Prevalence of retinal hemorrhages in pediatric patients after in-hospital cardiopulmonary resuscitation: a prospective study.

OBJECTIVE: Child abuse occurs in 1% of children in the United States every year; 10% of the traumatic injuries suffered by children under 5 years old are nonaccidental, and 5% to 20% of these nonaccidental injuries are lethal. Rapid characterization of the injury as nonaccidental is of considerable benefit to child protection workers and police investigators seeking to safeguard the child care environment and apprehend and prosecute those who have committed the crime of child abuse. Physically abused children present with a variety of well-described injuries that are usually easily identifiable. In some cases, however, particularly those involving children with the shaken baby syndrome, obvious signs of physical injury may not exist. Although external signs of such an injury are infrequent, the rapid acceleration-deceleration forces involved often cause subdural hematomas and retinal hemorrhages, hallmarks of the syndrome. Frequently, retinal hemorrhages may be the only presenting sign that child abuse has occurred. Complicating the interpretation of the finding of retinal hemorrhages is the belief by some physicians that retinal hemorrhages may be the result of chest compressions given during resuscitative efforts. The objective of this study is to determine the prevalence of retinal hemorrhages after inpatient cardiopulmonary resuscitation (CPR) in pediatric patients hospitalized for nontraumatic illnesses in an intensive care unit.
DESIGN: Prospective clinical study.
SETTING: Pediatric intensive care unit. PATIENTS: Forty-three pediatric patients receiving at least 1 minute of chest compressions as inpatients and surviving long enough for a retinal examination. Patients were excluded if they were admitted with evidence of trauma, documented retinal hemorrhages before the arrest, suspicion of child abuse, or diagnosis of near-drowning or seizures. All of the precipitating events leading to cardiopulmonary arrest occurred in our intensive care unit, eliminating the possibility of physical abuse as an etiology.
INTERVENTIONS: None. MEASUREMENTS: Examination of the retina was performed by one of two pediatric ophthalmologists within 96 hours of CPR. The chart was reviewed for pertinent demographic information; the platelet count, prothrombin time, and partial thromboplastin time proximate to the CPR were recorded if they had been determined.
RESULTS: A total of 43 pediatric patients hospitalized with nontraumatic illnesses survived 45 episodes of inpatient CPR. The mean age was 23 months (range, 1 month to 15.8 years), and 84% of the patients were under 2 years old. The majority of the patients (44%) were admitted to the intensive care unit after surgery for congenital heart disease, and another 21% were admitted for respiratory failure. The mean duration of chest compressions was 16.4 minutes +/- 17 minutes with 58% lasting between 1 and 10 minutes. Five patients had chest compressions lasting >40 minutes, and two patients had open chest cardiac massage. All patients survived their resuscitative efforts. Ninety-three percent of patients had an elevated prothrombin time and/or partial thromboplastin time while 49% were thrombocytopenic. Sixty-two percent of the patients had low platelet counts and an elevated prothrombin time and/or partial thromboplastin time. Small punctate retinal hemorrhages were found in only one patient.
CONCLUSIONS: Retinal hemorrhages are rarely found after chest compressions in pediatric patients with nontraumatic illnesses, and those retinal hemorrhages that are found appear to be different from the hemorrhages found in the shaken baby syndrome. Despite the small number of patients in this prospective study, we believe that these data support the idea that chest compressions do not result in retinal hemorrhages in children with a normal coagulation profile and platelet count. A larger number of patients should be evaluated in a prospective multi-institutional study to achieve statistical significance