Object: Because there is no gold standard for preoperative diagnosis of shunt failure, understanding the sensitivity, specificity, and predictive values of symptoms, signs, and diagnostic tests enables practitioners to make logical clinical decisions. Parents of children with shunts undergo educational instruction to enable them to recognize shunt failure. The authors prospectively investigated parental ability to recognize shunt failure.
Methods: Data were prospectively collected on 205 consecutive encounters in 153 children with shunted hydrocephalus presenting to the emergency department or clinic, or as an inpatient consultation, to the Children's Hospital of Alabama between April and October 2010. Regardless of the complaint, all parents were asked if they believed the shunt was in failure. Six children were excluded from analysis because a parental response was lacking. Using the Shunt Design Trial definitions, shunt failure was diagnosed intraoperatively or ruled out if the child did not undergo shunt revision within 1 week of presentation. Sensitivity, specificity, predictive values, and accuracy were calculated using the parental response and shunt failure diagnosis. Secondarily, parents were compared based on their experience with shunt failure in their children; experienced parents were defined as having experienced at least 3 shunt failures. Post hoc analysis evaluated diagnostic test characteristics among hydrocephalus causes and compared parental recognition of shunt failure to head CT and shunt series diagnostic test characteristics. Parents also completed a standardized shunt failure survey regarding their shunt teaching education and symptom tracking.
Results: Children enrolled were a mean age of 6.9 years old, 92 (46%) of the encounters were with male patients, and most patients were Caucasian (69%) and had undergone an average of 2.8 previous shunt revisions. Seventy-one children (36%) were diagnosed with shunt failure. Parental response diagnostic test characteristics were: positive predictive value (PPV) of 41%, negative predictive value (NPV) of 79%, sensitivity of 83%, specificity of 34%, and accuracy of 52% for shunt failure. Sixty-three parents were considered experienced and responded with a PPV of 29%, NPV of 92%, sensitivity of 94%, specificity of 23%, and accuracy of 41%. One hundred thirty-six parents were considered inexperienced and responded with a PPV of 48%, NPV of 75%, sensitivity of 80%, specificity of 41%, and accuracy of 57%. When statistically compared, experienced parents had significantly lower PPV (29% vs 48%, respectively; p = 0.035) and accuracy (41% vs 57%, respectively; p = 0.049) than inexperienced parents. On post hoc analysis, parental recognition of shunt failure was inferior to head CT and shunt series diagnostic tests with a lower specificity (20% vs 88%, respectively; p < 0.0005), PPV (44% vs 84%, respectively; p < 0.0005), NPV (61% vs 85%, respectively; p = 0.006), and accuracy (47% vs. 85%, respectively; p < 0.0005).
Conclusions: The overall parental response had the greatest value in ruling out shunt failure, reflected in the high NPV, particularly in experienced parents. The head CT and shunt series provide more favorable diagnostic test characteristics than the parental response. Although educational interventions have decreased shunt-related deaths, parents have difficulty differentiating shunt failure from alternative diagnoses.