Number of acute respiratory hospitalizations by week and discharge diagnosis among Medicaid enrollees with chronic lung disease for 4 successive years: 1995 to 1996, 1996 to 1997, 1997 to 1998, and 1998 to 1999, with week 1 beginning in August and week 52 ending July 31. Brackets indicate the weeks when respiratory syncytial virus (RSV) and influenza virus were circulating in the region of middle Tennessee. The annual peak weeks for RSV and influenza viral isolates are indicated by asterisks.
Adjusted age-specific rates, per 1000 persons with chronic lung disease, of acute respiratory hospitalizations (A), outpatient visits (C), and antibiotic prescriptions (E) by summer, respiratory syncytial virus (RSV), and influenza seasons and excess of acute respiratory hospitalizations (B), outpatient visits (D), and antibiotic prescriptions (F) attributable to influenza virus and RSV, Tennessee Medicaid program August 1995 to July 1999. Rates were adjusted for year, race, sex, and residence. Calculation of excess events included terms for rate differences between seasons and duration of virus circulation.
Griffin MR, Coffey CS, Neuzil KM, Mitchel EF, Wright PF, Edwards KM. Winter VirusesInfluenza- and Respiratory Syncytial Virus–Related Morbidity in Chronic Lung Disease. Arch Intern Med. 2002;162(11):1229-1236. doi:10.1001/archinte.162.11.1229
Chronic lung disease predisposes to serious consequences of respiratory viruses. While increasing influenza immunization rates in older adults signals an awareness of the impact of influenza, children with asthma are infrequently immunized. While respiratory syncytial virus (RSV) is recognized as an important target of vaccine development for infants, its impact on adults is underappreciated.
We performed a retrospective cohort study to estimate rates of hospitalizations, deaths, outpatient visits, and antibiotic courses due to influenza and RSV in persons with chronic lung disease in the Tennessee Medicaid program from 1995 to 1999. Differences between study event rates when influenza was cocirculating with RSV and event rates when RSV was circulating alone were used to calculate influenza-attributable morbidity. Differences in rates when RSV was circulating alone and during summer months were calculated to assess the effect of RSV.
Influenza- and RSV-associated hospitalizations were highest at the extremes of age. There were an estimated 8 and 23 hospitalizations per 1000 children younger than 5 years annually due to influenza and RSV, respectively. There were 23 and 18 hospitalizations, as well as 2 and 5 deaths per 1000 persons 65 years or older annually due to influenza and RSV, respectively. Both viruses were associated with an excess of outpatient visits in children, and antibiotic prescriptions in all age groups.
Among persons with chronic lung disease, influenza virus and RSV accounted for 15% to 33% of acute respiratory hospitalizations in children, 7% to 9% of such hospitalizations in adults, and 9% of deaths in those 65 years or older.
SEASONAL PATTERNS of viruses that cause frequent respiratory tract infections are distinct. Rhinoviruses, adenoviruses, and parainfluenza viruses occur throughout the year, with periodic increases in frequency. Influenza virus and respiratory syncytial virus (RSV), on the other hand, are usually confined to winter.1 Historically, cold temperature has been linked to health events, and winter increases in acute respiratory disease have been shown to presage increases in overall mortality.2 The recognition of the association between influenza epidemics and deaths over a century ago was facilitated by the explosive nature of influenza epidemics. Although RSV is recognized to cause seasonal increases in respiratory hospitalizations in children,3- 5 its influence in adults is underappreciated. Emerging evidence suggests that both influenza and RSV cause substantial illness at both ends of the age spectrum during most winters.1,2,6,7
Persons with chronic lung disease contribute substantially to the overall burden of acute respiratory hospitalizations.8,9 Such persons are at increased risk for serious morbidity related to influenza infection, yet influenza immunization rates remain low in younger persons with high-risk conditions, at 38% in those aged 50 to 64 years and less than 30% in children and young adults.10 The impact that a safe and effective vaccine against RSV would have on children's health is now well recognized; however, there are few data on the potential benefits in adult high-risk populations. Although RSV has been convincingly associated with exacerbations of chronic lung disease in adults,11- 13 there are few data on the magnitude of this effect.
To assess the impact of influenza and RSV across the age spectrum, we performed a retrospective cohort study of persons of all ages with chronic lung disease enrolled in the Tennessee Medicaid program to determine rates of acute cardiopulmonary hospitalizations, deaths, outpatient visits, and antibiotic courses during 4 consecutive years, 1995 to 1999. We estimated rates of these events due to influenza and RSV using active viral surveillance to determine times when these viruses were circulating in Tennessee.
The Medicaid program has computerized files that permitted cohort assembly and identification of medical events of interest. These files include the enrollment file (a central registry of all enrollees), the pharmacy file (records of prescriptions filled at the pharmacy), the inpatient file (records of hospitalizations [including Medicare data for those jointly enrolled in this program]), and the outpatient file (encounter records for emergency department, hospital outpatient departments, and physician visits). In addition, Medicaid files have been linked to state death certificates to determine date of death. Annual differences between event rates when influenza virus was cocirculating with RSV and event rates when RSV was circulating alone were used to calculate influenza-attributable morbidity and mortality. Differences in event rates when RSV was circulating alone and event rates during summer months were calculated to assess the effect of RSV.
Vanderbilt University and the State of Tennessee Department of Health institutional review boards reviewed and approved this study.
Noninstitutionalized persons enrolled in the Tennessee Medicaid program from birth or for at least 1 year from August 1, 1995, through July 31, 1999, were the base population. Those of black or white race contributed nearly 4 million person-years; those in other racial/ethnic groups were few and therefore excluded. Persons with chronic lung disease (about 16% of base population) were identified for this study. Persons were defined as having chronic lung disease if they had at least 1 hospitalization or emergency department visit within the past year for chronic lung disease or at least 2 prescriptions in the past year for medications used to treat chronic lung disease. Hospital visits with a discharge diagnosis of cystic fibrosis (International Classification of Diseases, Ninth Revision, Clinical Modification [ICD-9-CM], 277 or 277.0), chronic respiratory disease of the newborn (770.7), or chronic obstructive pulmonary disease and allied conditions (490.xx-496.xx), including asthma (493.xx), were included. Medications included β-agonists, ipratropium bromide, theophylline, inhaled corticosteroids, cromolyn sodium, montelukast sodium, and zafirlukast. Persons being treated for human immunodeficiency virus, cancer, chronic renal failure, or liver failure were excluded from this analysis. Under a special program initiated in Tennessee in 1994, Medicaid was expanded to include low-income persons who were uninsured as well as persons who were unable to obtain insurance because of preexisting medical conditions. The study population included those enrolled in Medicaid under traditional criteria including Aid to Families with Dependent Children, Blind and Disabled, and Aged, as well as the new group of uninsured.
Study outcomes were (1) acute respiratory hospitalizations with an ICD-9-CM coded discharge diagnosis of pneumonia (480.xx-486.xx), influenza (487.xx), acute respiratory conditions (460.xx-466.xx), other respiratory conditions (490.xx-519.xx), as well as heart failure or myocarditis (422.xx, 427.xx, and 428.xx); (2) all-cause mortality (as determined by the Medicaid enrollment file or death certificates); (3) all outpatient visits regardless of diagnosis; and (4) filled outpatient antibiotic prescriptions.
Two sources of viral isolation were used to define seasons. About 150 children, 5 years or younger, participated in a vaccine clinic at Vanderbilt University during each of the 4 study years. Viral cultures for RSV, influenza, and parainfluenza were obtained when these children developed a fever or respiratory symptoms.5 We also identified all positive tests for these viruses from the Vanderbilt Hospital Virology Laboratory, the majority of which were performed in children.
The start of influenza season was defined as the first day of the first of 2 consecutive weeks with at least 2 influenza isolates per week; the end of the season was the last day of the second of 2 consecutive weeks with at least 2 influenza isolates per week. Respiratory syncytial virus season was similarly defined, but excluded time classified as influenza season. During the 4 study years, influenza seasons were always encompassed by RSV seasons (Figure 1). Thus, all person-time and events in the study occurred in 1 of 3 exposure categories: RSV (during which only RSV circulated), influenza (during which RSV and influenza cocirculated), and summer (all time not in RSV or influenza seasons).
We ascertained these additional characteristics from computerized Medicaid files: age group (<5, 5-14, 15-49, 50-64, and ≥65 years), sex, residence (urban [Tennessee's 4 largest cities], other standard metropolitan statistical area, or rural), and race (black or white).
Persons entered the study on the first day after July 31, 1995, that they met entry criteria, and were followed up until loss of enrollment, death, or July 31, 1999. Crude age-specific rates of hospitalization due to acute cardiopulmonary events were calculated by dividing the number of these hospitalizations in the age group by the age-specific person-years during influenza, RSV, and summer seasons, and expressed per 1000 person-years of follow-up for persons with chronic lung disease. Differences between these crude rates during influenza and RSV seasons were calculated as measures of influenza-attributable risk. Differences between rates during RSV and summer seasons were calculated as a measure of risk attributable to RSV. Standardized rate differences were calculated as the weighted average of the strata-specific differences in rates, where the weights were defined as the corresponding strata-specific person-years in all seasons combined.
The excess number of hospitalizations due to influenza per 1000 persons with chronic lung disease was estimated separately in all strata by multiplying the strata-specific differences in rates for influenza and RSV seasons by the proportion of the corresponding study year classified as influenza season. The excess number of hospitalizations due to RSV per 1000 persons with chronic lung disease was computed by multiplying the strata-specific differences in rates for RSV and summer seasons by the proportion of the study years classified as either RSV or influenza season (RSV circulated during both these seasons). Standardized estimates were calculated with a weighted average of strata-specific values, as was done for the standardized differences in rates. Confidence intervals were computed based on large sample properties.14 Similar analyses were performed for deaths (expressed per 10 000 person-years), outpatient visits, and antibiotic courses. Estimates of virus-specific hospitalization and mortality rates were divided by total event rates (Table 1) to determine the proportion of hospitalizations and deaths due to theses viruses.
The 2 viral surveillance systems detected 1105 RSV-positive cultures, 253 influenza-positive tests (antigen or culture), and 77 positive parainfluenza type 1, 2, or 3 cultures during the 4 study years. The relative frequency of positive tests does not reflect the actual level of viral activity since RSV tests were done more commonly. The onset of influenza season ranged from late November to late January with an average duration of 12 weeks (Figure 1). The onset of RSV season ranged from late September to early November and averaged 26 weeks in duration. The 1995 to 1996 season was exceptionally long due to cocirculation of 2 influenza A strains and a late influenza B outbreak. During the first 2 years, peak influenza activity closely followed peak RSV activity, as indicated in Figure 1; whereas, the 2 peaks were more widely separated in the subsequent 2 years. There were few influenza isolates outside the defined influenza season. By definition, RSV activity occurred during both RSV and influenza seasons with means of 9.6, 11.2, and 0.3 positive tests per week in RSV, influenza, and summer seasons, respectively. Parainfluenza isolates were distributed throughout the year but were on average higher in summer (0.5 per week) than during RSV (0.3 per week) or influenza (0.2 per week) seasons.
Persons with asthma and other chronic pulmonary diseases contributed 625 509 person-years of follow-up and are the subject of this study. Because Medicaid preferentially enrolls children and persons 65 years or older, 36% of the study population were younger than 15 years, and 13% were 65 years or older (Table 1). The higher prevalence of males among children reflects the increased prevalence of asthma in boys, whereas the predominance of women in the older age groups reflects the higher enrollment of women in the Medicaid program. Thirty-one percent of the population resided in urban areas, 25% in other standard metropolitan statistical areas, and 44% in more rural locations, though this differed by age group.
Total acute respiratory hospitalization rates were lowest in children aged 5 to 14 years, were similar in those younger than 5 years and those 15 to 49 years, and increased markedly in older age groups (Table 1). Whereas hospitalizations for pneumonia and acute respiratory conditions, including bronchiolitis, predominated in children, chronic lung disease and congestive heart failure constituted a greater proportion of admissions in older age groups. Death rates increased with age. Patients of all ages had a high frequency of both outpatient visits and antibiotic prescriptions filled.
In all 4 years, acute respiratory hospitalizations were higher in winter than summer weeks and peaked during influenza seasons (Figure 1). The distribution of hospitalizations was similar in all age groups and except for those aged 5 to 14 years, rates were highest in influenza season, when by definition, influenza and RSV cocirculated (Figure 2 A). The excess hospitalization rates (Figure 2 B) account for the fact that influenza seasons averaged 12 weeks, whereas RSV circulated an average of 26 weeks. There were an estimated 8, 0, 3, 13, and 23 excess hospitalizations due to influenza per 1000 persons with chronic lung disease per year in the age groups younger than 5, 5 to 14, 15 to 49, 50 to 64, and 65 years or older, respectively. In addition, there were an estimated 23, 5, 3, 11, and 18 excess hospitalizations due to RSV per 1000 persons with chronic lung disease in these 5 age groups, respectively. For children younger than 5 years, RSV accounted for about 3 times the number of hospitalizations as influenza. Thirty-three percent of all acute respiratory hospitalizations were due to one of these viruses (Table 2). We detected no excess hospitalizations due to influenza in those aged 5 to 14 years, but RSV accounted for 15% of all such hospitalizations. In older age groups, influenza and RSV accounted for a remarkably similar estimated number of hospitalizations. Both influenza and RSV individually accounted for 3% to 5% of all cardiopulmonary hospitalizations in those 15 years or older (Table 2).
Among those with chronic lung disease 65 years or older, there were 23.8 (95% confidence interval, 10.1-37.5) and 46.5 (95% confidence interval, 23.7-69.3) deaths due to influenza and RSV, respectively, per 10 000 persons (Table 3 note deaths expressed per 10 000 not per 1000). During the study years, the excess in deaths during the winter respiratory virus seasons represented 9% of deaths from all causes in this age group. There was also an estimated 15.3 deaths per 10 000 among those aged 50 to 64 years from RSV. There was insufficient power to detect excess deaths of 1 per 10 000 or lower in the other age groups.
Seasonal variation in outpatient visits was more evident in children than in adults (Figure 2 C). A significant excess in outpatient visits due to influenza was observed only in those younger than 65 years and ranged from 66 to 173 per 1000 persons with chronic lung disease per year, with the highest excess in those younger than 5 years. Significant excesses in outpatient visits attributable to RSV were observed only in those younger than 15 years, and were 515 and 401 per 1000 persons with chronic lung disease per year in those younger than 5 years and 5 to14 years, respectively (Figure 2 D).
There was a consistent winter excess of antibiotic prescriptions in all age groups (Figure 2 E). The estimated increase in antibiotic prescriptions attributed to influenza ranged from 64 to 108 prescriptions per 1000 persons with chronic lung disease (Figure 2 F). Excess antibiotic prescriptions attributable to RSV were higher, ranging from 190 per 1000 in those 65 years or older to 641 per 1000 in those younger than 5 years.
In this comprehensive examination of winter virus-related illness among persons with chronic lung disease, both influenza and RSV were associated with a high burden of disease in all age groups. Respiratory syncytial virus, recognized to be the leading cause of lower respiratory tract infection in infants and young children,15 was associated with a greater burden of hospitalizations in children younger than 15 years than was influenza. However, in adults, the burden of hospitalizations due to influenza and RSV was remarkably similar, consistent with other recent reports.2,8,16 In addition, both viruses were associated with relatively high mortality in older adults, and with considerable outpatient morbidity in all age groups. Influenza morbidity and mortality reported here does not take into account influenza vaccination, and thus underestimates events that would occur in the absence of immunization. In 1999, 65.5% of persons in Tennessee 65 years or older reported influenza immunization17; rates were likely lower in younger groups, including those with high-risk conditions.
Approximately 98% of infants are infected with RSV in their first year of life,18 resulting in hospitalization rates of 3% to 4%.3,4 In their next 2 years of life, the majority of children have less severe symptomatic reinfections, as do 40% of older children and adults who are exposed to sick family members.18 Our estimate that RSV causes 25% of all acute respiratory hospitalizations in children younger than 5 years with chronic lung disease is consistent with a prior estimate that RSV causes 23% of lower respiratory tract disease in children younger than 5 years.18
Our study provides the first evidence from a large population-based group of adults with chronic lung disease that influenza and RSV account for similar rates of excess acute respiratory hospitalizations. In addition, in this population, RSV was associated with substantial mortality in adults. Although RSV epidemics with associated morbidity and mortality have been documented in nursing homes,19,20 RSV in noninstitutionalized elderly persons is less well studied. A review of 8 hospital-based studies estimated that RSV is responsible for 2% to 9% of all pneumonia admissions among elderly persons, a frequency similar to influenza,16 and also consistent with our results. There are several reasons why RSV may be underappreciated in adults. Because lower quantities of virus are shed at time of illness, RSV is harder to detect in adults than children. Unlike bronchiolitis in children, there is no specific syndrome that clinically distinguishes RSV infection in adults. Finally, the long RSV season and the imposition of influenza epidemics on this season makes the contribution of RSV more difficult to visualize (Figure 1).
Influenza accounted for excess hospitalizations in all age groups except the 5 to 14 years group. Thus, in children with lung disease, as in healthy children, the greatest burden of hospitalizations due to influenza appears to occur in the very young.21,22 Reported hospitalization rates due to influenza in persons 65 years and older range from 1 to more than 10 per 1000.23,24 Thus, our rates of 3 per 1000 in younger adults and 13 to 23 per 1000 in older adults with medically treated chronic lung disease appear to be reasonable estimates.
Both influenza and RSV were associated with a substantial number of outpatient visits in children, but not adults. We measured all outpatient visits, regardless of associated diagnosis because of uncertainty about the quality of diagnostic information. Since chronic lung disease results in frequent outpatient visits (Table 1), it is possible that acute illness caused cancellations of regularly scheduled visits as well as an increase in sick visits resulting in no net increase in visits among older adults. Respiratory syncytial virus was associated with a greater excess of outpatient visits in children than influenza, consistent with the prominence of this pathogen in association with asthma exacerbations in young children.25 In all age groups, there was a substantial excess of antibiotic prescriptions (Figure 2), which may be a more specific indicator of excess illness due to viral respiratory infections than outpatient visits. Respiratory syncytial virus and influenza are associated with otitis media in children,5 and with exacerbations of asthma and chronic pulmonary disease in children and adults,26- 29 all of which may result in antibiotic treatment. Our estimates indicate that RSV and influenza infections accounted for 20% to 24% of antibiotic use in children and 12% to 14% of use in adults with chronic lung disease.
Our definition of chronic lung disease was a pragmatic one that included persons who were hospitalized in the past year with a discharge diagnosis of chronic lung disease and persons who received at least 2 prescriptions for medications used primarily for asthma and chronic obstructive pulmonary disease. This definition constituted about 16% of noninstitutionalized persons enrolled in the Tennessee Medicaid program and included a wide spectrum of patients with chronic lung disease. All patients in this study were enrolled in Medicaid and thus may be poorer and sicker than other populations. However, previous estimates of influenza-attributable disease in this population were similar to those of persons enrolled in health maintenance organizations.30 In addition, during the study period, about 25% of persons in Tennessee were enrolled in this program.
We assumed study events that occurred during the 6 months when influenza and RSV were circulating in excess of events that occurred in the remainder of the year were due to these 2 viruses. This assumes that other viruses were distributed equally during these two 6-month time periods. Viral surveillance data indicated that parainfluenza occurred more frequently in summer months during the study years. This would have the effect of our underestimating the winter excess. Uneven distribution of other respiratory viruses such as coronaviruses and rhinoviruses, which are also associated with exacerbations of asthma and chronic lung disease, but were not part of our viral surveillance, would also affect the accuracy of our estimates. However, other respiratory viruses tend to be distributed throughout the year, and likely account for a substantial portion of exacerbations not due to RSV or influenza.11 It is unknown whether other winter factors, such as temperature, humidity, or time spent indoors increase respiratory illness independent of their role in the spread of respiratory viruses. To the degree that there are other independent "winter" factors that cause respiratory illness, our figures would overestimate viral-associated disease. Our definition of seasons relied on viral surveillance from the middle Tennessee region, primarily in children. Thus, these definitions were rough estimates of when these viruses were circulating in the state. Misclassification of seasons likely resulted in underestimation of the effect of these 2 viruses.
Morbidity from influenza and RSV will vary by year and for influenza with the number of persons vaccinated and the vaccine effectiveness. Our estimates of the burden of illness due to winter viruses are consistent with many other smaller and primarily hospital-based investigations. Among persons with chronic lung disease, we estimate these viruses are responsible for 15% to 33% of acute respiratory hospitalizations in children, 7% to 9% of such hospitalizations in adults, and 8% to 9% of total adult deaths. In addition, infections with these viruses cause a substantial increase in outpatient visits and antibiotic courses. These data reinforce the importance of influenza immunization in persons with pulmonary disease of all ages. In addition, the data suggest that vaccines against RSV and other strategies to decrease morbidity associated with RSV will be important for adults as well as children.
Accepted for publication October 2, 2001.
This study was supported in part by an unrestricted educational grant from GlaxoSmithKline, Research and Development, Greenford, England, and a cooperative agreement with the Centers for Disease Control and Prevention, Atlanta, Ga (UR6 CCU417579).
We would like to thank Yuwei Zhu for analysis of viral isolation data and Cindy Naron for editorial assistance.
Corresponding author and reprints: Marie R. Griffin, MD, MPH, Vanderbilt University School of Medicine, Department of Preventive Medicine, A-1124 MCN, Nashville, TN 37232 (e-mail: email@example.com).