A literature review of severity scores for adults with influenza or community-acquired pneumonia – implications for influenza vaccines and therapeutics

ABSTRACT

Influenza vaccination and antiviral therapeutics may attenuate disease, decreasing severity of illness in vaccinated and treated persons. Standardized assessment tools, definitions of disease severity, and clinical endpoints would support characterizing the attenuating effects of influenza vaccines and antivirals. We review potential clinical parameters and endpoints that may be useful for ordinal scales evaluating attenuating effects of influenza vaccines and antivirals in hospital-based studies. In studies of influenza and community-acquired pneumonia, common physiologic parameters that predicted outcomes such as mortality, ICU admission, complications, and duration of stay included vital signs (hypotension, tachypnea, fever, hypoxia), laboratory results (blood urea nitrogen, platelets, serum sodium), and radiographic findings of infiltrates or effusions. Ordinal scales based on these parameters may be useful endpoints for evaluating attenuating effects of influenza vaccines and therapeutics. Factors such as clinical and policy relevance, reproducibility, and specificity of measurements should be considered when creating a standardized ordinal scale for assessment.

KEYWORDS:

• Pneumonia
• influenza
• community-acquired pneumonia
• prediction score
• infectious disease
• severity
• influenza vaccine
• antiviral therapeutics

Introduction

Influenza causes substantial morbidity and mortality in persons of all ages each year.¹ Vaccines are partially protective against influenza illness requiring medical care, and protection can vary year to year, with pooled vaccine effectiveness estimates ranging from 22% to 64%.² Some data suggest that vaccination might attenuate influenza disease, whereby the clinical severity of influenza is less in vaccinated persons as compared with unvaccinated persons.³ Among patients with influenza illness, early antiviral treatment also might attenuate disease severity.^4–7 However, data on the attenuating effects of influenza vaccination and therapeutics are mixed, in part due to the lack of a standardized definition of disease severity or endpoint.

Fundamentally, influenza vaccines and antiviral treatments may limit the physiologic progression of respiratory disease and systemic spread by reducing viral replication and enhancing viral clearance. Prior studies of influenza vaccine attenuation and therapeutics have primarily applied binary endpoints as proxies of disease severity, including hospitalization, admission to intensive care units (ICUs), need for mechanical ventilation, or in-hospital death.^3,8,9 However, these crude binary measures may not accurately reflect disease severity that can be attenuated by vaccination or therapeutics because they are influenced by health care seeking practices, admission biases, and/or underlying health status of patients. Additionally, these binary measures do not capture the full spectrum of influenza progression, from asymptomatic infection to severe illness.

The U.S. Food and Drug Administration recognizes that in absence of a single, gold standard endpoint for seriously ill hospitalized influenza patients, clinical trials should consider endpoints that include clinical signs and symptoms, duration of hospitalization, time to normalization of vital signs and oxygenation, requirements for supplemental oxygen or assisted ventilation, and mortality.¹⁰ A recent influenza expert Working Group proposed the development of an ordinal scale that comprehensively considers a wide range of clinical measures and outcomes to better differentiate the range of disease severity in hospital-based influenza therapeutic trials. Ordinal scales that favor objective clinical measures of influenza severity may offer endpoints that are more specific for trials and observational studies of influenza vaccines and therapeutics.¹¹

Identifying clinical inputs for an ordinal scale specific to influenza is challenging because only a limited number of large, representative hospital-based studies have comprehensively evaluated the clinical manifestations of influenza, particularly as they relate to attenuation mediated by vaccines and antiviral therapeutics. However, many studies have defined and validated “severity scores” for identifying exposure variables at hospital presentation that can be used to predict clinical outcomes and risk-stratify patients with community acquired pneumonia (CAP). These severity scores likely share clinical overlaps with influenza.¹²

We reviewed these studies of CAP and influenza severity in adults to describe potential inputs for ordinal scales specific to the goal of evaluating the attenuating effects of influenza vaccines and antiviral therapeutics in hospital-based studies.

Methods

Search strategy

We performed an initial literature search of PubMed from 1996 to 2021. Relevant search terms included “influenza,” “community acquired pneumonia,” “severity,” and “scale,” “score,” “model,” “tool,” and “prediction.” Results were limited to studies published in English. Types of studies targeted by the review included those which performed systematic sampling of patients such as prospective and retrospective observational studies, randomized control trials, triage cases, and severity assessments of seasonal/pandemic influenza viruses or CAP. References of included studies and relevant reviews were manually inspected to identify any missing studies and scores, determined based on the authors’ clinical and research expertise in respiratory disease.

Study selection

Eligible studies were defined as those reporting on severity scores for influenza and CAP. We excluded studies that investigated specific subpopulations based on underlying risk conditions (e.g., immunocompromise, renal failure), pediatric studies, animal studies, case studies or series, ecological studies, or literature reviews. In addition, given that the focus of this review was on CAP from viral pathogens including influenza, we excluded studies that focused specifically on confirmed bacterial pneumonia, as well as hospital-acquired or aspiration pneumonia. SC and BL manually reviewed the remaining studies to eliminate those that did not mention severity predictors or comorbidities. We excluded studies related to COVID-19 because unique aspects related to the COVID-19 pandemic may reduce generalizability to seasonal influenza.

Data extraction and management

Data from included studies were jointly extracted into a spreadsheet by SC and BL. KA then performed an independent data extraction and verified agreement between the datasets. Expert reviewers were not blinded to study authors, affiliations, or journal name. Variables recorded from each article included study locations, number of patients, average age of patients, percentage of male patients, severity predictors, severity scale used, and outcomes.

Results

The initial literature search yielded 1891 potential articles. An additional 11 studies were included after review of references (Figure 1). Of these, 1535 articles were excluded based on eligibility criteria and 89 because they were duplicates. Of the 278 remaining full-text articles that were manually reviewed, an additional 160 were excluded based on missing severity predictors and/or comorbidities. A total of 118 studies were determined as eligible for inclusion in the review (Table 1).

Table 1. Characteristics of studies included in the review of severity scores among patients with community acquired pneumonia or influenza

First author	Publication year	Scale(s)	Disease	Outcome(s)	Country	Study design	Setting	Number	Mean age	% Male
Adeniji^Citation13	2011	SOFA, STSS	H1N1	mortality, ICU admission, mechanical ventilation	UK	Retrospective	Critical care/ICU	62	41	56.5
Agapakis^Citation14	2010	CURB-65	CAP	hospitalization	Greece	Prospective	Internal Medicine	77	65.9	62.3
Ahn^Citation15	2018	CURB-65, PSI, A-DROP	CAP	mortality (28 days)	South Korea	Retrospective	Emergency or Outpatient	1031	69.2	66.7
Ahnert^Citation16	2019	CRB-65, CURB-65, Halm, IDSA/ATSmc, PSI, qSOFA, SCAP, SIRS, SMART-COP, SOFA	CAP	mortality, ICU admission	Germany, Austria	Prospective	Admitted, Emergency/Acute Care, Critical care/ICU	1532	59	59
Alan^Citation17	2015	CURB-65, PSI	CAP	mortality (6 years)	Switzerland	Prospective	Emergency/Acute Care	925	73	58.8
Aliberti^Citation18	2011	CURB-65	CAP	hospitalization	Italy	Retrospective	Emergency/Acute Care	580	68	56
Andersen^Citation19	2017	CURB-65	CAP	mortality (30 days)	Denmark	Retrospective	Emergency/Acute Care	814	72	47
Andrijevic^Citation20	2014	CURB-65, PSI, MEWS	CAP	mortality (30 days)	Serbia	Prospective	Pulmonary/Respiratory	101	63.7	75.2
Angus^Citation21	2002	PSI, ATS, BTS	CAP	mortality, disease complications, ICU admission, mechanical ventilation	USA and Canada	Prospective	Admitted	1339	N/A	N/A
Arminanzas^Citation22	2013	CURB-65, Charleson Index	CAP	mortality (30 days)	Spain	Prospective	Internal Medicine	539	78	51
Asai^Citation23	2019	A-DROP, CURB-65,I-ROAD, PSI, qSOFA, and SOFA	CAP, HCAP	mortality (30 days), in-hospital mortality	Japan	Retrospective	Admitted	406	79	63
Bello^Citation24	2012	CURB-65, PSI	CAP	mortality	Spain	Prospective	Emergency/Acute Care	228	73	61
Bloos^Citation25	2011	APACHE II, SOFA	CAP, VAP, HAP	mortality (28 days)	Canada, USA, Europe	Prospective	Critical care/ICU	175	62	61.7
Buising^Citation26	2007	CURB-65, PSI, BTS, CURB, CORB	CAP	mortality, ICU admission, IRVS, mechanical ventilation	Australia	Prospective	Emergency/Acute Care	392	74	59.7
Capelastegui^Citation27	2006	CURB-65, CRB-65	CAP	mortality (30 days)	Spain	Prospective	Emergency/Acute Care	1776	61.8	63.3
Chalmers^Citation28	2008	CURB-65, PSI, SMART-COP, CRB-65	CAP	mortality (30 days), IRVS, mechanical ventilation	UK	Prospective	Emergency/Acute Care	335	36	N/A
Chen^Citation29	2010	CURB-65, PSI	CAP	mortality (30 days)	Taiwan	Prospective	Emergency/Acute Care	987	68	61.9
Chen^Citation30	2020	CURB-65, PSI	H1N1, H3N2	mortality (30 days), ICU admission, mechanical ventilation	China	Retrospective	Admitted	693	N/A	66.5
Chiang^Citation31	2013	APACHE II, CURB-65, PSI	CAP	disease severity	Taiwan	Prospective	Emergency or Outpatient	60	59.7	60
Cho^Citation32	2015	CURB-65, PSI	CAP	mortality	South Korea	Prospective	Emergency/Acute Care	56	64.4	69.6
Choi^Citation33	2011	CURB-65, PSI	H1N1 Associated Pneumonia	mortality	South Korea	Retrospective	Admitted	269	48	53.2
Chou^Citation34	2016	CURB-65, PSI	CAP	mortality, disease severity	Taiwan	Prospective	Admitted	279	65.4	58.4
Christ-Crain^Citation35	2006	PSI	CAP	mortality	Switzerland	Prospective	Emergency/Acute Care	302	69.6	61.9
Christ-Crain^Citation36	2008	PSI	CAP	mortality, treatment failure	Switzerland	Prospective	Emergency/Acute Care	302	69	61.9
Courtais^Citation37	2013	PSI	CAP	mortality (30 days)	France	Prospective	Emergency/Acute Care	109	71	65.1
Davis^Citation38	2010	SMART-COP	pneumonia	mortality (30 days), IRVS	Australia	Prospective	Admitted	184	50.1	56
Dwyer^Citation39	2014	CURB-65	CAP	mortality (30 days)	Sweden	Retrospective	Admitted or Outpatient	1172	64	49.6
Ehsanpoor^Citation40	2019	SMART-COP	CAP	mortality (30 days), ICU admission, length of stay, IRVS	Iran	Prospective	Emergency/Acute Care	143	68.1	52.4
Espana^Citation41	2006	CURB-65, PSI, SCAP	CAP	mortality (30 days)	Spain	Prospective	Emergency/Acute Care	1776	64.7	63.3
Espana^Citation42	2012	CURB-65	CAP	disease severity	Spain	Prospective	Emergency or Family Physician	344	53.5	57.3
Espana^Citation43	2015	CURB-65, PSI, SCAP	CAP	mortality (30 days), disease complications	Spain	Prospective	Emergency/Acute Care	491	N/A	63.1
Ewig	1998	ATS, BTS	CAP	mortality, ICU admission	Spain	Prospective	Admitted	696	67.8	66
Flanders^Citation44	1999	PSI	CAP	mortality	USA	Retrospective	Admitted	1024	67	43
Fukuyama^Citation45	2011	CURB-65, PSI, SMART-COP, Espana rule, IDSA/ATS rule	CAP	mortality, ICU admission	Japan	Prospective	Admitted	505	76	67.1
Gearhart^Citation46	2019	IDSA/ATS	CAP	ICU admission	USA	Retrospective	Admitted	8284	68	46
Golcuk^Citation47	2015	CURB-65	CAP	mortality (28 days)	Turkey	Prospective	Emergency/Acute Care	174	66.7	66.1
Gordo-Remartinez^Citation48	2015	PSI	CAP	mortality (30 days, 90 days), ICU admission, hospital readmission	Spain	Prospective	Emergency/Acute Care	226	75.6	55.3
Grudzinska^Citation49	2019	CURB-65, Lac-CURB-65, NEWS, qSOFA	CAP, HCAP	mortality (30 days, 90 days, 1 year), in-hospital mortality, ICU admission, length of stay	UK	Retrospective	Admitted	1545	76	50.8
Gunaydin^Citation50	2019	CURB-65, PSI	CAP	mortality	Turkey	Prospective	Internal Medicine	63	72.1	57.1
Guo^Citation51	2012	PSI, IDSA/ATS	CAP	mortality	China	Retrospective	Admitted	1230	47.5	49.3
Gwak^Citation52	2015	PSI	CAP	mortality	South Korea	Retrospective	Emergency/Acute Care	397	71.4	57.7
Hamaguchi^Citation53	2018	CURB-65, PSI	CAP, HCAP	mortality	Japan	Prospective	Outpatient	1772	77	58.7
Huaman^Citation54	2011	CURB-65, SMART-COP, SMRT-CO	CAP	ICU admission	USA	Retrospective	Admitted	115	61.6	48.7
Huang^Citation55	2008	CURB-65, PSI	CAP	mortality (30 days)	USA	Prospective	Emergency/Acute Care	1651	65	52
Huang^Citation56	2009	CURB-65, PSI	CAP	mortality (30 days)	USA	Prospective	Emergency/Acute Care	1653	65	52
Ito^Citation57	2017	CURB-65, PSI, IDSA/ATS, A-CROP	CAP	mortality (30 days), ICU admission	Japan	Retrospective	Admitted	1834	73.5	69.8
Jain^Citation58	2012	N/A	Pandemic H1N1-associated pneumonia	mortality, ICU admission	USA	Retrospective	Admitted	451	29	48
Jeong^Citation59	2013	PSI	CAP, HCAP	mortality	South Korea	Retrospective	Emergency/Acute Care	247	68	67.4
Jo^Citation60	2016	CURB-65, NEWS, PSI	CAP	mortality	South Korea	Retrospective	Admitted	553	70.1	62.6
Justel^Citation61	2013	APACHE II	Pandemic influenza	mortality	Spain	Prospective	Critical care/ICU	40	N/A	70
Kasamatsu^Citation62	2012	CURB-65, PSI, A-DROP	CAP	mortality (30 days)	Japan	Prospective	Internal Medicine	170	67.9	57.6
Katsoulis^Citation63	2005	PSI	CAP	disease severity	Greece	Prospective	Pulmonary/Respiratory	40	N/A	75
Kim^Citation64	2016	CURB-65, PSI	CAP	mortality (30 days)	South Korea	Prospective	Emergency/Acute Care	386	75	60.2
Kim^Citation65	2017	APACHE II, CURB-65, PSI, qSOFA, SOFA	CAP	mortality (28 days)	South Korea	Retrospective	Emergency/Acute Care	125	N/A	62.4
Ko^Citation66	2014	APACHE II, CURB-65, PSI	CAP	mortality	Taiwan	Prospective	Admitted	121	59	60
Kolditz^Citation67	2010	PSI	CAP	mortality, ICU admission, clinical instability after 72 h	Germany	Prospective	Emergency or Outpatient	59	72	52
Lee^Citation68	2011	PSI	CAP	mortality (28 days)	South Korea	Prospective	Emergency/Acute Care	424	70.4	35
Lee^Citation69	2016	PSI	CAP	mortality (30 days)	South Korea	Prospective	Emergency/Acute Care	470	71.5	68.7
Leroy^Citation70	1996	SAPS II, Glasgow score, pneumonia specific prognostic index	CAP	mortality	France	Prospective & Retrospective	Critical care/ICU	335	63	63
Li^Citation71	2015	CURB-65, CUR-65	CAP	mortality	China	Retrospective	Pulmonary/Respiratory	1230	47.5	49.3
Li^Citation72	2015	IDSA/ATS	CAP	mortality	China	Retrospective	Pulmonary/Respiratory	1230	47.5	49.3
Liapikou^Citation73	2009	IDSA/ATS guidelines	CAP	mortality (7 days, 30 days), ICU admission	Spain	Prospective	Emergency/Acute Care	2102	N/A	61.4
Lim	2000	BTSr, mBTSr	CAP	mortality	UK	Retrospective	Admitted	122	76	56
Lim^Citation74	2003	CURB-65, PSI, CURB, CRB-65	CAP	mortality (30 days)	UK, New Zealand, Netherlands	Prospective	Admitted	1068	64.1	51
Liu^Citation75	2014	CURB-65	SCAP/CAP	mortality (28 days), Acute Respiratory Distress Syndrome (ARDS), disseminated intravascular coagulation (DIC)	China	Prospective	Emergency/Acute Care	573	N/A	62.1
Liu^Citation76	2016	CURB-65, PSI, SMART-COP, expanded CURB-65, A-DROP	CAP	mortality (30 days), length of stay	China	Retrospective	Admitted	1640	64	59.6
Loh	2004	BTS	CAP	mortality, length of stay	Malaysia	Prospective	Admitted	108	55	58
Luo^Citation77	2018	CURB-65, PSI	CAP	mortality, hospital discharge, disease severity, in-hospital mortality	China	Prospective	Admitted	188	68	64.4
Mahendra^Citation78	2018	CURB-65	CAP	ICU admission	India	Prospective	Pulmonary/Respiratory	100	54	66
Mamani^Citation79	2017	CURB-65	CAP	mortality (30 days), ICU admission, length of stay	Iran	Prospective	Admitted	149	N/A	69.9
Masia^Citation80	2005	PSI, PORT PSI	CAP	mortality	Spain	Prospective	Admitted	240	59	62.5
Meier^Citation81	2017	CURB-65, PSI, qSOFA	CAP	mortality (30 days, 6 years), ICU admission	Switzerland	Retrospective	Emergency/Acute Care	268	N/A	60.1
Mendez^Citation82	2019	CURB-65, SOFA	L-CAP	mortality (30 days), length of stay, treatment failure	Spain	Prospective	Admitted	217	N/A	62.2
Menendez^Citation83	2009	CURB-65, PSI, CRB65	CAP	mortality (30 days)	Spain	Prospective	Admitted	453	67.3	62.2
Miller^Citation84	2012	CURB-65, PSI, SIRS	H1N1	mortality, ICU admission	USA	Prospective	Admitted	43	36.4	19
Miyazaki^Citation85	2018	PSI, A-DROP	CAP	mortality (30 days)	Japan	Prospective	Pulmonary/Respiratory	534	76.2	63.3
Muller^Citation86	2006	PSI	CAP	mortality	Switzerland	Prospective	Emergency/Acute Care	545	67.1	62.6
Muller^Citation87	2007	PSI	LRTI, CAP	mortality, disease cure	Switzerland	Retrospective	Emergency/Acute Care	545	67.1	62.6
Murcia^Citation88	2010	PSI, Barthel index	CAP	mortality (30 days)	Spain	Prospective	Emergency/Acute Care	518	60.7	65.8
Naderi^Citation89	2015	CURB-65, PSI, SMART-COP, CRB-65, IDSA/ATS 2007	CAP	mortality, IRVS	Iran	Prospective	Admitted	120	50.4	14.3
Nastasijevic Borovac^Citation90	2014	PSI	CAP	mortality	Serbia	Prospective	Admitted	129	N/A	N/A
Nicolini^Citation91	2012	Opravil XRC	H1N1-associated pneumonia	severity of pulmonary infiltrates	Italy	Retrospective	Critical care/ICU	28	31.7	53.6
Nowak^Citation92	2012	CURB-65, PSI	CAP	mortality (30 days, long-term)	Switzerland	Prospective	Emergency/Acute Care	341	75	61
Oh^Citation93	2011	APACHE II	H1N1	disease severity, length of stay, antiviral regimens	South Korea	Prospective	Admitted	709	50	39.5
Papadimitriou-Olivgeris^Citation94	2020	CURB-65, PSI, qSOFA, SIRS, SOFA	Influenza A and B	mortality (30 days), disease complications, disease severity, hospitalization, ICU admission, length of stay, mechanical ventilation	Switzerland	Retrospective	Admitted	441	N/A	46
Parsonage^Citation95	2009	CURB-65	CAP	mortality (30 days), ICU admission, in-hospital mortality	United Kingdoms	Prospective	Emergency/Acute Care	428	N/A	47
Pereira	2012	CURB-65, PSI, PIRO-CAP	H1N1-associated CAP	mortality	31 different countries	Prospective	Admitted	265	42	54
Phua^Citation96	2009	CURB-65, PSI, IDSA/ATS minor criteria	CAP	mortality, ICU admission	Singapore	Retrospective	Admitted	1242	65.7	61.3
Querol-Ribelles^Citation97	2004	APACHE II, PSI	CAP	mortality	Spain	Prospective	Emergency/Acute Care	302	73	74.8
Remmelts^Citation98	2012	CURB-65, PSI	CAP	mortality (30 days), ICU admission	Netherlands	Prospective	Emergency/Acute Care	272	63.5	56
Renaud^Citation99	2012	REA-ICU	CAP	mortality (30 days), ICU admission, IRVS, length of stay, mechanical ventilation	Switzerland	Retrospective	Emergency/Acute Care	877	73	58.8
Richards^Citation100	2011	APACHE II, CURB-65, PSI	CAP	mortality (28 days), in-hospital mortality	11 different countries	Retrospective	Admitted	278	64	N/A
Robins-Browne^Citation101	2012	PSI, SMART-COP, SMARTACOP, CORB	CAP	mortality (30 days), IRVS	Australia	Prospective	Emergency/Acute Care	367	50	52
Ronan^Citation102	2010	CURB-65, CRB-65	CAP	mortality (30 days)	Scotland	Retrospective	Emergency/Acute Care	428	N/A	47
Salluh^Citation103	2011	APACHE II, CURB-65, SOFA	CAP	mortality, in-hospital mortality	Brazil	Prospective	Critical care/ICU	90	73.5	44.4
Schuetz^Citation104	2008	CURB-65, PSI	CAP	mortality, ICU admission	Switzerland	Retrospective	Emergency/Acute Care	281	74	62
Schuetz^Citation105	2008	CURB-65, PSI	CAP	mortality, ICU admission, outpatient management	Switzerland	Retrospective	Emergency/Acute Care	373	73	60.1
Schuetz^Citation106	2010	CURB-65, PSI	LRTI, CAP	mortality, disease complications, ICU admission	Switzerland	Retrospective	Emergency/Acute Care	925	72	58.8
Shindo^Citation107	2008	CURB-65, A-DROP	CAP	mortality (30 days)	Japan	Retrospective	Admitted	329	75	59.9
Siljan^Citation108	2018	CURB-65	CAP	mortality (30 days), ICU admission	Norway	Prospective	Internal Medicine	247	64	51.8
Song^Citation109	2019	qSOFA	CAP	mortality, in-hospital mortality	South Korea	Retrospective	Emergency/Acute Care	443	66.5	57.1
Spoorenberg^Citation110	2017	CURB-65, PSI	CAP	mortality (30 days, long-term), ICU admission, in-hospital mortality	Netherlands	Retrospective	Admitted	289	64	55.7
Talmor^Citation111	2007	Triage rule	H5N1	mortality, in-hospital mortality	USA	Retrospective	Emergency/Acute Care	5133	N/A	48.2
Valencia^Citation112	2007	CURB-65, PSI, ATS, CURB	CAP	mortality, ICU admission	Spain	Prospective	Admitted	457	79	70
Vecchiarino^Citation113	2004	PSI	CAP	mortality (30 days), hospital discharge, length of stay, hospital readmission	USA	Prospective	Admitted	213	72.5	47.4
Vicco^Citation114	2015	CURB-65, SCAP	CAP	mortality	Argentina	Prospective	Admitted	272	52	48.2
Wang^Citation115	2013	APACHE II, CURB-65, PSI	CAP	levels of YKL-40 concentration	Taiwan	Prospective	Admitted	121	59.5	60.3
Wang^Citation116	2017	APACHE II, CURB-65, PSI, SOFA, CLCGH	SCAP	mortality (28 days), length of stay, mechanical ventilation, duration of ICU stay	China	Retrospective	Critical care/ICU	37348	70.9	51.5
Wu^Citation117	2009	APACHE II	CAP	mortality (28 days)	Taiwan	Prospective	Emergency/Acute Care	63	70	63.5
Yang^Citation118	2014	PSI	H1N1/H3N2	disease severity	China	Prospective	Admitted	88	52	54.5
Yeon Lee^Citation119	2016	CURB-65, PSI	CAP	mortality (30 days)	South Korea	Retrospective	Pulmonary/Respiratory	797	N/A	64.1
Yin^Citation120	2014	CURB-65, PSI	CAP	mortality (30 days)	China	Prospective	Emergency/Acute Care	573	74	62.1
Zhang^Citation121	2018	CURB-65, PSI	CAP	mortality (30 days)	Singapore	Retrospective	Emergency/Acute Care	1902	73	55.5
Zhou^Citation122	2012	genetic variations	H1N1	disease severity	Hong Kong	Prospective	Admitted or Outpatient	425	N/A	48.9
Zhou^Citation123	2018	CURB-65, PSI, qSOFA, SOFA	CAP	mortality (28 days)	China	Retrospective	Emergency/Acute Care	226	65	84.1
Zhu^Citation124	2018	APACHE II	H1N1/H7N9 Influenza	N/A	China	Prospective	Admitted	93	N/A	54.8
Zimmerman^Citation125	2010	CRP levels	H1N1 Influenza A virus	ICU admission, mechanical ventilation	Israel	Retrospective	Emergency/Acute Care	191	43	50
de Jager^Citation126	2012	CURB-65	CAP	mortality, ICU admission	Netherlands	Prospective	Emergency/Acute Care	395	63	61

N/A = Overall mean age and/or percent male of study participants was not available in the study report.

Figure 1. Study selection flow diagram.

The majority of studies were performed in Europe (n = 50, 42.4%) and Asia (n = 43, 36.4%) (Table 2). One hundred three (87.3%) studies investigated CAP, 11 (9.3%) studied influenza, and four (3.4%) focused on both. Median study size was 370 patients (IQR 175–709), median percentage of male participants was 59.3 (IQR 52.0–62.6), and median age was 67.0 years (IQR 59.7–72.1). Nine (7.6%) studies included information on the underlying influenza and pneumococcal vaccination status of their subjects.^{33,42,43,58,78,82,88,108,118} Of the 15 studies that evaluated influenza patients, six (40.0%) included information on antiviral treatments administered during hospitalization, including oseltamivir, zanamivir, amantadine, ribavirin, and peramivir.^{30,33,58,93,118,124} Only study specifically discussed in-hospital treatment with a drug (statins) that might modify the host response to CAP and influenza.⁸²

Table 2. Summary of included study characteristics

Characteristics	N (%)
Total studies	118
Study Origin
Asia	43 (36.4)
Europe	50 (42.4)
Global/multi-country	5 (4.2)
Middle East	6 (5.1)
North America^a	9 (7.6)
Oceania	3 (2.5)
South America	2 (1.7)
Publication Year
1996–2000	4 (3.4)
2001–2005	7 (5.9)
2006–2010	28 (23.7)
2011–2015	47 (39.8)
2016–2020	32 (27.1)
Study Design
Prospective	74 (62.7)
Retrospective	43 (36.4)
Both	1 (0.9)
Disease
Pneumonia^b	103 (87.3)
Influenza^c	11 (9.3)
Both^d	4 (3.4)
Scores^e
A-DROP	7 (5.9)
APACHE II	13 (11.0)
CURB-65	73 (61.9)
IDSA/ATS	9 (7.6)
NEWS	2 (1.7)
PSI	74 (62.7)
qSOFA	8 (6.8)
SAPS II	1 (0.9)
SMART-COP	9 (7.6)
SOFA	10 (8.5)
Other^f	38 (32.2)
Outcomes^e
Discharge	2 (1.7)
Disease complications/severity	13 (11.0)
Hospitalization	3 (2.5)
ICU admission	33 (28.0)
Intensive respiratory or vasopressor support (IRVS)	7 (5.9)
Length of hospital/ICU stay	11 (9.3)
Mechanical ventilation	9 (7.6)
Mortality	103 (87.3)
Readmission	2 (1.7)
Treatment failure	2 (1.7)
Other	8 (6.8)
	Median (IQR)
Patients Evaluated	370 (175–709)
% Male	59.3 (52–62.6)
Mean age	67 (59.7–72.1)

^aAll studies from United States of America.

^bIncludes community-acquired pneumonia (CAP), hospital-acquired pneumonia (HAP), healthcare-associated pneumonia (HCAP), lymphopenic community-acquired pneumonia (L-CAP), lower respiratory tract infections (LRTIs), severe community-acquired pneumonia (SCAP), and ventilator-associated pneumonia (VAP).

^cIncludes H1N1, H3N2, H5N1, and H7N9.

^dAll studies on H1N1-associated pneumonia.

^eNot mutually exclusive.

^fOthers include modified or earlier versions of scores described.

Scores

A total of 11 commonly used severity scores for influenza and community acquired pneumonia were identified, including 10 found from the literature review (Figure 2). A Simple Clinical Score was added based on the authors’ knowledge of clinical and respiratory research. Several additional scales were identified (e.g., REA-ICU and the Espana rule).^45,99 However, due to significant overlap with more commonly applied scales and few published studies validating these scales in different settings, they were not included in our review.

Figure 2. Predictors of poor outcomes by severity scores.

Legend: BP = blood pressure, RR = respiratory rate, HR = heart rate/pulse, O2 = oxygen saturation, BUN = blood urea nitrogen, WBC = white blood cell^aAge adjustments for RR and O2, ^bSystolic, ^cMean arterial pressure (mmHg), ^dHypotension, ^eVentilated or non-ventilated, ^fRespiratory rate or complaining of breathlessness, ^gAbnormal ECG (does not include bradycardia or tachycardia), ^hTachycardia, ⁱOxyhemoglobin saturation measured by pulse oximetry or partial pressure of oxygen in arterial blood (PaO2), ^jAaDO2 or PaO2, ^kPaO2/FiO2 ratio, ventilated or CPAP, ^lPartial pressure of oxygen, ^mPaO2, SaO2, or PaO2/FiO2, ⁿhypothermia, ^oBUN or dehydration, ^pThrombocytopenia, ^qleukopeniac, ^rConfusion, ^sGlasgow Coma Score, ^tLevel of consciousness or new confusion using Alert Voice Pain Unresponsive (AVPU) scale, ^uComa (responds only to pain or unresponsive) and altered mental status (not intoxicated), ^vConfusion (new onset), ^wFactored into respiratory rate, ^xMechanical ventilation or CPAP, ^yUnscheduled surgical, scheduled surgical, or medical, ^zseptic shock with the need for vasopressors, ^aaCerebrovascular disease, congestive heart failure, hepatic disease, renal disease, neoplastic disease, ^abHIV, hematologic malignancy, metastatic cancer, ^acDisability: stroke (new presentation), unable to stand unaided, prior illness (some part of daytime in bed), diabetes (type 1 or 2), ^adMultilobar chest x-ray infiltrates, ^aePleural effusion.

For the included scores, a combination of indicators on demographic, vital signs, laboratory results, mental status, in-hospital interventions, medical history, and radiographic findings were used to assess a variety of outcomes, including mortality, disease complications, disease severity, hospital discharge, hospital readmission, hospitalization, ICU admission, intensive respiratory or vasopressor support (IRVS), hospital or ICU length of stay, mechanical ventilation, and treatment failure.

Acute physiology and chronic health evaluation (APACHE II), 1985

Thirteen studies used APACHE II, including ten (76.9%) that were focused on CAP and 3 (23.1%) that investigated influenza.^{25,31,61,65,66,93,97,100,103,115–117,124} APACHE II is a disease severity classification score that has been subsequently revised (APACHE III and IV); however, studies identified in this review primarily used APACHE II. APACHE II is designed to evaluate the risk of hospital death in acutely ill patients admitted to the ICU and is used in clinical decision-making, resource allocation, and evaluation of ICU performance.¹²⁷ Patients are assigned a three-part score using 12 physiological variables, age, and chronic health status. Physiological variables range from high to low abnormality, including temperature, mean arterial pressure, heart rate, respiratory rate (non-ventilated or ventilated), oxygenation, arterial pH, serum sodium, serum potassium, serum creatinine, hematocrit, white blood cell count, and Glasgow Coma Score. Age points range from zero (≤44 years) to six (≤75 years). Chronic health conditions are evaluated to determine history of severe organ (liver, cardiovascular, respiratory, or renal) insufficiency or immunocompromised status. Studies using the APACHE II score predicted outcomes such as mortality, length of stay, and disease severity.

A-DROP, 2006

Seven CAP studies used the A-DROP score, which was developed by the Japanese Respiratory Society to predict CAP patient outcomes and inform clinical decision-making.^{15,23,57,62,76,85,107,128,129} A-DROP uses a combination of measures obtained through clinical assessment and laboratory testing, including sex-adjusted age (male ≥70 years and female ≥75 years), blood urea nitrogen (≥21 mg/dL or dehydration), oxyhemoglobin saturation (pulse oximetry ≤90% or partial pressure of oxygen in arterial blood ≤60 mmHg), confusion, and systolic blood pressure (≤90 mmHg). A-DROP has been used evaluate and validate outcomes for predicting mortality, ICU admission, and length of hospital/ICU stay.

CURB-65, 2003

CURB-65 was used in 73 studies, including 68 (93.2%) focused on pneumonia, three (4.1%) influenza, and two (2.7%) studies that examined both diseases.^{14–20,22–24,26–34,39,41–43,45,47,49,50,53–57,60,62,64–66,71,74–79,81–84,89,92,94–96,98,100,102–108,110,112,114–116,119–121,123,126,130} The score is a tool for emergency departments or primary care settings to triage patients into mortality risk groups.⁷⁴ It is based on the modified British Thoracic Society (mBTS) severity assessment tool which uses clinical features to identify severe CAP patients at high risk of mortality.¹³¹ A six-point score (0–5) is assigned based on clinical presentation of mental confusion (abbreviated Roth-Hopkins mental test ≤8 or disorientation), elevated urea (>19 mg/dL), respiratory rate (≥30 breaths/min), blood pressure (diastolic ≤60 mmHg or systolic <90 mmHg), and age (>65 years).¹³² Patients are categorized into mortality risk groups, with corresponding disposition recommendations, including home treatment (low risk), short-stay inpatient or hospital supervised outpatient treatment (intermediate risk), and ICU admission for hospital management of severe pneumonia (high risk).

In most studies, the CURB-65 score was evaluated/validated to predict outcomes including mortality, disease complications, hospitalization or ICU admission, duration of hospital or ICU stay, intensive respiratory or vasopressor support, mechanical ventilation, and treatment failure.

Infectious diseases society of America/American thoracic society (IDSA/ATS) consensus guidelines, 2007

Nine CAP studies used the IDSA/ATS consensus guidelines.^{16,45,46,51,57,72,73,89,96} These guidelines were proposed to improve clinical management of adult CAP patients by defining severe CAP and providing major and minor criteria to assess the need for ICU admission.¹³³ Major criteria include needing invasive mechanical ventilation and septic shock with the use of vasopressors. Minor criteria include respiratory rate (≥30 breaths/min), PaO2/FiO2 ratio (≤250), multilobar infiltrates, confusion/disorientation, uremia (BUN level, ≥20 mg/dL), leukopenia (<4000 white blood cells/mm³), thrombocytopenia (platelet count, <100,000 cells/mm³), hypothermia (core temperature, <36°C), and hypotension requiring aggressive fluid resuscitation. Studies using the IDSA/ATS criteria predicted outcomes of mortality, ICU admission, and intensive respiratory or vasopressor support.

National early warning score (NEWS), 2012

Two CAP studies used the NEWS standardized scoring system for assessment of acutely ill patients requiring critical care interventions.^49,60 NEWS was developed by a working group in the United Kingdom to standardize early detection of clinical deterioration.¹³⁴ Patients are evaluated at the time of presentation or after hospital admission based on six physiological parameters: respiratory rate, oxygen saturation, temperature, systolic blood pressure, pulse rate, and level of consciousness (Alert Voice Pain Unresponsive scale). Individual parameters are then combined with an additional indicator of patients requiring supplemental oxygen, and the total score is stratified into three trigger levels with varying clinical evaluation responses: low (nurse assessment for enhanced clinical monitoring or critical care escalation), medium (urgent physician assessment for critical care escalation), and high (emergency assessment by a critical care team). Studies have used NEWS to predict outcomes of mortality, ICU admission, and duration of inpatient stay.

Pneumonia severity index (PSI), 1997

Seventy-four studies used the PSI score: 68 (91.9%) CAP, four (5.4%) influenza, and two (2.7%) both diseases.,^{15–17,20,23,24,26,28–34,41,43,45,50,51,53,55–57,60,62,64–66,74,76,77,81,83–85,88,89,92,94,96–98,100,104–106,115,118,110,112,119–121,21,35–37,123,130,44,48,52,59,63,67–69,80,86,87,90,101,113,135136}PSI was derived and validated using data from a 1991 statewide database of adult CAP inpatients, as well as the Pneumonia Patient Outcomes Research Team (PORT) cohort study of CAP inpatients and outpatients. PSI is a prediction tool used to evaluate risk of death (30-day hospital mortality) in adult CAP patients using a combination of clinical and laboratory features with the goal of developing consistent criteria for hospitalization by eliminating subjectivity in clinical assessment. In the first PSI step, predictor variables obtained from intake history and physical examination stratify patients into low risk of mortality (class I) and high risk (classes II–V) patients requiring laboratory testing. Indicators include age (>50 years), preexisting conditions (cerebrovascular disease, congestive heart failure, hepatic disease, renal disease, neoplastic disease), altered mental status, heart rate (≥125 beats per minute), respiratory rate (≥30 breaths/min), systolic blood pressure (<90 mmHg), and temperature (<35°C or ≥40°C). For high risk patients, demographic, laboratory, and radiographic factors are used for further classification into risk classes II–V, including sex, nursing home residency, blood urea nitrogen concentration (≥30 mg/dL or ≥11 mmol/L), glucose concentration (≥250 mg/dL or ≥14 mmol/L), hematocrit (<30%), sodium concentration (<130 mmol/L), partial pressure of oxygen (<60 mmHg or <8 kPa), arterial pH (<7.35), and radiographically-determined pleural effusions. Studies using PSI predicted outcomes of mortality, clinical instability, disease complications, hospital discharge and readmission, ICU admission, intensive respiratory or vasopressor support, duration of hospital or ICU stay, mechanical ventilation, and treatment failure.

Sequential organ failure assessment (SOFA), 1996

Ten studies used SOFA, including eight (80.0%) CAP studies and two (20.0%) influenza.^{13,16,23,25,65,82,94,103,116,123} SOFA is an outcome prediction evaluation tool for ICU morbidity and mortality that was developed for sepsis, with the goal of improving ICU performance, therapeutic decision making, and resource allocation.^137,138 Organ dysfunction or failure is monitored over time, with SOFA scores calculated at time of admission and every 48 hours until discharge. SOFA assigns a score for each organ ranging from normal to high dysfunction using physiological parameters including respiration (PaO2/FiO2 with and without mechanical ventilation), coagulation (platelets x 10³/mm³), liver (bilirubin, mg/dL), cardiovascular (hypotension, use of vasopressors), central nervous system (Glasgow Coma Score), and renal (creatinine, mg/dl). SOFA has been used to predict mortality, ICU admission, duration of stay, mechanical ventilation, and treatment failure.

qSOFA, 2016

Seven CAP studies (87.5%) and one (12.5%) influenza study used qSOFA, a “quick” form of the SOFA scale that eliminates laboratory tests from scoring parameters.^{16,23,49,65,81,94,109,123} qSOFA is used to identify patients presenting outside of the ICU who have a suspected infection and are at high risk for sepsis and in-hospital mortality.¹³⁹ Three scoring measures are used: altered mental status (Glasgow coma score), systolic blood pressure (≤100 mmHg), and respiratory rate (≥22 breaths/min). Studies used qSOFA to predict outcomes of mortality, ICU admission, and duration of inpatient stay.

Simple clinical score (SCS), 2010

SCS is a tool that was not used by any studies but identified by the authors for inclusion. It was developed for use in severely ill patients presenting as emergencies to predict risk of mortality.^140,141 Independent predictors of mortality are used to assign points: age (>75 years, male ≥50 and ≤75, female ≥55 and ≤75), airway (coma – responding only to pain or unresponsive, oxygen saturation <90% or ≥90% and <95%), breathing (respiratory rate >30 breaths/min, >20 breaths/min and ≤30 breaths/min, complaining of breathlessness), circulation (systolic blood pressure ≤70 mmHg, >70 mmHg and ≤80 mmHg, >80 mmHg and ≤100 mmHg, or pulse > systolic blood pressure), disability (stroke, altered mental status, unable to stand unaided or a nursing home resident, prior illness, diabetes type 1 or 2), ECG (abnormal, does not include bradycardia or tachycardia), and temperature (<35°C or ≥39°C).

Simplified acute physiology score (SAPS) II, 1993

One CAP study used SAPS II to predict mortality.⁷⁰ SAPS II is designed to estimate risk of in-hospital mortality specifically within an ICU setting.^142,144 Measures included in SAPS II are age, heart rate, systolic blood pressure, temperature, respiration (PaO2/FiO2, ventilated or continuous pulmonary artery pressure), urinary output, serum urea, white blood cell count, serum potassium, serum sodium, serum bicarbonate, bilirubin level, Glasgow Coma Score, type of admission (unscheduled surgical, scheduled surgical, or medical), and history of chronic diseases (HIV, hematologic malignancy, metastatic cancer).

SMART-COP, 2008

Nine pneumonia/CAP studies used the SMART-COP severity scale.^{16,28,38,40,45,54,76,89,101} SMART-COP predicts the need for IRVS in CAP patients, with the aim of identifying candidates for ICU admission.¹⁴⁵ SMART-COP uses a combination of eight clinical, laboratory, and radiographic indicators associated with IRVS to assign a severity score: systolic blood pressure (<90 mmHg), multilobar chest radiography, albumin level (<3.5 g/dL), respiratory rate (age-adjusted: ≥25 breaths/min for ≤50 years old, ≥30 breaths/min for >50 years old), tachycardia (heart rate ≥125 beats per minute), confusion (new onset), oxygenation (age-adjusted: PaO2 < 70 mmHg/SaO2 < 93%/PaO2/FiO2 < 333 for ≤50 years old, PaO2 < 60 mmHg/SaO2 < 90%/PaO2/FiO2 < 250 for >50 years old), and arterial pH (<7.35). Scores are stratified into risk of intensive respiratory and vasopressor support brackets of low, moderate, high, and very high. Study outcomes included mortality, ICU admission, intensive respiratory or vasopressor support, length of stay, and mechanical ventilation.

Discussion

In our review of the literature, we found several objective measures of clinical severity that can be used as endpoints for evaluating the attenuating effects of influenza vaccines and antiviral therapeutics. Common parameters across eleven severity scores included vital signs (hypotension, fever, hypoxia) laboratory results (blood urea nitrogen, platelets, serum sodium), medical interventions (oxygen use, vasopressor treatment, and mechanical ventilation) and radiographic findings. The severity scores comprised a mix of these parameters of acute disease severity and other indicators such as age, sex, and underlying conditions. These scores were designed to predict outcomes such as mortality, ICU admission, and duration of hospital/ICU stay primarily for risk stratification, provision of clinical care, and patient disposition.

Crude outcomes such as hospitalization, mechanical ventilation, and ICU admission are often used as endpoints in studies of attenuation mediated by vaccines and therapeutics because they are meaningful targets for prevention.^146,147 However, these surrogate outcomes for disease severity may not reflect progression of disease related to viral replication, the host response to infection, or the full range of severe illness. Moreover, these outcomes are also influenced by factors such as age and preexisting medical conditions which vaccines do not affect.¹⁴⁸ In turn, they may diminish specificity for measuring vaccine and antiviral attenuation and offer explanations for inconsistent findings of attenuation in clinical studies of vaccines and therapeutics.^{3,8,9,146–151}

Lack of standardization in definitions of common parameters also limits their objectivity and reproducibility, as with mental status, health care practices, and medical history. Because individual severity scores were not designed for evaluation of attenuation by vaccines or therapeutics either in randomized controlled trials or observational studies, some scores may be either overly narrow (e.g., qSOFA, CURB-65) or be cumbersome to administer in critical care settings (e.g., PSI, SAPS II). The fidelity of implementing is also influenced by the temporal sequence of measuring exposure variables to predict clinical outcomes. Studies often use exposures to predict outcomes which may have already occurred (e.g., intubation). Additionally, many of the scores used by these studies were developed to assess overall disease severity (APACHE II, NEWS, SCS, SAPS II) and might be too generic to be applied only to influenza.

A Working Group established by the US Health and Human Services noted the need for a standardized ordinal severity scale to facilitate trials of hospital-based influenza therapeutic trials.¹¹ A similar need also has been noted for studies of vaccine attenuation of clinical disease.^142,143 Creation of a unified scale should consider objective indicators and endpoints that have high clinical and public health policy relevance (e.g., markers of acute lung injury), are reproducible across clinical settings, and are specific to vaccine and antiviral attenuation. The development of such scales has been tremendously valuable for other vaccine preventable diseases such as rotavirus and pertussis.^152,153 We believe that use of the physiologic parameters of disease severity common across the severity scores identified in our review might offer practical, valid, and specific measures of vaccine attenuation. Moreover, most of these parameters of acute disease severity are also on the pathway to clinically meaningful outcomes such as death and ICU admission.

Lacking a single gold standard endpoint for evaluating vaccine and antiviral attenuation, validation of such an ordinal scale may follow a Delphi approach to rank ordering measures, as has been proposed in other studies.^11,154 Additionally, adjustments must be considered for factors such as older age and preexisting conditions, which may falsely increase severity scores due to lower thresholds for poor outcomes.¹⁵⁵ When the prevalence of such conditions is high and they are not adjusted for, severity scores may lead to biased results in evaluations of vaccine and antiviral attenuation. For example, older adults with poor respiratory reserve from underlying chronic obstructive pulmonary are more likely than healthy younger adults to have respiratory failure requiring mechanical ventilation from infection that is similarly severe.¹⁵⁶

One key limitation of this literature review is that of the 118 identified studies, only 15 (12.7%) evaluated influenza.^{13,30,33,58,61,84,91,93,94,111,118,122,124,125,130} Of these 15 studies, six (40.0%) used PSI, five (33.3%) used CURB-65, three (20.0%) used APACHE II, two (13.3%) used SOFA, and one (6.7%) used qSOFA. Clinical outcomes predicted included mortality (overall, ICU, in-hospital), ICU admission, mechanical ventilation, length of stay, and hospitalization. Scoring systems examined in the context of influenza demonstrate limitations in applicability toward predicting clinical outcomes, including potential underestimation of mortality particularly in low-risk patients (e.g., younger age, no comorbidities). While there are clinical overlaps between CAP and influenza, the applicability of these severity scores may be limited. Scores also might use severity at different points in the disease progression and standardization to use scores on arrival, prior to confounding by treatment, would be useful.

Studies also showed significant heterogeneity across clinical parameters and research endpoints, and we were unable to pool results to perform a meta-analysis of findings. Furthermore, our review criteria included scores that have been validated for use in specific clinical settings (ICU, emergency/acute care, pulmonary/respiratory departments) and at various points along the patient care pathway, which may not represent optimal timing for endpoint measurements in evaluating disease severity.

Conclusion

Reducing the severity of influenza illness is an important benchmark for success of influenza vaccination and antiviral therapeutics. In future years, an increasing number of novel influenza vaccines, antivirals, and additional influenza therapeutics such as monoclonal or polyclonal antibodies, immune plasma, small-molecule inhibitors, or immune modulators are likely to be available for prevention and control of severe influenza. Evaluation of these agents for influenza can be complicated for influenza due to the lack of standardized endpoints and definitions for disease severity. Most studies of vaccine protection and antiviral effectiveness in hospitalized patients have focused on attenuation of severe outcomes such as influenza-associated hospitalization, intensive care unit admission, hospital or ICU length of stay, or mortality. While these outcomes are relevant and practical endpoints to consider, they may bias results when outcomes are influenced by factors that are not necessarily related to virus-mediated disease severity such as underlying conditions, treatments, and age. An ordinal scale that focuses on physiologic parameters which can be practical to measure, standardized across studies, and less biased by factors such as underlying conditions, health care seeking, or admission practices is needed to accurately assess disease severity. Our review of studies of acute respiratory infection identified physiological parameters of disease severity common across studies conducted in various settings worldwide. Future development of an ordinal scale that incorporates these common physiologic parameters of acute respiratory illness severity could be useful for evaluation of disease attenuation through influenza vaccination and therapeutics in observational studies and trials.¹⁴³

Disclaimer

The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention (CDC).

Disclosure statement

The authors declare no Conflicts of Interest.

Additional information

Funding

The author(s) reported there is no funding associated with the work featured in this article.