Table 3.
Studies Included in Systematic Review
| Study, Year | Region, Setting | Study Design | Age | n | Limitations and Potential Biases |
|---|---|---|---|---|---|
| Florin et al [17], 2018 | Single center, US, ED | Retrospective | 3 months–18 years old | 518 | Single center, use of ICD coding may have introduced misclassification bias |
| Williams et al [18], 2016 | Multicenter, US, inpatient | Prospective | <18 years old | 2319 | Definition of mild, moderate, and severe pneumonia based on author opinion, limited to hospitalized patients |
| Araya et al [19], 2016 | Single center, Paraguay, inpatient | Retrospective | <15 years old | 860 | Single center, retrospective, exclusively inpatient, significant rates of comorbidities |
| Duke et al [22], 2001 | Single center, Papua New Guinea, inpatient | Prospective | 28 days–5 years old | 703 | High rates of malnourished children, lack of invasive interventions available, elevation of 1600 m, use of WHO severity criteria |
| Djelantik et al [23], 2003 | Single center, Indonesia, inpatient | Retrospective | <24 months | 4531 | Possible selection bias, high rates of malnourishment, lack of invasive interventions, use of WHO severity criteria |
| Nantanda et al [24], 2008 | Single Center, US, ICU and General Wards | Retrospective | 2–59 months | 157 | Retrospective, single center, developing nation with high rates of malnourishment limiting generalizability, used WHO-definitions but required either CXR confirmation or excluded children with wheezing with negative CXR |
| Reed et al [25], 2012 | Single center, South Africa, inpatient | Retrospective | <24 months | 4148 | Limited to young infants, high rates of malnourished children |
| Wolf et al [26], 2015 | Single center, US, inpatient | Post hoc analysis of prospective population-based study | <18 years old | 336 | Single-site analysis of a multicenter study introducing possible selection bias |
| Neuman et al [27], 2012 | Large-scale Multicenter, US, inpatient | Retrospective | <18 years old | 82 566 | Children may have been readmitted at nonincluded facility, admission decisions can show great variation between institutions |
| Mamtani et al [28], 2009 | Multicenter, 8 developing countries, inpatient | Post hoc analysis of previous RCT | 3–59 months | 889 | Developing countries, use of WHO severity criteria, treatment dose of amoxicillin 45 mg/kg per day, did not comment on percentage of children with wheezing (bronchiolitis, reactive airway disease) or HIV status |
| Tiewsoh et al [34], 2009 | Single center, India, inpatient | Prospective | 2–60 months | 200 | Use of WHO severity criteria, high rates of malnourishment and overcrowding, high proportion of children had wheezing, possible recall bias |
| Basnet et al [35], 2006 | Single center, Nepal, outpatient/ED | Retrospective | 2–60 months | 250 | Use of WHO severity criteria, developing country, 1300+m above sea level likely influencing degree of hypoxemia |
| Kuti et al [36], 2013 | Single center, Gambia, inpatient | Prospective | 2–59 months | 420 | Used WHO definitions for pneumonia diagnosis and severity, and did not require CXR confirmation (however, did exclude children with wheezing or cough for >2 weeks), single center and developing nation limiting generalizability |
| Demers et al [37], 2000 | Single center, Central African Republic, inpatient | Prospective | <5 years old | 395 | Developing nation with limited resources and high rates of malnourishment, possible observer bias as “alteration of general status” based on physician opinion and not validated scoring system, possible selection bias (significant proportion of patients absconded due to military uncertainty during the study, and not all patients had CXR performed) |
| Chisti et al [38], 2013 | Single center, Bangladesh, ICU | Retrospective | <5 years old | 140 | Developing nation, high rates of malnourishment, retrospective, limited to ICU |
| Hsu et al [41], 2015 | Multicenter, Taiwan, ICU | Retrospective | <18 years old | 12577 | Retrospective, dependent upon ICD-9 coding, only ICU setting limiting generalizability |
| Hirsch et al [42], 2016 | Multicenter, US, inpatient | Retrospective | Children | 12097 | Retrospective, used administrative database relying on ICD coding, all sites are tertiary care referral centers limiting generalizability |
| Champatiray et al [43], 2017 | Single center, India, inpatient | Prospective | 2 months– 5 years old | 141 | Single center, exclusively inpatient, use of WHO definitions, CXR were obtained on admission, but study does not mention whether cases were radiographically confirmed, study did not comment on rate of wheezing, history of present illness/social history subject to recall bias, much longer LOS (8–9 days) and much higher mortality rate (22%) than US studies |
| Muszynski et al [44], 2011 | Single Center, US, ICU | Retrospective | <18 years old | 23 | Retrospective, single center, small n, limited to ICU setting |
| Grafakou et al [48], 2004 | Single center, Greece, inpatient | Retrospective | 1–14 years old | 167 | Single center and only inpatient limiting generalizability, markers for severity were duration of fever and LOS |
| Kin et al [49], 2009 | Single center, Brazil, inpatient | Prospective | <5 years old | 113 | Single center, excluded bilateral pulmonary infiltrates that could present with more severe disease, severity criteria were WHO and BTS guidelines that are not validated, CXR interpreted by single radiologist |
| Patria et al [50], 2013 | Single center, Italy, ED | Retrospective | <14 years old | 335 | Single center, not all children with CAP during the study period had CXR performed introducing possible selection bias for more severe disease, CXR interpreted by single radiologist, higher than expected mean age (7.5 years old) |
| Mclain et al [51], 2014 | Multicenter, US, inpatient | Retrospective | 60 days–18 years old | 406 | Retrospective, CXR interpreted by single radiologist, potential selection bias as this was a sampling of a larger cohort |
| Ferrero et al [52], 2010 | Multicenter, Developing nations, inpatient | Prospective | 3–59 months | 2536 | Exclusively inpatient, use of WHO severity criteria, patients not vaccinated against pneumococcus significantly limiting generalizability, each CXR interpreted by 1 reviewer (possible interobserver bias) |
| Tapisiz et al [53], 2011 | Single center, Turkey, inpatient | Retrospective | <18 years old | 501 | Single center, retrospective, pre- pneumococcal and Haemophilus vaccination era in this region |
| Erlichman et al [56], 2017 | Multicenter, Jerusalem, inpatient | Retrospective | <18 years old | 144 | Retrospective, demographics limit generalizability |
| Langley et al [57], 2008 | Multicenter, Canada, inpatient | Retrospective | <18 years old | 251 | Retrospective, reliant upon chart review and ICD-coding, did not require specific WBC cutoff in pleural fluid to verify diagnosis of empyema |
| Goldbart et al [58], 2009 | Single center, Israel, inpatient | Retrospective | ≤8 years old | 112 | Retrospective, single center, and patients not vaccinated against pneumococcus significantly limiting generalizability, each CXR interpreted by 1 reviewer (possible interobserver bias) |
| Sawicki et al [61], 2008 | Single center, US, inpatient | Retrospective | Children | 80 | Single-center and retrospective design limit generalizability |
| Bender et al [62], 2008 | Single center, US, inpatient | Retrospective | <18 years old | 33 | Small n, single-center and retrospective design limit generalizability |
| Krenke et al [63], 2015 | Single center, Poland, inpatient | Retrospective | 1 months–18 years old | 32 | Small n, single-center and retrospective design limit generalizability |
| Donnelly et al [64], 1998 | Single center, US, inpatient | Retrospective | 6 months–16 years old | 17 | Very small n, single center, retrospective |
| Hacimustafaoglu et al [65], 2004 | Singe center, Turkey, inpatient | Prospective | 6 months–14 years old | 108 | Single center, each radiographic study interpreted by 1 reviewer (possible interobserver bias) |
| Hsiesh et al [66], 2011 | Single center, Taiwan, inpatient | Retrospective | <18 years old | 112 | Retrospective, single center, relatively few number of cases with BPF (18) |
| Hsiesh et al [67], 2015 | Multicenter, Taiwan, inpatient | Prospective | <18 years old | 94 | All cases limited to 1 region, images were not independently reviewed by 2 pediatric radiologists |
| Chen et al [70], 2017 | Single center, Taiwan, inpatient | Retrospective | 6 months–18 years old | 142 | Single center, retrospective, exclusively inpatient, not all patients during study had an ultrasound performed introducing possible selection bias, intrinsic limitation of ultrasound is that quality of images are operator-dependent (significant number of cases were excluded due to suboptimal images) |
| Lai et al [71], 2015 | Single center, Taiwan, inpatient | Retrospective | Children | 236 | Retrospective, single center, potential selection bias as children who had lung ultrasound were likely to have more severe pneumonia, ultrasound inherently is operator-dependent |
| Williams et al [72], 2015 | Multicenter, US, inpatient | Retrospective | <18 years old | 153 | Sampling from larger cohort that only included patients that had CRP and WBC performed thus introducing possible selection bias, retrospective |
| Don et al [74], 2009 | Single center, Italy, ED | Prospective | Children | 100 | Used hospital admission (institutional and provider differences can play a role) and alveolar infiltrate (vs interstitial) as markers for severity, single center |
| Prat et al [75], 2003 | Single center, Spain, ED | Prospective | 6 months–10 years old | 85 | Primarily studied PCT, ESR, and WBC’s ability to predict etiology of CAP; however, secondary analyses revealed no association between WBC and bacteremic patients, which may indicate a more severe disease course, limitations include a relatively small n at a single center limiting generalizability |
| Wu et al [76], 2015 | Single center, China, inpatient | Retrospective | Children | 865 | Use of WHO definition and severity criteria that are not specific, single center, no mention of exclusion criteria, no mention of how many patients had CXR and what the results of those potential imaging studies may have been, no mention of additional outcomes of cases (ie, mortality, ICU admission, invasive interventions) |
| Agnello et al [78], 2015 | Single center, Italy, inpatient | Retrospective | 1–14 years old | 119 | Single center, excluded patients who were hospitalized for more than 48 hours introducing selection bias against more severe cases, clinical markers for severity were hypoxemia (SpO2 <92%), dyspnea and tachycardia but not more severe markers or outcomes |
| Stockmann et al [79], 2017 | Multicenter, US, inpatient | Post hoc analysis of prospective study | <18 years old | 532 | Possible selection bias as only those with residual serum available for analysis were included (patients in the ICU were more likely to have residual serum), median time during admission PCT obtained was 1 day, thus limiting applicability to risk stratification on initial presentation |
| Yadav et al [80], 2015 | Single center, India, inpatient | Prospective | 2 months–5 years old | 50 | Single center, small n |
| Korppi et al [81], 2003 | Multicenter, Finland, Primary Care | Retrospective | ≤15 years old | 190 | Retrospective, serum samples run for PCT over 15 years after they were collected, pre-pneumococcal and Haemophilus vaccination era, performed in 1 region of Finland limiting generalizability |
| Singhi et al [82], 1992 | Single center, India, inpatient | Prospective | Children | 264 | Single-center study in a developing nation during the pre-routine vaccination era significantly limiting generalizability, LOS much longer than average LOSs in current studies in the developed world |
| Wrotek et al [83], 2013 | Single center, Poland, inpatient | Retrospective | <18 years old | 312 | Significant number of patients did not have sodium measured introducing possible selection bias, retrospective, single center, severity assessment based on clinical factors and inflammatory markers but did not evaluate for more severe outcomes, average hospitalization length (8–9 days), significantly longer than other current studies in the developed world |
| Don et al [84], 2008 | Single center, Italy, ED | Prospective | Children | 108 | Small percentage of patients did not have sodium samples introducing possible selection bias, single center, severity assessment based on clinical factors and inflammatory markers but did not evaluate for more severe outcomes |
| Wang et al [85], 2013 | Single center, Taiwan, inpatient | Retrospective | <18 years old | 84 | Single center, retrospective, pre- pneumococcal and Haemophilus vaccination era |
| Shah et al [86], 2011 | Multicenter, US, ED | Retrospective | ≤18 years old | 291 | Few number of bacteremic patients (6) limits statistical power of assessment of severity |
| Neuman et al [88], 2017 | Multicenter, US, ED | Retrospective | 3 months–18 years old | 2568 | Retrospective, wide variation between sites in rates of obtaining blood cultures, which introduces possible selection bias, primary objective of study was to evaluate rate of bacteremia in hospitalized children with CAP and determine susceptibility of pathogens to standard care, evaluation of +blood culture impact on severity came via rates of +cultures in complicated vs noncomplicated CAP and did not evaluate for other outcomes |
| Myers at al [90], 2013 | Multicenter, US, inpatient | Retrospective | 60 days–18 years old | 369 | Retrospective, inpatient study limiting generalizability, evaluated severity based on hypoxemia, LOS, ICU admission, and complicated pneumonia including respiratory failure but no mention of assessment on mortality |
| Banerjee et al [91], 2011 | Multicenter, North America, inpatient | Retrospective survey | Children | 37 | Retrospective survey of pediatric infectious disease physicians thus potential for reporting bias with potential preference to report more severe cases |
| Muñoz et al [93], 2011 | Single center, Spain, inpatient | Prospective | <18 years old | 206 | Single center, evaluated severity by LOS and ICU admission but not by other clinical factors or more severe outcomes |
| Shen et al [94], 2011 | Single center, Taiwan, inpatient | Retrospective | Children | 119 | Single center, retrospective, only pneumococcal and exclusively inpatient limiting generalizability, urinary antigen tests may have detected colonization and not acute infection in some cases |
| Pettigrew et al [95], 2016 | Single center, US, inpatient | Retrospective | 6 months to <18 years old | 363 | Single center, retrospective, exclusively inpatient, approximately 50% of children had asthma/reactive airway disease, ethnic distribution not representative of general population, 20% had received antibiotics before sputum collection, limited generalizability |
Abbreviations: BPF, bronchopleural fistulas; BTS, British Thoracic Society; CAP, community-acquired pneumonia; CRP, C-reactive protein; CXR, chest radiographs; ED, emergency department; ESR, erythrocyte sedimentation rate; HIV, human immunodeficiency virus; ICD, International Classification of Diseases; ICU, intensive care unit; LOS, length of stay; PCR, polymerase chain reaction; PCT, procalcitonin; RCT, randomized clinical trial; SpO2, blood oxygen saturation; US, United States; WBC, white blood cells; WHO, World Health Organization.