In this large register based capture-recapture analysis we estimated the population living with a diagnosis of chronic hepatitis C in Denmark to be 0.21% of the population. Including undiagnosed patients the total estimate of patients with chronic hepatitis C corresponded to 0.38% of the adult population. The higher prevalence among men, persons 40–49years of age and residents in the capital region was in agreement with previous population surveys and the geographic distribution of hepatitis C infected Danish blood donors [3, 18, 19]. The South region had the highest proportion of young men. Here systematic screening for hepatitis C had been performed amongst drug users since 1996. The clinical registers (hospital discharge register and the hepatitis database) had a lower prevalence of persons <40years of age, males and drug users in the treatment register. This could indicate that these groups had less clinical illness due to hepatitis C, or that young males were less likely to seek clinical care for hepatitis C. The latter explanation is in agreement with a recent Danish survey reporting that drug users diagnosed with clinically significant chronic hepatitis C were unlikely to enter hospital based clinical care . It was surprising to us that only about a third of patients diagnosed with chronic hepatitis C attended specialised clinical care. With the improved therapies becoming available for hepatitis C these years, we suggest that new strategies must be developed in order to assure all patients with hepatitis C have proper care, once the disease has been diagnosed . It was also unexpected that half of all patients were not present in the laboratory register as the hepatitis C diagnosis cannot be made without a positive HCVRNA test. There may be several explanations for this: HCV tests became available in 1991–2, but most laboratories did not import old test results when testing was computerised in the late nineties and we had only access to electronic test result. In addition, the participating laboratories only covered 85% of the Danish population.
Our study had several limitations: Capture-recapture analysis requires a closed study population and the same case definition in all registers. To fulfil this we excluded patients who died, and extracted data from all registers at the same day (31.12.2007), but we could not identify patients who cleared chronic hepatitis C, either spontaneously or by treatment except in the laboratory register. Spontaneous clearance of hepatitis C after development of chronic infection is rare and it is estimated that only 2% of hepatitis C patients in Denmark have been treated [13, 14]. This bias could lead to overestimation of the hidden population. As for the case definition this may have changed over time. In the early nineties the clinical diagnosis had in some cases been based on positive anti-HCV (and elevated liver enzymes) in the communicable disease and hospital register. In contrast all cases in the laboratory and the hepatitis database (DANVIR and DANHEP) were based on positive HCV-RNA. Furthermore 9% of patients in the hepatitis database had cleared hepatitis C as a result of treatment . With different case definitions between registers, the hidden population will be over estimated. On the other side, we excluded 4,928 anti-HCV positive persons from the laboratory register as these had no available HCV-RNA test, but it is likely that 62.2% (3064 patients 95%CI 3006–3124) of this group had chronic HCV infection. Our register estimate is lower than the general population estimates from Scandinavia, so most likely the bias mentioned let to an underestimation of the HCV infected population .
Another limitation was that the capture-recapture method only estimated the number of patients diagnosed with hepatitis C and not the population that had never been tested for HCV. A direct estimate would require a general population survey which had not been performed in Denmark. Instead, we estimated the test coverage within the population of hepatitis C infected in the drug treatment register and applied this (46% undiagnosed) to calculate the total population with chronic hepatitis C in Denmark. If the prevalence of hepatitis C was lower among the drug users who had not been tested, this could overestimate the hepatitis C population. As the drug users that had not been tested for hepatitis C were younger such a bias would be expected but an independent population survey, performed in 2007 among drug users in treatment in the county of Funen, found a 40% hepatitis C prevalence, very similar to what we estimated from the national drug treatment register . Furthermore the hepatitis C prevalence among tested ever and never injectors in the treatment register were identical to the survey results (53% and 5%) giving further credibility to our estimate. An explanation of why an age difference did not correspond to a difference in hepatitis C prevalence could be that drug users becomes infected very rapidly after start of injection (50% within the first year) but on average do not enter treatment until after 4years of drug use and therefore the age difference is of less importance .
Another independent source was a national survey from 2004–2008, among 1009 drug related deaths, who were tested post mortem for hepatitis C. Their age and gender distribution was comparable to the drug treatment register (median age 38 years and 22% were female) and the estimated prevalence of chronic hepatitis C was 35% - slightly lower than our 40% treatment register estimate [23, 24]. If the true hepatitis C prevalence was 35% in the treatment register, 62% would have been identified and the corresponding national estimate of hepatitis C infected would decline to 14,783 (0.34% of the adult population).
The estimates of test coverage for hepatitis C (54% and 62%) among drug users in Denmark used two different methods but both were higher than a 42% test coverage reported among drug users in Scotland, suggesting that our national hepatitis C prevalence estimates could be too low .
Another limitation in the above calculations was that we assumed the same test coverage among hepatitis C patients outside the drug treatment register (former and never injectors). We found practically the same diagnostic coverage for a cohort of never injectors (patients infected by blood transfusion), but a cohort of former drug injectors could not be identified.
A comparable hepatitis C estimate from England and Wales (using multiple surveys, register data and mathematical modelling) found that former injectors constituted 47%, current injectors 38% and never-injectors 15% of the total British hepatitis C estimate (0.60% of the adult population) [26, 27]. Corresponding Danish estimates were 15-17% never users and 41% current injectors, deducing that former injectors may constitute 43% of the hepatitis C population [4, 28, 29]. A large proportion of former injectors with unknown but presumably low test coverage would make our estimate more uncertain and most likely too low. Our estimate was also lower than reported from two general population surveys from Sweden and Norway, both reporting an anti-HCV prevalence of 0.7% (corresponding to 0.45% and 0.5% with chronic hepatitis C infection) in the general population [2, 30]. If the true Danish prevalence of hepatitis C infection was 0.45% (corresponding to 19,645 patients), and our estimate was too low due to lower test coverage among infected former drug users, these would constitute 51% of the estimate (10,050) and the diagnosed proportion with hepatitis C amongst former injectors would be 39% (3948).