Accuracy of Screening Instruments: Professional and Paraprofessional-Administered Screeners

Instrument and ComponentDecision Cutoff PointReferenceUSPSTF Quality RatingChild AgenReference InstrumentSensitivity, % (95% CI)Specificity, % (95% CI)PrevalenceaPPVa,bNPVa,bPLRaNLRa
Battelle Developmental Inventory Screening Test, Receptive<1 SDFrisk et al 200927Fair4.5 y110PLS-4 Receptive56 (33–78)70 (60–79)1626891.80.89
PLS-4 Expressive68 (49–88)86 (79–94)2056925.00.37
Brigance Preschool Screen<1 SDFrisk et al 200927Fair4.5 y110
 ReceptivePLS-4 Receptive61 (39–84)60 (50–70)1623894.21.5
 ExpressivePLS-4 Expressive91 (79–100)78 (70–87)2051970.120.65
Davis Observation Checklist for TexasNRAlberts et al 199536Fair52–67 mo59MSCA, GFTA, informal language sample80 (55–100)98 (94–100)17899639.20.20
Denver Articulation Screening Test<15th percentileDrumwright et al 197341Fair30–72 mo150Henja Articulation Test92 (—d)97 (—d)dddd
Denver Developmental Screening Test Language SectorNRBorowitz and Glascoe 198639Fair18–66 mo71PLS46 (34–58)100 (100–100)9210015e0.53
Developmental Nurse ScreenNRStokes 199748Fair34–40 mo378SLP rating using language sample, RDLS Comprehension76 (—d)96 (—d)d8096dd
Early Screening Profile Verbal Concepts<1 SDFrisk et al 200927Fair4.5 y110PLS-4 Auditory94 (84–100)68 (59–78)1640983.00.08
PLS-4 Expressive86 (72–100)81 (72–89)2053964.50.17
Fluharty Preschool Screening TestFailure ≥1 subtestsAllen and Bliss 198737Fair36–47 mo182SICD60 (41–79)81 (75–87)1433933.10.49
FPSLST ArticulationNRSturner et al 199350 study 1Fair53–68 mo51AAPS-R74 (—d)96 (—d)4c50ddd
FPSLST LanguageNRSturner et al 199350 study 1Fair53–68 mo51TACL-R38 (—d)85 (—d)17c42ddd
FPSLST ArticulationNRSturner et al 199350 study 2Fair55–69 mo147TD43 (—d)93 (—d)5c26cdd
FPSLST LanguageNRSturner et al 199350 study 2Fair55–69 mo147TOLD-P17 (—d)97 (—d)22c50ddd
Northwestern Syntax Screening TestFailure ≥1 subtestsAllen and Bliss 198737Fair36–47 mo182SICD92 (81–100)48 (41–56)1422971.80.16
SKOLDBliss and Allen 198438Fair
 Standard English
  SKOLDS30<1130–36 mo47SICD100 (100–100)98 (93–100)67510044.00
  SKOLDS37<1037–42 mo93SICD100 (100–100)91 (85–97)113310011.10
  SKOLDS43<1943–48 mo100SICD100 (100–100)93 (88–98)96010015.20
 African American dialect
  SKOLDB30<930–36 mo75SICD89 (68–100)86 (78–95)1247986.50.13
  SKOLDB37<1437–42 mo91SICD88 (65–100)86 (78–92)937996.00.15
  SKOLDB43<1943–48 mo54SICD94 (84–100)78 (64–91)3368974.20.07
Sentence Repetition Screening Test<20th percentileSturner et al 199651Fair54–66 mo323iAAPS-R57 (45–69)95 (93–98)19c12.5NR0.45NR
ITPA, Bankson62 (45–78)91 (87–94)11c6.6NR0.42NR
Structured Screening Test<10Laing et al 200244Good282RDLS66 (54–77)89 (85–94)2365906.2c0.38c
Hackney Early Language Screening Test, earlier version≤10Law 199445Good30 mo189RDLS98 (94–100)69 (61–77)2653993.170.03
Trial Speech Screening Test<12 elementsRigby and Chesham 198131Fair54 mo438SLP evaluation of Renfrew, RDLS, Edinburgh Articulation80 (68–92)93 (91–96)10589812.10.21
  • AAPS-R, Arizona Articulation Proficiency Scale: Revised; FPLST, Fluharty Preschool Speech and Language Screening Test; GFTA, Goldman-Fristoe Test of Articulation; ITPA, Illinois Test of Psycholinguistic Abilities; MSCA, McCarthy Scales of Children's Abilities; NR, not reported; PLS, Preschool Language Scale; RDLS, Reynell Developmental Language Scales; SICD, Sequenced Inventory of Communication Development; SKOLD, Screening Kit of Language Development; SLP, speech language pathologist; TACL-R, Test for Auditory Comprehension of Language – Revised; TD, Templin-Darley Tests of Articulation Consonant Singles Subtest; TOLD-P, Test of Language Development Primary.

  • a Calculated by EPC authors unless otherwise noted that study investigators provided data. Prevalence values were not estimated or weighted to reflect sampling

  • b Predictive values may be questionable for studies in which prevalence exceeded 10%; the problem arises when investigators choose a random sample of children with negative screens to complete the reference measures.

  • c Study investigators provided data.

  • d Could not calculate because of lack of data in article.

  • e Calculated as infinity.