Abstract
Purpose
The incidence of pulmonary embolism (PE) in hospitalized children has increased in recent years. This study sought to characterize factors and outcomes associated with PE using a national pediatric cohort.
Methods
The Nationwide Readmissions Database was queried (2016-2018) for patients (<18 years) with a diagnosis of PE. Index and prior hospitalizations (PHs) within 1 year were analyzed. A binary logistic regression utilizing 37 covariates (demographics, procedures, comorbidities, etc.) was constructed to examine a primary outcome of in-hospital mortality.
Results
3440 patients were identified (57% female) with the majority >12 years old (77%). One-third had a known deep vein thrombosis (69% lower and 31% upper extremity). Nineteen percent underwent central venous catheter (CVC) placement. Twenty-one percent had a PH within 1 year. Nine percent underwent an operation with the majority being cardiothoracic (5%). Overall mortality was 5%. Neurocranial surgery, cardiothoracic surgery, and CVC placement were associated with the highest odds of inpatient mortality after logistic regression.
Conclusion
Pediatric patients with PE have a high rate of PHs, CVC placement, and inpatient operations, which may be associated with higher mortality. This information can be utilized to improve screening measures and clinical suspicion for PE in hospitalized children.
Keywords
Introduction
Pulmonary embolism (PE) in children is an uncommon but emergent condition associated with high morbidity and mortality. The incidence of hospitalized children with PE has increased over the past 20 years and is expected to continue to rise, with recent reports indicating an increased prevalence of PE and venous thromboembolic (VTE) complications in the pediatric population up to 15.5%.1,2 This rising trend is likely due in part to the increased survival of children with predisposing conditions, frequent use of central venous catheters (CVCs), as well as expanded screening and recognition. 3 Historically, suspected risk factors, preclinical prediction rules, and management guidelines have been derived from adult studies with limited pediatric-specific adaptations.3,4
Diagnosis of pediatric PE can often be delayed or missed, which results in higher mortality. 5 Two commonly used adult PE algorithms, the Wells criteria and Pulmonary Embolism Rule-out Criteria (PERC), have been cautioned for use in children due to their low specificity and increased risk of radiation exposure in this population. 6 Risk factors for pediatric PE extrapolated from adult studies have typically included contraceptive medications, thrombophilia, obesity, and autoimmune conditions. 5 However, multiple recent studies have concluded that certain risk factors are associated with worse outcomes in children, including CVC placement, immobility, and critical or complex chronic illness.5,7-11 Carpenter et al 1 noted that African American and Hispanic patients were significantly more likely to experience recurrent PE than white patients, and it is currently unknown what factors influence recurrence rates in non-white populations.
Although some previous studies have attempted to describe risk factors for PE in children, the majority of these analyses are limited by small sample sizes or their single institution study design.5,7-11 The purpose of this study was to characterize operative and hospitalization-associated risk factors and outcomes including mortality associated with PE using a nationally representative pediatric cohort. We hypothesize that factors such as CVC placement and operation type are associated with higher in-hospital mortality among children with PE.
Methods
The Nationwide Readmissions Database (NRD) was employed for analysis as it comprises information on more than 15 million admissions over 28 states. The NRD is the most robust database containing annual hospital readmission information in the United States and includes over 17 million unweighted discharges (36 million weighted). This data source contains demographic, diagnosis, and procedure codes documented per the International Classification of Disease, Tenth Revision Clinical Modification (ICD10-CM) coding system in addition to length of stay and hospital characteristics for each patient’s admission. A single admission record is created in the NRD for patients with same-day readmission, discharges, or transfers across different institutions. Given that the NRD records annual data from individual State Inpatient Databases, the NRD does not track readmissions over separate states or multiple calendar years. Furthermore, index admissions occurring in the month of December are not recorded so that an accurate 30-day readmission rate can be calculated.
The NRD converted from the ICD9 to the ICD10 revision coding schema in 2015. As a result, the most recent contiguous releases (2016-2018) were queried to identify patients (<18 years old) with a diagnosis of pulmonary embolism (I26.0, I26.02, I26.9, I26.92, I26.93, I26.94, I26.99). Those with septic PE (I26.01) were excluded. Index hospitalization refers to the hospitalization in which the PE diagnosis was identified, and all prior hospitalizations within the prior year of the index hospitalization were examined. ICD10-CM diagnosis codes were utilized to identify patients with a previous upper or lower extremity DVT (Supplemental Table 1). Procedure codes from the ICD10 system were used to identify patients who underwent central venous catheter placement, and the full list of all surgical procedures classified by type (cardiothoracic, gynecologic, etc.) is available in Supplemental Table 2.
Categorical data were analyzed utilizing Fisher’s exact test or chi-squared analysis, as appropriate. The normality of continuous data distributions was evaluated using the Kolmogorov-Smirnov test. For those data with violation of the normality assumption, continuous variables are presented as median [interquartile range (IQR)]. These data were analyzed using the Mann-Whitney U test. There were no parametric data in the present analysis. A binomial logistic regression model utilizing 37 covariates (demographics, procedures, comorbidities, and hospital characteristics) was constructed in a stepwise fashion to identify factors associated with a primary outcome of in-hospital mortality and reported as odds ratios (OR) with corresponding 95% confidence intervals (95% CI). Categorical variables found to be significant (P < .05) on univariate analysis were identified for inclusion in the regression model. IBM SPSS Statistics version 28 (International Business Machines Corp, Armonk, New York) was utilized for statistical analysis. Statistical significance was defined as P < .05.
Results
Cohort Demographics
Cohort Demographics and Outcomes.
Cells marked with an asterisk (*) represent values less than or equal to 10 censored from publication in accordance with the Healthcare Cost and Utilization Program (HCUP) Data Use Agreement. HENT = head, ears, nose, and throat; DVT = deep vein thrombosis.
aHypercoagulable disorders include deficiencies of antithrombin III, protein C, protein S as well as mutations in factor V Leiden and prothrombin.
bData presented as median [interquartile range].
Characteristics Among Patients With Prior Hospitalizations
Characteristics and Outcomes of Patients Admitted Within One Year Prior of Index Hospitalization.
Cells marked with an asterisk (*) represent values less than or equal to 10 censored from publication in accordance with the Healthcare Cost and Utilization Program (HCUP) Data Use Agreement. DH = different hospital; HENT = head, ears, neck, and throat; OMFS = oral maxillofacial surgery.
aPercent given as breakdown of overall subcategory.
bData presented as median [interquartile range].
Analysis of Factors Associated With Inpatient Mortality
Univariate Analysis of Factors Associated With Inpatient Mortality After PE.
Cells marked with an asterisk (*) represent values less than or equal to 10 censored from publication in accordance with the Healthcare Cost and Utilization Program (HCUP) Data Use Agreement. DH = different hospital; DVT = deep vein thrombosis; GU = genitourinary; GYN = gynecologic; HENT = head, ears, neck, and throat; OMFS = oral maxillofacial surgery.
Bolded values represent statistical significance defined as P <.05.
Factors Associated With Inpatient Mortality After PE by Binary Regression.
DH = different hospital.
Bolded values represent statistical significance defined as P <.05.
Discussion
The present study is one of the largest series to examine pulmonary embolism utilizing a nationally representative pediatric cohort. Pediatric patients with PE have a high rate of prior hospitalizations within 1 year, with a significant portion admitted to a different hospital than their admission for a PE. Patients with PE experience a substantial frequency of CVC placement and inpatient operations, both in their prior and index hospitalizations. Concomitant medical comorbidities such as congestive heart failure, extremity fracture, and coagulopathies as well as CVC placement, neurocranial, and cardiothoracic operations, may be associated with higher odds of mortality.
The incidence of VTE in the pediatric population has been increasing over the last few decades. 12 This is partially driven by the more liberal use of screening with ultrasound imaging technology as well as increased procedural interventions associated with provoked thrombotic events, such as CVC placement.13-15 Despite the increased recognition and incidence of VTE, there continue to be few clinical prediction rules and understanding of pediatric-specific risk factors associated with acute complications of VTE, including PE.16,17
Hennelly et al 9 utilized a large multihospital cohort of more than 1100 patients to retrospectively evaluate factors associated with PE diagnosis in children presenting to the emergency department. After multivariate analysis, they demonstrated 2 significant factors associated with PE, including hospitalization for trauma or surgery within the past 4 weeks before presentation (OR 2.1, 95% CI 1.1-3.7) and prior VTE (OR 11.8, 95% CI 6.3-21.9). Within our cohort, there was also a high rate of patients with a prior admission in the same calendar year (21%) and a significant proportion of these patients who were admitted to a different hospital than their index hospitalization which has not been delineated in prior studies (17%). There was also a significant proportion of patients with multiple prior admissions in the same calendar year (36%). Given that the NRD only tracks patients within a single calendar year and does not include patients admitted in December such that a true 30-day readmission rate can be calculated, the number and effect of prior hospitalizations for patients is likely underestimated in our sample particularly for patients admitted early in the calendar year. This was an additional reason that multivariate analysis was utilized to account for multiple contributory factors as well as more than 1 individual NRD sample year. While Hennelly et al 9 examined only factors associated with PE diagnosis; our results further interrogated inpatient mortality after PE diagnosis and found prior hospitalization status, particularly at a different hospital, to be associated with a greater odds of mortality. This could potentially result from the lack of follow-up for patients between different centers concerning known DVT and/or factors associated with a greater propensity to develop recurrent VTE. This may represent a target for future quality improvement initiatives and study measures to ensure pediatric patients readmitted across different hospitals are evaluated for risk factors associated with PE. Previous studies have estimated the incidence of DVT in pediatric patients with PE to be around 60%-72%.17,18 Other studies that have attempted to characterize the recurrence rate of thromboembolic events in children have reported ranges from 7% to 19%, including PE among all VTE.17,19-21 There was a high rate of DVT documented in those patients with prior admissions in our study. Taken together, these results underscore the need to interrogate prior hospitalization history, as well as the procedures performed (surgery, CVC placement, etc.), among pediatric patients with suspected PE.
Another factor that has been given attention in prior work related to PE in both the pediatric and adult populations is its association with CVC placement. Previous studies have observed thrombotic event occurrence rates among children with CVCs as high as 33%-64% and even higher among younger age groups such as neonates (89%-94%).17,19,20 Within our study, there was a significant rate of CVC placement in patients during index (19%) and prior hospitalizations (21%). Given that the NRD is unable to provide granular factors regarding procedure details and complications, we were unable to analyze if complications associated with CVC placement in these patients contributed to the development of PE. Furthermore, CVC placement was significantly associated with greater odds of inpatient mortality in this cohort. However, it is important to recognize that mortality among those with PE is often a multi-hit phenomenon. Namely, patients with more medical comorbidities and procedural factors associated with PE development are also more likely to require CVC placement. 22 Few studies have investigated factors associated with mortality after PE. Other series have reported mortality of PE in children around 10%; however, the number of children who underwent testing for PE was relatively low.16,17,19 This was somewhat higher than our overall inpatient mortality rate of 5%. Given the high risk of mortality in these series, mechanical prophylaxis such as sequential compression devices in high risk pediatric patients undergoing orthopedic or major abdominal surgery who are non-mobile should be considered.23,24 There is a paucity of literature regarding the widespread utility of routine pharmacologic VTE prophylaxis in the pediatric population with the majority of these studies examining pediatric trauma patients and demonstrating a benefit of prophylactic anticoagulation in reducing VTE complications.25-27 Data from our study can help better inform future prospective investigations in identifying high risk pediatric surgical patients to obtain higher level evidence on the benefits of mechanical and chemical VTE prophylaxis.
Few studies have attempted to characterize surgical factors associated with VTE. Sherrod et al 22 utilized the National Surgical Quality Improvement Program-Pediatric (NSQIP-P) database to analyze over 153 000 inpatient pediatric surgical patients. They found the highest incidence of VTE occurring among patients receiving cardiothoracic surgery independent of elective vs urgent procedure status. Similarly, cardiothoracic operations were the most frequently represented procedures in our study population and were significantly associated with greater odds of mortality. Within our study, common general surgical procedures (abdominal, soft tissue, gynecologic, vascular, etc.) were less frequently represented and were not associated with a higher odds of mortality. Concomitant medical comorbidities in this population, such as valvular disease, alterations in flow, and immobility, as well as the increased need for CVCs in these patients, may also explain their development of PE and have been previously studied as independent factors for all VTE.14,28-30 Neurocranial operations were also associated with greater odds of mortality, which may be partially explained by the acuity of this patient population and similar contributing factors (postoperative immobility and need for CVCs). 22 These results indicate the need for future risk-prediction models to consider the effect of specific surgical procedures to maintain a high index of clinical suspicion for diagnosing PE in children.
There were several significant limitations in the present study primarily related to the use of administrative datasets such as the NRD. Since the NRD relies on ICD10 diagnosis and procedure codes to capture data, more granular patient factors that are not encoded cannot be measured. Medication administration, particularly with regard to the timing and initiation of anticoagulation, is a significant factor influencing outcomes in children with PE.16,17,21 Although the NRD does describe a code for “long-term anticoagulation” that we could utilize, this code lacks information on the specific type of anticoagulation therapy, its duration, or its timing. Previous studies have found that surgical factors such as longer operative times are associated with a greater probability of VTE development. 22 Duration of both ventilator dependence and CVC placement have also been suggested in the pediatric population to be positively correlated with VTE risk.16,31 Admission to the intensive care unit has also been implicated with a greater likelihood of VTE in children.28,32 Features associated with hemodynamic and oxygenation status also play a major role in dictating the course of thrombolytic and anticoagulation management strategies for patients with PE. 21 These factors as well as factors regarding symptomatic vs asymptomatic status of patients were unavailable for analysis in this database. This was an additional reason that multivariate analysis was performed to account as much as possible for the effect of multiple contributory factors associated with patient acuity (need for CVC placement, medical comorbidities, and operation type) and mortality from PE. Further limitations associated with retrospective datasets such as the NRD include errors related to administrative data sampling, entry, and incorrect utilization of ICD diagnosis codes. An additional limitation of the 2018 NRD dataset release is that discharges from patients age less than 1 year old are inconsistently reported due to reporting of linkage numbers during 2018 to the NRD by 14 of the 28 centers in the NRD, which could skew the representation of this age demographic in our sample. Additionally, the NRD cannot track patients across contiguous calendar years or when individuals are readmitted across separate states, which can underestimate readmission outcomes and incidence. Fortunately, previous studies suggest these factors affect less than 5% of patients. 33
Conclusions
The current investigation is one of the largest series to examine pulmonary embolism utilizing a nationally representative pediatric cohort. Overall, pediatric patients with PE have a high rate of previous hospital admissions within 1 year, with a significant portion who are admitted to a different hospital than their index hospitalization. Those with PE have a high frequency of CVC placement and inpatient operations both in their prior and index hospitalizations. Concomitant medical comorbidities such as congestive heart failure, extremity fracture, and coagulopathies, as well as CVC placement, neurocranial, and cardiothoracic operations, may be associated with higher odds of mortality. Future studies should be undertaken to consider these and other factors in developing pediatric-specific risk stratification tools for the diagnosis and management of PE in children.
Supplemental Material
Supplemental Material - Uncovering Risk Factors and Outcomes of Pulmonary Embolism in a Nationwide Cohort of Hospitalized Children
Supplemental Material for Uncovering Risk Factors and Outcomes of Pulmonary Embolism in a Nationwide Cohort of Hospitalized Children by Carlos T. Huerta, Walter A. Ramsey, Cindy Rodriguez, Joshua P. Parreco, Chad M. Thorson, Juan E. Sola, and Eduardo A. Perez in The American Surgeon™
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The author(s) received no financial support for the research, authorship, and/or publication of this article.
Supplemental Material
Supplemental material for this article is available online.
References
Supplementary Material
Please find the following supplemental material available below.
For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.
For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.
