Criteria abstracted from The Users' Guide to Medical Literature, from the Health Information Research Unit and Clinical Epidemiology and Biostatistics, McMaster University

Highlighted lines and questions below provide links to the pertinent description of criteria in The EBM User's Guide, now available at the Canadian Centres for Health Evidence

Article Reviewed:

Please visit the new Evidence Based Journal Club Reviews

Risk of seizures in survivors of newborn heart surgery using deep hypothermic circulatory arrest.

Clancy RR, McGaurn SA, Wernovsky G, et al.

Pediatrics. 2003;111(3):592-601. [abstract]

Reviewed by Girish Deshpande MD, Al Torres MD MS, Department of Pediatrics, Children's Hospital of Illinois/University of Illinois, Peoria, IL

Review posted April 7, 2005

I. What is being studied?

Study objective:

To identify the pre- and intraoperative factors those were associated with postoperative acute neurologic events (ANEs) (seizures, coma or both) in an anatomically diverse population of survivors of newborn heart surgery using deep hypothermic cardiac arrest (DHCA).

To risk-stratify this population on the basis of preoperative risk variables for the purpose of designing future neuroprotection trials.

Study design

This is a retrospective, observational cohort study. This study is an analysis of an existing database of survivors of newborn heart surgery who were enrolled in a previously conducted neuroprotection trial (1). The neuroprotective trial was conducted as a single-center, stratified, randomized, blinded, placebo-controlled trial with study drug administered before, during and after newborn heart surgery using DHCA.

II. Are the results of the study valid?

Note: These questions follow from Randolph AG et al. Understanding articles describing clinical prediction tools. Crit Care Med 1998;26:1603-1612. [abstract]

1. Was a representative group of patients completely followed up? Was follow-up sufficiently long and complete?

1. A total of 217 non-HLHS infants were enrolled in the neuroprotection study from June 1992 to September 1997.
2. Thirty infants did not undergo surgery or DHCA and 6 who underwent DHCA < 20 minutes were excluded.
3. There were 17 deaths, with 41% (7 of 17) having ANEs. The authors do not explain why these infants were excluded from the analysis. This makes the sample less representative of post-operative non-HLHS patients at risk for ANEs.

Consequently, 164 (76%) of 217 surviving infants' data were available for statistical analyses.

All patients who received neuromuscular blocking agents were deemed to be unsuitable for assessment of coma. The authors do not mention the number of such infants.

Data collection: The database included clinical observations, laboratory values, and imaging examinations and was organized into 3 periods: (1) preoperative, (2) intraoperative, and (3) postoperative. The data were further categorized to reflect preoperative sources of data.

Yes, follow-up was sufficiently long and complete.

1. Clinical endpoints were recorded during the postoperative period in the intensive care unit, up to 6 weeks after surgery.
2. There was no outpatient follow up of these patients, nor was there any developmental outcome mentioned.

2. Were all potential predictors included?

Maybe.

Preoperative Data Included:

1. Demographics: Mother's age, race, parity and socioeconomic status (Hollingshead index), the presence of maternal health conditions; and family history of epilepsy.
2. Infant's characteristics included gestational age, birth weight, gender, Apgar scores, and microcephaly and early neonatal course.
3. The presence of genetic condition was based on a clear constellation of clinical findings to name a specific genetic syndrome, an abnormal chromosomal analysis (e.g. 22q11 deletion), or an isolated dysmorphic feature (e.g. cleft palate)
4. The infants' congenital heart diseases (CHD) were classified into standard anatomic categories on the basis of preoperative echocardiogram. Furthermore, each infant was classified in a simplified scheme by presence or absence of aortic arch obstruction. Aortic arch obstruction was considered to be present for any hypoplasia or atresia of the ascending, transverse, or descending aorta of sufficient degree to require surgical repair.

Intraoperative Data Included

1. The duration of surface cooling
2. Lowest recorded nasopharyngeal temperature
3. Duration of cardiopulmonary bypass (CPB) cooling
4. DHCA time
5. Duration of rewarming on CPB
6. The use of modified ultrafiltration (MUF)
7. The need for additional periods of CPB

The predictors chosen for inclusion in the model were based on the investigators clinical knowledge and results of statistical analyses. The investigators determined a priori that the model would be based on a maximum of 3 independent variables due to the sample size.

3. Did the investigators test the independent contribution of each predictor variable?

Yes.

• The logistic regression model was used for selecting variables and characterizing their strength of associations with ANEs using the odds ratio (OR) and associated 95% confidence intervals (CIs). Logistic regression models the proportion of ANEs and how these proportions are influenced by independent model variables. Mean values were substituted for candidate independent variables with 5% or fewer missing values.
• After final selection of the independent variables to model risk of ANEs, exact logistic regression methods were used to derive the final models. Model fit was determined using the Hosmer and Lemeshow method and a rejection of the model at p < 0.05. ORs and associated exact 95% CIs are provided for each independent model variable.
• Risk stratification trees are displayed to help describe the relationships among the independent variables obtained from the logistic regression analysis and how these variables interrelate to influence risk of ANEs.

4. Were outcome variables clearly and objectively defined?

1. Seizures were identified using established clinical criteria of tonic, clonic, or myoclonic activity of a limb, trunk, or cranial muscle that could not be aborted by restraining the area.
2. All cardiac intensive care unit nurses were specifically trained in the recognition of clinical neonatal seizures by instructional videotapes, with pre and posttest examinations.
3. Coma was identified from clinical examination by a child neurologist: comatose infant exhibited no spontaneous or responsive eye opening on repeated observations. Infants who were administered neuromuscular blocking drugs were not assessable for this endpoint.

Potential Drawback: Nurses may not have recognized subtle seizures and EEGs were not used to confirm the presence or absence of seizures.

III. What are the results?

1. What is(are) the prediction tool(s)?

1. ANEs occurred in 31 (18.9%) including "seizures alone" (n=28), "coma alone" (n=2), or "seizures and coma" (n=1).
2. A preoperative risk model was constructed demonstrating that infants with a genetic condition and aortic arch obstruction had a 47.8% risk of ANEs compared to all other remaining infants, who had a 9.9% risk.
3. It was also found that prolonged DHCA time (> 60-minutes) could be a significant risk for infants who have a preexisting genetic condition; however, infants who have genetic condition and do not undergo prolonged DHCA time or have an aortic arch obstruction are not at increased risk of ANEs

A preoperative and operative prediction model was developed.

Preoperative model:

Variable	OR (95% CI)	P value
Genetic syndrome	2.9 (1.2-7.1)	0.02
Arch obstruction	2.1 (0.9-5.1)	0.10

Operative model:

Variable	OR (95% CI)	P value
Genetic syndrome	3.1 (1.2-7.8)	0.01
Complex*	3.9 (1.5-11.6)	0.004

*Complex is a derived variable of aortic arch obstruction or DHCA time > 60 minutes. Neither of these variables was statistically significant when entered into the model individually.

2. How well does the model categorize patients into different levels of risk?

Three categories of risk are observed.

1. The risk of ANEs was low (9%; 95% CI: 3.7%-17.6%) when there was no aortic arch obstruction and the DHCA time was < 60 minutes, regardless of the genetic status.
2. In those without a genetic condition, the risk of ANEs was intermediate (17.2%; 95% CI: 8.6%-29.4%) when there was an aortic arch obstruction or DHCA > 60 minutes.
3. The risk of ANEs was highest (50.0%; 95% CI: 30.7%-69.4%) in those who had a genetic condition and aortic arch obstruction or DHCA time > 60-minutes.

The amount of risk in the low and intermediate risk groups is not statistically different.

3. How confident are you in the estimates of risk?

The presence of a genetic syndrome appears to be a good predictor of ANEs since the CIs for the ORs do not cross 1 in either model. The variables aortic arch obstruction and DHCA time > 60 minutes are only useful as the derived variable "complex" that includes one or the other.

IV. Will the results help me in caring for my patients?

1. Does the tool maintain its prediction power in a new sample of patients?

No, these risk factors were not validated in a new population.

2. Are your patients similar to those patients used in deriving and validating the tool(s)?

Yes our patients have similar demographic and clinical characteristics to the study population. However, the findings cannot be applied to all newborns with CHD that are postoperative. The study was done in infants undergoing surgery for non-HLHS. Although presence of a genetic condition was identified as a predictor of ANEs, patients with multiple congenital anomalies or potentially lethal genetic disorders were excluded. DHCA time > 60 minutes was a potential predictor of ANEs but patients with DHCA time < 20 minutes were not studied.

3. Does this tool improve your clinical decisions?

No. However, authors have identified three major variables and identifying these three variables does not limit our ability to evaluate other variables in future.

4. Are the results useful for reassuring or counseling patients?

The evidence is useful in providing our patients families with the information about possibility of seizures or coma or both as a important complication especially if their neonate has genetic condition, is likely to require prolonged DHCA or has aortic arch anomaly significant enough to require surgical repair. In this study, the presence of ANE is used as a surrogate endpoint for lasting neurologic deficits. It is the strength of this association that determines the usefulness of the findings.

References

1. Clancy RR, McGaurn SA, Goin JE, et al. Allopurinol neuro-cardiac protection trial in infants undergoing heart surgery utilizing deep hypothermic circulatory arrest. Pediatrics. 2001; 107: 61-70. [abstract]: PedsCCM EBJC Review

 

Comments

Back to the EB Journal Club Index