Accuracy of Using Medicaid Claims to Identify Children with Asthma for Influenza Vaccination

Background: Influenza vaccination rates are consistently low among children with asthma, a group at increased risk for influenza complications. Michigan public health officials implemented a new initiative in 2006, using administrative data to identify children with a high-risk indicator in the statewide immunization registry.

Objectives: Assess the accuracy of claims from state-administered programs to identify children with asthma for influenza vaccination.

Methods: Michigan Department of Community Health administrative data (2005-2006) were analyzed to identify children 2-18 years with asthma. Children were classified using administrative data into three mutually-exclusive asthma case definition groups: Title V Program qualifying diagnosis of disease (Title V), >1 asthma medication claims (Rx), or >1 health services claim reporting an asthma diagnosis without medication (Dx).  Children were randomly selected from each asthma case definition group; parents were invited to participate in a telephone interview documenting physician diagnosis of asthma, symptoms, medications, and health services use.  Asthma control was classified using National Heart, Lung, and Blood Institute (NHLBI) criteria.  The positive predictive value (PPV) was determined for each asthma case definition group using parent report of a physician diagnosis of asthma, wheezy-cough, or reactive-airway disease (RAD) as the gold standard; differences were assessed using chi-square tests.

Results: In total, 440 interviews were completed (78% participation); 89% of parents confirmed the child's high-risk status, reporting physician diagnosis of asthma (83%), wheezy-cough (5%), and RAD (0.5%). The PPV varied (p<0.0001) between the Title V (100%), Rx (91%) and Dx (73%) asthma case definition groups. Though control levels were similar among cases identified from Title V and Rx (p=0.94), cases identified based solely on Dx had a higher degree of asthma control (p=0.02).

Conclusions: Data from state-administered programs can be used to accurately identify children with asthma and may represent a feasible approach for Medicaid programs and health plans to identify asthma cases for targeted influenza vaccination reminders.