Does Point-of-Care Ultrasonography Improve Clinical Outcomes in Emergency Department Patients With Undifferentiated Hypotension? An International Randomized Controlled Trial From the SHoC-ED Investigators

Atkinson RP. Ann Em Med (Ahead of print). DOI:

Clinical Question

  • In selected non-trauma patients presenting to the Emergency Department (ED) with undifferentiated hypotension, does a standardised point-of-care ultrasound protocol, compared with usual care, improve 30 day or hospital survival?


  • Patients with undifferentiated shock are a diagnostic challenge in the emergency department, with an increased mortality
  • Point-of-care ultrasound (POCUS) is currently being used to diagnose numerous pathologies in the emergency department
  • Single centre studies and case series incorporating POCUS into the management of shock have shown an improvement in both diagnostic accuracy and mortality, with ultrasound directing changes in management in up to 50% of patients (such as this trial by Volpicelli)
  • There have been no multicentre prospective randomised control trials of POCUS in shock in the emergency department


  • Multicentre randomised control trial
  • Non-blinded
  • All centres had accredited EM physicians and active POCUS programmes
  • Scans performed either by physicians who had demonstrated training and competence in POCUS, or by residents supervised by those physicians
  • Competencies required for scanning agreed by investigators before initiation of study
  • Physicians from North America visited the South African sites to ensure comparable competency standards and vice versa
  • Patient selection was a convenience sample of eligible patients when appropriately trained staff were available in ED
  • Randomised convenience-sampling blocks performed by site, using computer-generated randomisation into 100 sealed envelopes- 50 POCUS, 50 control
  • Sealed envelopes ensured allocation concealment
  • Randomisation secured by having pre-randomised sequentially numbered envelopes, matched for size and weight
  • To provide 80% power to detect a 10% difference in primary outcome at alpha= 0.05, sample size of 400 required
    • assuming baseline mortality of 30% in group


  • 6 Centres
    • 3 in North America (all tertiary referral/treatment centres)
    • 3 in South Africa (1 tertiary centre, one large regional hospital, one large district hospital)
  • September 2012-December 2016


  • Inclusion:
    • Age 19 or over
    • Sustained systolic BP <100mmHg OR
    • Shock index (systolic BP/HR) of >1.0 (with systolic BP <120)
  • Exclusion:
    • Pregnancy
    • Necessity of CPR before screening or enrolment
    • History of significant trauma in last 24 hours
    • 12 lead ECG diagnostic of myocardial infarction
    • Clear evident mechanism for shock identifiable by clinician
    • Previously known diagnosis from another hospital
    • Vagal episode
    • Non-pathological low blood pressure
  • 347 patients assessed, 273 patients enrolled (138 POCUS, 135 Control)
    • terminated at interim analysis due to slow recruitment and futility
  • 3 patients lost to follow up
  • Similar baseline characteristics, with a slight male preponderance in the POCUS group (73% vs 65%)
  • There were significant differences in final diagnoses in the North American and South African cohorts:
    • Much more sepsis in South African than in North American group (71% vs 42%)
    • More severe dehydration in North America vs South Africa( 20% vs 3%)


  • POCUS protocol, combining core elements of Abdominal and Cardiothoracic Evaluation with Sonography in Shock (ACES) and Rapid Ultrasound in Shock and Hypotension (RUSH) protocols
  • Performed within 60 minutes of patient visit, after normal physician clinical assessment
  • Documented on standardised form
  • Views:
    • Cardiac views: PLAX, PSAX, Apical and Subcostal to assess LV/RV function, size and presence of pericardial effusion
    • Lung views: Base of lung thoracic scans looking for lung sliding and pleural effusions
    • Abdominal views: RUQ, LUQ, and pelvic views for free fluid and bladder size; Abdominal aorta for AAA; IVC for size and collapsibility


  • Normal physician clinical assessment

Management common to both groups

  • Secondary clinical assessment performed by physician 60 minutes after primary assessment


  • Primary outcome: no significant difference in either 30 day or hospital discharge survival
      • POCUS: 104/136= 76.5% (95% CI 68.4%-83.3%)
      • Control: 102/134= 76.1% (95% CI 68.0- 83.1%)
        • Difference = 0.35%, 95% CI= -10.2 -11.1%
        • Analysis by site showed a higher mortality in South African sites but still no significant benefit from POCUS
  • Secondary outcomes: no significant differences seen between POCUS and control in:
    • Median volumes of fluid administered (median= 1609ml vs 1683 ml)
    • Hospital admission rates (81% vs 83%)
    • Rates of inotrope administration (12% vs 9%)
    • Rates of CT scanning (26% vs 24%)
    • ICU admission rates  (18% vs 14%)
    • ICU length of stay (median LOS 7.16 vs 5.14 days; 95% CI for difference -0.85 to 4.63 days)
    • The lack of significant differences in any outcomes persisted when the North American and South African cohorts were analysed separately

Authors’ Conclusions

  • In ED patients with undifferentiated hypotension, the addition of a point of care ultrasound protocol did not translate to additional benefit


  • Randomised
  • Good allocation concealment
  • Registered with
  • Very few patients lost to follow up
  • POCUS protocol standardised, limiting inter-observer bias
  • A pragmatic approach to an important research question – the rapid adoption of POCUS calls into question the appropriateness of its use


  • POCUS is a diagnostic test – expecting a diagnostic test to have a mortality benefit may be a cognitive fallacy; thus 30 day mortality may have been an inappropriate outcome measure
  • Convenience sampling would have led to many patients not being included in the study, and slowed down the rate of recruitment
  • Patients with a “high suspicion” of certain diagnoses, in particular ruptured AAA and ectopic pregnancy, were excluded from enrolment; this would have biased in favour of the control group
  • Study terminated early, reducing its statistical power (although interim analysis did suggest no difference)
  • Some patients had incomplete POCUS studies
  • There is no suggestion in the protocols that the POCUS findings resulted in an alteration in management. This is the whole point of the test, and would have biased outcomes in favour of the control group
  • The majority of patients were diagnosed with “occult sepsis”- the RUSH/ACES protocol do not include certain views that would have allowed increased diagnostic accuracy:
    • Cardiac – colour doppler for valvular abnormalities
    • Lung – anterior pneumothorax, interstitial syndrome, consolidation
    • Abdomen – hydronephrosis, appendix, biliary, intestinal
    • Those diagnoses would be missed in this trial and would further bias in favour of the control group
  • There was no image storage and independent review of POCUS scans- diagnoses may have been missed

The Bottom Line

  • In this trial of selected ED patients with undifferentiated shock, the use of a POCUS protocol did not affect 30 day or in hospital mortality
  • Using this limited protocol of POCUS, in a limited subset of patients, without acting on the results of the scans is not representative of my practice in this group
  • I will continue to use appropriate targeted POCUS to both diagnose and manage patients with shock, and await further trials on the efficacy and effectiveness of POCUS in the critically unwell patient

External Links


Summary author: Segun Olusanya
Summary date: 27/6/2018
Peer-review editor: Adrian Wong


  • Nice commentary on this trial by Crager and Hoffman here:

    They highlight that:

    “The study by Atkinson et al thus cannot definitively answer whether some ideal version of POCUS could be efficacious in improvingPOE outcomes in undifferentiated shock. What it does show is that the version of POCUS used in this trial was not effective. While answering questions about (real-life)effectiveness are ultimately more important, future studies that wish to answer the former question about efficacy (under ideal conditions) need to address how well POCUS measurements can approximate the relevant physiologic parameters being sought, how accurate and reliable is the data collection and interpretation, and how appropriate and predictable is the response to the data that is gathered. If such an ideal version of POCUS could be developed, it could then be possible to test whether or not its application improves POEs.”

  • Thanks for this excellent discussion. Just a few points to add.

    First – all authors are PoCUS proponents, before the study, during, and subsequently, despite these neutral (not negative) findings.

    Second – we accept the limitations outlined in the discussion and in fact we describe them all in the paper.

    Next – the PoCUS protocol was very standardized and typical of what is used internationally – in fact, it is taught on most PoCUS courses and curricula.

    Next – we excluded the list of patients you highlight, because IRB/REB and our moral compass would not allow us to randomize a young pregnant patient to NO PoCUS, or an obvious clinical AAA (a shocky 80 year old male with abdominal and back pain and pale legs) etc.

    Also – the PoCUS was completed before other tests such as CXR, labs etc

    We acknowledge may clinicans’ unwillingness to engage and enroll. Attendings were of mixed experience.
    We also acknowledge that repeat PoCUS would have been a better approach than a single exam.

    We have now analyzed more secondary outcomes, and I think we have uncovered possible reasons there was no difference:

    1. The accuracy of diagnosis in the clinical group was 94% – very hard to improve on that.
    2. The diagnostic performance of PoCUS was good, similar to other published studies, accuracy 94%, sens 75%, spec 96%, but so was the clinical assessment.
    3. The interventions including fluid did not vary between Control and PoCUS for shock subcategories

    Finally, our conclusion was “We did not FIND any benefits for survival, length of stay, rates of CT scanning, inotrope use, or fluid administration. The addition of point-of care ultrasonography protocol to standard care MAY not translate into a survival benefit in this group”

    We also state that there was NO HARM, and support continued use asking SPECIFIC questions, in fact proposing a Bayesian approach in the SHoC protocol

    Thanks again – at least we hope we have stimulated the discussion and perhaps a more robust prospective RCT will be initiated to look for the specific benefits we all believe in.

    best regards,


    Dr. Paul Atkinson MB MA FRCPC
    Professor and Research Program Director
    Emergency Medicine
    Dalhousie University
    Saint John Regional Hospital
    Saint John, NB E2L 4L2

    Chair, Department of Emergency Medicine Research Committee,
    Dalhousie University in New Brunswick

    Research Project Co-ordinator
    Integrated Family Medicine Emergency Medicine Residency Program
    Dalhousie University

    Chief Medical Officer, WorkSafeNB

    Deputy Editor, Canadian Journal of Emergency Medicine


  • Segun Olusanya

    Dear Paul and Chris,

    Thanks both for your great comments and taking this discussion further. Paul, you and the team deserve congratulations on performing this trial.

    I find it really interesting that the proponents of SHOC-ED have not changed their practice of POCUS integration after the trial. Is this a representation of our a higher understanding of medicine, or just showing our “sunk cost” cognitive bias as people trained in POCUS?

    Your comments about the protocol are also interesting. RUSH is taught as an international standard, yes. However, if it’s not answering the questions we seek- maybe we should change the protocol?

    Lastly, clinician diagnostic accuracy in the protocol was 94%- greater than Pat Croskerry’s 85% estimates. Could there have been an effect of being included in the trial, where everyone paid a lot more attention to clinical signs and symptoms purely as a result of enrolment? In which case, entering into a POCUS trial becomes good for POEs…

    I’m looking forward to being part of future trials!

    God bless you
    ICM Registrar, St Mary’s Hospital, London
    TBL Editor
    Self-confessed POCUS addict

  • I think that is the only possible conclusion to have for this trial. As you point out, trialing a diagnostic test for mortality in a heterogeneous disease grab-bag is a recipe for failure, not even mentioning that the very management has not been sorted out.

    The only questions around POCUS are the training requirements, and the optimal management of the diagnoses that are made.

    Questioning POCUS as a concept is challenging the belief that a prompt and accurate diagnosis is a worthwhile concept. It will take more than a meta-analysis to change my mind about that, except for entirely benign and self limited diseases!!!

    Philippe @ThinkingCC

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.