Background: Robust PID surveillance is important for monitoring STD outcomes. Administrative data analysis may provide a valuable method for assessing PID burden, but must be validated.
Methods: We analyzed 2009-2010 claims from a large publicly-funded family planning program to identify females within four PID case definitions based on ICD-9 codes and treatment claims: PID diagnosis with (1)CDC-recommended treatment, (2)non-standard antibiotics, (3)no treatment, or (4)rule-out PID (non-specific abdominal pain). A sampling frame based on PID case numbers in each of the groups was determined and stratified by public versus private provider (N=1,593). A subset (N=500) was selected for which medical records were requested and reviewed by a clinician. Sensitivity and specificity of claims-based case definitions were determined using abstractor diagnosis as the gold standard. The proportion of validated PID cases with chlamydia/gonorrhea testing, appropriate treatment, and follow-up <7 days was assessed and compared by provider type.
Results: Of 239 records abstracted, 185 (65%) were validated as PID. PID claims with CDC-recommended treatment had the highest specificity (98%), but low sensitivity (36%). Expanding to include all PID diagnosis claims (regardless of treatment) and rule-out PID with CDC-recommended treatment increased sensitivity (75%), but decreased specificity (85%). Of validated PID cases, 39 (21%) had no provider PID diagnosis; alternate diagnoses included lower abdominal pain and urinary tract infection. Sixty-one percent received CDC–recommended treatment, 28% received non-standard antibiotic treatment; 85% were tested for chlamydia/gonorrhea and 30% had follow-up <7 days. Private providers had lower testing (70% vs. 96%) and follow-up rates (9% vs. 47%) than public providers.
Conclusions: Although a large proportion of claims with PID diagnosis were validated, many true cases may still be missed and the total burden underestimated. Partnerships between health departments, training institutions, and academia should be developed to monitor and improve the quality of PID diagnosis and management.