To reduce missed cases of pediatric abusive head trauma (AHT), Pediatric Brain Injury Research Network investigators derived a 4-variable AHT clinical prediction rule (CPR) with sensitivity of .96. Our objective was to validate the screening performance of this AHT CPR in a new, equivalent patient population.
We conducted a prospective, multicenter, observational, cross-sectional study. Applying the same inclusion criteria, definitional criteria for AHT, and methods used in the completed derivation study, Pediatric Brain Injury Research Network investigators captured complete clinical, historical, and radiologic data on 291 acutely head-injured children <3 years of age admitted to PICUs at 14 participating sites, sorted them into comparison groups of abusive and nonabusive head trauma, and measured the screening performance of the AHT CPR.
In this new patient population, the 4-variable AHT CPR demonstrated sensitivity of .96, specificity of .46, positive predictive value of .55, negative predictive value of .93, positive likelihood ratio of 1.67, and negative likelihood ratio of 0.09. Secondary analysis revealed that the AHT CPR identified 98% of study patients who were ultimately diagnosed with AHT.
Four readily available variables (acute respiratory compromise before admission; bruising of the torso, ears, or neck; bilateral or interhemispheric subdural hemorrhages or collections; and any skull fractures other than an isolated, unilateral, nondiastatic, linear, parietal fracture) identify AHT with high sensitivity in young, acutely head-injured children admitted to the PICU.
The pediatric diseases can be analyzed not only through the old- fashioned cause/effect, linear paradigm typical of the deterministic systems, but also in another, promising way. In touch with the recent "nonlinear" theories (1) , each child can be mathematically treated as a particle which crosses an abstract landscape made of valleys, peaks and basins, following an individual trajectory represented by a vector. In the landscape there are whirlpools, corresponding to diseases, in which the particles can be attracted, depending either on random walks, or on their peculiar characteristics (2): to make an example, an iron particle will be attracted by a magnet located in the whirlpool. The trajectories of the children who do not develop the disease follow a straight line, while the affected children's trajectories enter the whirlpools' spiral orbit. When and if a child recovers from the disease, his trajectory leaves the orbit and comes back to the landscape. However, if he does not recover, his trajectory ceaselessly persists into the disease zone. The nonlinear techniques could be useful for researchers investigating epidemiology and course of known (3) and unknown pediatric diseases. Indeed, such techniques are able to predict the evolution of trajectories in time: for each of the rounds around the "attractor", there is a characteristic nonlinear scaling law (4) that governs the amplitude and the period of the orbit and that can be quantified via differential equations.
BIBLIOGRAPHY:
1) Friston K, Ao P. Free energy, value, and attractors. Comput Math Methods Med. 2012; 2012:937860. doi: 10.1155/2012/937860.
2) Strogatz SH. Nonlinear Dynamics And Chaos: With Applications To Physics, Biology, Chemistry, And Engineering (Studies in Nonlinearity). 2001. Westview Press, 2001
3) Jiang Z, Du C, Jablensky A, Liang H, Lu Z, Ma Y, Teo KL. Analysis of schizophrenia data using a nonlinear threshold index logistic model. PLoS One. 2014 ; 9(10):e109454. doi: 10.1371/journal.pone.0109454.
4) Newman MEJ. Power laws, Pareto distributions and Zipf's law. Contemporary Physics. 2005; 46, 323-351. arXiv:cond-mat/0412004 [cond- mat.stat-mech]. DOI: 10.1016/j.cities.2012.03.001
Conflict of Interest:
None declared
Sir,
We read the article recently published on-line in the Pediatrics by Kent P. Hymel et al (1) with great interest, and we appreciate the authors' efforts to construct the clinical prediction rule and to validate it. However, if clinicians apply the clinical prediction rule in diagnostic practice, they should recognize the following two pitfalls.
First, the reported sensitivity is 0.93. That is excellent for a screening test, but the sensitivity may be overestimated due to review bias (2). Review bias may occur when the index test is not compared with the reference standard blindly each other (2). On the other hand, we consider that both reference standard and index test in the current study are difficult to assess blindly to each other due to characteristic of the elements. Therefore, we suggest that the reference standard is switched to patient's final diagnosis of external organization without knowledge of the clinical prediction rule. As a result, such bias would be minimum.
Secondly, spectrum bias (3) may occur due to the method of selecting the patient's population that is limited to participants only admitted in PICU. Given the inherent features of a screening test, such a test should be an adequate one to detect the disease at certain cutoff points, not solely in the severe injured population, but in the general population.
Hence, we would like to suggest that the current study needs validation in general outpatient population to overcome the spectrum bias, and additional research is needed to overcome the review bias.
References
1. Hymel KP, Armijo-Garcia V, Foster R, Frazier TN, Stoiko M, Christie LM, et al. Validation of a clinical prediction rule for pediatric abusive head trauma. Pediatrics. 2014 Dec;134(6):e1537-44.
2. Whiting P. Sources of Variation and Bias in Studies of Diagnostic Accuracy. Ann Intern Med. 2004 Feb 3;140(3):189.
3. Willis BH. Spectrum bias-why clinicians need to be cautious when applying diagnostic test studies. Fam Pract. 2008 Oct;25(5):390-6.
Conflict of Interest:
None declared