Skip to main content
NCBI home page
Journal of the American College of Emergency Physicians Open logoLink to Journal of the American College of Emergency Physicians Open
. 2025 Sep 9;6(5):100245. doi: 10.1016/j.acepjo.2025.100245

A Host-Protein Test for Differentiating Bacterial From Viral Infection: Diagnostic Accuracy in Elderly Patients

Tanya M Gottlieb 1, Yaly Orr 1, Hagai Hamami 1, Roy Navon 1, Lior Kellerman 1, Eran Eden 1, Daniel Haber 2, Neta Petersiel 3, Ami Neuberger 4,5, Adam J Singer 6, Mical Paul 2,5, Richard E Rothman 7,∗
PMCID: PMC12529685  PMID: 41114128

Abstract

Objectives

Older adults are vulnerable to infection and are difficult to diagnose. This study assessed the performance of MeMed BV (MMBV), a host-protein test for differentiating bacterial from viral infection, in adults ≥65 years.

Methods

Post hoc pooled and meta-analysis of adults with suspected acute infections enrolled in 3 prospective studies. MMBV results were interpreted as bacterial/viral/equivocal per manufacturer’s instructions. Reference standard infection etiology was adjudicated by experienced physicians who were blinded to MMBV. Diagnostic accuracy for bacterial infection was calculated for MMBV results (area under the receiver operating characteristic curve [AUC], bin analysis) and for unequivocal MMBV results (sensitivity/specificity). MMBV’s potential impact on antibiotic use was estimated by comparing MMBV-guided decisions to actual practice.

Results

A total of 754 younger (18-64 years) and 248 older (≥65 years) adults were included. Among older adults, the median age was 75.0 years (IQR, 69.0, 82.0), 53.2% were male, 68.1% were hospitalized, and 79.0% had ≥3 comorbidities. Respiratory tract infections were common (76.2%), and 85.1% were prescribed antibiotics. A total of 111 patients were assigned a bacterial reference standard infection etiology, 77 a viral etiology, and 60 an indeterminate etiology. In pooled analysis, MMBV attained comparable AUC in older (0.95; 95% CI, 0.92-0.98) vs younger adults (0.95; 0.93-0.97). Focusing on older adults, 96.2% (90.3-98.8) sensitivity and 85.7% (74.8-92.5) specificity with 10.6% equivocal results were observed. MMBV could reduce potentially unwarranted antibiotic prescriptions 2.5-fold (from 62.3% to 24.7%; P < .0001). Using a bivariate model, MMBV similarly attained AUC 0.92 (0.81-0.97).

Conclusions

MMBV demonstrated high diagnostic accuracy in older adults, supporting its potential to optimize antibiotic use in this population. Further studies are needed to evaluate real-world utility.

Keywords: older adults, CRP, TRAIL, IP-10, host-protein diagnostic test, bacterial infection, antibiotic stewardship

The Bottom Line.

Older adults (≥65 years) are vulnerable to infection. Discriminating bacterial vs viral causes is challenging because of comorbidities and similar symptoms. Accurate, rapid tools are needed that enable reduced antibiotic prescription to patients with viral infections without causing bacterial infections to be missed. MeMed BV (MMBV), a blood test that measures 3 immune system proteins, helps physicians determine whether an infection is bacterial or viral. This study shows that also among older adults MMBV attains high diagnostic accuracy and could potentially reduce unnecessary antibiotic prescriptions 2.5-fold. Further studies are needed to evaluate clinical utility in real-world practice for older adults.

1. Introduction

1.1. Background

Infectious diseases, in particular community-acquired pneumonia (CAP), are a leading cause for hospital referrals, hospitalizations, morbidity, and mortality in older adults (≥65 years of age).1,2 In the United States, the incidence of CAP is over 2.5-fold greater among adults aged 65 to 79 vs younger adults aged <65 years old, with the incidence of CAP in individuals ≥80 years of age reaching 16.4 cases per 1000 adults.3

Diagnosing the etiology of infection in the elderly can be challenging because their clinical presentations can be atypical.1 Traditional laboratory and imaging tests that are used and are available during the emergency department (ED) stay frequently fail to identify a definitive pathogen in real time and can be subjective.4, 5, 6 It was estimated that the etiology of ∼50% of all CAP cases is never identified.7, 8, 9 Although the advent of broad molecular testing has increased the yield of identified pathogens,10 for CAP, these techniques rely on the acquisition of sputum specimens, which can be challenging in the elderly. Further, in the elderly population, laboratory biomarkers such as white blood cell count or C-reactive protein (CRP) are known to be less reliable in guiding clinical practice.11,12

1.2. Importance

The diagnostic uncertainty in evaluating elderly patients with suspected infection, along with the considerable risk for morbidity and mortality, leads to particularly high rates of antibiotic use (>50% per capita vs younger populations).13 It is estimated that 1 out of 3 antibiotic prescriptions for elderly patients is unwarranted14 and approximately half of those unwarranted prescriptions are given for respiratory tract infections (RTIs).14,15 Antibiotics can cause significant adverse effects along with allergic reactions, drug interactions, and increased risk for Clostridioides difficile infections and ultimately contribute to antimicrobial resistance.16,17 All of these effects are compounded in the elderly because of higher rates of comorbidity. Accordingly, there is an unmet need for a tool that has both high sensitivity and high specificity for bacterial infections, so that unnecessary antibiotic prescriptions to patients with viral infections can be avoided while ensuring patients with bacterial infections are still accurately identified.18

1.3. Goals of This Investigation

A host-protein test (MeMed BV [MMBV]), based on tumor necrosis factor-related apoptosis-inducing ligand (TRAIL), interferon γ-induced protein-10 (IP-10), and CRP, has demonstrated high performance for differentiating bacterial from viral infections in children and adults.19, 20, 21, 22, 23, 24, 25, 26 Here, the performance of MMBV is evaluated in adults, focusing on elderly patients with suspected infectious disease, and specifically elderly patients that may be considered at high risk for complications of infection: patients with multiple comorbidities (≥3 chronic illnesses), patients aged ≥75 years, and those with RTI and their respective lower-risk mirror subgroups. A combined dataset for this post hoc pooled analysis and meta-analysis was derived from 3 previously completed prospective studies of patients recruited from EDs, urgent care centers, and/or inpatient hospital ward settings.19,24,25

2. Methods

2.1. Study Design

This is a post hoc analysis of adult patients who were enrolled in 3 previous prospective studies evaluating the diagnostic accuracy of MMBV:

  • (1)

    The Apollo study (NCT04690569), which recruited patients aged >3 months with suspected acute infection between May 2019 and August 2020 from 9 EDs and 3 urgent care centers in the United States and Israel.25

  • (2)

    The Curiosity study (NCT01917461), which recruited patients aged >3 months with suspected acute infections between August 2009 and November 2013 from the ED, internal medicine, and surgical departments of 2 medical centers in Israel.19

  • (3)

    The Observer study (NCT03011515), which recruited patients aged >18 years with lower RTI (LRTI) signs or symptoms between March 2017 and October 2018 at 3 EDs in Israel.24

The pooling of the individual patient data is outlined in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis flow diagram (Fig S1). Studies were approved by the ethics committees of each participating center (Table S1), and informed consent was required.

2.2. Eligibility Criteria

The inclusion criteria for this post hoc analysis were individuals aged 18 years and older along with eligibility in the parent study. The eligibility criteria for each parent study are as described previously.19,24,25 Briefly, patients were included who exhibited symptoms of acute infectious disease and experienced fever within the last 7 days. Patients with ≥1 of the following conditions were excluded: active inflammatory disease, congenital or acquired immunodeficiency, chronic fungal or parasitic infection, hepatitis B virus, hepatitis C virus, infection with active tuberculosis, significant trauma or burns in the last 7 days, major surgery within the last 7 days, pregnancy, and active malignancy. The complete eligibility criteria are given in Table S1.

2.3. Index Test (MMBV)

The MMBV result is a score, ranging from 0 to 100, that computationally integrates TRAIL, IP-10, and CRP measurements using a previously derived algorithm,19 which was employed in multiple previous studies.20, 21, 22,24, 25, 26 Higher scores are indicative of bacterial infection (or coinfection); for cutoffs, see section 2.6 Statistical Analysis.

In the Observer and Curiosity studies, the serum concentrations of TRAIL and IP-10 were measured using ImmunoXpert (MeMed), CRP was measured on a Cobas platform, and the MMBV score was computed using the ImmunoXpert software (MeMed). In the Apollo study, TRAIL, IP-10, and CRP were measured, and the MMBV score was calculated using the MeMed Key platform. Of note, the scores generated by the 2 measurement platforms (ImmunoXpert and Key) have been established as comparable.27

2.4. Reference Standard Infection Etiology Based on Adjudication

Given the absence of a gold standard for determining bacterial or viral infection etiology, the reference standard for the cause of infection was generated based on panel adjudication as described previously.28 The adjudicators were physicians with relevant clinical experience (Table S1). Briefly, each adjudicator independently reviewed anonymized electronic case report forms containing comprehensive clinical, laboratory, radiologic, microbiological, and follow-up data and assigned to each patient an etiologic label (bacterial, viral, or “I don’t know”), along with their subjective level of confidence (≥90% or 70%-90%). For this study, the original adjudication labels were combined according to the following rubric: to classify a case as bacterial or viral reference standard infection etiology, at least 2 out of 3 adjudicators were required to have assigned a bacterial or viral label, respectively, with high confidence (≥90%). Cases not meeting this condition were assigned an “indeterminate” reference standard infection etiology.

2.5. Blinding

The adjudicators were blinded to MMBV results. Those performing MMBV were blinded to the adjudications. CRP could be ordered according to the physician’s discretion in all studies and, when measured, was available to the adjudicators.

2.6. Statistical Analysis

Descriptive statistics were used to summarize the data. Continuous data were summarized as medians and IQRs. Categoric data were summarized as numbers and frequency of occurrence.

MMBV’s diagnostic accuracy was evaluated against the adjudication-based reference standard infection etiology, with indeterminate reference standard infection etiology cases excluded, using the following statistical frameworks:

  • 1.

    Area under the receiver operating characteristic curve (ROC-AUC) analysis: ROC-AUC was calculated including equivocal patients using all possible thresholds.

  • 2.

    Bin analysis: the likelihood of bacterial infection was analyzed as an increasing function of MMBV score. Patients were assigned to 1 of 5 predetermined MMBV score bins19, 20, 21, 22, 23, 24, 25, 26: 0 ≤ score ≤ 10, high likelihood of viral infection (or other nonbacterial etiology); 10 < score < 35, moderate likelihood of viral infection (or other nonbacterial etiology); 35 ≤ score ≤ 65, equivocal; 65 < score < 90, moderate likelihood of bacterial infection (including coinfection); and 90 ≤ score ≤ 100, high likelihood of bacterial infection (including coinfection). Within the bins, patients were assigned according to their reference standard infection etiology as described previously.28 Equivocal patients were included.

  • 3.

    Sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), positive likelihood ratio, and negative likelihood ratio: sensitivity was defined as the number of patients who received a bacterial reference standard infection etiology and a bacterial MMBV result (score > 65), divided by the number of patients who received a bacterial reference standard infection etiology. Specificity was defined as the number of patients who received a viral reference standard infection etiology and a viral MMBV result (score < 35), divided by the number of patients who received a viral reference standard infection etiology. PPV was defined as the number of patients who received a bacterial reference standard infection etiology and a bacterial MMBV result (score > 65), divided by the number of patients who received a bacterial MMBV result. NPV was defined as the number of patients who received a viral reference standard infection etiology and a viral MMBV result (score < 35), divided by the number of patients who received a viral MMBV result. Cases with equivocal MMBV results were excluded from these calculations. Of note, an equivocal MMBV score (35 ≤ score ≤ 65) represents a valid test result (ie, it is not a failed test).

The distribution of MMBV results for the cases with an indeterminate reference standard infection etiology is presented along with the proportions that indicate a high likelihood for bacterial or viral infection.

MMBV’s diagnostic accuracy was evaluated across the younger and older adults and, for older adults, across subgroups that may be considered at high risk for complications of infection: patients with multiple comorbidities (≥3 chronic illnesses), patients aged ≥75 years, and those with RTI, as well as across the mirror subgroups (patients with <3 comorbidities and patients aged 65-74 years).

For older adults, in addition to assessing performance based on pooling of individual patient data, a post hoc meta-analysis was conducted that incorporates study-level adjustment. Briefly, 2 formal meta-analytic frameworks were used to validate the assumption of low heterogeneity in the main pooled analysis.29,30 For more details, refer to Supplementary Appendix 1.

Although MMBV results were not provided to attending physicians, the potential impact of physicians’ theoretical use of MMBV for decision-making regarding antibiotic use was estimated based on the reference standard infection etiology and actual antibiotic prescription practice. Potentially unwarranted antibiotic use was defined as cases in which antibiotics were prescribed, but the reference standard infection etiology was viral. Potentially missed bacterial infections that may benefit from antibiotics were defined as cases in which antibiotics were not prescribed, but the reference standard infection etiology was bacterial. MMBV’s conjectured impact on antibiotic use was calculated based on the assumption that the physician would have changed their prescription practice to align with MMBV (eg, if MMBV was bacterial, then the physician would have prescribed antibiotics, and in cases with equivocal MMBV results, their practice would not change).

3. Results

3.1. Study Population

The 3 parent studies enrolled a total of 2055 potentially eligible patients. Of them, 248 were eligible and ≥65 years of age (Fig 1): 111 were assigned a bacterial reference standard infection etiology by the adjudication panel, 77 a viral etiology, and 60 an indeterminate etiology (Table 1). There were 754 eligible younger adults (18-64 years), of whom 205 were assigned a bacterial reference standard infection etiology by the adjudication panel, 409 a viral etiology, and 140 an indeterminate etiology (Table S2).

Figure 1.

Figure 1

Open in a new tab

Patient enrollment flow.

Table 1.

Characteristics of older adult patients (≥65 years).

Category Older adults (n = 248) Older adults with bacterial reference standard infection etiology (n = 111) Older adults with viral reference standard infection etiologya (n = 77) P valueb Older adults with indeterminate reference standard infection etiology (n = 60)
Demographics
 Age (y), median (IQR) 75.0 (69.0, 82.0) 76.0 (71.0, 82.0) 73.0 (68.0, 82.0) .105 77.0 (70.0, 82.0)
 Sex (male), n (%) 132 (53.2%) 74 (66.7%) 27 (35.1%) <.001 31 (51.7%)
Symptoms onset
 Time from symptoms onset (d), median (IQR) 3.0 (2.0, 5.0) 3.0 (2.0, 5.0) 3.0 (2.0, 5.0) .636 3.0 (2.0, 4.0)
 Onset: ≤2 d, n (%) 80 (32.3%) 35 (31.5%) 21 (27.3%) .627 24 (40.0%)
 Onset: >2 d, n (%) 168 (67.7%) 76 (68.5%) 56 (72.7%) .627 36 (60.0%)
Hospitalization
 Hospitalized, n (%) 169 (68.1%) 104 (93.7%) 22 (28.6%) <.001 43 (71.7%)
 Hospitalization duration (d), median (IQR) 5.0 (3.0, 8.0) 5.5 (3.0, 8.0) 4.0 (3.0, 6.8) .332 4.0 (4.0, 6.5)
Treatment
 Antibiotics use, n (%) 211 (85.1%) 111 (100.0%) 48 (62.3%) <.001 52 (86.7%)
No. of comorbiditiesc
 0, n (%) 9 (3.6%) 0 (0.0%) 9 (11.7%) <.001 0 (0.0%)
 1, n (%) 13 (5.2%) 2 (1.8%) 8 (10.4%) .017 3 (5.0%)
 2, n (%) 30 (12.1%) 13 (11.7%) 13 (16.9%) .391 4 (6.7%)
 +3, n (%) 196 (79.0%) 96 (86.5%) 47 (61.0%) <.001 53 (88.3%)
Comorbiditiesc
 Hypertension, n (%) 190 (76.6%) 93 (83.8%) 50 (64.9%) .005 47 (78.3%)
 Other comorbidity, n (%) 87 (35.1%) 42 (37.8%) 26 (33.8%) .644 19 (31.7%)
 Hyperlipidemia, n (%) 139 (56.0%) 70 (63.1%) 29 (37.7%) <.001 40 (66.7%)
 Diabetes mellitus, n (%) 105 (42.3%) 54 (48.6%) 24 (31.2%) .024 27 (45.0%)
 Ischemic heart disease, n (%) 76 (30.6%) 36 (32.4%) 15 (19.5%) .066 25 (41.7%)
 COPD, n (%) 41 (16.5%) 18 (16.2%) 12 (15.6%) 1.000 11 (18.3%)
 Chronic kidney disease, n (%) 45 (18.1%) 24 (21.6%) 6 (7.8%) .014 15 (25.0%)
 Heart disease, n (%) 35 (14.1%) 13 (11.7%) 9 (11.7%) 1.000 13 (21.7%)
 Asthma, n (%) 26 (10.5%) 13 (11.7%) 8 (10.4%) .819 5 (8.3%)
 Obesity, n (%) 26 (10.5%) 11 (9.9%) 5 (6.5%) .596 10 (16.7%)
Diagnosisd
 URTIe, n (%) 38 (15.3%) 5 (4.5%) 28 (36.4%) <.001 5 (8.3%)
 LRTIf, n (%) 157 (63.3%) 74 (66.7%) 39 (50.6%) .034 44 (73.3%)
 UTI, n (%) 14 (5.6%) 12 (10.8%) 0 (0.0%) .002 2 (3.3%)
 Unspecified viral infection, n (%) 12 (4.8%) 1 (0.9%) 7 (9.1%) .009 4 (6.7%)
 Bacteremia, n (%) 12 (4.8%) 11 (9.9%) 0 (0.0%) .003 1 (1.7%)
 Cellulitis, n (%) 8 (3.2%) 8 (7.2%) 0 (0.0%) .022 0 (0.0%)
 Other diagnosisg, n (%) 23 (9.3%) 9 (8.1%) 6 (7.8%) 1.000 8 (13.3%)
Open in a new tab

COPD, chronic obstructive pulmonary disease; LRTI, lower respiratory tract infection; URTI, upper respiratory tract infection; UTI, urinary tract infection.

a

Includes both “viral” and “non-infectious” reference standard diagnoses.

b

P values refer to “bacterial reference standard etiology” versus “viral reference standard etiology.” For “age,” “time from symptoms onset” and “hospitalization duration” P values were calculated using the Mann-Whitney U test. The remaining P values were calculated using Fisher exact test.

c

See Table S3 for the full list of comorbidities.

d

A patient can have more than 1 discharge diagnosis, except for “other diagnosis.”

e

URTI includes the following discharge diagnoses: URTI and pharyngitis/tonsillitis.

f

LRTI (lower respiratory tract infection) includes the following discharge diagnoses: pneumonia, acute bronchitis, COPD (chronic obstructive pulmonary disease) exacerbation, LRTI, bronchiolitis, asthma exacerbation, suspected pneumonia, and asthma unspecified.

g

“Other diagnosis” includes the following discharge diagnoses: acute cholecystitis (n = 1), diverticulitis (n = 1), gastroenteritis (n = 2), fever without source (n = 2), acute appendicitis (n = 1), non-infectious (n = 1), other (n = 6), peritonitis (n = 1), septic shock (n = 2), CHF exacerbation (n = 2), pulmonary edema (n = 1), unknown (n = 2), fever (n = 1), morbid obesity (n = 1), chills (n = 1), headache (n = 1), and cough (n = 2).

3.2. Characteristics of Elderly Cohort

Among the older adults, the median age was 75.0 years (IQR, 69.0, 82.0; Table 1), 53.2% (132/248) were male, 76.6% (190/248) had hypertension, 56.0% (139/248) had hyperlipidemia, 42.3% (105/248) had diabetes, 16.5% (41/248) had chronic obstructive pulmonary disease, 18.1% (45/248) had chronic kidney disease, and 14.1% (35/248) had heart disease. A total of 196 patients (79.0%) had ≥3 comorbidities (See Table S3 for the full list of comorbidities in older adults). The time from symptom onset at the time of recruitment was >2 days for 67.7% (168/248) of the patients. Most patients, 68.1% (169/248), were hospitalized with a median hospitalization duration of 5.0 days (IQR, 3.0, 8.0). The most common discharge diagnosis was lower RTI (63.3%; 157/248). There was at least 1 bacterial detection in 37.5% (93/248) of the patients, at least 1 viral detection in 52.8% (131/248) of the patients, and 1 fungal detection (0.4% [1/248]). Overall, at least 1 pathogen was detected in 70.2% (174/248) of the population (this includes routine care as well as study-specific testing; Table S4). Of the 248 eligible elderly patients, 151 had bacterial, 73 had viral, and 24 had equivocal MMBV results.

3.3. MMBV Performance

To assess diagnostic performance, we compared MMBV results to the reference standard infection etiology. When analyzed across all older adult patients with a bacterial or viral reference standard infection etiology (n = 188), including those with equivocal MMBV results, MMBV attained an ROC-AUC of 0.95 (95% CI, 0.92-0.98). Higher MMBV scores correlated with a higher likelihood of a bacterial infection (Table S5). The bacterial prevalence, ie, the number of patients assigned a bacterial reference standard infection etiology out of the analysis cohort, was 59.0% (111/188).

For older adult patients with unequivocal MMBV results (n = 168), MMBV exhibited a sensitivity for bacterial infection of 96.2% (101/105, 95% CI, 90.3-98.8), a specificity of 85.7% (54/63, 74.8-92.5), a PPV of 91.8% (101/110, 85.0-95.8), and an NPV of 93.1% (54/58, 83.1-97.8) (Table 2). The equivocal rate was 10.6% (20/188). Four older adult patients with a bacterial reference standard infection etiology had a viral MMBV result (ie, false negatives; Table S6); none were bacteremic or experienced septic shock. Nine older adult patients with a viral reference standard infection etiology had a bacterial MMBV result (false positives; Table S7). Notably, all 12 older adult patients with documented bacteremia or septic shock (n = 12) received high-confidence bacterial MMBV results (90 ≤ score ≤ 100).

Table 2.

MMBV performance in older adults.

Statistic 95.0% CI
Sensitivity (%) 96.2 90.3-98.8
Specificity (%) 85.7 74.8-92.5
PPV (%) 91.8 85.0-95.8
NPV (%) 93.1 83.1-97.8
Total (n) 188
Open in a new tab

Patients with equivocal MMBV results were excluded from this analysis; 10.6% (20/188) of the population.

MMBV, MeMed BV; NPV, negative predictive value; PPV, positive predictive value.

In addition to the pooled analyses, to address potential heterogeneity among the 3 studies, we performed a bivariate random-effects meta-analysis and a split component synthesis (SCS) analysis in older adults. Notably, similar results were found using both types of analyses. For example, the AUC calculated using the bivariate model was 0.92 (0.81-0.97) and using the SCS model was 0.92 (0.85-0.96), compared with 0.95 (0.92-0.98) calculated in the pooled analysis (Supplementary Appendix 1).

Among the older adult patients with indeterminate reference standard infection etiology, 68.3% (41/60) had bacterial MMBV results; of them, 73.2% (30/41) received MMBV results indicative of a high likelihood for a bacterial infection (score ≥ 90) (Fig S2). In the same group, 25% (15/60) had viral MMBV results, of whom 66.7% (10/15) received MMBV results indicative of a high likelihood for a viral infection (score ≤ 10). Four patients with an indeterminate reference standard infection etiology (6.7%) received equivocal MMBV results.

The performance of MMBV in older adults was comparable to its performance in younger adults (aged 18-64 years) recruited in the same 3 studies (Table S8).

3.3. MMBV Performance in Older Adult Subgroups

MMBV’s diagnostic accuracy was evaluated in 3 older adult subgroups that may be considered at high risk for complications of infection and, accordingly, received high rates of antibiotic treatment in the present study: patients with multiple comorbidities (≥3 chronic illnesses, n = 143; 90.2% [129/143] received antibiotics), patients aged ≥75 years (n = 93; 92.5% [86/93] received antibiotics), and those with RTI (n = 147; 83.7% [123/147] received antibiotics). MMBV performed with sensitivity ranging from 94.8% to 96.0% and specificity ranging from 81.1% to 86.2% (Tables S9 and S10). The equivocal rate was 9.5% to 14.0%. Of note, MMBV performed robustly also in the mirror subgroups of patients with <3 comorbidities and patients aged 65-74 years.

3.4. Conjectured Impact of MMBV on Optimizing Antibiotic Prescription in Older Adults

Among the 60 older adult patients with indeterminate reference standard infection etiology, there were high prescription rates both for those with high likelihood bacterial MMBV results (27/30) and those with high likelihood viral MMBV results (8/10).

Among the 77 older adult patients who were assigned a viral reference standard infection etiology, 48 were prescribed antibiotics (62.3%) by the physician at the ED or ward (Fig 2). Assuming full adoption of MMBV by the physician (see the “Methods” section), MMBV could reduce this potentially unwarranted antibiotic prescription 2.5-fold to 24.7% (19/77; P value < .0001). All 111 patients assigned a bacterial reference standard infection etiology were treated with antibiotics; 4 of them (3.6%) had a viral MMBV result. These cases represent potentially missed bacterial infections by MMBV (false negatives). Among the subgroups, MMBV could reduce potentially unwarranted antibiotic use 2.2-fold in patients with ≥3 comorbidities (from 70.2% to 31.9%; P value < .001; Fig 2) and 2.7-fold in patients with RTIs (from 65.7% to 24.3%; P value < .0001).

Figure 2.

Figure 2

Open in a new tab

Conjectured impact of MeMed BV (MMBV) on antibiotic use. For potentially unwarranted antibiotic use, “current practice” represents the proportion of patients assigned a reference standard viral infection etiology and prescribed antibiotics according to the medical record. “Current practice + MMBV” represents an extrapolation of what would happen if MMBV was available and adopted. Respiratory tract infection (RTI) diagnosis included lower RTI, upper RTI, and unspecified viral infection.

4. Limitations

A limitation of this study is that direct utility could not be assessed because the studies were observational. Another intrinsic methodologic limitation is associated with using pooled retrospective data from multiple previous studies,19,24,25 which constrains analyses to data points that were originally collected. For example, data relating to the status of each comorbidity were not available, preventing a more detailed analysis of MMBV performance in older adult patients with active disease. Data missing from the original studies may have impacted their adjudication processes, but it is noteworthy that each study provided a wealth of routine and study-specific patient data to the adjudicators and that this methodology for reference standard etiology is considered the best possible in the absence of a gold standard. Notably, this study included only patients presenting with fever and did not evaluate MMBV performance in elderly patients who may have an infection without a febrile response, potentially introducing a bias. Further studies are warranted in elderly patients with systemic indications of infection but without a fever. Lastly, the adjudicators were blinded to MMBV but provided with the CRP value if available, potentially introducing incorporation bias in their adjudication of infection etiology. This notwithstanding, the conjectured impact of MMBV on optimizing appropriate antibiotic use in older adults demonstrates superiority to the standard of care, which included CRP.

5. Discussion

This study assessed MMBV’s diagnostic accuracy in older adults. The elderly are a growing sector of the global population,31 for whom healthcare costs are the highest.32 This population is particularly susceptible to infection.33, 34, 35, 36, 37 There is a dearth of accurate tools that give actionable information on the patient’s immune response to pathogens, and it remains unclear whether such tests would perform well in the elderly due to immunosenescence.34 Here, we show that a host-protein test called MMBV for differentiating between bacterial and viral infections based on integration of 3 immune proteins exhibits high diagnostic accuracy in the elderly population, with a sensitivity for bacterial versus viral infection of 96.2% and a specificity of 85.7%. The performance is on par with that observed in younger adults aged 18 to 64 years enrolled in the same studies, as well as with prior validation studies in pediatric patients, in which MMBV demonstrated sensitivity of 92% to 94% and specificity of 89% to 94%.19, 20, 21, 22, 23,25,26 In addition, based on MMBV’s accuracy in this elderly population, we show that MMBV could optimize antibiotic prescribing, reducing potentially unwarranted antibiotic use 2.5-fold without significantly increasing the number of missed bacterial infections.

MMBV’s performance was maintained in older adult subgroups that may be considered at increased risk for complications from severe infection based on the presence of ≥3 comorbidities or being aged ≥75 years, with all 12 older adult patients diagnosed with bacteremia or septic shock receiving high-confidence bacterial MMBV results (90 ≤ score ≤ 100). MMBV performed well also in older adult subgroups with potentially lower risk for infection complications (<3 comorbidities or aged 65-74 years), representing patients for whom physicians may more readily consider adoption of a new test.

Overall, the vast majority (85.1%) of the older adult patients in this study population were prescribed antibiotics, including 62.3% of those who were adjudicated as having a viral reference standard infection etiology. This finding likely reflects the high level of physician diagnostic uncertainty when evaluating elderly patients with fever and suspected infection, as well as the conservative practice patterns given the increased risk of complications in the event of a “missed” bacterial diagnosis. The conjectured analysis supports that MMBV could significantly reduce potentially unwarranted antibiotic use in the elderly in general and specifically in patients with a high likelihood of receiving antibiotics. MMBV’s actual impact on practice and safety associated with its use is being evaluated in ongoing studies. Of note, for 94% of patients, CRP was typically available as part of the standard of care. Therefore, the conjectured impact of MMBV on appropriate antibiotic use in older adults can be considered as on top of that enabled by CRP. Taken together, the findings of this study suggest that MMBV has the potential to add value beyond the standard of care.

Multiple studies have reported health economic models that support MMBV’s costeffectiveness.38, 39, 40, 41 For example, a literature-based economic model focusing on CAP in adults reported per-patient savings of $293 for payers and $809 for providers when MMBV was added to standard diagnostics.38 Additional studies are warranted to explore the cost impact of MMBV specifically in the elderly.

A strength of this diagnostic accuracy study is the rigorous reference standard for infection etiology based on panel adjudication. Additionally, the study assessed elderly patients who may be considered at high risk for complications of infection in whom assessing etiology is known to be more challenging and for whom the test is anticipated to have important clinical value in real-world practice. Another strength is that the study population included patients enrolled in multiple previous studies, recruited in both the United States and Israel over a period spanning 10 years, supporting the generalizability of the findings.

In conclusion, MMBV demonstrated high diagnostic accuracy for differentiating between viral and bacterial infections in adults ≥65 years of age, supporting its potential to optimize appropriate antibiotic use in this population. Future clinical utility studies that include assessing the safety of using the test to guide decision making are necessary to determine whether a reduction in unwarranted antibiotic prescriptions can be realized in real-world practice.

Author Contributions

TMG, YO, HH, EE, AS, AN, MP, and RER took part in the study conception and design, analysis and interpretation of the data, and preparation of the manuscript.

LK, RN, DH, and NP took part in the acquisition of subjects and/or data, analysis, and interpretation of the data.

All authors reviewed and approved the final manuscript.

Funding and Support

The Observer parent study (Observer) was supported by the H2020 Innovation In SMEs (684589). MeMed funded 2 of the parent studies (Curiosity and Apollo).

Conflict of Interest

TMG, YO, HH, RN, LK, and EE are employees of MeMed, some of whom have options. AJS is participating in a randomized controlled trial of MMBV performance sponsored by MeMed. RER participated in a scientific board on health care economic modeling for MeMed (without compensation) and participated in the Apollo parent diagnostic accuracy study that was sponsored by MeMed. DH, NP, AN, and MP have no conflicts of interest to declare.

Acknowledgments

We thank our colleagues for their input: Boris Lebedenko, MSc, Jeroen Stas, PhD, Peter Antkowiak, MD, Amir Nakar, PhD, and Efrat Flashner-Abramson, PhD.

Footnotes

Drs Gottlieb, Orr, and Hamami contributed equally to this study.

Supervising Editor: Faheem Guirgis, MD

Supplementary material associated with this article can be found in the online version at https://doi.org/10.1016/j.acepjo.2025.100245.

Supplemental material

Supplementary Material
mmc1.docx (429.6KB, docx)

References

  • 1.Beckett C.L., Harbarth S., Huttner B. Special considerations of antibiotic prescription in the geriatric population. Clin Microbiol Infect. 2015;21(1):3–9. doi: 10.1016/j.cmi.2014.08.018. [DOI] [PubMed] [Google Scholar]
  • 2.Nelson R.E., Hyun D., Jezek A., Samore M.H. Mortality, length of stay, and healthcare costs associated with multidrug-resistant bacterial infections among elderly hospitalized patients in the United States. Clin Infect Dis. 2022;74(6):1070–1080. doi: 10.1093/cid/ciab696. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Jain S., Self W.H., Wunderink R.G., et al. Community-acquired pneumonia requiring hospitalization among U.S. adults. N Engl J Med. 2015;373(5):415–427. doi: 10.1056/NEJMoa1500245. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4.Yoshikawa T.T., Norman D.C. Geriatric infectious diseases: current concepts on diagnosis and management. J Am Geriatr Soc. 2017;65(3):631–641. doi: 10.1111/jgs.14731. [DOI] [PubMed] [Google Scholar]
  • 5.Esme M., Topeli A., Yavuz B.B., Akova M. Infections in the elderly critically-ill patients. Front Med (Lausanne) 2019;6:118. doi: 10.3389/fmed.2019.00118. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Prendki V., Scheffler M., Huttner B., et al. Low-dose computed tomography for the diagnosis of pneumonia in elderly patients: a prospective, interventional cohort study. Eur Respir J. 2018;51(5) doi: 10.1183/13993003.02375-2017. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Shoar S., Musher D.M. Etiology of community-acquired pneumonia in adults: a systematic review. Pneumonia (Nathan) 2020;12(1):11. doi: 10.1186/s41479-020-00074-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Torres A., Cilloniz C., Niederman M.S., et al. Pneumonia. Nat Rev Dis Primers. 2021;7(1):25. doi: 10.1038/s41572-021-00259-0. [DOI] [PubMed] [Google Scholar]
  • 9.GBD 2016 Lower Respiratory Infections Collaborators Estimates of the global, regional, and national morbidity, mortality, and aetiologies of lower respiratory infections in 195 countries, 1990-2016: a systematic analysis for the Global Burden of Disease Study 2016. Lancet Infect Dis. 2018;18(11):1191–1210. doi: 10.1016/S1473-3099(18)30310-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Gadsby N.J., Russell C.D., McHugh M.P., et al. Comprehensive molecular testing for respiratory pathogens in community-acquired pneumonia. Clin Infect Dis. 2016;62(7):817–823. doi: 10.1093/cid/civ1214. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Tang Y., Fung E., Xu A., Lan H.Y. C-reactive protein and ageing. Clin Exp Pharmacol Physiol. 2017;44(suppl 1):9–14. doi: 10.1111/1440-1681.12758. [DOI] [PubMed] [Google Scholar]
  • 12.American College of Emergency Physicians Clinical Policies Subcommittee (Writing Committee) on Community-Acquired Pneumonia, Smith M.D., Fee C., Mace S.E., et al. Clinical policy: critical issues in the management of adult patients presenting to the emergency department with community-acquired pneumonia. Ann Emerg Med. 2021;77(1):e1–e57. doi: 10.1016/j.annemergmed.2020.10.024. [DOI] [PubMed] [Google Scholar]
  • 13.Portero de la Cruz S.P., Cebrino J. Prevalence and determinants of antibiotic consumption in the elderly during 2006–2017. Int J Environ Res Public Health. 2020;17(9):3243. doi: 10.3390/ijerph17093243. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Fleming-Dutra K.E., Hersh A.L., Shapiro D.J., et al. Prevalence of inappropriate antibiotic prescriptions among US ambulatory care visits, 2010-2011. JAMA. 2016;315(17):1864–1873. doi: 10.1001/jama.2016.4151. [DOI] [PubMed] [Google Scholar]
  • 15.Pulia M.S., Keller S.C., Crnich C.J., Jump R.L.P., Yoshikawa T.T. Antibiotic stewardship for older adults in ambulatory care settings: addressing an unmet challenge. J Am Geriatr Soc. 2020;68(2):244–249. doi: 10.1111/jgs.16256. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 16.Giarratano A., Green S.E., Nicolau D.P. Review of antimicrobial use and considerations in the elderly population. Clin Interv Aging. 2018;13:657–667. doi: 10.2147/CIA.S133640. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.CDC C. diff: Facts for Clinicians. August 18, 2025. https://www.cdc.gov/c-diff/hcp/clinical-overview/index.html
  • 18.Caterino J.M., Leininger R., Kline D.M., et al. Accuracy of current diagnostic criteria for acute bacterial infection in older adults in the emergency department. J Am Geriatr Soc. 2017;65(8):1802–1809. doi: 10.1111/jgs.14912. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Oved K., Cohen A., Boico O., et al. A novel host-proteome signature for distinguishing between acute bacterial and viral infections. PLoS One. 2015;10(3) doi: 10.1371/journal.pone.0120012. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Srugo I., Klein A., Stein M., et al. Validation of a novel assay to distinguish bacterial and viral infections. Pediatrics. 2017;140(4) doi: 10.1542/peds.2016-3453. [DOI] [PubMed] [Google Scholar]
  • 21.Papan C., Argentiero A., Porwoll M., et al. A host signature based on TRAIL, IP-10, and CRP for reducing antibiotic overuse in children by differentiating bacterial from viral infections: a prospective, multicentre cohort study. Clin Microbiol Infect. 2022;28(5):723–730. doi: 10.1016/j.cmi.2021.10.019. [DOI] [PubMed] [Google Scholar]
  • 22.van Houten C.B., de Groot J.A.H., Klein A., et al. A host-protein based assay to differentiate between bacterial and viral infections in preschool children (OPPORTUNITY): a double-blind, multicentre, validation study. Lancet Infect Dis. 2017;17(4):431–440. doi: 10.1016/S1473-3099(16)30519-9. [DOI] [PubMed] [Google Scholar]
  • 23.Mor M., Paz M., Amir L., et al. Bacterial vs viral etiology of fever: a prospective study of a host score for supporting etiologic accuracy of emergency department physicians. PLOS One. 2023;18(1) doi: 10.1371/journal.pone.0281018. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 24.Halabi S., Shiber S., Paz M., et al. Host test based on tumor necrosis factor-related apoptosis-inducing ligand, interferon gamma-induced protein-10 and C-reactive protein for differentiating bacterial and viral respiratory tract infections in adults: diagnostic accuracy study. Clin Microbiol Infect. 2023;29(9):1159–1165. doi: 10.1016/j.cmi.2023.05.033. [DOI] [PubMed] [Google Scholar]
  • 25.Bachur R.G., Kaplan S.L., Arias C.A., et al. A rapid host–protein test for differentiating bacterial from viral infection: Apollo diagnostic accuracy study. J Am Coll Emerg Physicians Open. 2024;5(3) doi: 10.1002/emp2.13167. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Klein A, Shapira M, Lipman-Arens S, et al. Diagnostic accuracy of a real-time host-protein test for infection. Pediatrics. 152(6):e2022060441. https://doi.org/10.1542/peds.2022-060441. [DOI] [PubMed]
  • 27.Hainrichson M., Avni N., Eden E., et al. A point-of-need platform for rapid measurement of a host-protein score that differentiates bacterial from viral infection: analytical evaluation. Clin Biochem. 2023;117:39–47. doi: 10.1016/j.clinbiochem.2022.04.012. [DOI] [PubMed] [Google Scholar]
  • 28.Allen C., Deanehan J.K., Dotan Y., et al. Development of a reference standard to assign bacterial versus viral infection etiology using an all-inclusive methodology for comparison of novel diagnostic tool performance. Clin Infect Dis. 2025:30;80(4):735–743. doi: 10.1093/cid/ciae656. 30. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Reitsma J.B., Glas A.S., Rutjes A.W.S., Scholten R.J.P.M., Bossuyt P.M., Zwinderman A.H. Bivariate analysis of sensitivity and specificity produces informative summary measures in diagnostic reviews. J Clin Epidemiol. 2005;58(10):982–990. doi: 10.1016/j.jclinepi.2005.02.022. [DOI] [PubMed] [Google Scholar]
  • 30.Furuya-Kanamori L., Kostoulas P., Doi S.A.R. A new method for synthesizing test accuracy data outperformed the bivariate method. J Clin Epidemiol. 2021;132:51–58. doi: 10.1016/j.jclinepi.2020.12.015. [DOI] [PubMed] [Google Scholar]
  • 31.OECD Elderly Population. https://www.oecd-ilibrary.org/social-issues-migration-health/elderly-population/indicator/english_8d805ea1-en
  • 32.NHE Fact Sheet | CMS. https://www.cms.gov/data-research/statistics-trends-and-reports/national-health-expenditure-data/nhe-fact-sheet
  • 33.Rodríguez-Villodres Á., Martín-Gandul C., Peñalva G., et al. Prevalence and risk factors for multidrug-resistant organisms colonization in long-term care facilities around the world: a review. Antibiotics (Basel) 2021;10(6):680. doi: 10.3390/antibiotics10060680. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 34.Lee K.A., Flores R.R., Jang I.H., Saathoff A., Robbins P.D. Immune senescence, immunosenescence and aging. Front Aging. 2022;3 doi: 10.3389/fragi.2022.900028. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Ibarz M., Haas L.E.M., Ceccato A., Artigas A. The critically ill older patient with sepsis: a narrative review. Ann Intensive Care. 2024;14(1):6. doi: 10.1186/s13613-023-01233-7. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Falcone M., Bauer M., Ferrer R., et al. Biomarkers for risk stratification and antibiotic stewardship in elderly patients. Aging Clin Exp Res. 2023;35(5):925–935. doi: 10.1007/s40520-023-02388-w. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 37.Froom P., Shimoni Z. The uncertainties of the diagnosis and treatment of a suspected urinary tract infection in elderly hospitalized patients. Expert Rev Anti Infect Ther. 2018;16(10):763–770. doi: 10.1080/14787210.2018.1523006. [DOI] [PubMed] [Google Scholar]
  • 38.Schneider J.E., Cooper J.T. Cost impact analysis of novel host-response diagnostic for patients with community-acquired pneumonia in the emergency department. J Med Econ. 2022;25(1):138–151. doi: 10.1080/13696998.2022.2026686. [DOI] [PubMed] [Google Scholar]
  • 39.Houshmand H., Porta C., Pradelli L., Pinciroli M., Sotgiu G. Cost-impact analysis of a novel diagnostic test to assess community-acquired pneumonia etiology in the emergency department setting: a multi-country European study. Int J Environ Res Public Health. 2023;20(5):3853. doi: 10.3390/ijerph20053853. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Humphries B., Sun Y., Pernica J., Xie F. Model-based cost-impact analysis of a diagnostic test for patients with community-acquired pneumonia in Canada. BMC Infect Dis. 2025;25(1):305. doi: 10.1186/s12879-025-10608-z. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 41.Gregg E., Graziadio S., Green W., et al. Host-response testing with MeMed BV in community-acquired pneumonia: an economic evaluation from the UK NHS perspective. JAC Antimicrob Resist. 2025;7(1) doi: 10.1093/jacamr/dlaf016. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Material
mmc1.docx (429.6KB, docx)

Articles from Journal of the American College of Emergency Physicians Open are provided here courtesy of American College of Emergency Physicians

RESOURCES