicity-stratified screening criteriaN=1434 · 339 Hindustani Surinamese, 605 African Surinamese, 490 DutchAmsterdam adults aged 35–60 (Hindustani Surinamese, African Surinamese, Dutch)Diabetes prevalence and risk-score AUC / numbers-needed-to-screen by ethnicity
  • Diabetes risk scores trade accuracy against how many people you must screen locally.

    Diabetes risk scores in an Amsterdam sample showed AUCs about 0.74–0.80 with numbers-needed-to-screen of 3–7 depending on the outcome definition—accuracy metrics tied to local prevalence and cut-offs.

    1 supporting · 2 qualifying

    Qualifies

    1. 1How well does CA125 find ovarian cancer in GP care?different disease and prevalence regime — ovarian cancer screening PPV
    2. 2Blood GFAP and NfL predict dementia years aheaddifferent task — prognostic plasma proteins, not cross-sectional diabetes scoring

    Study comparison

    StudyRoleDesignNPopulationOutcome
    Diabetes prevalence and risk-score accuracy2008SupportsCross-sectionalSUNSET Amsterdam population sample; ethnicity-stratified screening criteriaN=1434 · 339 Hindustani Surinamese, 605 African Surinamese, 490 DutchAmsterdam adults aged 35–60 (Hindustani Surinamese, African Surinamese, Dutch)Diabetes prevalence and risk-score AUC / numbers-needed-to-screen by ethnicity
    How well does CA125 find ovarian cancer in GP care?2020Qualifiesdifferent disease and prevalence regime — ovarian cancer screening PPVCohortPopulation-based primary-care EHR cohortN=50780 · Women with CA125 tested in English primary careWomen having CA125 measured in English general practicePPV, sensitivity, specificity, and AUC of CA125 ≥35 U/ml for ovarian cancer
    Blood GFAP and NfL predict dementia years ahead2024Qualifiesdifferent task — prognostic plasma proteins, not cross-sectional diabetes scoringCohortUK Biobank prospective plasma GFAP/NfL with registry dementia follow-upN=48542 · 1,312 incident all-cause dementia over ~13 yearsUK Biobank adults with baseline Olink plasma GFAP and NfLIncident all-cause dementia / ADRD HRs and prediction AUC for GFAP and NfL
  • Higher blood GFAP and NfL associate with later dementia risk — prognosis, not same-day diagnosis.

    In UK Biobank, higher baseline plasma GFAP and NfL were associated with roughly doubled hazards of incident dementia over ~13 years, with incremental discrimination when added to demographic risk scores—prognostic biomarker performance, not a bedside diagnostic rule-in test.

    1 supporting · 2 qualifying

    Qualifies

    1. 1Smartphone PPG to detect atrial fibrillationdifferent task — detecting current AF vs ECG, not forecasting dementia
    2. 2Rising frailty flags one-year death riskrelated prognostic framing — frailty trajectory vs death, not plasma NfL

    Study comparison

    StudyRoleDesignNPopulationOutcome
    Blood GFAP and NfL predict dementia years ahead2024SupportsCohortUK Biobank prospective plasma GFAP/NfL with registry dementia follow-upN=48542 · 1,312 incident all-cause dementia over ~13 yearsUK Biobank adults with baseline Olink plasma GFAP and NfLIncident all-cause dementia / ADRD HRs and prediction AUC for GFAP and NfL
    Smartphone PPG to detect atrial fibrillation2019Qualifiesdifferent task — detecting current AF vs ECG, not forecasting dementiaCross-sectionalDiagnostic accuracy vs 12-lead ECG in primary careN=223 · 241 enrolled; 223 analysed after exclusionsPrimary-care patients screened for atrial fibrillation with FibriCheck PPGSensitivity and specificity of mobile PPG for AF vs 12-lead ECG
    Rising frailty flags one-year death risk2018Qualifiesrelated prognostic framing — frailty trajectory vs death, not plasma NfLCohortEnglish primary-care EHR; monthly eFI trajectories before death vs matched controlsN=26298 · 13,149 decedents and 13,149 matched controlsEnglish primary-care patients with electronic frailty index trajectories12-month mortality risk by rapidly rising vs stable frailty class
  • Other tools (sickle cell, pain subtype, frailty trajectories) answer their own accuracy questions.

    Other validated tools in this set include a point-of-care sickle-cell device detecting HbS/HbC at low percentages suitable for neonates, StEP pinprick signs highly sensitive/specific for neuropathic vs non-neuropathic pain in specialist clinics, and rapidly rising electronic frailty index trajectories associated with doubled 12-month mortality versus stable frailty.

    3 supporting · 2 qualifying

    Qualifies

    1. 1How well does CA125 find ovarian cancer in GP care?different prevalence and pathway — GP ovarian-cancer screening
    2. 2Blood GFAP and NfL predict dementia years aheaddifferent prognostic biomarker — plasma proteins for dementia

Open questions

Tensions and limits

Some items are genuine disagreements on the same question. Others mark different assays, populations, or outcomes — limits on how far one study travels — not a forced fight between papers.

  • Scope / different questions

    Point-of-care classification accuracy (AF, blood culture ID, sickle cell, pain subtype) is not the same claim as low-prevalence screening PPV or long-horizon prognostic discrimination. Mixing sensitivities with hazards or PPVs invents a false head-to-head.

    3 studies
    1. 1Smartphone PPG to detect atrial fibrillation
    2. 2How well does CA125 find ovarian cancer in GP care?
    3. 3Blood GFAP and NfL predict dementia years ahead

    Study comparison

    StudyRoleDesignNPopulationOutcome
    Smartphone PPG to detect atrial fibrillation2019SupportsCross-sectionalDiagnostic accuracy vs 12-lead ECG in primary careN=223 · 241 enrolled; 223 analysed after exclusionsPrimary-care patients screened for atrial fibrillation with FibriCheck PPGSensitivity and specificity of mobile PPG for AF vs 12-lead ECG
    How well does CA125 find ovarian cancer in GP care?2020SupportsCohortPopulation-based primary-care EHR cohortN=50780 · Women with CA125 tested in English primary careWomen having CA125 measured in English general practicePPV, sensitivity, specificity, and AUC of CA125 ≥35 U/ml for ovarian cancer
    Blood GFAP and NfL predict dementia years ahead2024SupportsCohortUK Biobank prospective plasma GFAP/NfL with registry dementia follow-upN=48542 · 1,312 incident all-cause dementia over ~13 yearsUK Biobank adults with baseline Olink plasma GFAP and NfLIncident all-cause dementia / ADRD HRs and prediction AUC for GFAP and NfL
  • Scope / different questions

    High sensitivity/specificity in a selected clinic sample does not guarantee useful predictive values in a low-prevalence screening population—the CA125 GP analysis is the cautionary case in this library.

    2 studies
    1. 1How well does CA125 find ovarian cancer in GP care?
    2. 2Smartphone PPG to detect atrial fibrillation

    Study comparison

    StudyRoleDesignNPopulationOutcome
    How well does CA125 find ovarian cancer in GP care?2020SupportsCohortPopulation-based primary-care EHR cohortN=50780 · Women with CA125 tested in English primary careWomen having CA125 measured in English general practicePredictive values of CA125 for ovarian cancer in low-prevalence primary careclaim outcome
    Smartphone PPG to detect atrial fibrillation2019SupportsCross-sectionalDiagnostic accuracy vs 12-lead ECG in primary careN=223 · 241 enrolled; 223 analysed after exclusionsPrimary-care patients screened for atrial fibrillation with FibriCheck PPGSensitivity/specificity of PPG for AF in a selected primary-care sampleclaim outcome

Common misconceptions

  • If sensitivity and specificity are high, positive predictive value will be high too.

    CA125 in GP care kept relatively high specificity while PPV stayed about 10% because ovarian cancer was rare (~0.9%). Prevalence drives predictive values.

    1. 1How well does CA125 find ovarian cancer in GP care?
  • Plasma NfL/GFAP dementia associations are the same kind of result as a smartphone AF diagnostic study.

    NfL/GFAP here are long-horizon prognostic associations with incident dementia. The PPG study classifies current AF against ECG. Different tasks and metrics.

    1. 1Blood GFAP and NfL predict dementia years ahead
    2. 2Smartphone PPG to detect atrial fibrillation
  • Assay accuracy automatically means better patient outcomes.

    Blood-culture ID and similar papers evaluate analytic/clinical classification performance; they do not by themselves prove that faster or more accurate reporting reduces mortality.

    1. 1Rapid gram-positive blood culture ID

Exam-style questions

Short-answer questions that ask you to explain or compare, not recall.

Why can a test with 94% specificity still have a disappointing PPV?

When disease is rare, even modest false-positive rates create many false positives relative to true positives. CA125 in GP care is the worked example (PPV ~10% at ~0.9% incidence).

How should you label NfL/GFAP dementia results on a diagnostic-accuracy page?

As prognostic discrimination / hazard associations for future dementia—not as a point-of-care rule-in test with sensitivity/specificity against a current gold-standard diagnosis.

The studies

8 studies in this library bear on Diagnostic Accuracy, ordered by citations.

  • StEP pain-subtype assessment

    A structured interview-plus-exam tool (StEP) was developed and validated to separate neuropathic/radicular from non-neuropathic axial low-back pain.

    PLoS medicine · 2009 · 190 citations

  • How well does CA125 find ovarian cancer in GP care?

    In UK primary care, CA125 ≥35 U/ml had high NPV but only about 10% PPV for ovarian cancer, and elevated values also flagged other cancers.

    PLoS medicine · 2020 · 102 citations

  • Blood GFAP and NfL predict dementia years ahead

    In 48,542 UK Biobank participants followed ~13 years, elevated plasma GFAP and NfL preceded dementia up to 15 years and improved prediction beyond CAIDE/DRS risk scores (AUC up to ~0.89).

    BMC medicine · 2024 · 98 citations

  • Smartphone PPG to detect atrial fibrillation

    FibriCheck PPG showed ~96% sensitivity and specificity versus cardiologist ECG on analysable traces.

    JMIR mHealth and uHealth · 2019 · 93 citations

  • Rising frailty flags one-year death risk

    Rapid monthly rises in the electronic frailty index marked older adults twice as likely to die within a year, with high specificity.

    BMC medicine · 2018 · 89 citations

  • Diabetes prevalence and risk-score accuracy

    Hindustani Surinamese had ~26% diabetes prevalence; a clinical risk score showed moderate-to-good AUCs (~0.74–0.80) across ethnic groups.

    BMC public health · 2008 · 87 citations

  • Rapid gram-positive blood culture ID

    A microarray panel identified Staphylococcus from positive blood cultures with ~99% sensitivity and specificity versus culture.

    PLoS medicine · 2013 · 86 citations

  • Can a rapid test diagnose sickle cell at the bedside?

    Sickle SCAN, a point-of-care immunoassay, was validated against laboratory gold standards for detecting hemoglobin S and related genotypes.

    BMC medicine · 2015 · 81 citations

Learn alongside