Viral respiratory infections are common in children, and practice guidelines do not recommend routine testing for typical viral illnesses. Despite results often not impacting care, nasopharyngeal swabs for viral testing are frequently performed and are an uncomfortable procedure. The aim of this initiative was to decrease unnecessary respiratory viral testing (RVT) in the emergency department (ED) and the pediatric medicine wards (PMWs) by 50% and 25%, respectively, over 36 months.
An expert panel reviewed published guidelines and appropriate evidence to formulate an RVT pathway using plan-do-study-act cycles. A multifaceted improvement strategy was developed that included implementing 2 newer, more effective tests when testing was deemed necessary; electronic order modifications with force functions; audit and feedback; and education. By using statistical process control charts, the outcomes analyzed were the percentage of RVT ordered in the ED and the rate of RVT ordered on the PMWs. Balancing measures included return visits leading to admission and inpatient viral nosocomial outbreaks.
The RVT rate decreased from a mean of 3.0% to 0.5% of ED visits and from 44.3 to 30.1 per 1000 patient days on the PMWs and was sustained throughout the study. Even when accounting for the new rapid influenza test available in the ED, a 50% decrease in overall ED RVT was still achieved without any significant impact on return visits leading to admission or inpatient nosocomial infections.
Through implementation of a standardized, electronically integrated RVT pathway, a decrease in unnecessary RVT was successfully achieved. Audit and feedback, reminders, and biannual education all supported long-term sustainability of this initiative.
Viral respiratory tract infections (VRTIs), such as the common cold, bronchiolitis, and influenza, routinely impact children’s health and are a frequent reason for seeking medical care. The burden of VRTIs leads to significant use of health care resources internationally.1–6 Although the diagnosis of a VRTI is generally made clinically and usually does not require confirmatory viral testing through collection of a nasopharyngeal swab for diagnosis, this test is commonly performed, even in previously healthy children. In some outpatient settings, such as emergency departments (EDs), there may not be a reliable system in place to notify patients and families of the results, which often are only reported after the child has been discharged from the hospital.
The results of respiratory viral testing (RVT) generally have little impact on patient management for typical viruses,2 except for specific instances, such as at-risk children needing influenza treatment or in pandemics (eg, coronavirus disease 2019), in which a public health need for infection control in the community supersedes decisions surrounding individual testing. There are many guidelines, including those from the American Academy of Pediatrics and the Canadian Paediatric Society, promoting evidenced-based best practice recommendations for specific viral illnesses, such as bronchiolitis, that do not recommend routine RVT as part of management.7,8 Additionally, because children often have asymptomatic colonization and postviral shedding, a positive test result for a respiratory virus does not always provide evidence for an etiologic role or clinical relevance. Up to 27% of children <6 years of age and 44% of children <1 year of age test positive for a respiratory virus without any respiratory symptoms or fever.9 Furthermore, bacterial coinfection is still possible when a child has a positive RVT result, potentially leading to false reassurance and premature closure of a viral diagnosis when there is a secondary bacterial infection. The process for performing nasopharyngeal sampling is also uncomfortable for children and can contribute to added stress and anxiety for the patient, family, and health care provider, in addition to the required point-of-care and laboratory resources. Following best practice recommendations, published quality benchmarks from the best-performing hospitals demonstrate RVT rates as low as 0.6% for bronchiolitis and 1.5% for pneumonia without negatively impacting clinical outcomes.10
Despite ample evidence, these best practices are not always integrated into the clinical environment. In 2014, a baseline analysis of our hospital’s practice revealed that ∼6000 nasopharyngeal swabs were ordered for RVT, with 60% of those swabs from the ED or pediatric medicine wards (PMWs), ∼25% from the PICU, and the remaining 15% from other wards or outpatient clinics. Of the 2500 swabs ordered from the ED, ∼63% were performed on children who were discharged from the hospital from the ED. Because the test performed at that time had a turnaround time of 24 hours, the results were not available at the time of discharge, and unlike bacterial culture results, there was not a formal follow-up process in the ED for outstanding viral tests. Therefore, families and the primary care providers were not provided with the results by the ED or from the laboratory. Given the prolonged turnaround time and lack of reliable follow-up of the RVT, high-risk patients with positive results for the influenza virus were not prescribed oseltamivir, despite recommendations from authoritative national organizations.11,12
The aim of this study was to improve value by decreasing unnecessary nasopharyngeal RVT in the ED by 50% and in the PMWs by 25% at a large children’s hospital and sustain these results over 3 years.
Methods
Study Design and Setting
This quality improvement (QI) study was conducted over a 5-year period from January 2014 to December 2018 at The Hospital for Sick Children in Toronto, Ontario, Canada. The Hospital for Sick Children is an urban, 370-bed quaternary care pediatric academic hospital caring for patients from birth to age 18, with an annual ED volume of ∼80 000 patients. The study received approval as a QI study by our institutional quality risk management team and met requirements for research ethics board exemption.
QI Approach
By using the model for improvement methodology,13 a multifaceted QI strategy was used to address the quality gap of unnecessary RVT. First, a team of key hospital stakeholders was assembled, including clinicians from emergency medicine, pediatric medicine, microbiology, infectious disease, and infection prevention and control (IPAC). The expert panel reviewed the evidence to determine indications for RVT for all common pediatric VRTIs, including influenza. Baseline RVT data from the previous 2 years was obtained to better understand expected seasonal and annual fluctuations in testing practices. Root cause analysis with a cause and effect diagram was used to understand the various contributors to unnecessary RVT in our setting and helped inform development of our interventions. The identified barriers included preexisting practice patterns, diagnostic uncertainty associated with febrile illnesses in children, a reluctance to prescribe oseltamivir without a confirmed influenza diagnosis, clinician workload, and patient and family desire for a definitive diagnosis for their symptoms.2
Interventions
New Technology
At baseline, all nasopharyngeal swabs (flocked swabs; COPAN Diagnostics, Murrieta, CA) were processed by using the direct fluorescent antibody (DFA) test, which detects 8 common respiratory viruses with results generally available within 24 hours (Diagnostic HYBRIDS, Athens, OH). Given advances in technology, 2 new tests were introduced in January 2016 to replace the older DFA test because their testing characteristics appeared to be more effective for patient care and had also been associated with some reduction in antibiotic and radiography use2,14 First, the respiratory multiplex polymerase chain reaction test (RMPCR) (NexTAG Respiratory Pathogen Panel; Luminex Molecular Diagnostics, Toronto, Ontario, Canada) expanded detection to up to 15 respiratory viruses with a turnaround time of ∼24 hours. Second, a point-of-care rapid influenza isothermal amplification test (Alere i Influenza A & B; Alere Inc [now Abbott], Waltham, MA) provided influenza A and B results within 15 minutes with improved sensitivity and specificity compared with traditional rapid antigen direct tests. The point-of-care rapid influenza test (FLU) was made available in the ED annually during influenza season (generally December through April).
Respiratory Virus Pathway Development
On the basis of consensus from the literature review, an RVT pathway was created by using plan-do-study-act cycles for both the ED and PMWs outlining evidence-based indications for ordering the viral tests (Table 1, see also Supplemental Fig 6 for full pathway). Plan-do-study-act cycles helped further refine the pathways through qualitative feedback obtained from the identified stakeholders, then at divisional staff meeting presentations, and finally, through trialing draft versions of the pathways in the clinical environment for a few shifts. For the ED, the RMPCR was limited to specific patient populations, such as children with high-risk comorbidities being discharged from the hospital in whom follow-up was secured to review the outstanding test results. For ED patients requiring hospitalization, ordering of RMPCR was deferred to the admitting team to decide whether the results would impact management on the basis of testing indications (Table 1). FLU was only routinely available in the ED and was restricted to high-risk patients in whom guidelines advised timely antiviral therapy11,12 or patients with prolonged fever symptoms in whom a positive test result would preclude the need for additional testing or treatments, such as bloodwork, radiologic investigations, or antibiotics.
RVT Indications for ED and General Pediatric Inpatients
| Rapid Influenza A and B Testinga (Routinely Available in ED) | RMPCRb (Routinely Available in ED and PMWs) |
|---|---|
| Immunocompromised status | Sickle cell disease with fever |
| Malignancy | Oncologic diagnosis with fever |
| Respiratory disease (including asthma, as clinically indicated) | Immunosuppression or neutropenia with fever |
| Cardiovascular disease | Febrile infants <90 d of age |
| Neurodevelopmental condition and poor handling of secretions | |
| Hemoglobinopathies (ie, sickle cell disease) | |
| Metabolic disease or diabetes | |
| Chronic renal disease | |
| Chronic liver disease | |
| Rheumatologic disease | |
| Indigenous persons or living in group home | |
| Aged 12–24 mo with severe disease | |
| Prolonged fever and influenza symptoms for which influenza diagnosis will likely change management (ie, bloodwork, disposition, or antibiotics) |
| Rapid Influenza A and B Testinga (Routinely Available in ED) | RMPCRb (Routinely Available in ED and PMWs) |
|---|---|
| Immunocompromised status | Sickle cell disease with fever |
| Malignancy | Oncologic diagnosis with fever |
| Respiratory disease (including asthma, as clinically indicated) | Immunosuppression or neutropenia with fever |
| Cardiovascular disease | Febrile infants <90 d of age |
| Neurodevelopmental condition and poor handling of secretions | |
| Hemoglobinopathies (ie, sickle cell disease) | |
| Metabolic disease or diabetes | |
| Chronic renal disease | |
| Chronic liver disease | |
| Rheumatologic disease | |
| Indigenous persons or living in group home | |
| Aged 12–24 mo with severe disease | |
| Prolonged fever and influenza symptoms for which influenza diagnosis will likely change management (ie, bloodwork, disposition, or antibiotics) |
Patients with influenzalike and fever symptoms along with the following risk factor(s).
Patients with respiratory symptoms along with the following risk factor(s).
Electronic Medical Record Modifications
To better integrate the RVT pathway into clinical care, force functions were added in the electronic medical record (EMR), requiring clinicians to select an appropriate indication for testing from the available options (Table 1). The hospital EMR was Sunrise Clinical Manager (Allscripts, Richmond, British Columbia, Canada) until June 2018, when the EMR transitioned to Epic (Epic Systems Corporation, Verona, WI). Additionally, a hard stop was implemented so that FLU could only be ordered on the PMWs with microbiologist approval because the majority of FLUs would already have been completed in the ED. Lastly, all applicable divisional order sets, such as the bronchiolitis order set, were reviewed and modified to ensure that they did not include routine RVT. All EMR modifications were implemented in February 2016.
Education and Strategic Alignment
A multipronged educational campaign was implemented at the beginning of the QI initiative and sustained on a biannual basis to coincide with the beginning and peak of each viral respiratory season. Divisional educational rounds and business meetings in the fall and midwinter months were used to remind faculty and trainees of the initiative, to address barriers, and to share improvement results. The project was also launched as 1 of 5 initiatives introduced in the hospital’s new Choosing Wisely campaign. Reminders of the initiative and the campaign were regularly posted on the hospital’s intranet site and delivered in group e-mails.
Audit and Feedback
After the first year of the project, ED physicians were provided individualized data on their RVT ordering patterns together with anonymized aggregated data for all ED physicians for peer comparison (Fig 1). Audit and feedback could not be implemented for inpatient faculty because of EMR limitations in determining accurate ordering attribution among various physicians on the wards.
Measures and Analysis
The primary outcome measures were the percentage of RVT ordered in the ED and the rate of RVT ordered on the PMWs. Two ED tests were measured. First, the percentage of monthly ED patient visits with a laboratory-based test (new RMPCR) ordered was compared with that for the historical DFA test. To ensure the old DFA test was not simply being replaced by the new FLU, the second measurement was the total ED RVT (FLU plus RMPCR) per monthly ED patient volumes. For the PMWs, only 1 test was measured because FLU was not readily available on the PMWs. The rate of monthly RVT was calculated by measuring the total RMPCRs ordered compared with the historical DFA tests ordered per 1000 inpatient days. A RVT could be ordered more than once during a hospitalization, as clinically indicated.
As a process measure, the monthly hospital oseltamivir days of therapy (DOT) per 1000 inpatient days was selected to assess adherence to the RVT pathway because the pathway emphasized testing indications for timely antiviral therapy for high-risk patients with influenza. For balancing measures, we monitored viral nosocomial infections and outbreaks on the PMWs and the percentage of return visits leading to admission (RVs) within 72 hours of the initial ED visits. The hospital’s IPAC team tracked nosocomial infections and potential outbreaks and was a stakeholder in the QI project.
By using statistical process control (SPC) methods, the primary outcome measures were tracked retrospectively from January 2014 to December 2015 and prospectively from January 2016 to December 2018. Given seasonal variation, 2 years of retrospective data were used to establish baseline testing patterns. All data were extracted from the EMR, and SPC charts were created by using QI Macros (QI Macros for Excel, version 2015; KnowWare International, Inc, Denver, CO). Given the large and varying size of the monthly ED volumes, Prime SPC charts were selected to display the percentage of RVT. Special cause variation was monitored by using accepted SPC chart methods.15
Results
All outcome measures achieved special cause variation and showed sustained improvement during the study period. At baseline, the percentage of testing for respiratory viruses (DFA) was 3.0% of 139 463 ED visits over a 24-month period (January 2014 to December 2015). Post implementation in January 2016, testing decreased to 0.5% of ED visits (83.3% absolute reduction) with the RMPCR and was sustained for 3 years (Fig 2A). Special cause variation was detected with the official launch of the RVT pathway in January 2016. When factoring in the new FLU, total ED RVT (RMPCR and FLU) still decreased to 1.5% of patient visits (50.0% absolute reduction) (Fig 2B). For total ED RVT, special cause variation was detected in April 2016 and annually in the winter months because of expected seasonal variation (Fig 2B).
P-prime SPC charts for the ED. A, Change in percentage of ED patients who received a comprehensive RVT (DFA used in 2014–2015 versus multiplex polymerase chain reaction used in 2016–2018). B, Change in percentage of ED patients who received any RVT (multiplex + rapid influenza). Annotations indicate intervention components. The lighter curved lines indicate the upper control limit (UCL) and the lower control limit (LCL). The dark straight line indicates the center line (CL) (mean).
P-prime SPC charts for the ED. A, Change in percentage of ED patients who received a comprehensive RVT (DFA used in 2014–2015 versus multiplex polymerase chain reaction used in 2016–2018). B, Change in percentage of ED patients who received any RVT (multiplex + rapid influenza). Annotations indicate intervention components. The lighter curved lines indicate the upper control limit (UCL) and the lower control limit (LCL). The dark straight line indicates the center line (CL) (mean).
For the PMWs, the baseline RVT rate over 24 months was 44.3 per 1000 inpatient days. Post implementation, the testing rate decreased to an average of 30.1 per 1000 inpatient days (32.1% absolute reduction) and was similarly sustained (Fig 3). The SPC chart reveals a shift in the centerline in February 2016, which coincided with the EMR modifications after the RVT pathway launch.
U SPC chart for PMWs: change in rate of PMW patients who received a comprehensive RVT (DFA used in 2014–2015 versus multiplex polymerase chain reaction used in 2016–2018). Annotations indicate intervention components. The lighter curved lines indicate the upper control limit (UCL) and the lower control limit (LCL). The dark straight line indicates the center line (CL) (mean).
U SPC chart for PMWs: change in rate of PMW patients who received a comprehensive RVT (DFA used in 2014–2015 versus multiplex polymerase chain reaction used in 2016–2018). Annotations indicate intervention components. The lighter curved lines indicate the upper control limit (UCL) and the lower control limit (LCL). The dark straight line indicates the center line (CL) (mean).
At baseline, oseltamivir was less often prescribed in 2 influenza seasons, with a mean of 4.8 DOT per 1000 monthly patient days. Special cause variation was detected January 2016 after the launch of the QI interventions, with increased appropriate oseltamivir use to a mean of 18.7 DOT per 1000 patient days (74.3% absolute increase) (Fig 4), and was sustained for most of the study. The final month (December 2018) of oseltamivir data was not retrievable with migration to the new hospital EMR.
U SPC chart displaying monthly inpatient oseltamivir DOT per 1000 patient days. Annotations indicate intervention components. The lighter curved lines indicate the upper control limit (UCL) and the lower control limit (LCL). The dark straight line indicates the center line (CL) (mean).
U SPC chart displaying monthly inpatient oseltamivir DOT per 1000 patient days. Annotations indicate intervention components. The lighter curved lines indicate the upper control limit (UCL) and the lower control limit (LCL). The dark straight line indicates the center line (CL) (mean).
During the study period, no nosocomial viral outbreaks on the PMWs or ED were reported by IPAC. Additionally, special cause variation was not detected in the percentage of ED RVs (Fig 5).
Balancing measure: P SPC chart displaying monthly percentage of ED RVs within 72 hours of the initial visit. The lighter curved lines indicate the upper control limit (UCL) and the lower control limit (LCL). The dark straight line indicates the center line (CL) (mean).
Balancing measure: P SPC chart displaying monthly percentage of ED RVs within 72 hours of the initial visit. The lighter curved lines indicate the upper control limit (UCL) and the lower control limit (LCL). The dark straight line indicates the center line (CL) (mean).
Discussion
We present a QI study that has established a large scale, sustained reduction in RVT in a quaternary care children’s hospital, regardless of underlying diagnosis. RVT was successfully decreased beyond our initial aim and was sustained by using improvement methodology with a multifaceted intervention strategy. Additionally, oseltamivir use for high-risk inpatients with an influenza diagnosis increased, demonstrating clinician adherence to our intervention of an RVT pathway that outlined best practice standards for influenza management.7,8 With an estimated 1200 fewer nasopharyngeal swab tests annually, children avoided the discomfort and stress of a test that would likely add little to no value,16 and laboratory and clinical resources were freed up for more appropriate uses. Recognizing the finite resources of health care systems, high-value care requires allocating resources toward tests and treatments that directly impact decisions and ultimately support better patient outcomes.17,18 All of the outcomes in this study enhanced the value of care provided to children with VRTIs without unintended consequences on ED RV rates or nosocomial outbreaks.
Recognizing the barriers to RVT stewardship uncovered with the cause and effect diagram, we were able to develop mitigation strategies through targeted interventions, which were essential to improvement success. Our adoption of 2 newer, more effective RVTs that benefit select patients addressed several barriers, specifically diagnostic uncertainty. The ability to rapidly diagnose influenza using the newer, highly sensitive molecular tests can influence treatment decisions and ensure timely oseltamivir treatment and may even reduce unnecessary antibiotic prescriptions and return visits.19 By implementing an effective rapid test in the ED, clinicians could more easily decrease RMPCR testing because delayed results often did not impact ED decision-making.
New technology alone, however, did not explain the entire shift in total ED and PMW RVT. Multifaceted interventions that include system-based process changes are generally known to be more effective than single-component, person-based interventions.18 The new RVT pathway outlined the diagnostic indications for the 2 respiratory viral tests, such as febrile patients with immunosuppression in whom an RMPCR test identifying a respiratory virus could lead to discontinuation of broad-spectrum antibiotics with negative blood culture results,20 and included appropriate oseltamivir use recommendations. Once integrated into the EMR order, the pathway essentially promoted knowledge translation and made it easier for clinicians to do “the right thing” because it aligned with workflows, ensured that testing recommendations would be reviewed with the force functions, and helped mitigate the impact of staff and trainee turnover. Integrated bedside tools, such as order sets and algorithms that highlight best practice recommendations, educate and can also change ingrained practices.21,22 Lastly, audit and feedback, a well-studied improvement strategy and a powerful driver of individual practice change,23 identified clinicians one year into the study who were high users of RVT and were outliers when compared with their peers. Audit and feedback, along with ongoing education and reminders, supported the sustainability of the achieved improvements.
There were several other factors that contributed to this initiative’s success, starting with the project team. An engaged clinician champion who was motivated to lead this work was an important enabler, as was identification of key stakeholders from across the organization to involve early in project planning. Recognizing that patients often flow through several services of the hospital, it was critical to develop synergy on RVT strategies among stakeholders so that a reduction in ED testing would not inadvertently lead to increased inpatient or infectious disease consultant testing requests. Additionally, a strong working relationship between frontline clinicians and clinicians in microbiology allowed for ongoing dialogue on the best, most up-to-date testing modalities that support high-value care. Lastly, given the seasonal pattern of respiratory viruses, as demonstrated in the results in the winter months, detailed sustainability planning was essential to ensure ordering behaviors did not revert to older practices patterns over time. This was achieved through regular biannual educational rounds, business meeting updates, and e-mail reminders.
This study does have important limitations. First, the appropriateness of RVT is impacted by several factors, including the seasonality of viruses, local context, and, most recently, a pandemic that requires a much greater need to identify and isolate patients with specific viruses for infection control. Additionally, because many interventions were implemented initially over a short time frame, it is difficult to determine with certainty which intervention(s) had the greatest impact. A formal economic analysis was also not part of this study design, so the true cost impact of these testing changes is not yet determined. Lastly, the balancing measures for monitoring unintended consequences did have limitations. Monitoring of outbreaks relied on individual reporting and testing, so smaller viral spreads on individual hospital units might not have been identified. For the RVs measure, visits were only captured for patients who returned to our hospital ED; patients could have returned to outside hospitals, but our hospital is the only children’s hospital in the city.
Conclusions
A multifaceted iterative QI strategy with electronic integration of best practice recommendations for RVT is an effective method for successfully reducing RVT in pediatric patients in the ED and PMWs. Reducing unnecessary routine RVT promotes high-value care and allows for more effective allocation of resources without any clear unintended consequences. Clinicians should order RVT on children not because they can, but because it will add value to the patient’s care. This study demonstrates the value of using improvement methodology to drive change and ultimately deliver high-value care.
Dr Ostrow conceptualized and designed the study, collected data, conducted the analyses, drafted the initial manuscript, and reviewed and revised the manuscript; Drs Savlov and Richardson conceptualized and designed the study, collected data, and reviewed and revised the manuscript; Dr Friedman conceptualized and designed the study, helped draft the initial manuscript, and reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.
FUNDING: No external funding.
- DFA
direct fluorescent antibody
- DOT
days of therapy
- ED
emergency department
- EMR
electronic medical record
- FLU
point-of-care rapid influenza test
- IPAC
infection prevention and control
- PMW
pediatric medicine ward
- QI
quality improvement
- RMPCR
respiratory multiplex polymerase chain reaction test
- RV
return visit leading to admission
- RVT
respiratory viral testing or tests
- SPC
statistical process control
- VRTI
viral respiratory tract infections
Comments