Effect of an electronic medical record nudge to improve quality improvement program tracking of neuraxial catheter replacements in obstetric patients

Introduction

In order to track key obstetric anaesthesiology-related complications and/or events at Lucile Packard Children’s Hospital, Stanford, we developed quality improvement (QI) metrics and integrated their collection into our electronic medical record (EMR) (figure 1A). Before an anaesthesiologist can close a patient encounter, they must complete the QI database charting for the case by documenting any relevant complications and/or events or select ‘None’ if no events occurred (figure 1A). The accuracy of the QI data in this system relies on the anaesthesiologist to correctly enter the data at the time of event or recall any complications that occurred during the case prior to closing the record. Because of the demands of a busy obstetric service and the potential for several handovers to occur, the anaesthesiologist closing the chart may not be aware or may not remember to chart pertinent data. Literature regarding the effectiveness of protocols that rely on human recall and motivation to document complications in this manner conclude that these means are often inadequate, variable and not sustainable.1

Request permission

Behavioural economic concepts such as ‘choice architecture’ and ‘nudge theory’ are used across industries to better facilitate decision-making and reduce human error and misjudgment.2 3 The incorporation of nudge theory and choice architecture into EMRs to influence decision-making by clinicians has a growing body of support and success in delivering quality, evidence-based care.4 5 To improve the documentation of the replaced neuraxial block metric in our QI database, we implemented an electronic nudge that would display a reminder when two or more neuraxial block procedure notes were detected within a single patient’s encounter (figure 1B). This nudge would alert the anaesthesiologist that a block replacement may have occurred and suggest that, if true, this be logged in the QI database.

The primary aim of this study was to determine if this system improved documentation of neuraxial block replacements in the obstetric setting. We hypothesised that the introduction of an EMR nudge would improve the documentation of replaced neuraxial blocks by anaesthesiologists in the QI database at our institution.

Methods

This report conforms to the SQUIRE V.2.0 guidelines for reporting new work to improve healthcare safety.6 The project did not meet the definition of human subjects research as defined in 45 CFR 46, nor the FDA definition of a clinical investigation as defined in 21 CFR 56, and was approved for IRB review exemption.

Primary outcome

To compare adherence rates with charting replaced blocks in the QI database before and after implementation of the EMR nudge, we collected charts that had more than one neuraxial procedure note listed, and assessed whether these had been charted as replaced blocks in our QI database. Charts that were detected via the two-note detection rule but not entered in the QI database were audited to confirm whether a neuraxial block replacement had occurred. We only included those confirmed to have had a block replacement as part of our analysis (figure 2). We then calculated the incidence of adherence to replaced block charting in the QI database before and after implementation of the EMR nudge. The replaced block rate was plotted in a p-chart using QI Macros for Excel (Denver, Colorado, USA) to show adherence rate over time, which can be seen in figure 3.

Request permission

Figure 2

The flow chart of the anaesthetic records that were reviewed for the study along with the outcomes. QI, quality improvement.

Request permission

Figure 3

A p-chart displaying the proportion of patient charts with failed/replaced blocks appropriately charted in the quality improvement database before and after implementation of the nudge in the electronic medical record. CL, central line; LCL, low control limit; UCL, upper control limit.

Results

Between January 2019 and December 2020, there were a total of 494 patient encounters in which a replaced epidural block occurred, based on detection of two or more neuraxial procedure notes per encounter. Figure 2 outlines the validation performed to confirm that patients truly had a replaced block. Five patients in chart review were determined not to have had a replaced block. Prior to introduction of the nudge, the replaced block rate was 6.2% and after the introduction of the nudge, the replaced block rate was unchanged at 6.3% (0.1% difference, 95% CI: −0.0119 to 0.0099). Thus, the true block replacement rate in the study period was 6.3% (489 out of 7768 neuraxial catheters placed) over the course of the 2 years. Two hundred of these replaced neuraxial blocks occurred prior to our intervention (October 2019), and 289 were detected after our intervention. Prior to the introduction of the nudge in October 2019, the appropriately charted replaced block rate in our QI database was 80.0% (160 out of 200), and after the intervention, the rate was 96.2% (278 out of 289) (p value <0.00001, OR=6.32, 95% CI: 3.15 to 12.66). The adherence rates reported monthly are shown in the p-chart (figure 3). After intervention, there is a shift in the median with sustained improvement in the adherence rate to charting replaced blocks.

Discussion

We demonstrated that an EMR nudge improved and sustained adherence to quality metric monitoring of neuraxial block replacement in obstetric patients. This is significant, as accurately monitoring quality metrics such as the rate of failed and replaced neuraxial blocks for labour analgesia provides insight into the quality of obstetric anaesthesiology care delivered. The ability to produce such data reliably and quickly is also important, as frequent peer review by quality assurance teams may allow detection of concerning trends and allow surveillance of any changes in practice that may have been employed. This single-site, pre–post QI project demonstrates that the use of ‘nudge technology’ and ‘choice architecture’ in obstetric anaesthesiology may positively influence the behaviour of clinicians and, in turn, positively influence the effectiveness of other QI efforts.

National and international agencies continue to pursue the development of quality metrics to measure individuals and institutional care. The Obstetric Anaesthetists’ Association and the National Perinatal Epidemiology Unit used Delphi methodology to determine ‘key indicators to drive quality improvement in obstetric anaesthesia’.7 One of the five key metrics was ‘percentage of epidurals for labour analgesia that provided adequate pain relief within 45 min of placement’. The Society for Obstetric Anesthesia and Perinatology (SOAP) offers a designation titled ‘Centers of Excellence’ to hospitals offering obstetric anaesthesia care.8 The SOAP Centers of Excellence expects centres to monitor their neuraxial block failures/replacements, and suggest aiming for a failure/replacement rate of 3%–6%.9 Monitoring one’s block replacement rate is important. A labour epidural block replacement rate that is too low may indicate that an insufficient number of inadequate blocks are being appropriately replaced, while a block replacement rate that is too high may imply poor epidural insertion technique, inadequate supervision of trainees or suboptimal local anaesthetic/opioid maintenance solution or epidural pump settings. Additionally, one of the SOAP Centers of Excellence essential criteria is the presence of ‘quality assurance and patient follow-up systems’, in addition to a description of ‘systems used to track labour epidural replacement’. These international agencies recognise the importance of accurately tracking the rate of failed and replaced neuraxial blocks, further supporting the importance of QI initiatives such as ours. Our results show that institutions relying on clinicians to remember to document neuraxial block failures and replacements may be under-reporting without an EMR-specific nudge. Implementation of this or similar nudges in EMRs may be necessary to improve quality metric reporting of failed and replaced blocks and other key quality metrics.

Due to the high acuity and/or fluctuating workload on labour and delivery, even those who are motivated to change or improve compliance with best practice guidelines and protocols may fall short in their efforts. This ‘intention–action gap’ is a phenomenon well described by behavioural scientists.10 Across medical specialties, methodology used to influence physician behaviour to close this gap has garnered a wide array of interest. Most of the research surrounding clinical decision-making and implementation science concludes that the promotion of change with simple audit and feedback, didactic educational tools and individually targeted interventions yield poor results with waning sustainability.11 12 Examination of behaviour patterns when analysing clinician decision-making highlights the influence of rationality, habits and implicit processes/biases that impact clinicians’ choices.1 The assumption that these cognitive processes are always sound and effective is flawed, and clinicians may require influence in their decision-making on a more granular cognitive level.13

The act of ‘nudging’ is defined as a ‘function of any attempt at influencing people’s judgement, choice or behaviour in a predictable way, that is made possible because of cognitive boundaries, biases, routines and habits in individual and social decision-making’.13 Similarly, the concept of ‘choice architecture’ involves ‘the design of different ways in which choices can be presented to individuals’.13 The origins of these theories stem from the schools of economics and behaviour science. Their applications have become very popular in present-day medicine, especially when used in combination with EMRs and information systems. Despite the accelerated advancements in our knowledge of disease and innovation in healthcare technology over the past century, the ‘evidence-to-practice’ gap (ie, the time between description of an effective and evidence-based intervention and large-scale uptake into clinical practice), is on average, 17 years.14

A recent systematic review of trials using nudge technology within the Cochrane network found that, of the outcomes impacted by a clinician nudge, 86% influenced clinician behaviour in the hypothesised direction, with 53% of these achieving statistical significance.5 Among the studies included in the Last et al systematic review analysing the efficacy of different nudge strategies employed in healthcare, the most used nudge strategy was ‘framing’, and the most effective nudge strategies were those that employed ‘default’ selections.15 The authors describe a clever graphic depicting a ‘nudge ladder’, with the least desirable/effective nudge strategies at the bottom (those that simply provide information), and the most desirable/effective at the top (those that guide choice through defaults; figure 4). The nudge strategy described in our study strongly aligns with the second highest nudge strategy on the nudge ladder, those that ‘enable choice’. Although several aspects of EMR nudges commonly use forcing functions and default selection to achieve clinician compliance,16 the unique benefit of the use of a nudge such as ours is the maintenance of physician autonomy, space for nuance and freedom from a forcing function to make alternative choices. For example, despite using an electronic tool to highlight all patient records that contained two or more neuraxial procedure notes, our audit revealed that some of these charts did not meet our clinical criteria for block failure (figure 2). Using the nudge allowed us to maintain clinician oversight in keeping these records sound and accurate. Additionally, the reminder to consider charting a replaced block was designed to be unobtrusive and not directly interfere with routine workflow.

Request permission

Although the data we collect on failed and replaced neuraxial blocks are for the purpose of review and analysis for our own division, a simple intervention like this has potential to be implemented on a much larger scale and to make significant impact on patient care and cost-savings strategies for hospitals and healthcare systems. Nudge theory and choice architecture have been successfully used in the promotion of smoking cessation in pregnancy17 and pre-eclampsia prevention in obstetric patients.18 Particular to anaesthesiology, similar principles have been used to encourage adherence with guidelines for lung protective ventilation strategies during surgery,19 safer opioid prescribing practices,20 sedation minimisation and liberation from mechanical ventilation in the ICU21 and antibiotic stewardship.22 Obstetric anaesthesiology is a unique subspecialty in that a provider may be caring for several patients at once in different stages of their pregnancies and deliveries. Accurate record keeping may be increasingly challenging given the potential acuity of a labour and delivery unit. Reducing the cognitive load of the clinician and absolving them of the need to ‘remember’ what to chart is one of the greatest theorised benefits in using a real-time, EMR nudge such as ours.23 However, EMR nudges and similar reminder messages may contribute to another rising phenomenon in EMR charting—‘click fatigue’ or ‘alert burden’. In addition to their association with physician burnout, fatigue from these alerts can also lead to active ignoring of nudges such as ours, especially if not clinically relevant.24 Assessing the effects of any added EMR tools and their contribution to burnout should be a balancing measure assessed alongside QI initiatives, and clinical informaticists implementing additional alerts should ensure they are only active when clinically relevant and quiescent when not.

There are several limitations to our study. We relied on retrospective data collection as this study aimed to assess routine clinical practice that would be impacted if a prospective study design were applied. We only collected data on replaced blocks and acknowledge that inadequate blocks that were not replaced would not have been detected in this analysis. We appreciate that failed or replaced neuraxial block trends over time could be impacted by staffing and institutional changes. However, we have made no significant changes in our clinical care during this study period, and our p-chart analysis shows a real and sustained impact of the nudge on replaced block detection. The success of this project highlights the potential to apply equivalent EMR nudges to the collection of other QI data, for example, a nudge to document haemorrhage on detection of blood product charting, or a nudge to document a patient’s ICU admission with detection of an ICU intake note or transfer of care in the EMR.

In conclusion, the addition of an EMR nudge improved the accuracy of replaced neuraxial block QI metric documentation in our obstetric patients. The results show that institutions attempting to gather quality metric data and relying on clinician memory to do so are likely under-reporting neuraxial block failures and replacements. If national or institutional QI metric comparisons are made, the accuracy of data collection should be a priority. The study findings support widespread implementation of this or similar nudges in EMRs to improve obstetric anaesthesiology quality metric reporting. However, this technology should be appropriate and applied sparingly to minimise click fatigue and alert burden. Further research exploring techniques to optimise QI metric monitoring should address the ideal number and key QI metrics to document, without increasing the potential for physician cognitive overload or burnout.