Quality,not just data: Implementing NELA for safer emergency laparotomy care

Abstract

Background:

The National Emergency Laparotomy Audit (NELA) is credited with improving outcomes after emergency laparotomy in England and Wales. Yet audit data do not save lives on their own; human effort to implement change does. This article examines how perioperative teams can convert NELA metrics into meaningful improvements, drawing on three implementation studies to separate genuine effect from ‘audit spotlight’ artefact.

Key insights:

First: outcome gains depend on implementation fidelity with local adaptation. Where bundle adherence was high and teams had senior champions, effective feedback, and practical enablers, mortality and length of stay improved. Where fidelity to the bundle faltered, benefits disappeared. Second: early wins plateau, so teams should expect diminishing returns and target the remaining vital few barriers (e.g. timely imaging, senior presence, postoperative critical care). Third: equity matters. Smaller hospitals lag on several NELA indicators, suggesting a need for focused support and shared learning. Finally, what matters to patients should be tracked, using patient-reported outcome measures (PROMs) and follow-up where feasible.

Conclusion:

For perioperative services, the message is practical: NELA data should be paired with near-real-time feedback, pathway steps under clinicians’ control need resources, what matters to patients needs to be measured, and local adaptation helps to sustain gains.

Introduction

Emergency laparotomy is by definition an urgent, lifesaving operation: bleeding, sepsis or bowel ischaemia must be dealt with promptly, particularly as patients are often older, frail and/or physiologically deranged. Baseline observational data from England and Wales prior to the introduction of the National Emergency Laparotomy Audit (NELA) suggested a 30-day mortality of approximately 15% (Saunders et al 2012). Since 2013, the figure has fallen below 9% (NELA Project Team 2023). The NELA is frequently credited with this advance, showcasing how a national audit can stimulate quality improvement (QI). The causal link is nuanced, hinging on implementation fidelity, not audit process alone. This article reviews NELA, looking at three implementation studies: ELPQuiC (Huddart et al 2015), Enhanced Peri-Operative Care of High-risk Patients (EPOCH) (Peden et al 2019), and the Emergency Laparotomy Collaborative (ELC) (Aggarwal et al 2019), to separate true effect from Hawthorne artefact. By approaching the NELA this way, we see that it has likely improved patient outcomes, but the causal link cannot be assumed to be robust. Implementing care bundles such as NELA recommendations is challenging, with diminishing returns and future adaptations needed.

Audit & feedback

Audit data do not save lives or improve outcomes simply by existing. NELA faces challenges in its implementation: feedback must be interpreted by clinicians who subsequently decide to change their practices, and who have the resources to do so. The 2012 Cochrane review of audit and feedback showed only a median 4.3% absolute improvement in guideline adherence, in the context of considerable heterogeneity (I² = 98.8%) (Ivers et al 2025). Improvements were contingent on low baseline performance, repeated feedback, and explicit goals with action plans provided (Ivers et al 2025). NELA combines frequent data feedback, peer benchmarking, and financial levers in the form of best practice tariffs to maximise impact.

The same review found that ‘backfire’ can occur, worsening performance (Ivers et al 2025). Factors contributing to this backfire are summarised in Table 1. These challenges should be acknowledged as improvements continue to be sought.

Table 1

Contributory factors to audit worsening outcomes, based on Ivers et al’s meta-analysis (Ivers et al 2025)

Factor	Description
Poorly-targeted, non-actionable feedback	Feedback on practices that aren’t immediately relevant or within clinicians’ control can feel irrelevant or threatening rather than motivating.
Psychological ‘backfire’ from self- focused feedback	Ranking or comparing individuals shifts attention to the ‘self’ over the specific task, distracting from concrete behaviour-change goals.
Goodhart’s Law and metric gaming (Mattson et al 2021)	‘When a measure becomes a target, it ceases to be a good measure.’(Strathern 1997: 308) Sometimes leading to worse outcomes.
Unaddressed contextual barriers and low engagement	Lack of facilitation, discussion, or clear next steps means clinicians may ignore or resist feedback, reducing uptake or provoking resistance.
Complexity of targeted behaviour	For multi-step or intricate processes, feedback alone can introduce confusion or mis-prioritisation, potentially worsening performance.

NELA: Scope, methods, and headline results

Commissioned by the Healthcare Quality Improvement Partnership, hosted by the Royal College of Anaesthetists, NELA records every emergency laparotomy performed in England and Wales (National Emergency Laparotomy Audit (NELA) – About, n.d.). Four aims are specified as follows:

• i. Enable providers to improve care delivery.

• ii. Benchmark hospitals, providing comparative information.

• iii. Provide comparative data on outcomes.

• iv. Facilitate QI and spread best practice.

NELA benchmarks against several key factors, which are summarised, with some comment on trends, in Table 2.

Table 2

NELA key indicators from annual NELA reports (NELA project team, 2015, 2023, 2024)

Indicator (definition abridged)	Baseline Year 1 (2013–2014)	Year 8 (2020–2021)	Year 9 (2021–2023)	Trend since Year 1	Year 8 to Year 9
Risk assessment documented – preop	56%	86.8%	84.6%	+ 28.6 pts	−2.2 pts
High-risk pts with consultant surgeon & anaesthetist in theatre	70%	91.3%	90.4%	+ 20.4 pts	−0.9 pts
High-risk pts admitted directly to critical care *	64%	79.1%	80.3%	+ 16.3 pts	+ 1.2 pts
Older/frail pts reviewed by CGA (comprehensive geriatric assessment) trained periop team	10%	31.8%	33.2%	+ 23.2 pts	+ 1.4 pts
Formal frailty assessment (⩾ 65 yrs)	n/a	86.5%	84.8%	—	−1.7 pts
Mortality*	11.7%	9.2%	9.3%	−2.4 pts	≈ (-0.1 pts)
Median postop length of stay	18.1 d	10 d	11 d	−7.1 d	+ 1 d

*

Subsequent to Year 1 NELA reported in-hospital mortality; in Year 1, they reported 30-day mortality, so interpretation cannot be direct.

The fall in mortality of 2.4% is to be celebrated. With 27,863 patients included, it represents an additional 641 survivors. However, causation cannot be assumed, since practice has changed in the last decade outside of NELA recommendations, such as enhanced critical care capacity (-NHS Hospital Bed Numbers: Past, Present, Future -The King’s Fund n.d.), evolving anaesthetic practice (Jessen et al 2022); (Edwards et al 2023), and enhanced recovery after surgery (Scott et al 2023). Implementation studies are required to provide firmer evidence.

In the most recent report, mortality, length of stay, and consistent consultant presence have shown a plateau or reduction (NELA Project Team 2024). There are significant challenges with continued implantation, explored in the following sections.

Strengths and limitations of the NELA data set

Mandatory participation gives NELA rare near-population-wide coverage. This allows outlier detection and adjusting for patient factors (Oliver et al 2018) such as age, ASA Score (American Society of Anaesthesiologists Physical Status Score), and P-POSSUM score (Portsmouth Physiological and Operative Severity Score for the enUmeration of Mortality and morbidity) while retaining hospital-level random effects. Problems with fidelity still exist, and it should be noted that both P-POSSUM and the NELA score (developed after the second report) can overestimate mortality risk in underrepresented groups and the most at-risk patients (Darbyshire et al 2022).

Links to Hospital Statistics and the Office for National Statistics improve completeness and accuracy of survival analysis. However, risk adjustment still relies on manually entered physiological variables, raising the theoretical possibility of data drift or gaming. The ninth report reveals 4%–6% missing data for sepsis (NELA Project Team 2024).

Governance and patient involvement

NELA benefits from a multidisciplinary steering group, lay representatives, and an open data ethos, promoting transparency and secondary analyses. Yet patients’ voices remain underrepresented in metric selection. Patient-reported outcomes (PROMs) and functional outcomes are only now entering the pilot phase with Soylu et al’s recent publication (Í Soylu et al 2025).

Real-world influence

NELA reports have shaped national policy, most noticeably a Best Practice Tariff. It is possible to see how NELA might drive NHSE Getting it Right First Time (GIRFT) visits and investment. Funding CT scanners and overnight consultant rotas are examples which would directly influence achievement of NELA recommendations. Nevertheless, with a significant number of national QI programmes and audits, there is room for improvement in how they are acted upon (Antonacci et al 2023).

Controlled implementation studies

A succinct comparison of bias risk is presented in Table 3. Critical appraisal highlights follow in the narrative below. A detailed risk of bias assessment is available in Supplemental Appendix 1.

Table 3

Risk of bias summary across three implementation studies

Domain	ELPQuiC	EPOCH	ELC
Design hierarchy	Controlled before-after	Stepped-wedge cRCT	Interrupted time series (ITS) quasi-experimental
Risk of bias tool applied	CASP Cohort	Cochrane RoB 2 (cluster)	ROBINS-I
Selection bias	Moderate; volunteer centres	Low (randomised order)	Moderate; self-selecting
Performance bias	High; unblinded QI	Moderate; cannot blind but randomisation dilutes the risk	High; unblinded
Detection bias	Low; objective mortality	Low	Low
Confounding control	Nil	Randomisation & modelling	ITS-segmented regression
Overall rating	Fair	Good	Fair-good

ELPQuiC (2011–2013)

A four-centre, controlled before-and-after study introduced a six-element Emergency Laparotomy Pathway Quality Improvement Care (ELPQuiC) bundle that closely aligns with NELA standards (Huddart et al 2015):

1. Consultant surgeon and anaesthetist review.

2. Sepsis Six.

3. CT (where indicated) before theatre.

4. Surgery within 6 h of the decision to operate.

5. Goal-directed intraoperative fluid therapy.

6. Routine postoperative admission to critical care.

The study shows moderate selection bias (as assessed using the CASP cohort checklist): enthusiastic sites volunteered, and it remains vulnerable to uncontrolled secular trends. Risk-adjusted 30-day mortality fell from 15.6% to 9.6% (relative risk 0.61), while overall bundle compliance rose from 24% to 78%. Importantly, eight neighbouring non-intervention hospitals showed no concurrent fall in mortality, supporting a genuine effect.

EPOCH (2014–2017)

The EPOCH stepped-wedge cluster RCT enrolled 93 NHS hospitals, each contributing 15 preintervention and 20 intervention months (Peden et al 2019). Assessed using the Cochrane RoB 2 (cluster) tool, the trial carries ‘some concerns’ – chiefly implementation deviations. Implementation proved difficult; 62% of patients received ⩾70% of the 37-item pathway. Qualitative work by Stephens et al (2018) traced the shortfall in bundle fidelity to competing QI initiatives, theatre bottlenecks, and data-entry burden. The 90-day mortality was unchanged (16.0% control vs. 16.9% intervention). Illustrating that fidelity, not just clinical content, may be a critical driver of outcome change.

ELC (2015–2018)

The ELC involved 28 hospitals, implementing a six-element bundle alongside structured learning sessions, monthly run charts and peer-to-peer site visits (Aggarwal et al 2019). Risk-adjusted monthly data were examined with an approach endorsed by the Cochrane EPOC group. Application of ROBINS-I (Risk Of Bias In Non-Randomized Studies–of Interventions) tool rates the study at moderate risk of bias, chiefly due to residual confounding from the lack of a concurrent control, although modelling of the preintervention trend mitigates this. No downwards trend was evident before roll-out; bundle launch coincided with an immediate 0.9% fall in risk-adjusted mortality, followed by a sustained −0.05% monthly decline. Bundle compliance exceeded 80%, and length of stay shortened by 1.3 days. Follow-up over 2 years (Aggarwal et al 2019) confirmed that these gains were largely maintained.

Addressing the Hawthorne effect

All audits and studies are susceptible to the Hawthorne effect, whereby behaviour changes when people know they are under observation, and brief improvements may be exhibited for a period after the observation starts (Adair 1984). Study designs modelling preintervention trends (interrupted timeseries) or including contemporaneous controls (cluster randomised trials) work to mitigate this (Hulley 2013).

The pattern of improvement with NELA recommendations is inconsistent with a significant Hawthorne effect:

• ELPQuiC: Mortality fell in the four intervention hospitals yet remained flat in eight neighbouring trusts that were exposed to the same audit spotlight. A Hawthorne response should have lifted outcomes in both groups.

• EPOCH: All 93 sites were under identical audit scrutiny throughout the trial, preintervention and postintervention. Where fidelity proved low, the mortality signal disappeared. If being observed were enough, an across-the-board improvement would have surfaced regardless of fidelity.

• ELC: The immediate 0.9-percentage-point mortality drop persisted for at least 24 months and was still evident in a 2-year follow-up, after central coaching ceased.

Taken together, three key factors are drawn out: being observed, by providence of being under audit, is insufficient; controls and low-fidelity phases did not improve outcomes; and sustained gains require high bundle adherence. Importantly, improvements endure beyond the initial implementation period.

The decade-long NELA trend mirrors this: early step-changes have been followed by steady gains and then plateau (NELA Project Team, 2015, 2023, 2024). The largest yield comes from early efforts. Diminishing returns after that are to be expected. This principle, which is often traced back to Pareto, (1897), applies so universally to both individual and organisational improvement that it should not be surprising to see signs of it in the NELA data (Alkiayat 2021). However, Pareto has been turned into effective tools in other surgical fields. Using live Pareto charts to openly track incidents has been seen to improve outcome in other surgical fields, such as neurosurgery (Shi et al 2023).

Overview

Effect size

Meta-analysis shows an odds ratio of 0.76, with an absolute reduction of 1%–2%, when baseline mortality is 10% (Trangbæk et al 2023). The EPOCH result tempers enthusiasm, implying that achieving high bundle compliance is critical.

Implementation lessons

Success factors repeatedly cited include senior clinical champions, protected QI time, near real-time data dashboards and collaborative learning. Barriers are theatre access, human factors, and data entry (Stephens et al 2018). It has been seen that following a successful implementation the changes can be sustained out to 2 years (Aggarwal et al 2019).

Generalisability and equity

Across consecutive NELA reports, smaller district general hospitals (DGHs) continue to lag behind major teaching centres: only 52% of DGH patients receive a consultant-reported CT scan before surgery, and just 76% benefit from both consultant surgeon and anaesthetist presence, compared with 71% and 88% respectively in large teaching hospitals (NELA Project Team, 2018, 2020).

Challenges and future priorities

1. Adherence versus flexibility: mandating a rigid bundle may stifle local innovation; process-evaluation work from the EPOCH trial shows that successful hospitals adapted the bundle locally (Stephens et al 2018). Improvements will be context dependent, and standardisation needs contextual adaptation.

2. Functional recovery and PROMs after emergency laparotomy (Saunders et al 2021) should become headline metrics. They are reportedly feasible and well received by patients.(McLean et al 2020)

3. Frailty: only 38% of patients aged ⩾75 have documented clinical frailty scores, despite frailty predicting discharge destination more strongly than age (Parmar et al 2021).

4. Real-time analytics: live data dashboards could shift audit from yearly retrospectives to continuous improvement cycles (Leung et al 2021).

5. Expansion to a non-surgical pathway: approximately a third of patients considered for laparotomy are managed nonoperatively (McIlveen et al 2020). Auditing decision quality, conservative, and palliative pathways is logical.

Conclusion

NELA will improve outcomes only where perioperative teams translate its measures into reliable processes. The message is practical: pair audit data with real-time feedback, target the vital few steps on your pathway, and adapt the bundle to local constraints. Resource the pathway elements clinicians’ control, monitor progress with run charts, and include patient-centred outcomes. Use dashboards that keep everyone close to the clinical work and widen success measures beyond mortality and length of stay. This creates an actionable route from national audit to safer care at theatre doors, in recovery, and on the ward.

Supplemental Material

sj-docx-1-ppj-10.1177_17504589251393188 – Supplemental material for Quality, not just data: Implementing NELA for safer emergency laparotomy care

Supplemental material, sj-docx-1-ppj-10.1177_17504589251393188 for Quality, not just data: Implementing NELA for safer emergency laparotomy care by Alexander Gaspar in Journal of Perioperative Practice