Khalid Al Sulaiman, MBA, BCCCP, BCNSP, FCCM https://orcid.org/0000-0002-5547-2043, Ohoud Aljuhani, PharmD, […], and Ghassan Alghamdi, MD
Abstract
Doxycycline has revealed potential effects in animal studies to prevent thrombosis and reduce mortality. However, less is known about its antithrombotic role in patients with COVID-19. Our study aimed to evaluate doxycycline’s impact on clinical outcomes in critically ill patients with COVID-19. A multicenter retrospective cohort study was conducted between March 1, 2020, and July 31, 2021. Patients who received doxycycline in intensive care units (ICUs) were compared to patients who did not (control). The primary outcome was the composite thrombotic events. The secondary outcomes were 30-day and in-hospital mortality, length of stay, ventilator-free days, and complications during ICU stay. Propensity score (PS) matching was used based on the selected criteria. Logistic, negative binomial, and Cox proportional hazards regression analyses were used as appropriate. After PS (1:3) matching, 664 patients (doxycycline n = 166, control n = 498) were included. The number of thromboembolic events was lower in the doxycycline group (OR: 0.54; 95% CI: 0.26-1.08; P = .08); however, it failed to reach to a statistical significance. Moreover, D-dimer levels and 30-day mortality were lower in the doxycycline group (beta coefficient [95% CI]: −0.22 [−0.46, 0.03; P = .08]; HR: 0.73; 95% CI: 0.52-1.00; P = .05, respectively). In addition, patients who received doxycycline had significantly lower odds of bacterial/fungal pneumonia (OR: 0.65; 95% CI: 0.44-0.94; P = .02). The use of doxycycline as adjunctive therapy in critically ill patients with COVID-19 might may be a desirable therapeutic option for thrombosis reduction and survival benefits.
Introduction
Coronavirus disease 2019 (COVID-19) affects multiple organ systems and has a wide spectrum of clinical manifestations. Besides pulmonary complications, COVID-19 is associated with a prothrombotic state, which may increase thromboembolic risk,1 specifically, large-vessel thrombosis, venous thromboembolic events (VTE), and stroke.2–4 The prevalence of COVID-19-associated VTE is estimated to be between 25% to 30% in severely ill patients and those on mechanical ventilation (MV).3 The thrombolytic process may involve endothelial inflammation and injury, leading to in situ microvascular thrombosis.2 The thrombotic events associated with COVID-19 compromise patient’s clinical outcomes and may increase mortality.1,5
The global COVID-19 case count has increased to over 500 million people, and more than 6 million deaths.6 The need for effective therapies against COVID-19 has been a top priority in controlling the COVID-19 spread. Several approved drugs, including immune suppressors and modulators, convalescent plasma, and neutralizing antibodies, have been repurposed and tested to treat or control COVID-19.7 Patients with COVID-19 may have secondary bacterial pneumonia that warrants antibiotic therapy.8 A meta-analysis reported the prevalence of antibiotic prescribing in patients with COVID-19 to be up to 75%.9 The commonly prescribed antibiotics were fluoroquinolones (20%) and macrolides (18.9%).9 Due to safety concerns related to aortic aneurysm, QTc prolongation, and neurological toxicities associated with fluoroquinolone and macrolide, doxycycline may be used as an alternative.9
Doxycycline has modest antiviral activity,10 anti-inflammatory activity,11 antioxidant,12 and tissue-protective effects,13 which could mitigate the severity and improve the clinical outcomes of patients with COVID-19. The potential impacts of doxycycline, either alone or in combination with other medications such ivermectin, lopinavir/ritonavir, hydroxychloroquine, colchicine, dapsone, azithromycin, famotidine, and vitamin C, have been the subject of various studies.14–23 Moreover, doxycycline has shown to prevent acute pulmonary embolism (PE)-induced lung injury and reduces mortality in animal studies.24 A study by Alam et al. included patients with moderate to severe COVID-19 admitted to long-term care facilities in the United States revealed that doxycycline was associated with early clinical recovery, decreased hospitalization, and decreased mortality.15 Therefore, the objective of this study was to assess the potential doxycycline impacts on clinical outcomes, such as thrombosis and mortality in critically ill patients with COVID-19.
Methods
Study Design
This study is part of the Saudi Critical Care Pharmacy Research (SCAPE) platform, which conducted several studies that evaluate the safety and effectiveness of multiple therapies in critically ill patients.25 Our current study design and methods were based on our previous studies. Details of previous studies’ designs and methodology have been published previously 26–30. This is a multicenter retrospective cohort study of adult patients with COVID-19 admitted to the intensive care units (ICUs) at five centers in Saudi Arabia between March 1, 2020, and July 31, 2021. The diagnosis of COVID-19 was confirmed using Reverse Transcriptase-Polymerase Chain Reaction (RT-PCR) nasopharyngeal or throat swabs. Eligible patients were categorized into two groups based on doxycycline therapy during their ICU stay (doxycycline vs control). Controls were patients with COVID-19 admitted to ICU and did not receive doxycycline.
Study Setting
Five Saudi Arabian medical facilities participated in the study. The centers were chosen based on their abilities to use the standardized national (MOH) COVID-19 care protocol, advanced ICU settings, the ability to handle critically ill patients with COVID-19, geographic variation, accessibility of electronic records, or willingness to contribute. The primary site for this multicenter study was King Abdulaziz Medical City (KAMC-RD), a tertiary care center in Riyadh.
Study Participants
All adult patients (aged ≥18 years) admitted to the ICUs of study centers with confirmed COVID-19 were assessed for eligibility. Patients were excluded if they received doxycycline prior to ICU admission or after 48 h of ICU admission, had a history of VTE, stroke, left ventricular clot, antiphospholipid syndrome, or systemic lupus erythematosus. In addition, patients with a history of cancer, who died within the first 24 h of ICU admission, ICU length of stay (LOS) ≤one day, or were designated as “Do-Not-Resuscitate” or “No code” at ICU admission were excluded from the study (Figure 1).
Data Collection
Demographic information, comorbidities, laboratory results, vital signs, and baseline severity scores were among the variables and data collected utilizing the Research Electronic Data Capture (REDCap®) platform hosted by King Abdullah International Medical Research Center; more information about the data collection may be found in the Supplementary material.
Outcomes
The primary outcome was composite thrombotic events (arterial and venous). The secondary outcomes were 30-day and in-hospital mortality, ICU/hospital LOS, ventilator-free days (VFD), and ICU-acquired complications (i.e., new-onset atrial fibrillation [Afib], AKI, liver injury, hospital/ventilator-acquired pneumonia (bacterial/fungal), and secondary fungal infection. All patients were followed until they were discharged or died during their in-hospital stay. The details of the definition of the outcomes are available in the Supplementary material.
Statistical Analysis
We reported continuous variables as means with standard deviations (SD), or medians with Quartile 1 (Q1) and Quartile 3 (Q3) based on data distribution using the Shapiro–Wilk test and graphical representation (ie, histograms and Q–Q plots), while, categorical variables were reported as counts and percentages. Baseline categorical variables were compared between the two groups using either Chi-square or Fisher’s exact test as appropriate, and student t-test or Mann–Whitney U test for normally and non-normally distributed continuous variables, respectively.
Propensity score (PS) matching procedure (Proc PS match) (SAS Version 9.4, SAS Institute Inc. Cary, NC, USA) was used to match patients who received doxycycline therapy (active group) to patients who did not (control group) (1:3 ratio) based on patient’s age, APACHE II scores, baseline platelets count, aPTT, early use of dexamethasone within 24 h of ICU admission, and chronic kidney disease as comorbidity. A greedy nearest neighbor matching method was used in which one patient who received doxycycline (active) matched with three patients who did not (control), which eventually produced the smallest within-pair difference among all available pairs with treated patients. Patients were matched only if the difference in the logits of the PSs for pairs of patients from the two groups was less than or equal to 0.1 times the pooled estimate of the SD (Supplementary material – Detailed statistical analysis).
Results
A total of 1592 critically ill patients admitted to the ICUs with confirmed COVID-19 were screened; among those, 1302 patients met the inclusion criteria (Figure 1). Of those, 203 patients (15.6%) received doxycycline, and 1099 were controls during ICU admission. After PS matching (1:3 ratio) based on the selected criteria, 664 patients were included, with 166 patients receiving doxycycline and 498 patients being in the control group. The daily doxycycline dose used in our study was 200 mg/day for a median duration of 5.0 (2.00, 6.00) days.
Demographic and Clinical Characteristics
Before PS matching, the study population was predominantly male (64.1%) with a mean age of 61.5 (±14.66) years. The most common comorbidities were diabetes mellitus (59.4%), hypertension (55.8%), and dyslipidemia (20.7%). There was a notable difference in baseline characteristics between the groups. Before PS matching, patients who received doxycycline had higher baseline platelet count, C-reactive protein (CRP), ferritin levels, and received more nephrotoxic medications during ICU stay compared to the control group. Conversely, the control group had a higher AKI at admission, APACHE II and SOFA scores, blood urea nitrogen, INR, and aPTT. After PS matching, most demographic and baseline characteristics were similar in both groups except for the total white blood cells, ferritin, CRP, and CPK levels that were higher in the doxycycline group, and higher aPTT in the control group (Table 1).Table 1. Summary of Demography and Baseline Characteristics.
| Baseline Characteristics | Before propensity score | After propensity score | ||||||
|---|---|---|---|---|---|---|---|---|
| Overall (N = 1302) | Control (N = 1099) | Doxycycline (N = 203) | P-value | Overall (N = 664) | Control (N = 498) | Doxycycline (N = 166) | P-value | |
| Age (years), Mean (SD) | 61.5 (14.66) | 61.6 (14.83) | 60.6 (13.69) | .2513b | 59.8 (14.84) | 59.7 (15.18) | 60.1 (13.80) | .7287a |
| Gender—Male, n (%) | 825 (64.1) | 690 (63.3) | 135 (68.2) | .1881c | 441 (66.8) | 324 (65.5) | 117 (70.9) | .1975c |
| Body mass index, Mean (SD) | 31.1 (8.74) | 31.1 (9.02) | 31.0 (7.06) | .5296b | 31.0 (7.86) | 31.0 (8.25) | 31.0 (6.61) | .5401b |
| APACHE II score, Median (Q1,Q3) | 13.0 (9.00, 21.00) | 14.0 (9.00, 22.00) | 11.0 (7.00, 17.00) | <.0001a | 12.0 (7.00, 18.00) | 12.0 (8.00, 18.00) | 11.0 (7.00, 18.00) | .4711b |
| SOFA score, Median (Q1,Q3) | 4.0 (3.00, 7.00) | 5.0 (3.00, 7.00) | 4.0 (2.00, 7.00) | .0463b | 4.0 (2.00, 6.00) | 4.0 (2.00, 6.00) | 4.0 (2.00, 6.00) | .7486b |
| Multiple organ dysfunction score, Median (Q1,Q3) | 5.0 (4.00, 7.00) | 5.0 (4.00, 7.00) | 5.0 (4.00, 7.00) | .1769b | 5.0 (4.00, 7.00) | 5.0 (4.00, 7.00) | 5.0 (4.00, 7.00) | .5908b |
| Early use of dexamethasone within 24 h, n (%) | 833 (64.0) | 709 (64.5) | 124 (61.1) | .3497c | 383 (57.7) | 288 (57.8) | 95 (57.2) | .8918b |
| Early use of methylprednisolone within 24 h, n (%) | 134 (10.3) | 117 (10.6) | 17 (8.4) | .3278c | 74 (11.1) | 59 (11.8) | 15 (9.0) | .3189c |
| Early use of tocilizumab within 24 h, n (%) | 304 (23.3) | 264 (24.0) | 40 (19.7) | .1816c | 127 (19.1) | 102 (20.5) | 25 (15.1) | .1240c |
| Proning at admission, n (%) | 327 (25.9) | 277 (26.0) | 50 (25.4) | .8588c | 187 (28.9) | 143 (29.4) | 44 (27.3) | .6214c |
| Serum creatinine (mmol/L), Median (Q1,Q3) | 88.0 (68.00, 130.00) | 88.2 (67.00, 133.50) | 84.0 (69.00, 116.00) | .2802b | 84.0 (67.00, 117.56) | 84.0 (66.00, 117.11) | 84.0 (69.00, 118.00) | .5346b |
| Blood urea nitrogen (mmol/L), Median (Q1,Q3) | 6.9 (4.80, 11.30) | 6.9 (4.80, 11.60) | 6.6 (4.40, 9.80) | .0464b | 6.6 (4.65, 10.10) | 6.7 (4.70, 10.20) | 6.4 (4.40, 9.80) | .4686b |
| Acute kidney injury within 24 h of ICU admission, n (%) | 367 (28.5) | 322 (29.7) | 45 (22.5) | .0396c | 170 (25.8) | 135 (27.2) | 35 (21.5) | .1458c |
| Mechanical ventilation within 24 h of ICU admission, n (%) | 900 (69.6) | 757 (69.4) | 143 (70.4) | .7637c | 455 (68.5) | 342 (68.7) | 113 (68.1) | .8849c |
| Lowest PaO2/FiO2 ratio at admission, Median (Q1,Q3) | 81.5 (61.00, 131.50) | 82.5 (61.00, 132.92) | 78.2 (59.24, 113.45) | .2689b | 83.0 (62.40, 132.73) | 84.7 (63.22, 137.00) | 79.9 (57.15, 117.00) | .1217b |
| Inotropes/vasopressors use within 24 h of admission, n (%) | 288 (22.4) | 247 (22.7) | 41 (20.3) | .4434c | 137 (20.6) | 103 (20.7) | 34 (20.5) | .9558c |
| Vasoactive inotropic score, Mean (SD) | 7.3 (44.55) | 7.5 (45.05) | 6.0 (41.97) | .8587b | 5.5 (35.75) | 5.7 (36.39) | 4.7 (33.94) | .3061b |
| Lactic acid (mmol/L), Median (Q1,Q3) | 1.7 (1.29, 2.44) | 1.7 (1.24, 2.39) | 1.7 (1.37, 2.55) | .2595b | 1.7 (1.28, 2.30) | 1.7 (1.21, 2.23) | 1.7 (1.35, 2.55) | .1396b |
| Platelets count (109), Median (Q1,Q3) | 243.0 (190.00, 316.00) | 242.0 (187.50, 314.00) | 252.0 (199.00, 329.00) | .0597b | 260.0 (204.00, 331.00) | 260.5 (205.00, 331.00) | 255.5 (201.00, 331.00) | .9627b |
| Total WBCs (109), Median (Q1,Q3) | 9.3 (6.57, 12.71) | 9.2 (6.54, 12.60) | 9.9 (6.70, 13.30) | .1248b | 9.4 (6.63, 12.60) | 9.0 (6.61, 12.25) | 10.0 (6.74, 14.40) | .0433b |
| International normalized ratio, Median (Q1,Q3) | 1.1 (1.01, 1.20) | 1.1 (1.01, 1.20) | 1.1 (1.00, 1.14) | .0129b | 1.1 (1.00, 1.18) | 1.1 (1.00, 1.19) | 1.1 (1.01, 1.14) | .1292b |
| Activated partial thromboplastin time (aPTT), Median (Q1,Q3) | 30.0 (26.60, 33.70) | 30.2 (26.80, 34.00) | 29.0 (26.00, 32.30) | .0012b | 29.1 (26.00, 32.85) | 29.3 (26.30, 33.20) | 28.7 (26.00, 32.00) | .0449b |
| Total bilirubin (umol/l), Median (Q1,Q3) | 9.2 (6.60, 13.40) | 9.0 (6.60, 13.00) | 9.8 (6.60, 14.00) | .2437b | 10.3 (7.40, 15.50) | 10.4 (7.30, 15.10) | 10.2 (7.60, 16.40) | .4858b |
| Alanine transaminase (ALT) (U/L), Median (Q1,Q3) | 37.0 (24.00, 58.50) | 37.0 (24.00, 58.00) | 37.0 (24.00, 63.00) | .9159b | 37.0 (24.00, 58.00) | 37.0 (23.00, 58.00) | 37.0 (24.00, 62.00) | .8108b |
| Aspartate transaminase (AST) (U/L), Median (Q1,Q3) | 51.0 (34.00, 77.00) | 51.0 (34.00, 78.00) | 51.5 (33.00, 73.00) | .9038b | 48.0 (32.00, 74.00) | 47.0 (32.00, 75.00) | 52.0 (32.50, 72.50) | .3487b |
| Albumin (gm/l), Median (Q1,Q3) | 33.0 (29.00, 36.00) | 33.0 (28.65, 36.00) | 33.0 (31.00, 35.00) | .2117b | 33.0 (29.00, 36.00) | 32.5 (28.00, 36.00) | 33.0 (31.00, 36.00) | .2046b |
| Hematocrit (L/L), Mean (SD) | 0.4 (0.34, 0.43) | 0.4 (0.34, 0.43) | 0.4 (0.36, 0.43) | .0247b | 0.4 (0.35, 0.43) | 0.4 (0.35, 0.43) | 0.4 (0.36, 0.43) | .3029b |
| Creatine phosphokinase (CPK) (U/L), Median (Q1,Q3) | 172.0 (77.00, 428.00) | 169.0 (77.00, 416.50) | 239.0 (78.00, 531.00) | .1121b | 163.0 (75.00, 382.50) | 150.0 (72.00, 292.00) | 256.0 (87.00, 536.00) | .0034b |
| C-reactive protein (CRP) (mg/L), Median (Q1,Q3) | 130.0 (74.00, 201.00) | 126.3 (71.00, 193.96) | 147.5 (84.00, 253.00) | .0016b | 136.0 (76.00, 209.00) | 130.2 (69.80, 198.00) | 150.0 (83.00, 246.00) | .0101b |
| Fibrinogen level (gm/L), Median (Q1,Q3) | 5.5 (3.83, 7.05) | 5.5 (3.77, 7.09) | 5.5 (4.03, 7.02) | .8189b | 5.5 (3.77, 7.22) | 5.5 (3.72, 7.23) | 5.4 (4.03, 7.02) | .9210b |
| D-dimer level (mg/l), Median (Q1,Q3) | 1.3 (0.71, 3.10) | 1.3 (0.72, 3.20) | 1.2 (0.67, 2.51) | .6417b | 1.2 (0.70, 2.50) | 1.2 (0.70, 2.50) | 1.3 (0.72, 2.51) | .4235b |
| Ferritin level (ug/l), Median (Q1,Q3) | 662.7 (378.00, 1456.60) | 642.0 (374.40, 1395.10) | 791.0 (406.00, 2026.4) | .0177b | 688.3 (387.00, 1387.90) | 652.0 (379.00, 1269.60) | 775.9 (421.40, 2026.40) | .0220b |
| Blood glucose level (mmol/L), Median (Q1,Q3) | 10.9 (7.70, 15.40) | 10.9 (7.60, 15.30) | 11.4 (8.20, 16.00) | .1754b | 11.0 (7.80, 15.00) | 10.8 (7.51, 14.80) | 11.5 (8.20, 15.95) | .1272b |
| Best GCS at admission, Median (Q1,Q3) | 15.0 (14.00, 15.00) | 15.0 (14.00, 15.00) | 15.0 (14.00, 15.00) | .3679b | 15.0 (14.00, 15.00) | 15.0 (14.00, 15.00) | 15.0 (14.00, 15.00) | .8313b |
| Highest heart rate at admission (BPM), Median (Q1,Q3) | 102.0 (90.00, 114.00) | 102.0 (90.00, 114.00) | 103.0 (90.00, 116.00) | .5629b | 102.0 (91.00, 114.00) | 102.0 (91.00, 113.00) | 102.5 (90.00, 115.00) | .9285b |
| High dose of pharmacological DVT prophylaxis, n (%)f | 474 (39.9) | 382 (38.7) | 92 (45.3) | .2148c | 255 (41.4) | 181 (40.2) | 74 (44.6) | .6223c |
| Standard dose of pharmacological DVT prophylaxis, n (%)f | 614 (51.6) | 518 (52.5) | 96 (47.3) | 314 (51.0) | 234 (52.0) | 80 (48.2) | ||
| Low dose of pharmacological DVT prophylaxis, n (%)f | 101 (8.5) | 86 (8.7) | 15 (7.4) | 47 (7.6) | 35 (7.8) | 12 (7.2) | ||
| Patient received nephrotoxic drugs during ICU stay, n (%)e | 1091 (84.5) | 908 (83.4) | 183 (90.6) | .0092c | 553 (84.0) | 407 (82.6) | 146 (88.5) | .0718c |
| Comorbidity, n (%) | ||||||||
| Atrial fibrillation | 47 (3.6) | 43 (3.9) | 4 (2.0) | .1729c | 20 (3.0) | 17 (3.4) | 3 (1.8) | .2943c |
| Heart Failure | 108 (8.3) | 97 (8.8) | 11 (5.4) | .1058c | 52 (7.8) | 44 (8.8) | 8 (4.8) | .0953c |
| Hypertension | 726 (55.8) | 616 (56.1) | 110 (54.2) | .6233c | 345 (52.0) | 257 (51.6) | 88 (53.0) | .7536c |
| Diabetes mellitus | 773 (59.4) | 654 (59.5) | 119 (58.6) | .8129c | 384 (57.8) | 286 (57.4) | 98 (59.0) | .7166c |
| Dyslipidemia | 270 (20.7) | 210 (19.1) | 60 (29.6) | .0007c | 137 (20.6) | 91 (18.3) | 46 (27.7) | .0093c |
| Ischemic heart disease | 123 (9.4) | 102 (9.3) | 21 (10.3) | .6340c | 51 (7.7) | 39 (7.8) | 12 (7.2) | .8007c |
| Chronic kidney disease | 141 (10.8) | 130 (11.8) | 11 (5.4) | .0069c | 48 (7.2) | 40 (8.0) | 8 (4.8) | .1663c |
| Liver disease (any type) | 25 (1.9) | 23 (2.1) | 2 (1.0) | .2907d | 7 (1.1) | 6 (1.2) | 1 (0.6) | .5105d |
APACHE, Acute Physiology and Chronic Health Evaluation; DVT, deep vein thrombosis; GCS, Glasgow Coma Score; ICU, intensive care unit; PaO2/FiO2, partial pressure of oxygen/fraction of inspired oxygen; SOFA, Sequential Organ Failure Assessment; UFH, unfractionated heparin; VTE, venous thromboembolic events; WBCs, white blood cells. a T Test/bWilcoxon rank sum test was used to calculate the P-value.c Chi square/d Fisher’s exact teat was used to calculate P-value.
Nephrotoxic medications included intravenous vancomycin, gentamicin, amikacin, contrast, colistin, furosemide, and/or sulfamethoxazole/trimethoprim. f Patients who received either enoxaparin 40 mg daily or UFH 5000 units 3 times daily were grouped under the standard dose VTE prophylaxis. Any patient who received higher than standard dose but not as treatment dose (enoxaparin 1 mg/kg q12 h or 1.5 mg/kg q24 h or UFH infusion) was categorized as receiving high VTE prophylaxis dose. On the other hand, lower VTE prophylaxis was considered for patients who received enoxaparin <40 mg/day or UFH <5000 units 3 times daily/day.
Thrombosis
Composite thrombotic events were lower in patients who received doxycycline compared to the control group (6.1% vs 11.2%, P = .06). Moreover, logistic regression analysis showed lower odds of composite thrombotic events in patients who received doxycycline, but it failed to reach the statistical significance (OR: 0.54; 95% CI: 0.26-1.08; P = .08) (Table 2). The follow-up peak levels of D-dimer were lower in patients who received doxycycline than in the control group; however, did not reach the statistical significance (beta coefficient [95% CI]: −0.22 [−0.46, 0.03], P = .08) (Table 3).Table 2. The Outcomes of Critically Ill Patients With COVID-19 After PS Matching.
| Outcomes | Number of outcomes/Total number of patients | OR (95%CI) | P-valued | ||
|---|---|---|---|---|---|
| Control | Doxycycline | P-valuec | |||
| Composite thrombotic events, n (%)a | 55 (11.2) | 10 (6.1) | .06c | 0.54 (0.266, 1.083) | .08 |
| New onset atrial fibrillation, n (%) | 42 (8.4) | 16 (9.6) | .63c | 1.24 (0.673, 2.284) | .49 |
| Acute kidney injury, n (%)a | 194 (39.0) | 62 (37.3) | .71c | 1.01 (0.693, 1.468) | .96 |
| Liver injury, n (%)a | 55 (11.0) | 12 (7.2) | .16c | 0.66 (0.344, 1.273) | .22 |
| Pneumonia (bacterial/fungal), n (%)a | 197 (39.6) | 48 (28.9) | .01c | 0.65 (0.440, 0.949) | .03 |
| Secondary fungal infection, n (%)a | 62 (13.0) | 27 (16.6) | .26c | 1.36 (0.828, 2.223) | .23 |
| Outcomes | Control | Doxycycline | P-valuec | HR (95%CI) | P-valuee |
| 30-day mortality, n (%)a | 190 (41.7) | 49 (32.2) | .04c | 0.73 (0.52, 1.00) | .05 |
| In-hospital mortality, n (%)a | 208 (45.2) | 56 (36.4) | .05c | 0.79 (0.59, 1.07) | .13 |
| P-valueb | beta coefficient (Estimates) (95%CI) | P-valuef | |||
| Ventilator-free days (Days), Mean (SD)a | 12.3 (12.64) | 14.2 (12.27) | .08b | 0.09 (−0.25, 0.44) | .59 |
| ICU Length of Stay (Days), Median (Q1,Q3)a | 11.0 (6.00, 20.00) | 11.0 (6.00, 21.00) | .93b | 0.01 (−0.12, 0.14) | .83 |
| Hospital Length of Stay (Days), Median (Q1,Q3)a | 18.0 (11.00, 28.00) | 17.0 (11.00, 30.00) | .86b | 0.01 (−0.12, 0.13) | .92 |
Abbreviations: CI, confidence interval; HR, hazard ratio; ICU, intensive care unit; OR, odds ratio; PS, propensity score; Q1, quartile 1; Q3, quartile 3. a Denominator of the percentage is the total number of patients. b Wilcoxon rank sum test was used to calculate the P-value. c Chi-square test was used to calculate the P-value. d Logistic regression analysis is used to calculate the OR and P-value. e Cox proportional hazards regression analysis was used to calculate HR and P-value. f Generalized linear model was used to calculate estimates and P-value.
Table 3. COVID-19 Surrogate Markers.
| Follow-up markers | Beta coefficient (Estimates) (95%CI) | P-valueb |
|---|---|---|
| D-dimer level (mg/L), Median (Q1, Q3)a | −0.22 (−0.46, 0.03) | .08 |
| Fibrinogen level (gm/l), Median (Q1, Q3)a | −0.08 (−0.18, 0.03) | .14 |
| Creatine phosphokinase level (U/L), Median (Q1, Q3)a | −0.03 (−0.29, 0.23) | .83 |
| C-reactive protein (CRP) (mg/l), Median (Q1, Q3)a | 0.13 (−0.04, 0.29) | .13 |
Abbreviations: CI, confidence interval; Q1, quartile 1; Q3, quartile 3. Denominator of the percentage is the total number of patients. b Generalized linear model was used to calculate estimates and P-value.
30-day and Hospital Mortality
In our crude analysis, patients who received doxycycline during their ICU stay had a statistically significant reduction in 30-day mortality (32.2% vs. 41.7%, P = .04) and in-hospital mortality (36.4% vs. 45.2%, P = .05) compared to the control group, respectively. After the Cox proportional hazard regression analysis, the hazard of 30-day mortality was non-significantly lower in patients who received doxycycline (HR: 0.73; 95% CI: 0.52-1.00; P = .05). Similarly, the in-hospital mortality was lower than the control group; but was not statistically significant (HR: 0.79; 95% CI: 0.59-1.07; P = .13) (Table 2).
VFDs and LOS
The mean VFDs was longer in patients who received doxycycline compared to the control group (14.2 ± 12.2 vs 12.3 ± 12.6; P = .08). However, it was not statistically significant in the regression analysis (beta coefficient [95% CI]: 0.09 [−0.25, 0.44], P = .59). The ICU and hospital LOS were similar between the two groups (beta coefficient [95% CI]: 0.01 [−0.12, 0.14], P = .83, and beta coefficient [95% CI]: 0.01 [−0.12, 0.13], P = .92, respectively) (Table 2).
ICU Acquired Complications
Patients who received doxycycline had statistically significantly lower odds of pneumonia during their stay (OR: 0.65; 95% CI: 0.44-0.94; P = .03). Other complications during the stay were similar between the two groups; the new onset of Afib (OR: 1.24; 95% CI: 0.67-2.28; P = .49), AKI (OR: 1.01; 95% CI: 0.69-1.46; P = .96), liver injury (OR: 0.66; 95% CI: 0.34-1.27; P = .22), and secondary fungal infection (OR: 1.36; 95% CI: 0.82-2.22, P = .22) (Table 2).
Discussion
Doxycycline has potential preliminary data for reducing thromboembolic events and mortality. Our study evaluated the clinical outcomes of doxycycline in critically ill patients with COVID-19. We found a signal for reduction of thromboembolic events in patients with COVID-19 who received doxycycline concurrently during their hospital stay, yet not statistically significant. Notably, 30-day mortality was statistically significantly lower in the doxycycline group. Furthermore, ICU complications were numerically lower in the doxycycline group but did not reach a statistically significant difference.
On the other hand, hospital/ventilator-acquired pneumonia (bacterial/fungal) was statistically significantly lower in the doxycycline group. Patients with severe COVID-19 infection can progress to direct and indirect cardiovascular complications, including acute MI, arrhythmia, and vascular thrombosis.15,31 To our knowledge, this is the first observational study investigating the doxycycline effects on the incidence of thrombosis and mortality in critically ill patients with COVID-19.
In the early phases of the COVID-19 pandemic, doxycycline was introduced as an alternative to azithromycin, particularly for those at risk of severe QTc prolongation caused by azithromycin. It was proposed as a potential treatment for high-risk hospitalized patients with severe COVID-19 who require ICU admission.15 Similar to other COVID-19 studies, our cohort used a daily dose of 200 mg of doxycycline.24,32 In addition to its safety profile, doxycycline has pleiotropic, anti-inflammatory, anti-oxidative, and tissue-protective effects.33 Therefore, initiating doxycycline earlier in the course of the disease’s progression may be useful to target the inflammatory process at its early stages; the population included in our study have received doxycycline within 48 hours after ICU admission for a median length of five days.
In our study, among adults with COVID-19, adding doxycycline during ICU stay had lower odds of thrombosis, but after logistic regression analysis did not significantly reduce the incidence of arterial/venous thrombosis compared with standard of care alone. No other clinical study is available to compare our findings to, despite the two groups having comparable baseline fibrinogen, D-dimer, and platelet levels as well as similar dose intensity of pharmacological DVT prophylaxis. Nevertheless, a preclinical study found that doxycycline’s matrix metalloproteinase (MMP) inhibition significantly decreased acute PE-induced mortality (P < .05).24 Besides, the anti-inflammatory, immunomodulatory, and cardio-protective properties of doxycycline may be responsible for MMP gene inhibition, which is related to the severity and mortality of COVID-19.15,33–35 In addition to the ability of doxycycline to lower the odds of thrombosis, we found a lower percentage of 30-day mortality in the doxycycline group. Also, this difference was statistically significant compared to the other group (P = .04).
Yates et al. reported rapid clinical improvement in COVID-19 patients who received doxycycline for 5 to 14 days.14 Simultaneously, out of the 89 patients with confirmed moderate to severe COVID-19, 85% of the patients demonstrated clinical recovery after being treated with doxycycline for a week.15 These studies provide some evidence that doxycycline is an effective treatment in patients with COVID-19. However, one prospective observational study involved 315 patients hospitalized with COVID-19, of whom 47% received doxycycline. In the study, they found no evidence that doxycycline was associated with decreased 30-day mortality compared with standard care alone.36
In a randomized open-label clinical trial evaluating COVID-19 treatments in 2689 randomly assigned patients, 827 received 100 mg of doxycycline once a day for six days after receiving 200 mg on the first day. Compared to standard therapy alone, doxycycline did not accelerate overall recovery time, reduce hospitalizations, or prevent COVID-19-related deaths.37 Our findings are consistent with this study that found no difference between the usual care and doxycycline groups in terms of ICU or hospital LOS (P = .83 and .92, respectively).
We noted relatively comparable high VFD in those treated with doxycycline (14.2 ± 12.27), compared with usual care (12.3 ± 12.64), which could be attributed to a lower rate of hospital/ventilator acquired pneumonia (bacterial/fungal) in our patients, but the difference did not reach a significance (P = .59). Conforti et al. described the potential benefit of doxycycline therapy in patients with severe COVID-19 through inhibiting pro-inflammatory cytokines engaged with the development of lung injury and acute respiratory distress syndrome.11 Also, a randomized trial by Butler et al. found no statistically significant difference in MV and oxygen administration within 28 days with doxycycline plus standard care compared with standard care alone in people with COVID-19 in a community based settings.37 On the other hand, a cohort study for high-risk patients in long-term care facilities with moderate to severe COVID-19 treated with doxycycline showed improvement in oxygen saturation.15 At the same time, we found that the doxycycline group has a significantly low rate of hospital/ventilator-acquired pneumonia in patients with COVID-19 compared to usual care (28.9% vs 39.6%; P = .02). A similar finding by Muge Cevik et al. included 49 non-ICU patients with low severity hospital-acquired pneumonia who were successfully treated with doxycycline.32 In our study, no significant difference was observed between the two groups regarding secondary fungal infection, new-onset Afib, AKI, or liver injury.
Our study has some strengths. We included a relatively large sample size of patients from a wide variety of centers, evaluated endpoints relevant to clinical practice, and had clear outcomes definitions to limit variability in data collection. The ratio of doxycycline and the other group of patients was 3:1. The use of PS matching allowed us to obtain a good balance between our treatment and control groups on measured covariates.
In contrast, this study has limitations. Since this is a retrospective observational study, the results may be considered a hypothesis that serves as the basis for randomized controlled trials. Moreover, it should be considered that all patients underwent similar treatment and interventions, according to the approved internal guide for patients with COVID-19 admitted to the study centers. Also, we included only patients who started on doxycycline ≤48 hours of ICU admission. We presumed that patients who were discharged from the hospital survived, which might affect the result of 30-day mortality. Finally, the analysis of the effects of doxycycline should be interpreted cautiously because it was not conducted on randomized groups and might therefore be affected by several measured and unmeasured confounding factors.
Conclusion
Doxycycline use as adjunctive therapy in critically ill patients with COVID-19 may be a desirable therapeutic option for thrombosis reduction and survival benefits. Additional randomized control studies are needed to evaluate doxycycline’s potential therapeutic effects and to confirm this study findings.