Initiation of antihypertensive drugs to patients with confirmed COVID-19-A population-based cohort study in Sweden.

PURPOSE: Hypertension is an important risk factor for severe outcomes in patients with COVID-19, and antihypertensive drugs may have a protective effect. However, the pandemic may have negatively impacted health care services for chronic diseases. The aim of this study was to assess initiations of antihypertensive medicines in patients infected by COVID-19.
METHODS: A cohort study including all Swedish residents 20-80 years old with a COVID-19 positive test compared with an unexposed group without COVID-19 matched for age, sex, and index date (date of confirmed COVID-19). Data were collected within SCIFI-PEARL, a study including linked data on COVID tests, hospital diagnoses, dispensed prescriptions, and socioeconomic data from Swedish national registers. Initiations of different antihypertensive drugs were studied from March 2020 until October 2020. Associations between COVID-19 and initiation of antihypertensives were assessed by a multivariable Cox proportional hazards model.
RESULTS: A total of 224 582 patients (exposed and unexposed) were included. After adjusting for cardiovascular comorbidities and education level, ACEi was the most commonly initiated antihypertensive agent to patients with COVID-19. Hazard ratio and 95% confidence interval for initiation of drug therapy was 1.83 [1.53-2.19] for ACEi, followed by beta-blockers 1.74 [1.55-1.95], calcium channel blockers 1.61 [1.41-1.83], angiotensin receptor blockers 1.61 [1.40-1.86], and diuretics 1.53 [1.32-1.77].
CONCLUSION: All antihypertensive medicines were initiated more frequently in COVID-19 patients. This can either be associated with hypertension caused by the COVID-19 infection, more frequent diagnosis of hypertension among people with COVID-19 since they consult health care, or residual confounding factors not adjusted for in the study.

INTRODUCTION

Coronavirus Disease 2019 (COVID‐19) was officially declared a pandemic by the World Health Organization (WHO) on 11 March 2020, and the virus rapidly spread across the world affecting the health of millions of people. , Many studies have shown that patients with cardiovascular risk factors such as hypertension, diabetes mellitus, and obesity experience more severe outcomes of the infection. Consequently, cardiovascular prevention and treatment have become even more important during the pandemic.

However, the intensified focus on COVID‐19, as well as lockdown measures and physical distancing restrictions, may have negatively impacted the management of patients with chronic diseases. , In May 2020, WHO conducted a survey in 155 countries on health care services for chronic diseases during the pandemic. The respondents claimed that COVID‐19 negatively impacted health care delivery in all regions and three quarters of countries reported problems in the management of chronic diseases, including hypertension.

There are currently five major antihypertensive drug classes available on the market: Angiotensin‐converting enzyme (ACE) inhibitors, angiotensin receptor blockers (ARBs), calcium channel blockers (CCBs), beta‐blockers, and diuretics. All have proven to be effective in reducing the risk of myocardial infarction, stroke and heart failure in patients with hypertension, but they are also used for other indications. There has been an intensive discussion on the appropriate choice of antihypertensive agent for patients with COVID‐19. SARS‐Cov‐2 is known to use the angiotensin‐converting enzyme 2 (ACE2), a component of the Renin Angiotensin Aldosterone System (RAAS), and the transmembrane serine protease 2 (TMPRS2) as co‐receptors in order to gain the entry to the host cell. Activation of the RAAS with a reduced expression of ACE2 leads to an increase in the inflammatory cascade, leading to cell fibrosis. These mechanisms raised the debate on both harmful and beneficial effects of RAAS inhibitors (RAASi), that is, angiotensin‐converting enzyme inhibitors (ACEi) and angiotensin receptor blockers (ARB), in patients with COVID‐19. To date, there are several reviews summarizing the evidence, suggesting either no risk or a benefit of using RAAS inhibitors. , , , , One challenge, though, is the fact that most studies have included hospitalized people with a confirmed SARS‐CoV‐infection, thus potentially introducing bias and confounding by indication. These potential limitations were recently addressed in a large nationwide Swedish study, confirming that RAAS inhibitor use in primary prevention does not increase the risk of severe COVID‐19 outcomes.

Overall, the existing literature emphasizes the importance of continuing to use antihypertensive agents in patients with hypertension who develop COVID‐19. , Given the challenges maintaining treatment during the pandemic, it is important to study the prescribing and use of antihypertensives. The aim of this study was to assess initiations of different antihypertensive medicines to patients infected with COVID‐19 compared with matched controls from the general population.

METHODS

Study design

We conducted a cohort study to examine initiations of different antihypertensive drug classes after infection with COVID‐19. The exposed group was defined as the patients who had a positive COVID‐19 test, and the nonexposed group comprised patients without COVID‐19, that is, either people in the general population who had not been tested for COVID‐19 or those who had had been tested negative for COVID‐19. The unexposed group was matched to the exposed group on age, sex, and sampling time (date of positive test in the exposed) to minimize the confounding effect as much as possible. We matched the exposed and unexposed group with a 1:1 proportion.

Data sources

Data were collected from the database of the SCIFI‐PEARL (Swedish COVID‐19 Investigation for Future Insights—a Population Epidemiology Approach using Register Linkage) project. Data were used from the following registers, linked through the Swedish personal identification numbers :

SmiNet: the national database of notifiable diseases ‐ People who have positive SARS‐CoV‐2 polymerase chain reaction (PCR) test results (exposure)

Swedish Prescribed Drug Register —Dispensed prescriptions of antihypertensive drugs (outcome)

National Patient Register —Selected comorbidities recorded in inpatient care or during consultations in specialist ambulatory care (covariates, potential confounders) and information about hospitalization and ICU care (censoring)

Sociodemographic registers from Statistics Sweden (covariates, potential confounders)

Cause‐of‐death register (date of death for censoring)

Study population

The population used in this study comprised patients aged 20–80 years, with a positive COVID‐19 test. The study was restricted to this age group since most of the COVID‐19‐positive cases were among these ages. Furthermore, hypertension is uncommon in children and multimorbidity is common in the oldest, having a substantial impact on the prescribing of antihypertensives for other indications. , , Study participants had a “washout period” of 1 year before getting a positive test of COVID‐19 or corresponding date in comparison group. The washout period was defined for each antihypertensive drug class as the period of not being dispensed any prescription of a drug belonging to that class. The index date for matching of the unexposed group was set to the date that a corresponding patient in the exposed group had a positive COVID‐19 test. The follow‐up period was between the index date and the earliest of outcome, censoring, or end of study follow‐up (October 2020). Both exposed and unexposed groups were censored if receiving ICU care or hospitalized for more than 1 month, because we had no access to drug prescribing data in the inpatient setting. Furthermore, the unexposed group was censored if they tested positive for COVID‐19.

Study variables

The exposure in this cohort study was a positive SARS‐CoV‐2 polymerase chain reaction (PCR) test result recorded in the national database of notifiable diseases.

The outcome was the initiation of antihypertensive medicine following the exposure to COVID‐19 or corresponding index date in the unexposed. Initiation was defined as claiming a first prescription at a pharmacy. The following antihypertensive drug classes were recorded (with corresponding Anatomic Therapeutic Chemical classification system [ATC] codes): angiotensin‐converting enzyme inhibitors (ACE inhibitors; C09A), angiotensin receptor blockers (ARB; C09C), beta‐blockers (C07), calcium channel blockers (CCB; C08), and diuretics (C03) as well as fixed combinations of ACE inhibitors/thiazides (C09B) and fixed combinations of ARB/thiazides (C09D).

The following covariates were used both to describe the population and as potential confounders to adjust for in the analysis. All variables were measured at or up to the index date.

Age, sex, time (month) of COVID test (used in matching);

Prior hospitalization(s) from January 2015 to index date (yes/no);

Prior diagnosis of hypertension (ICD I10‐I15) at least recorded once from January 2015 to index date, recorded during hospitalization or consultation in specialist ambulatory care;

Selected comorbidities recorded at least once during hospitalization or consultation in specialist ambulatory care between January 2015 and index date; diabetes (E10,E11,E13), stroke/TIA (G45,I63), ischaemic heart disease (I20‐I25), atrial fibrillation (I48), heart failure (I50), asthma, and COPD (J40‐J45);

Previous dispensing of antihypertensive drugs defined as at least one prescription dispensed (same ATC‐classes as defined above, up to 365 days before index date);

Educational level, which was subdivided into four main subgroups: Primary school <9 years, primary school 9 years, secondary school, and postsecondary school.

Statistical analysis

Baseline characteristics are presented as frequencies and percentages for categorical variables, and as mean values and standard deviation (SD) for continuous variables. The standardized mean difference (SMD) was used to investigate the balance of covariates between the groups of the study. SMD is defined being independent to the unit of measurement; therefore, this allows it to be used in comparison between variables with different units and prevalences. We used 0.1 (10%) as the threshold for SMD indicating imbalance.

The crude cumulative proportion initiated on each drug class was assessed using Kaplan–Meier analysis. Incidence rates were expressed per 1000 person‐year for the exposed and unexposed group. Crude and adjusted hazard ratios for initiation of different antihypertensive drugs were estimated using Cox regression. Models were adjusted for diabetes mellitus, stroke/TIA, hypertension, ischaemic heart disease, atrial fibrillation, asthma/COPD, prior drug usage, and education.

A sensitivity analysis was conducted to further assess potential confounding. In this analysis, we excluded patients with a prior hospitalization 1 and 5 years before index date, respectively.

To illustrate the differences, Kaplan–Meier curves were used to show the change in the prescribing patterns over time up to 200 days after index date. All analyses were conducted using R version 1.3.1073.

RESULTS

We identified a total number of 112 278 patients infected with and tested positive for COVID‐19 between March 2020 and October 2020. After matching for age, sex, and time of COVID‐19 test, a total of 112 278 people without COVID‐19 were identified for the unexposed group.

Baseline characteristics of COVID‐19 test‐positive subjects and the comparison group are shown in Table 1. A total of 7.8% of all study subjects who tested positive for COVID‐19 and 5.7% in the comparison group, respectively, had previously been diagnosed with hypertension in specialist care. The most frequent other co‐morbidity was diabetes mellitus, with 4.5% in the COVID‐19 positive group and 2.8% in the comparison group. The two matched cohorts were relatively similar in terms of education, but prior use of other antihypertensive drug classes, some other comorbidities as well as the proportion being hospitalized prior to the COVID test tended to be more frequent in the test‐positive group, although all SMDs were below 0.1 (Table 1).

TABLE 1

Characteristics of 112 291 persons 20–80 years old who tested positive for COVID‐19 in Sweden between March and October 2020 and their comparison group matched by age, sex and calendar time

Baseline characteristics	Covid‐19 positive group	Unexposed group	SMD	Missing (%)
Total number	112 291	112 291
Age (mean (SD))	44.5 (15.6%)	44.5 (15.6%)	<0.001	0.0%
Sex = M (%)	51 284 (45.7%)	51 284 (45.7%)	<0.001	0.0%
Education level (%)			0.080	3.4%
Primary school <9 years	5017 (4.6%)	4488 (4.2%)
Primary school 9 year	8967 (8.2%)	10 764 (10.0%)
Secondary school	48 056 (44.1%)	49 352 (45.7%)
Postsecondary	46 912 (43.1%)	43 368 (40.2%)
History of medication use (%)
Diuretics	5437 (4.8%)	4365 (3.9%)	0.047	0.0%
Betablockers	10 171 (9.1%)	9039 (8.0%)	0.036	0.0%
CCB	8055 (7.2%)	7497 (6.7%)	0.020	0.0%
ACE‐inhibitors	5876 (5.2%)	5281 (4.7%)	0.024	0.0%
ACEi + thiazide (fixed comb)	835 (0.7%)	830 (0.7%)	0.001	0.0%
ARB‐inhibitors	6981 (6.2%)	6569 (5.8%)	0.015	0.0%
ARB + thiazide (fixed comb)	2249 (2.0%)	2076 (1.8%)	0.011	0.0%
Comorbidities (%)
Diabetes mellitus	5045 (4.5%)	3172 (2.8%)	0.089	0.0%
Stroke/TIA	955 (0.9%)	540 (0.5%)	0.045	0.0%
Hypertension	8782 (7.8%)	6454 (5.7%)	0.083	0.0%
Ischemic heart disease	2709 (2.4%)	1921 (1.7%)	0.049	0.0%
Atrial fibrillation	2497 (2.2%)	1745 (1.6%)	0.049	0.0%
Asthma/COPD	4036 (3.6%)	2620 (2.3%)	0.074	0.0%
Previous hospitalization (%)
Within 1 year	10 920 (9.7%)	8125 (7.2%)	0.089	0.0%
Within 5 years	32 059 (28.5%)	28 471 (25.4%)	0.072	0.0%

Abbreviations: ACE‐inhibitor, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; CCB, calcium channel blockers; COPD, chronic obstructive pulmonary disease; SMD, standardized mean difference.

Incidence rates of antihypertensive drugs to patients not previously dispensed any drug from that pharmacological group are shown in Table 2. Overall, incidence rates were higher in exposed group compared with unexposed group for all drugs except for ACEi in fixed combination with thiazide. Betablockers (30.4 patients per 1000 person‐years) were the most commonly used drug for initiation to exposed (COVID‐19 test‐positive) subjects, while ACEi in fixed combination with thiazides were the least common drug used for initiation in this group (1.5 patients per 1000‐person years). However, in total, RAAS inhibitors (alone or in fixed combinations) were initiated more frequently than betablockers to study subjects who had tested positive for COVID‐19.

TABLE 2

Incidence rates of initiating different antihypertensive drug classes (cases per 1000‐person‐years with 95% confidence intervals), in both COVID‐19‐infected patients naïve to each specific antihypertensive drug class in the prior year and their matched comparison group

ATC	Drugs	Exposed group			Unexposed group			IRR
		Initiated on treatment	Person years	Incidence rate (per 1000‐person years)	Initiated on treatment	Person years	Incidence rate (per 1000‐person years)
C03	Diuretics	472	27610	17.1 [15.6–18.7]	323	29658	10.9 [9.7–12.1]	1.6
C07	Betablockers	802	26361	30.4 [28.3–32.6]	480	28369	16.9 [15.4–18.5]	1.8
C08	CCB	590	26933	21.9 [20.2–23.7]	402	28705	14.0 [12.7–15.4]	1.6
C09A	ACE‐inhibitors	355	27463	12.9 [11.6–14.3]	201	29372	6.8 [5.9–7.9]	1.9
C09B	ACEi/thiazide	42	28576	1.5 [1.1–2.0]	33	30501	1.1 [0.7–1.5]	1.4
C09C	ARB	511	27136	18.8 [17.2–20.5]	348	28919	12.0 [10.8–13.4]	1.6
C09D	ARB/thiazide	116	28264	4.1 [3.4–4.9]	107	30156	3.5 [2.9–4.3]	1.2

Abbreviations: ACE inhibitor, angiotensin converting enzyme inhibitor; ARB, angiotensin receptor blocker; CCB, calcium channel blockers; IRR, incidence rate ratio.

The largest absolute difference in incidence rates between subjects with COVID‐19 positive test and their comparison group was observed for betablockers (13.5 patients/1000 person‐years) and the smallest absolute difference was observed from ARB in fixed combination with a thiazide (0.6 patients/1000 person‐years). The largest relative difference was observed for ACE inhibitors (incidence rate ratio 1.9).

Survival analysis

Our results showed an increased proportion of people initiated on all antihypertensive drug classes after COVID‐19 infection compared with a matched group without COVID‐19 (Kaplan–Meier curves for all antihypertensive drug classes are found in Appendix S1). However, the increase varied between the different pharmacological groups. After adjusting for diabetes mellitus, stroke/TIA, hypertension, ischaemic heart disease, atrial fibrillation, asthma/COPD, prior drug usage, and education, the associations were persistently significant for all studied hypertensive drugs except for ACEi and ARB in fixed combinations. Hazard Ratios (95%CI) of initiation of drug consumption were 1.83 [1.53–2.19] for ACEi, followed by beta‐blockers 1.74 [1.55–1.95], calcium channel blockers 1.61 [1.41–1.83], angiotensin receptor blockers 1.61 [1.40–1.86] and diuretics 1.53 [1.32–1.77] (Figure 1). The differences for the fixed ACEi/ARB and thiazide combinations were smaller and not statistically significant.

FIGURE 1

Unadjusted and adjusted hazard ratios of being initiated on different antihypertensive drugs for patients tested positive for COVID‐19. Comparisons up 200 days after the date of the test with a comparison group who had tested negative or not been tested, matched by age, sex and calendar time. Adjusted for difference in prior use of other antihypertensive drugs, selected comorbidities, and educational level

Despite some covariate imbalances in the two cohorts (although all SMDs were below 0.1), adjustment for the potential confounders did not affect the HR estimates appreciably. The sensitivity analysis excluding patients with a prior hospitalization had no major impact on the HR estimates (see Supplementary Table in Appendix S1).

DISCUSSION

Our study showed that, during the initial phase of the pandemic, antihypertensives were more commonly initiated to persons with a positive test for COVID‐19 compared with those who had only tested negative or were never tested. Moreover, it was slightly more likely that patients with COVID‐19 received a prescription with a RAAS‐active drug compared with controls. However, the differences between COVID‐19 positive patients and their controls were rather similar for the different antihypertensive drug classes.

The higher proportion of patients infected with COVID‐19 initiated on antihypertensive drugs is a positive sign, given that effective management of chronic diseases generally has a protective effect against severe outcomes of COVID‐19 infection. It is not surprising that the crude proportion is higher since patients with COVID‐19 had more comorbidities. However, even after adjustments for comorbidity, education and prior drug treatment, the risk of initiating treatment was higher for test‐positive COVID‐19 patients in our study. There are several potential reasons behind it. One potential explanation is that COVID‐19 leads to hypertension. A more likely reason is that patients infected with COVID‐19 consult health care and have their blood pressure measured. According to the so‐called “rule of halves,” half of all people with hypertension are not diagnosed. Therefore, there might have been people with high blood pressure in the COVID‐19 positive group, identified after being infected. It could also be that too few people in the control group were initiated on antihypertensives, if they were reluctant to seek care during the first phase of the pandemic. In December 2020, the EU‐funded European Network to Advance Best practices & technoLogy on medication adherencE (ENABLE) COST Action conducted a survey in 39 European countries to assess barriers and facilitators for patients accessing their chronic medication during the pandemic. , The survey indicated significant disruption of chronic disease services, especially in countries with a greater number of COVID‐19 cases per 100 000 inhabitants, and a large variation between countries in measures taken to ensure adequate drug management. This is of great concern, and even more critical since the pandemic may have a negative impact on blood pressure levels in the population.

It is also important to acknowledge that there may be other differences between those being infected and the comparison group. Our findings showed a slight change between unadjusted and adjusted results in the HRs when we included the potential confounders in the model. The small difference between unadjusted and adjusted results and the fact that the sensitivity analysis had no major impact on the results could indicate that the effect of confounding factors is relatively small. However, we had no data for other potential confounders, such as smoking, alcohol consumption and BMI. Furthermore, we only used data on diagnoses from the national patient register including diagnoses registered in hospitals and in specialist outpatient care, while most patients with hypertension are diagnosed and treated in primary care.

Our results showed differences compared with previously reported studies of antihypertensive drugs in Sweden, as betablockers and ACE‐inhibitors were initiated more in our study compared with previous findings. It is important to recognize that there have been several changes on the drug market since that time with, for example, new guidelines and formularies as well as changes in price after patent expiries and introduction of generics. The current prescribing patterns correspond rather well to the recent European guidelines for the treatment of hypertension. Furthermore, we assessed initiation of antihypertensives for any indication, because explicit information of indication for prescription was not available. The high rate of initiations of betablockers might partly have been for the treatment of heart failure or as secondary prevention after myocardial infarction, as recommended by current guidelines. ,

The fact that ACE inhibitors were the drugs with the highest likelihood of being initiated to a patient infected with COVID‐19 is not surprising since according to some hypotheses, it may have some benefit compared with the other antihypertensive agents. , In a meta‐analysis, Ren et al. concluded that the mortality and severity of COVID‐19 were significantly lower in patients taking ACE inhibitors/ARBs than in controls. No association was found between using other antihypertensive drugs including CCBs, beta‐blockers, and diuretics and the incidence and severity of COVID‐19. In contrast to other reviews, Nozari and Hamidizadeh investigated the effects of antihypertensive drugs on COVID‐19 only in patients with essential hypertension. They suggest that ACE inhibitors and ARBs may be better choices to treat hypertension in this population. Conversely, diuretics can be considered the least effective drug in the setting of concomitant hypertension and COVID‐19. It should be noted, though, that these reviews are quite recent, and most of the discussion in media focused on the RAAS inhibitors at the time of this study.

In this study, we restricted our analysis to initiations of antihypertensive therapy. It is well known from previous studies that many patients do not take their medicines as prescribed and a review found that persistence rates for antihypertensives can be as low as 35% 1 year after initiation. The largest decline in persistence occurs early after initiation and in a previous study, we found that a large proportion of Swedish patients with hypertension in primary care only claimed one prescription. It will be important to study if the COVID‐19 pandemic has further negatively impacted patient adherence and persistence.

A major strength in our study was the large study population and nationwide coverage of our datasets. Our study was based on nationwide registers with complete coverage and high validity. The Swedish prescribed drug register is unique with 99.7% coverage, including individual patient data for all dispensed prescription drugs in the country. The patient register has also showed high validity, enabling us to adjust for a range of potential confounders. There were, however, some limitations. Diagnoses recorded in primary care are not available in the national patient register. Some variables such as BMI, alcohol consumption and smoking that may have been possible confounders were not included in the analysis either, due to the lack of information in the used registers. Another limitation was that COVID‐19 test was not available to everyone in the beginning of the pandemic, and people with milder symptoms were never tested, likely resulting in some misclassification between the test‐positive and comparison group, which may have affected the results of our study. It is also possible that patients with risk factors and chronic diseases had less access to health care in Sweden during the pandemic. This may have resulted in a general under‐reporting of comorbidities in our data. Finally, it is important to acknowledge that we defined co‐morbidities and drug treatment based on only one registration. This might have introduced some information bias, but it would most likely be nondifferential between groups.

In conclusion, we found that all antihypertensive medicines were more frequently initiated in people with a positive COVID‐19 test compared with a matched study group from the general population. This difference may either be associated with hypertension caused by the COVID‐19 infection, more people with COVID‐19 diagnosed with hypertension since they consult health care, or residual confounding factors not adjusted for in this study.

CONFLICT OF INTEREST

The authors have declared that they have no conflict of interest.

Appendix S1. Kaplan‐Meier graphs for initiation of new antihypertensive medication for different antihypertensive drug classes, in COVID‐19‐infected patients naïve to each hypertensive drug class in the prior year and their matched comparison group.

Click here for additional data file.