The Effect of Rapid Respiratory Viral Diagnostic Testing on Antibiotic Use in a Children's Hospital FREE

ACUTE VIRAL respiratory disease is the most common reason for pediatric hospitalization in the United States.¹ Respiratory syncytial virus (RSV), influenza A and B, and parainfluenza are common causes of disease, and result in syndromes that may require hospitalization (bronchiolitis, pneumonia, and croup).^{2 - 5} In young children, viral illnesses often result in fever, leading to diagnostic evaluations for bacterial disease and to empiric antibiotic use. Pediatricians prescribe antibiotics in 20% of cases of upper respiratory infection, 40% of cases of bronchiolitis, and as many as 100% of cases of pneumonia, even though the majority of these illnesses are thought to be viral.^{6 - 9} Recent data indicate that patients with documented viral illnesses are less likely to have a bacterial infection than those without viral illness.^{10 - 13} Empiric antibiotic therapy is, therefore, usually unnecessary in patients with viral illness. This inappropriate use contributes to antimicrobial resistance among bacterial pathogens encountered in pediatrics.^{7 - 9 ,14}

Accurate rapid detection methods for viruses have recently become available.^{15 - 17} The effect of viral diagnostics on hospital length of stay and treatment costs has been assessed in single studies of adult and pediatric populations.^{16 - 17} Likewise, data regarding the effects of viral diagnostics on antibiotic prescribing practices for hospitalized children are limited.^{17 - 18} We sought to investigate whether the results of rapid viral testing for multiple respiratory viruses would affect antibiotic use among hospitalized patients in an urban children's hospital.

The study was conducted during 2 consecutive winter seasons at Primary Children's Medical Center (PCMC), a 232-bed children's hospital that serves as a community hospital for Salt Lake County, Utah, and as a tertiary referral center for the intermountain West. The emergency department evaluates 33 000 children per year, and there are approximately 10 000 admissions to the hospital each year.

All admitting physicians and house staff received a mailing in December 2000 notifying them of the availability of direct fluorescent assay (DFA) testing for respiratory viruses in the clinical microbiology laboratory. We hypothesized that clinicians would be less likely to prescribe antibiotics and more likely to discontinue antibiotics if rapid and reliable viral diagnostic tests were available.

We reviewed the effects of testing during 2 consecutive winter seasons. Study period 1 was the first winter season during which DFA testing was offered, encompassing December 20, 2000, through February 13, 2001; study period 2 was in the second winter season, after DFA testing had been available for 1 full year, and included 2 weeks between December 27, 2001, and January 9, 2002. Both study periods encompassed weeks of high viral activity in our community (positive DFA results for PCMC, by week, can be viewed online¹⁹ ). During study period 2, DFA testing was ordered more frequently, and the peak of the respiratory viral season was shorter; therefore, we were able to enroll the needed patients for comparison for a shorter period.

The clinical microbiology laboratory began testing nasal wash specimens for respiratory viruses on December 20, 2000, using the SimulFluor Respiratory Screen (Chemicon International Inc, Temecula, Calif).¹⁵ The assay tests for respiratory syncytial virus (RSV); influenza A and B; parainfluenza viruses 1, 2, and 3; and adenovirus were performed by DFA. The manufacturer reports an overall sensitivity of between 98% and 100%, and specificity between 84% and 100% when compared with viral culture.²⁰

Specimens were prepared by cytospin and tested using the SimulFluor Respiratory Screen, followed by SimulFluor influenza A/influenza B; and SimulFluor parainfluenza 1, 2, and 3; or adenovirus stains if the screen was positive. During study period 1, viral respiratory culture testing was performed on all specimens using 4 shell vials, with 72-hour and 10-day exit stains. Sensitivity and specificity of the DFA was based on comparison with viral cultures. Because of the excellent sensitivity and specificity of the assay observed during the first year of use,²¹ during study period 2, viral respiratory culture was performed only for parainfluenza DFA-positive specimens and for DFA-negative specimens. Testing with DFA was performed 3 times daily during study period 1, and 5 times daily during study period 2. The laboratory charge was US $32 for DFA testing during study periods 1 and 2. There were no institutional protocols in place regarding DFA testing during either of the study periods.

The microbiology records were queried for all patients who had respiratory viral DFA testing performed during the 2 study periods. Patients who had DFA testing were eligible for inclusion in the study if they were admitted to the hospital. During study period 1, review of paper medical records for enrolled patients was performed. Data were abstracted from the discharge summary, medication administration record, and microbiology records. Data included demographics, admitting and discharge diagnosis, results of DFA and bacterial cultures, inpatient antibiotic therapy, and discharge prescriptions for antibiotics. Patients with positive and negative DFA results were compared. During study period 2, the medical records at PCMC were computerized. The computerized discharge summaries, medication administration records, and microbiology records were reviewed, and data collected included admitting diagnosis, DFA and bacterial culture results, whether intravenous antibiotics were administered during the hospital stay, and discharge prescriptions for antibiotics. Patients admitted during study period 2 were compared with those in study period 1.

All data were entered into a computerized database (Access 97; Microsoft Inc, Seattle, Wash). Comparisons between DFA-positive and DFA-negative patients were made using either a 2-sample inference for proportions (Statit Custom QC; Statware Inc, Corvalis, Ore), the t test, Fisher exact test, or χ² analysis (SPSS Inc, Chicago, Ill). Multiple logistic regression was performed to analyze variables related to intravenous antibiotic use. In the model, intravenous antibiotic use was the dependent variable, and DFA result, time to DFA result, and patient age were the independent variables. The effect of specific viral diagnoses on antibiotic prescribing was analyzed using χ² analysis.

During study period 1 (8 weeks), 229 admitted patients had DFA testing performed. During study period 2 (2 weeks), 109 admitted patients had DFA testing performed. The most common admitting diagnoses for patients in both study periods were bronchiolitis, pneumonia, and fever, representing 61% in study period 1 and 71% in study period 2. The ages of patients in study period 1 ranged from 4 days to 19 years (mean age, 29 months), and in study period 2, age ranged from 11 days to 18 years (mean age, 17.3 months, P = .02). Additional patient information is given in Table 1.

During study period 1, 104 (45%) of 229 patients had positive DFA results. The viruses identified included RSV (66%); influenza B (16%); parainfluenza 1, 2, and 3 (10%); and influenza A (7%). During study period 2, 57 (52%) of 109 had positive DFA results, and the viruses identified included RSV (74%); influenza A (16%); parainfluenza 1, 2, and 3 (7%); and adenovirus (3%). Patients who were 3 years old or younger were more than twice as likely to have a positive DFA result when compared with those who were older than 3 years (odds ratio [OR], 2.4; 95% confidence interval [CI], 1.2-4.8; P = .005).

The turnaround time for DFA results for patients during study period 1 was 8.6 hours, and during study period 2, it was 4.5 hours. The sensitivity of DFA testing for 1112 specimens processed at PCMC during the first 9 months of testing was 87%, and the specificity was 94%.²¹ Information regarding the sensitivity and specificity of the DFA testing for individual viruses can be found online.²²

During study period 1, 1 (<1%) of 104 DFA-positive patients had a culture-confirmed bacterial infection, compared with 16 (12.8%) of 125 in the DFA-negative group (P<.001). During study period 2, 1 (1.8%) of 57 DFA-positive patients had confirmed bacterial infection, compared with 7 (13.5%) of 52 DFA-negative patients (P = .02). When both study periods are combined, 2 (1.2%) of 161 DFA-positive patients had culture-proven bacterial infection, compared with 23 (13%) of 177 DFA-negative patients (P<.001). Patients who tested negative for viral infection by DFA were 11.9 times more likely to have culture-confirmed bacterial infection than those who tested positive (95% CI, 2.65-38.6).

The DFA-positive patients with dual viral and bacterial infections both, were diagnosed with pneumonia. The pathogens identified were RSV and Staphylococcus aureus bacteremia and influenza A and Haemophilus influenzae bacteremia.

During study period 1, 44% of DFA-positive patients received intravenous antibiotics during their hospital stay, compared with 45% of DFA-negative patients. However, the duration of intravenous antibiotic therapy was significantly shorter for DFA-positive patients compared with DFA-negative patients (2.5 days vs 4.0 days; P = .04). At the time of hospital discharge, DFA-positive patients were less likely to receive a discharge prescription for oral antibiotics (37% vs 52%; P = .02).

During study period 2, DFA-positive patients were significantly less likely to have intravenous antibiotic administration initiated when compared with DFA-positive patients from study period 1 (26% vs 44%; P = .008). At the time of hospital discharge, 37% of DFA-positive patients received prescriptions for oral antibiotics, which was the same percentage as during study period 1. Approximately 50% of the discharge prescriptions for oral antibiotics for DFA-positive patients during both winter seasons were written for otitis media, and approximately 50% were written for pneumonia.

In the multiple logistic regression model using data from both study periods, a negative DFA result was significantly associated with an increase in intravenous antibiotic administration. Patients who were DFA-negative were 2.3 times as likely to receive intravenous antibiotics than patients who were the same age and DFA-positive (95% CI, 1.5-3.6; P<.001). The turnaround time for the test was not a significant variable in the model. Children with RSV infection were 3.2 times less likely to receive intravenous antibiotics than patients with other viruses or no viruses identified (95% CI, 1.8-5.4; P<.001).

Overprescribing of antibiotics for viral respiratory infections is a significant problem in pediatrics and is believed to play a major role in increasing antibiotic resistance among common bacterial pathogens in the United States. Professional organizations, including the American Academy of Pediatrics (Elk Grove Village, Ill), the American Academy of Family Physicians (Leawood, Kan), and the Centers for Disease Control and Prevention (Atlanta, Ga) have published guidelines on the judicious use of antimicrobials in pediatric upper respiratory infections.²³ The Infectious Disease Society of America (Alexandria, Va) and the World Health Organization (Geneva, Switzerland) have both issued policy statements regarding the containment of antimicrobial resistance, and both emphasize the importance of the clinical microbiological laboratory.^{24 - 25}

The availability of rapid, accurate diagnostic testing that can identify viral infections has tremendous potential to limit the inappropriate use of antimicrobials. Relatively few studies, however, have demonstrated this. A study among adults compared patients with viral infections detected by conventional culture-based methods, with those that had infection detected by rapid methods.¹⁶ During the rapid testing period, detection of viruses was associated with shorter hospital stays and improved antibiotic stewardship.¹⁶ In the United States, 2 studies have shown that the rapid detection of influenza A is associated with decreased antibiotic use in children.^{18 ,26} A study from Hong Kong also showed a decrease in antibiotic use for children with influenza A or B, parainfluenza, or adenovirus infection following the initiation of rapid viral diagnostic testing.¹⁷

Our study sought to evaluate the effect of the availability of a single reliable diagnostic test for several common respiratory viruses on antimicrobial use. The DFA used in this study had excellent sensitivity and specificity, similar to what has been reported by others.¹⁵

Our study demonstrates that positive DFA results for respiratory viruses are common among patients tested, especially in younger children. The DFA results during both study periods indicated that approximately 50% of pediatric patients tested were positive for 1 of the 7 viruses identified by the DFA assay. Children younger than 3 years were 2.4 times more likely to have positive DFA results than older children, as is consistent with other studies documenting the importance of recognizing respiratory viruses in this age group.^{3 ,5 ,27}

Among children tested with the viral DFA panel, DFA-positive patients had a lower risk of bacterial infection that could be detected by culture than did DFA-negative patients. Only 1.2% of DFA-positive patients had culture-confirmed concomitant bacterial infection. Other studies have found similar low occurrences of bacterial infection in pediatric patients with viral illness.^{10 - 11 ,13 ,28}

Our study demonstrated an association between positive DFA test results and shorter duration of intravenous antibiotics during hospitalization for study period 1. During study period 2, when physicians were more familiar with DFA testing, there was a decrease in the rate of initiation of intravenous antibiotics for DFA-positive patients. This decrease occurred despite the fact that the age of patients during study period 2 was lower than in study period 1, and physicians might be more inclined to treat younger patients with antibiotics.^{14 ,29}

The availability of positive viral test results may have made it easier for physicians to explain to parents that antibiotics were not indicated for their child. Several studies have documented that physicians feel pressured by parents to prescribe antimicrobials for respiratory infections.^{30 - 34} Parents are likely to appreciate diagnostic certainty, and a physician with objective DFA results may be better able to communicate with parents, offer specific education about the expected course of the child's viral illness, and therefore, be more prepared to resist parental pressure for antibiotics. Several studies have documented that patient satisfaction is associated with communication with the physician, and not with antibiotic prescriptions.^{35 - 37}

The rate of prescribing oral antibiotics for DFA-positive patients at the time of hospital discharge was 37% during both study periods. Although this rate was lower than for DFA-negative patients, it was still substantial. Approximately 50% of the prescriptions were for the treatment of otitis media. Otitis media is a known complication of respiratory viral illness, especially RSV.^{38 - 39} Therefore, the prescribing of oral antibiotics in that situation may be warranted.⁴⁰ The remaining prescriptions were written for the treatment of pneumonia in patients with positive DFA tests for respiratory viruses. Dual infections with viruses and bacteria can occur in the case of pneumonia, but this is considered rare, especially with RSV infection.⁴¹ In our study, 2 children had pneumonia with both viral and bacterial pathogens isolated. These patients represented 5% of those with pneumonia. Another recent study using multiple diagnostic methods described dual infections with respiratory viruses and bacteria in 30% of children with pneumonia.²⁷ Our study and others indicate that physicians need more guidance regarding the use of antibiotics for children with pneumonia and a documented viral infection.^{7 - 8}

This study is limited by its retrospective design. The data reported are dependent on the accuracy of paper and computerized medical records. There were differences at PCMC in the way that medical care was delivered and recorded during the 2 study periods. However, the abstraction of medical record data from paper in season 1 and computer in season 2 was performed by the same investigator, using equivalent information sources, including the discharge summary, pharmacy records, and microbiology records. The number of times DFA testing was performed in season 2 increased; however, the difference in turnaround time for DFA results was not statistically significant in the logistic regression model. In addition, there may have been differences that affected physician management during the 2 study periods that could not be measured by a retrospective study (ie, the effect of preparations for the winter Olympics during season 2). Finally, the results are applicable only to patients who underwent viral DFA testing and were admitted to a single children's hospital. Despite these limitations, we believe the study does show the effect that viral testing can have on antibiotic prescribing for hospitalized children, and we believe that similar results could be seen in other hospitals that provide care for children.

In this era of increasing antimicrobial resistance, strategies for decreasing antibiotic use have become critically important. Physician and parent education about antibiotic use has been a cornerstone in most plans to decrease antimicrobial prescribing. These efforts are labor-intensive and costly, and may not completely remedy the problem.^{42 - 43} Other strategies must be examined to further limit inappropriate antimicrobial prescribing. The availability of the DFA test for respiratory viruses at our institution had a dramatic effect on physician prescribing of antimicrobials for hospitalized children with respiratory infections. The DFA test is sensitive and specific for the respiratory viruses included in the assay, is economical to perform, and gained wide acceptance by the admitting physicians at PCMC, as was demonstrated by the increase in ordering of the test during season 2. Hospitals, clinical microbiology laboratories, and other institutions should consider offering rapid diagnostic testing for common viruses. Newer technologies such as real-time polymerase chain reaction may further improve the performance and availability of rapid viral diagnostics in the future. Rapid and accurate viral diagnostic testing should be an important component of any plan for the containment of antimicrobial resistance.

Accepted for publication August 1, 2002.

Dr Byington is supported by the Robert Wood Johnson Generalist Physician Faculty Scholar Program.

Presented in part at the Annual Meeting of the Infectious Diseases Society of America, San Francisco, Calif, October 27, 2001.

We thank Charles Hoff, PhD, for his assistance with statistical analysis.

Corresponding author and reprints: Carrie L. Byington, MD, Department of Pediatric Infectious Diseases, University of Utah, 50 N Medical Drive, Salt Lake City, UT 84132 (e-mail: cbyington@med.utah.edu).

Overprescribing of antimicrobials for viral illness is a factor contributing to increasing antimicrobial resistance among bacterial pathogens encountered in pediatrics. This study demonstrates that the availability of an accurate and rapid diagnostic method for common respiratory viruses was associated with decreased use of antimicrobials in infants and children hospitalized for viral respiratory illnesses. Rapid and accurate viral diagnostic testing should be an important component of any plan for the containment of antimicrobial resistance.