Drug-utilisation profiles and COVID-19

Coronavirus disease 2019 (COVID-19) has substantially challenged healthcare systems worldwide. By investigating population characteristics and prescribing profiles, it is possible to generate hypotheses about the associations between specific drug-utilisation profiles and susceptibility to COVID-19 infection. A retrospective drug-utilisation study was carried out using routinely collected information from a healthcare database in Campania (Southern Italy). We aimed to discover the prevalence of drug utilisation (monotherapy and polytherapy) in COVID-19 versus non-COVID-19 patients in Campania (~ 6 million inhabitants). The study cohort comprised 1532 individuals who tested positive for COVID-19. Drugs were grouped according to the Anatomical Therapeutic Chemical (ATC) classification system. We noted higher prevalence rates of the use of drugs in the ATC categories C01, B01 and M04, which was probably linked to related comorbidities (i.e., cardiovascular and metabolic). Nevertheless, the prevalence of the use of drugs acting on the renin-angiotensin system, such as antihypertensive drugs, was not higher in COVID-19 patients than in non-COVID-19 patients after adjustments for age and sex. These results highlight the need for further case–control studies to define the effects of medications and comorbidities on susceptibility to and associated mortality from COVID-19.

www.nature.com/scientificreports/ archives are linked together by a unique anonymous identifier that is encrypted to protect patient privacy. Our research protocol adhered to the tenets of the Declaration of Helsinki 1975 and its later amendments. Permission to use anonymised data for this study was granted to the researchers of the Centro di Ricerca in Farmacoeconomia e Farmacoutilizzazione (CIRFF) by the governance board of Unità del Farmaco della Regione Campania. The research did not involve a clinical study, and all patients' data were fully anonymised and were analysed retrospectively. For this type of study, formal consent was not required according current national established by the Italian Medicines Agency, and according to the Italian Data Protection Authority, neither ethical committee approval nor informed consent was required for our study 21 .
Study population. People who had been dispensed medication according to CaReDB during 2019 were included in the study cohort. From regional surveillance system data, we obtained the information of patients with confirmed COVID-19 from the beginning of the epidemic (26 February 2020) to 30 March 2020 who were linked to the population identified in CaReDB. For the purposes of our investigation, the study population diagnosed with SARS-CoV-2 infection on or before the date of analysis was referred to as the 'COVID-19 group' (C19G). The remaining individuals were used as a comparator group in the analysis and were referred to as the 'general population group' (GPG).
Patient characteristics. The study population was categorised by sex and subdivided into four age groups: 0-39; 40-59; 60-79; and ≥ 80 years. The number of drug prescriptions, prevalence of drug use and polypharmacy regimens (classified as 'no-polypharmacy'; 'polypharmacy'; and 'excessive polypharmacy') were ascertained in 2019. Drugs were grouped according to the Anatomical Therapeutic Chemical (ATC) classification system. ATC II and ATC IV codes with a prevalence ≥ 3% in the C19G were included in the analysis.
Outcome. The drug-utilisation profile was evaluated as the prevalence of drug use. Drug use prevalence was estimated as the number of individuals dispensed ≥ 1 drug prescription per 100 inhabitants in 2019. The prevalence of drug use was evaluated in the C19G and GPG. Prevalence was stratified by age group and sex. Prevalence was probably influenced by the heterogeneous demographic distribution among the age groups, so we conducted direct standardisation.
Statistical methods. The baseline characteristics of the study population were analysed using descriptive statistics. Quantitative variables are described as means ± standard deviations. Categorical variables are described as counts and percentages. The chi-square test and t-test were performed to determine the difference between the C19G and GPG in terms of sex and age. A P-value of < 0.05 was considered significant. Crude and age-adjusted prevalence rates were calculated. Differences in the prevalence between the C19G and GPG are expressed as risk ratios (RRs) adjusted for sex and age with 95% confidence intervals (CIs). Standardisation was performed using a direct method whereby the Italian population up to 1 January 2019 was used as the standard population (available on the Demo Istat website 22 ).
where (T i = n i /n) = rate in stratum 'i' of the study population; n i = number of cases in stratum 'i' of the study population; N = size of the study population in stratum 'i'; w i = size of stratum 'i' of the reference population; m = number of considered strata; k = multiplicative constant. The age-adjusted RRs and 95% CIs were computed using standard methods. Data management was performed with SQL server v2018 (Microsoft, Redmond, WA, USA). Analyses were carried out with SPSS v17.1 (IBM, Armonk, NY, USA).

C19G characteristics.
A total of 1,532 individuals in Campania who tested positive for COVID-19 on 30 March 2020 were identified. Of these, 926 (60.4%) were males, and the median age of the entire sample was 55 ± 19 years. Among the C19G patients, 20.8% were aged 0-39 years, 36.1% were aged 40-59 years, 33.6% were aged 60-79 years, and 9.5% were aged > 80 years. The percentage of males was higher than that of females in all age groups except the > 80 years age group (43.8% males). Differences in age and the sex ratio between the C19G and GPG were statistically significant (p-value < 0.001).
Half of the COVID-19 patients aged 0-39 years had no exposure to any medication, whereas 45.5% of the COVID-19 patients were prescribed ≤ 4 medications, and 4.1% had polypharmacy regimens (5-9 drugs). The percentage of participants receiving polypharmacy increased with increasing age, at 18.3% in those aged 40-59 years and 34.8% in those aged 60-79 years; moreover, ~ 80% of participants aged > 80 years were prescribed polypharmacy or excessive polypharmacy (≥ 10 drugs) regimens. The C19G characteristics are shown in Table 1.
Drug-utilisation profiles of the C19G. Twenty-three pharmacological ATC II groups and 39 ATC IV groups had a prevalence > 3% in the C19G. The highest unadjusted and adjusted prevalence rates of drug use in the ATC II groups were observed for drug categories J01, A02, C09, M01, B01 and R03 in the C19G and GPG (Fig. 1). www.nature.com/scientificreports/ Crude differences (at least ± 20% in the overall prevalence of drug use between the C19G and GPG) were found in all 23 pharmacological ATC II groups and in 30 of 39 ATC IV groups included in the analysis (Fig. 1, Table 2). After adjustment, differences remained in six ATC II groups and eight ATC IV groups. With respect to drugs acting on the renin-angiotensin system (RAS) (C09), including beta-blockers (C07), antibacterial drugs for systemic use (J01) and anti-inflammatory and antirheumatic drugs (M01), the differences disappeared after adjustment. The large differences in antithrombotic agents (B01), cardiac therapy drugs (C01) and anti-epileptics (N03) diminished after adjustment, even though they were more common in the C19G than in the GPG after adjustment.
The other ATC II therapeutic group that pertained to the cardiovascular system did not show a relevant difference in the overall adjusted prevalence between the C19G and GPG (Fig. 1). Nevertheless, after stratification by age group, a higher RR (C19G/GPG) in people aged < 60 years was noted. In those older than 80 years, the differences disappeared or reversed, specifically for agents acting on the RAS (ATC II: C09) and lipid-modifying agents (ATC II: C10) (65.6% vs. 71.2% and 34.6% vs. 42.7% in the C19G vs. GPG, respectively) (Fig. 2).
Focusing on chemical subgroups, other antidepressant (ATC IV: N06AX) use had a high prevalence in COVID-19 patients in all age groups except for those aged 40-59 years (Fig. 3).

Discussion
The COVID-19 pandemic has imposed great challenges on healthcare systems worldwide. Some literature has been published on the clinical aspects of, possible treatments for and risk factors in patients with COVID-19 [23][24][25][26] . Nevertheless, apart from a few studies, the epidemiology of and drug use profiles in patients with COVID-19 has not been studied. To our knowledge, this is the first study addressing this topic.
In general, our results revealed four profiles. The first comprises an age range of 0-39 (median age, 27 ± 9) years, male sex, no exposure to any drug in approximately half of the patients and a very low prevalence of polytherapy. The second comprises an age range of 40-59 (median age, 51 ± 5) years, male sex, use of 1-4 drugs in nearly half of the patients and a low prevalence of polytherapy (< 25%). The third comprises an age range of 60-79 (median age, 68 ± 6) years, male sex, use of ≥ 1 drug in 90% of the patients and polytherapy in more than half of the patients. The final profile comprises an age > 80 (median age 85 ± 4) years, female sex, use of ≥ 1 drug in 94% of the patients and polytherapy in 78% of the patients.
The analyses of drug-utilisation profiles highlighted differences between the C19G and GPG in terms of the prevalence of drug exposure. The drug categories with a difference ≥ 30% were antithrombotic agents (B01), antiepileptics (N03), anti-hyperuricaemics/anti-gout (M04) and cardiac therapy agents (C01). The higher prevalence rates associated with drugs in categories C01, B01 and M04 indicate a frequent pattern of cardiovascular and metabolic comorbidities in COVID-19 populations, as reported in other studies 4,5,8 . It is of some relevance that B01 drug use had the largest difference between the COVID-19 and general populations. This therapeutic profile indicates the presence of cardiovascular complications (including venous thromboembolism), supporting the hypothesis of an increased risk associated with COVID-19 infection in these patients 8 .
With regard to the higher prevalence of use of drugs in the M04 category, a retrospective cohort study including 131,565 patients and 252,763 controls, using data from the UK Clinical Practice Research Datalink, reported an increased risk of pneumonia (hazard ratio, 1.27; 95% CI 1. 18-1.36) in patients with gout 27 .
There is no clear association between epilepsy and the risk of developing COVID-19. Nevertheless, epilepsy may be associated with other comorbidities or a component of congenital/inherited syndromes that may affect the immune system. Additionally, anti-epileptic agents can be used in association with other medications that can influence the immune system (e.g., adrenocorticotropic hormones, corticosteroids, everolimus, immunotherapy agents), and this may increase the infection risk 28 . Moreover, these patients may require frequent clinical evaluation, which may explain (at least in part) their increased risk of healthcare-associated infections.
Notably, the adjusted prevalence of the use of drugs acting on the RAS (C09) was not different between the C19G and GPG (RR, 1.02; 95% CI, 1.01-1.02). This result is in accordance with evidence from a retrospective study involving a COVID-19 cohort in Italy 29 and supports the position of the European Society of Cardiology 30 . Furthermore, no major differences were noted for any category of antihypertensive drugs. Corroborating our results, a recent study carried out in the United States revealed no association between ACEI or ARB use and COVID-19, supporting the recommendation of continuing ACEI and ARB use in the setting of the COVID-19 pandemic 31 . This was further explored in a recent Brazilian study that confirmed that among patients hospitalised www.nature.com/scientificreports/ with mild to moderate COVID-19 who were taking ACEIs or ARBs before hospital admission, there were no significant differences in the mean number of days alive and out of the hospital between those who discontinued and continued these medications 32 .
Stratification by age showed a higher prevalence of use of drugs in categories B01, B03, C09 and C10 in people aged < 40 years. This evidence should be interpreted with caution because the number of such patients was very small. Nevertheless, a morbidity pattern similar to that in older patients was observed in these patients. Conversely, in COVID-19 patients aged > 60 years, there was no significant difference in the prevalence of drug use for cardiometabolic diseases compared with that in the GPG, but the prevalence rates of drug use for respiratory diseases and neurological diseases were increased in the C19G.
A large number of males took analgesics (N02) and drugs for cardiac therapy (C01). A high number of females took anti-anaemia agents (B03) and anti-epileptic agents (N03). Early descriptions of COVID-19 suggested a male preponderance 23,24,33 . Sex-based immunological, genetic, and lifestyle differences (e.g., tobacco smoking) have been postulated to explain the male preponderance of COVID-19 34 . In a population of 507 patients with COVID-19 between 13 and 31 January 2020 (including 364 from mainland China), 281 patients were male (55%), and the median age was 46 (IQR, 35-60) years 35 . Zhou and colleagues described 191 COVID-19 patients from Wuhan (Hubei Province, China) during the first month of the outbreak. That cohort had a median age of 56 (IQR, 46.0-67.0) years, with 62% being male and 48% with comorbidities 23 . Additionally, data from Italy revealed a higher prevalence of COVID-19 in males than in females 36,37 . However, sex-and age-disaggregated data revealed the opposite to be true for women aged > 80 years in Campania. National data from Italy revealed that in those aged 20-29 years, 56.5% of the diagnosed patients were female, and only after the age of 50 years does the male preponderance of COVID-19 increase. Thus, the male preponderance of COVID-19 should be interpreted with caution because sex-disaggregated data are incomplete, and more robust evidence is needed.
Our study was not designed to define the association between drug use, comorbidities, risk of adverse outcomes and outcomes in COVID-19 patients. The associations between the use of certain drugs and susceptibility to SARS-CoV-2 infection (e.g., predictive factors for poor outcome) must be studied in a large cohort with a control group and robust clinical data. This was a retrospective study of health records. Additional detailed patient information (mainly regarding clinical outcomes) was not available at the time of the analysis. Despite these limitations, we delineated the drug use profiles and epidemiological and demographic characteristics of 1532 Italian patients with COVID-19. This information provides the first evidence of the association between drug utilisation and COVID-19 risk, giving us a solid background for further analyses and interpretations using new data.

Conclusions
In conclusion, the current data provide baseline information about the complexity of patients affected by COVID-19, showing frequencies and differences in drug utilisation profiles in COVID-19 patients compared with the general population. The higher prevalence rates of C01, B01 and M04 use were probably linked to related comorbidities (i.e., cardiovascular, metabolic). Nevertheless, the prevalence of the use of drugs acting on RAS, such as other anti-hypertensive drugs, did not show a higher prevalence among COVID-19 patients than among the general population. Since these pilot data were derived from the first month of documented COVID-19 cases in the Campania region (southern Italy), our results highlight the need for further case-control studies to define the effects of medications and comorbidities on susceptibility to and associated mortality from COVID-19 infection. Finally, to better understand the global epidemiology of COVID-19, reproducible and comparable results from cohorts from multiple countries and regions are needed for further investigation and meta-analysis. License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http:// creat iveco mmons. org/ licen ses/ by/4. 0/.