Intended for healthcare professionals

Practice Safety Alerts

National Patient Safety Agency: combining stories with statistics to minimise harm

BMJ 2009; 339 doi: (Published 19 November 2009) Cite this as: BMJ 2009;339:b4489
  1. Tara Lamont, head of response1,
  2. John Scarpello, deputy medical director1
  1. 1National Reporting and Learning Service, National Patient Safety Agency, London W1T 5HD
  1. Correspondence to: T Lamont tara.lamont{at}

    In a busy endoscopy clinic, you draw up part of a high strength midazolam ampoule and realise that you have oversedated the patient. The patient has been harmed. What stops you reporting it? What would prevent it from happening again in your trust or elsewhere?

    An article about avoiding midazolam overdose is the first in a series of BMJ summaries of recommendations to improve patients’ safety, based on reports of safety concerns, incident analysis, and other evidence.1 The midazolam article is from the National Patient Safety Agency (NPSA) in England and Wales, which each week receives reports from healthcare professionals of serious incidents that have the potential for serious harm.2 This article is a summary of the agency’s rapid response report (RRR) on midazolam overdose. RRRs are one-page notices with clear recommendations for actions that will reduce risks to patients, issued to all relevant NHS organisations in England and Wales, from hospitals to primary care organisations (

    Previous national safety interventions took time to develop3 and proved challenging to implement.4 And so over the past two years, the NPSA has developed RRRs as more timely alerts to staff about new or under-recognised risks. The trigger for an RRR is reported deaths or serious events that may be seen as “one-off” events by local organisations but which at a national level indicate to the agency that system weaknesses exist for which fixes are available that could benefit all trusts. RRRs are issued in England through the Department of Health’s central alerting system (, and organisations are required to act on these within a given time (usually six months) and report compliance, which is monitored by regulators. In the case of midazolam overdose, a safer low dose product had become available but was not widely used. Since the agency’s RRR was issued, there has been an encouraging twofold to threefold increase in use of safer doses.1 2

    Some risks highlighted in RRRs have been well publicised—for example, the RRR on strengthening systems for dealing with major haemorrhage (including access to blood and vital equipment) was triggered by the death of a patient after routine laparoscopic cholecystectomy in a stand-alone surgical unit, as featured in a recent BBC Panorama programme.5 In England 95% of trusts now report compliance with actions in this RRR through the central alerting system. But although trusts may have acted to improve systems, individual doctors and nurses may not be aware of RRRs, the evidence of harm from reported incidents, and what can be done by individuals to make practice safer. The new BMJ series is aimed at clinicians, highlighting key messages on safer practice.

    Box 1 outlines the process for developing rapid response reports and the criteria for prioritising.

    Box 1 Process for developing rapid response reports (RRRs)6

    • 300 deaths and serious incident reports are reviewed individually each week by clinical staff (these are incidents reported locally by doctors, nurses, and other healthcare professionals) and uploaded to the national database

    • After initial screening, the NPSA gathers evidence on selected topics from its wider database (including data on “low or no harm” events and “near miss” incidents), research literature, litigation and coroners’ data, and on local investigations by trusts in which serious incidents occurred

    • Issues are prioritised using explicit criteria (substantive harm or potential for harm; risk not well known; clear actions to prevent) at a weekly multidisciplinary meeting with clinical and other input, including expertise in human factors7 and design

    • Draft RRRs are drawn up with input from clinicians to identify key actions to reduce risks; the drafts are then issued for comment to external stakeholders and shared with a few hospitals to test the feasibility and impact of recommended actions

    • About 10 RRRs are issued each year as one-page notices to the health service through a central system, with required actions for organisations within a deadline

    The process for generating RRRs6 is driven by patient safety incidents reported by healthcare staff. This is a voluntary, national reporting system set up in 2003 for the NHS in England and Wales—one of the largest and most comprehensive reporting systems in the world; it has recorded over 3.9 million incidents to date, with encouraging increases in reporting year on year.8 Staff report incidents through their hospital or primary care organisation so that local action can be taken when needed (for example, supplying missing resuscitation equipment). These incidents are automatically uploaded to the NPSA for national learning—over 99% of the agency’s incident data come from that route. Healthcare staff, patients, and other members of the public can also report incidents independently through the NPSA website (

    Each incident reported as leading to death or serious harm is reviewed individually by trained clinical staff (box 1). Important information is often found in the free text—for example, “[Patient] attended ED [emergency department] with haematemesis. Bleeding +++++. Unable to contact any physicians at any site or at home (night-time) to carry out emergency endoscopy. [Patient] bled and died in [the department].”

    Although the reporting system is voluntary, with inherent bias and known under-reporting,9 it at least gives an indication of the scale of the problem, with valuable learning from all harm categories (including important “near miss” events in “no harm” incidents).

    To date, we have issued 22 RRRs (box 2). Some are developed very quickly—the RRR on risks to haemodialysis from water supplies contaminated by hydrogen peroxide was issued two weeks after receiving the trigger incident—but most are produced over two to four months.

    Box 2 Some risk areas highlighted by rapid response reports (RRRs)*

    • Confusion between lipid and non-lipid amphotericin

    • Infusions and sampling from arterial lines

    • Neurosurgery (burr holes) done on the wrong side

    • Insertion of chest drains

    • Resuscitation in mental health settings

    • Insertion of suprapubic catheters

    • Oral products used for bowel cleansing

    RRRs focus on fixing systems to reduce harm but also contain reminders for individual clinicians on safer practice. The reports usually reinforce evidence based clinical guidelines developed by others, especially when the agency’s data indicate that these are not being followed everywhere. After 44 reported incidents in which patients with glaucoma experienced some loss of vision after delayed follow-up appointments, a RRR reinforced recent NICE guidelines for individual clinicians on the need to adhere to optimal follow-up intervals, as well as identifying system fixes for hospitals to tag patients on their appointment systems.10

    Early feedback suggests that RRRs have been well received, although more detailed feedback studies are under way. One medical director noted that “the one-page format is good . . . and helpful in alerting staff to safety issues” (unpublished survey of senior NHS managers by the University of York, commissioned by the NPSA, 2008).

    Organisations have to report that they have complied with action requests, and these compliance data are monitored by regulators. But as well as using the self reporting data from trusts, the agency looks for evidence of changed practice—for example, by monitoring purchasing trends on low strength midazolam products as recommended in the RRR.2

    The NPSA national database of patient safety incidents helps to identify risks and weaknesses in the underlying system that may not be apparent locally. But getting to the key issues in such a large database is not easy. The agency’s process tries to combine the “power of stories”11 in the individual incident with evidence from the wider database and elsewhere. Numbers alone will not change clinical practice (“analysing a small number of incidents thoroughly is probably more valuable than a cursory overview of hundreds of incidents”12). Without numbers, stories are just anecdotes, but without stories, numbers are just dry statistics.

    Behind every RRR is a trigger incident that started the trail. Our thanks to the first busy gastroenterologist who took the time to report an incident of midazolam overdose and helped to prevent other patients being harmed.


    Cite this as: BMJ 2009;339:b4489


    • doi:10.1136/bmj.b4459
    • Contributors: TL wrote the first draft, which was reviewed and amended by JS. TL is the guarantor.

    • Funding: No special funding.

    • Competing interests: None declared.

    • Provenance and peer review: Commissioned; not externally peer reviewed.