This study is a retrospective register study applying Danish national health and administrative registers for the period 1 January 2013 to 31 December 2022. The registers include The Civil Registration System (CRS) for information on age, sex and residence region [15]; The Danish Microbiology Database (MiBa); the National Patient Register (NPR) for information on ICD-10 codes and procedure codes, including codes for specific treatments, and information on Diagnosis Related Groups (DRG) tariffs [16, 17]; the Register of Selected Chronic Diseases (RUKS) for information on asthma; the National Health Insurance Service Register for information on contacts, services received and fees in the primary healthcare sector [18]; the Cause of Death Register for information on date of death [19]; the Register of Medicinal Product Statistics for information on expedition date and prices for all prescription drugs sold in Danish community pharmacies [20]; The Danish Education Register for information on parents education [21].
Each register contains individual level observations. Individuals can be identified through a unique 10-digit personal identification number, making it possible to securely merge data from the different registers. Furthermore, individuals are listed with mothers’ and fathers’ personal identification number in the civil registration system, allowing for merging of intergenerational data.
Study population
RSV cases were identified using NPR. Individuals with RSV-related hospital admissions were cross-referenced with MiBa, a national microbiology register used for the surveillance of infectious diseases. If a positive RSV test in MiBa was observed within 10 days (+/-) of a hospital admission where an RSV related diagnoses code was listed as primary or secondary diagnosis, the individual was classified as an RSV case. The 10 days leeway between hospital admission and positive RSV test was to allow for potential frictions between date registrations in the two different registers. The criteria for RSV related diagnoses are presented in Table 1. As most of the infants with RSV in Denmark are admitted in their first six months [5] and the majority of infants are admitted before the age of one year, only RSV cases who were younger than six months at time of hospital admission were included in the base case scenario. In a sensitivity analysis, this inclusion criterion was expanded to including all cases who were younger than 12 months at time of hospital admission.
Each RSV case was subsequently pairwise matched with five individuals who were not observed with an RSV hospitalisation to constitute a reference group. The five control individuals were unique for the RSV case, but the controls could be matched with multiple RSV cases – i.e., matching with replacement. The matching covariates were age (+/- 30 days), sex, region of residence, birth (pre/full term), number of siblings < 7 years old, and parents’ education (highest education of any parent). Each individual in the reference group was assigned a pseudo index date to reflect the age at index of the matched RSV case.
Outcomes
The case and reference population were compared in terms of use and costs of primary and secondary healthcare services as well as use of prescription drugs. The use and cost of healthcare services attributable to RSV was calculated as the mean difference between the case and reference group.
Hospital contacts were described as inpatient and outpatient contacts, based on length of stay. Hospital contacts < 12 h were classified as outpatient contacts. Hospital contacts ≥ 12 h were classified as inpatient contacts. Cost of hospital care were calculated based on DRG tariffs. The DRG tariffs are a function of the patients age, diagnosis, length of stay, procedures and treatments provided in a hospital contact. Based on economic and activity data from all Danish hospitals, the tariffs are estimated each year to reflect updated national mean costs for inpatient and outpatient hospital contacts.
Cost of primary care were estimated based on the reimbursement fees to primary care practitioners (general practitioners (GPs) and privately practicing specialists). Cost of prescription medications were based on the pharmacy sales prices. All costs associated with prescription medications and primary care contacts financed by the government, including routine child examinations, were included in the analysis.
All prices were converted to 2023 prices through the consumer price index provided by Statistics Denmark. Danish currency (DKK) was transformed to EUR using the conversion rate 1 EUR = 7.5 DKK.
Finally, previous studies have demonstrated an association between RSV hospitalisation in infancy and the development of asthma – with the causal link being unaccounted for. To place the results of the study in context, we provide an overview of asthma in the RSV case and control population, and for the parents of RSV cases and parents of the controls. Asthma was defined in the register of select chronic diseases, where an algorithm depending on age, hospital contacts and collection of prescription medication is used to determine asthma [22]. Any registration of asthma between hospital contact until end of follow-up was identified in the RSV and control population, while any registration of asthma since 1994 onwards in the adult population was used to determine prevalence of asthma.
Ethical statement
The study was conducted in accordance with legal and regulatory requirements and with scientific purpose, value, and rigour. The study followed the generally accepted research practices described in International Ethical Guidelines for Epidemiological Studies, issued by the Council for International Organizations of Medical Sciences (CIOMS).
Data was stored on Statistics Denmark’s research computers – an environment maintained by Statistics Denmark, fulfilling all requirements for storage and handling of personal sensitive data. Data was accessed through a log-in solution using a two-step validation and administrative access control. While data was available at individual level, all personal identification numbers were encrypted by Statistics Denmark before data was made available to the researchers.
No data were sent from the server to open environments. Only aggregated tables containing at least five observations were exported as part of the research process.
Statistical analysis
An RSV episode was defined as the seven days prior to hospitalisation to 30 days after initial hospitalisation. Furthermore, we studied the year following the RSV episode, divided into two six-month periods.
The data was analysed in a difference-in-difference with variation in treatment timing framework [23], using a linear ordinary least square regression model. The difference-in-difference approach is frequently used in instances where it is unethical or not possible to conduct a randomized control trial. The method can identify either the within-group differences (before and after event – in this case RSV event), in outcome, or between-group differences (difference between case and control group). In this study, the focus was to identify the between-group differences – i.e., differences attributable to the RSV event. When identifying the between-group differences, the underlying assumption is that the two groups are equal prior to the event (parallel trend assumption).
The young age of the study participants in this instance makes it difficult to assess the similarity between the RSV cases and their controls prior to the RSV event. The parallel trend assumption is therefore, in theory, applied by matching on both health determinants at birth, and socioeconomic position of the parents. Therefore, if no event occurs, the two groups will remain equal over time. Any differences observed between the groups will therefore be attributable to the event of interest. To ensure a balanced analysis in all periods, each reference individual was weighted. The weights were calculated such that the sum of weights for controls to each RSV case was equal to 1.
Statistically significant differences between the two groups were evaluated by testing the difference in the mean for each group in each period. A two-sided t-test was applied with a significance level of 0.05. When analysing the differences in asthma between RSV cases and controls, or the parents of the two groups, a chi-square test was conducted.
As a sensitivity test, the inclusion criteria were expanded to all individuals younger than one year at time of RSV hospitalisation, in the same analytical framework. An analysis identifying RSV cases only using ICD-10 codes was also conducted. Furthermore, the outcomes were evaluated for up to nine years after the initial RSV episode.
Add Comment