Aim
The life cycle of iteration to implementation. Develop, validate, transpose, refine, measure usability and finally implement a diagnostic support tool (DST) for identification of causes of delirium.
Outcomes
Creation: A series of eight steps were formulated by an expert group to identify causes of delirium.
Validation: 40 inpatients admitted to a general medical unit with a consultant physician/geriatrician diagnosis of delirium were reviewed against the AiD-DST. Median number of causes of delirium detected on AiD-DST was 3 (IQR 3-4) versus 5 (IQR 3-6) using the reference standard diagnosis. Sensitivity of 88.8% (95% confidence interval, 81.6-93.9%) and specificity of 71.8% (63-79.5%).
Transposition: The final electronic version was built on Microsoft Blazor technology.
Refinement: software issues were identified, and modifications made.
Three cycles of feedback were obtained from 29 doctors. Content was grouped into themes of; 'style and grammar', 'formatting', 'missed diagnosis' and 'other concerns.'
Usability: 20 participants completed a standard usability (MAUQ) questionnaire. Average score was 6.36/7 (SD=0.8). This translates to agreement up to strong agreement concerning usability of AiD-DST.
Implementation: Reach: fifty-three out of 87 (61%) eligible doctors consented to participation in the study.
Effectiveness: A mean of 4.3 diagnoses were generated per patient with no difference in frequency when compared with historical control (z = 1.36; p = .17). Average usability score was 5.86 (SD = 1.15), with 93% of respondents being satisfied with the AiD-DST. Free text feedback comprised themes of accessibility, ergonomics, diagnostic accuracy and applicability of AiD-DST to related conditions.
Implementation: Instrument completion rate was 98% (n = 49/50), with a median completion time of 90 s. Maintenance: Sixty-seven % of uses of AiD-DST occurred in the second half of the study (p = .3). Following the initiation period there was an increase in use (r = .79; p = 02).
Background
Delirium represents an episode of new confusion caused by usually acute medical problems. Delirium affects one in four of hospitalized older patients or over 130,00 cases per year in Australia.
Delirium is associated with a tripling of mortality in relation to aged matched controls and costs over $8 billion per year in Australia alone. The causes of delirium are multifactorial and heterogeneous with correct identification representing a clinical challenge with evidence that causes are missed in every other patient associated with a quadrupling of mortality.
The Australian national safety and quality standard for delirium identifies that the cause of delirium is an area of quality focus but one for which there are up to now no currently available diagnostic support tools.
The challenge of creating a diagnostic support tool in which the combination of causes in any given patient is almost infinite renders the problem beyond the scope of traditional lists or heuristics.
Using Bayesian theory applied to clinical method we crafted a diagnostic support tool from first principles and have since successfully undertaken a complete cycle of validation and translation.
