Abstract

Background

We assessed the association between early oseltamivir treatment and influenza complications in hospitalized patients.

Methods

A retrospective cohort study, including adults with laboratory-confirmed 2009 influenza A (H1N1) in three hospitals in Israel, was performed between July 2009 and January 2010, when admission was limited to high-risk patients. We compared patients treated with oseltamivir early versus late (>48 h after symptom onset). We analysed risk factors for complications, defined as radiographic pneumonia, hypoxia, mechanical ventilation, intensive care unit admission, haemodynamic support or in-hospital death. Risk factors for complications on univariate analysis were entered into a multivariable logistic regression analysis. Odds ratios (ORs) with 95% confidence intervals (CI) are reported.

Results

Four hundred and forty-nine inpatients treated with oseltamivir were included, 189 (42.1%) of whom were treated early. Influenza complications occurred significantly more frequently among patients treated late with oseltamivir [150/260 (57.7%) versus 67/189 (35.4%), P < 0.001]. Late oseltamivir remained significantly associated with complications in the adjusted analysis (OR 2.37, 95% CI 1.52–3.70). Other independent risk factors included dyspnoea, disease severity on admission, lower sodium and treatment at one hospital; rhinorrhoea was protective. In an analysis adjusted for the propensity for early treatment the association remained significant (OR 2.21, 95% CI 1.41–3.46). Initiation of oseltamivir >48 h after admission was associated with a higher rate of complications documented after admission (OR 4.09, 95% CI 1.55–10.80). Severe complications (excluding hypoxia and uncomplicated pneumonia) occurred more frequently with late oseltamivir (adjusted OR 3.28,95% CI 1.56–6.89).

Conclusions

Initiation of oseltamivir within 48 h of symptom onset was associated with fewer complications in patients hospitalized with 2009 influenza A (H1N1).

Introduction

Data on the efficacy of neuraminidase inhibitors in influenza derive from trials recruiting healthy outpatients with mild influenza. The relevant outcome in this patient population is the number of days of symptomatic illness and the most recent systematic review reports a significant reduction with zanamivir or oseltamivir if taken within 48 h of symptom onset, with a hazard ratio of 1.22 (1.14–1.31).1,2 In this patient population complications are rare, leading to uncertainty with regard to the effects of neuraminidase inhibitors on influenza complications. For example, the pooled risk ratio for the composite outcome of pneumonia, bronchitis, sinusitis or otitis media requiring antibiotics was 0.55 (0.22–1.35) with only three trials included.1 There are no data on severe complications, such as those encountered in hospitalized patients during the recent 2009 A (H1N1) pandemic.

In the absence of randomized controlled trials assessing the efficacy of neuraminidase inhibitors on outcomes in hospitalized patients with influenza, our current knowledge relies on observational studies. The recent 2009 influenza A (H1N1) pandemic, associated with high complication and mortality rates,3–6 highlighted the need to understand the effects of oseltamivir and its time of initiation on patient outcomes.

We sought to analyse the effects of early initiation of oseltamivir on complications among adults with 2009 influenza A (H1N1) requiring hospitalization. As a policy during the study period, all inpatients with suspected influenza received oseltamivir. Thus, our analysis focuses on the time of initiation of oseltamivir relative to symptom onset rather than on treated versus untreated patients. Our hypothesis was that if oseltamivir is effective, late-onset treatment would be associated with a higher rate of complications.

Methods

We conducted an observational cohort study. The study protocol was planned prospectively. Included patients were identified using a prospective registry of all influenza test results at each hospital. Data collection was performed retrospectively through patient chart review and electronic databases by a single investigator. The study was approved by the local ethics committee at each hospital with a waiver for informed consent, given the non-interventional nature of the study. All data were stored anonymously.

We included all adults with laboratory-confirmed pandemic 2009 influenza A (H1N1) in three hospitals in central Israel between 22 July 2009 and the end of the influenza pandemic in January 2010. During this period, the national policy was to admit only patients requiring hospitalization (defined by symptoms or exacerbation of chronic conditions) and to provide oseltamivir treatment for all inpatients with suspected or confirmed influenza (and for outpatients with risk factors for complications). The standard protocol was 75 mg of oseltamivir twice daily for 5 days. Laboratory confirmation was performed by RT–PCR of respiratory samples at the Ministry of Health's Central Virology Laboratory.7

The exposure variable of interest was late treatment with oseltamivir, defined as oseltamivir initiation later than 48 h after symptom onset. The primary outcome assessed was influenza complications defined as follows: pulmonary infiltrates visualized on chest X-ray or CT scan; documentation of hypoxia defined as arterial saturation <90%; mechanical ventilation; intensive care unit (ICU) admission; need for haemodynamic support; or in-hospital death. In the primary outcome analysis we included all outcomes occurring after symptom onset (before or after start of oseltamivir), to avoid the immortal time bias that would occur when assessing only outcomes occurring after treatment initiation.8 Secondary outcomes included events occurring only after initiation of oseltamivir and those presenting after admission. For the latter analysis, we defined early and late treatment in relation to admission date (within or later than 48 h after admission, respectively). In addition, we defined severe complications as radiographic pneumonia accompanied by hypoxia, need for mechanical ventilation, any ICU admission, haemodynamic support or in-hospital death, omitting from the outcome definition hypoxia or uncomplicated pneumonia alone. The time of symptom onset was recorded by caring physicians routinely and the timing of oseltamivir treatment was extracted from the patients’ charts in the primary care clinics or in hospital. Patients were followed until final discharge (including readmissions) or death.

Potential confounders considered for prediction of the outcome included: demographic patient characteristics; background conditions, including identified risk factors for complications of influenza in general and pandemic 2009 A (H1N1) specifically, the Charlson co-morbidity score and functional status; concurrent medications; influenza vaccinations (2009 A-H1N1 vaccines became available only in November 2009); period of hospitalization (segregated into two periods, before or after October 2009); influenza disease characteristics; and severity at admission [including the Sequential Organ Failure Assessment (SOFA) score].

We compared baseline and outcome data for patients given early versus late oseltamivir, developing a propensity score for early treatment. A multivariable logistic regression analysis was performed including the outcome as the dependent variable and risk factors significantly associated with the outcome on univariate analysis (P < 0.05) as independent variables. We excluded from the multivariable analysis highly correlated risk factors using clinical interpretation to select included risk factors. Early versus late oseltamivir treatment was forced into the model. Other variables were entered using backward stepwise regression. Adjusted odds ratios (ORs) with 95% confidence intervals (CIs) are reported. In sensitivity analyses we included the propensity for early oseltamivir in the model and imputed missing laboratory values. Missing values were imputed using multiple imputation analysis with fully conditional specification.9 The model's fit was assessed using the Hosmer–Lemeshow goodness of fit test and its prediction was assessed using a receiving operating characteristic (ROC) curve. Dichotomous data were compared using the χ2 test or Fisher's exact test as appropriate. All continuous data are presented as medians with ranges and were compared using the Mann–Whitney U-test. Analyses were conducted using IBM SPSS Statistic 19 (IBM Corporation, NY, USA).

Results

Between the end of July 2009 and the end of January 2010, 506 patients with documented 2009 Influenza A (H1N1) were hospitalized in the three participating hospitals. Four patients were transferred to one of the hospitals for extracorporeal membrane oxygenation; these patients were excluded from the analysis since data on symptom onset and timing of treatment were missing. Fifty-three patients who were not treated with oseltamivir were excluded. These patients had fewer co-morbidities and milder disease compared with other patients; none required ICU admission or died during follow-up (Tables S1 and S2, available as Supplementary data at JAC Online).

Of 449 patients included, 189 (42.1%) were provided with oseltamivir within 48 h of symptom onset (early oseltamivir) and 421 were treated within 48 h of admission (93.8%). The comparison between early versus late oseltamivir is presented in the supplementary tables. Among underlying conditions, early treatment was associated only with pregnancy or post-partum state (Table S1). Of the presenting symptoms and signs, early treatment was associated with myalgia and lower lymphocyte count (Table S2). The presence of pneumonia and higher lactate dehydrogenase (LDH) on admission were associated with late treatment. Treatment onset differed significantly between the hospitals; early treatment was most frequent in the smallest hospital, which is a primary care hospital where none of the patients required ICU admission (Table S1). The propensity score for early oseltamivir treatment was not highly predictive, given the paucity of variables significantly associated with early treatment (hospital, myalgia, pregnancy/post-partum and total lymphocyte count; ROC area 0.66, 95% CI 0.60–0.71).

Early treatment with oseltamivir was associated with fewer complications as defined by the primary outcome [67/189 (35.4%) early versus 150/260 (57.7%) late, P < 0.001]. Early oseltamivir was associated with a lower rate of all secondary outcomes and most individual components of the composite primary outcome (Table 1), including in-hospital mortality [1/189 (0.5%) early versus 13/260 (5.0%) late, P = 0.006]. Risk factors for influenza complications (primary outcome) on univariate analysis are shown in Tables 2 and 3. Risk factors included the winter months, treatment in one of the hospitals (on account of a higher rate of documented hypoxia), older age, several chronic diseases and chronic steroid treatment. Obesity was not a risk factor for complications in our cohort. Pregnant/post-partum women experienced a lower outcome rate, but were treated early with oseltamivir (Table S1). Presenting symptoms associated with complications were dyspnoea as a risk factor and rhinorrhoea and myalgia as protective factors. Higher urea was associated with complications, as were lower sodium and albumin levels and higher LDH. Albumin and LDH measurements were available on admission only in a small proportion of patients (Table 3). Both the Charlson score for background conditions and the SOFA score for disease severity on admission predicted complications. Variables in Tables S1 and S2 omitted from Table 2 were not significantly associated with complications.

Table 1.

Influenza complications for patients given early versus late oseltamivir (>48 h after symptom onset)

 Early oseltamivir (N = 189) Late oseltamivir (N = 260) P value 
Any complication (primary outcome)a 67 (35.4%) 150 (57.7%) <0.001 
Any complication developing after admission 15 (7.9%) 42 (16.2%) 0.010 
Any complication developing after start of oseltamivir 13 (6.9%) 33 (12.7%) 0.045 
Severe complicationsb 15 (7.9%) 60 (23.1%) <0.001 
Any pneumonia 42 (22.2%) 122 (46.9%) <0.001 
Pneumonia with hypoxaemia 15 (7.9%) 52 (20.0%) <0.001 
Hypoxaemia 38 (20.1%) 73 (28.1%) 0.053 
ICU admission 6 (3.2%) 24 (9.2%) 0.011 
Mechanical ventilation 6 (3.2%) 21 (8.1%) 0.031 
Mortality in hospital 1 (0.5%) 13 (5.0%) 0.006 
Mortality 30 days 2 (1.1%) 9 (3.5%) 0.129 
No. of hospitalization days for patients discharged alive, median (range), [mean]/no. evaluated 4 (1–37), [5]/188 4 (1–112), [7]/247 0.001 
 Early oseltamivir (N = 189) Late oseltamivir (N = 260) P value 
Any complication (primary outcome)a 67 (35.4%) 150 (57.7%) <0.001 
Any complication developing after admission 15 (7.9%) 42 (16.2%) 0.010 
Any complication developing after start of oseltamivir 13 (6.9%) 33 (12.7%) 0.045 
Severe complicationsb 15 (7.9%) 60 (23.1%) <0.001 
Any pneumonia 42 (22.2%) 122 (46.9%) <0.001 
Pneumonia with hypoxaemia 15 (7.9%) 52 (20.0%) <0.001 
Hypoxaemia 38 (20.1%) 73 (28.1%) 0.053 
ICU admission 6 (3.2%) 24 (9.2%) 0.011 
Mechanical ventilation 6 (3.2%) 21 (8.1%) 0.031 
Mortality in hospital 1 (0.5%) 13 (5.0%) 0.006 
Mortality 30 days 2 (1.1%) 9 (3.5%) 0.129 
No. of hospitalization days for patients discharged alive, median (range), [mean]/no. evaluated 4 (1–37), [5]/188 4 (1–112), [7]/247 0.001 

aAny complication: pulmonary infiltrates visualized on chest X-ray or CT scan; documentation of hypoxia defined as arterial saturation <90%; mechanical ventilation; ICU admission; need for haemodynamic support; or in-hospital death.

bSevere complications are radiographic pneumonia with hypoxia, need for mechanical ventilation, ICU admission, haemodynamic support or in-hospital death (omitting hypoxia alone and uncomplicated pneumonia).

Table 2.

Demography and background conditions

 Without complications (N = 232), n (%)a With complications (N = 217), n (%)a P value 
Admission period   0.029 
 July 2009–October 2009 90 (38.8) 63 (29)  
 November 2009–January 2010 142 (61.2) 154 (71)  
Hospitalb   <0.001 
 Beilinson, n = 178 97 (41.8) 81 (37.3)  
 Hasharon, n = 116 82 (35.3) 34 (15.7)  
 Meir, n = 155 53 (22.8) 102 (47)  
Age (years), median (range) 39 (16–89) 48 (16–93) <0.001 
Female sex 117 (50.4) 118 (54.4) 0.403 
Poor functional capacityc 14 (6) 32 (14.7) 0.002 
Obesity   0.801 
 body mass index >30 28 (12.1) 27 (12.4)  
 body mass index >40 7 (3) 9 (4.1)  
Pregnancy or post-partum 21 (9.1) 7 (3.2) 0.011 
Congestive heart failure 9 (3.9) 23 (10.6) 0.006 
Diabetes mellitus 30 (12.9) 44 (20.3) 0.036 
Chronic lung disease 63 (27.2) 87 (40.1) 0.004 
Cancer 10 (4.3) 23 (10.6) 0.011 
Immune suppression 23 (9.9) 20 (9.2) 0.802 
Chronic steroid treatment 14 (6) 26 (12) 0.027 
Charlson score, median (range) 0 (0–7) 1 (0–8) 0.001 
Influenza vaccination in the past 59/232 (25.4) 71/214 (33.2) 0.072 
 Without complications (N = 232), n (%)a With complications (N = 217), n (%)a P value 
Admission period   0.029 
 July 2009–October 2009 90 (38.8) 63 (29)  
 November 2009–January 2010 142 (61.2) 154 (71)  
Hospitalb   <0.001 
 Beilinson, n = 178 97 (41.8) 81 (37.3)  
 Hasharon, n = 116 82 (35.3) 34 (15.7)  
 Meir, n = 155 53 (22.8) 102 (47)  
Age (years), median (range) 39 (16–89) 48 (16–93) <0.001 
Female sex 117 (50.4) 118 (54.4) 0.403 
Poor functional capacityc 14 (6) 32 (14.7) 0.002 
Obesity   0.801 
 body mass index >30 28 (12.1) 27 (12.4)  
 body mass index >40 7 (3) 9 (4.1)  
Pregnancy or post-partum 21 (9.1) 7 (3.2) 0.011 
Congestive heart failure 9 (3.9) 23 (10.6) 0.006 
Diabetes mellitus 30 (12.9) 44 (20.3) 0.036 
Chronic lung disease 63 (27.2) 87 (40.1) 0.004 
Cancer 10 (4.3) 23 (10.6) 0.011 
Immune suppression 23 (9.9) 20 (9.2) 0.802 
Chronic steroid treatment 14 (6) 26 (12) 0.027 
Charlson score, median (range) 0 (0–7) 1 (0–8) 0.001 
Influenza vaccination in the past 59/232 (25.4) 71/214 (33.2) 0.072 

aAll patients evaluated unless otherwise noted.

bParticipating hospitals: Rabin Medical Center, Beilinson Hospital; Rabin Medical Center, Hasharon Hospital; and Sapir Medical Center, Meir Hospital.

cAssistance required for everyday activities.

Table 3.

Disease presentation

 Without complications (N = 232), n (%) With complications (N = 217), n (%) P value 
Maximal fever (°C) 39 (36.1–41)/232 38.9 (36–40.5)/213 0.261 
Chills 55 (23.7) 37 (17.1) 0.081 
Rhinorrhoea 81 (34.9) 52 (24) 0.011 
Myalgia 98 (42.2) 60 (27.6) 0.001 
Sore throat 74 (31.9) 47 (21.7) 0.015 
Cough 193 (83.2) 184 (84.8) 0.644 
Dyspnoea 81 (34.9) 119 (54.8) <0.001 
Chest pain 19 (8.2) 27 (12.4) 0.138 
SOFA score on admission 0 (0–6)/231 0 (0–9)/215 0.001 
SOFA score at start of oseltamivir 0 (0–6)/229 0 (0–10)/216 0.001 
Total white blood cell count (103/μL) 7.51 (1.72–23.82)/229 7.46 (1.92–26.07)/215 0.829 
Total lymphocyte count (103/μL) 0.9 (0.1–23)/225 0.9 (0.1–17.1)/212 0.032 
Platelets (103/μL) 198 (4–862)/229 198 (23–748)/215 0.214 
Albumin (g/dL) 4.1 (0.3–5.1)/100 3.9 (0.6–5)/76 0.014 
Sodium (mEq/L) 137 (122–151)/224 136 (123–147)/201 0.001 
Creatinine (mg/dL) 0.88 (0.36–56)/192 0.9 (0.2–23)/162 0.166 
Urea (mg/dL) 27 (7–132)/219 30 (8–189)/211 0.002 
LDH (U/L) 389 (227–2218)/96 461 (247–2989)/65 0.001 
 Without complications (N = 232), n (%) With complications (N = 217), n (%) P value 
Maximal fever (°C) 39 (36.1–41)/232 38.9 (36–40.5)/213 0.261 
Chills 55 (23.7) 37 (17.1) 0.081 
Rhinorrhoea 81 (34.9) 52 (24) 0.011 
Myalgia 98 (42.2) 60 (27.6) 0.001 
Sore throat 74 (31.9) 47 (21.7) 0.015 
Cough 193 (83.2) 184 (84.8) 0.644 
Dyspnoea 81 (34.9) 119 (54.8) <0.001 
Chest pain 19 (8.2) 27 (12.4) 0.138 
SOFA score on admission 0 (0–6)/231 0 (0–9)/215 0.001 
SOFA score at start of oseltamivir 0 (0–6)/229 0 (0–10)/216 0.001 
Total white blood cell count (103/μL) 7.51 (1.72–23.82)/229 7.46 (1.92–26.07)/215 0.829 
Total lymphocyte count (103/μL) 0.9 (0.1–23)/225 0.9 (0.1–17.1)/212 0.032 
Platelets (103/μL) 198 (4–862)/229 198 (23–748)/215 0.214 
Albumin (g/dL) 4.1 (0.3–5.1)/100 3.9 (0.6–5)/76 0.014 
Sodium (mEq/L) 137 (122–151)/224 136 (123–147)/201 0.001 
Creatinine (mg/dL) 0.88 (0.36–56)/192 0.9 (0.2–23)/162 0.166 
Urea (mg/dL) 27 (7–132)/219 30 (8–189)/211 0.002 
LDH (U/L) 389 (227–2218)/96 461 (247–2989)/65 0.001 

Continuous variables presented as median (range)/no. assessed. All patients evaluated unless otherwise noted.

On multivariable analysis, late initiation of oseltamivir remained significantly associated with complications (OR 2.37, 95% CI 1.52–3.70). Other variables independently and significantly associated with complications included treatment in one hospital, dyspnoea, no rhinorrhoea, SOFA score, lower sodium and lower lymphocyte count on admission (Table 4). In sensitivity analyses, late oseltamivir remained significantly associated with the occurrence of complications when adjusting for the propensity for early versus late treatment and when including all patients in the analysis using missing value imputations (Table 4).

Table 4.

Multivariable logistic regression analysis for the occurrence of influenza complications

Variable OR (95% CI) 
Main model (418 patients included)a 
Late oseltamivir (>48 h after symptom onset) 2.37 (1.52–3.70) 
Hospital  
 Beilinson reference 
 Hasharon 0.83 (0.46–1.5) 
 Meir 2.30 (1.39–3.82) 
Congestive heart failure 2.08 (0.87–4.96) 
Rhinorrhoea 0.49 (0.30–0.79) 
Dyspnoea 2.05 (1.32–3.18) 
SOFA score at admission 1.26 (1.08–1.47) 
Sodium at admissionb 0.93 (0.88–0.99) 
Total lymphocyte count at admissionb 0.86 (0.7–1.05) 
  
Hosmer–Lemeshow test 0.272 
Model constant P = 0.028 
ROC area 0.76 (0.71–0.80) 
Sensitivity analyses (only ORs for early oseltamivir shown) 
Including the propensity for early oseltamivir (N = 418)c 2.21 (1.41–3.46) 
Multiple imputation analysis for missing values (N = 449)d 2.33 (1.51–3.57) 
Variable OR (95% CI) 
Main model (418 patients included)a 
Late oseltamivir (>48 h after symptom onset) 2.37 (1.52–3.70) 
Hospital  
 Beilinson reference 
 Hasharon 0.83 (0.46–1.5) 
 Meir 2.30 (1.39–3.82) 
Congestive heart failure 2.08 (0.87–4.96) 
Rhinorrhoea 0.49 (0.30–0.79) 
Dyspnoea 2.05 (1.32–3.18) 
SOFA score at admission 1.26 (1.08–1.47) 
Sodium at admissionb 0.93 (0.88–0.99) 
Total lymphocyte count at admissionb 0.86 (0.7–1.05) 
  
Hosmer–Lemeshow test 0.272 
Model constant P = 0.028 
ROC area 0.76 (0.71–0.80) 
Sensitivity analyses (only ORs for early oseltamivir shown) 
Including the propensity for early oseltamivir (N = 418)c 2.21 (1.41–3.46) 
Multiple imputation analysis for missing values (N = 449)d 2.33 (1.51–3.57) 

aVariables included in the model but not retained in the final equation were admission period, age, poor functional capacity, pregnancy/post-partum, diabetes mellitus, cancer, sore throat, myalgia and urea levels.

bOR given for each 1 mEq/L increase in sodium and each 103/μL increase in lymphocyte count.

cThe propensity for early treatment was retained in the final equation and was significantly associated with the outcome (OR 14.79, 95% CI 1.28–171.28).

dVariables shown in Tables 2 and 3 were used for imputation of the following missing values (number of patients): SOFA score (3); lymphocytes (12); sodium (24); and urea (19). Only the lowest OR obtained out of five different imputation cycles is shown.

Risk factors for complications occurring after admission are shown in Table S3 (available as Supplementary data at JAC Online). In the adjusted analysis, initiation of oseltamivir >48 h after admission was significantly associated with complications developing after admission (OR 4.09, 95% CI 1.55–10.80). Other risk factors were treatment in one hospital and the SOFA score on admission. The model had good predictive value (ROC area 0.77, 95% CI 0.69–0.84). Complications developing after initiation of oseltamivir were rare. The OR for these complications adjusted only for the treating hospital was 1.97 (95% CI 1.01–3.85).

Severe complications, as defined by our secondary outcome, were observed in 75/449 (16.7%) of patients. In this group of patients in-hospital mortality was 18.7% (14/75), 40% were admitted to the ICU (30/75) and the median hospital stay for those discharged alive was 8 days (range 2–112). Oseltamivir was started late in 60/75 (80%) of patients with severe complications compared with 200/374 (53.5%) without severe complications (P < 0.001). Haemoptysis, lower systolic blood pressure and higher creatine phosphokinase (CPK) on admission were significantly associated with severe complications. Other risk factors were mostly similar to those shown for any complication (Table S4, available as Supplementary data at JAC Online). Late oseltamivir was significantly associated with severe complications on multivariable analysis (OR 3.28, 95% CI 1.56–6.89). Other independent risk factors were treatment at one hospital, congestive heart failure, dyspnoea, lower sodium levels and increasing SOFA score on admission (Table S4).

Discussion

In our dataset, initiation of oseltamivir within 48 h after symptom onset was associated with a lower rate of 2009 influenza (H1N1) complications, including hypoxia, pneumonia, need for respiratory or haemodynamic support, ICU admission and in-hospital death. The association remained significant when adjusted for other risk factors for complications and in several sensitivity analyses, focusing on outcomes occurring after initiation of oseltamivir, severe complications only (omitting hypoxia alone and uncomplicated pneumonia from the outcome definition) and adjusting for the differences between patients given early versus late oseltamivir using a propensity score-adjusted analysis. Initiation of oseltamivir >48 h after admission was significantly associated with complications whose onset was after admission.

Beneficial effects of oseltamivir in hospitalized patients during the 2009 H1N1 pandemic were reported in other studies. In critically ill patients, studies reported a higher rate of any treatment or early treatment among survivors compared with non-survivors.4,6 Receipt of antiviral treatment within 2 days of symptom onset was the only variable significantly associated with survival without ICU admission on multivariable analysis in a cohort of 272 patients hospitalized with 2009 H1N1 influenza in the USA early during the epidemic (April–June 2009).10 The effects of oseltamivir were reported for two nationwide cohorts from China.11,12 In the early period, when all patients with suspected influenza were hospitalized regardless of disease severity, treatment with oseltamivir was associated with 0.12 the odds (95% CI 0.08–0.18) for developing pneumonia (number needed to treat four patients)12 Similar ORs were observed for early and late start of oseltamivir (after 2 days of symptoms) and none of the patients in this cohort required ICU admission or mechanical ventilation. Later on, when only patients who required medical treatment for complications were hospitalized, as in our study, a more stringent outcome of death or ICU admission was assessed.11 Oseltamivir started >4 days after symptom onset was associated with a higher complication rate (adjusted OR 1.42, 95% CI 1.20–1.67). In a study from France during the same period, early oseltamivir (≤2 days after symptom onset) was given to 45 (39%) patients requiring ICU admission compared with 153 (74%) other hospitalized patients (unadjusted P < 0.001).13 In a cohort of solid organ transplant recipients, delayed antiviral treatment (>48 h) was associated with increased risk of ICU admission (adjusted OR 3.03, 95% CI 1.24–7.39).

Given that all these data are derived from observational studies, the marked benefit of oseltamivir reported in these studies should be viewed with caution and potential sources of bias should be investigated. An obvious problem occurs if early versus late treatment is the consequence of disease presentation and complications rather than vice versa. It might seem plausible that early complications would lead to early treatment, thus attenuating its observed benefit. Rather, we observed that early treatment was associated with classical flu symptoms, such as myalgia, while physicians delayed oseltamivir initiation for patients with pneumonia or other unexpected influenza presentations. In the Chinese study assessing the outcome of pneumonia, only pneumonia developing after starting oseltamivir in treated patients was considered as an outcome.12 The immortal time bias might explain part of the large benefit reported in this study; untreated patients had a longer time to experience the outcome.8 In our study, we assessed all complications rather than only those commencing after admission or after the start of oseltamivir. In all analyses oseltamivir appeared beneficial, although the analysis of complications developing after the start of oseltamivir masks the benefit attributable to complications developing between early and late initiation. All studies focusing on inpatients lack information on patients treated with oseltamivir that were not hospitalized. These patients probably all recovered. If a large proportion of late-onset treatment occurs in this patient population, the effects might be biased in favour of early treatment. However, the consistency of the clinical benefit of oseltamivir reported in all recent studies, supported by data on earlier virus clearance with oseltamivir,14–17 confers reasonable confidence that oseltamivir treatment is beneficial and that early treatment (within 2 days of symptom onset) is advantageous.

Limitations specific to our study include the lack of complete data for laboratory values that might have been significantly associated with the occurrence of complications, including LDH as a marker of lung injury and CPK. Only a few variables could be included in the analysis of complications developing after admission, because of the low outcome rate in this analysis. Our primary outcome included minor complications. We selected it considering that for the individual patient these minor events (hypoxia, uncomplicated pneumonia) represented some burden, including anxiety and loss of work days. In an analysis restricted to more significant complications, early oseltamivir remained associated with benefit.

In summary, we observed an association between early initiation of oseltamivir, within 48 h of symptom onset, and a lower rate of complications related to influenza. This association was not explained by differences between patients treated early versus late and was stable in several analyses. Other observational studies support a benefit for oseltamivir treatment (versus no treatment) and early oseltamivir initiation. Since oseltamivir is a relatively safe drug,2,18 we encourage its early use among patients with suspected oseltamivir-susceptible influenza at risk of complications and for patients presenting with complications.

Funding

The study was supported by internal funding.

Transparency declarations

None to declare.

Supplementary data

Tables S1 to S4 are available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/).

References

1
Jefferson
T
Jones
M
Doshi
P
, et al.  . 
Neuraminidase inhibitors for preventing and treating influenza in healthy adults: systematic review and meta-analysis
BMJ
 , 
2009
, vol. 
339
 pg. 
b5106
 
2
Jefferson
T
Jones
M
Doshi
P
, et al.  . 
Neuraminidase inhibitors for preventing and treating influenza in healthy adults
Cochrane Database Syst Rev
 , 
2010
, vol. 
issue 2
 pg. 
CD001265
 
3
Davies
A
Jones
D
Bailey
M
, et al.  . 
Extracorporeal membrane oxygenation for 2009 influenza A(H1N1) acute respiratory distress syndrome
JAMA
 , 
2009
, vol. 
302
 (pg. 
1888
-
95
)
4
Dominguez-Cherit
G
Lapinsky
SE
Macias
AE
, et al.  . 
Critically ill patients with 2009 influenza A(H1N1) in Mexico
JAMA
 , 
2009
, vol. 
302
 (pg. 
1880
-
7
)
5
Kumar
A
Zarychanski
R
Pinto
R
, et al.  . 
Critically ill patients with 2009 influenza A(H1N1) infection in Canada
JAMA
 , 
2009
, vol. 
302
 (pg. 
1872
-
9
)
6
Louie
JK
Acosta
M
Winter
K
, et al.  . 
Factors associated with death or hospitalization due to pandemic 2009 influenza A(H1N1) infection in California
JAMA
 , 
2009
, vol. 
302
 (pg. 
1896
-
902
)
7
Mendelson
E
Mandelboim
M
Grossman
Z
, et al.  . 
[Laboratory diagnosis of influenza H1N1 2009 at the Central Virology Laboratory in Israel during the first 12 weeks of the pandemic]
Harefuah
 , 
2009
, vol. 
148
 (pg. 
677
-
81
735
8
Suissa
S
Immortal time bias in observational studies of drug effects
Pharmacoepidemiol Drug Saf
 , 
2007
, vol. 
16
 (pg. 
241
-
9
)
9
van Buuren
S
Multiple imputation of discrete and continuous data by fully conditional specification
Stat Methods Med Res
 , 
2007
, vol. 
16
 (pg. 
219
-
42
)
10
Jain
S
Kamimoto
L
Bramley
AM
, et al.  . 
Hospitalized patients with 2009 H1N1 influenza in the United States, April-June 2009
N Engl J Med
 , 
2009
, vol. 
361
 (pg. 
1935
-
44
)
11
Yu
H
Feng
Z
Uyeki
TM
, et al.  . 
Risk factors for severe illness with 2009 pandemic influenza A (H1N1) virus infection in China
Clin Infect Dis
 , 
2011
, vol. 
52
 (pg. 
457
-
65
)
12
Yu
H
Liao
Q
Yuan
Y
, et al.  . 
Effectiveness of oseltamivir on disease progression and viral RNA shedding in patients with mild pandemic 2009 influenza A H1N1: opportunistic retrospective study of medical charts in China
BMJ
 , 
2010
, vol. 
341
 pg. 
c4779
 
13
Fuhrman
C
Bonmarin
I
Paty
AC
, et al.  . 
Severe hospitalised 2009 pandemic influenza A(H1N1) cases in France, 1 July-15 November 2009
Euro Surveill
 , 
2010
, vol. 
15
  
pii = 19463
14
Cao
B
Li
XW
Mao
Y
, et al.  . 
Clinical features of the initial cases of 2009 pandemic influenza A (H1N1) virus infection in China
N Engl J Med
 , 
2009
, vol. 
361
 (pg. 
2507
-
17
)
15
Li
IW
Hung
IF
To
KK
, et al.  . 
The natural viral load profile of patients with pandemic 2009 influenza A(H1N1) and the effect of oseltamivir treatment
Chest
 , 
2010
, vol. 
137
 (pg. 
759
-
68
)
16
Ling
LM
Chow
AL
Lye
DC
, et al.  . 
Effects of early oseltamivir therapy on viral shedding in 2009 pandemic influenza A (H1N1) virus infection
Clin Infect Dis
 , 
2010
, vol. 
50
 (pg. 
963
-
9
)
17
Na
S
Chong
YP
Kim
MN
, et al.  . 
Duration of viral shedding in patients admitted to hospital with pandemic influenza A/H1N1 2009 infection
J Med Virol
 , 
2010
, vol. 
83
 (pg. 
5
-
9
)
18
Toovey
S
Rayner
C
Prinssen
E
, et al.  . 
Assessment of neuropsychiatric adverse events in influenza patients treated with oseltamivir: a comprehensive review
Drug Saf
 , 
2008
, vol. 
31
 (pg. 
1097
-
114
)