Causes and consequences of 'delays' in the diagnosis of spinal dural arterio-venous fistula; An illustrative case series, review and commentary.

BACKGROUND

Delays in management and misdiagnosis are common in patients with spinal dural arteriovenous fistula (sdAVF). In this study we review the incidence of delays, where they occur in the diagnostic process, who is responsible and what are the consequences.

METHODS

We report three recent cases that demonstrate the diagnostic challenges of sdAVF. We utilise, and extend, recent systematic and scoping reviews on sdAVF to obtain recent data on time to diagnosis and treatment. We propose a hierarchical flow-chart to document the diagnostic pathway. Using t his template where diagnostic errors and systemic shortcomings occur can be categorized and who is responsible and consequences identified. We comment on why delays occur and how practice can be improved.

RESULTS

The three patients had symptoms and signs suggestive of serious spino-radicular neurological dysfunction for between 15 and 24 months before referral to a spinal specialist led to diagnosis of sdAVF. In 58 series of sdAVF published between 2014-24 mean time to diagnosis was 14.6 months. There were no reports concerning misdiagnosis in primary care. The incidence of clinical 'misdiagnosis' in secondary (specialist) care varied between 17 and 79 % and radiological misdiagnosis (secondary and tertiary care) between 15-60 %. Misdiagnosis delayed correct diagnosis by between 6 and 12 months if clinical and 6-17 months if radiological and was associated with clinical deterioration. Many patients were initially diagnosed with either degenerative spinal disorders or other neurological disorders of the spinal cord and underwent surgical and medical treatments that worsened their clinical status. Where reported, time to treatment after diagnosis of sdAVF varied from 21-36 days. Many case series had insufficient information to either adjudge timelines or allocate where delays and misdiagnosis occurred. Most series recorded myelopathic dysfunction using the modified Aminoff-Logue scale for which no psychometric data (reliability, validity, reproducibility) exists.

CONCLUSIONS

Misdiagnosis of sdAVF occurs at all levels of medical care because its signs and symptoms mimic more common disorders. Transparent reporting of timelines to diagnosis and treatment, documentation of impairment and disability using validated assessment tools together with the incidence of and reasons for misdiagnosis in future studies of sdAVF are important to understand where and how time to diagnosis can be reduced. Population based studies with case ascertainment to national registries would facilitate unbiased collection of this data. Future application of Artificial Intelligence in primary, secondary and neuroradiological settings may reduce the incidence of misdiagnosis.