Discussion
Error
Arguably the most important benefit associated with patient records is the positive impact on patient safety, and this is what drives the majority of programmes in healthcare in the UK. Error during calculation of volumes, rates and values is high on paper and this is known from both the prescribing literature13–15 and from previous studies in intensive care settings.7 We demonstrated a high error rate in our own paper charts, there is heterogeneity in the error rates and in order to establish the precise rate a larger study would be needed, however, even by the most conservative estimates, the error rate remains high. Although most of these are trivial if, for example, 1 in every 10 000 errors was significant, the technology would stop 1 significant error every 3 weeks. Humans are not routinely good at obsessive checking but computers are.
EHR does not completely remove error, as calculations within it are dependent on correct data entry and well-built rules, but by introducing pragmatic limits, users can be warned when unlikely values are entered. Calculations can also evoke algorithms which detect patterns of deteriorating physiology deployed in early warning scores.16–18
Time
Data entry is sometimes said to be slower in electronic environments but this was not substantiated here. For some values, we used automatic download of data, as with arterial blood gases, so that these no longer needed to be manually entered. Calculation of volumes and values is laborious in ICU, and transformed by digitalisation. There was no need in the electronic chart, to record drug infusion rates as these were automatically recorded on the chart. Nursing time saved equated to a several whole nursing shifts each day. Significant time was also saved for ward clerks and for staff investigating patient complaints. It is interesting in user feedback that staff report data entry is not quicker. Users quickly forget the inconvenience of preparing charts and searching for notes, and do not take these into account when considering data entry into digital systems.
The time-savings measured does not include the additional benefits; research studies are screened for more quickly, notes are accessible for audit, quality improvement, complaints and incident reporting. The labour of sorting through notes and addressing content is alleviated to some extent by the legibility and ability to electronically search, further improving ability to answer complaints or participate in research. The translation of time saved into economic benefit is not direct since each patient requires standard staffing ratios. However, this study does give some normalised of in kind benefit with respect to the ability of nursing staff to deliver direct care to the patient.
Additional benefits
In addition to demonstrating considerable time-savings and improvement in patient safety, there are other benefits that are not easily quantifiable but are obvious from user expectations and feedback. The charts are remotely available and this allows timely review of a patient’s status without the need to physically arrive at the ICU. For example, a surgeon in the operating theatre who has limited time between surgical cases can remotely check the status of a patient and even prescribe drugs remotely if necessary. In these situations, it is clearly important that patient review is not compromised, but there are many clinical situations where new microbiology results or biochemistry results necessitate changes in current treatment regimens and rapid treatment change is beneficial to the patient. Remote viewing of charts is also useful prospectively, for example, nurses awaiting transfer of patients from the ICU to the ward can view the clinical state of the patient before they arrive.
Many clinicians can simultaneously access the same patient’s electronic record and interact with it. Thus, on a busy ward round blood pressure can be viewed on one screen, while on a tablet another doctor adjusts a medication dose and a nurse documents notes on a third device. All members of the multidisciplinary team can simultaneously view the record during discussion.
Another unquantifiable advantage of an electronic chart is the ability to easily view historic charts. In the electronic environment, these are always available and have not been filed or removed from the patient record. This advantage continues after the time the patient is transferred out of the intensive care ward, where often, historic events within the ICU stay are opaque to the clinical teams subsequently caring for the patient. Even in current EHRs there is often discontinuity between ward care (where observations may be documented electronically) and ICU or the operating theatre (where often they are not electronically recorded or are recorded into separate systems).
A simple but powerful advantage of electronic charts is the legibility of the chart, with clear accountability of what is, and is not, documented. The chart does not get lost or misfiled.
Audit and accountability are improved. No-one can add retrospective notes and imply they were written contemporaneously. Contemporaneous notes can be added remotely (eg, by a remote consultant discussing a medical plan with a junior). It is clear from the record who has documented in the record, exactly when, and what that user’s specialty and designation is.
Research
All data entered into the EHR within our organisation is archived for audit for improvement in clinical care and for retrospective clinical investigation, and can be used, with the appropriate ethics and information governance in audit and research. Research teams were asked about their opinion on whether the electronic record had made any difference to workloads or data gathering and report changes in practice, efficiencies and disadvantages after changing to an electronic environment.
Research studies required a screening process and this had previously required a visit to view patients’ paper charts. After launch of electronic charts, screening could be performed remotely. This meant that researchers were focused on patients who fulfilled study criteria, thus reducing the time spend in ICU by two research nurses from 20–40 min to 10–20 min per patient per day. The research nurses reported not feeling as if they were in the way, but also reported a feeling of less presence within the ICU, as they no longer needed to visit
Reporting and artificial intelligence
Across the rest of the organisation, we are realising the benefits of being able to use data to inform practice. Thus alerts can be sent to appropriate clinical teams when patients with particular diagnoses are admitted, patients with high glucose are reported to the diabetes nurse specialists and patients with abnormal early warning scores are alerted to the outreach teams. These benefits are now used in ICU, with our informatics department being asked to set up regular reports, and stream alerts to mobile devices in a similar fashion to the rest of the organisation.
This will undoubtedly drive behavioural change as we have done elsewhere,19 so that reaching full compliance with set targets of care is possible with constant use of data to drive good practice.
Staff attitudes
The project was introduced into an institution where EHR was already ubiquitously used and indeed electronic prescribing and medicines administration was already in use in the ICU. In general, we have found that introduction of more detailed elements of the EHR was not only welcomed but was expected and driven by users, with demand for the software predating the roll out. This is clearly a different situation from de novo EHR roll out and is consequently met with less resistance. Initial staff surveys during roll out demonstrated that staff were generally receptive to the idea of EHR and thought that it would reduce calculation errors. Nurses were concerned that it would not improve patient care overall or patient safety, this view was not shared by doctors or AHPs and has been observed previously in EHR adoption20. The reasons for nurses being less convinced about EHR improving patient care and safety are likely to be complex, and include the confidence that nurses themselves feel interacting with digital systems.
Staff also respond differently over time as they become used to the new technology. In our institution, there is very high expectation that EHR will solve problems as this has been the experience of users in previous projects. Therefore, although error rate is demonstratively reduced by the introduction of electronic charts, some staff answered ‘I don’t know’ to this question, this may be a reflection of doctors who look for published evidence before answering ‘yes’ to questions of this type, where they assume the question is about EHR worldwide rather than the doctor’s own local experience.
Surveys of the established system were overwhelmingly supportive of the software despite the fact that nearly half of the staff had never worked in ICU, and therefore, were using the charts for the first time. The nature of the COVID-19 crisis meant that support, training and supervision of staff members was reduced, attesting that the software supported processes and patient care even in extenuating circumstances. Importantly in the question ‘Do you think electronic charts are a good idea, no doctor or nurse answered ‘no’ to this question.
Harm
We believe that EHR, if used incorrectly, can cause harm. We are cautious, for example, about the introduction of automatic download of patient data as it is likely that manual entry allows for consideration of parameters. On the other hand, we cannot know if all of the manually entered data is correct (indeed this would seem unlikely). We are well aware of the dangers of ‘alert fatigue’ and have published on this previously.21 22 Alerts within the system must be well designed, infrequent and relevant, to have an impact. Finally, we are aware that operating within EHR in a tertiary referral system is unhelpful if prescriptive. Complex medicine means that often practitioners are working outside ‘standard care’ and EHR which is restrictive or dictatorial is unhelpful and likely to be boycotted. We aim to provide safe limits and warn if parameters in prescribing, observations or values are outside these limits, but to avoid a railroad on which we require practitioners to remain. CDS is a powerful tool and used wisely can considerably reduce patient risk; multifaceted EHR provides enriched data all in one place to enable complex CDS rules. Already in the EHR we have built hundreds of complex rules which by combining the diagnostic lists, medications and results, can warn users of potential interactions or changes in the patient’s condition requiring reconsideration of current pathways of care. Finally, we intend to use automatic download of data from monitors, ventilators and pumps and will use these in complex algorithms, we have not yet explored this, or whether in itself this introduces error.