5.5. Prediction of neurosensory outcome

Fourteen of the preterm infants (27%) had at least one neurosensory disability, including 12 with CP, four of whom, together with two infants without CP, had permanent hearing loss. In terms of intrauterine growth, two of the 16 SGA infants had CP and two others permanent hearing loss. The mean (SD) birth weights of the infants with and without neurosensory disability were 942 (211) g and 1232 (274) g, respectively (p=0.01, independent samples t test), and their mean (SD) gestational ages at birth were 27.5 (1.5) weeks and 30.0 (2.0) weeks (p<0.01, independent samples t test).

The findings in the MRI and SPET examinations were compared to findings in US examinations for predicting CP. Numbers of the cases in examinations are summarised in table 11.

Table 11. Numbers (n) of preterm infants with normal and abnormal findings in brain magnetic resonance imaging (MRI) and single photon emission tomography (SPET) examinations at term age in relation to findings in ultrasound (US) examinations in preterm infants with and without cerebral palsy (CP and non-CP) as an outcome at a corrected age of 18 months.

Groups of infants with examinationsUS (n=51) MRI (n=50) SPET (n=34)
CPNon-CPNormal (n=34)Abnormal (n=16)Normal (n=18)Abnormal (n=16)
CPNon-CPCPNon-CPCPNon-CPCPNon-CP
Normal US without SPET114 01410 
Normal US with SPET3171141311225
Total (n)4311282311225
Abnormal US without SPET020101
Abnormal US with SPET8604821  472
Total (n)8805831  472
All (n)123913310621697
– = not done

Prediction of CP was most sensitive in the case of parenchymal lesion visible in MRI. Out of the ventricular-brain ratios, only an abnormal trigone index predicted CP in both US and MRI examinations. Wide extracerebral spaces or delayed myelination as seen in MRI did not predict CP. Brain perfusion SPET attained a sensitivity and specificity of the same magnitude as US for predicting all forms of CP, but its sensitivity in predicting moderate or severe CP was better (100% vs. 71%, but with a specificity of 67% vs. 74%, p=0.002 vs. 0.070). The predictive values of the different imaging methods for CP as recorded in the present series are summarised in Table 12.

Table 12. Performance of brain magnetic resonance imaging (MRI) and brain perfusion single photon emission tomography (SPET) at term age for the prediction of cerebral palsy as an outcome at a corrected age of 18 months in very low birth weight preterm infants with reference values for ultrasounds (US).

Imaging method and defects to be predictedNumber of infantsSensitivity (%)Specificity (%)Odds ratio95 % confidence intervalp-value
MRI50     
Parenchymal lesion1910079  <0.001
Parenchymal lesion, excluding subependymal haemorrhages108297171.013.9–2100.0<0.001
Abnormal trigone index4279714.21.3–155.20.029
Reference US51     
Parenchymal lesion14678511.02.5–48.40.001
Parenchymal lesion, excluding subependymal haemorrhages758100  <0.001
Abnormal trigone index1658764.51.1–17.70.031
MRI and/or US51     
Parenchymal lesion in both11679224.04.5–129.0<0.001
Parenchymal lesion in one21837212.72.4–67.60.002
SPET34     
Perfusion defect16827010.31.8–60.40.010
Cranio-caudal defect1373789.61.8–50.30.007
Cranio-caudal sensomotor or thalamic hypoperfusion1164838.31.6–42.60.011
Reference US34     
Parenchymal lesion12738312.72.3–70.00.005

Prediction of neurosensory disability was based on abnormal brainstem dimensions in MRI and unilateral or bilateral failures in components of BAEP. Brainstem dimensions in MRI gave poor sensitivity values of 23–31%, but their specificities reached 100%. BAEP components gained more significant predictive values, with sensitivities of 79–93% and specificities of 49–91%. The significant predictive values of brainstem dimensions and BAEP for neurosensory disability are summarised in Table 13.

Table 13. Brainstem dimensions in magnetic resonance imaging (MRI) and absence or abnormal peak and inter-peak latencies and the amplitude ratio V/I in brainstem auditory evoked potentials (BAEPs) at term age for the prediction of neurosensory disability in very low birth weight preterm infants at a corrected age of 18 months.

Imaging method and abnormal parametersNumber of infantsSensitivity (%)Specificity (%)Odds ratio95 % confidence intervalp-value
Brainstem MRI50     
Sagittal dimensions      
Mesencephalon 239710.81.0–115.40.049
Pons 31100  0.003
Axial dimensions      
Mesencephalon 31100  0.003
Pons 23100  0.015
BAEP51     
Peak latency I 935919.12.3–161.60.001
Peak latency III 79687.61.8–32.60.005
Peak latency V 86495.71.1–29.00.029
Inter-peak latency I–V 79687.61.8–32.60.005
Inter-peak latency III–V 935717.12.0–144.40.001
Amplitude ratio V/I 798930.35.8–156.7<0.001
Inter-peak latency III–Vand amplitude ratio V/I 799141.67.3–236.5<0.001
Inter-peak latency III–Vor amplitude ratio V/I 935717.12.0–144.40.001

Prediction of permanent hearing loss succeeded best with ABR, which had a sensitivity of 100%, whereas the TEOAE and FF auditory examinations achieved only a half of this figure. The predictive values of the various hearing assessments for permanent hearing loss in the series are summarised in Table 14.

Table 14. Auditory brainstem responses (ABR), transient evoked otoacoustic emissions (TEOAEs), and free-field auditory examination (FF) at term age for the prediction of permanent hearing loss in very low birth weight preterm infants at a corrected age of 18 months.

Screening methodNumber of infantsSensitivity (%)Specificity (%)Odds ratio95 % confidence intervalp-value
ABR51     
Bilateral fail  710098  <0.001
TEOAE44     
Bilateral fail  9  5084  5.30.9–33.00.089
FF51     
Fail  4  509842.03.3–536.80.004