Agreement in neonatal x-ray interpretation: a comparison between consultant paediatric radiologists and a reporting radiographer

X-rays are frequently used by clinicians to aid in patient diagnosis and management; infants which require specialised neonatal care are no exception. Accurate and prompt x-ray interpretation is of fundamental importance; however significant observer variation is reported in the literature. Increasing workloads in developed nations and limited specialized radiology resources in the developing world can act as barriers to the availability of a radiology report. In the United Kingdom, radiographer reporting is often used as a solution to manage radiology workload while maintaining a patient focused service.

In order to influence patient care the radiology report needs to aid the clinician in arriving at an accurate diagnosis. A paucity of recent evidence exists in the literature related to this aspect of x-ray interpretation, particularly for radiographer produced x-ray reports. The paediatric literature reveals a small evidence base; the majority of studies examine agreement between radiologists and paediatricians in the interpretation and diagnosis of pneumonia in children. There is little research evidence which examines agreement in neonatal x-ray interpretation. Previous studies have examined pathology scoring systems rather than clinical reports produced in routine practice.

Royal College of Radiologists and College of Radiographers guidance states that radiographers who interpret images must perform at a level comparable to radiologists; this hypothesis will be tested. If radiographers are trained to interpret neonatal x-rays and do so to a standard comparable to radiologists, they could provide additional reporting resource across healthcare settings worldwide.

Methods
Reports (n=101) independently produced by 2 paediatric radiologists(PCR1/2) and a reporting radiographer(RR) were compared for agreement by 2 neonatal consultants, blinded to the origin of the report. Agreement was assessed using Kappa statistic.

Results/Discussion
A clinically significant discrepancy between reports was found in 1/3 of cases. Kappa(κ) demonstrated fair agreement for RR-PCR1/PCR2 [κ=0.51, κ=0.35] and inter-PCR [κ=0.36](p<0.001). Fair agreement was found between neonatologists; RR/PCR1/PCR2[κ=0.31, κ=0.39] and inter-PCR[κ=0.33] reports(p<0.001). This study demonstrates fair agreement, both for x-ray interpretation and neonatologists’ interpretation of the report. There was comparable agreement between the radiographer and radiologists.