5 minute read
Review of Clinical Decision Support Tools for use in Electronic Patient
Records at Manchester University Foundation Trust.
At our institution, a new EPR system was introduced in September 2022 (Epic Healthcare Systems Limited) The software includes a highly configurable, clinical decision support tools(CDSTs) called “Best Practice Advisories” (BPAs) In the first 7 months, we observed that the BPAs configured in our system had generated > 25 million alerts to our user base of 28,000 staff
Advertisement
CDSTs have the intended purpose of improving healthcare delivery (1) It can reduce patient safety incidents, reduce errors, increase adherence to guidelines, streamline administrative processes and more However, there are also possible drawbacks, and alert fatigue is a commonly reported phenomenon (2,3)
This occurs when too many in-significant alerts are presented, and users may start to dismiss alerts regardless of importance One solution to combat this is to prioritise alerts of critical importance and tailor alerts to avoid overwhelming users with clinically insignificant alerts
We report on a QIP designed to evaluate and refine our configured BPAs to promote accurate, appropriate, and practical clinical decision support whilst minimising alarm fatigue
Since the introduction of the EPR system in our trust in September 2022, there have been more than 25 million BPA alerts Of these, only a minor proportion (7 4%) have been actioned, with the majority (92 6%) being dismissed (Figure
2) Given the sheer volume of alerts, the trend observed could be perpetuated by alarm fatigue, which can cause some significant alerts to be inappropriately dismissed
On reviewing individual BPAs, it is clear there is scope to refine these CDSTs The common themes highlighted, corresponding to the ‘CDS five rights’, are summarised in Figure 3 Some BPAs were incorrectly configured meaning that despite an acknowledgement option being selected, this registered as ‘no action taken’ Discrepancies such as this will need to be rectified before re-auditing to improve data quality and aid in future reviews As outlined in Figure 1, these suggestions will be proposed to the relevant governance boards (e g acute care/ pharmacy/ nursing boards) to gather further opinions before working with buildanalysts to make changes on the live platform
We adopted a similar methodology employed at Jurong Health campus, Singapore (existing EPIC users) (4) Data was collected on each BPA, focussing on the following parameters:
• Number of BPA alerts (% of total BPA alerts)
• Triggers for BPA alert
• End-users alerted
• Departments alerted
• Acknowledgement options
• % age Appropriate actions taken
The collected data allowed an MDT approach to evaluate each BPA against the Healthcare Information and Management Systems Society’s (HIMSS) “CDS five rights” framework: (5)
The right information
To the right people
In the right intervention formats
Through the right channels
• At the right points in workflow
Recommendations were presented at respective hospital governance boards, before being trialled in the live system with the help of the EPR programming team (Figure 1)
1) Sutton RT, Pincock D, Baumgart DC, Sadowski DC, Fedorak RN, Kroeker KI. An overview of clinical decision support systems: benefits, risks, and strategies for success. NPJ Digit Med. 2020 Feb 6;3:17. doi: 10.1038/s41746-020-0221-y.
PMID: 32047862; PMCID: PMC7005290.
2) Sirajuddin AM, Osheroff JA, Sittig DF, Chuo J, Velasco F, Collins DA. Implementation pearls from a new guidebook on improving medication use and outcomes with clinical decision support. Effective CDS is essential for addressing healthcare performance improvement imperatives. J Healthc Inf Manag. 2009 Fall;23(4):38-45. PMID: 19894486; PMCID: PMC3316472.
3) Scheepers-Hoeks AM, Grouls RJ, Neef C, Korsten HH. Strategy for implementation and first results of advanced clinical decision support in hospital pharmacy practice. Stud Health Technol Inform. 2009;148:142-8.
PMID: 19745244.
4) Taking a thoughtful approach to clinical decision support: The right information to the right person at the right time [Internet]. EpicShare. [cited 2023Apr13]. Available from: https://www.epicshare.org /share-andlearn/juronghealth-bpa-overhaul
5) CDS/Pi Collaborative: Getting better faster-together(sm) - CDS 5 rights [Internet]. [cited 2023Apr13]. Available from: https://sites.google.com/site/cdsforpiimperativespublic/cds
6) Khan S, Richardson S, Liu A, Mechery V, McCullagh L, Schachter A, et al. Improving provider adoption with Adaptive Clinical Decision Support
Surveillance: An observational study. JMIR Human Factors. 2019;6(1).
7) (Kawamoto K, Houlihan CA, Balas EA, Lobach DF. Improving clinical practice using clinical decision support systems: A systematic review of trials to identify features critical to success. BMJ. 2005;330(7494):765)
The process of reviewing a vast amount of BPAs required a standardised and reproducible approach, and thus a few core principles were identified This led to the development of a ‘Build Guide’, which was created to promote a rigorous approach to evaluating and creating BPAs
This includes several checkpoints, focussing on 5 main areas; user groups targeted, circumstances of triggering (e.g. location of encounter), build characteristics (e.g. jargon-free, acknowledgement options points in workflow), trigger, and inclusion/exclusion criteria (e.g. departments, patient age, deceased patients). This can be used as a checklist to standardise the process of building BPAs in future.
CDS five rights (5) Common themes
The right information
Some disparities in the wording and response options. Sometimes unclear in how responses would be interpreted (e.g. action vs no action taken).
To the right people Sometimes targeting inappropriate staff (prescribers vs non-prescribers, nurses vs clinicians).
In the right intervention formats
Through the right channels
At the right points in workflow
Acknowledgement options need refining- some BPAs are configured to register as ‘no action taken’ regardless of option selected (difficult to audit data unless resolved). Could include activity links to relevant pages e.g. sepsis screen order set/ safeguarding referral links on relevant BPAs.
Different types of BPAs- silent vs pop up vs banner BPAs. Some may be more appropriate than others depending on context.
Some BPAs alert at inappropriate times in workflow; e.g. when opening chart/ administering medicines / recording observations etc
Some lockout options need refining to prevent repeated burden of BPAs if already acknowledged.
The guiding principles for introducing BPAs into the trust were that they should provide harm free care, increase quality of care, and it should meet MFTs current standards of secondary care These principles defined the scope when reviewing BPAs
One of the key challenges to implementing CDSTs is the sheer volume of alerts which contribute to alarm fatigue To mitigate this, we propose that organisations limit the number of BPAs and undertake regular reviews to ensure their clinical relevance (1,6)
It is also important to note that individual trusts can vary; including in patient demographics, organisation culture, and local practices Therefore, evaluating CDSTs to ensure they reflect this is critical when introducing EPRs (1) Continuous evaluation of CDSTs identifies discrepancies which can be amended For example, we discovered BPAs that were alerting even in deceased patients and some which were configured to not recognise an ‘action’ regardless of what the user clicks Secondly, this iterative process of re-evaluation will provide feedback to build-analysts on how BPAs are impacting end-users and clinical outcomes (7)
The process of re-auditing after making changes is highlighted above in Figure 1
Throughout our process, we identified some key aspects to building effective BPAs This has fed into the creation of a ‘Build Guide’ that contains common concepts discussed across stakeholders such as clinicians and build-analysts for building and validating BPAs This guide can be used as a checklist to refine and standardise the process of building BPAs in the future
Moreover, clinicians being directly involved can increase engagement and foster a sense of ownership and trust in CDSTs, which can improve integration of changes (6)
Having input from a large team can come with challenges To overcome these, it is important to have a MDT approach of healthcare professionals and buildanalysts This combines insights to include expertise in multiple fields, including clinical decision making and information technology
With these principles in mind, the team can develop BPAs that are evidence-based, user-friendly and tailored to the needs of particular individuals/ trusts To complement this, a ‘Build Guide’ can provide a more standardised and rigorous approach when considering evaluating and building BPAs
C O N C L U S I O N
Our process illustrates that CDSTs need careful, iterative scrutiny by clinical staff working with analytics teams Such processes can lead to significant improvements in the clinical decision support burden experienced by end users
Clinical informatics teams should proactively plan for such QIPs as part of new EPR implementation programmes Furthermore, utilising a standardised checklist such as a ’Build Guide’ can help to validate CDSTs