Skip to main content

Table 3 Studies that investigated subgroup variations*

From: A methodological framework to distinguish spectrum effects from spectrum biases and to assess diagnostic and screening test accuracy for patient populations: Application to the Papanicolaou cervical cancer smear test

Study, Year

Approach to the analysis of spectrum bias/effect

Disease (number of subjects)

Diagnostic or screening test

Factors investigated

Effect on sensitivity and/or specificity

Effect on Likelihood ratios

Van der schouw, 1995 [5]

Subgroup analysis

Epididymitis (372)

Ultrasonography

Disease prevalence (severity according to clinician)

Disease prevalence (severity according to clinician)

Disease prevalence (severity according to clinician)

Morise, 1995 [36]

Subgroup analysis

Coronary disease (4467)

Exercise electrocardiography

Sex Verification by the gold standard (angiography) or not

Sex Verification bias by the gold standard (angiography) or not

Not considered

O'Connor, 1996 [6]

Subgroup analysis

Multiple sclerosis (303)

Magnetic resonance imaging and evoked potentials

Study group (two pooled studies considered), disease prevalence, clinical subjective disease probability

Study group, disease prevalence, clinical subjective disease probability

Not considered

Egglin, 1996 [37]

Subgroup analysis

Pulmonary embolism (24)

Pulmonary arteriogram

Disease prevalence

Disease prevalence

Not considered

Roger, 1997 [8]

Subgroup analysis

Coronary disease (3679)

Exercise echocardiography

Sex

Sex

Not considered

Curtin, 1997 [7]

Subgroup analysis

Obesity (226)

Body mass index

Weight, sex

Weight, sex

Not considered

Moons, 1997 [9]

Subgroup analysis and modelling of sensitivity

Coronary disease (295)

Exercise test

Patient history and clinical examination, various disease-related factors (maximal load, relative load, systolic blood pressure, number of diseased vessels)

Sex, expected load, maximal load, relative load, systolic blood pressure (baseline and peak), number of diseased vessels

Age, sex, symptoms, smoking, beta blocker use, cholesterol level, expected load, maximal load, relative load, systolic blood pressure (baseline and peak)

Santana-Bodao, 1998 [38]

Subgroup analysis

Coronary disease (702)

Single-photon emission computed tomography

Sex

Sex

Not considered

Steinbauer, 1998 [10]

Subgroup analysis

Alcohol abuse (1333)

Various screening tests

Race and sex

Race and sex

Race and sex

DiMatteo, 2001 [39]

Subgroup analysis

Group A beta haemolytic streptococcal pharyngitis (498)

Rapid antigen detection test

Disease severity

Disease severity

Not considered

Filly, 2002 [48]

Subgroup analysis and modelling

Cirrhosis (100)

Nodularity of surfaces of the liver

Deep versus superficial surfaces

Deep versus superficial surfaces

Age, sex, pathological type of cirrhosis

Hall, 2004 [40]

Subgroup analysis and modelling

Group A beta haemolytic streptococcal pharyngitis (561)

Rapid antigen detection test

Disease severity

Disease severity

Not considered

Meideros, 2005 [49]

Subgroup analysis

Glaucoma (136)

Scanning laser polarimetry

Two forms of glaucomatous optic neuropathy

Two forms of glaucomatous optic neuropathy

Two forms of glaucomatous optic neuropathy

  1. * Papers published between 1978 and 2000 were identified from the systematic review by Whiting et al [14]. Papers published between 2000 and 2005 were similarly selected through Medline® using the keywords: diagnostic, test, screening, performance, accuracy, sensitivity, specificity, likelihood ratio, spectrum, subgroup, bias, prevalence, accuracy. Only primary studies investigating variations in diagnostic test performances between subgroups were considered. Letters were excluded.