Can Latent Class Analysis Be Used to Improve the Diagnostic Process in Pediatric Patients with Chronic Ataxia?

Chronic ataxia is a relatively common symptom in children. There are numerous causes of chronic ataxia, making it difficult to derive a diagnosis in a timely manner. We hypothesized that the efficiency of the diagnostic process can be improved with systematic analysis of clinical features in pediatric patients with chronic ataxia. Our aim was to improve the efficiency of the diagnostic process in pediatric patients with chronic ataxia. A cohort of 184 patients, aged 0–16 years with chronic ataxia who received medical care at Winnipeg Children’s Hospital during 1991–2008, was ascertained retrospectively from several hospital databases. Clinical details were extracted from hospital charts. The data were compared among the more common diseases using univariate analysis to identify pertinent clinical features that could potentially improve the efficiency of the diagnostic process. Latent class analysis was then conducted to detect unique patterns of clinical features and to determine whether these patterns could be associated with chronic ataxia diagnoses. Two models each with three classes were chosen based on statistical criteria and clinical knowledge for best fit. Each class represented a specific pattern of presenting symptoms or other clinical features. The three classes corresponded to a plausible and shorter list of possible diagnoses. For example, developmental delay and hypotonia correlated best with Angelman syndrome. Specific patterns of presenting symptoms or other clinical features can potentially aid in the initial assessment and diagnosis of pediatric patients with chronic ataxia. This will likely improve the efficiency of the diagnostic process.