Potential Cost Savings Through Rapid Diagnosis of Enteroviral Meningitis

Viral diagnostic tests have the potential to reduce antibiotic use, length of hospital stay and costs, although few studies have demonstrated this. 1 Data from a statewide enterovirus epidemic in Rhode Island suggested that rapid viral detection could have improved patient care and reduced health care expenditures. 2 Another study with the use of cell culture techniques demonstrated that patients benefitted from identification of enteroviruses as the cause of aseptic meningitis. 3 This finding notwithstanding, viral isolation from cerebrospinal fluid (CSF) or rectal swabs can take up to 2 weeks, beyond the time when bacterial infection has already been ruled out and the illness has resolved. 3 PCR could have a greater impact on patient care, given its rapidity and sensitivity. Rotbart et al. 4, 5 and others6, 7 alluded to potential cost savings from routine use of this technique, and one study estimated achievable reductions in hospital stay. 8 We looked at potential cost savings from routine use of enterovirus PCR on CSF during a seasonal outbreak of aseptic meningitis. Methods. In September and October, 1995, all infants ≤18 months of age hospitalized for "rule out sepsis, " undifferentiated febrile illness or suspected meningitis were identified. Patients with obvious bacterial meningitis, those ultimately diagnosed with other bacterial infections and those with noninfectious illnesses were excluded. Cerebrospinal fluid specimens were recovered from the clinical laboratory after routine studies had been performed (these were considered waste specimens, permitting waiver of informed consent by the University of Louisville Human Studies Committee). Specimens were stored at −70°C until they were tested in bulk by the AMPLICOR® EV PCR kit (Roche Diagnostic Systems, Inc. , Branchburg, NJ) according to the manufacturer's instructions and as previously reported. 4 Both physical9 and chemical10 precautions against contamination were taken, and positive and negative template RNA controls supplied by the manufacturer were used in each assay. Diagnostic tests were performed at the discretion of treating physicians. No attempt was made to evaluate rigorously PCR results using virus isolation or serology, mirroring what would occur in clinical practice. Cerebrospinal fluid pleocytosis was defined by the following age-adjusted norms for CSF white blood cell count: >35 for age 0 months; >25 for age 1 month; >5 for age ≥2 months. 11 Cerebrospinal fluid was considered uninterpretable if the number of white blood cells exceeded the upper limit of normal for age but contained >1000 red blood cells/ml. Hospital charts were reviewed and itemized bills were examined to determine charges assessed on each hospital day (defined as calendar days during which the patient was in the hospital for any amount of time). Polymerase chain reaction results and clinical information were used to identify infants with CSF pleocytosis who would have been eligible for early discharge. Charges that would have been eliminated by early diagnosis (e. g. room charges and antibiotics), as well as charges for viral cultures, were calculated. Revised hospital charges were determined by subtracting these amounts, but 150 (estimated charge) was added per patient for performing PCR. Results. Fifty-three infants were initially identified (all had negative bacterial cultures of CSF). Five patients with uninterpretable CSF studies had negative PCR results, and 25 patients without CSF pleocytosis had negative PCR results. The analysis focused on 18 patients with CSF pleocytosis, because these were the patients most likely to remain hospitalized until bacterial meningitis was ruled out (Table 1). Nine of these patients were PCR-positive and 9 were PCR-negative (one of these was clearly false negative because enterovirus was isolated from CSF). Collectively these patients generated 79 hospital days and 91 689 in charges. However, PCR-positive patients were afebrile for 18 of those hospital days and did not have seizures, apnea, persistent vomiting or oxygen dependence that might have necessitated prolonged hospitalization. Under the conservative assumption that clinicians would discharge afebrile patients with an established viral diagnosis (Table 1, Scenario A), routinely performing PCR on all infants with CSF pleocytosis would have resulted in total charges of 75 966, a reduction of 17%. Under the more liberal assumption that clinicians would discharge PCR-positive patients regardless of fever (Table 1, Scenario B) and assuming that PCR results would be available at 24 h, charges would have been reduced by 35%. TABLE 1: Characteristics of infants with CSF pleocytosis Two PCR-positive patients (Patients 5 and 8) and one PCR-negative patient (Patient 14) received antibiotics before lumbar puncture. Each of these was hospitalized for only 4 days, indicating that partially treated bacterial meningitis was not considered likely. Five patients with positive PCR results but either no pleocytosis (n = 4) or uninterpretable CSF (n = 1) were analyzed separately. Assuming these to be true positives, 5 total hospital charges of 27 575 would have been reduced 23% under Scenario A and 67% under Scenario B. Discussion. Infants with fever and CSF pleocytosis are usually hospitalized and expectantly treated with antibiotics until bacterial infection is ruled out, which traditionally takes 72 h from the time of culture inoculation. We used a manufactured PCR kit to test stored CSF specimens for evidence of enteroviral infection. The intent was to determine what cost benefits would have accrued had the results been available in real time. According to a recent large study, 5 the sensitivity and resolved specificity of this assay are 100 and 97%, respectively, for the diagnosis of enteroviral meningitis. Test performance was not specifically addressed in our study; rather it was assumed that enterovirus PCR would be applied without concomitant viral cultures and that positive results would most likely be true positives. Under study assumptions routine use of enterovirus CSF PCR on infants with pleocytosis during enterovirus season would have reduced hospital charges substantially. In making this determination it was assumed that establishing a diagnosis of enterovirus would obviate the need to hospitalize the patient until bacterial cultures were finalized as negative and that early discharge would ensue if there were no clinical complications. Scenario A, which resulted in a 17% reduction, assumed that clinicians would wait until the patient was afebrile before discharge. Scenario B, which resulted in a 35% reduction, assumed that patients would be discharged even if still febrile. Potential reductions in charges for PCR-positive patients without CSF pleocytosis were also shown, although the costs of doing PCR on all CSFs (as opposed to only those with abnormal cell counts) were not taken into account. The finding that one-half of infants with pleocytosis were PCR-positive is consistent with data from other studies. 7 One of the PCR-negative infants did have an enterovirus isolated from CSF (false negatives have been reported in other studies as well). 4, 6, 7 Had this and perhaps other false negatives been correctly identified, even greater cost savings might have been realized (alternatively caution should be raised about the possible consequences of early discharge of patients with false positive PCR results). The possibility for cost savings in infants with meningitis who receive antibiotic therapy before lumbar puncture should be emphasized. Other potential benefits of rapid molecular diagnosis include reducing unnecessary diagnostic tests such as computerized tomography and electroencephalography and obviating the need for other empiric therapies such as acyclovir. 12 In addition reducing antibiotic use through molecular viral diagnosis might impede the emergence of resistant bacteria. Acknowledgments. This work was supported by the Alliant Community Trust Fund and Roche Diagnostic Systems, Inc. Gary S. Marshall, M. D. Mary Anne Hauck, B. A. George Buck, Ph. D. Gerard P. Rabalais, M. D. Department of Pediatrics; University of Louisville School of Medicine; (GSM, MAH, GRP) Clinical Microbiology Laboratory; Alliant Health Systems (GB) Louisville, KY