full title

Using Cognitive Task Analysis in a Medical Setting

keywords

Cognitive task analysis; decision making; neonatal intensive care; respiratory distress syndrome; expert systems

summary

New medical systems often fail to gain acceptance by by staff because they do not integrate with local work practices. An expert system, FLORENCE (Fuzzy LOgic REspiratory Neonatal Care Expert), is being developed to help clinical staff in a neonatal intensive care unit (NICU) make decisions about ventilator settings when treating babies with Respiratory Distress Syndrome. For FLORENCE to succeed it must be dependable—that includes being both clinically useful and acceptable to staff—in the context of local work practices. The aim of this work was to identify those contextual factors that would affect FLORENCE’s success.

Methods:
A cognitive task analysis (CTA) of the NICU was performed. First, work context analysis was carried out to identify how work is performed in the NICU using a combination of interviews and rich pictures. Second, the critical decision method (CDM) was used to analyse in detail how staff make decisions about changing the ventilator settings during clinical incidents. Third, naturalistic observation of staff’s routine use of the ventilator in situ was performed.

Results:
A. The work context analysis uncovered a hierarchical communication structure in the NICU and highlighted the importance of clinical records in communication.
B. It also identified important ergonomic and practical requirements for designing the displays and locating the computer equipment.
C. The CDM interviews suggested instances where problems could arise if the data used by FLORENCE, which is automatically read, is not manually verified by staff.
D. Observation showed that most of the equipment alarms cleared automatically. When FLORENCE raises an alarm, staff will usually be required to intervene and make a clinical judgement, even if this does not involve changing the ventilator settings.

Conclusions:
The particular combination of methods used in the CTA (work context analysis, CDM and observation) proved to be very successful in identifying issues in the NICU that have implications for FLORENCE. These implications go beyond the functional capability of FLORENCE to calculate changes to the ventilator settings. The fact that several of the implications were identified as the results of applying more than one of the methods serves to confirm the importance of the implications.

Work context analysis is not normally mentioned in texts on CTA. On the basis of its usefulness here, in identifying the underlying work structures in the practice of neonatal care and generating implications for the acceptability of FLORENCE, it should be considered more often when carrying out a CTA. In particular it has helped to identify the areas where FLORENCE can provide support for current practice (such as prompting staff to follow the DOPE mnemonic to check the ventilator equipment), where it could change existing practice (such as the need for the data records produced by FLORENCE to be incorporated into the patient records, and highlighting possible data redundancy issues), and more traditional ergonomic issues (such as the need for space to accommodate FLORENCE, and the need for clear, legible displays that can be unambiguously read when staff are standing at the ventilator).

The implications for FLORENCE are manifold. FLORENCE must not undermine the existing hierarchical communication channels in the NICU (A). The re-design of working practices to incorporate FLORENCE, reinforced through its user interface, must ensure that expert help is called on as appropriate (A). The procedures adopted with FLORENCE should ensure the validity of the data on which the advice is based (C). For example, FLORENCE could prompt staff to manually check the data before implementing any suggested changes. FLORENCE’s audible alarm should be clearly distinguishable from existing NICU alarms (D); new procedures should be established to ensure that FLORENCE alarms receive appropriate attention (D), and FLORENCE should minimise the generation of unnecessary false alarms (B, D). FLORENCE should always provide the data and reasoning to support its suggestions (A, C, D). The CTA identified several contextual issues that could affect FLORENCE’s acceptance. These issues, which extend beyond FLORENCE’s capability to suggest changes to the ventilator settings, are being addressed in the design of the user interface and plans for FLORENCE’s subsequent deployment.

Author: Gordon Baxter

links

Timeliness theme

Timing and collaboration in the neonatal intensive care unit

papers

Using Cognitive Task Analysis to facilitate the integration of decision support systems into the neonatal intensive care unit.,
Baxter, G.D., Monk, A.F., Tan, K., Dear, P.R.F., & Newell, S.J.
Artificial Intelligence in Medicine , Volume 35 , Issue 3 , pp. 243-257 , 2005.