Clinical Yield of Multiple Testing with Respiratory Pathogen Panels

Abstract

Though multiplex respiratory pathogen panels (RPP) have high sensitivity, multiple tests are occasionally performed simultaneously or in rapid succession in an attempt to increase the yield. The purpose of this study was to assess the impact of this practice.

Methods:

“Multiple testing” was defined as >1 RPP performed within 12 hours on the same patient and specimen type. All cases of multiple testing for adults at two hospitals over a 5-year period were included. Chart review was performed to determine whether discordant results led to a clinical diagnosis or change in clinical management.

Results:

Of 18,779 RPPs, 462 (2.5%) represented cases of multiple testing. 26 of 462 cases (5.6%) produced discordant results. 5 discordant results (1.1% of 462 multiple testing episodes) were associated with a clinical diagnosis, and 4 (0.9%) influenced clinical management.

Conclusion:

Multiple RPP testing facilitates clinical management in <1% of cases. Medical centers may consider de-implementing this practice.

Background

Respiratory pathogen panels (RPP, also known as respiratory virus panels or RVPs) use multiplex PCR to test a specimen for up to 20–25 pathogens simultaneously. In general, RPPs have excellent performance characteristics, including high sensitivity and specificity[1, 2] RPPs also tend to be remarkably consistent, with >99% concordance between duplicate tests performed for validation on the same specimen and 95% concordance between tests performed on bronchoalveolar lavage (BAL) fluid and bronchial washings collected during the same procedure.[3–5]

Nonetheless, despite the high sensitivity provided by a single RPP, clinicians may choose to submit multiple specimens for testing in order to increase the yield. For instance, multiple testing is a common feature of many hospital-based Coronavirus disease 2019 (COVID-19) testing algorithms.[6, 7] However, the value of multiple testing is unproven, particularly for non-COVID-19 respiratory viruses. The purpose of this study was to evaluate the yield of multiple RPP testing and the impact of multiple testing on diagnosis and clinical management.

Methods

We conducted a retrospective study of all adult patients who underwent RPP testing in acute care at two hospitals in Baltimore (including a tertiary care center and a community hospital) between November 1, 2015 and December 31, 2020. Structured electronic health record data was pulled from the University of Maryland Medical System Epic Clarity Database (Epic) and Clinical Data Repository (CDR). Epic and the CDR are Oracle relational databases residing on secure servers. Data were extracted using Structured Query Language, Python, and linux bash scripts. This study was approved by the Institutional Review Board of the University of Maryland, Baltimore.

The RPP in use during this period was the FilmArray respiratory panel (BioFire Diagnostics, Salt Lake City, UT, used per manufacturer’s instruction), which detects adenovirus, seasonal coronaviruses (HKU1, NL63, 229E, OC43), human metapneumovirus, rhinovirus/enterovirus, influenza (A, A/H1, A/H3, A/H1N1–2009, and B), parainfluenza types 1–4, respiratory syncytial virus (RSV), Bordetella pertussis and B. parapertussis, Chlamydia pneumoniae, and Mycoplasma pneumoniae. Tests performed on bronchial wash specimens were excluded, since the FilmArray respiratory panel has not been validated for use with this specimen type.

Multiple testing was defined as two or more RPPs performed within a 12-hour period for one patient on the same specimen type. The 12-hour period was selected to capture same-day testing that was performed without significant changes in the patient’s clinical condition. Multiple tests were categorized as simultaneous if collected at the same moment in time as the index test and near-simultaneous if collected after the index test but within 4 hours.[4]

Chart review was performed to determine whether discordant results facilitated clinical diagnosis or changes in clinical management. RPP testing was considered associated with a clinical diagnosis if the pathogen detected was referenced as the cause of symptoms, pathology, or disease. In cases where infection due to a respiratory pathogen was already known, persistent positivity by the same pathogen was not counted as facilitating clinical diagnosis. Statistical comparisons were performed using chi-square testing for 2×2 tables.

Results

From 18,569 RPPs collected and resulted during the study period, 462 (2.5%) represented cases of multiple testing. 84% of multiple tests were performed on BAL specimens, including 388 of all 5,685 (6.8%) RPPs performed on BAL fluid. The remaining 74 multiple tests were performed on nasopharyngeal specimens. 246 (53% of 462) cases of multiple testing were simultaneous, 140 (30%) were near-simultaneous, and 76 (16%) were performed >4 hours after the index test.

Clinical impact of multiple testing is summarized in Table 1. Overall, 26 of 462 (5.6%) multiple tests produced discordant results compared to results from initial testing. The most common discordant pairs were a positive result for rhinovirus/enterovirus along with a negative result (9 of 26 discordant pairs, 35%). Rates of discordant results were 5.3%, 6.4%, and 5.3% for simultaneous tests, near-simultaneous tests, and tests performed after 4–12 hours, respectively (p = 0.94 by chi-square test).

Table 1.

Discordant results from episodes of multiple testing ranked by time between index result and follow-up results. Specimen source types and impact of discordant results on diagnosis and management are described.

Respiratory Virus Panel Results			Source	Clinical Impact
Index	Follow-up	Time
Adenovirus, human metapneumovirus	Adenovirus	0 min.	BAL	Symptomatic with diarrhea and rhinorrhea of unknown duration. Diagnosed with “minimally symptomatic” adenovirus. Positive result for human metapneumovirus not acknowledged. Immunosuppression for history of lung transplant adjusted.
Coronavirus 229E	Negative	0 min.	BAL	Intermittently febrile during prolonged hospitalization for bilateral lung transplant, with most recent febrile episode beginning 5 days prior to testing. +Coronavirus 229E disregarded by the care team. Positive for the same pathogen twice in the previous 30 days.
Coronavirus 229E, B. parapertussis	Negative	0 min.	BAL	Surveillance bronchoscopy.† Both positive results disregarded by the care team. Developed transient bradycardia and hypotension after the procedure.
Coronavirus 229E	Negative	0 min.	BAL	Surveillance bronchoscopy.† Positive result disregarded by the care team. Previously positive for Coronavirus 229E one month earlier.
Influenza A	Equivocal for Influenza A	0 min.	BAL	Bronchoscopy performed to evaluate worsening chest imaging in setting of known pneumonia. Diagnosed with influenza pneumonia. Had already completed antiviral treatment, no changes made to clinical management.
Influenza B	Negative	0 min.	BAL	Presented 1 week prior to testing with respiratory failure and was unable to provide a history. Previously diagnosed with influenza B. Persistent positive noted as evidence of stable infection. Antibiotics stopped.
Parainfluenza 3	Negative	0 min.	BAL	Symptomatic with fever, cough and diarrhea for 2–3 days. Diagnosed with Parainfluenza infection. Antibiotics stopped.
Rhinovirus*	Negative	0 min.	BAL	Surveillance bronchoscopy.† Positive result disregarded by the care team.
Rhinovirus*	Negative	0 min.	BAL	Surveillance bronchoscopy.† Positive result disregarded by the care team.
Rhinovirus*	Negative	0 min.	BAL	Surveillance bronchoscopy.† Positive result disregarded by the care team. Subsequently diagnosed with and treated for invasive fungal pneumonia.
Rhinovirus*	Negative	0 min.	BAL	Surveillance bronchoscopy.† Positive result disregarded by the care team. Diagnosed with pseudomonal pneumonia.
RSV	Negative	0 min.	BAL	Symptomatic with isolated fever 3 days. Diagnosed with RSV infection, but not treated. Subsequently diagnosed with enterococcal bacteremia. History of lung transplant.
RSV, parainfluenza 4	Negative	0 min.	BAL	Previously diagnosed with RSV and parainfluenza by respiratory pathogen panel 6 days earlier. Bronchoscopy performed while already on antiviral treatment.
Negative	Rhinovirus, Coronavirus 229E	1 min.	BAL	Bronchoscopy performed at the time of lung transplantation. Positive results disregarded by the care team. No changes in clinical management.
Parainfluenza 4	Parainfluenza 4, Parainfluenza 3	1 min	BAL	Surveillance bronchoscopy.† Positive results acknowledged but not associated with a diagnosis. No changes in clinical management.
Negative	Coronavirus OC43	5 min.	BAL	Surveillance bronchoscopy.† Positive result disregarded by the care team.
Negative	Rhinovirus*	5 min.	BAL	Surveillance bronchoscopy.† Positive result disregarded by the care team.
Influenza A / H3	Negative	13 min.	BAL	Symptomatic for 17 days at time of testing. Previously diagnosed with influenza pneumonia. Repeat positive considered persistent shedding. No changes in clinical management.
Rhinovirus*	Negative	19 min.	BAL	Surveillance bronchoscopy.† Positive result disregarded by the care team.
RSV	Negative	23 min.	BAL	Surveillance bronchoscopy.† Positive result acknowledged but not suspected to be a cause of clinical disease.
Negative	Rhinovirus*	28 min.	BAL	Symptomatic with 3 days of dyspnea. Positive result disregarded by the care team, but oseltamivir stopped in response to initial negative. No changes to management associated with second test. Subsequently diagnosed with MRSA pneumonia.
Adenovirus	Negative	3.8 hr	NP	Symptomatic with 4 days of productive cough and hemoptysis. Positive result disregarded by primary team. No changes in clinical management. Subsequently diagnosed with recurrent malignancy
Rhinovirus*	Negative	4.3 hr	NP	Symptomatic with cough and dyspnea for less than 1 week. Diagnosed with rhinovirus infection as an underlying cause of COPD exacerbation and treated for COPD. Negative result on second test disregarded. No changes in clinical management.
Rhinovirus*	Negative	6.3 hr	NP	Symptomatic with fever for 1 day at time of testing. Diagnosed with rhinovirus as a possible cause of neutropenic fever and placed on droplet precautions. Negative result on second test disregarded. No changes in clinical management attributed to viral infection.
RSV	Negative	6.8 hr	NP	Symptomatic with cough and sore throat for 3–4 days. Discharged from the emergency department with suspected viral illness as a cause of cough. +RSV not acknowledged. No changes in clinical management.
Negative	Influenza A H1	7.5 hr	NP	Symptomatic with upper respiratory syndrome for 5 days after known influenza exposure. Diagnosed with influenza A and secondary Group A Streptococcal bacteremia. Treated with oseltamivir, baloxavir, and antibiotics.

RSV, respiratory syncytial virus; BAL, bronchoalveolar lavage; NP, nasopharyngeal; COPD, chronic obstructive pulmonary disease. Rows highlighted in grey reflect cases in which multiple testing did not contribute to either diagnosis or clinical management.

^*Rhinovirus/enterovirus

^†Surveillance bronchoscopy performed in the routine evaluation of a lung transplant patient. Symptoms were not documented at the time of bronchoscopy.

Five discordant results (1.1% of 462 multiples) were acknowledged by the care team as providing new information related to the cause of clinical disease. Clinically relevant viral pathogens identified based on discordant results from duplicate testing included influenza (3 cases), parainfluenza, and RSV. Four discordant results (0.9% of 462 multiples) led to a change in clinical management, including two instances in which antibiotics were discontinued, one instance in which combination therapy for influenza was initiated, and one instance in which immunosuppressive medications were adjusted.

Discussion

Testing a single specimen type with multiple RPPs in rapid succession is not known to provide clinical value but nonetheless occurs in real-world practice.[8] In this retrospective analysis of observational data from two hospitals, we found that multiple RPP testing infrequently provides new information and rarely facilitates clinical diagnosis or management. In other words, multiple testing increases the cost of care, occupies the time of staff in the clinical microbiology laboratory, and adds clutter to the electronic health record, but does not typically benefit the patient.

Multiple RPPs may be performed on the same specimen type from a single patient for several reasons. Clinicians may submit multiple tests because they believe initial testing is falsely negative, because specimens are collected from both sides of the body (i.e., BAL from left and right lung), or because duplicate orders are entered unintentionally, as may occur at time of hand-off from the emergency department to the inpatient team. Whether intentional or unintentional, our results demonstrate that multiple testing within a short time frame does not improve the diagnostic value of the RPP.

Multiple testing yielded discordant results in 5.8% of cases. Though the low rate of discordant results confirm the remarkable consistency of RPP testing, clinicians may be nonetheless surprised that 1 in 20 RPP results are potentially not replicable. Varying quality of specimen collection and detection of borderline or low-level quantities of microbial material are potential reasons why multiple tests from the same patient may produce different results.

This study is subject to the limitations associated with a retrospective, observational analysis. Though we had specimen type data, the laterality or specific pulmonary lobe from which the BAL fluid was collected was not reliably recorded. Chart review was only performed on patients with discordant results and was unable to determine why multiple testing was performed in most cases. In the absence of a gold standard, we used the initial RPP as the comparator and did not compare the performance of RPPs to other testing modalities.

Conclusion

Multiple RPP testing is a low-value practice that facilitated clinical diagnosis in 1.1% of cases and influenced clinical management in 0.9% of cases. Medical centers may consider implementing protocols or policies to reduce this practice.

Figure 1.

Proportion of multiple tests that produced discordant results, along with the frequency of instances in which multiple testing influenced clinical diagnosis, management, or both.