Author + information
- Received March 9, 2015
- Revision received June 12, 2015
- Accepted July 6, 2015
- Published online September 15, 2015.
- Michael L. O’Byrne, MD, MSCE∗,†∗ (, )
- Andrew C. Glatz, MD, MSCE∗,†,
- Brian D. Hanna, MD, PhD∗,
- Russell T. Shinohara, PhD†,
- Matthew J. Gillespie, MD∗,
- Yoav Dori, MD, PhD∗,
- Jonathan J. Rome, MD∗ and
- Steven M. Kawut, MD†,‡,§
- ∗Division of Cardiology, The Children’s Hospital of Philadelphia, and Department of Pediatrics, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania
- †Center for Clinical Epidemiology and Biostatistics, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania
- ‡Penn Cardiovascular Institute, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania
- §Department of Medicine, Perelman School of Medicine at the University of Pennsylvania, Philadelphia, Pennsylvania
- ↵∗Reprint requests and correspondence:
Dr. Michael L. O’Byrne, Division of Cardiology, The Children’s Hospital of Philadelphia, 34th Street and Civic Center Boulevard, Philadelphia, Pennsylvania 19104.
Background Cardiac catheterization is the standard of care procedure for diagnosis, choice of therapy, and longitudinal follow-up of children and adults with pulmonary hypertension (PH). However, the procedure is invasive and has risks associated with both the procedure and recovery period.
Objectives The purpose of this study was to identify risk factors for catastrophic adverse outcomes in children with PH undergoing cardiac catheterization.
Methods We studied children and young adults up to 21 years of age with PH undergoing 1 or more cardiac catheterization at centers participating in the Pediatric Health Information Systems database between 2007 and 2012. Using mixed-effects multivariable regression, we assessed the association between pre-specified subject- and procedure-level covariates and the risk of the composite outcome of death or initiation of mechanical circulatory support within 1 day of cardiac catheterization after adjustment for patient- and procedure-level factors.
Results A total of 6,339 procedures performed on 4,401 patients with a diagnosis of PH from 38 of 43 centers contributing data to the Pediatric Health Information Systems database were included. The observed risk of composite outcome was 3.5%. In multivariate modeling, the adjusted risk of the composite outcome was 3.3%. Younger age at catheterization, cardiac operation in the same admission as the catheterization, pre-procedural systemic vasodilator infusion, and hemodialysis were independently associated with an increased risk of adverse outcomes. Pre-procedural use of pulmonary vasodilators was associated with reduced risk of composite outcome.
Conclusions The risk of cardiac catheterization in children and young adults with PH is high relative to previously reported risk in other pediatric populations. The risk is influenced by patient-level factors. Further research is necessary to determine whether knowledge of these factors can be translated into practices that improve outcomes for children with PH.
- extracorporeal membrane oxygenation
- outcomes research
- pediatric cardiology
- Pediatric Health Information Systems Database
Pulmonary hypertension (PH) affects 2.1 to 3.7 children per million (1–3) but remains an extremely morbid condition, with a 5-year survival of 65% to 75% (1,2,4–6). Right-sided heart catheterization is an important tool in the diagnosis, classification, and longitudinal care of these patients (7,8); however, cardiac catheterization in children with PH carries a risk of cardiac arrest of 4.5 to 5.7 per hundred (9,10), which is >10 times the risk of catheterization in children with other diagnoses and in adults with PH (11–20).
The determinants of periprocedural morbidity and mortality are not well defined, particularly in children. Relatively small procedural volumes at single centers and differences in practice at different centers are obstacles to the study of outcome in children with PH. Identification of risk factors for adverse outcomes could provide an opportunity to intervene and improve the safety of catheterization in children with PH.
We performed a multicenter retrospective cohort study assessing risk factors for catastrophic adverse events using the Pediatric Health Information Systems (PHIS) database, an administrative database. We hypothesized that patient-level risk factors, such as age, pathogenesis of PH, and indicators of severity of illness, would influence the periprocedural risk of death and catastrophic outcome.
The PHIS is an administrative database that contains data from inpatient, emergency department, ambulatory surgery, and observation encounters from 43 nonprofit tertiary care pediatric hospitals affiliated with the Children’s Hospital Association (Overland Park, Kansas) in the United States. Encounters in PHIS include inpatient and observation admissions but exclude outpatient procedures (those without overnight observation). Data quality and reliability are ensured through a joint effort between the Children’s Hospital Association and participating hospitals. The data warehouse function for the PHIS database is managed by Truven Health Analytics (Ann Arbor, Michigan). For the purposes of external benchmarking, participating hospitals provide discharge/encounter data including demographics, diagnoses, and procedures. Forty-two of these hospitals also submit resource utilization data (e.g., pharmacy products, radiologic studies, and laboratory studies) to the PHIS. Data are deidentified at the time of data submission and are subjected to a number of reliability and validity checks before being included in the database. A data-use agreement was signed between study investigators and the Children’s Hospital Association. The institutional review board of The Children’s Hospital of Philadelphia reviewed the proposed project and determined that it did not represent human subjects research in accordance with the Common Rule [45 CFR 46.102(f)].
Procedures and diagnoses were identified by International Classification of Disease-9th Revision (ICD-9) codes. We included children and adults, age 0 to 21 years, who had a diagnosis of PH and were undergoing cardiac catheterization at any of the 43 PHIS centers between January 1, 2007, and December 31, 2012. We excluded subjects from centers reporting <25 cardiac catheterization procedures per year or not reporting cardiac catheterization procedures in at least 4 of 6 years during the study period to ensure that only centers with stable reporting practices were included. Subjects for whom the date of catheterization was missing were also excluded. Subjects undergoing electrophysiology studies and those undergoing cardiac catheterization on mechanical circulatory support were excluded because their risk of adverse event was considered to be qualitatively different.
Data were extracted from the PHIS database by direct query with use of ICD-9 codes and clinical transaction codes for pharmaceutical products as described previously (Online Table 1). The primary outcome was a composite of death or initiation of mechanical circulatory support (extracorporeal membrane oxygenation [ECMO], percutaneous ventricular assist device, or balloon pump) within 1 day of cardiac catheterization. Patient-level data included subject age, sex, race, insurance payer (private, public, other), presence of genetic syndrome (21), presence of noncardiac congenital anomalies, history of prematurity (defined as gestational age <34 weeks in patients <1 year of age), location of patient before the procedure (outpatient, neonatal intensive care unit, intensive care unit, cardiac intensive care unit, and step-down unit/cardiac unit), receipt of mechanical ventilation before catheterization, and receipt of inotropic agents, systemic vasodilators, and pulmonary vasodilators. Subclassification of PH was limited in this database to: 1) idiopathic pulmonary arterial hypertension (IPAH); 2) pulmonary arterial hypertension associated with congenital heart disease (APAH-CHD); 3) PH with cardiomyopathy, 4) PH in the context of a heart transplant, and 5) chronic thromboembolic pulmonary hypertension (CTEPH). Diagnoses were based on ICD-9 codes extracted by coders based on physician documentation and sent to the Children's Hospital Association. Further details regarding diagnoses and clinically relevant data such as oximetry, pressure data, laboratory data, and imaging data were not available. Procedural data included whether a transcatheter intervention was performed during the case.
For some covariates identified before analysis (such as race and cardiac diagnosis), data were missing in <10% of cases. For these subjects, data were coded as “missing” and included in the analysis. There were no other missing data.
Descriptive statistics were expressed as mean ± SD, median (range and interquartile range [IQR]), and percentages and counts as appropriate. Multiple catheterizations were performed on individual subjects over the study period. All eligible procedures were included, and all statistics are reported per procedure except where noted.
The association between patient-level characteristics and composite outcome was assessed with mixed-effects multivariate generalized linear models (with logistic link) (22). Fixed effects for the pre-specified covariates (listed above) were included. To account for covariance within centers and between multiple procedures performed in a single patient, random intercepts were added to the model for each center and patient (23). An additional analysis restricted to the first catheterization for each subject during the study period was performed as an alternative means of accounting for bias introduced by the inclusion of multiple catheterizations per subject. Adjusted risks of outcomes were estimated with conditional standardization (i.e., the risk estimated by the model if variables were set at either the mean values for the cohort for continuous variables or at the referent group for categorical variables) to provide a more clinically applicable estimation of risk. On the basis of previous research (24), an interaction term for age category and history of prematurity was included.
Several pre-specified sensitivity and subset analyses were performed. A model that included PH medications by drug class (calcium-channel blockers [CCBs], phosphodiesterase [PDE]-5 inhibitors, endothelin receptor antagonists [ERAs], and prostacyclin analogues) was performed. We also adjusted for the number of classes of PH medications to assess whether this provided pre-procedural risk stratification.
Post-hoc analyses included: 1) characteristics and outcomes of subjects in whom a catheter-based intervention was performed versus those without an intervention; 2) outcomes stratified by quintiles of center annual catheterization volume (quintiles determined as described previously ); and 3) outcomes excluding subjects who underwent a cardiac operation within 1 day of catheterization. The latter analysis was performed to assess the effect of pre-operative catheterizations at centers that used routine post-operative ECMO in high-risk patients on observed results. This may also have eliminated subjects who experienced clinical decompensation and subsequent ECMO.
All analyses were performed with Stata MP version 13 (StataCorp, College Station, Texas). The threshold for statistical significance was set at p < 0.05.
The initial query yielded data from 7,049 procedures performed in 4,818 subjects during 6,180 hospitalizations at 43 centers (Figure 1). Twenty-two procedures (0.3%) on 18 subjects from 5 centers were excluded because the centers reported fewer than 25 catheterizations per year or did not report any catheterization procedures in at least 4 of the 6 years of the study period. Twenty-five procedures (0.4%) in 14 subjects were excluded because of missing date of catheterization, and 663 procedures (9%) on 385 subjects were excluded because of mechanical circulatory support before and on the day of catheterization. Thus, the study sample included 6,339 procedures in 4,401 unique subjects from 5,651 hospitalizations across 38 centers.
The median age at catheterization was 2.2 years (range 2 days to 21 years) (Table 1). Fifty-one percent of catheterizations were performed in subjects who were male, 62% in whites, and 38% in those with private insurance. Twenty-one percent of procedures were performed in subjects with IPAH, 69% with APAH-CHD, 6% with PH and cardiomyopathy, 4% with PH after an orthotopic heart transplant, and 1% with CTEPH. Before catheterization, 5% were receiving CCBs, 10% were receiving PDE-5 inhibitors, 6% were receiving ERAs, 1.1% were receiving continuous intravenous prostacyclin analogues, and 0.1% were receiving inhaled nitric oxide. Overall, 13% were receiving inotropes and 9% systemic vasodilators. During the same hospitalization and before catheterization, 5% of subjects had undergone a cardiac operation. During the catheterization, 23% of subjects underwent a transcatheter intervention.
Catheterizations that included a transcatheter intervention were performed in subjects who were younger (p < 0.001), more likely male, and of “other” race. Such procedures also had a higher proportion of APAH-CHD (89% vs. 62.8%) and a lower proportion of IPAH (6.9% vs. 24.6%; p < 0.001), a lower proportion receiving pulmonary vasodilators (14.9% vs. 19.0%; p < 0.001), a higher proportion receiving systemic vasodilators (12.7% vs. 9.4%; p < 0.001), and a higher proportion who had undergone a cardiac operation during the same hospitalization (8.7% vs. 4.1%; p < 0.001) compared with catheterizations without an intervention (Online Table 2). The risks of composite outcome, death, and initiation of ECMO were not significantly different on the day of catheterization or within 1 day of catheterization.
The risk of the composite outcome within 1 day of catheterization was 3.5% (n = 222) (Table 1). A total of 206 subjects (3.3%) underwent initiation of mechanical circulatory support, and 17 died (0.3%). The risk of the composite outcome was 1.0% (n = 61) on the day of cardiac catheterization; 52 (0.8%) underwent initiation of mechanical circulatory support, and 9 died (0.1%). The risk of death before discharge was 6.6% (n = 416), with an increased risk of death if the patient was placed on mechanical support within 1 day of catheterization (15.1%; p < 0.001). In the subgroup of procedures in subjects with IPAH, 1.8% (25 of 1,416) underwent ECMO, and 36% of these (9 of 16) died before discharge.
Multivariate model for risk factors associated with death or mechanical circulatory support within 1 day of cardiac catheterization
Table 2 and Central Illustration show the results of a mixed-effects multivariate regression model of the risk factors for death or mechanical circulatory support within 1 day after catheterization. Using conditional standardization, the adjusted risk of composite primary outcome within 1 day of catheterization was 3.3% (95% confidence interval [CI]: 1.9% to 5.5%). Several factors were independently associated with an increased risk. Neonates with history of prematurity (odds ratio [OR]: 4.95; p = 0.02) and infants without prematurity (OR: 1.61; p = 0.007) were associated with an increased risk of the composite outcome relative to subjects between >1 and ≤8 years of age. Similarly, older children (age >8 and ≤18 years) had a reduced risk of the composite outcome (OR: 0.45; p = 0.02). Age >18 years was not associated with a significant change in the risk of adverse events. In terms of diagnosis, there were no significant differences in risk of composite outcome for subjects with APAH-CHD, cardiomyopathy, or CTEPH compared with IPAH. However, PH after heart transplantation was associated with a significantly increased risk (OR: 2.78; p = 0.005).
Several aspects of pre-procedural medical care were independently associated with the risk of catastrophic adverse outcome. Hemodialysis was associated with increased risk (OR: 19.40; p < 0.001), as was receipt of systemic vasodilators (OR: 1.85; p = 0.007). Receipt of any PH medications was associated with reduced risk of adverse outcome (OR: 0.38; p < 0.001). Mechanical ventilation and receipt of inotropes were not associated with the risk of composite outcome.
We also assessed the associated risk of catastrophic adverse outcome with treatment with PH medications by class (Online Table 3). The point estimates for CCBs, PDE-5 inhibitors, and ERAs were similar to that of PH medications overall (but with decreased precision). The association between PDE-5 inhibitors and reduced risk remained significant (OR: 0.44; p = 0.02), whereas the observed reductions in risk with CCBs (OR: 0.47; p = 0.08) and ERAs (OR: 0.63; p = 0.34) were not statistically significant. The number of PH medications received did not appear to affect the results (Online Table 4).
Several post-hoc analyses were performed. The exclusion of patients who underwent transcatheter interventions did not change the standardized risk of composite outcome or the previously observed associations with other covariates (data not shown). An analysis restricted to the first catheterization for each subject during the study period demonstrated no significant changes in the adjusted risk or the ORs for selected covariates, which suggests that the initial method for accounting for multiple catheterizations per subject was effective (Online Table 5). A sensitivity analysis was performed that restricted analysis to procedures in subjects who had not undergone an operation within 1 day of catheterization (n = 6,207 of 6,339 cases). The intention of this analysis was to eliminate subjects who had undergone a pre-operative catheterization who then underwent subsequent post-operative ECMO support. The standardized risk of composite outcome was 2.6% (95% CI: 1.5% to 4.7%), with similar predictors as seen in the main analysis (Online Table 6).
The association between catheterization laboratory volume and adverse outcomes was measured. The risk of composite outcome was highest in the lowest-volume quintile (6.1%) and lowest in the highest-volume quintile (2.4%), with a statistically significant association across quintiles (p < 0.001) (Online Table 7). After adjustment for other covariates, adjusted risk of death was significantly higher at low-volume centers (quintiles 1 and 2) than at higher-volume centers. There was no significant difference between outcomes at centers with higher volume (quintiles 3, 4, and 5), although a nonsignificant trend toward reduced risk with higher volume was observed (Online Tables 7 and 8).
Despite advances in noninvasive imaging technology, cardiac catheterization remains the gold standard for initial diagnosis, choice of initial pharmacotherapy, and longitudinal assessment of patients with PH (7,8). The risk of mortality with catheterization in children (across all diagnoses) has been measured in several single-center series, with estimates between 0.3% and 0.8% (12–15,19,24,25). Catheterization in adults with PH has been associated with a risk of mortality of 0.05% (26). The risk of cardiac arrest during cardiac catheterization in children with PH in single-center series is >10-fold higher than in other children undergoing cardiac catheterization (9,10,27). The single multicenter registry study that measured the risk of adverse events in cardiac catheterizations in children with PH reported no deaths in >177 cases, but the relatively small study population introduces uncertainty, with a 1-sided 97.5% CI for the mortality risk of 0% to 2.1% (28).
These studies emphasize the challenge of measuring the risk of mortality in PH. Because PH is a rare condition, even large-volume primary children’s hospitals perform relatively few cardiac catheterization procedures in children and young adults with PH. In addition, the wide spectrum of disease severity makes appropriate adjustment for pre-procedural risk necessary, which is challenging in small study samples. Several registries have collected data on patients with PH (4,6). A single study from the TOPP (Tracking Outcomes in Pediatric Pulmonary Hypertension) registry demonstrated a risk of death after catheterization of 0.6% (95% CI: 0.2% to 1.3%) (29) but was not sufficiently powered to define risk factors for catastrophic adverse outcome. Also, TOPP is a voluntary registry that requires informed consent for inclusion, which potentially excluded the most severely affected patients. In the current study, we used administrative data to perform a retrospective cohort study, which, with some limitations, allowed us to measure risk of cardiac catheterization without this bias from a broad range of pediatric hospitals and to adjust for patient characteristics. Our population may differ in ways from those in other series, especially those focused on IPAH, but may be more representative of the burden of disease that is seen at the tertiary care centers that submit data to PHIS. The adjusted risk of death or initiation of mechanical circulatory support within 1 day of cardiac catheterization was 2.6% to 3.3%, significantly higher than the previously reported risk of similar outcomes with cardiac catheterization in children (24) and that reported in adults with PH (11–20). The discrepancy between the high risk of catastrophic adverse outcome and the relative technical simplicity of these procedures (unlike right-sided heart catheterization and endomyocardial biopsy in children after orthotopic heart transplantation, which has a very low risk of adverse outcome [18,24,30]) highlighted that patient-associated factors are more influential than procedural characteristics in determining overall risk in this case.
This study identified patient-specific factors that may be useful in stratifying patients according to risk. As noted in previous studies (17,31–33), infants and especially premature infants with PH had a higher risk than other children with PH, with generally lower risks with increasing age. Exceptions to this pattern were neonates without prematurity, infants with history of prematurity, and subjects >18 years of age, in whom risk was not significantly different from school-aged children. These subpopulations were relatively small (<5% of the total study population), and the failure to demonstrate differences may have been caused by insufficient statistical power.
In addition, several markers of increased severity of illness before catheterization were independently associated with adverse outcome, including hemodialysis, receipt of systemic vasodilators, and previous cardiac operation during the same admission. These presumably are markers of hemodynamic instability and general frailty.
Transcatheter intervention was associated with a lower risk of adverse outcome. This is counterintuitive, because conventional wisdom suggests that risk should be tied to intervention. However, we found a similar decrease in risk in a broader cohort of children undergoing catheterization (24). Interventions may have been more likely in patients who were relatively well (with less chance of attempting an intervention in patients with more severe illness). Performance of an intervention may have indicated a reversible anatomic issue, with the performance of an intervention improving patient outcome.
The association of chronic medications with decreased risk of adverse events should be interpreted with care. Diagnostic and treatment algorithms recommend that patients undergo catheterization before initiation of these therapies, so patients receiving these treatments before catheterization have necessarily survived at least 1 catheterization and have likely demonstrated some responsiveness to pulmonary vasodilators, which leads to selection bias. Patients receiving PH medications and undergoing subsequent catheterization may also have a different risk profile. Further measures to account for confounding by indication are necessary to clarify the relationship between pulmonary vasodilators and risk of catastrophic adverse outcome.
A secondary analysis identified an association between center volume and the risk of early catastrophic adverse outcome. The protective effect of center catheterization volume for catheterization has been demonstrated previously across the range of procedures and diagnoses (24). Additional research is necessary to identify whether this protective effect is attributable to dispersible best practices or economy of scale.
Identification of subjects relied on ICD-9 codes from administrative data. As evidenced by a large number of subjects not receiving medications and a large number of APAH-CHD patients, the population described in this study may differ from those cited in other studies (especially those focused on IPAH). The detail available, especially in regard to the specific anatomy of patients with APAH-CHD, was limited. Post-hoc analyses were performed to determine the degree to which APAH-CHD patients influenced the risk of catastrophic outcome. The PHIS database is limited to catheterizations that are part of an observation or inpatient admission, though it notably includes procedures with subsequent hospital admission or death in the catheterization laboratory. This may have resulted in an overestimation of risk. Although most procedures for PH are performed with post-procedural inpatient observation, there are centers that do not admit patients for observation after catheterization, especially those with mild disease. Hemodynamic and oximetric data from the procedure, as well as additional procedural data (e.g., method of anesthesia) and additional clinical data (e.g., imaging studies), all of which might affect risk, were not available. The database has a temporal resolution of a single day of service. Choosing death or ECMO initiation on the day of catheterization and the following date of service as the primary outcome allowed us to capture all events that occurred from the catheterization and the ensuing 24 h (while including events that might have occurred as much as 48 h later). Finally, it was not possible to determine the degree to which observed adverse events were the direct result of the catheterization (as opposed to the result of progressive disease or other events). In a previous study, death before discharge was attributable to catheterization in 10% of cases (33). Thus, the risks calculated are necessarily inflated, but we hoped to mitigate this by limiting the time horizon as described previously (24).
The risk of catastrophic adverse outcome after cardiac catheterization in children with PH was significantly higher than that in children with other diagnoses. Younger age, use of pre-procedural systemic vasodilators, and cardiac operation within the same admission all increased the risk of an adverse event, whereas treatment with PH medications was associated with reduced risk. Further study is necessary to determine how these factors interrelate and whether a predictive model based on them would improve outcomes of children and young adults undergoing cardiac catheterization for PH. Although administrative databases have limitations, this study provides information that may inform the design and interpretation of future studies, whether they use multicenter registry data or datasets that combine clinical and administrative data.
COMPETENCY IN MEDICAL KNOWLEDGE: Cardiac catheterization in children and young adults with pulmonary hypertension is associated with higher risks of death or circulatory collapse than when performed for other diagnoses. Adverse outcomes are more frequent in younger patients and less common in those receiving pulmonary vasodilator medications.
COMPETENCY IN INTERPERSONAL AND COMMUNICATION SKILLS: When discussing cardiac catheterization procedures for children with pulmonary hypertension, physicians should explain the risk of catastrophic outcomes to patients (when appropriate) and their families in the context of balancing risks against potential benefits.
TRANSLATIONAL OUTLOOK: This multicenter cohort study may lead to a better understanding of the factors that are associated with risk of adverse events with catheterization in children with pulmonary hypertension.
The authors acknowledge Jin Long (Center for Pediatric Clinical Effectiveness Healthcare Analytics Unit at The Children’s Hospital of Philadelphia) for his role as a programmer who performed the query from the Pediatric Health Information Systems Database.
For supplemental tables, please see the online version of this article.
Dr. O’Byrne has received support from the National Institutes of Health (T32 HL007915) and an ENTELLIGENCE Young Investigator grant. Dr. Kawut has received support from the National Institutes of Health (K24 HL103844). Dr. Dori has received research support from Siemens. Dr. Gillespie has served as a consultant for Medtronic. Dr. Glatz has served as a consultant for Bristol-Myers Squibb. Dr. Rome has served as a consultant for Bristol-Myers Squibb. Dr. Hanna has received research support from Eli Lilly, United Therapeutics, Gilead Sciences, and Actelion. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose. Drs. Rome and Kawut are co-senior authors of this work.
- Abbreviations and Acronyms
- pulmonary arterial hypertension associated with congenital heart disease
- calcium-channel blocker
- chronic thromboembolic pulmonary hypertension
- endothelin receptor antagonist
- International Classification of Diseases-9th Revision
- idiopathic pulmonary arterial hypertension
- interquartile range
- pulmonary hypertension
- Pediatric Health Information Systems
- Received March 9, 2015.
- Revision received June 12, 2015.
- Accepted July 6, 2015.
- 2015 American College of Cardiology Foundation
- van Loon R.L.E.,
- Roofthooft M.T.R.,
- Hillege H.L.,
- et al.
- Moledina S.,
- Hislop A.A.,
- Foster H.,
- Schulze-Neick I.,
- Haworth S.G.
- Barst R.J.,
- McGoon M.D.,
- Elliott C.G.,
- Foreman A.J.,
- Miller D.P.,
- Ivy D.D.
- Simonneau G.,
- Galiè N.,
- Rubin L.J.,
- et al.
- Cassidy S.C.,
- Schmidt K.G.,
- Van Hare G.F.,
- Stanger P.,
- Teitel D.F.
- Kirkpatrick S.E.,
- Takahashi M.,
- Petry E.L.,
- Stanton R.E.,
- Lurie P.R.
- Vitiello R.,
- McCrindle B.W.,
- Nykanen D.,
- Freedom R.M.,
- Benson L.N.
- Bergersen L.,
- Gauvreau K.,
- Foerster S.R.,
- et al.
- Bergersen L.,
- Gauvreau K.,
- Marshall A.,
- et al.
- Pasquali S.K.,
- Hall M.,
- Li J.S.,
- et al.
- ↵(2011) Applied Longitudinal Analysis (Kindle Edition), Overview of linear models for longitudinal data (Wiley, Hoboken, NJ), 2nd ed. pp 1961–2888.
- ↵(2011) Applied Longitudinal Analysis (Kindle Edition), Generalized linear mixed effects models (Wiley, Hoboken, NJ), 2nd ed. pp 10334–11453.
- Hoeper M.M.,
- Lee S.H.,
- Voswinckel R.,
- et al.
- Barst R.J.,
- Beghetti M.,
- Berger R.,
- et al.
- Schneider D.J.,
- Moore J.W.
- Backes C.H.,
- Bergersen L.,
- Rome J.J.,
- et al.