Evaluation of Young Adult Hepatitis C Surveillance in Philadelphia: January 1-December 31, 2012 *

BACKGROUND: Recent studies have indicated increasing rates of hepatitis C (HCV) infection among young adults.  To characterize clinical and social risk factors for HCV among young adults in Philadelphia, the health department initiated an enhanced surveillance effort.   Young adult HCV cases are investigated through two approaches:  case patient interviews and mailed provider questionnaires.  The surveillance system was evaluated to assess performance and to gauge the utility of the two investigative approaches.  

METHODS: The evaluation examined newly reported HCV cases from January 1 through December 31, 2012 among individuals less than 31 years of age.  Completeness of patient interviews and provider questionnaires was assessed by the proportion of responses marked as missing or unknown.  Kappa coefficients (k) of agreement were calculated between risk factor information reported by patients and providers. Chi-square was used to examine system sensitivity and representativeness by comparing demographic characteristics of responders versus non-responders.

RESULTS:  Investigations were completed for 216 of 320 eligible cases.  The analysis of representativeness and sensitivity found a significant difference in the gender distribution of responders relative to non-responders (48% vs. 34% female, p<0.03). The 133 patient interviews, with 60 clinical and risk factor questions, demonstrated an overall completeness of 70%, whereas the 162 provider interviews, with 32 questions, displayed a completeness of 41%.  Completeness was lowest for clinical symptoms and vaccination histories among patient interviews, and for vaccination and exposure histories among provider questionnaires.  Among the 79 cases in which both patient and provider investigations were completed, high agreement (κ >0.8) was seen for gender of sex partners and history of injection drug use.  Slight to poor agreement (κ <0.2) was seen for underlying illness and treatment for HCV.  Acceptability of the system among providers was low, with only 50% of the 226 linked providers returning at least one questionnaire.  The system demonstrated substantial flexibility, with additional variables being added or dropped following interviewer feedback or new data requirements.

CONCLUSIONS:  The low levels of completeness seen among provider questionnaires, in conjunction with the relatively high agreement seen among fields of interest such as injection drug use and sexual behavior suggests the need for a shift toward a patient interview-only approach.  Focusing upon patient interviews would increase data quality, although this methodology would be more resource-intensive for health departments.  Valuable risk factor information was ascertained both through patient interviews and provider questionnaires, indicating substantial surveillance system utility.  This information has informed education and prevention strategies.