Table 4.

Accuracy of Clinical Prediction Rules for Influenza

Study or DatasetType of Validation and Number of PatientsAUROCCOverall Prevalence of Flu% Flu and LR by Risk Group (Derivation)% Flu and LR by Risk Group (Validation)% of Patients in Each Risk Group
FluScore risk score
    Ebell, 2012Split sample internal: derivation n = 326, validation n = 1330.79 *34.2%Low: 9.2%, LR 0.20Low: 5.0%, LR 0.10Low: 32.5% *
Mod: 27.8%, LR 0.76Mod: 35.9%, LR 1.02Mod: 28.1%
High: 59.1%, LR 2.83High: 57.4%, LR 2.47High: 39.4%
    van Vugt, 2015 (flu season)Prospective external: n = 5050.71 †23.6%NALow: 13.6%, LR 0.51Low: 60% ‡
Mod: 32.1%, LR 1.53Mod: 27%
High: 50.0%, LR 3.24High: 14%
Classification and Regression Tree (CART) models
    Afonso, 2012: Model 1Split sample internal: derivation n = 322, validation n = 1340.82/0.80 *34.2%Low: 5.6% flu, LR 0.12Low: 8%, LR 0.15Low: 21% *
Mod: 29% flu, LR 0.83Mod: 37%, LR 0.98Mod: 62%
High: 78% flu, LR 7.2High: 82%, LR 7.8High: 17%
    Afonso, 2012: Model 2Split sample internal: derivation n = 322, validation n = 1340.75/0.76 *34.2%Low: 5.6% flu, LR 0.12Low: 8%, LR 0.15Low: 21% *
Mod: 18% flu, LR 0.44Mod: 18%, LR 0.37Mod: 30%
High: 55% flu, LR 2.4High: 62%, LR 2.7High: 49%
    Afonso, 2012: Model 3Split sample internal: derivation n = 322, validation n = 1370.76/0.77 *34.2%Low: 7.5% flu, LR 0.16Low: 3%, LR 0.06Low: 24% *
Mod: 26% flu, LR 0.72Mod: 34%, LR 0.89Mod: 46%
High: 63% flu, LR 3.4High: 70%, LR 3.9High: 30%
    Anderson, 2018Prospective internal: derivation n = 3782 (2009 to 2013),validation n = 790 (2014)0.69 *32.7%Low: 21.9% flu, LR 0.52Low: 20.2% flu, LR NRLow: 71% †
High: 59.7% flu, LR 2.73High: 63.8%, LR NRHigh: 29%
    Vuichard-Gysin, 2019 (children)Split sample internal: derivation n = 819, validation n = 4220.77/0.74 $12.3%Low: 8% flu, LR 4.5Low: 8%, LR 0.72Low: 88.6% †
High: 36% flu, LR 4.5High: 34%, LR 4.3High: 11.4%
    Vuichard-Gysin, 2019 (adults)Split sample internal: derivation n = 627, validation n = 3230.80/0.75 $7.1%Low 4% flu, LR 0.59Low: 5%, LR 0.68Low: 89.3%†
High 30% flu, LR 5.8High: 26%, LR 4.6High: 10.7%
    Zimmerman, 2016Split sample internal: derivation n = 2087, validation n = 20860.68/0.69 $15.4%Low: 6% flu, LR 0.33Low 5%43% *
High: 23% flu, LR 1.63High: 23%57%
Other risk scores
    Dugas, 2019Split sample internal: derivation n = 1553, validation n = 388NR9.4%Low (0 to 2): 1.5%, LR 0.16Low (0 to 2): 3.3%, LR 0.25Low: 33.4% *
High (3+): 12.5%, LR 1.48High (3+): 16.2%, LR 1.40High: 66.6%
    Govaert, 1998 (post hoc)None, derivation only,n = 1791NR6.8%Low (0): 3.2%, LR 0.46;NALow: 64.5%†
Mod (1 to 2): 6.2%, LR 0.91Mod: 8.1%
High (≥ 3): 15%, LR 2.49High: 27.4%
    Ranjan, 2012None, derivation only,n = 638NR19.9%Low (0 to 6): 4%NANR
High (7+): 64%
    Vuichard-Gysin, 2019 (children)Split sample internal: derivation n = 819, validation n = 4220.76/0.70 $12.3%Low (0 to 4): 6% flu, LR 0.53Low (0 to 4): 7%, LR 0.59Low: 80.1% *
High (5+): 31% flu, LR 3.7High (5+): 28%, LR 3.1High: 19.9%
    Vuichard-Gysin, 2019 (adults)Split sample internal: derivation n = 627, validation n = 3230.78/0.79 $7.1%Low (0 to 3): 6% flu, LR 0.59;Low (0 to 3): 5%, LR 0.67Low: 94.2% *
High (4+): 34% flu, LR 5.8High (4+): 26%, LR 4.8High: 5.8%
    Woolpert, 2012 (original)None, derivation only,n = 523NR30.0%0 pts: 0.0%, LR 0.0NA0 pts: 0.6% *
1 pts: 10.1%, LR 0.261 pt: 15.1%
2 pts: 20.9%, LR 0.622 pts: 53.0%
3 pts: 50.0%, LR 2.333 pts: 26.0%
4 pts: 82.1%, LR 10.74 pts: 5.4%
    Woolpert, 2012 (post hoc)None, derivation only,n = 523NR30.0%Low (0 to 1): 9.8%, LR 0.25NALow: 15.7% *
Mod (2): 20.9%, LR 0.61Mod: 53.0%
High (3+): 55.5%, LR 2.87High: 31.4%

  • AUROCC, area under the receiver operating characteristic curve; EAST-PC, Enhancing Antibiotic Stewardship in Primary Care dataset; Mod, moderate; LR, likelihood ratio; NA, not applicable; NR, not reported.

  • ↵* All subjects.

  • ↵† Derivation group only.

  • ↵‡ Validation group only.

  • ↵$ Derivation/Validation.