Study of the Impact of Oseltamivir on the Risk for Pneumonia and Other Outcomes of Influenza, 2000–2005

Abstract

Context

Influenza results in large numbers of secondary complications and hospitalizations.

Objective

To assess the impact of oseltamivir on influenza-related complications and hospitalizations by analyzing health insurance claims data for 6 influenza seasons.

Design

A retrospective cohort study utilizing claims data from the 2000–2005 influenza seasons.

Setting

Claims data were obtained from Thomson Healthcare MarketScan Research Databases.

Patients

Patients prescribed oseltamivir within 1 day of influenza diagnosis were compared to those prescribed no antiviral therapy (controls).

Outcomes

Frequencies of pneumonia, other respiratory illnesses, and otitis media, and rates of hospitalization, were compared for the oseltamivir and no antiviral groups. Expenditure was also analyzed. Relative risk (RR) for each outcome was assessed using Cox proportional hazards regression.

Results

Overall, 31,674 patients received oseltamivir and were propensity matched to patients with no antiviral prescription. Oseltamivir reduced the risk of diagnosis of pneumonia by 15% (RR 0.85, 95% confidence interval [CI] 0.73, 0.98), other respiratory illnesses by 20% (RR 0.80, 95% CI: 0.76, 0.83), and otitis media and its complications by 31% (RR 0.69, 95% CI: 0.61, 0.79). The greatest reductions in pneumonia risk, of 57% and 52%, were observed for children aged 6–12 years (RR 0.43, 95% CI: 0.26, 0.71) and 1–2 years (RR 0.48, 95% CI: 0.24, 0.99), respectively. Overall, hospitalization rates were reduced by 38% (RR 0.62, 95% CI: 0.52, 0.74) in the oseltamivir group compared with the no antiviral group. Total adjusted expenditures in the oseltamivir and no antiviral groups were not significantly different.

Conclusions

Oseltamivir reduced the relative risk of influenza-related complications and hospitalization when prescribed immediately upon presentation of influenza.

Introduction

Influenza affects between 10% and 20% of the US population annually,[1] resulting in approximately 226,000 hospitalizations[2] and up to 36,000 deaths.[3,4] These hospitalizations and deaths typically result from secondary complications such as pneumonia and otitis media or from exacerbation of underlying medical conditions such as heart failure or asthma,[5,6] and are highest in children, the elderly, and people with certain existing health conditions.[5]

Current treatment guidelines recommend vaccination for at-risk groups including the elderly and young children, and treatment with antiviral agents in case of infection.[7] Four antiviral medications are currently approved for use in the United States: the adamantanes, rimantadine and amantadine, and the neuraminidase inhibitors, oseltamivir and zanamivir.[1,8] Adamantanes are effective against susceptible influenza A viruses[9] but were not recommended for use in the United States for the 2006–2007 influenza season due to increased reports of viral resistance.[10] Oseltamivir and zanamivir are approved for treatment of type A and B influenza viruses and have been shown to reduce the severity and duration of influenza and to decrease viral shedding if administered within 48 hours of symptom onset.[11–16]

Oseltamivir is also effective in preventing secondary complications of influenza. Oseltamivir treatment reduced the incidence of otitis media in children,[17] lower respiratory tract infections such as pneumonia in children[6] and the elderly,[18] and bronchitis and sinusitis in adults,[12] and also reduced hospitalization rates in adults and children and the number of prescriptions for antibiotics.[6,17,18] Similarly, zanamivir treatment decreased the number of respiratory tract complications and the number of events leading to prescriptions for antibiotics.[19]

The aim of this study was to assess the impact of oseltamivir on the risk of developing select influenza-related complications and subsequent hospitalization rates. This broad-based US national study was undertaken using a large cohort of patients with clinically diagnosed influenza in a retrospective database analysis.

Materials and Methods

Patients

This retrospective cohort study utilized health insurance claims data from the 2000–2005 influenza seasons (October 1 through March 31). Data were obtained for persons with both medical and pharmacy benefits from the Thomson Healthcare MarketScan Research Databases (including the Commercial Claims and Encounters Database and Medicare Supplemental and Coordination of Benefits Databases). The database is derived from employer and government-funded (Medicare) healthcare insurance plans and includes approximately 25 million individuals of all ages who are covered under a variety of fee-for-service and capitated provider reimbursement schemes. Inpatient and outpatient claims files are linked with outpatient prescription drug data via unique encrypted patient identifiers. Data from these databases have been used previously in a number of other health outcomes and health economics studies.[20–23]

Claims were screened to identify all patients with an influenza diagnosis (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9]-CM diagnosis code 487.xx) recorded on a claim for an outpatient physician visit (office visit, emergency room [ER] visit, or outpatient clinic visit). The first claim indicating evidence of influenza was used to establish the study index date (first date of influenza diagnosis). If a patient had more than 1 influenza episode in the same season, only the first was included, but the same individual may have contributed cases of influenza in different seasons.

Patients were excluded if they were not continuously insured from 3 months prior to the index date for the clinical outcomes study and 1 year prior to the index date for the safety analysis; and in both analyses, at least 1 month after the index date, except in cases where death occurred; and if both medical and pharmacy data were not available. As the outcomes evaluated in this study included pneumonia diagnosis and hospitalization (for pneumonia, other respiratory diagnosis, and any cause) following influenza diagnosis, patients were excluded if they presented with both influenza and pneumonia, or were admitted to, or already residing in, a healthcare institution (hospital, nursing home, or other facility) on the index date. In addition, patients were excluded if they received antiviral medication other than oseltamivir at any point during the baseline or follow-up period or if there was evidence of pregnancy.

Study Design

Eligible patients were grouped according to whether they had a prescription for oseltamivir. The treatment group was defined as having a clinical influenza diagnosis and a prescription for oseltamivir within 1 day before or after the index date. The treatment group was compared with a no antiviral group comprising patients with a diagnosis of influenza who did not have a claim for oseltamivir at any time during the influenza season. Propensity score methodology was used to match patients according to demographic characteristics and clinical histories, to ensure that the composition of the 2 groups was as similar as possible.

Study Outcome Measures

Clinical outcomes. The primary outcome was the occurrence of pneumonia, other respiratory conditions, or otitis media and its complications within 14 days of onset of influenza. Diagnoses were based on ICD-9 codes (Table 1). Pneumonia, including influenza pneumonia, other viral pneumonia, and bacterial pneumonia, was identified from claims evidence of a primary diagnosis of pneumonia recorded from inpatient hospitalizations or outpatient visits. Rates of hospitalization for pneumonia, respiratory conditions, and overall hospitalizations within 14 days of the index date were also examined as hospital admissions may have been due to influenza-related complications.

Table 1.

ICD-9-CM Codes for Diagnosis of Pneumonia, Other Respiratory Conditions, or Otitis Media and Its Complications Within 14 Days of Onset of Influenza

Description	ICD-9-CM Code
Otitis media
Unspecified acute nonsuppurative otitis media	381.00
Acute serous otitis media	381.01
Acute mucoid otitis media	381.02
Acute sanguinoid otitis media	381.03
Acute nonsuppurative otitis media without spontaneous rupture of eardrum	381.4×
Acute nonsuppurative otitis media with spontaneous rupture of eardrum	382.00
Acute nonsuppurative otitis media in diseases classified elsewhere	382.01
Acute suppurative otitis media in diseases classified elsewhere	382.02
Unspecified suppurative otitis media	382.4×
Unspecified otitis media	382.9×
Unspecified eustachian salpingitis	381.50
Acute eustachian salpingitis	381.51
Unspecified obstruction of eustachian tube	381.60
Unspecified Eustachian tube disorder	381.9×
Acute mastoiditis without complications	383.00
Subperiosteal abscess of mastoid	383.01
Acute mastoiditis with other complications	383.02
Unspecified petrositis	383.20
Acute petrositis	383.21
Unspecified mastoiditis	383.9×
Unspecified acute myringitis	384.00
Bullous myringitis	384.01
Other acute myringitis without mention of otitis media	384.09
Unspecified perforation of tympanic membrane	384.20
Central perforation of tympanic membrane	384.21
Attic perforation of tympanic membrane	384.22
Other marginal perforation of tympanic membrane	384.23
Multiple perforations of tympanic membrane	384.24
Total perforation of tympanic membrane	384.25
Other respiratory conditions
Acute respiratory infections	460.×× – 466.××
Other disease of the upper respiratory tract	470.×× – 478.××
Chronic obstructive pulmonary disease and allied conditions	490.×× – 496.××
Pneumoconiosis and other lung disease due to external agents	500.×× – 508.××
Other diseases of the respiratory system	510.×× – 519.××
Pneumonia
Pneumonia, influenza	487.0×
Pneumonia, other viral	480.××
Pneumonia, bacterial	481.××, 482.××
Pneumonia, other and unspecified	483.××, 484.××, 485.×× or 486.××

ICD-9-CM = International Classification of Diseases, Ninth Revision, Clinical Modification

Healthcare expenditures. Healthcare utilization and expenditures for inpatient, outpatient medical, and outpatient pharmacy services were assessed for 30 days following the index date. For patients with capitated insurance coverage, expenditures were estimated from encounter records using fee-for-service equivalents defined by procedure code and region. Expenditures were adjusted for inflation using the Consumer Price Index Medical Component[24] and expressed in December 2005 US dollars.

Safety analysis. Events of interest for the safety analysis were defined by the presence of ICD-9 diagnosis codes in the following categories: unspecified adverse effect of drug, dermatologic and serious allergic reactions (angioedema, skin rash, toxic shock syndrome/Stevens-Johnson syndrome), musculoskeletal, central nervous system (CNS) and neuropsychiatric (psychiatric, encephalitis, consciousness, movement, seizure, stroke, other), cardiac (acute myocardial infarction, arrhythmia, congestive heart failure), renal, metabolic, gastrointestinal, and sudden death.

Statistical Analyses

Individuals in the treatment cohort were propensity matched to the pool of patients with no antiviral prescription (1:1). In observational studies, the assignment of participants to treatment and no antiviral groups is not random. Estimation of treatment effects may therefore be biased by confounding factors. Propensity score matching techniques can be utilized in analyses of such studies to ensure that matching of nonrandomized patient groups is as close as possible in order to minimize confounding variables. Formally, the propensity score for a patient is the probability of a patient being treated on the basis of covariates such as the patient's demographic and clinical characteristics. For example, if 2 patients, 1 in the treatment group and 1 in the no antiviral group, have the same (or a similar) propensity score, they can be considered as equivalent as if they had been randomly assigned to either group.

In this analysis, a logistic regression model was developed to predict the probability of a patient being treated with oseltamivir. The ‘nearest-neighbor’ method was used subsequently to select patients treated with oseltamivir and those who were not.[25] Matching was conducted separately by influenza season for those enrollees with a private pay commercial insurer and those enrollees with Medicare as the primary payer (5 seasons, 2 payers = 10 matched cohorts). The matching variables for the commercially insured patients were age (< 12, 13–18, 19–49, and ≥ 50 years), gender, regional location, plan benefit design, vaccination status, Charlson Comorbidity Index (CCI) score (to classify comorbidity for risk adjustment), chronic disease score, and baseline cost. The matching variables for the Medicare patients were age (< 65, 65–74, and ≥ 75 years), gender, regional location, chronic disease score, vaccination status, and baseline expenditure.

Comparisons of hospitalization frequencies were expressed in terms of relative risk (RR), while comparisons of healthcare utilization and expenditures were expressed in terms of differences in means and proportions. For all comparisons, 95% confidence intervals (CIs) were estimated. Differences between oseltamivir-treated and untreated cases were considered statistically significant (P < .05 in a 2-tail test) if the CI did not include 1.0 for relative risk or $0 for expenditures. For the safety analysis, comparison of frequency of events was expressed in terms of odds ratios (OR).

To adjust for differences in baseline characteristics between oseltamivir-treated and untreated cases, multivariate statistical models were estimated for each outcome measure. Cox proportional hazards regression was used to adjust the RR of any clinical outcome within 14 days for differences between treatment groups in influenza season, age category, gender, urban/rural location, chronic lung disease, cardiac disease, neurologic disease, immunocompromised status, pregnancy, influenza vaccination, CCI,[26] and baseline cost of all medical services. The final models included only baseline characteristics that were significant predictors of the outcome, identified by a stepwise process with a probability value of .05 for entering and staying in the model.

Expenditure differences were adjusted for the same set of baseline characteristics using a generalized linear model with an exponential link function. The adjusted incremental expenditure was obtained by using the model to generate a predicted difference for each member of the oseltamivir group and calculating the mean. Unadjusted expenditure differences were calculated for inpatient acute care hospitalizations, office and ER visits, total outpatient costs, and outpatient pharmacy costs.

In the safety analysis, logistic regression was used to calculate the odds of having an adverse event within 30 days, adjusting for differences between treatment groups in influenza season, age category, gender, urban/rural location, or a medical history of chronic lung disease, diabetes, renal disease, other respiratory conditions, or select conditions compromising immune status (cancer, HIV, transplant) in the 365 days prior to index.

Results

Clinical Outcomes Study

Patient characteristics. In total, 235,295 patients were diagnosed with influenza during 1 of the 5 seasons from 2000–2005 and had employer-sponsored primary or supplemental insurance coverage throughout the study period. Of these, 81,098 were excluded from the study as they failed the continuous enrollment requirement (39%), received other antiviral therapy (34%), had inadequate pharmacy data (15%), and/or had presented with pneumonia at the index date (16%). A further 2097 patients were excluded for having oseltamivir claims more than 1 day from the index date. There were 43,920 patients who met all the study criteria and had a claim for oseltamivir within 1 day of influenza diagnosis. These patients were propensity matched, in a 1:1 ratio, to patients who received no antiviral treatment, generating treatment and no antiviral groups each with 31,674 patients (Figure 1).

Figure 1.

Patient disposition for clinical outcomes and healthcare expenditures, and safety analyses.

Following propensity matching, there were no differences between the treatment and no antiviral group with regard to sex, age, population density, Medicare eligibility, or CCI; there was a significant difference between groups in region, history of congestive heart failure, chronic obstructive pulmonary disorder (COPD), diabetes mellitus, renal disease, pneumonia, or other respiratory disease (Table 2).

Table 2.

Demographic Characteristics of Influenza Cases for the Clinical Outcomes Study

	Oseltamivir (N=31,674)	No Antiviral (N=31,674)	P Value
Gender			.968
Male (%)	14283 (45.1)	14288 (45.1)
Female (%)	17391 (54.9)	17386 (54.9)
Mean age (SD)	30.50 (19.97)	30.45 (20.36)	.756
Age group (years), n (%)			< .0001
< 1	142 (0.4)	535 (1.7)
1–2	1303 (4.1)	1379 (4.4)
3–5	1820 (5.7)	1961 (6.2)
6–12	4649 (14.7)	4049 (12.8)
13–49	17128 (54.1)	17139 (54.1)
50–64	5485 (17.3)	5459 (17.2)
≥ 65	1147 (3.6)	1152 (3.6)
Geographic region, n (%)			.049
Northeast	1628 (5.1)	1745 (5.5)
North Central	8329 (26.3)	8132 (25.7)
South	19609 (61.9)	19751 (62.4)
West	2108 (6.7)	2046 (6.5)
Population density, n (%)			.625
Metropolitan	21811 (68.9)	21868 (69.0)
Rural	9863 (31.1)	9806 (31.0)
Medicare eligible, n (%)			1.000
Yes	1162 (3.7)	1162 (3.7)
No	30512 (96.3)	30512 (96.3)
Comorbidities, n (%)
Asthma	536 (1.7)	569 (1.8)	.317
Congestive heart failure	56 (0.2)	91 (0.3)	.004
COPD	115 (0.4)	184 (0.6)	.000
Cystic fibrosis	1 (0.0)	3 (0.0)	.317
Diabetes	490 (1.5)	558 (1.8)	.034
Emphysema	4 (0.0)	14 (0.0)	.018
Immunocompromised*	292 (0.9)	299 (0.9)	.772
Acute myocardial infarction	11 (0.0)	13 (0.0)	.683
Respiratory diagnoses	7594 (24.0)	8204 (25.9)	.000
Pneumonia	202 (0.6)	292 (0.9)	.000
Renal disease	26 (0.1)	53 (0.2)	.002
Stroke	21 (0.1)	23 (0.1)	.763
Prior therapy, n (%)
Respiratory therapy	2719 (8.6)	2831 (8.9)	.116
Oxygen therapy	88 (0.3)	124 (0.4)	.013
Influenza test on or before index date	7973 (25.2)	4308 (13.6)	.000
Vaccination before index date	1112 (3.5)	1153 (3.6)	.380
Charlson Comorbidity Index (SD)	0.13 (0.74)	0.13 (0.70)	.895

COPD = chronic obstructive pulmonary disease; SD = standard deviation

^*Includes cancer, HIV, and transplant patients.

Predictors of pneumonia and other outcome measures. Several factors within the model were found to be predictive of the risk for pneumonia, other respiratory complications, or hospitalizations at the P < .05 level. Chronic lung disease was associated with an increased risk for pneumonia, any respiratory disease, and hospitalization for either any respiratory disease or any reason. CCI score was also associated with an increased risk for pneumonia and otitis media, and an increased risk for hospitalization for pneumonia or for any reason. Higher baseline medical costs were associated with an increased risk for pneumonia, respiratory disease, otitis media, and hospitalization for pneumonia, respiratory disease, and for any reason.

Oseltamivir was associated with reduced risks for pneumonia, respiratory disease, otitis media, hospitalization for respiratory disease, and hospitalization for any reason.

Effect of oseltamivir use on risk for pneumonia and other respiratory outcomes – Total population. In the multivariate analysis, oseltamivir use was associated with statistically significant reductions in the risk of developing influenza complications during the 30-day follow-up period. The RR was reduced by 15% for diagnosis of pneumonia (RR 0.85, 95% CI: 0.73, 0.98), 20% for respiratory illnesses (RR 0.80, 95% CI: 0.76, 0.83), and 31% for otitis media and its complications (RR 0.69, 95% CI: 0.61, 0.79; Figure 2). Oseltamivir use was also associated with a 38% reduction in the risk for subsequent hospitalization for any reason (RR 0.62, 95% CI: 0.52, 0.74) and 57% for hospitalization due to respiratory illnesses (RR 0.43, 95% CI: 0.27, 0.69; Figure 2).

Figure 2.

Adjusted relative risks for the diagnosis of selected illness within 14 days of diagnosis of influenza in patients receiving oseltamivir vs patients who received no antiviral therapy – overall population.*

*Primary diagnoses for respiratory illness were categorized as follows: acute respiratory infections, other disease of the upper respiratory tract, COPD, and allied conditions, pneumoconiosis and related lung diseases, other diseases of the respiratory system, all respiratory diagnoses. Diagnoses of otitis media and its complications were categorized as follows: eustachian salpingitis, mastoiditis, myringitis, and any otitis media or complication.

Pediatric population (≤ 12 years). In children aged ≤ 12 years, oseltamivir treatment significantly reduced the adjusted risks for pneumonia by 53% (RR 0.47, 95% CI: 0.33, 0.66), respiratory illnesses by 28% (RR 0.72, 95% CI: 0.65, 0.80), and otitis media and its complications by 39% (RR 0.61, 95% CI: 0.54, 0.71; Figure 3). There were also significant reductions in the adjusted risk for hospitalization for respiratory illnesses (91% reduction [RR 0.09, 95% CI: 0.01, 0.70]) and hospitalization for any reason (50% reduction [RR 0.50, 95% CI: 0.31, 0.81]; Figure 4).

Figure 3.

Adjusted relative risks for diagnosis of pneumonia, respiratory illnesses, and otitis media and its complications in patients receiving oseltamivir vs patients who received no antiviral therapy, by age group.*

*Primary diagnoses for respiratory illness were categorized as follows: acute respiratory infections, other disease of the upper respiratory tract, chronic obstructive pulmonary disorder COPD, and allied conditions, pneumoconiosis and related lung diseases, other diseases of the respiratory system, all respiratory diagnoses. Diagnoses of otitis media and its complications were categorized as follows: eustachian salpingitis, mastoiditis, myringitis, and any otitis media or complication.

Figure 4.

Adjusted relative risks for hospitalization for pneumonia, respiratory illnesses, and/or any reason in patients receiving oseltamivir vs patients who received no antiviral therapy, by age group.*

*Primary diagnoses for respiratory illness were categorized as follows: acute respiratory infections, other disease of the upper respiratory tract, chronic obstructive pulmonary disorder COPD, and allied conditions, pneumoconiosis and related lung diseases, other diseases of the respiratory system, all respiratory diagnoses. Diagnoses of otitis media and its complications were categorized as follows: eustachian salpingitis, mastoiditis, myringitis, and any otitis media or complication.

In children aged 1–2 years, oseltamivir significantly reduced the adjusted risks for diagnosis of pneumonia by 52% (RR 0.48, 95% CI: 0.24, 0.99), respiratory illnesses by 31% (RR 0.69, 95% CI: 0.54, 0.87), and otitis media or its complications by 32% (RR 0.68, 95% CI: 0.52, 0.88; Figure 3). In children aged 3–5 years, oseltamivir significantly reduced the adjusted risk for diagnosis of respiratory illnesses by 26% (RR 0.74, 95% CI: 0.59, 0.92) and otitis media or its complications by 29% (RR 0.71, 95% CI: 0.53, 0.95; Figure 3). There was a trend toward a reduction in the adjusted risk for pneumonia, but this did not reach significance (RR 0.63, 95% CI: 0.34, 1.18). In children aged 6–12 years, oseltamivir significantly reduced the risk for diagnosis of pneumonia by 57% (RR 0.43, 95% CI: 0.26, 0.71), respiratory illnesses by 20% (RR 0.80, 95% CI: 0.75, 0.85), and otitis media or its complications by 29% (RR 0.71, 95% CI: 0.54, 0.95).

Adolescent and adult population (13–64 years). Oseltamivir treatment significantly reduced the adjusted risk for diagnosis of respiratory illnesses by 20% (RR 0.80, 95% CI: 0.75, 0.85) and of otitis media and its complications by 32% (RR 0.68, 95% CI: 0.54, 0.86) in patients aged 13–49 years (Figure 3), and also significantly decreased overall hospitalization by 44% (RR 0.56, 95% CI: 0.43, 0.74; Figure 4). Hospitalizations for respiratory illnesses were reduced by 51% (RR 0.49, 95% CI: 0.24, 1.00), but this was not significant (Figure 4).

In patients aged 50–64 years, oseltamivir treatment significantly reduced the adjusted risks for respiratory illness by 19% (RR 0.81, 95% CI: 0.73, 0.89; Figure 3) and hospitalization for respiratory illnesses by 64% (RR 0.36, 95% CI: 0.15, 0.84; Figure 4).

Elderly (aged > 65 Years) population. Oseltamivir did not affect the incidence of any assessed secondary complications (Figure 3) or hospitalization rates (Figure 4) in elderly patients. However, the number of elderly patients in each subgroup was low.

Healthcare expenditures. After multivariate adjustment, there was no significant difference in overall expenditures between the oseltamivir and the no antiviral groups ($533 vs $532; Diff $1, 95% CI: −$61, $63). Unadjusted healthcare expenditures for inpatient acute care hospital costs, office visits, ER visits, and total outpatient costs were slightly and significantly lower for the oseltamivir group compared with the no antiviral group (Table 3). However, unadjusted outpatient pharmacy costs were higher for the oseltamivir group compared with the no antiviral group.

Table 3.

Costs of Healthcare Services in Patients Prescribed Oseltamivir vs No Antiviral Within 30 Days of Diagnosis With Influenza

Expenditure (US $)	Oseltamivir N = 31,674		No Antiviral N = 31,674
	N (%)	Mean (SD)	N (%)	Mean (SD)	Unadjusted Difference (95% CI)
Inpatient acute care	279 (0.9)	85 (1710)	447 (1.4)	111 (1525)	−26 (−52, 1)
Office visits	29,524 (93.2)	84 (66)	28,754 (90.8)	87 (75)	−3 (−4, −2)
ER visits	3599 (11.4)	42 (205)	4577 (14.5)	56 (228)	−14 (−17, −10)
Total outpatient	31,477 (99.4)	273 (826)	31,429 (99.2)	346 (1176)	−74 (−90, −58)
Outpatient pharmacy	31,445 (99.3)	161 (200)	21,039 (66.4)	96 (225)	65 (62, 68)
Total healthcare	31,543 (99.6)	519 (1879)	31,487 (99.4)	554 (2020)	−35 (−65, −5)

Safety Analysis Study

Patient Characteristics

A total of 218,766 patients met the safety analysis criteria; of these, 41,938 patients had a claim for oseltamivir within 1 day of influenza diagnosis. One-to-one propensity matching of these patients with those who were prescribed no antiviral treatment generated 2 cohorts of 29,620 patients each (Figure 1). There were no demographic or clinical differences between the groups with regard to sex, age, region, payer population density, Medicare eligibility, or CCI; there was a significant difference between groups with regard to history of asthma, COPD, pneumonia, or other respiratory disease.

Safety Evaluations

Oseltamivir was associated with a lower likelihood of CNS and psychiatric (OR 0.80, 95% CI: 0.73, 0.88), cardiac (OR, 0.80, 95% CI: 0.69, 0.94), and gastrointestinal (OR, 0.81, 95% CI: 0.73, 0.91) events compared with the no antiviral group. There were no significant differences for any other events. Overall, there were 73 hospitalizations associated with these diagnoses, with little difference between oseltamivir-treated patients (0.1%) and patients without an antiviral prescription (0.2%).

Discussion

This retrospective cohort study utilizing health insurance claims in patients with influenza indicates that oseltamivir reduces the risk of influenza-related secondary complications and rates of hospitalization for non-pneumonia respiratory conditions. Reductions in the risk of developing influenza-related complications in adolescents and adults are consistent with previous reports,[6,17,18] as are the reductions in hospitalization rates observed in oseltamivir-treated patients.[6,18]

The effect of oseltamivir on influenza-related outcomes was particularly marked in the pediatric population, with reductions in the risk for diagnosis of pneumonia, respiratory illness, and otitis media and its complications, and of hospitalization for respiratory illness or hospitalization for any reason. Oseltamivir was highly effective in reducing the incidence of otitis media consistent with previous studies.[17,27] The finding that a higher percentage of children < 2 years of age developed otitis media compared with children > 2 years of age is also in line with previous reports.[28] Otitis media is the most common reason for administration of antibiotics in children,[29] so reductions in the risk of developing this complication may reduce the duration and severity of influenza and the administration of antibiotics. Neuraminidase inhibitors have been found to reduce antibiotic use in children in other studies.[27]

Prescribing oseltamivir did not reduce the adjusted overall healthcare costs. This can be explained in terms of the overall incidence of secondary complications. Given that the frequency of complications was low, the cost savings expected to be associated with a reduction in events in absolute terms would have been small. However, with the exception of outpatient pharmacy costs, all other unadjusted healthcare costs were lower for patients prescribed oseltamivir. The higher outpatient pharmacy costs in the oseltamivir group can mostly be attributed to the cost of prescribing oseltamivir. In a recent cost-effectiveness study for otitis media treatment, costs of a 5- to 10-day antibiotic prescription plus an office consultation were estimated at between US$44.03 and US$45.44[30] vs an approximate US$67 cost for a 5-day course of oseltamivir.[31]

The impact of influenza in children is considerable. Children have the highest use of inpatient and outpatient healthcare services, high levels of antibiotic use, and significant school absenteeism during influenza epidemics.[32] Rates of hospitalization due to influenza in very young children are high and equal to those observed in high-risk adults.[33,34] Children also have the highest rates of infection[32] and have a longer duration of viral shedding than adults.[28]

Both neuraminidase inhibitors, oseltamivir and zanamivir, have beneficial effects in children with influenza; oseltamivir has previously been shown to reduce the duration of influenza symptoms[35] as well as the duration of viral shedding.[13] Zanamivir has been shown to reduce the severity and duration of influenza-like illness in children aged 5–12 years.[24] However, while oseltamivir is currently licensed for use in children aged ≥ 1 year, zanamivir is only licensed for treatment of influenza in children ≥ 7 years of age, since the inhaled route of administration has limited clinical utility in young children. Oseltamivir was the subject of this study since this drug accounts for approximately 92% of the current market for antiviral influenza medication in the United States.[36]

Results in elderly patients are inconclusive as small sample size precluded any meaningful analyses. Difficulties in the diagnosis of influenza in elderly patients may partly explain the low sample size. Clinical symptoms in the elderly differ from those in other age groups, typically being more respiratory in nature without accompanying fever.[37] Furthermore, a higher proportion of elderly patients were excluded from this study due to presentation of pneumonia or admission to a healthcare institution on the index date compared with patients < 65 years of age. Elderly patients have a high rate of infection and are at increased risk for complications, hospitalization, and death.[4,38,39] Although vaccination is recommended in this population and vaccination rates are high,[40] vaccine effectiveness may be variable.[41,42] Effective antiviral treatment may therefore be required in cases of infection. Oseltamivir has previously been shown to be effective for influenza treatment in elderly patients.[43]

Oseltamivir was generally well tolerated in this study. The odds of experiencing a number of adverse events were lower in oseltamivir-treated patients compared with the no antiviral cohort.

This study confirms and extends previous findings. However, some limitations should be noted. Patient selection required claim-based evidence of influenza, as evidenced by an ICD-9-CM diagnosis code. In some cases the diagnosis may have been confirmed by virologic testing, but may have been based solely on the clinical judgment of the physician. The claims data were not verified through medical record review; thus, within this study it should be assumed that patients have either influenza or influenza-like illness. Additionally, the index date was based on the first diagnosis of influenza associated with a medical service and not the exposure to infection or onset of symptoms. Therefore, although study inclusion required prescription of oseltamivir within 1 day of a clinical diagnosis of influenza, oseltamivir may have been administered outside of the 48-hour time window within which it has been shown to be effective.[44] However, the inclusion of patients who may not have influenza and patients who may have received treatment outside of the 48-hour window would lead to a more conservative estimate of the effects of oseltamivir.

Oseltamivir therapy should be initiated within 48 hours of symptom onset. Some patients with an influenza diagnosis who were included within the no antiviral group may not have been prescribed oseltamivir because the period between the onset of symptoms and their visit to the physician was more than 48 hours. Although it is not possible to ascertain the patient numbers within this category from the database, this factor is unlikely to have an effect on the outcomes of this study.

Lack of randomized patient allocation to the oseltamivir and no antiviral groups could be viewed as a limitation in the design of the study. However, propensity score matching was used to reduce the risk of selection bias. Propensity scores represent a patient's probability of receiving a given treatment option, and this technique helps to reduce bias and improve precision in observational studies.[45]

Although multiple comparisons made within the study were not statistically tested, the number of outcomes was sufficiently small to allow for such comparisons. Finally, when assessing outcomes from observational research, it may not always be possible to elucidate clearly some information relevant to these populations, such as living status (ie, independent vs chronic care) and administration of therapy due to a lack of electronic health records. However, the findings of this claims database analysis are consistent with those observed in previous clinical trials.[6,12,17]

Conclusion

This study shows that treatment of influenza patients with oseltamivir significantly reduces the risk for influenza-related complications and hospitalization. These effects were particularly marked in the pediatric population. The results of this study underscore the need to widen the appropriate use of neuraminidase inhibitors for treatment of influenza in children, and highlight the need for further research in elderly patients.