Study design and data collection
We conducted a retrospective observational study with the year 2015 as the baseline and the year 2018 as the first year of follow-up.
We identified all cases of hepatitis B and C reported to the National System for Notifiable Diseases (SmiNet) and with a permanent personal identification number (see definition below) from the years 1969 and 1990 (the time-points when hepatitis B and C became notifiable by law, respectively [4, 5]) until 31 December 2018 (Fig. 1). In Sweden, both hepatitis B and C are notifiable by law, both by the clinicians and by the laboratories [14]. The two notifications are then merged into a “case” in the national database for notifiable infections “SmiNet”. The clinical notification should include, among other clinical data, information about acute or chronic infection, plausible route of transmission as well as the country of infection. Though not mandatory, the laboratory method (serology or PCR) can be reported as well as if the result was positive, negative or indeterminate. Subsequent tests are reported under the same personal identification number but are not automatically performed. We extracted the personal identification number, age at time of notification, sex, country of infection, date of notification in SmiNet, and chronicity of infection as reported by the clinician and laboratory method used for diagnosis (serology or PCR). Cases were linked by personal identification number to the population registry at the Swedish Tax Agency to add date of death, emigration, or removal from the population registry for other reasons [15].
These data were further linked to the national registers at the National Board of Health and Welfare, the National Patient Register, the Prescribed Drug Register, and the Cause of Death Register (Fig. 1). In the Patient Register all inpatient hospital visits and all outpatient visits to specialist care are recorded with diagnoses coded according to the 10th International Classification of Diseases (ICD-10) [16]. We used the ICD-10 codes suggested in the WHO framework [2]. The following specific codes were extracted for diagnoses of diseases attributable to HBV and HCV infections: HCC or liver cancer of unknown type (C22.0 and C22.9), liver cirrhosis (K74.4- K74.6), and chronic liver disease (K72, K73, K74, K75). In the Prescribed Drug Register all prescribed drugs have been recorded on an individual level since July 2005 [17], and we extracted treatment for HCV infection using the following Anatomical Therapeutic Chemical Classification System (ATC) codes: L03AB for IFNs (Interferons—IFNs and pegylated (peg)-IFNs) and J05AP for DAAs and ribavirin. Finally, we linked cases to the Cause of Death Register [18]. We specifically extracted ICD-10 codes for liver disease as defined above, and causes of death with these ICD-10 codes will be further referred to as “liver-related deaths” in this study. Aggregated data on size and demographics of the general population were extracted from Statistics Sweden [19].
Operational definitions and assumptions
Personal identification number
The personal identification number is a unique national official identifier given to each Swedish resident and follows a person for their entire life. It is obtained at birth or when immigrating to Sweden. Because all national registers use this unique number, it is possible to link different data sources. After arriving in Sweden, it usually takes at least 3 months to be registered and obtain the permanent personal identification number for migrants. In the meantime temporary registration numbers are generated at the local level when visiting health care facilities, and one person could get more than one temporary number. These cannot be linked across registers. Once a person obtains a permanent personal identification number, it is not possible to link data associated with the temporary registration number to the permanent personal identification number. We refer to this temporary registration number as the “temporary personal identification number” in this study. In this group we also included the notifications where the full permanent personal identification number was missing for unknown reason. We assumed cases reported with a temporary personal identification number would be reported again under a permanent personal identification number if they remained in Sweden. We therefore focused our analyses on those with permanent personal identification number to avoid duplicate notifications and to be able to link to other national registers.
Clearance and cure rates
We estimated clearance and cure rates in order to calculate the prevalence of chronically infected individuals with HBV or HCV diagnosis. The percentage of spontaneous clearance after notification for acute HBV infections was estimated to be 5% for infants (< 1 year of age) and 95% for the remaining cases [2]. We assumed no spontaneous clearance when the clinician notified the infection as a “chronic” HBV infection.
In Sweden, HCV infections used to be reported based on a positive anti-HCV IgG serology test according to the case definition used until 2020. Thus, we could not know if these infections were chronic or had cleared spontaneously. Based on a meta-analysis [20], we assumed a 37% spontaneous clearance rate for all reported and non-treated HCV infections. But estimations of spontaneous clearance vary. The WHO estimates it to be 15% -45%, the ECDC 20%, United States Centers for Disease Control and Prevention 50% and The Swedish Reference Group for Antiviral Therapy estimates it to be 25%–50%. Due to this variation in the estimates, a sensitivity analysis around the prevalence was conducted assuming a spontaneous clearance rate for all reported and non-treated HCV infections of 25% and 50%.
For treated HBV-infected individuals, the functional cure rate is low [19], and for the analyses we therefore assumed no cure rates after antiviral treatment [11], and thus did not collect medication data. For treated HCV-infected individuals, we assumed treatment cure rates of 50% for those who had received IFNs in combination with ribavirin [21] and 95% for those who had received DAAs [13]. For the analyses, we assumed that there were no re-infections after HCV-treatment.
These clearance and cure rates were applied to the number of cases reported in SmiNet as described in the Data Analysis paragraph.
Diagnosis of liver disease
Liver disease at the time of HBV or HCV diagnosis was defined as the occurrence of one of the codes specified by the WHO (see above) in the Patient Register within one year before or after the date of HBV/HCV notification, for those notified in 2015 and 2018.
Data analysis
For continuous data we calculated the median and interquartile range. We followed the WHO and ECDC guidelines for calculating the prevalence, notification rate, liver-related morbidity, and liver-related mortality [2, 3].
The prevalence of diagnosed chronic HBV and HCV infections in 2015 and in 2018 was calculated by dividing the total number of individuals living with a diagnosis of chronic HBV or HCV infection by the total population in Sweden in the respective years. In 2015 and 2018, 9,851,017 and 10,230,185 people were residing in Sweden, respectively. The numerator included the total number of cases reported in SmiNet since HBV and HCV infections became notifiable diseases, excluding cases where deaths and emigrations had occurred as of January 1st of the next year (2016 and 2019), and the estimated number of cleared infections as described above (spontaneously or by treatment).
For both hepatitis types, we calculated the notification rates for 2015 and 2018, respectively, as the number of new cases of hepatitis reported in SmiNet in each year divided by the total number of residents in Sweden in that year.
For both prevalence and notification rates, we made additional calculations also including those with a temporary personal identification number.
We calculated the rates of liver-related morbidity at the time of HBV or HCV diagnosis for each year (2015 and 2018) as the proportion of newly reported cases who also had received at least one of the ICD-10 codes indicating liver disease (as specified by the WHO) within one year before or after the date of diagnosis of the infection.
We calculated liver-related mortality attributable to HBV and HCV as the number of deaths with at least one ICD-10 code indicating liver disease (as specified by the WHO) as the cause of death, divided by the total number of deaths among people living with HBV or HCV infection for 2015 and 2018, respectively.
We calculated Wilson 95% confidence intervals (CIs) for proportions and Poisson 95% CIs for notification rates. P-values for the differences in proportions were calculated using Fisher’s exact test. All analyses were conducted in STATA 15.1.