Association of α1-Blocker Receipt With 30-Day Mortality and Risk of Intensive Care Unit Admission Among Adults Hospitalized With Influenza or Pneumonia in Denmark

Key Points Question Is the receipt of α1–adrenergic receptor blocking agents (α1-blockers) associated with protective benefits against adverse outcomes, such as mortality and intensive care unit admission, among adult patients with severe respiratory tract infections? Findings In this cohort study of 528 467 Danish adults hospitalized with influenza or pneumonia, current receipt of α1-blockers was associated with a 14% reduction in the relative risk of 30-day mortality. Meaning This study’s findings suggest that the receipt of α1-blockers may have a clinically relevant association with protective benefits against adverse outcomes among patients with severe respiratory tract infections.


Introduction
Acute respiratory syndrome coronaviruses are associated with severe viral pneumonia and death. 1 In the ongoing coronavirus disease 2019 (COVID- 19) pandemic, mortality appears to be associated with acute respiratory distress syndrome and a dysregulated immune response with hyperinflammation and cytokine storm syndrome, 2,3 factors that have also been observed in patients with other severe respiratory tract infections and sepsis. 4,5[8] Given the safety profile and low cost of treatment with α1-blockers, any benefits associated with protection against adverse outcomes among patients hospitalized with COVID-19 or other severe respiratory tract infections would have substantial clinical and public health importance. 9Studies of the association of α1-blockers with outcomes among human study participants with respiratory tract infections are scarce or nonexistent. 8To address this gap, we conducted a large population-based study using data from Danish national registries to investigate the association of the receipt of α1-blockers with intensive care unit (ICU) admission and 30-day mortality among patients hospitalized with influenza or pneumonia.

Study Design and Setting
The population for this nationwide cohort study included all patients 40 years and older who were hospitalized with influenza or pneumonia in Denmark between January 1, 2005, and November 30, 2018, with follow-up through December 31, 2018 (eFigure 1 in the Supplement).Data collection and processing were reported to the Danish Data Protection Agency through Aarhus University.Ethics review board approval and informed consent are not required for registry-based observational studies in Denmark.This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline for cohort studies. 10nmark has a tax-supported health care system that provides health care services, including acute care and hospital care for influenza and pneumonia, to all residents. 11All Danish residents receive a personal identity number at birth or immigration that allows individual-level linkage across the extensive Danish registry system, which includes national-level information on residence; prescriptions; vital status (dead or alive); and primary, specialty, and hospital-based care. 12 assessed hospitalizations (including direct inpatient hospital admissions and emergency department visits leading to either inpatient hospital admission or discharge to home) among patients 40 years and older (because receipt of α1-blockers is rare among individuals younger than 40 years) who had either a primary or secondary diagnosis of influenza or pneumonia recorded in the Danish National Patient Registry (Figure 1). 13This registry includes data on primary and secondary Hospitalizations that were preceded by an influenza or pneumonia diagnosis within the previous 3 months were excluded to avoid the inclusion of readmissions.We predefined subgroups based on an influenza diagnosis or a diagnosis of pneumonia that specified a bacterial or nonspecific pathogen.

JAMA Network Open | Infectious Diseases
Specific diagnostic codes used in the study are listed in eTable 1 in the Supplement.

Outcomes and Exposures
Primary study outcomes were 30-day mortality and 30-day ICU admission during the index hospitalization associated with an influenza or pneumonia diagnosis.Secondary outcomes included the receipt of organ-supportive treatment (mechanical ventilation, noninvasive ventilation, and treatment with inotropic and/or vasopressor medications) during ICU admission.Dialysis-treated acute kidney injury was defined as treatment with acute renal replacement therapy among patients with no history of previous dialysis for the treatment of chronic kidney disease.[14] Data on all filled prescriptions were obtained from the Danish National Prescription Registry. 15is registry contains data on all prescription drugs obtained by Danish residents at any community pharmacy in Denmark since 1995.The main exposure of interest was current receipt of α1-blockers, which are primarily used for 2 indications: BPH and hypertension (eTable 1 in the Supplement).We defined current receipt of α1-blockers as a prescription filled within 90 days before a hospitalization for influenza or pneumonia.This definition was consistent with the 3-month supply of α1-blockers that is dispensed most often in Danish pharmacies.
In our main analyses, patients currently receiving α1-blockers were compared with those not receiving α1-blockers (defined as patients with no prescription for an α1-blocker filled within 365 days before the index date).In a secondary analysis, patients who formerly received α1-blockers (defined as patients with a prescription for an α1-blocker that was filled 91-365 days before the index date) were compared with those who were not receiving α1-blockers to address potential confounding by treatment indication.In an additional analysis to account for confounding by treatment indication, patients currently receiving α1-blockers were compared with those currently receiving a different type of medication, 5α-reductase inhibitors (finasteride and dutasteride), for the treatment of BPH.
This analysis was restricted to male patients who did not receive the 2 types of drugs (α1-blockers and 5α-reductase inhibitors) as combination therapy for BPH.

Potential Confounders
We considered a range of potential confounders in our study. 16We obtained patient age and sex from the population registry, 12 and we collected information from the patient registry on the presence of a range of comorbidities that required inpatient or outpatient hospital contact within 10 years before the index hospitalization (Table 1; eTable 1 in the Supplement). 13escriptions for relevant concurrent medications that were filled within 90 days before hospital admission were also ascertained; these medications included angiotensin-converting enzyme inhibitors, angiotensin II receptor blockers, calcium channel blockers, thiazides, β-blockers, other antihypertensive medications, statins, aspirin, loop diuretics, antibiotics, antiviral agents, glucocorticoids, other immunosuppressants, nonsteroidal anti-inflammatory drugs, opioids, vitamin K antagonists, proton pump inhibitors, antidepressants, hypnotics or sedatives, and antipsychotics. 15cause lifestyle and social factors are associated with health, we included information on obesity, alcohol use, smoking, marital status, and urban vs rural residence.

Statistical Analysis
We applied propensity score balancing of potential confounders across treatment groups. 17ntinuous covariates were included as a cubic spline with 7 knots.We used propensity score weighting to generate a population of relevant comparison groups (eg, patients not receiving α1-blockers or 5α-reductase inhibitors) that resembled the number and covariate distribution of patients receiving α1-blockers.The exposed patients were assigned a weight of 1, and the unexposed patients were assigned a weight equivalent to their estimated propensity score divided by the difference between 1 and their estimated propensity score (ie, the weight was the individual's estimated odds of being exposed).If successful, this weighting method produces a comparison population with size and covariate distribution resembling that of the exposed population. 17variate balance was assessed using standardized differences and was deemed acceptable.
Follow-up started on the date of the first hospital admission associated with an influenza or pneumonia diagnosis and continued until a specific outcome of interest, emigration, or the completion of 30 days, whichever occurred first.The 30-day risks (both unadjusted and weighted by propensity score) of death and ICU admission were computed and plotted.Risk differences were calculated for all outcomes by subtracting propensity score-weighted risks.Risk ratios (RRs) were estimated as the ratios of propensity score-weighted risk estimates.All estimates were accompanied by 95% CIs that were obtained using bootstrapping with 200 bootstrap samples.
Several subgroup analyses were conducted that were stratified by (1) patient age, (2) restriction to patients with a diagnosis of influenza or pneumonia that was listed first in the hospital discharge summary, (3) restriction to patients with BPH or hypertension as underlying conditions, and (4) the 3 most frequently prescribed α1-blocker medications (doxazosin, alfuzosin, and tamsulosin).We also performed an analysis of 30-day risk of death associated with the receipt of α1-blockers that was restricted to patients transferred to the ICU during their hospitalization, with follow-up beginning on the date of ICU admission.In additional analyses, current receipt and nonreceipt of α1-blockers were compared among male patients only.
Statistical analyses were performed using SAS software, version 9.4 (SAS Institute).Data were analyzed from April 21 to December 21, 2020.

Results
The     After propensity score weighting of patients receiving α1-blockers (eFigure 2 in the Supplement), treatment groups were well balanced on all measured covariates, with absolute standardized differences for all covariates decreasing from between 0 and 1.78 before propensity score balancing to less than 0.10 after propensity score balancing (Table 1).The final cohorts included in the propensity score-weighted outcome analysis consisted of 21 772 patients currently receiving α1-blockers and 22 117 patients not receiving α1-blockers (weighted to the propensity score distribution of the patients currently receiving α1-blockers).

Patient Outcomes
In the unadjusted analyses before propensity score weighting, patients receiving α-1 blockers had higher 30-day mortality (15.9%) than those not receiving α-1 blockers (14.5%) (eTable 3 in the Supplement), which was likely associated with the older age and greater comorbidity burden of this cohort compared with the cohort not receiving α-1 blockers.After covariate balancing by propensity score weighting, among all patients with influenza or pneumonia, 30-day mortality was 15.9% for patients receiving α1-blockers and 18.5% for patients not receiving α1-blockers, with a corresponding risk difference of −2.7% (95% CI, −3.2% to −2.2%) and an RR of 0.85 (95% CI, 0.83-0.88)(Table 2).
b To ensure anonymity, Danish legislation prohibits the reporting of exact values for instances in which low numbers of participants (eg, n <5) are observed or can be inferred using information from other categories.2).

JAMA Network Open | Infectious Diseases
In the outcome analysis of the subgroup of 7636 patients diagnosed with influenza (which included 327 patients receiving α1-blockers and 336 propensity score-weighted patients not receiving α1-blockers), 30-day mortality was 7.3% among those receiving α1-blockers and 7.6% among those not receiving α1-blockers (risk difference, −0.2% [95% CI, −3.7% to 3.2%]; RR, 0.97 [95% CI, 0.61-1.55])(Table 2).The risk of ICU admission was slightly higher among patients with influenza who were receiving vs not receiving α1-blockers (RR, 1.34; 95% CI, 0.88-2.02),whereas the risk of dialysis-treated acute kidney injury was lower (RR, 0.50; 95% CI, 0.19-1.36);however, these estimates were imprecise owing to the limited number of outcomes available for analysis.Outcomes for those receiving vs not receiving α1-blockers were only similar when restricting the analysis to male patients (eTable 4 in the Supplement).
The subgroup analyses (stratified by age, patients with a first-listed diagnosis of influenza or pneumonia, and patients with BPH or hypertension) were generally consistent with our main findings (Figure 3).An analysis of α1-blockers by type indicated that tamsulosin, alfuzosin, and doxazosin were all associated with reductions in the risk of mortality (for tamsulosin, RR, 0.84 [95%

Discussion
In this large nationwide population-based cohort study of 528 467 Danish patients 40 years or older who were hospitalized with influenza or pneumonia, receipt of α1-blockers was associated with a decreased risk of death, compared with nonreceipt of α1-blockers and receipt of 5α-reductase inhibitors.
This study provides novel information about the association of α1-blockers with protective benefits against adverse outcomes among patients with severe respiratory tract infections.These results are consistent with and extend the findings of a preliminary epidemiological analysis of this association conducted in the US. 7That study found that among men aged 45 to 85 years (108 956 men from the MarketScan database and 252 708 men from the Optum database) who were hospitalized with pneumonia, the propensity score-matched odds ratio for in-hospital ventilation or death among those receiving vs not receiving α1-blockers was 0.91 (95% CI, 0.87-0.96). 7This finding is consistent with the RRs in the current study of 0.85 (95% CI, 0.83-0.88)for 30-day mortality and 0.92 (95% CI, 0.86-0.99)for mechanical ventilation associated with receipt vs nonreceipt of α1-blockers among patients with influenza or pneumonia.
Of interest, the US analysis also included 13 125 men from the Market Scan database and 6534 men from the Optum database who had a diagnosis code for acute respiratory distress, which is a potential precursor of acute respiratory distress syndrome.In this group, the association between α1-blocker receipt and protective benefits was clearer, with an adjusted odds ratio for in-hospital ventilation or death of 0.67 (95% CI, 0.46-0.96). 7 studies of mouse models, it has recently been reported that macrophages secrete and respond to catecholamines through adrenergic receptors when exposed to inflammatory stimuli such as bacteria.Catecholamines orchestrate cytokine production and severity of inflammation injury, 6 and catecholamine synthesis inhibition reduces cytokine responses.When mice injected with bacterial lipopolysaccharide were pretreated with pharmacologic catecholamine blockade through metyrosine therapy, they were protected from the fatal complications of cytokine release syndrome. 6erging data from human studies suggest that a subset of patients with COVID-19 develops cytokine storm syndrome that is associated with increased production of proinflammatory cytokines (including interleukin 6, interleukin 2R, interleukin 8, tumor necrosis factor α, and granulocyte colony-stimulating factor)

Limitations
This study has several limitations.One limitation was the study's reliance on diagnostic coding of influenza and pneumonia.Although some patients with these infections may not receive a diagnosis, we believe that restriction to physician-coded influenza and pneumonia discharge diagnoses ensured inclusion of only those patients with clinically relevant infections.The positive predictive value of pneumonia diagnoses in the Danish patient registry is 90%. 21,22Overall 30-day mortality after pneumonia diagnosis in this study's population-based cohort was similar to that reported in other parts of the world. 23Any selection bias in this study should be minimal because follow-up was almost complete, and different associations between included and nonincluded patients would not be expected.Given the chronic receipt of the drugs included in the study, any misclassification from sporadic use that was not captured before hospitalization should be minor and not associated with the outcome of interest.
This study lacked data on continuous in-hospital use of α1-blockers, and possible drug discontinuation during acute illness may have produced an underestimation of any association.Deaths are accurately recorded in the Danish population registry and updated daily. 12Intensive care unit admissions and treatments are also accurately recorded, as the Danish patient registry is used for financial reimbursement to hospitals and for mandatory reporting to quality-of-care databases. 14Because conditions treated with α1-blockers(eg, BPH and hypertension) and severe influenza and pneumonia infections may both lead to acute hospitalization, conditioning the analyses on hospitalized patients may, in theory, introduce collider bias.However, in cases of influenza and pneumonia in which death occurs shortly after diagnosis, the infection is likely to be a main factor associated with hospitalization and death rather than an inconsequential variable.Moreover, this study found robust estimates among patients with a primary diagnosis of influenza or pneumonia and among patients with severe infections who were admitted to the ICU.Potential confounding by drug treatment indication was handled by using an active comparator in 1 analysis, using propensity score weighting that included a large number of potential confounders, and restricting analyses to subgroups according to treatment indication.Nevertheless, it is possible that unmeasured confounding factors impacted the study's risk estimates.
Healthy-user bias is an unlikely explanation for the findings, given that the comorbidities and lifestyle factors captured did not indicate healthier lifestyles among patients receiving α1-blockers.Moreover, patients who formerly received α1-blockers did not experience improved outcomes.Although the study included more than 500 000 patients, the precision of risk estimates was limited in some subgroups.
Because more than 95% of patients receiving α1-blockers were men and more than 90% of people in Denmark are White, it remains uncertain whether these results will also apply to women and non-White individuals.

Conclusions
In this study, patients receiving α1-blockers who were hospitalized with influenza or pneumonia had lower mortality after confounding factors were controlled for compared with those not receiving α1-blockers and those receiving 5α-reductase inhibitors.Thus, these data support the hypothesis that α1-blockers may have a clinically relevant association with outcomes among patients with acute respiratory tract infections.These findings will need to be reproduced among patients with confirmed COVID-19 infection.Randomized clinical trials may enable more definitive conclusions to be reached regarding the association between α1-blockers and ICU admission and mortality among patients with COVID-19 and other respiratory tract infections.Such clinical trials may evaluate any benefits associated with initiating treatment with α1-blockers early in the course of severe respiratory infection rather than the more chronic treatment examined in this study, and clinical trials may also consider possible adverse effects associated with α-1 blocker therapy.

Figure 1 .
Figure 1.Patient Flow Diagram 647 738 Patients with hospital-diagnosed influenza or pneumonia infection b A population of patients not receiving α1-blockers was weighted to the propensity score distribution of patients currently receiving α1-blockers.cIncludes antihypertensive medications with the exception of α1-blockers.

Figure 2 .
Figure 2. Cumulative Risk of 30-Day Mortality and Intensive Care Unit (ICU) Admission

Table 1 .
Characteristics of Patients Currently Receiving α1-Blockers vs Patients Not Receiving α1-Blockers a (continued) a A total of 9119 patients (1.7%) who formerly received α1-blockers were not included in the table.