What clinical features or scoring system, if any, might best predict a benefit from hospital admission for patients with COVID-19?

April 20, 2020

Ms Rebekah Burrow, Dr Julian Treadwell, Ms Nia Roberts

On behalf of the Oxford COVID-19 Evidence Service Team
Centre for Evidence-Based Medicine, Nuffield Department of Primary Care Health Sciences
University of Oxford
Correspondence to julian.treadwell@phc.ox.ac.uk

---

VERDICT
No reliable, applicable, or useable clinical model or scoring system currently exists to predict outcomes or inform decisions regarding hospital admission for patients in the community with COVID-19.

Patient characteristics with reported associations with poor outcomes are:  increasing age, male sex, smoking and a number of co-morbidities including hypertension, diabetes, cardiovascular disease, chronic respiratory disease and cancer. These are reported in one systematic review and five clinical guidelines. However, these are based on what is likely to be lower quality evidence and may not be reliable. In addition, all may not be applicable to a UK primary care population.

We did not find any evidence regarding physical signs associated with mortality or need for hospital admission.

BACKGROUND
General Practitioners (GPs) are faced with making decisions about the care of patients with suspected COVID-19 in the community. They may be doing this via telephone or video consultation, or via a face to face consultation allowing a physical examination but without access to laboratory or radiological tests.

COVID-19 is a new disease with atypical clinical features. There are pressures to avoid hospital admission (to conserve resources and avoid transmission of infection) whilst ensuring prompt treatment of the severely ill in the context of significant mortality risk.

We sought evidence to answer the following question:

In people in the community with suspected or confirmed COVID-19, are there demographic features, other conditions, symptoms, or signs that are prognostic factors for 30-day all-cause mortality, ventilation, or ICU care?

SEARCH STRATEGY
We searched PubMed, TRIP, Google and medRxiv on 30^th March 2020.

We found five clinical guidelines and four systematic reviews that included either guidance for primary care clinicians on either identifying or referring patients with suspected or confirmed COVID-19, or that linked clinical features to patient outcomes in a way that could define prognostic factors.

We also found 60+ potentially relevant studies of other designs, including but not limited to: cohort, case-control, case series, modelling, editorials.

We chose to focus our analysis on the works most likely to be impactful on practice; containing the information most relevant to answer our questions, and of high quality – the clinical guidelines and systematic reviews.

CRITICAL APPRAISAL CHECKLIST
We used AMSTAR 2 to appraise two systematic reviews [1].

Systematic Reviews
Two of the four systematic reviews linked clinical features of patients with COVID-19 to patient outcomes, focusing on children and the role of thrombocytopenia respectively [2, 3]. We chose to focus on the other two systematic reviews which were wider in scope and collated more evidence [4, 5].

Wynants et al. collated 27 studies containing 31 prediction or prognostic models, of which 13 were relevant to answer our question [4].

We used AMSTAR 2 to carry out a critical appraisal of this review; our confidence in the findings of this review are high.

One study created three models to identify people at risk of hospital admission in the general population. They used Medicare claims data, proxy outcomes and patients who did not have COVID-19. These results may not be applicable to our clinical context.

Nine studies reported 10 prognostic models. Six models estimated mortality risk in patients with suspected or confirmed COVID-19, two models aimed to predict a hospital stay of more than 10 days from admission, two models aimed to predict progression to a severe or critical state. All data was from patients with COVID-19 in China. 9 models used data from inpatient populations, the 10^th population was unclear. Sample sizes ranged from 26 – 478 patients, one was unknown. The populations were generally poorly described and the applicability to our clinical context is unclear.

In addition to our concerns about applicability, the reviewers describe that the models were poorly reported, all were at a high risk of bias (likely too optimistic, likely to have a large degree of bias), some models had not been validated internally, none had been validated externally, and that the predictions of the models could be unreliable when applied in daily practice.

Another consideration is pragmatic; whether or not these studies and model describe accurate prognostic factors, can we use them? Many of them are not useable in a UK primary care context, where remote consultations are currently preferred, access to CT scanners is lacking, and time to receive blood test results may be longer than the time before which a patient deteriorates. Age and sex, however, are pragmatic predictors that could be used.

Zhao et al. conducted a wide ranging systematic review exploring predictors of disease severity and mortality in patients with COVID-19 and in addition generating comparisons with two other coronavirus infections: Middle East Respiratory Syndrome (MERS) and Severe Acute Respiratory Syndrome (SARS) [5].

We used AMSTAR 2 to carry out a critical appraisal of this review; our confidence in the findings of this review is low.

The majority of the analysis was performed on 30 retrospective cohort studies, 27 of which were from China. This generated data on patient characteristics associated with severe and non-severe illness, but does not link characteristics to outcomes. These findings have been summarised in the CEBM COVID-19 Signs and Symptoms Tracker.

The data of interest to our question applied to risk of mortality only, and was derived separately from the Chinese Centre for Disease Control, rather than the individual studies. The review authors provide little detail regarding the nature of this data but report a series of relative risks for mortality in patients with COVID-19 disease as follows:

Age>60 years RR 9.45 (95%CI 8.09-11.04), Male sex RR 1.67 (95%CI 1.47-1.89), Hypertension RR 4.48 (95%CI 3.69-5.61), Diabetes RR 4.43 (95%CI 3.49-5.61), CVD RR 6.75 (95%CI 5.4-8.43), Cancer RR 2.93 (95%CI 1.34-6.41), Respiratory Disease RR 3.43 (95%CI 2.42-4.86).

In summary, this review provides some information on patient characteristics which are associated with increased mortality but are uncertain and at high risk of bias. In addition, as they are derived from a Chinese population in unclear clinical settings, the findings may be of limited applicability to a UK, primary care setting.

Guidelines
Four guidelines highlight different but overlapping clinical features that may correlate with worse outcomes, sometimes in the context of considering which patients may require hospital-based care [6-9]. None of these guidelines discuss use of a tool or scoring system.

It is not within the scope of this review to formally appraise these guidelines, however two guidelines from producers who employ robust evidence searching and appraisal methods – UpToDate and BMJ Best Practice, report similar and overlapping patient characteristics which have some evidence that they are associated with poorer prognoses. These are: increasing age, male sex, smoking, co-morbidities including cardiovascular disease, diabetes, hypertension, chronic lung disease, chronic kidney disease, chronic liver disease, obesity, immunosuppression and cancer [7, 8].

These are derived mainly from case series and cohort studies predominantly from China, mainly from secondary care and are therefore likely to represent lower quality evidence and have limited applicability to a UK primary care setting.

There is no data presented in the guidelines to give a quantitative estimate of the increased risk of any of these characteristics on an individual’s chance of death or need for intensive care.

None of the guidelines report evidence on clinical signs and symptoms which might predict prognosis that are specific to COVID-19 illness.

A fifth guideline identified by our search from NICE does not report clinical features that may correlate with patient outcomes, but does include guidance for clinicians in primary care on the use of three specific tools: CRB65, ROTH and NEWS2 [10]. In short, CRB65 and ROTH are described as unvalidated in COVID-19 patients, with measurements that require an in-person assessment creating a risk of infection (CRB65 and ROTH) and remote use exacerbating potential inaccuracies (ROTH). NEWS2 has a recommendation that it “may be useful… however, a face-to-face consultation should not be arranged solely to calculate a NEWS2 score.” However, NEWS2 also has not been validated in COVID-19 patients, nor in primary care populations, has not been validated for remote use, and requires an in-person assessment. We undertook a separate review on NEWS scores in the context of COVID-19 primary care [11].

In summary, this rapid review suggests that whilst some demographic features and comorbid conditions are associated with poorer outcome from covid-19, these is not yet a reliable and practicable set of predictors to indicate which patients require hospital admission. GPs will have to rely on clinical judgement combined with an understanding of what evidence is available.

Two other reviews in this series by CEBM may be helpful for clinicians to support their understanding:

Rapid diagnosis of community acquired pneumonia  – drawing on evidence for the management of non-COVID community acquired pneumonia including the CRB-65 score.

In patients of COVID-19, what are the features of mild and moderate cases?

We hope that more accurate, more useful answers may be on the horizon. We found two planned living systematic reviews that could answer these questions if primary research studies of sufficiently high quality design, applicability and reporting are carried out [4][12].

Our thanks to Professor Trisha Greenhalgh for editorial input to this manuscript.

Disclaimer:  the article has not been peer-reviewed; it should not replace individual clinical judgement and the sources cited should be checked. The views expressed in this commentary represent the views of the authors and not necessarily those of the host institution, the NHS, the NIHR, or the Department of Health and Social Care. The views are not a substitute for professional medical advice.

Search Terms
((coronavirus*[Title] OR coronovirus*[Title] OR coronoravirus*[Title] OR coronaravirus*[Title] OR corono-virus*[Title] OR corona-virus*[Title] OR “Coronavirus”[Mesh] OR “Coronavirus Infections”[Mesh] OR “Wuhan coronavirus” [Supplementary Concept] OR “Severe Acute Respiratory Syndrome Coronavirus 2″[Supplementary Concept] OR COVID-19[All Fields] OR COVID-19[All Fields] OR “2019nCoV”[All Fields] OR “2019-nCoV”[All Fields] OR WN-CoV[All Fields] OR nCoV[All Fields] OR “SARS-CoV-2”[All Fields] OR HCoV-19[All Fields] OR “novel coronavirus”[All Fields]) AND ((((“Mortality”[Subheading] OR “Mortality”[MeSH Terms] OR “Mortalit*”[All Fields] OR “Hospital Mortality”[MeSH] OR “Death*”[All Fields] or “Case fatalit*”[All Fields] OR “all cause mortalit*”[All Fields] OR “all-cause mortalit*”[All Fields] or fatalit*[All Fields])) OR ((Ventilat*[All Fields] OR “Ventilators, Mechanical”[MeSH] OR “continuous positive airway pressure”[MeSH] OR “continuous positive airway pressure”[All Fields] OR “mechanical ventilat*”[All Fields] OR “pulmonary ventilat*”[All Fields] OR respirator[All Fields] OR respirators[All Fields] OR cpap[All Fields] OR ncpap[All Fields] OR APRV[All Fields]))) OR ((“critical care”[MeSH] OR “critical care”[All Fields] OR “Intensive care units”[MeSH] OR “intensive care”[All Fields] OR ICU[All Fields] OR ITU[All Fields] OR CCU[All Fields] OR “intensive treatment unit*”[All Fields] OR “intensive therapy unit*”[All Fields])))) AND ((((“Prognosis”[Mesh]) OR “Severity of Illness Index”[Mesh]) OR “Signs and Symptoms”[Mesh]) OR (presentation OR signs OR symptoms OR manifestations OR markers OR determinants OR features OR prognosis OR prognostic))) Filters: English, from 2019 – 2020

((Algorithm OR Pathway OR Protocol OR Formula OR Criteria OR Standardization OR triage OR stratification OR score AND (english[Filter])) AND ((coronavirus*[Title] OR coronovirus*[Title] OR coronoravirus*[Title] OR coronaravirus*[Title] OR corono-virus*[Title] OR corona-virus*[Title] OR “Coronavirus”[Mesh] OR “Coronavirus Infections”[Mesh] OR “Wuhan coronavirus” [Supplementary Concept] OR “Severe Acute Respiratory Syndrome Coronavirus 2″[Supplementary Concept] OR COVID-19[All Fields] OR CORVID-19[All Fields] OR “2019nCoV”[All Fields] OR “2019-nCoV”[All Fields] OR WN-CoV[All Fields] OR nCoV[All Fields] OR “SARS-CoV-2”[All Fields] OR HCoV-19[All Fields] OR “novel coronavirus”[All Fields]) AND (english[Filter])) AND ((english[Filter]) AND (2019:2020[pdat]))) AND (“Mortality”[Subheading] OR “Mortality”[MeSH Terms] OR “Mortalit*”[All Fields] OR “Hospital Mortality”[MeSH] OR “Death*”[All Fields] or “Case fatalit*”[All Fields] OR “all cause mortalit*”[All Fields] OR “all-cause mortalit*”[All Fields] or fatalit*[All Fields] OR Ventilat*[All Fields] OR “Ventilators, Mechanical”[MeSH] OR “continuous positive airway pressure”[MeSH] OR “continuous positive airway pressure”[All Fields] OR “mechanical ventilat*”[All Fields] OR “pulmonary ventilat*”[All Fields] OR respirator[All Fields] OR respirators[All Fields] OR cpap[All Fields] OR ncpap[All Fields] OR APRV[All Fields] OR “critical care”[MeSH] OR “critical care”[All Fields] OR “Intensive care units”[MeSH] OR “intensive care”[All Fields] OR ICU[All Fields] OR ITU[All Fields] OR CCU[All Fields] OR “intensive treatment unit*”[All Fields] OR “intensive therapy unit*”)

Search: (coronavirus*[Title] OR coronovirus*[Title] OR coronoravirus*[Title] OR coronaravirus*[Title] OR corono-virus*[Title] OR corona-virus*[Title] OR “Coronavirus”[Mesh] OR “Coronavirus Infections”[Mesh] OR “Wuhan coronavirus” [Supplementary Concept] OR “Severe Acute Respiratory Syndrome Coronavirus 2″[Supplementary Concept] OR COVID-19[All Fields] OR COVID-19[All Fields] OR “2019nCoV”[All Fields] OR “2019-nCoV”[All Fields] OR WN-CoV[All Fields] OR nCoV[All Fields] OR “SARS-CoV-2”[All Fields] OR HCoV-19[All Fields] OR “novel coronavirus”[All Fields]) AND ((((“Mortality”[Subheading] OR “Mortality”[MeSH Terms] OR “Mortalit*”[All Fields] OR “Hospital Mortality”[MeSH] OR “Death*”[All Fields] or “Case fatalit*”[All Fields] OR “all cause mortalit*”[All Fields] OR “all-cause mortalit*”[All Fields] or fatalit*[All Fields])) OR ((Ventilat*[All Fields] OR “Ventilators, Mechanical”[MeSH] OR “continuous positive airway pressure”[MeSH] OR “continuous positive airway pressure”[All Fields] OR “mechanical ventilat*”[All Fields] OR “pulmonary ventilat*”[All Fields] OR respirator[All Fields] OR respirators[All Fields] OR cpap[All Fields] OR ncpap[All Fields] OR APRV[All Fields]))) OR ((“critical care”[MeSH] OR “critical care”[All Fields] OR “Intensive care units”[MeSH] OR “intensive care”[All Fields] OR ICU[All Fields] OR ITU[All Fields] OR CCU[All Fields] OR “intensive treatment unit*”[All Fields] OR “intensive therapy unit*”[All Fields]))) Filters: Systematic Reviews

References

1. Shea, B.J., et al., AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both. BMJ, 2017. 358: p. j4008.
2. Ludvigsson, J.F., Systematic review of COVID-19 in children shows milder cases and a better prognosis than adults. Acta Paediatrica. n/a(n/a).
3. Lippi, G., M. Plebani, and B.M. Henry, Thrombocytopenia is associated with severe coronavirus disease 2019 (COVID-19) infections: A meta-analysis. Clin Chim Acta, 2020. 506: p. 145-148.
4. Wynants, L., et al., Prediction models for diagnosis and prognosis of covid-19 infection: systematic review and critical appraisal. BMJ, 2020. 369: p. m1328.
5. Zhao, X., et al., Incidence, clinical characteristics and prognostic factor of patients with COVID-19: a systematic review and meta-analysis. medRxiv, 2020: p. 2020.03.17.20037572.
6. Jin, Y.H., et al., A rapid advice guideline for the diagnosis and treatment of 2019 novel coronavirus (2019-nCoV) infected pneumonia (standard version). Mil Med Res, 2020. 7(1): p. 4.
7. UpToDate. Coronavirus disease 2019 (COVID-19): Epidemiology, virology, clinical features, diagnosis, and prevention. 2020 [cited 2020 14 April 2020].
8. Beeching, N.J., T.E. Fletcher, and R. Fowler. Coronavirus disease 2019 (COVID-19). 2020 [cited 2020 14th April 2020].
9. Organization., W.H., Clinical management of severe acute respiratory infection (SARI) when COVID-19 disease is suspected. 2020, World Health Organization.: https://www.who.int/publications-detail/clinical-management-of-severe-acute-respiratory-infection-when-novel-coronavirus-(ncov)-infection-is-suspected.
10. Excellence, N.I.f.H.a.C. COVID-19 rapid guideline: managing suspected or confirmed pneumonia in adults in the community NICE guideline [NG165]. NICE guideline 2020 [cited 2020 14th April 2020].
11. Greenhalgh, T., et al. NEWS (or NEWS2) score when assessing possible COVID-19 patients in primary care? Oxford COVID-19 Evidence Service 2020 [cited 2020 14th April 2020].
12. Izcovich, A., et al., Prognostic factors for severity and mortality in patients infected with COVID-19: A living systematic review protocol. medRxiv, 2020: p. 2020.04.08.20056598.