OBJECTIVES:

The Roseville Protocol modifies the Rochester Protocol by adding a high-risk temperature criterion of >38.5°C for infants 7 to 28 days old and by allowing febrile infants 29 to 60 days old with abnormal urinalysis but reassuring complete blood cell counts to be discharged home on oral antibiotics without receiving a lumbar puncture (LP). In this study, we define the Roseville Protocol test characteristics to detect invasive bacterial infection (IBI) and retrospectively compare its performance to that of the Rochester, Philadelphia, and Boston protocols.

METHODS:

In this retrospective study, we examine all cases of fever in infants aged 7 to 60 days presenting to a large health maintenance organization from 2007 to 2016 and having requisite laboratory tests for protocol analysis. The 4 protocols were retrospectively applied to this cohort to calculate each protocol’s sensitivity and specificity to detect IBI. Protocols were compared regarding recommended LPs, admissions, and parenteral antibiotics.

RESULTS:

In 627 infants 7 to 28 days old, the Roseville Protocol had a sensitivity of 96.7% and a negative predictive value of 99.5%. It identified 2 IBIs missed by the Rochester Protocol but recommended an absolute increase of 19% in LPs and admissions. In 1176 infants 29 to 60 days old, the Roseville Protocol had a sensitivity of 91.4% and a negative predictive value of 99.6%. There was an absolute reduction in LPs by 18% to 44% compared to the Rochester Protocol and by 74% to 100% compared to the Philadelphia and Boston protocols. There was an absolute reduction in admissions by 18% to 44% compared to the Rochester Protocol, by 25% to 51% compared to the Philadelphia Protocol, and by 10% to 36% compared to the Boston Protocol.

CONCLUSIONS:

The Roseville Protocol has sensitivity and specificity comparable to that of existing protocols for IBI in febrile infants 7 to 60 days old, while allowing for fewer invasive procedures and hospitalizations in infants ≥29 days old.

The management of fever in well-appearing infants <60 days old remains controversial. Clinicians often follow guidelines, such as the Rochester, Philadelphia, and Boston protocols, to determine which febrile infants require lumbar puncture (LP), hospital admission, and parenteral antibiotics (PAbx). Protocols differ in the age ranges to which they apply, exclusion criteria, laboratory parameters defining high versus low risk, and management recommendations (Table 1). Notably, the Philadelphia and Boston protocols require LP, PAbx, and admission for all infants ≤28 days old, and all 3 protocols require these for infants ≤60 days old with an abnormal urinalysis result.13 

TABLE 1

The Roseville, Rochester, Boston, and Philadelphia Protocols

RosevilleRosevilleRochester3 Boston4 Philadelphia5 
Age 7–28 d 29–60 d ≤60 d 28–89 d 29–60 d 
Temperature ≥38°C ≥38°C ≥38°C ≥38°C ≥38.2°C 
Inclusion criteria      
 Historical      
 Term infant; no maternal prenatal or perinatal fever or antibiotics; no neonatal fever or antibiotics; no technology-dependent or congenital anomalies — Term infant; no perinatal antibiotics; no antimicrobial agents; no previous admission; no underlying disease No immunizations within 48 h; no antimicrobial within 48 h Not specified 
 Examination      
 Well appearing; no evident infection — Well appearing; no ear, soft tissue, or bone infection Well appearing; no ear, soft tissue, or bone infection; not dehydrated Well appearing; unremarkable examination 
Temperature defining low risk ≤38.5°C None None None None 
Laboratory parameters defining low risk WBC count >5000 and <15 000 cells per mm3; ABC <1500 cells per mm3 (if obtained); urinalysis: <5 WBCs per HPF, negative for LE, negative for nitrites; CSF <20 WBCs per mm3 (if obtained), CSF Gram-stain negative result (if obtained) — WBC count >5000 and <15 000 cells per mm3; ABC <1500 cells per mm3 (if obtained); urinalysis: ≤10 WBCs per HPF; stool smear: ≤5 WBCs per HPF (if obtained) WBC count <20 000 cells per mm3; urinalysis: <10 WBCs per HPF; CSF: <10 WBCs per mm3; CXR: no infiltrate (if obtained) WBC count <15 000 cells per mm3; band cell/neutrophil ratio <0.2; urinalysis: <10 WBCs per HPF; urine Gram-stain negative result; CSF: <8 WBCs per mm3, CSF Gram-stain negative result; CXR: no infiltrate; stool smear: no blood, few or no WBCs (if obtained) 
Management: high risk      
 LP + hospitalize and give parenteral antibiotics (laboratory high risk) or no antibiotics (laboratory low risk + temperature high risk) CBC high risk: parenteral antibiotics + consider LP and hospitalization; CBC low risk + urinalysis high risk: home + oral antibiotics LP + hospitalize + parenteral antibiotics Hospitalize + parenteral antibiotics Hospitalize + parenteral antibiotics 
Management: low risk Home + follow-up — Home + follow-up Parenteral antibiotics + home + follow-up Home + follow-up 
RosevilleRosevilleRochester3 Boston4 Philadelphia5 
Age 7–28 d 29–60 d ≤60 d 28–89 d 29–60 d 
Temperature ≥38°C ≥38°C ≥38°C ≥38°C ≥38.2°C 
Inclusion criteria      
 Historical      
 Term infant; no maternal prenatal or perinatal fever or antibiotics; no neonatal fever or antibiotics; no technology-dependent or congenital anomalies — Term infant; no perinatal antibiotics; no antimicrobial agents; no previous admission; no underlying disease No immunizations within 48 h; no antimicrobial within 48 h Not specified 
 Examination      
 Well appearing; no evident infection — Well appearing; no ear, soft tissue, or bone infection Well appearing; no ear, soft tissue, or bone infection; not dehydrated Well appearing; unremarkable examination 
Temperature defining low risk ≤38.5°C None None None None 
Laboratory parameters defining low risk WBC count >5000 and <15 000 cells per mm3; ABC <1500 cells per mm3 (if obtained); urinalysis: <5 WBCs per HPF, negative for LE, negative for nitrites; CSF <20 WBCs per mm3 (if obtained), CSF Gram-stain negative result (if obtained) — WBC count >5000 and <15 000 cells per mm3; ABC <1500 cells per mm3 (if obtained); urinalysis: ≤10 WBCs per HPF; stool smear: ≤5 WBCs per HPF (if obtained) WBC count <20 000 cells per mm3; urinalysis: <10 WBCs per HPF; CSF: <10 WBCs per mm3; CXR: no infiltrate (if obtained) WBC count <15 000 cells per mm3; band cell/neutrophil ratio <0.2; urinalysis: <10 WBCs per HPF; urine Gram-stain negative result; CSF: <8 WBCs per mm3, CSF Gram-stain negative result; CXR: no infiltrate; stool smear: no blood, few or no WBCs (if obtained) 
Management: high risk      
 LP + hospitalize and give parenteral antibiotics (laboratory high risk) or no antibiotics (laboratory low risk + temperature high risk) CBC high risk: parenteral antibiotics + consider LP and hospitalization; CBC low risk + urinalysis high risk: home + oral antibiotics LP + hospitalize + parenteral antibiotics Hospitalize + parenteral antibiotics Hospitalize + parenteral antibiotics 
Management: low risk Home + follow-up — Home + follow-up Parenteral antibiotics + home + follow-up Home + follow-up 

CXR, chest radiograph; LE, leukocyte esterase.

Several recent studies reveal that urinary tract infection (UTI) carries a lower risk of sequelae than invasive bacterial infection (IBI) (bacteremia and meningitis).47  In well-appearing infants ≥29 days old, UTI may safely be treated with oral antibiotics, even with associated bacteremia,6,8  and the risk of concomitant meningitis is low.7,911  This work suggests that an abnormal urinalysis result may not mandate an LP, PAbx, and admission.

Additionally, research suggests that the IBI rate is significantly higher in infants with a fever >38.5°C,1214  suggesting that incorporating a high-risk temperature threshold may better identify infants with IBI.

In 2015, Kaiser Permanente Roseville Medical Center successfully piloted a protocol for the management of febrile infants, first presented at the Clinical Pediatric Hospital Medicine conference (San Diego; January 16–18, 2015). This protocol is based on the Rochester Protocol and has 2 modifications: infants 7 to 28 days old with a temperature >38.5°C (101.3°F) are considered high risk for IBI and should undergo full evaluation, and infants 29 to 60 days old with abnormal urinalysis results but otherwise low-risk laboratory test results can be discharged home on oral antibiotics without receiving an LP (Fig 1). In this study, we seek to validate that protocol retrospectively (hereinafter called the Roseville Protocol), and to compare its performance to that of the Rochester, Philadelphia, and Boston protocols.

FIGURE 1

The Roseville Protocol. A, Febrile infants 7 to 28 days old. B, Febrile infants 29 to 60 days old. CBC w diff, complete blood count with differential; UCx, urine culture.

FIGURE 1

The Roseville Protocol. A, Febrile infants 7 to 28 days old. B, Febrile infants 29 to 60 days old. CBC w diff, complete blood count with differential; UCx, urine culture.

Close modal

This study was approved by the Kaiser Permanente Northern California (KPNC) Institutional Review Board. Informed consent was waived.

In this study, we retrospectively compared the performance of the Roseville, Rochester, Boston, and Philadelphia protocols in 2478 febrile infants 7 to 60 days old evaluated at KPNC clinical sites from January 1, 2007, to December 31, 2016. By using the cohort of infants meeting each protocol’s own inclusion criteria, the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (PLR), and negative likelihood ratio (NLR) for IBI were calculated (with Wilson score 95% confidence intervals [CIs]), along with the percentage of infants recommended to receive an LP, PAbx, and admission. The same analysis was repeated by using the subset of infants meeting the inclusion criteria for all 4 protocols. Statistical analyses were performed by using SAS 9.4 (SAS Institute, Inc, Cary, NC).

Kaiser Permanente is the largest health care maintenance organization in the United States, with >4 million members receiving care in >40 pediatric clinics, 19 emergency departments, and 10 pediatric hospital wards in Northern California. It is a closed health care system, so most members receive all their care internally.

Term infants aged 7 to 60 days born within KPNC were included if they had a complete blood cell count (CBC), blood culture, urinalysis, and urine culture collected within a 24-hour time frame. Charts were manually reviewed for a temperature ≥38°C measured by any means at home or on presentation to the medical setting. Cases were excluded for clinical bronchiolitis, repeat episodes of fever during the same illness, or incomplete evaluation. Cases were excluded if urine was obtained by using a perineal bag. See Fig 2.

FIGURE 2

Study flow diagram, Roseville Protocol. A, Seven to 28 days old. B, Twenty-nine to 60 days old. KP, Kaiser Permanente; BCx, blood culture; UCx, urine culture.

FIGURE 2

Study flow diagram, Roseville Protocol. A, Seven to 28 days old. B, Twenty-nine to 60 days old. KP, Kaiser Permanente; BCx, blood culture; UCx, urine culture.

Close modal

Data were collected for all subjects by either electronic extraction (gestational age, age on presentation, sex, neonatal or maternal antibiotic treatment for infection [reliably available after 2010], presenting temperature, urinalysis result, serum white blood cell [WBC] count and absolute band count [ABC] [if obtained], cerebrospinal fluid [CSF] red blood cell and WBC counts and Gram-stain result [if obtained], and blood, urine, and CSF culture results) or manual chart review (well versus ill appearance, clinically evident infection, known congenital anomaly, maximum temperature at home, method of temperature measurement, maximum temperature during initial evaluation for fever, method of urine collection, and chest radiograph result [if obtained]).

UTI

UTI was defined as isolation of a single urinary bacterial pathogen with a colony count ≥50 000 colony forming units per mL.

Bacteremia

Bacteremia was defined as isolation of a bacterial pathogen from blood. Coagulase-negative Staphylococcus, viridans group Streptococcus, Micrococcus species, and diphtheroids were considered contaminants unless treated as a pathogen on the basis of guidance from a pediatric infectious diseases specialist.

Bacterial Meningitis

Bacterial meningitis was defined as definite with the isolation of a single bacterial pathogen from CSF or as probable with sterile CSF with pleocytosis (CSF WBC count ≥20 cells per high-power field [HPF]) obtained after administration of antibiotics and a diagnosis confirmed by a pediatric infectious disease specialist, according to criteria outlined by Schnadower et al.6  Infants who did not undergo an LP were presumed not to have meningitis.

Ill Appearance

To be included in the “ill” category, one of the following terms was documented in a physician examination or assessment: cold, decreased mental status, difficult to arouse, floppy, hypotonic, ill appearing, inconsolable, irritable, listless, toxic, nonresponsive, lethargic, poorly perfused, sick, or shock.

Roseville Protocol: Clinical Management of Infants

Infants 7 to 28 days old are classified as low risk if they have a temperature ≤38.5°C; have a reassuring WBC count (5–15 000 cells per mm3), ABC (<1500 cells per mm3), and urinalysis result (negative for leukocyte esterase and nitrites and WBC count <5 cells per HPF); and are discharged home without antibiotics (Fig 1). Infants classified as high risk are admitted for an LP and are further stratified into 2 management groups: those with any laboratory test result abnormality who receive PAbx and those with reassuring laboratory test results who are observed off antibiotics. Infants 29 to 60 days old are classified as low risk if they have a reassuring WBC count, ABC, and urinalysis result and are discharged home without antibiotics. High-risk infants with an abnormal urinalysis result but otherwise reassuring laboratory test results are stratified to be discharged home with oral antibiotics. High-risk infants with a nonreassuring WBC count or ABC are stratified to receive PAbx, with case-by-case consideration of LP and admission.

In Table 1, we detail inclusion and high- and low-risk criteria for all 4 protocols.

The Roseville Protocol includes term (37–43 weeks’ gestational age), well-appearing febrile infants 7 to 60 days old presenting with a temperature ≥38°C and excludes infants who are ill appearing or have a history of perinatal maternal or neonatal fever or antibiotic treatment of a presumed infection, evident infection at presentation, congenital anomalies, or technology dependence. Maternal fever or antibiotics was defined as a maternal temperature ≥38°C during delivery or antibiotic administration with Gram-negative coverage (excluding antibiotics for surgical or group B Streptococcus [GBS] prophylaxis). Neonatal fever or antibiotics was defined as a temperature ≥38°C or antibiotic administration during the newborn hospitalization. Congenital anomaly was defined as any inborn condition known on presentation with implications for infection risk. Evident infection included abscess, cellulitis, bacterial conjunctivitis, mastitis, omphalitis, and perforated otitis media.

There were 879 febrile infants 7 to 28 days old and 1599 febrile infants 29 to 60 days old who would have been eligible for all protocols. The overall prevalence of IBI was 4.5% (4.2% bacteremia and 0.73% meningitis). Organisms identified in infants with bacteremia included Escherichia coli (62); GBS (17); Staphylococcus aureus (8); Enterococcus faecalis (3); Salmonella spp (3); Streptococcus pyogenes (2); Klebsiella pneumoniae (2); and Pseudomonas aeruginosa, Citrobacter freundi, Citrobacter koseri, Enterobacter cloacae, Moraxella catarrhalis, Streptococcus pneumoniae, and Staphylococcus warnerii (1 each). Organisms identified in infants with bacterial meningitis included GBS (8), E coli (7), Neisseria meningitidis (1), K pneumoniae (1), and Salmonella spp (1).

See Table 2 for demographic information among infants who would have qualified for the Roseville Protocol.

TABLE 2

Patient Characteristics for Infants Meeting Roseville Protocol Inclusion Criteria

VariableAge 7–28 d (n = 627)Age 29–60 d (n = 1176)
Age, d, median (IQR) 19 (13–24) 44 (36–52) 
Female sex, n (%) 265 (42.3) 520 (44.2) 
Health history, n (%)   
 Gestational age, wk   
  37–38 171 (27.3) 326 (27.7) 
  39–40 379 (60.5) 746 (63.4) 
  41–42 77 (12.3) 104 (8.8) 
 Antibiotics last 48 h 1 (0.2) 4 (0.3) 
 Vaccination last 48 h 1 (0.2) 28 (2.4) 
 Previous admission 23 (3.7) 52 (4.4) 
Presentation, median (IQR)   
 Home temperaturea 38.3 (38.1–38.7) 38.3 (38.1–38.8) 
 First clinical temperature 38.2 (37.8–38.6) 38.3 (37.7–38.6) 
Laboratory test results   
 WBC count, median (IQR) 11.2 (8.0–14.7) 9.6 (6.9–13.0) 
 ABC, median (IQR)b 0.4 (0.13–0.98) 0.27 (0.07–0.76) 
 Urinalysis result positive, n (%) 201 (32.1) 316 (26.9) 
LP, n (%)   
 Yes 364 (58.1) 314 (26.7) 
 No 81 (12.9) 754 (64.1) 
 Complicated 182 (29.0) 108 (9.2) 
Culture results, n (%)   
 UTI alone 142 (22.6) 201 (17.1) 
 Bacteremia alone 10 (1.6) 10 (0.9) 
 Meningitis alone 3 (0.5) 
 UTI + bacteremia 16 (2.6) 23 (2.0) 
 UTI + meningitis 1 (0.1) 
 Bacteremia + meningitis 1 (0.1) 
 UTI + bacteremia + meningitis 1 (0.2) 
VariableAge 7–28 d (n = 627)Age 29–60 d (n = 1176)
Age, d, median (IQR) 19 (13–24) 44 (36–52) 
Female sex, n (%) 265 (42.3) 520 (44.2) 
Health history, n (%)   
 Gestational age, wk   
  37–38 171 (27.3) 326 (27.7) 
  39–40 379 (60.5) 746 (63.4) 
  41–42 77 (12.3) 104 (8.8) 
 Antibiotics last 48 h 1 (0.2) 4 (0.3) 
 Vaccination last 48 h 1 (0.2) 28 (2.4) 
 Previous admission 23 (3.7) 52 (4.4) 
Presentation, median (IQR)   
 Home temperaturea 38.3 (38.1–38.7) 38.3 (38.1–38.8) 
 First clinical temperature 38.2 (37.8–38.6) 38.3 (37.7–38.6) 
Laboratory test results   
 WBC count, median (IQR) 11.2 (8.0–14.7) 9.6 (6.9–13.0) 
 ABC, median (IQR)b 0.4 (0.13–0.98) 0.27 (0.07–0.76) 
 Urinalysis result positive, n (%) 201 (32.1) 316 (26.9) 
LP, n (%)   
 Yes 364 (58.1) 314 (26.7) 
 No 81 (12.9) 754 (64.1) 
 Complicated 182 (29.0) 108 (9.2) 
Culture results, n (%)   
 UTI alone 142 (22.6) 201 (17.1) 
 Bacteremia alone 10 (1.6) 10 (0.9) 
 Meningitis alone 3 (0.5) 
 UTI + bacteremia 16 (2.6) 23 (2.0) 
 UTI + meningitis 1 (0.1) 
 Bacteremia + meningitis 1 (0.1) 
 UTI + bacteremia + meningitis 1 (0.2) 

ABC, absolute band count; IQR, interquartile range.

a

Missing data for home temperature: age 7–28 d, n = 100; age 29–60 d, n = 178.

b

Missing data for ANC: age 7–28 d, n = 1; age 29–60 d, n = 7.

Infants 7 to 28 Days Old

Of 879 febrile infants 7 to 28 days old, 627 would have met Roseville Protocol inclusion criteria (Fig 2). The Roseville Protocol would have classified 437 infants (69.7%) as high risk and 190 (30.3%) infants as low risk. IBI was identified in 6.6% of high-risk and 0.5% of low-risk infants. Among high-risk infants, all would have received an LP. Among these, 353 (80.8%) would have been further stratified by nonreassuring laboratory test results for admission with PAbx, and of these, 150 had UTI (42.4%), 25 had bacteremia (7.1%), and 4 had meningitis (1.1%). Eighty-four (19.2%) high-risk infants with a fever >38.5°C but reassuring laboratory test results (including CSF analysis) would have been stratified to admission and observation without antibiotics. In this group, 4 had UTI (4.8%), 1 had E coli bacteremia (1.2%), and 0 had meningitis. Among low-risk infants, 5 had UTI (2.6%), 1 had GBS bacteremia (0.5%), and 0 had meningitis.

The Roseville Protocol test characteristics for IBI were was follows: sensitivity, 96.7% (95% CI 82.8%–99.9%); specificity, 31.7% (95% CI 27.9%–35.6%); PPV, 6.6% (95% CI 6.1%–7.2%); NPV, 99.5% (95% CI 96.5%–99.9%); PLR, 1.41 (95% CI 1.30–1.54); and NLR, 0.11 (95% CI 0.02–0.73). LP and admission would have been recommended in 70% of cases, and PAbx would have been recommended in 56% of cases (Table 3).

TABLE 3

Comparison of the Roseville, Rochester, Philadelphia, and Boston Protocols: Test Characteristics for the Detection of IBI and Recommended Interventions

Total, NHigh RiskLow RiskTest CharacteristicsRecommended Interventions
nBacteremia, n (%)Meningitis, n (%)nBacteremia, n (%)Meningitis, n (%)Sensitivity, % (95% CI)Specificity, % (95% CI)PPV, % (95% CI)NPV, % (95% CI)PLR (95% CI)NLR (95% CI)LP, %PAbx, %Admit, %
7–28 d                 
 Roseville 627 437 26 (5.9) 4 (0.9) 190 1 (0.5) 0 (0) 96.7 (82.8–99.9) 31.7 (27.9–35.6) 6.6 (6.1–7.2) 99.5 (96.5–99.9) 1.41 (1.30–1.54) 0.11 (0.02–0.73) 70 56 70 
 Rochester 639 327 25 (7.6) 3 (0.9) 312 2 (0.6) 1 (0.3) 90.0 (73.5–97.9) 50.7 (46.7–54.8) 8.3 (7.2–9.4) 99.0 (97.2–99.7) 1.83 (1.58–2.11) 0.20 (0.07–0.58) 51 51 51 
29–60 d                 
 Roseville 1176 505 31 (6.1) 2 (0.4) 671 3 (0.4) 0 (0) 91.4 (76.9–98.2) 58.5 (55.6–61.4) 6.3 (5.7–7.1) 99.6 (98.7–99.9) 2.21 (1.95–2.49) 0.15 (0.05–0.43) 0–26 26 0–26 
 Rochester 1209 537 29 (5.4) 2 (0.4) 672 2 (0.3) 0 (0) 93.8 (79.2–99.2) 56.9 (54.0–59.8) 5.6 (5.0–6.2) 99.7 (98.9–99.9) 2.18 (1.95–2.43) 0.11 (0.03–0.42) 44 44 44 
 Philadelphia 1448 732 34 (4.6) 3 (0.4) 715 3 (0.4) 0 (0) 92.3 (79.1–98.4) 50.6 (47.9–53.2) 4.9 (4.5–5.4) 99.6 (98.8–99.9) 1.87 (1.68–2.07) 0.15 (0.05–0.45) 100 51 51 
 Boston 1365 493 29 (5.9) 3 (0.6) 871 5 (0.6) 0 (0) 86.1 (70.5–95.3) 65.2 (62.6–67.8) 6.3 (5.5–7.2) 99.4 (98.7–99.7) 2.48 (2.13–2.88) 0.21 (0.09–0.48) 100 100 36 
Total, NHigh RiskLow RiskTest CharacteristicsRecommended Interventions
nBacteremia, n (%)Meningitis, n (%)nBacteremia, n (%)Meningitis, n (%)Sensitivity, % (95% CI)Specificity, % (95% CI)PPV, % (95% CI)NPV, % (95% CI)PLR (95% CI)NLR (95% CI)LP, %PAbx, %Admit, %
7–28 d                 
 Roseville 627 437 26 (5.9) 4 (0.9) 190 1 (0.5) 0 (0) 96.7 (82.8–99.9) 31.7 (27.9–35.6) 6.6 (6.1–7.2) 99.5 (96.5–99.9) 1.41 (1.30–1.54) 0.11 (0.02–0.73) 70 56 70 
 Rochester 639 327 25 (7.6) 3 (0.9) 312 2 (0.6) 1 (0.3) 90.0 (73.5–97.9) 50.7 (46.7–54.8) 8.3 (7.2–9.4) 99.0 (97.2–99.7) 1.83 (1.58–2.11) 0.20 (0.07–0.58) 51 51 51 
29–60 d                 
 Roseville 1176 505 31 (6.1) 2 (0.4) 671 3 (0.4) 0 (0) 91.4 (76.9–98.2) 58.5 (55.6–61.4) 6.3 (5.7–7.1) 99.6 (98.7–99.9) 2.21 (1.95–2.49) 0.15 (0.05–0.43) 0–26 26 0–26 
 Rochester 1209 537 29 (5.4) 2 (0.4) 672 2 (0.3) 0 (0) 93.8 (79.2–99.2) 56.9 (54.0–59.8) 5.6 (5.0–6.2) 99.7 (98.9–99.9) 2.18 (1.95–2.43) 0.11 (0.03–0.42) 44 44 44 
 Philadelphia 1448 732 34 (4.6) 3 (0.4) 715 3 (0.4) 0 (0) 92.3 (79.1–98.4) 50.6 (47.9–53.2) 4.9 (4.5–5.4) 99.6 (98.8–99.9) 1.87 (1.68–2.07) 0.15 (0.05–0.45) 100 51 51 
 Boston 1365 493 29 (5.9) 3 (0.6) 871 5 (0.6) 0 (0) 86.1 (70.5–95.3) 65.2 (62.6–67.8) 6.3 (5.5–7.2) 99.4 (98.7–99.7) 2.48 (2.13–2.88) 0.21 (0.09–0.48) 100 100 36 

Infants 29 to 60 Days Old

Of 1599 febrile infants 29 to 60 days old, 1176 would have met Roseville Protocol inclusion criteria (Fig 2).

The Roseville Protocol classified 505 (42.9%) infants as high risk and 671 (57.1%) infants as low risk. IBI was detected in 6.3% of high-risk and 0.45% of low-risk infants. Of 310 high-risk infants stratified by a nonreassuring CBC to receive PAbx, 101 had UTI (32.6%), 20 had bacteremia (6.5%, including 14 with bacteremic UTI), and 1 had meningitis (0.3%). Of 195 high-risk infants with abnormal urinalysis results but a reassuring CBC who were stratified to discharge on oral antibiotics, 109 had UTI (55.9%), 11 had bacteremia (5.6%, including 9 with bacteremic UTI), and 1 had meningitis (a single colony of E coli, considered equivocal but treated as meningitis; 0.5%). Among low-risk infants, 15 had UTI (2.2%), 3 had bacteremia (0.45%), and 0 had meningitis.

The Roseville Protocol test characteristics for IBI were as follows: sensitivity, 91.4% (95% CI 76.9%–98.2%); specificity, 58.5% (95% CI 55.6%–61.4%); PPV, 6.3% (95% CI 5.7%–7.1%); NPV, 99.6% (95% CI 98.7%–99.9%); PLR, 2.21 (95% CI 1.95–2.49); and NLR, 0.15 (95% CI 0.05–0.43). LP and admission would have been recommended in 0% to 26% of cases, and PAbx would have been recommended in 26% of cases (Table 3).

See Table 3 for a comparison of test characteristics for detection of IBI and interventions that would have been recommended by the Rochester (7–60 days), Philadelphia (29–60 days), and Boston (29–60 days) protocols.

The Roseville Protocol had sensitivity and NPV comparable to that of the Rochester Protocol in the 7- to 28-day-old group (Roseville: sensitivity 96.7%, NPV 99.5%; Rochester: sensitivity 90.0%, NPV 99.0%) but inferior specificity and PPV (Roseville: specificity 31.7%, PPV 6.6%; Rochester: specificity 50.7%, PPV 8.3%). Sensitivity and NPV were also similar between the 4 protocols in the 29- to 60-day-old group (Roseville: sensitivity 91.4%, NPV 99.6%; Rochester: sensitivity 93.8%, NPV 99.7%; Philadelphia: sensitivity 92.3%, NPV 99.6%; Boston: sensitivity 86.1%, NPV 99.4%). The Roseville and Rochester protocols had similar specificity (Roseville specificity 58.5%; Rochester specificity 56.9%), but the Philadelphia Protocol was inferior (50.6%), and the Boston Protocol was superior (65.2%).

In the 7- to 28-day-old age range, the Roseville Protocol would have recommended an LP and admission for 70% of cases and PAbx for 56% of cases compared to the Rochester Protocol, in which an LP, admission, and PAbx would have been recommended for 51% of cases. In the 29- to 60-day-old age range, the Roseville Protocol recommended an LP, admission, and PAbx in up to 26% of cases, compared to 51% as recommended by the Rochester Protocol. The Philadelphia Protocol would have recommended an LP for all cases and PAbx and admission for 51% of cases. The Boston Protocol would have recommended an LP and PAbx for all cases and admission for 36% of cases.

Test characteristics for the 4 protocols did not differ when analyzed by using the subset of infants meeting inclusion criteria for all 4 protocols, compared to when analyzed by using each protocol’s own inclusion criteria (Supplemental Table 4).

All 4 protocols misclassified a small number of infants with IBI as low risk (Table 3 and Supplemental Table 5).

Every febrile infant protocol strives to minimize the harms of overtesting and overtreatment while maximizing sensitivity for IBI. The Roseville Protocol is unique in its incorporation of temperature >38.5°C as a risk criterion and is the first protocol to allow infants >28 days old with abnormal urinalysis results and otherwise low-risk laboratory test results to be discharged home. In this retrospective analysis, it performed comparably to the Rochester, Philadelphia, and Boston protocols, with the benefit of reduced interventions in the 29- to 60-day-old age group.

Compared to the Rochester Protocol, the inclusion of a high-risk temperature criterion (>38.5°C) for 7- to 28-day-old infants in the Roseville Protocol revealed a trend toward increased sensitivity for IBI (96.7% vs 90.0%) at the expense of specificity (31.7% vs 58.5%) and an absolute increase in recommended interventions (19% more infants would undergo an LP and hospitalization, and 5% more infants would receive PAbx). The temperature criterion allowed the Roseville Protocol to correctly identify 1 case each of meningitis and bacteremia, which were not identified by the Rochester Protocol.

By discharging infants with abnormal urinalysis results but otherwise reassuring laboratory test results on oral antibiotics, LPs and PAbx would have been reduced in infants 29 to 60 days old, while sensitivity and NPV comparable to that of the Rochester, Philadelphia, and Boston protocols would have been retained. The Roseville Protocol does not direct which infants with a nonreassuring WBC count or ABC require an LP or hospitalization, allowing room for clinical judgment depending on multiple factors, such as the degree of laboratory abnormality, social milieu, and availability of close follow-up. Should every infant in this category undergo these interventions, there would be an absolute reduction of 18% compared to the Rochester Protocol and even greater reductions compared to the Philadelphia and Boston protocols. Given the 40% to 50% failure rate for LPs15  and the resultant diagnostic uncertainty, reducing the number of unneeded LPs would independently reduce unnecessary treatment, resource use, and family disruption.

Recent studies reveal that infants with bacteremic UTI are not at an increased risk of meningitis.6,7,911  However, the presence of bacteremia may prompt some clinicians to perform an LP to evaluate for meningitis. Of the 23 infants 29 to 60 days old with bacteremic UTI, 18 underwent an LP, 15 had a recorded CSF WBC count, 2 had CSF pleocytosis, and 0 were diagnosed with bacterial meningitis. The rare scenario of missed CSF pleocytosis would be more than balanced out by the large reduction in infants requiring an LP, admission, and PAbx in this age group.

This study is limited by the rarity of bacterial meningitis (<1% in our cohort). The retrospective design limited the precision of variables, such as clinical appearance, which relied on chart review of documentation not specifically intended for research. Like the Rochester Protocol, the Roseville Protocol excludes a significant proportion of infants on the basis of historical or clinical criteria: 28.6% of infants 7 to 28 days old and 26.5% of infants 29 to 60 days old. Further study is underway to determine if these exclusions are justified. Finally, unlike recently derived clinical prediction rules, the Roseville Protocol is not statistically derived and instead uses predefined cutoffs. Additional study is underway to determine optimal cutoffs and to determine if the addition of a temperature criterion in the 29- to 60-day-old age group would improve protocol performance.

In many recent studies, researchers are focusing on additional laboratory biomarkers, such as absolute neutrophil count, C-reactive protein, procalcitonin, and RNA biosignatures, in the evaluation of febrile young infants.1618  It is unclear how much value the additional biomarkers add, especially when they require a rapid turnaround time to be clinically useful. The Step-by-Step Protocol requires both C-reactive protein and procalcitonin and has sensitivity and NPV comparable to that of the Roseville Protocol. A clinical prediction rule outlined by Kuppermann et al17  revealed slightly better sensitivity and specificity than the Roseville Protocol but continues to require LPs in patients with positive urinalysis results. With clinical assessment and widely available laboratory tests, the Roseville Protocol offers an effective alternative approach to the febrile young infant.

Following the Roseville Protocol, febrile infants 7 to 28 days old may be risk-stratified to be discharged home without a LP or antibiotics; the addition of a high-risk temperature criterion (>38.5°C) leads to a trend toward improved sensitivity and NPV compared to the Rochester Protocol. Febrile infants 29 to 60 days old with abnormal urinalysis results may safely be discharged home on oral antibiotics with fewer interventions and no loss in sensitivity for IBI as compared to the Rochester, Philadelphia, and Boston protocols.

Drs Nguyen, Young, and Poggel conceptualized and designed the study, collected data, drafted the initial manuscript, and reviewed and revised the manuscript; Ms Alabaster designed the data collection instruments, coordinated data collection, conducted the initial analyses, and reviewed and revised the manuscript; Dr Greenhow advised on the study design, collected data, reviewed diagnoses of bacterial meningitis, and critically reviewed and revised the manuscript; and all authors approved the final manuscript as submitted and agree to be accountable for all aspects of the work.

FUNDING: Supported by a grant from the Kaiser Permanente Northern California Community Benefit Program.

1
Jaskiewicz
JA
,
McCarthy
CA
,
Richardson
AC
, et al
;
Febrile Infant Collaborative Study Group
.
Febrile infants at low risk for serious bacterial infection–an appraisal of the Rochester criteria and implications for management
.
Pediatrics
.
1994
;
94
(
3
):
390
396
2
Bachur
RG
,
Harper
MB
.
Predictive model for serious bacterial infections among infants younger than 3 months of age
.
Pediatrics
.
2001
;
108
(
2
):
311
316
3
Baker
MD
,
Bell
LM
,
Avner
JR
.
Outpatient management without antibiotics of fever in selected infants
.
N Engl J Med
.
1993
;
329
(
20
):
1437
1441
4
de Louvois
J
,
Halket
S
,
Harvey
D
.
Neonatal meningitis in England and Wales: sequelae at 5 years of age
.
Eur J Pediatr
.
2005
;
164
(
12
):
730
734
5
Pruitt
CM
,
Neuman
MI
,
Shah
SS
, et al
;
Febrile Young Infant Research Collaborative
.
Factors associated with adverse outcomes among febrile young infants with invasive bacterial infections
.
J Pediatr
.
2019
;
204
:
177
182.e1
6
Schnadower
D
,
Kuppermann
N
,
Macias
CG
, et al
;
American Academy of Pediatrics Pediatric Emergency Medicine Collaborative Research Committee
.
Febrile infants with urinary tract infections at very low risk for adverse events and bacteremia
.
Pediatrics
.
2010
;
126
(
6
):
1074
1083
7
Young
BR
,
Nguyen
THP
,
Alabaster
A
,
Greenhow
TL
.
The prevalence of bacterial meningitis in febrile infants 29-60 days with positive urinalysis
.
Hosp Pediatr
.
2018
;
8
(
8
):
450
457
8
Hoberman
A
,
Wald
ER
,
Hickey
RW
, et al
.
Oral versus initial intravenous therapy for urinary tract infections in young febrile children
.
Pediatrics
.
1999
;
104
(
1, pt 1
):
79
86
9
Vuillermin
PJ
,
Starr
M
.
Investigation of the rate of meningitis in association with urinary tract infection in infants 90 days of age or younger
.
Emerg Med Australas
.
2007
;
19
(
5
):
464
469
10
Paquette
K
,
Cheng
MP
,
McGillivray
D
,
Lam
C
,
Quach
C
.
Is a lumbar puncture necessary when evaluating febrile infants (30 to 90 days of age) with an abnormal urinalysis?
Pediatr Emerg Care
.
2011
;
27
(
11
):
1057
1061
11
Tebruegge
M
,
Pantazidou
A
,
Clifford
V
, et al
.
The age-related risk of co-existing meningitis in children with urinary tract infection
.
PLoS One
.
2011
;
6
(
11
):
e26576
12
Pantell
RH
,
Newman
TB
,
Bernzweig
J
, et al
.
Management and outcomes of care of fever in early infancy
.
JAMA
.
2004
;
291
(
10
):
1203
1212
13
Davis
J
,
Lehman
E
.
Fever characteristics and risk of serious bacterial infection in febrile infants
.
J Emerg Med
.
2019
;
57
(
3
):
306
313
14
Carmon
L
,
Goldbart
A
,
Greenberg
D
,
Ben-Shimol
S
.
Serious bacterial infections in hospitalized febrile infants in the first and second months of life
.
Pediatr Infect Dis J
.
2017
;
36
(
10
):
924
929
15
Neal
JT
,
Kaplan
SL
,
Woodford
AL
,
Desai
K
,
Zorc
JJ
,
Chen
AE
.
The effect of bedside ultrasonographic skin marking on infant lumbar puncture success: a randomized controlled trial
.
Ann Emerg Med
.
2017
;
69
(
5
):
610
619.e1
16
Gomez
B
,
Mintegi
S
,
Bressan
S
,
Da Dalt
L
,
Gervaix
A
,
Lacroix
L
;
European Group for Validation of the Step-by-Step Approach
.
Validation of the “step-by-step” approach in the management of young febrile infants
.
Pediatrics
.
2016
;
138
(
2
):
e20154381
17
Kuppermann
N
,
Dayan
PS
,
Levine
DA
, et al
;
Febrile Infant Working Group of the Pediatric Emergency Care Applied Research Network (PECARN)
.
A clinical prediction rule to identify febrile infants 60 days and younger at low risk for serious bacterial infections
.
JAMA Pediatr
.
2019
;
173
(
4
):
342
351
18
Mahajan
P
,
Kuppermann
N
,
Mejias
A
, et al
;
Pediatric Emergency Care Applied Research Network (PECARN)
.
Association of RNA biosignatures with bacterial infections in febrile infants aged 60 days or younger
.
JAMA
.
2016
;
316
(
8
):
846
857

Competing Interests

POTENTIAL CONFLICT OF INTEREST: The authors have indicated they have no potential conflicts of interest to disclose.

FINANCIAL DISCLOSURE: The authors have indicated they have no financial relationships relevant to this article to disclose.

Supplementary data