Abstract
Background
For infants aged ≤ 60-days evaluated for temperature instability, we sought to ascertain if use of a multiplex meningoencephalitis panel (MEP) decreased length of hospital stay (LOS), antimicrobial length of therapy (LoT), and use of other molecular testing.
Methods
This was a retrospective cohort study of infants aged ≤ 60-days hospitalized for temperature instability between 1/2012 and 1/2022. Patients were included if blood and cerebrospinal fluid (CSF) cultures were negative for bacterial infection. Patients were categorized into groups by MEP availability and use. LOS, LoT, and use of other molecular tests were compared by MEP usage group.
Results
A total of 2076 patients met study criteria; there was no difference between groups regarding sex, race, ethnicity, or age at admission. During the period of MEP availability (2017–2022), the test was performed for 156/1101 (14%) patients. Compared to eligible patients without the MEP performed, MEP use was associated with longer LOS (57.7 vs. 47.3 h, p < 0.0001) and longer LoT (37.4 vs. 35.5 h, p < 0.0001). Use of the MEP was also associated with performance of additional molecular tests. Patients with CSF obtained after antimicrobial initiation had longer LOS compared to those with CSF obtained before antimicrobials were initiated, and MEP use did not mitigate that.
Conclusions
Young infants without bacterial infection by blood or CSF culture that had the MEP performed had longer length of hospital stays, longer duration of antimicrobial therapy, and more molecular testing performed compared to those without the MEP performed. Further research focusing on clinical scenarios where MEP use may be most beneficial for infants with temperature instability is needed.
Clinical trial
Not applicable.
Keywords: Neonatal fever, Meningoencephalitis, Antimicrobials, Diagnostic stewardship, Multiplex PCR
Background
For infants with temperature instability, the prevalence of invasive bacterial infections is low [1–3], but risk of significant morbidity and mortality necessitates immediate evaluation with stratification by age, physical examination, and laboratory testing [4, 5]. After blood, urine, and cerebrospinal fluid (CSF) cultures are obtained, initiation of empiric, broad-spectrum antimicrobials [4] typically occurs quickly. The time needed to finalize culture results can prolong unnecessary antimicrobial use and hospital stays.
For patients with concern for bacterial or viral meningitis, use of molecular multiplex panels can lead to more rapid identification of specific infectious etiologies compared to CSF culture or monoplex polymerase chain reaction (mPCR) testing. This could possibly shorten the time to narrowing or discontinuing antimicrobials in patients with proven infectious meningitis. Previous studies including pediatric patients with suspected central nervous system infection found utilization of a multiplex meningoencephalitis panel (MEP) decreased the use and duration of antimicrobials, decreased hospital length of stay (LOS), and decreased hospitalization costs [6–10]. Although these studies included young infants, none focused on the use of MEP in evaluation and clinical decision-making for infants aged ≤ 60-days presenting with temperature instability.
The primary objective of this study was to measure the impact of MEP use on antimicrobial duration, hospital LOS, and other testing in infants aged ≤ 60-days hospitalized with temperature instability and without bacteremia or bacterial meningitis. The secondary objective was to evaluate the use of MEP in a sub-population of infants with antimicrobials initiated prior to obtaining CSF for culture.
Methods
Study design
The electronic health record (Epic systems, Verona, WI) was queried for infants aged ≤ 60-days hospitalized between 1/1/2012 and 1/31/2022 who underwent evaluation for temperature instability and were admitted to the acute care or intensive care unit of a single 298-bed, freestanding midwestern children’s hospital (Fig. 1). Temperature instability was identified by ICD-9 and ICD-10 codes corresponding to “fever,” “fever in neonate,” “fever in patient under 28 days,” “fever in patient 29–60 days,” “hypothermia,” and “hypothermia of newborn.” Inclusion in the study required blood and CSF cultures be performed for each patient. Exclusions were made for infants with a history of intracranial surgery or device, intracranial structural or neurovascular abnormality, head trauma, CSF obtained ≥ 72 h after admission to our institution, CSF obtained from patients transferred from a neonatal intensive care unit at an outlying hospital, positive CSF or blood cultures, and positive CSF bacterial pathogen molecular testing. Seven patients tested positive for herpes simplex virus (HSV) in ≥ 1 specimen from blood, lesion, skin, mucosal membrane, and/or CSF and were included in the analysis cohort.
Fig. 1.
Study population description. Abbreviations: cerebrospinal fluid (CSF), meningoencephalitis panel (MEP), number (N). 1Temperature instability was identified by ICD-9 and ICD-10 codes corresponding to “fever,” “fever in neonate,” “fever in patient under 28 days,” “fever in patient 29–60 days,” “hypothermia,” and “hypothermia of newborn.” 2Institutional restrictions precluded MEP from being performed on CSF labeled as normal for all measurements, including total nucleated cell count (TNC) ≤ 30 if aged ≤ 30-days or TNC ≤ 5 if aged > 30-days, protein 15–45 mg/dL, glucose 45–80 mg/dL, and red blood cell (RBC) count < 800 if aged ≤ 7-days, < 50 if aged 7–30 days, or 0 if aged > 30-days
Data was stored using Research Electronic Data Capture (REDCap), a secure, web-based software platform for data capture [11]. The dataset generated and analyzed during the current study is not publicly available as it contains protected health information, but the de-identified dataset is available from the corresponding author on reasonable request. This retrospective observational cohort study was reviewed by the Children’s Wisconsin Human Research Protection Program Institutional Review Board in accordance with the Belmont Report, was determined to not be human subject’s research, and the need for consent was waived. Clinical trial number: not applicable.
The meningoencephalitis panel (MEP) (BioFire FilmArray Meningitis/Encephalitis Panel, BioMérieux, Salt Lake City, Utah) is a molecular assay targeting fourteen pathogens (six bacteria, seven viruses, one yeast) causing central nervous system infections. MEP became available for in-house performance 7-days a week at our institution in December 2016. Institutional restrictions precluded MEP from being performed on CSF labeled as normal for all measurements, including total nucleated cell (TNC) count ≤ 30 if aged ≤ 30-days or TNC count ≤ 5 if aged > 30-days, protein 15–45 mg/dL, glucose 45–80 mg/dL, and red blood cell (RBC) count < 800 if aged ≤ 7days, < 50 if aged 7–30 days, or 0 if aged > 30-days. During our study timeframe, mPCR for herpes simplex virus was available 7-days per week and mPCRs for enterovirus and parechovirus were available weekdays.
Patients meeting study inclusion criteria were grouped by admission year and if MEP was performed (Fig. 1). Groups A and B included patients admitted 1/1/2017–1/31/2022, the study timeframe when MEP was available. Group A included patients with MEP performed while group B consisted of patients without MEP performed. Group B was further divided into two subgroups: “group B allowed” consisting of patients with abnormal CSF analysis allowing MEP to be performed if ordered and “group B not allowed” consisting of patients with normal CSF analysis precluding MEP performance even if ordered. Group C included patients admitted between 1/1/2012–12/31/2016, the study timeframe prior to MEP availability at our institution representing a historical control group. Group B + C represents all study participants without MEP performed. Patients with pre-treated CSF received antibiotics prior to obtaining CSF for analysis and culture, as determined by comparing antimicrobial administration date and time (by electronic health record query, see below) to the date and time of CSF collection.
Outcome measures
The primary outcomes investigated were hospital length of stay (LOS) and length of antimicrobial therapy (LoT). The electronic health record was queried for antimicrobial orders and administration dates and times. Antimicrobial orders included penicillin, ampicillin, cefepime, ceftriaxone, cefotaxime, meropenem, vancomycin, acyclovir, and gentamicin, representing those antimicrobials potentially used for empiric therapy in this population. LoT was defined as the total number of hours a patient received systemic antimicrobials regardless of how many different antimicrobials they received, calculated between the dates and times of the first and last doses of antimicrobial given. For patients receiving > 1 antimicrobial, LoT was the maximum number of hours they received any antimicrobials. Acyclovir LoT (hours) was calculated from the dates and times of the first and last doses administered. For patients receiving antimicrobials prior to arrival at our institution, dates and times of antimicrobial initiation were recorded if available in the referring hospital documentation. If not available, initiation dates and times were recorded as 4 h prior to the first dose of antimicrobials given at our institution. LOS was calculated as the number of hours between the date and time of initial presentation to our institution through the date and time of discharge from our institution.
Data collection
Variables collected by electronic health record query included sex, age at the time of admission, gestational age, race, ethnicity, complete blood cell count (CBC) with differential, initial inflammatory markers (C-reactive protein (CRP) and procalcitonin), urine culture results, blood culture results, CSF analysis and culture results, CSF MEP results, and mPCR results (CSF, blood, rectal, and nasopharyngeal samples) for enterovirus, parechovirus, and HSV. Normal values were defined as CRP ≤ 1.0 mg/dL and procalcitonin < 0.11 ng/mL. Chart reviews for every patient meeting inclusion criteria were not feasible, though two authors (BP, AH) reviewed a sample of data collected for accuracy.
Data analysis
Data were summarized as number n (%) or median and interquartile range (IQR). Chi-square or Fisher’s exact test was used to compare categorical variables and Kruskal-Wallis test or Mann-Whitney-Wilcoxon test was used to compare continuous variables. Logistic regression was used to model MEP use with the following predictors included in the model: age at admission (days), gestational age (weeks), year of admission, race, CSF pretreated (yes/no), CSF TNC, CSF RBC, and CSF protein. Firth’s penalized likelihood approach was used to minimize bias. Odds ratio (OR) and 95% confidence intervals (CI) were generated. Generalized linear model with normal distribution and log link function was also used to investigate the effect of predictors on the outcomes, LOS and LOT. Geometric mean and 95% CI were generated. P < 0.05 was considered statistically significant. Correlation analysis was used to assure the absence of collinearity in variables included in multivariable models. Because MEP use was not randomly assigned, propensity score matching was employed to reduce the risk of bias due to confounding in groups A and B allowed. Each patient who received MEP testing was matched to two control patients with a similar propensity score. The matching was done using nearest neighbor matching without replacement and the propensity score was based on the covariates listed above in the logistic regression. Balance was assessed by standardized mean difference. All standardized mean differences in the matched samples were within 0.15, which indicates good covariates balance between the two groups. We then compared the differences in LOS and LoT between the two matched groups using Generalized Estimating Equations. SAS version 9.4 (SAS Institute Inc., Cary, NC), SPSS version 28 (IBM Corp., Armonk, NY), and R version 4.5.0 (R Core Team, Vienna, Austria) were used for statistical analysis.
Results
Description of the whole cohort and analysis subgroups
A total of 2076 patients met criteria for inclusion in the whole cohort (Fig. 1; Table 1). Over half (53.7%) of the patients were male, and the majority (59%) were White/Caucasian. There was no significant difference between groups regarding sex or race. Median age at admission of the whole cohort was 24.7 days (IQR 14.7, 38.5; range 1.6–60.9 days). There was no significant difference in age at admission when comparing group A with group B or group B + C. However, gestational age of group A was lower than group B + C, 38 weeks (37, 39) vs. 39 weeks (37, 40), p = 0.04. Median age of group B patients (allowed + not allowed) was 22.6 days (12.7, 35.5), younger than group C patients (26.9 days, 16.8, 40.9), p < 0.0001. There was a significant difference in ethnicity between group C and group B (allowed + not allowed), p = 0.011.
Table 1.
Patient demographics, grouped by meningoencephalitis panel (MEP) availability and use
| Whole Cohort | Group A | Group B | Group C | Group B + C | ||
|---|---|---|---|---|---|---|
| Group B allowed | Group B not allowed | |||||
| Number of patients | 2076 | 156 | 909 | 36 | 975 | 1920 |
| Age at admission in days (median; IQR) | 24.7 (14.7, 38.5) | 23.2 (13, 33) | 22.5 (12.4, 35.1) | 26 (20.2, 38.8) | 26.9 (16.8, 40.9) | 24.8 (14.8, 38.8) |
|
Gestational age in weeks (median; IQR) Number of patients with gestation age documented Number of patients < 37-weeks gestation |
39 (37–40) 1377 197 (14.3%) |
38 (37–39) 1 97 19 (19.6%) |
39 (37–39) 622 102 (16.4%) |
39 (39–40) 25 0 |
39 (38–40) 633 76 (12.0%) |
39 (37–40) 1280 178 (13.9%) |
| Sex | ||||||
|
Male Female |
1115 (53.7%) 961 (46.3%) |
82 (52.6%) 74 (47.4%) |
494 (54.4%) 415 (45.6%) |
19 (52.8%) 17 (47.2%) |
520 (53.3%) 455 (46.7%) |
1033 (53.8%) 887 (46.2%) |
| Race | ||||||
|
Black or African American White or Caucasian Asian AI / AN / NH / PI Multiracial Unknown / Refused |
564 (27.2%) 1225 (59.0%) 85 (4.1%) 16 (0.8%) 81 (3.9%) 105 (5.1%) |
31 (19.9%) 101 (64.7%) 8 (5.1%) 0 8 (5.1%) 8 (5.2%) |
229 (25.2%) 557 (61.3%) 37 (4.1%) 4 (0.4%) 47 (5.2%) 35 (3.9%) |
15 (41.7%) 18 (50%) 1 (2.8%) 0 1 (2.8%) 1 (2.8%) |
289 (30%) 549 (57%) 39 (4.1%) 8 (0.8%) 25 (2.6%) 53 (5.6%) |
533 (27.9%) 1124 (58.9%) 77 (4%) 12 (0.7%) 73 (3.8%) 89 (4.6%) |
| Ethnicity | ||||||
|
Hispanic Non-Hispanic Unknown / Refused |
370 (17.8%) 1678 (80.1%) 41 (2.0%) |
30 (19.2%) 121 (77.6%) 5 (3.2%) |
141 (15.5%) 752 (82.7%) 16 (1.7%) |
4 (11.1%) 32 (88.9%) 0 |
195 (20%) 772 (79.2%) 8 (0.8%) |
340 (17.7%) 1556 (81%) 24 (1.3%) |
Abbreviations: interquartile range (IQR), American Indian (AI), Alaskan Native (AN), Native Hawaiian (NH), Pacific Islander (PI)
Group definitions: Group A: MEP available and performed, Group B: MEP available but not performed, Group B allowed: MEP was not done despite abnormal CSF parameters, Group B not allowed: had normal CSF precluding MEP from being performed, Group C patients admitted prior to MEP availability at our institution
1 Compared to group B + C, p = 0.04
Laboratory data for whole cohort and analysis subgroups
Table 2 provides laboratory data for the whole cohort and each analysis group. The whole cohort had a median peripheral white blood cell count of 10.1 × 103/µL, with no significant differences among analysis groups. For the whole cohort, median CRP was elevated at 1.3 mg/dL (< 0.5, 4.6) and procalcitonin was 0.2 ng/mL (0.1, 1.9), with no significant differences among groups.
Table 2.
Comparison of laboratory data by study cohort group (median, IQR)
| Whole Cohort | Group A | Group B | Group C | Group B + C | ||
|---|---|---|---|---|---|---|
| Group B allowed | Group B not allowed | |||||
| Peripheral white blood cell count (x103/µL) | 10.1 (7.5, 13.5) | 9.8 (7.3, 13.6) | 10 (7.6, 13.4) | 10.4 (7.4, 15.2) | 10.4 (7.5, 13.7) | 10.2 (7.5, 13.5) |
| First CRP (mg/dL) | 1.3 (< 0.5, 4.6) | 1.8 (< 0.5, 5.6) | 1 (< 0.5, 4.2) | 14.7 (14.7, 14.7)* | 1.7 (0.5, 5.3) | 1.1 (-<0.5, 4.5) |
| First Procalcitonin (ng/mL) | 0.2 (0.1, 1.9) | 0.5 (0.1, 6.8) | 0.1 (0.1, 1.5) | 0.4 (0.4, 0.4)* | - | 0.2 (0.1, 1.4) |
| CSF TNC count (per µL) | 6.0 (3, 14) | 22 (7, 108)1 | 5 (2, 10) | 2 (1, 4) | 6 (3, 17) | 6 (2, 12) |
| CSF % neutrophils | 7 (1, 28) | 18 (3, 46)1 | 4 (1, 20) | 0 (0, 2) | 10 (2, 31) | 6 (1, 26) |
| CSF % lymphocytes | 15 (8, 25) | 14 (7, 24) | 14 (8, 25) | 11 (6, 15) | 15 (9, 25) | 15 (8, 25) |
|
CSF RBC count (per µL) % >500 RBC (per µL) |
41 (2, 1835) 33.2% |
1345 (19, 17500)1 59.3% |
47 (2.5, 1357.5) 31.3% |
0 (0, 1) 0% |
33 (2, 1525) 32.1% |
35 (2, 1260) 31.0% |
| CSF glucose (mg/dL) | 46 (42, 51) | 46 (41, 52) | 46 (42, 51) | 50 (47.5, 54.5) | 46 (42, 51) | 46 (42, 51) |
| CSF protein (mg/dL) | 69 (53, 100) | 93 (65, 141)1 | 69 (53, 99) | 36 (32.5, 42) | 68 (53, 98) | 68 (52, 98) |
Abbreviations: C-reactive protein (CRP), cerebrospinal fluid (CSF), red blood cell (RBC), total nucleated cell (TNC)
Group definitions: Group A: meningoencephalitis panel (MEP) available and performed, Group B: MEP available but not performed, Group B allowed: MEP was not done despite abnormal CSF parameters, Group B not allowed: had normal CSF precluding MEP from being performed, Group C patients admitted prior to MEP availability at our institution
*Only 1 patient in group B not allowed had CRP and procalcitonin performed
1Compared to group B allowed, p < 0.0001
The whole cohort had a median CSF TNC count of 6 cells/µL with 7% neutrophils (1, 28) and 15% lymphocytes (8, 25), and RBC count of 41 cells/µL (2, 1835) (Table 2). Group A had a significantly higher CSF TNC count (22 cells/µL; 7, 108) compared to group B allowed (5 cells/µL; 2, 10), p < 0.0001. Group A had a significantly higher percentage of CSF neutrophils (18%; 3, 46) compared to group B allowed (4%; 1, 20), p < 0.0001. Group A showed a significantly higher CSF RBC count (1345 cells/µL; 19, 17500) and protein content (93 mg/dL; 65, 141) compared to group B allowed (RBC count 47 cells/µL; 2.5, 1357.5; protein 69 mg/dL, 53, 99), p < 0.0001 for both. There was no significant difference in the percent of CSF lymphocytes or CSF glucose between groups. In the whole cohort, 332 patients had positive urine cultures. There was no significant association between positive urine culture and MEP utilization group, p = 0.32.
Associations of MEP use and LOS, LoT, and other PCR testing: whole cohort
The cohort with MEP use (Group A) was compared to the other cohorts for the outcomes of interest (Table 3). Group A had LOS of 57.7 h (45.1, 89), longer than group B allowed 47.3 (41.6, 61.9) and group B + C 50.8 (43.9, 67.1) p < 0.0001 and p = 0.003, respectively. Group C had a significantly longer LOS compared to group B allowed, 54.4 (47.3, 71.5) versus 47.3 (41.6, 61.9) h, respectively, p < 0.0001. In addition to longer LOS, group A had a significantly longer LoT (37.4 h; 29.8, 50.8) compared to group B allowed (35.5 h; 29.4, 40), p < 0.0001. The number of patients in each group with CSF and non-CSF mPCRs performed were compared (Table 3). Group A had significantly more patients with additional CSF mPCRs performed (63/156 patients, 40.4%) compared to group B allowed (275/909 patients, 30.3%), p = 0.012; a significant difference that was also observed for non-CSF mPCR tests, 48.7 vs. 24.3, p < 0.0001.
Table 3.
Study group cohort LOS, LoT, acyclovir duration, monoplex PCR (mPCR) testing, and HSV PCR testing by study cohort group
| Whole Cohort | Group A | Group B | Group C | Group B + C | ||
|---|---|---|---|---|---|---|
| Group B allowed | Group B not allowed | |||||
| Median LOS, h (IQR) | 51 (44, 68.5) | 57.7 (45.1, 89)1, 2 | 47.3 (41.6, 61.9) | 47.3 (39.4, 53.4) | 54.4 (47.3, 71.5) 1 | 50.8 (43.9, 67.1) |
| Median LoT, h (IQR) | 37.5 (31.5, 46) | 37.4 (29.8, 50.8)1 | 35.5 (29.4, 40) | 31.6 (24.6, 36.3) | 42.2 (36.6, 48.4)1 | 37.6 (31.6, 45.6) |
| Median acyclovir, h (IQR) | 15.5 (8.1, 23.3) | 15.5 (7.9, 21.7) | 15.3 (8.1, 23.1) | 20.1 (15.5, 24.8) | 15.5 (8.1, 23.3) | 15.5 (8.1, 23.3) |
| Number of patients with CSF mPCR tests done* (%) | 723 (34.8%) | 63 (40.4%) 3 | 275 (30.3%) | 10 (27.8%) | 375 (38.5%) 1 | 660 (34.4%) |
| Number of patients with non-CSF mPCR tests done (%) | 547 (26.3%) | 76 (48.7%) 4 | 221 (24.3%) | 7 (19.4%) | 243 (24.9%) | 471 (24.5%) |
| Number of patients with CSF + non-CSF mPCR tests done (%) | 845 (40.7%) | 88 (56.4%) 4 | 322 (35.4%) | 12 (33.3%) | 423 (43.4%) 1 | 757 (39.4%) |
| Number of patients with HSV + (any specimen) | 7 (0.3%) | 1 (0.6%) | 3 (0.3%) | 0 | 3 (0.3%) | 6 (0.3%) |
Abbreviations: cerebrospinal fluid (CSF), hours (h), interquartile range (IQR), length of hospital stay (LOS), length of antimicrobial therapy (LoT),
Group definitions: Group A: meningoencephalitis panel (MEP) available and performed, Group B: MEP available but not performed, Group B allowed: MEP was not done despite abnormal CSF parameters, Group B not allowed: had normal CSF precluding MEP from being performed, Group C patients admitted prior to MEP availability at our institution
*For group A, this is in addition to the MEP. For groups B and C, this is patients who had a CSF PCR performed
1 Compared to group B allowed, p < 0.0005
2 Compared to group B + C, p = 0.003
3 Compared to group B allowed, p = 0.012
4 Compared to group B allowed, p < 0.0001 and group B + C, p < 0.0001
Associations of MEP use and LOS, LoT, and additional PCR testing: patients with pre-treated CSF
MEP use for patients with pre-treated CSF was evaluated for impact on LOS, LoT, acyclovir duration, and number of additional molecular tests performed (Table 4). LOS trended longer for patients with antimicrobial pre-treated compared to not pre-treated CSF in all groups. Among patients with pre-treated CSF, LOS trended longer in group A compared to group B and was comparable to group C. Patients with pre-treated CSF were highly represented in the MEP performed group, comprising 44.2% of group A versus 7.7% of group B + C. Neither group A nor group B showed a significant difference in median LoT based on pre-treatment status. Group A patients with pre-treated CSF had significantly more non-CSF mPCR testing done as compared to group B + C with pretreated CSF (p = 0.047).
Table 4.
Pretreated vs. not pretreated CSF: LOS, LoT, acyclovir duration, and additional tests by study cohort group
| Group A | Group B | Group C | Group B + C | |||||||
|---|---|---|---|---|---|---|---|---|---|---|
| Group B allowed | Group B not allowed | |||||||||
| Not pre-treated CSF | Pre-treated CSF | Not pre-treated CSF | Pre-treated CSF | Not pre-treated CSF | Pre-treated CSF | Not pre-treated CSF | Pre-treated CSF | Not pre-treated CSF | Pre-treated CSF | |
| Number patients (%) | 87 (55.8%) | 69 (44.2%) | 848 (93.3%) | 61 (6.7%) | 36 (100%) | 0 | 888 (91.1%) | 87 (8.9%) | 1772 (92.3%) | 148 (7.7%) |
| Median LOS, h (IQR) | 50.4 (43.6, 66.3) | 69.1 (48.5, 103.1) | 47 (41.4, 60.6) | 52.4 (43.5, 84.2) | 47.3 (39.4, 53.4) | - | 53.1 (47.3, 69.8) | 67 (52.9, 141.6) | 50.3 (43.7, 65.8) | 61.7 (48, 120.9) |
| Median LoT, h (IQR) | 37.1 (31.2, 43.6) | 39.4 (34.3, 54.3) | 35.2 (29.1, 39.4) | 36.7 (31.4, 48.4) | 31.6 (24.6, 36.3) | - | 42 (36.4, 48.1) | 48.6 (42.4, 62.8) | 37.1 (31.4, 43.5) | 45.3 (36.3, 55.4) |
| Median acyclovir duration, h (IQR) | 15.3 (8.0, 21.9) | 14.7 (8.5, 17.1) | 15.2 (8.0, 18.6) | 16.7 (15.1, 31.7) | 20.1 (15.5, 24.8) | - | 15.4 (8.0, 21.1) | 24.2 (14.4, 38.5) | 15.3 (8.0, 21.1) | 17.2 (14.4, 33.2) |
| Number patients with CSF mPCRs done (%) | 39 (44.8%) | 24 (34.8%) | 251 (29.6%) | 24 (39.3%) | 10 (27.8%) | 0 | 334 (37.6%) | 41 (47.1%) | 595 (33.6%) | 65 (43.9%) |
| Number patients with non-CSF mPCRs done (%) | 41 (47.1%) | 35 (50.7%) 1 | 199 (23.5%) | 22 (36.1%) | 7 (19.4%) | 0 | 211 (23.8%) | 32 (36.8%) | 417 (23.5%) | 54 (36.5%) |
| Number patients with CSF + non-CSF mPCRs done (%) | 50 (57.5%) | 38 (55.1%) | 293 (34.6%) | 29 (47.5%) | 12 (33.3%) | 0 | 380 (42.8%) | 43 (49.4%) | 685 (38.7%) | 72 (48.7%) |
Abbreviations: cerebrospinal fluid (CSF), hours (h), interquartile range (IQR), length of stay (LOS), length of therapy (LoT), monoplex polymerase chain reaction (mPCR)
Group definitions: Group A: MEP available and performed, Group B: MEP available but not performed, Group B allowed: MEP was not done despite abnormal CSF parameters, Group B not allowed: had normal CSF precluding MEP from being performed, Group C patients admitted prior to MEP availability at our institution
1 Compared to group B + group C, pre-treated, p = 0.047
Multivariable analysis: predictors of MEP use
Only 156 (14.4%) of the 1065 patients in groups A plus B allowed had MEP performed. Year of admission (per year; OR 1.9, 95% CI 1.7, 2.1, p < 0.0001), pre-treated CSF (yes vs. no; OR 1.7, 95% CI 1.1, 2.5, p = 0.011), CSF TNC (per 10 cell increase; OR 1.016, 95% CI 1.009, 1.022; p < 0.0001), and CSF protein (per 10-unit increase; OR 1.024, 95% CI 1.003, 1.045, p = 0.026) were all associated with MEP use, while race, age at admission, and CSF RBC were not (data not shown).
Multivariable analysis: effect of MEP use on LOS, lot: whole cohort
Adjusting for age, year of admission, pre-treated CSF, CSF TNC, and CSF RBC, MEP use was associated with 54.1% longer LOS (p < 0.0001) and 39.6% longer LoT (p = 0.0004), compared to those without MEP performed (Table 5). A 1-day increase in age was associated with 0.52% reduction in LOS (p = 0.002) and 0.54% reduction in LoT (p = 0.002). Each subsequent calendar year was associated with 5.8% decrease in LOS (p < 0.0001) and 6.7% reduction in LoT (p < 0.0001). Patients with pre-treated CSF had 11.5% longer LOS (p = 0.03) and 5.4% longer LoT (p = 0.33). CSF TNC was significantly associated with increased LoT but not LOS.
Table 5.
Multivariable analysis for LOS and lot
| Beta Estimate (95% CI) | p-value | |
|---|---|---|
| LOS (hours) | ||
| Age (days) | 0.995 (0.991–0.998) | 0.002 |
| Year of admission | 0.942 (0.921–0.964) | < 0.0001 |
| Pre-treated CSF | 1.115 (1.011–1.229) | 0.03 |
| MEP performed | 1.541 (1.310–1.812) | < 0.0001 |
| CSF TNC count | 1.0002 (1.000–1.0003) | 0.06 |
| CSF RBC count | 1.0 (1.0–1.0) | 0.09 |
| LoT (hours) | ||
| Age (days) | 0.995 (0.991–0.998) | 0.002 |
| Year of admission | 0.934 (0.911–0.956) | < 0.0001 |
| Pre-treated CSF | 1.054 (0.949–1.171) | 0.33 |
| MEP performed | 1.396 (1.161–1.680) | 0.0004 |
| CSF TNC count | 1.0004 (1.0003–1.0005) | < 0.0001 |
| CSF RBC count | 1.0 (1.0–1.0) | 0.06 |
Abbreviations: cerebrospinal fluid (CSF), length of hospital stay (LOS), length of antimicrobial therapy (LoT), meningoencephalitis panel (MEP), red blood cells (RBC), total nucleated cells (TNC)
Effect of MEP use on LOS, lot: propensity matched cohort
Of the 557 original samples without missing values in groups A and B allowed, 236 were able to be matched. Using the matched cohort data, patients with MEP panel ordered had 1.5 times longer LOS (hours): estimated mean and 95% CI 112.0 (88.7, 141.3) than that in patients without MEP 76.3 (62.7, 92.9), p = 0.015. Estimated mean LoT (hours) in patients with and without MEP ordered was 61.8 (45.7, 83.6) vs. 44.7 (35.2, 56.6), respectively, p = 0.11.
Discussion
In this cohort of infants aged ≤ 60-days hospitalized with temperature instability and with invasive bacterial infection excluded, use of MEP was associated with longer LOS, longer LoT, and increased use of additional molecular tests compared to those infants in whom MEP was not performed. MEP was used more frequently in patients with pre-treated CSF, greater CSF pleocytosis, and CSF with increased protein, but such use was not associated with shorter LOS or LoT, perhaps suggesting clinicians were more apt to use MEP in patients perceived to be sicker or in patients with increased diagnostic uncertainty. While MEP might be useful in narrowing empiric antimicrobial therapy in patients with clinical suspicion for bacterial meningitis, it did not show utility in shortening LOS or LoT in this high-risk group of patients without laboratory evidence of bacterial CSF infection.
In our cohort, LOS decreased over time corresponding with changing clinical practice toward shorter LOS for young infants with temperature instability. MEP use, however, was associated with longer LOS and longer LoT, even when controlling for other variables including year of admission. These findings contradict previous studies including older pediatric patients which demonstrated shorter LOS and LoT with MEP performed [6–10]. The limited available literature investigating MEP use in evaluation of young infants with temperature instability includes two studies that evaluated use of the molecular testing in this patient population presenting to the emergency department [10, 12]. Crook, et al. investigated the combined impact of respiratory multiplex and MEP testing plus guidelines for the management of infants with temperature instability. Adjusting for temporal trends in management, they noted no difference in antimicrobial duration [12], though the impact of MEP versus respiratory multiplex testing was not delineated. Despain, et al. found no difference in LOS or acyclovir duration for febrile, well-appearing infants aged ≤ 60-days without focal or systemic bacterial infection but excluded hypothermic or ill-appearing infants [10]. Similarly, our study, with a focus on young infants not selected for a strong clinical suspicion of CSF infection but including hypothermic and ill-appearing infants, showed longer LOS and LoT with use of MEP.
In our cohort, MEP was more likely to be used in patients with higher CSF TNC and higher CSF protein. This suggests that clinician concern for bacterial meningitis or increased severity of patient illness resulted in use of the MEP to guide clinical care. Even in the propensity matched analysis, MEP use did not result in shorter LOS or LoT, suggesting that uncertainty about test accuracy and/or clinician concern for serious illness led to longer LOS and LoT in spite of negative MEP results.
While MEP was frequently performed in patients who were treated with antimicrobials prior to obtaining CSF for culture (“pre-treated CSF”), there was no apparent benefit in shortening either LOS or LoT. For patients with pre-treated CSF, MEP provides a potential mechanism to identify an infectious source not identified on culture, but clinical use in this setting requires a test with low rates of false-negative results. Several studies have evaluated discordance in results of MEP and CSF cultures [12–15]. Tansarli and Chapman demonstrated a 2% false-negative and 11% false-positive rate of MEP with the highest discordance noted for Streptococcus pneumoniae, Streptococcus agalactiae, and HSV 1/2 [15]. Perhaps it is concern about test accuracy that could explain the lack of MEP benefit on LOS or LoT in our cohort of patients with pre-treated CSF.
In addition to longer LOS and LoT, our study results demonstrated that MEP use was not associated with decreased additional mPCR testing. One possible explanation is that providers lack sufficient knowledge or guidelines for MEP utilization. Rajbhandari, et al. performed a national survey of providers regarding MEP use in pediatric patients. Among survey participants reporting use of MEP, they found 4% still used mPCRs in all scenarios, over half still used mPCRs in certain scenarios, and three-quarters of participants have no institutional guidelines regarding use or interpretation of MEP results [16]. In conjunction with education of ordering providers, Crook, et al. implemented clinical guidelines for febrile, young infants including use of both respiratory and meningoencephalitis multiplex panels which resulted in an overall decrease in ancillary testing performed [17]. In our study, MEP use increased over the years of the study, suggesting evolving clinician awareness of its availability. Still, in our cohort, MEP use was associated with increased ancillary testing, perhaps suggesting an opportunity for diagnostic stewardship related to mPCR and MEP use in this patient population.
Despite our institution limiting MEP use to samples of CSF with abnormalities, our cohort demonstrated significantly longer LOS and LoT for patients with MEP suggesting abnormal CSF alone is not an effective diagnostic stewardship gateway. This is supported by Precit, et al. who found abnormal CSF biomarkers had a poor predictive value and sensitivity for predicting positive MEP results [18]. If CSF biomarkers cannot serve as a gatekeeper, perhaps the clinical scenario may dictate when MEP is most useful. When Despain, et al. excluded patients diagnosed with gastroenteritis or bronchiolitis, MEP use was associated with longer LOS and antimicrobial LoT [19]. Potential explanations for these findings may be that MEP use was most beneficial in confirming clinician suspicion for an alternative explanation of temperature instability or that MEP provided little benefit in cases of diagnostic uncertainty. While abnormal CSF may be a reasonable gateway for MEP use, the variables in the current study suggest other factors controlling LOS or LoT are still heavily weighted by clinicians. These other factors might represent clinician’s assessment of a patient’s severity of illness, and diagnostic uncertainty. This may explain why MEP was performed on only a subset of those patients that met eligibility for the testing. However, inherent to the nature of a retrospective cohort review of this size relying heavily on concrete data extracted from the electronic health record, the exact factors influencing clinician decision-making remain unmeasured.
The strengths of our study include the large cohort of patients. The long timeframe of the study period allows for observation of secular trends of LOS and LoT and puts MEP use in the context of changing practice patterns. There are, however, a few study limitations. First, well- versus ill-appearing infants were not distinguished, which may have influenced decision-making in which patients had MEP performed. Second, alternate diagnoses that may impact LOS were not delineated. Lastly, the study’s retrospective nature precludes randomization so selection bias may impact the outcomes. Despite these limitations, our study provides important data regarding use of this diagnostic test in young infants with temperature instability.
Conclusion
In our study, infants aged ≤ 60-days hospitalized with temperature instability and had the MEP performed compared to those without had longer LOS, longer LoT, and more molecular testing performed. Our findings support the need for additional research focusing on specific clinical scenarios where MEP use may be most beneficial for young infants with temperature instability.
Acknowledgements
Not applicable.
Abbreviations
- CBC
Complete blood cell count
- CI
Confidence interval
- CRP
C-reactive protein
- CSF
Cerebrospinal fluid
- HSV
Herpes simplex virus
- IQR
Interquartile range
- LOS
Length of stay
- LoT
Length of therapy
- MEP
Meningoencephalitis panel
- mPCR
Monoplex polymerase chain reaction
- N
Number
- OR
Odds ratio
- PCR
Polymerase chain reaction
- RBC
Red blood cell
- TNC
Total nucleated cell
Author contributions
BP: conceptualization, methodology, investigation, data curation, formal analysis, writing original draft preparation, writing review and editing, visualizationPH: supervision, data curation, formal analysis, writing original draft preparation, writing review and editingMM: conceptualization, methodology, investigation, data curation, writing review and editingSS: conceptualization, methodology, investigation, data curation, writing review and editingAP: formal analysis, data curation, writing review and editingSAC: formal analysis, data curation, writing review and editingGB: data extraction, organizing, and quality assuranceAH: conceptualization, supervision, formal analysis, writing original draft preparation, writing review and editing.
Funding
This work was supported by the Medical College of Wisconsin, Department of Pediatrics.
Data availability
The dataset generated and analyzed during the current study are not publicly available as it contains protected health information, but a de-identified dataset is available from the corresponding author on reasonable request.
Declarations
Ethics approval and consent to participate
The need for informed consent was waived by the Children’s Wisconsin Human Research Protection Program Institutional Review Board because of the retrospective nature of the study.
Financial disclosure statement
Dr. Huppler received an honorarium from Elsevier for a Clinical Overview on Fever of Unknown Origin in Children. This honorarium was not associated with this study.
Consent for publication
Not applicable.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Biondi EA, Lee B, Ralston SL, Winikor JM, Lynn JF, Dixon A, McCulloh R. Prevalence of bacteremia and bacterial meningitis in febrile neonates and infants in the second month of life: A systematic review and Meta-analysis. JAMA Netw Open. 2019;2(3):e190874. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Raffaele JL, Sharma M, Berger S, Mitchell M, Lee C, Morrison J, et al. Prevalence of invasive bacterial infection in hypothermic young infants: A multisite study. J Pediatr. 2023;258:113407. [DOI] [PubMed] [Google Scholar]
- 3.Perry MC, Yaeger SK, Noorbakhsh K, Cruz AT, Hickey RW. Hypothermia in young infants: frequency and yield of sepsis workup. Pediatr Emerg Care. 2021;37(8):e449–55. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Pantell RH, Roberts KB, Adams WG, Dreyer BP, Kuppermann N, O’Leary ST et al. Evaluation and management of well-appearing febrile infants 8 to 60 days old. Pediatrics. 2021;148(2). [DOI] [PubMed]
- 5.Baraff LJ, Bass JW, Fleisher GR, Klein JO, McCracken GH Jr., Powell KR, Schriger DL. Practice guideline for the management of infants and children 0 to 36 months of age with fever without source. Agency for Health Care Policy and Research. Ann Emerg Med. 1993;22(7):1198 – 210. [DOI] [PubMed]
- 6.Mina Y, Schechner V, Savion M, Yahav D, Bilavsky E, Sorek N, et al. Clinical benefits of filmarray meningitis-encephalitis PCR assay in partially-treated bacterial meningitis in Israel. BMC Infect Dis. 2019;19(1):713. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.McDonald D, Gagliardo C, Chiu S, Di Pentima MC. Impact of a rapid diagnostic meningitis/encephalitis panel on antimicrobial use and clinical outcomes in children. Antibiot (Basel). 2020;9(11). [DOI] [PMC free article] [PubMed]
- 8.Hagen A, Eichinger A, Meyer-Buehn M, Schober T, Huebner J. Comparison of antibiotic and acyclovir usage before and after the implementation of an on-site filmarray meningitis/encephalitis panel in an academic tertiary pediatric hospital: a retrospective observational study. BMC Pediatr. 2020;20(1):56. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Posnakoglou L, Siahanidou T, Syriopoulou V, Michos A. Impact of cerebrospinal fluid syndromic testing in the management of children with suspected central nervous system infection. Eur J Clin Microbiol Infect Dis. 2020;39(12):2379–86. [DOI] [PubMed] [Google Scholar]
- 10.Nabower AM, Miller S, Biewen B, Lyden E, Goodrich N, Miller A, et al. Association of the filmarray Meningitis/Encephalitis panel with clinical management. Hosp Pediatr. 2019;9(10):763–9. [DOI] [PubMed] [Google Scholar]
- 11.Harris PA, Taylor R, Thielke R, Payne J, Gonzalez N, Conde JG. Research electronic data capture (REDCap)--a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inf. 2009;42(2):377–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Arora HS, Asmar BI, Salimnia H, Agarwal P, Chawla S, Abdel-Haq N. Enhanced identification of group B Streptococcus and Escherichia coli in young infants with meningitis using the biofire filmarray meningitis/Encephalitis panel. Pediatr Infect Dis J. 2017;36(7):685–7. [DOI] [PubMed] [Google Scholar]
- 13.Ekambaram M, Nabower A, Rajbhandari P, Eisenberg J, Goodrich N, Ampofo K, et al. Evaluation of discordant results between filmarray Meningitis/Encephalitis panel and conventional testing in pediatric patients: A multisite retrospective cohort study. J Pediatr Infect Dis Soc. 2022;11(4):134–41. [DOI] [PubMed] [Google Scholar]
- 14.Graf EH, Farquharson MV, Cardenas AM. Comparative evaluation of the filmarray meningitis/encephalitis molecular panel in a pediatric population. Diagn Microbiol Infect Dis. 2017;87(1):92–4. [DOI] [PubMed] [Google Scholar]
- 15.Tansarli GS, Chapin KC. Diagnostic test accuracy of the BioFire® FilmArray® meningitis/encephalitis panel: a systematic review and meta-analysis. Clin Microbiol Infect. 2020;26(3):281–90. [DOI] [PubMed] [Google Scholar]
- 16.Rajbhandari P, Goodrich N, Nabower AM, Brown MF, Ekambaram M, Eisenberg J, et al. Current state and practice variation in the use of Meningitis/Encephalitis (ME) filmarray panel in children. BMC Infect Dis. 2022;22(1):811. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Crook J, Xu M, Slaughter JC, Willis J, Browning W, Estrada C, et al. Impact of clinical guidance and rapid molecular pathogen detection on evaluation and outcomes of febrile or hypothermic infants. Infect Control Hosp Epidemiol. 2020;41(11):1285–91. [DOI] [PubMed] [Google Scholar]
- 18.Precit MR, Yee R, Pandey U, Fahit M, Pool C, Naccache SN, Dien Bard J. Cerebrospinal fluid findings are poor predictors of appropriate filmarray meningitis/encephalitis panel utilization in pediatric patients. J Clin Microbiol. 2020;58(3). [DOI] [PMC free article] [PubMed]
- 19.DesPain AW, Pearman R, Hamdy RF, Campos J, Badolato GM, Breslin K. Impact of CSF meningitis and encephalitis panel on resource use for febrile Well-Appearing infants. Hosp Pediatr. 2022;12(11):1002–12. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The dataset generated and analyzed during the current study are not publicly available as it contains protected health information, but a de-identified dataset is available from the corresponding author on reasonable request.