Children Hospitalized With Pneumonia Who Complete Oral Antibiotic Treatment Without Complications

Abstract

OBJECTIVE:

While initial oral antibiotics are likely as effective as intravenous (IV) antibiotics for most children hospitalized with community-acquired pneumonia (CAP), most still receive IV therapy. We evaluated factors associated with doing well with initial oral antibiotics among children hospitalized with CAP.

METHODS:

We performed a multicenter retrospective cohort study of children hospitalized with CAP who were started on oral antibiotics at 4 children’s hospitals between 2014 and 2020. We performed multivariable logistic regression to evaluate the association between hospital site and patient-level factors and doing well with oral antibiotics, defined as those who remained on oral antibiotics throughout the hospitalization, did not have escalation of care during the hospitalization, and did not have readmissions or emergency department revisits within 7 days of discharge.

RESULTS:

Among 419 children who received initial oral antibiotics, 76% (n = 317) did well with this regimen (ie, remained on orals and did not have escalation or revisits). Of those who did not do well (n = 102), 80% (n = 82) switched to IV antibiotics but had no escalation of care or revisits. In multivariable analyses, there were no patient-level factors associated with doing well with initial oral antibiotics, and only hospital site was statistically significant.

CONCLUSIONS:

Most children hospitalized with CAP given initial oral antibiotics continued oral therapy and did not have escalation of care or return visits. Switching from oral to IV antibiotics may be based more on nonpatient level factors (eg, clinician preference) than patient-level factors, although prospective studies are needed.

INTRODUCTION

Community-acquired pneumonia (CAP) accounts for more antibiotic days than any other infection among hospitalized children.¹ Clinicians can treat these children with intravenous (IV) or oral antibiotics. Three randomized controlled trials outside the United States suggest that initial oral antibiotics are as effective as IV antibiotics for many children hospitalized with CAP.^2–4 In a US cohort, we previously identified characteristics associated with receipt of oral vs IV antibiotics and similarly found that children treated initially with orals had shorter length of stay (LOS) without increased complications.⁵ Oral antibiotics can also reduce IV line-related pain/anxiety, harms (eg, extravasation), LOS, and costs.^6–14 Despite literature suggesting that oral antibiotics are a reasonable initial approach, given similar effectiveness to IV therapy and the additional benefits of oral antibiotics, most clinicians are not taking this approach. In a recent study of more than 30 000 children across 43 US hospitals, 84% received initial IV antibiotics.¹⁵ Identifying characteristics that might predict success or failure with initial oral antibiotics may be important for clinicians when deciding whether to treat these children with oral or IV antibiotics. We aimed to identify factors associated with remaining on oral antibiotics without complications among children hospitalized with CAP.

METHODS

Study Design and Population

We conducted a retrospective cohort study of children (aged 3 months-18 years) admitted to 4 children’s hospitals (1/1/2014–2/28/2020) with a diagnosis of CAP, chest radiograph (CXR) within 24 hours of arrival, and receipt of antibiotics for CAP (Supplemental Table 1). Definition of CAP was based on a validated algorithm that includes discharge diagnosis codes for viral, bacterial, or indeterminant CAP (Supplemental Table 2).¹⁶ In line with prior studies, we included patients with viral CAP codes, along with our requirement that they received antibiotics for CAP, to pragmatically capture children whom clinicians thought had bacterial CAP.^5,15 By requiring performance of CXRs, which is common for inpatient CAP,¹⁷ we captured children presenting with CAP rather than acquiring it later. We excluded children with complex chronic conditions,¹⁷ LOS of less than 48 hours, initial admission to the intensive care unit (ICU), moderate-to-large pleural effusions, antibiotics for atypical CAP only, or concomitant bacterial infections (Supplemental Table 3). Exclusion of stays of less than 48 hours ensured sufficient time to assess in-hospital outcomes resulting from antibiotic exposure rather than a reflection of illness severity on presentation.^5,18 Data were collected from the Pediatric Health Information System (PHIS) database (Children’s Hospital Association, Lenexa, KS) and each hospital’s electronic health record with study-specific training, validation, and quality checks.⁵ This study was deemed exempt from human subjects’ research by each site’s institutional review board.

Outcome

The outcome was “doing well” with initial oral antibiotics, which was defined as remaining on oral antibiotics without switching to IV during hospitalization and without pneumonia-related complications. Complications included escalation of care (ie, transfer to ICU after 48 hours, chest tube placed after 48 hours, or escalation in respiratory support after 48 hours) or CAP-related readmission/emergency department (ED) revisit within 7 days of discharge. Escalation of respiratory support was defined as moving from one level of respiratory support (ie, no oxygen, low flow, high flow, noninvasive positive pressure ventilation, or mechanical ventilation) to higher level after 48 hours. These complications, potentially indicative of treatment failure, align with prior CAP studies.^5,15,19 Children who switched to IV antibiotics or had complications were classified as “not doing well” with initial oral therapy.

Risk Factors

We assessed hospital site and patient factors potentially associated with illness severity on presentation or likelihood of bacterial CAP: age, fever presence/duration, antibiotic use in the prior 7 days, CXR findings (radiographic pneumonia vs equivocal vs negative), asthma or bronchiolitis codiagnosis, oxygen use in ED, time to initial antibiotic, and initial broad-spectrum antibiotic use (ie, anything other than amoxicillin).^20–25

Statistics

We examined the proportion of children with initial oral antibiotics (ie, received oral antibiotics as their initial antibiotic dose in the ED or inpatient setting, wherever antibiotics were first started) by site. Remaining analyses were conducted among children with initial oral antibiotics. We compared characteristics of children who did vs did not do well with initial oral antibiotics using χ² and Wilcoxon rank-sum tests. Multivariable logistic regression assessed associations between hospital/patient factors and doing well with oral antibiotics. Results were presented as adjusted odds ratios (aORs) with 95% CIs. We performed 4 sensitivity analyses: (1) excluded patients with viral CAP diagnosis code or virus detected on respiratory pathogen panel if performed as part of clinical care, (2) excluded patients aged younger than 2 years, (3) removed hospital site from model, and (4) redefined outcome to switching to IV antibiotics.

RESULTS

Among 1137 children hospitalized with CAP, 37% (n = 419) received initial oral antibiotics, and 63% received initial IV antibiotics (Supplemental Figure 1). Site-specific oral antibiotic use ranged from 14% to 58% (site#1: 315/540[58%], site#2: 21/146[14%], site#3: 60/304 [20%], site#4: 23/147[16%]). Characteristics and outcomes of children in this cohort with initial oral vs IV antibiotics are described elsewhere.¹⁵ Among children with initial oral antibiotics (n = 419), most did well with initial oral antibiotics (76% [n = 317]; Table 1).

TABLE 1.

Patient Characteristics of Children With Initial Oral Antibiotics

Demographics and History of Present Illness	Overall (n = 419)	Children Who Did Well With Initial Orals (n = 317)	Children Who Did Not Do Well With Initial Orals (n = 102)
Age, mean [IQR], y	3.2 [2.0–5.4]	3.3 [1.9–4.9]	3.2 [2.1–6.2]
Age (categorical), n (%)
≤2 y	110 (26)	85 (27)	25 (24)
>2 and <5 y	193 (46)	153 (48)	40 (39)
≥5 y	116 (28)	79 (25)	37 (36)
Female sex, n (%)	184 (44)	132 (42)	52 (51)
Insurance payer, n (%)
Government	191 (46)	138 (44)	53 (52)
Private	209 (50)	162 (51)	47 (46)
Other	19 (5)	17 (5)	2 (2)
Winter/fall season, n (%)	317 (76)	245 (77)	72 (71)
Hospital site, n (%)
1	315 (75)	254 (80)	61 (60)
2	21 (5)	14 (4)	7 (7)
3	60 (14)	34 (11)	26 (25)
4	23 (5)	15 (5)	8 (8)
Fever in ED or at homea	390 (93)	295 (93)	95 (93)
Total fever duration (including ED), n (%)
No fever	28 (7)	22 (7)	6 (6)
1 d	77 (18)	57 (18)	20 (20)
≥2 d	226 (54)	173 (55)	53 (52)
Fever but unknown duration	88 (21)	65 (21)	23 (23)
Antibiotic use in 7 d prior to arrival in ED, n (%)	82 (20)	64 (20)	18 (18)
Vital signs and initial diagnostics
Tachypnea (age adjusted) in the ED, n (%)	350 (84)	268 (85)	82 (80)
Hypoxemia in the EDb, n (%)	237 (57)	183 (58)	54 (53)
CXR resultsc, n (%)
Negative	76 (18)	63 (20)	13 (13)
Equivocal	228 (54)	164 (52)	64 (63)
Radiographic CAP	115 (27)	90 (28)	25 (25)
CXR—pleural effusion, n (%)
Small/trace	72 (17)	52 (16)	20 (20)
None	347 (83)	265 (84)	82 (80)
Performance of blood tests (CBC, CRP, PCT, or BCx), n (%)	66 (16)	36 (11)	30 (29)
Performance of respiratory pathogen testing, n (%)	117 (28)	79 (25)	38 (37)
Respiratory virus detectedd, n (%)	86 (21)	57 (18)	29 (28)
Respiratory atypical bacteria detectedd, n (%)	10 (9)	6 (8)	4 (11)
Initial interventions
Supplement oxygen started in ED, n (%)	326 (78)	248 (78)	78 (76)
Antibiotics given in ED, n (%)	284 (68)	205 (65)	79 (77)
Time to initial antibiotic, mean [IQR], h	3.52 [2.2–8.4]	3.67 [2.3–9.1]	3.00 [2.2–6.6]
Name of initial antibiotice, n (%)
Amoxicillin	363 (87)	276 (87)	87 (85)
Amoxicillin-clavulanate	23 (5)	20 (6)	3 (3)
Azithromycin	8 (2)	7 (2)	1 (1)
Cefdinir	6 (1)	2 (1)	4 (4)
Cefprozil	6 (1)	4 (1)	2 (2)
Cefuroxime	5 (1)	2 (1)	3 (3)
Cefpodoxime	3 (1)	3 (1)	0 (0)
Clindamycin	2 (0)	1 (0)	1 (1)
Levofloxacin	2 (0)	2 (1)	0 (0)
Metronidazole	1 (0)	0 (0)	1 (1)
Received broad-spectrum antibiotics initiallye, n (%)	56 (13)	41 (13)	15 (15)
Concurrent diagnoses
Treated for concurrent asthma exacerbationf, n (%)	86 (21)	68 (21)	18 (18)
Bronchiolitis codiagnosis, n (%)	77 (18)	57 (18)	20 (20)
Hospitalization and posthospitalization
Received supplemental oxygen, n (%)	398 (95)	302 (95)	96 (94)
Duration of oxygen, mean [IQR], h	55.0 [41.0,80.2]	51.3 [40.6,71.9]	76.1 [43.6124.8]
LOS, mean [IQR], h	72.4 [57.5,98.4]	68.5 [55.2,88.2]	97.6 (73.2152.5)
Switch to IV antibiotics, n (%)	96 (23)	0	96 (94)
In-hospital escalation of care (composite), n (%)	13 (3)	0	13 (13)
ICU transfer	6 (1)	0	6 (6)
Chest tube placement	4 (1)	0	4 (4)
Escalation of respiratory support	8 (2)	0	8 (8)
7-d CAP-related readmission or ED revisit, n (%)	7 (2)	0	7 (7)

Abbreviations: BCx, blood culture; CAP, community-acquired pneumonia; CBC, complete blood count; CRP, C-reactive protein; CXR, chest radiograph; ED, emergency department; ICU, intensive care unit; IV, intravenous; LOS, length of stay; PCT, procalcitonin.

^aFever defined as tactile or >100.3 °F at home or >38.0 °C in the ED.

^bHypoxemia defined as <90% at sea level or <88% at high altitude.

^cCXRs were classified as radiographic pneumonia (eg, focal consolidation, infiltrate), equivocal (eg, pneumonia vs atelectasis) vs negative (eg, normal, atelectasis only, peribronchial thickening without focal consolidation), based on radiology read as part of routine clinical care.

^dAmong those who had respiratory pathogen testing performed.

^eDefined as a nonamoxicillin antibiotic given within 4 h of antibiotic initiation.

^fDefined as receipt of albuterol and systemic corticosteroid on first hospital day.

Among children who did not do well with initial oral antibiotics (n = 102), 80% (n = 82) switched to IV antibiotics without complications, 13% (n = 13) switched to IV and had escalation of care, 1% (n = 1) switched to IV and were readmitted, and 6% (n = 6) remained on orals but were readmitted (Figure 1). For those with escalation of care, 8 had ICU transfer with increased respiratory support, 4 had chest tube only, and 1 had increased respiratory support only. Among children who switched to IV antibiotics (n = 96), switching occurred 12.3 hours (IQR, 8.2–17.9) after oral antibiotics, and 40% (n = 38) switched after just 1 oral dose.

FIGURE 1.

Children who did vs did not do well with initial oral antibiotics and why children were classified as not doing well with initial oral antibiotics based on 3 criteria (switching to IV antibiotics, escalation of care, revisit/readmission after discharge), listed in the order in which they occur clinically. This figure demonstrates that most patients did well with initial oral antibiotics, and most of those who did not do well were classified as such because they were switched to IV antibiotics without any pneumonia-related complications (ie, escalation of care or readmission). No patients had emergency department revisits.

Abbreviation: IV, intravenous.

Only hospital site was associated with doing well with initial oral antibiotics; no patient factors were significant (Table 2). When excluding children with viral CAP discharge diagnoses or viruses detected on respiratory pathogen panel (resulting n = 266 [63% of cohort]), excluding children aged at most 2 years (resulting n = 303 [74% of cohort]) and removing hospital from the model, findings remained consistent (Supplemental Tables 4–6). When assessing factors associated with switching to IV antibiotics, results persisted aside from a small statistically significant finding with age in months and odds of switching to IV antibiotics (aOR, 1.01 [1.001–1.013]; Supplemental Table 7).

TABLE 2.

Factors Associated With Doing Well With Initial Oral Antibiotics

Factor	Unadjusted OR (95% CI)	Adjusted OR (95% CI)
Age, mo	0.99 (0.99–1.00)	0.99 (0.99–1.00)
Site 2 (vs 1)	0.48 (0.19–1.24)	0.38 (0.14–1.06)
Site 3 (vs 1)	0.31 (0.18–0.56)	0.23 (0.12–0.44)a
Site 4 (vs 1)	0.45 (0.18–1.11)	0.29 (0.11–0.76)a
Fever for 1 d (vs no fever)	0.77 (0.28–2.19)	0.66 (0.22–2.00)
Fever for ≥2 d (vs no fever)	0.89 (0.34–2.31)	0.75 (0.27–2.08)
Fever for unknown duration (vs no fever)	0.77 (0.28–2.14)	0.49 (0.16–1.49)
Antibiotics within 7 d prior to arrival	1.18 (0.66–2.11)	1.33 (0.70–2.51)
Radiographic CXR (vs negative)	0.74 (0.35–1.56)	0.76 (0.34–1.71)
Equivocal CXR (vs negative)	0.53 (0.27–1.03)	0.57 (0.23–1.18)
Time to initial antibiotics, h	1.01 (0.99–1.03)	1.01 (0.99–1.03)
Initial broad-spectrum antibiotics	0.79 (0.42–1.51)	0.73 (0.36–1.51)
Supplemental oxygen in the ED	1.11 (0.65–1.88)	0.90 (0.51–1.61)
Bronchiolitis and/or asthma codiagnosis	1.15 (0.72–1.83)	1.22 (0.72–2.06)

Abbreviations: CXR, chest radiograph; ED, emergency department; OR, odds ratio.

^aStatistically significant finding.

DISCUSSION

Hospital site but no patient factors were associated with doing well with initial oral antibiotics. This suggests that clinician/hospital practice tendencies may drive antibiotic route decision-making more than individual patient characteristics. While there has been limited work on factors influencing antibiotic route selection, this finding is consistent with interhospital variation in antibiotic use and diagnostic testing in pediatric CAP.²⁶

Including a hospital with high oral antibiotic use (58%) provided one of the largest CAP cohorts treated with oral antibiotics. Consistent with prior studies, more than 75% of children who started oral antibiotics did well without IV therapy or complications.^5,15 Among those who did not do well, most switched to IV within 24 hours and without complications, likely reflecting clinician preference or concerns about illness progression or oral intolerance rather than true treatment failure as improvement often takes 24 to 48 hours.^18,27

Limitations include potential underestimation of success with initial oral antibiotics due to exclusion of LOS of less than 48 hours and our conservative definition of failure (including switching to IV, which may have prevented worse outcomes). By not evaluating outcomes in the first 48 hours or revisits outside PHIS hospitals, we may have missed complications. Similar to clinical practice, we could not confirm that all children had bacterial CAP—37% had viral CAP diagnosis or virus detected, which is likely an underestimate because not all were tested. However, all received antibiotics, reflecting real-world practice of treating bacterial CAP with concurrent viral illnesses. Additionally, findings were consistent after excluding children with viral CAP diagnosis/detection and those aged younger than 2 years. This underscores challenges and limitations in differentiating viral vs bacterial CAP in both practice and research. The predominance of children from one hospital may limit generalizability of site-level comparisons. We may have missed additional patient-level factors that influence the odds of doing well with oral antibiotics. We also did not evaluate oral intolerance or vomiting, which could prompt switching to IV antibiotics. Finally, findings may not generalize to other settings or patients.

In conclusion, hospital site—not patient characteristics—was the only factor associated with doing well with initial oral therapy. Additionally, most children who received initial oral antibiotics, representing more than one-third of the overall cohort, recovered without IV antibiotics or complications. This adds to existing evidence supporting the likely effectiveness of initial oral antibiotics for many children hospitalized with CAP. Future studies in larger, more generalizable cohorts should evaluate demographic (eg, age), clinical, and biomarker predictors of oral antibiotic success to identify subpopulations who can do well with orals and those who may benefit from IV therapy. Research should also examine clinical decision-making and contextual factors influencing antibiotic route selection at each institution, as cultural changes may be important to increase initial oral antibiotic therapy.

ABBREVIATIONS

aOR

adjusted odds ratio

CAP

community-acquired pneumonia

CXR

chest radiograph

ED

emergency department

ICU

intensive care unit

IV

intravenous

LOS

length of stay

PHIS

Pediatric Health Information System