Screening for duct-dependent congenital heart disease with pulse oximetry: A critical evaluation of strategies to maximize sensitivity
Version of Record online: 2 JAN 2007
2005 Taylor & Francis
Volume 94, Issue 11, pages 1590–1596, November 2005
How to Cite
GRANELLI, A. D.-W., MELLANDER, M., SUNNEGÅRDH, J., &, K. S. and ÖSTMAN-SMITH, I. (2005), Screening for duct-dependent congenital heart disease with pulse oximetry: A critical evaluation of strategies to maximize sensitivity. Acta Paediatrica, 94: 1590–1596. doi: 10.1111/j.1651-2227.2005.tb01834.x
- Issue online: 2 JAN 2007
- Version of Record online: 2 JAN 2007
- (Received 1 December 2004; revised 5 April 2005; accepted 25 April 2005)
- Congenital heart defect;
- duct dependent;
- newborn examination;
- pulse oximetry;
Aim: To evaluate the feasibility of detecting duct-dependent congenital heart disease before hospital discharge by using pulse oximetry. Design: Case-control study. Setting: A supra-regional referral centre for paediatric cardiac surgery in Sweden. Patients: 200 normal term newborns with echocardiographically normal hearts (median age 1.0 d) and 66 infants with critical congenital heart disease (CCHD; median age 3 d).
Methods: Pulse oximetry was performed in the right hand and one foot using a new-generation pulse oximeter (NGoxi) and a conventional-technology oximeter (CToxi).
Results: With the NGoxi, normal newborns showed a median postductal saturation of 99% (range 94–100%); intra-observer variability showed a mean difference of 0% (SD 1.3%), and inter-observer variability was 0% (SD 1.5%). The CToxi recorded a significantly greater proportion of postductal values below 95% (41% vs 1%) in the normal newborns compared with NGoxi (p < 0.0001). The CCHD group showed a median postductal saturation of 90% (45–99%) with the NGoxi. Analysis of distributions suggested a screening cut-off of < 95%; however, this still gave 7/66 false-negative patients, all with aortic arch obstruction. Best sensitivity was obtained by adding one further criterion: saturation of < 95% in both hand and foot or a difference of > ± 3% between hand and foot. These combined criteria gave a sensitivity of 98.5%, specificity of 96.0%, positive predictive value of 89.0% and negative predictive value of 99.5%.
Conclusion: Systematic screening for CCHD with high accuracy requires a new-generation oximeter, and comparison of saturation values from the right hand and one foot substantially improves the detection of CCHD.