This real-world observational study was conducted using the Medicare 5% national sample administrative database from January 2011 through December 2015. All patient identifiers in the database have been fully encrypted; therefore, neither institutional review board approval nor consent was necessary for this study, as it was conducted in the USA with depersonalized claims data and does not meet criteria for studies with human participants; it is therefore exempt from approval per the provision for unidentifiable personal data in the Federal Policy for the Protection of Human Subjects (1991).
The Medicare claims data files used for this study included inpatient and outpatient (Parts A and B), Medicare carrier, prescription (Part D) drug events, skilled nursing facility (SNF), home health agency (HHA), hospice, durable medical equipment (DME), and the Medicare denominator file, which contains demographic and enrollment information of Medicare beneficiaries.
Patients with a medical claim for RSV diagnosis (International Classification of Disease, Ninth Revision, Clinical Modification [ICD-9-CM] codes 079.6, 466.11, 480.01 and ICD-10-CM codes B97.4, J20.5, J12.1, J21.0) were identified between July 1, 2011 and June 30, 2015. The date of the first observed RSV diagnosis during this period was designated as the index date.
Patients were required to be aged 18 years or older at the time of diagnosis, with continuous Medicare medical and pharmacy benefits for at least 180 days prior to the index date (baseline period) and at least 180 days after the index date (follow-up period). Patients who died during the follow-up period were also included in the study. Patients were excluded if they had an RSV diagnosis during the baseline period or an influenza or human metapneumovirus (hMPV) diagnosis during the study period. Patients were categorized as hospitalized if they were diagnosed with RSV during hospitalization or hospitalized within 1 day of RSV diagnosis. The hospital admission date was captured as the start of the index hospitalization.
Among patients who were hospitalized, participants were further categorized as high-risk and non-high-risk patients. High-risk patients were identified if they met any of the following criteria: diagnosis of chronic lung disease (including asthma and COPD), prior pneumonia, congestive heart failure (CHF), or immune compromise on or within 180 days before the date of RSV diagnosis. A patient was considered to be immunocompromised if they had evidence of SOT, HSCT, or hematological malignancies (leukemia, lymphoma, and plasma cell neoplasms). All conditions were identified using ICD-9-CM and ICD-10-CM codes. The non-high-risk patients were the remaining hospitalized patients in the study population.
Patients were categorized as outpatients if they were diagnosed with RSV in the outpatient setting and not hospitalized within 1 day of diagnosis. Patients diagnosed in the outpatient setting were also categorized into high-risk and non-high-risk cohorts.
Baseline patient demographics including age, sex, US geographic region, comorbidities, and setting of diagnosis during the 180 days prior to the index date (baseline period) were assessed for all patients. The diagnosis setting included ER (combined ER and/or inpatient setting), physician’s office, or other Medicare settings (DME, HHA, hospice, SNF). Among outpatients, patients were further classified as later hospitalized (diagnosed in the outpatient setting and hospitalized at least 2 days from index diagnosis) or never hospitalized. Comorbidities including HSCT, SOT, hematological malignancies, stroke, osteoporosis, anxiety, asthma, cancer, depression, osteoarthritis, chronic kidney disease (CKD), coronary artery disease (CAD), CHF, diabetes, high cholesterol, COPD, hypertension, and previous evidence of pneumonia were identified using ICD-9-CM or ICD-10-CM codes during the baseline period (6 months prior to the index date).
Complications, all-cause mortality, death during inpatient stay, and 30-day readmission rates during the follow-up period were evaluated. Complications of interest—respiratory failure, chest pain, hypoxia, cardiac arrhythmia, cough, myocardial infarction, dyspnea, pneumonia, lower and upper respiratory tract infection (excluding influenza, RSV, and hMPV)—were identified using ICD-9-CM and ICD-10-CM codes. The proportion of patients who were discharged to an SNF as well as LOS in an SNF were also evaluated.
All-cause healthcare utilization and costs during the pre- (baseline) and post-index (follow-up) periods were analyzed, including hospital LOS, number of office visits, pharmacy use, ER, inpatient (and hospital LOS across all hospitalization), and outpatient visits. Costs were adjusted to 2015 US dollars using the medical care component of the Consumer Price Index (CPI).
All baseline and outcome variables were analyzed descriptively. Percentages and counts were provided for categorical variables. Means and standard deviations (SDs) were computed for continuous variables. Healthcare costs and utilizations were analyzed descriptively and compared between 180 days pre- and post-RSV diagnosis. Bivariate comparison between pre- and post-index date periods were made using a paired t test to evaluate the statistical significance for continuous variables, including average number of resource utilization and costs.
Multivariate logistic regression was conducted to examine potential predictors of hospitalization among hospitalized patients vs those who were never hospitalized. Purposeful model selection was used to identify variables to be included in the logistic regression. First, all independent variables with p < 0.25 in bivariate testing were initially included in the model. Secondly, variables with p > 0.10 were dropped from the model sequentially unless they were identified as confounders (i.e., variables that, when dropped from the model, resulted in at least a 20% change in parameter estimates for 1 or more of the other variables, when compared to the original model). The final model included confounders, variables that had p ≤ 0.10, and high-risk conditions. Covariates in the final model included demographics, comorbidities, previous evidence of pneumonia, number of conditions, and the number of inpatient and ER visits during the baseline period.
Statistical analyses were conducted using the Statistical Analysis System (SAS) v.9.3. (Cary, North Carolina, USA). The threshold for p value significance was set at α-level 0.05 for pre–post analysis and logistic regression.