Evaluation of a computer algorithm for promoting both annual hepatitis C screening and hepatitis B vaccination among HIV-infected individuals (ORCHESTRA programme)

Sirs: We read with interest the article by Garvey et al. ¹ published in the February 2011 edition of the International Journal of STD and AIDS, regarding the management of hepatitis C virus (HCV)/hepatitis B (HBV) co-infection in HIV-infected patients in the UK. The authors reported data on HBV and hepatitis A virus (HAV) vaccine cover and the results of regular screening for HCV in HCV-seronegative, HIV-seropositive patients. They reported that, in keeping with UK guidelines, two-thirds of sites proposed annual HCV serological screening to non-immune patients. ² This study confirmed that patients co-infected by HIV and a hepatitis virus are under-vaccinated, as 13% of HAV-non-immune, HIV/HCV-co-infected patients had not received HAV vaccination, while 27% of HIV/HCV-co-infected patients with no (5% of patients) or only partial immunity (22%) to HBV had not received HBV vaccination. Finally, 20% of HIV-infected patients had not been screened for HCV. ³ Garvey et al. suggested that the creation of a dedicated computer system could improve the rates of both hepatitis C screening and vaccine cover.

In our hospital, which has an active file of 3000 HIV-infected patients, 1100 of whom are co-infected by HCV, we encounter similar difficulties with vaccination and hepatitis C screening in HIV-infected, HCV-seronegative patients. Given the inadequate HBV and HAV vaccination of our patients, and the need to begin yearly screening for hepatitis C, we created a computer algorithm to analyse the history and dates of HCV serological screening, and to issue automatic reminders on the need to vaccinate patients lacking immunity to HBV and HAV. For each HIV-infected patient, the programme scans the patient's computer file for the relevant information (HCV serology, HCV polymerase chain reaction [PCR], HBV/HAV serology, etc.). If the information is lacking, a pop-up window prompts the clinician to test the patient. If the clinician accepts to do so, the programme adds the test order to the prescription that is printed automatically at the end of each consultation. The information is sought in the following order: existence and date of HCV serological testing, then existence and date of HCV PCR in case of HCV seropositivity. The same algorithm is applied to anti-HAV and anti-HBV vaccination. The programme then checks for anti-HAV, HBV surface antigen (HBsAg) and/or HBV surface antibody (anti-HBs) status. If this information is lacking, orders for the relevant serological tests are added to the prescription printed at the end of the consultation. The programme also prompts the clinician to enter the results during the following consultation.

One year after implementing this system, the proportion of HIV-infected patients whose last negative HCV screening test dated back more than three years fell from 46% to 24% (P < 0.001) among the 2074 patients initially concerned. Yearly HCV screening tests in patients who had not been tested for more than three years revealed that 1% of them were ‘newly infected’. It would be interesting to know the impact of such a measure in the UK, where it is recommended in national guidelines. Indeed, although we support regular HCV screening for this population, it is not currently included in French guidelines. In addition, the diagnostic yield and cost-effectiveness of annual HCV screening has not been studied.

We found that these measures failed to increase HBV and HAV vaccine uptake, which remains alarmingly low. Indeed, one year after their implementation there was no increase in immunization against hepatitis A (30% of patients were not immunized against hepatitis A) both before and after implementation, (NS) or against hepatitis B (7% versus 10%, NS). These results suggest that computer algorithms can improve regular screening for hepatitis C but are not sufficient to increase HBV or HAV vaccine cover among HIV/HCV-co-infected patients, at least during the first year after implementation.