Communicable Disease Surveillance Plagued by Low Reporting: Comprehensive Analysis of North Carolina Statewide Data

Background:   Communicable disease (CD) surveillance is the key method by which states measure endemic disease incidence, recognize disease outbreaks, evaluate prevention and control measures, allocate public health resources, and further describe emerging disease epidemiology. Despite the widespread usage of surveillance data, completeness of reporting has never been comprehensively assessed. This is the most comprehensive study to date of reporting completeness with an analysis of over 50 diseases over 10 years across NC (Population rank #11).

Objective:   To describe CD-specific reporting completeness

Methods: A retrospective cohort study was conducted for the years 1995-97 and 2000-06 for all patients assigned an ICD-9-CM diagnosis code for a state required reportable CD in 7 healthcare systems which represent 25% of inpatient and 17% of outpatient visits in NC. Case-patients were matched by unique identifiers to surveillance data.  Summary and disease-specific completeness proportions with 95% confidence and uncertainty intervals were estimated. Positive predictive values (PPV) of ICD-9-CM codes for CD surveillance were also estimated.

Results:   Disease-specific reporting completeness proportions ranged from 0-78.6% with an overall median of 8.5%.  Ten diseases accounted for 90% of disease reporting: salmonellosis, TB, meningococcal disease, campylobacteriosis, RMSF, shigellosis, acute hepatitis A, pneumococcal meningitis, legionellosis and malaria; preliminary PPVs for ICD-9-CM diagnosis codes ranged from 7.9- 71.4% (Table).  Aerosol and droplet transmitted diseases were reported half as often (5.8% [5.1,6.5] vs 9.8% [9.0,10.7]) as all other diseases.  Food and waterborne transmitted diseases were reported 4 times as frequently (18.8% [17.2, 20.5] vs 4.6% [4.1,5.1]) as all other diseases. Person-to-person transmitted diseases were reported about 2.5 times as frequently (10.6% [9.8, 11.5] vs 4.0% [3.4, 4.7]) as all other diseases.  Laboratory reportable diseases were reported 5 times as frequently (8.5% [8.0, 9.1] vs 1.8% [0.9, 3.0]) than diseases reportable only by physicians.  

Conclusions:   Disease reporting completeness was very low even for diseases with great public health importance and opportunity for preventing person-to-person transmission (e.g., meningococcal meningitis: 20.4%).  Low reporting proportions may be due to (1) perceived low importance of reporting by clinicians (2) assumption that someone else will report (3) over-coding of ICD-9-CM codes (low PPVs) for CD surveillance (eg, does not meet case definition, suspect disease, history of disease).  Improving communicable disease reporting is crucial because these data are used to inform prevention and control measures and allocate public health resources.  In addition, the accuracy of ICD-9-CM codes needs to be further validated for CD surveillance and research.