A mobile app for delirium screening

Abstract

Objective The objective of this study is to describe the algorithm and technical implementation of a mobile app that uses adaptive testing to assess an efficient mobile app for the diagnosis of delirium. Materials and Methods The app was used as part of a NIH-funded project to assess the feasibility, effectiveness, administration time, and costs of the 2-step delirium identification protocol when performed by physicians and nurses, and certified nursing assistants (CNA). The cohort included 535 hospitalized patients aged 79.7 (SD = 6.6) years enrolled at 2 different sites. Each patient was assessed on 2 consecutive days by the research associate who performed the reference delirium assessment. Thereafter, physicians, nurses, and CNAs performed adaptive delirium assessments using the app. Qualitative data to assess the experience of administering the 2-step protocol, and the app usability were also collected and analyzed from 50 physicians, 189 nurses, and 83 CNAs. We used extensible hypertext markup language (XHTML) and JavaScript to develop the app for the iOS–based iPad. The App was linked to Research Electronic Data Capture (REDCap), a relational database system, via a REDCap application programming interface (API) that sent and received data from/to the app. The data from REDCap were sent to the Statistical Analysis System for statistical analysis. Results The app graphical interface was successfully implemented by XHTML and JavaScript. The API facilitated the instant updating and retrieval of delirium status data between REDCap and the app. Clinicians performed 881 delirium assessments using the app for 535 patients. The transmission of data between the app and the REDCap system showed no errors. Qualitative data indicated that the users were enthusiastic about using the app with no negative comments, 82% positive comments, and 18% suggestions of improvement. Delirium administration time for the 2-step protocol showed similar total time between nurses and physicians (103.9 vs 106.5 seconds). Weekly enrollment reports of the app data were generated for study tracking purposes, and the data are being used for statistical analyses for publications. Discussion The app developed using iOS could be easily converted to other operating systems such as Android and could be linked to other relational databases beside REDCap, such as electronic health records to facilitate better data retrieval and updating of patient’s delirium status. Conclusion Our app operationalizes an adaptive 2-step delirium screening protocol. Its algorithm and cross-plat formed code of XHTML and JavaScript can be easily exported to other operating systems and hardware platforms, thus enabling wider use of the efficient delirium screening protocol that we have developed. The app is currently implemented as a research tool, but with adaptation could be implemented in the clinical setting to facilitate widespread delirium screening in hospitalized older adults.

Comparison of delirium detection tools in acute care

“…MOTYB to identify inattention 2. If inattention present → SPMSQ for identifying cognitive impairment, Comprehension subtest of CTD for identifying disorganized thinking CAM algorithm: 1 + 2 + (3 or 4) positive = suspected delirium < 5 min Total sample: – DSM-IV criteria [ 18 , 22 ] Two-step approach; operationalization of core features 0.90 0.98 Patients with dementia: 0.91 0.87 Patients without dementia: 0.89 0.99 UB-CAM (Ultra-brief-CAM) General medicine patients ≥ 75 years ( n = 201) (+ collateral history) I: trained physician/nurse Screening 2–15 items UB‑2, in the case of an incorrect answer followed by a modified 3D-CAM (assessment of each CAM feature is stopped after one incorrect answer or positive observation item of that feature) CAM algorithm: 1 + 2 + (3 or 4) positive = suspected delirium 2 min 0.93 0.95 – 3D-CAM [ 4 , 25 , 40 ] Retrospective simulation based on 3D-CAM and UB‑2 data of two studies Other tools than CAM 4AT (4 As test) Acute care and rehabilitation patients ≥ 70 years ( n = 234) + collateral history I: untrained geriatrician Screening 4 items: alertness AMT‑4 Attention (MOTYB) Acute change or fluctuation Score 0–12 0 = no CI or delirium 1–3 = possible CI ≥ 4 = possible delirium < 5 min 0.90 0.84 AUC 0.89–0.93 DSM-IV criteria [ 5 ] Information on acute change/fluctuation not mandatory for delirium diagnosis; no special training required; includes MOTYB BCS (bedside confusion scale) Palliative patients ( n = 31) I: rater (no requirements) Screening Tool 2 items: psychomotor activity + MOTYB Score 0–5 Cut-off: ≥ 2 = suspected delirium 2 min 1.0 0.85 ...…”

“…This smartphone app tests attention by presentation and recall of a sequence of symbols [ 56 ]. Other smartphone applications offer digital versions of common interview-based assessment tools facilitating documentation and practicability of the assessments [ 4 ].…”

Quality and Safety in the Literature: September 2022

“…In their cohort study, Marcantonio and colleagues sought to assess the feasibility and accuracy of the Researching Efficient Approaches to Delirium Identification (READI) app, a tablet computer-based application derived from the Research Electronic Data Capture programming14 to aid in the diagnosis of delirium, among three groups of healthcare providers: certified nursing assistants (CNAs), registered nurses and hospitalist physicians 15. The READI app contains two screening tools for delirium detection at the bedside: the ultra-brief 2-item screen (UB-2) and the 3-Minute Diagnostic Assessment for the Confusion Assessment Method (3D-CAM) 16 17.…”

A pilot study testing the iOS UB‐CAM delirium app

“…6 To promote ease of bedside use and integration into clinician workflow, 7 we previously developed, tested, and implemented a UB-CAM app using the Application Programming Interface (API) with the Research Electronic Data Capture (REDCap). 8 To further improve accessibility and sustainability in the clinical setting, our research team collaborated with a computer scientist to develop and refine an iOS-based UB-CAM app for the iPhone and iPad. Through iterative testing we refined the app to include explicit written instructions and color cueing to enhance the flow of screening (Figure 1).…”