Impact of the Rapid Diagnosis of Influenza on Physician Decision-Making and Patient Management in the Pediatric Emergency Department: Results of a Randomized, Prospective, Controlled Trial
Objective. To determine the impact of the rapid diagnosis of influenza on physician decision-making and patient management, including laboratory tests and radiographs ordered, patient charges associated with these tests, antibiotics/antivirals prescribed, and length of time to patient discharge from the emergency department.
Methods. Patients aged 2 months to 21 years presenting to an urban children’s teaching hospital emergency department were screened for fever and cough, coryza, myalgias, headache, and/or malaise. After obtaining informed consent, patients were randomized to 1 of 2 groups: 1) physician receives (physician aware of) the rapid influenza test result; or 2) physician does not receive (physician unaware of) the result. For patients in the physician aware group, nasopharyngeal swabs were obtained, immediately tested with the FluOIA test for influenza A and B, and the result was placed on the chart before patient evaluation by the attending physician. For the physician unaware group, nasopharyngeal swabs were obtained, stored according to manufacturer’s directions, and tested within 24 hours. Results for the physician unaware group were not disclosed to the treating physicians at any time. The 2 resultant influenza-positive groups (aware and unaware) were compared for laboratory and radiograph studies and their associated patient charges, antibiotic/antiviral prescriptions, and length of stay in the emergency department.
Results. A total of 418 patients were enrolled, and 391 completed the study. Of these, 202 tested positive for influenza. Comparison of the 96 influenza-positive patients whose physician was aware of the result with the 106 influenza-positive patients whose physician was unaware of the result revealed significant reductions among the former group in: 1) numbers of complete blood counts, blood cultures, urinalyses, urine cultures, and chest radiographs performed; 2) charges associated with these tests; 3) antibiotics prescribed; and 4) length of stay in the emergency department. The number of influenza-positive patients who received prescriptions for antiviral drugs was significantly higher among those whose physician was aware of the result.
Conclusions. Physician awareness of a rapid diagnosis of influenza in the pediatric emergency department significantly reduced the number of laboratory tests and radiographs ordered and their associated charges, decreased antibiotic use, increased antiviral use, and decreased length of time to discharge.
Influenza virus types A and B are common respiratory pathogens in the pediatric population. Depending on age, attack rates may be 1.5 to 3 times higher than for adults, with school-aged children having the highest attack rates.1,2 A retrospective cohort study of children under 15 years of age demonstrated outpatient visits attributable to influenza ranging from 6 to 15 per 100 children.3 Infection with influenza virus leads to a significant increase in primary care visits, and also increases in emergency department utilization during wintertime epidemics.2
Rapid diagnostic test kits for influenza types A and B are currently available for outpatient use and have proven to be both sensitive and specific.4–7 Few studies have been performed which analyze the impact of rapid diagnostic testing for influenza and subsequent effect on patient management.8–11 To date, there are no prospective, randomized studies analyzing use of rapid influenza testing and effect on patient management in the pediatric emergency department. Rapid diagnostic tests are not currently routinely incorporated in the work-up of infants and children with fever and vague symptoms, or with fever and no documented source.12 Use of rapid tests in the pediatric emergency department which are sensitive and specific for influenza could potentially decrease performance of other more invasive tests, thereby reducing associated patient charges, reducing patient length of stay in the emergency department, avoiding unnecessary antibiotic usage, and increasing appropriate antiviral use.
The aim of this randomized, prospective study was to evaluate the effect of knowledge of a positive rapid influenza test on physician decision-making and patient management in the pediatric emergency department. The hypothesis was that physicians would alter their decision-making and subsequent patient management when they were aware of a positive rapid influenza test before patient evaluation. Analysis of all study participants and the 2- to 36-month aged subgroup was planned as part of the study design.
All children who presented to the Children’s Hospital of Alabama emergency department were eligible if they: 1) were between the ages of 2 months and 21 years; 2) had a temperature of 100.4°F or greater in the emergency department; 3) had cough, coryza, malaise, headache, and/or myalgias; and 4) had symptom duration of 72 hours or less. Children with immunosuppression secondary to drugs, malignancy, or HIV infection, as well as all transplant recipients, were excluded from enrollment. Additionally, children who had already been evaluated for this illness or were on antibiotics at the time of evaluation in the emergency department were ineligible for enrollment.
Following approval by the University of Alabama at Birmingham Institutional Review Board, research team members were stationed at the emergency department triage area and screened patients for eligibility criteria. If eligibility was met, the study was described to the patient and parent or legal guardian, and informed consent for participation was obtained.
Rapid diagnostic testing for influenza virus infection is not the standard of care in our emergency department for evaluation of patients with fever and flu-like symptoms.
The only option for influenza diagnosis in our institution before this study was by a viral respiratory panel that is only available Monday through Friday mornings before 11:30 am. Several weeks into this study, rapid influenza testing was made available through our main laboratory; however, lack of physician familiarity with the test and a 1- to 2-hour turnaround time precluded its routine use in our emergency department patients.
This study was undertaken to determine the impact of the rapid diagnosis of influenza on physician decision-making and patient management. Study endpoints included: 1) reduction of antibiotic prescriptions of 40%; 2) reduction of laboratory and radiograph charges of 50%; 3) decrease length of time to discharge by 1 hour; and 4) increase antiviral use by 25%. Sample size calculations were based on a 40% reduction of antibiotic prescriptions written for known influenza-positive children. A total of 340 study patients would be required to meet these objectives.
This randomized, prospective, controlled trial evaluated 2 groups of pediatric patients: group 1, patients whose rapid influenza test results were known by the treating physician; and group 2, patients whose results were not known by the treating physician. The principal investigator did not provide care for any patients enrolled in the study in an effort to reduce potential study bias. All physicians (emergency department and fast track area) participating in this study are board-certified in pediatrics. All of the emergency department attending physicians are also sub-board-certified in pediatric emergency medicine or are engaged in pediatric emergency medicine fellowship training.
For patients randomized to group 1, nasopharyngeal swabs were obtained and immediately tested by study personnel using rapid diagnostic testing for influenza A and B. Test results for this group were available within 20 to 25 minutes and were placed on the patient’s chart before evaluation by the attending physician. Specimens obtained from patients in group 2 were stored according to manufacturer’s recommendations and the tests were run within 24 hours following collection. All study patients received care in the usual manner by attending physicians in the emergency department, the only difference between groups being whether or not the physician was given the influenza test result.
Influenza virus determination on viral respiratory panels run by the Children’s Hospital virology laboratory was monitored for the seasonal appearance of influenza virus in the pediatric community. Once influenza was documented to be in the community, patient screening and enrollment was begun. Enrollment began on January 19, 2002, and ended on March 5, 2002, after conclusion of the peak influenza season.
Influenza Virus Determination
The FluOIA test by Biostar was used to determine infection with influenza virus types A or B. The nasopharyngeal swab method was chosen for the highest combination of sensitivity (83%–96%) and specificity (64%–76%).7 Each patient had a nasopharyngeal swab specimen obtained and tested according to the manufacturer’s recommendations. Positive and negative test results were determined by use of the visual key provided within the test kits. The rapid influenza test results were not confirmed by other methods. The FluOIA test kits were provided by Biostar, but all other funding for the study was obtained from the Research Institute of the Children’s Hospital of Alabama.
Demographic information, including age, gender, and race, was collected for each patient on enrollment into the study. Individual influenza test results were recorded as positive or negative on the data collection forms. Randomization status was also recorded on each individual patient data form. Combination of the influenza result and randomization status ultimately yielded 4 groups of patients: 1) physician aware of result, influenza-positive; 2) physician unaware of result, influenza-positive; 3) physician aware of result, influenza-negative; and 4) physician unaware of result, influenza-negative. Patient records were obtained after discharge from the emergency department, and information regarding patient disposition, laboratory tests and radiographs ordered, antibiotic or antiviral use, and length of stay in the emergency department was recorded on a standardized form for subsequent data entry.
Telephone follow-up was performed for all patients who were enrolled and completed the entire study. At least 1 attempt was made to call the parent or legal guardian approximately 2 weeks after discharge, and the following patient data were collected: 1) number of return visits to primary physician or emergency department for this illness; 2) new prescriptions for this illness; 3) length of time patient missed school or day care; and 4) length of time parent or primary caregiver missed work.
Randomization was performed using a computerized randomization program, Rancode, in blocks of 4 patients with 2 patients allocated to each Group, 1 and 2. The computer-generated randomization list was produced for the entire study before study enrollment, eliminating the ability to maneuver patients into one group or the other. This block randomization technique was also used to more equally distribute patients into the 2 groups during the attending physicians’ shifts, thereby reducing potential for individual physician treatment bias.
Demographic information was categorized for each of the 4 groups using descriptive methods. The study was designed to evaluate only infants and children positive for influenza; however, influenza-negative patients were also analyzed for comparison of the study endpoints. Comparisons within the groups of influenza-positive patients and influenza-negative patients were made based on the randomization status (physician aware vs unaware of the influenza test result). The Student t test was used to compare means between the groups. For nonparametric methods, statistical comparisons were made using the Wilcoxon 2-sample test. All tests of significance were 2-sided using an α of 0.05. All analyses were performed using SAS Version 8.1 for Windows (SAS Inc, Cary, NC).
During the 46-day study period, 418 patients were enrolled into the study. Of these, 22 left before obtaining evaluation and treatment, and 5 additional patients were subsequently found to have exclusion criteria and thus were excluded from analysis. A total of 391 patients met enrollment criteria and completed the entire study. Information regarding patient demographics, influenza test result, and randomization status is located in Table 1. These 4 groups of patients did not differ significantly for age, gender, ethnicity, or temperature and respiratory rate at triage.
There were a total of 202 patients who were influenza positive and completed the emergency department visit. Randomization status subdivided these patients into 2 groups: physician aware of the result (n = 96), and physician unaware of the result (n = 106), during the visit. Comparison of these groups for laboratory tests and radiographs ordered and their associated charges billed to the patient are shown in Table 2. Effect of the test result on physician decision-making and patient treatment with regard to length of time to discharge and type of prescription given is also shown in Table 2. Analysis of the 189 influenza-negative patients is included in this table as well.
Within the 391 study patients, there were a total of 241 patients ages 2 to 36 months. Subgroup analysis of these patients for tests ordered, associated charges, type of prescription given, and time to discharge is shown in Table 3.
Follow-up of all laboratory tests and chest radiographs was performed to assure that serious bacterial illness was not missed in patients who had positive influenza test results. There were no influenza-positive patients in either group who had positive blood, urine, or cerebrospinal fluid bacterial cultures. Chest radiographic findings in all of the influenza-positive patients were read as either1 normal or2 consistent with viral lower respiratory tract disease. There were no cases of lobar pneumonia.
Telephone follow-up revealed no differences between the 2 influenza-positive groups for 1) number of return visits to primary physician or emergency department for this illness; 2) new prescriptions for this illness; 3) length of time patient missed school or day care; and 4) length of time parent or primary caregiver missed work.
This study evaluated the effect of knowledge of an influenza-positive rapid diagnostic test result on physician management and treatment of patients with influenza. It demonstrated that physician knowledge of rapid diagnostic test results for influenza-positive pediatric patients resulted in significant alteration of physician decision-making and management of the patient’s illness. Rapid diagnosis of influenza and physician knowledge of the result produced significant reductions in tests performed and their associated charges, decreased antibiotic use, increased antiviral use, and decreased length of stay in the emergency department.
The ethics of withholding the test results from treating physicians for half of the study patients was considered during the study design. Because rapid diagnostic testing for influenza was not available in our institution until several weeks into the study, it was not standard of care in our emergency department. When the rapid influenza test became available, physicians were able to order it if they chose to do so, although it was ordered in fewer than 10 study patients. All study patients received usual and customary treatment by the attending physicians.
For patients randomized to physician does not get the result, specimens were stored and the influenza test was run within 24 hours. This precluded having information that was being withheld from treating physicians that could potentially alter patient care. Physicians were not aware of patients’ enrollment into the physician does not receive results study arm during the patient visit and never received influenza test results for these patients. There was no coercion of physicians to alter treatment plans for study patients in the aware arm.
Physicians knew before initiation of the study that we would be evaluating differences in treatment between the aware and unaware groups of patients. There were no physicians who refused to participate in the study. The study hypothesis and specific aims were not discussed with the physicians in an effort to reduce introduction of management and treatment bias between the groups of patients.
Traditional diagnosis of influenza by viral culture or polymerase chain reaction is too lengthy to be useful in generating treatment options. Recent advances in technology have led to development of rapid diagnostic tests that are both sensitive and specific for diagnosis of influenza types A and B.4–7 To date, rapid diagnostic testing has not been incorporated into guidelines for the evaluation of febrile infants and young children.12 Factors such as lack of physician familiarity with rapid diagnostic test technology, cost of the test, reimbursement issues, and performance of tests in non-Clinical Laboratory Improvement Amendments settings are possible reasons for lack of use of these tests.
Morbidity associated with influenza is primarily in the form of upper and lower respiratory tract disease. Otitis media is a common finding with influenza infection and occurs in 20% to 42% of children.13,14 Children with influenza often appear quite ill and present with a variety of symptoms. In the setting of the emergency department, ill-appearing infants and children with fever and vague symptoms often have extensive testing performed to rule out serious bacterial illnesses such as bacteremia, pneumonia, meningitis, and urinary tract infection. Complete blood counts, urinalyses, blood cultures, lumbar punctures, and radiographs are frequently performed to diagnose these illnesses, especially in infants and younger children.12 These tests have the potential to significantly add to total patient charges and length of stay in the emergency department. One retrospective study demonstrated that fewer complete blood counts and urinalyses were ordered when patients were known to be influenza-positive before discharge from the emergency department.11 Our study confirms that there was a significant reduction in tests ordered and their associated charges when the physician had knowledge of the influenza-positive result. This was true for all study participants as well as the subgroup of patients 2 to 36 months of age.
Obtaining laboratory tests and radiographs in any age group would be expected to increase length of stay in the emergency department. In a teaching institution where both a resident and attending physician sees the patient, it is difficult to measure overall length of stay in the emergency department because of the numerous factors that influence patient flow. In our institution, it is customary for a resident to see the patient and then check out to an attending physician, who is ultimately responsible for patient management and disposition. For these reasons, we chose to measure the length of time from when the patient was first seen by the attending until discharge from the emergency department. We demonstrated a significant decrease in the length of time from when initially seen by an attending physician until discharge from the emergency department in the patients with influenza for whom the physician received the result.
Several studies have shown that patients with documented influenza infection have fewer antibiotics administered, thus reducing indiscriminant antibiotic use in proven viral illness.10,11 Decreased side effects and reduction in emergence of antibiotic resistance would naturally follow. Other investigators have demonstrated even further reduction in antibiotic use with appropriate diagnosis of influenza and use of antiviral agents early in the disease course.15 We observed statistically significant differences not only in tests ordered and their associated charges, but also demonstrated a significant decrease in antibiotic prescriptions between the influenza-positive groups where the differentiating variable between them was physician knowledge of the test result. Subgroup analysis of the 2 to 36 month age influenza-positive groups for antibiotic prescriptions also reached statistical significance. With substantial literature on antibiotic resistance and because of increasing pressure to utilize antibiotics only for proven bacterial disease, diagnosis of viral illness becomes more paramount to decrease inappropriate antibiotic usage.
Increased appropriate use of antiviral agents should also occur in patients with known influenza infection. Currently available antiviral agents for use in children include oseltamivir, zanamivir, amantadine, and rimantadine.16 Use of these agents has been proven to hasten resolution of illness, but effective utilization requires that the child be started on appropriate therapy within 48 hours of symptom onset.15,17
The older antiviral agents, amantadine and rimantadine, are synthetic cyclic amines with activity against influenza A and have the advantage of being relatively inexpensive. Disadvantages to these drugs include rapid development of viral resistance, ineffectiveness against influenza B virus, and adverse effects such as nervousness, jitteriness, and confusion. Newer agents are the neuraminidase inhibitors zanamivir and oseltamivir, which inhibit replication of both influenza A and B viruses. Zanamivir comes as an inhalational powder and is approved for use in patients 7 years and older. Oseltamivir is formulated as both a capsule and a suspension and is licensed for use down to 1 year of age.
Our study demonstrated a significant increase in antiviral use in known influenza positive patients. However, oseltamivir suspension was not produced for use in the 2001–2002 influenza season, which possibly resulted in fewer overall prescriptions being written for antiviral agents. Attending physicians who cared for known influenza-positive patients cited the side effect profiles and antiviral resistance as 2 major reasons for reluctance in prescribing amantadine and rimantadine.
The major limitation of this study was the relatively small number of influenza positive patients in the 2 to 36 month age group. Because this age group is most frequently targeted for testing for serious bacterial illness, one would suspect that differences in physician management of these patients could be altered by having knowledge of the test result. Even with smaller numbers of patients within these groups, significant differences were noted between the influenza-positive groups for many of the outcome variables. This suggests that further study of this subpopulation is warranted to determine the potential for use of rapid diagnostic testing in younger febrile children.
Our study demonstrates that rapid diagnostic testing for influenza has a basis for use in the point of care environment to reduce numbers of tests ordered and their associated charges, decrease antibiotic use for proven viral illness, decrease length of stay in the emergency department, and increase appropriate antiviral use in the pediatric population.
- Received December 20, 2002.
- Accepted May 15, 2002.
- Reprint requests to (A.B.B.) Department of Emergency Medicine, Scott & White Hospital, Texas A&M University Health Science Center, 2401 S 31st St, Temple, TX 76508. E-mail:
- Covalciuc KA, Webb KH, Carlson CA. Comparison of four clinical specimen types for detection of influenza A and B viruses by optical immunoassay and cell culture methods. J Clin Microbiol.1999;37 :3971– 3974
- Poehling KA, Griffin MR, Dittus RS, et al. Bedside diagnosis of influenzavirus infections in hospitalized children. Pediatrics.2002;110 :83– 88
- ↵FluOIA Test Product Package Insert. Boulder, CO: Thermo BioStar, Inc; 2001. Available at: http://www.thermo.com/eThermo/CDA/Products/Product_Detail/ 1,1075,13682-122,00.html. Accessed February 18, 2003
- ↵Woo PCY, Chiu SS, Seto W, Peiris M. Cost-effectiveness of rapid diagnosis of viral respiratory tract infections in pediatric patients. J Clin Microbiol.1997;35 :1579– 1581
- ↵Kimberlin DW, Prober CG. Antiviral agents and interferons. In: Long SS, Pickering LK, Prober CG, eds. Principles and Practice of Pediatric Infectious Diseases. 2nd ed. New York, NY: Churchill Livingstone; 2003:1527–1547
- Copyright © 2003 by the American Academy of Pediatrics