INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Chronic obstructive pulmonary disease (COPD) is the fourth leading cause of death in the United States (1). The cost of care for COPD patients is estimated to be $24 billion/yr (2). Although cigarette smoking is declining as a major risk factor, certain groups are showing an increase in tobacco use and associated increases in COPD case rates (3). Exacerbations of COPD are thought to be due to viral, bacterial, or mixed viral-bacterial infections of the respiratory tract. Since many COPD patients are older, it has been unclear how much age-related host factors contribute to illness frequency and severity. Moreover, age-related declines in host defenses have been argued to be important for the severity of respiratory tract viral infections (RTVIs) in the elderly (4, 5).

An association of RTVIs with exacerbations of COPD has been documented (6). RTVIs have been reported to contribute to acute as well as to chronic deterioration of pulmonary function in COPD patients (11). Although the viruses causing acute respiratory illnesses in these patients have been similar to those recovered from otherwise healthy adults, several studies have suggested increased illness rates and/or increased susceptibility of COPD patients to specific viruses (15). Many investigators have commented on the apparent increase in illness severity in COPD patients with RTVIs, but did not define the morbidity or impact of this increase on medical care utilization. The utilization of outpatient medical care services, physician services, and hospitalization by acutely ill COPD patients has been underappreciated. To extend understanding of the role of RTVIs in older adults with and without COPD, we performed a longitudinal cohort study.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

A cohort of patients with a history of COPD and a group of control subjects were recruited for this longitudinal study of RTVIs. The criteria for COPD were as defined by the American Thoracic Society (18). Subjects were placed in groups with and without respiratory dysfunction on the basis of a comparison of their highest measured FEV₁ during their first two scheduled visits (19). Control subjects had no history of chronic respiratory disease. The subjects were recruited through local private physicians, an urban public hospital (Ben Taub General Hospital) pulmonary clinic, and the Huffington Center Registry of Aged (Baylor College of Medicine). Enrollment began on September 5, 1991 and continued until April 26, 1994. The last follow-up date was May 10, 1995. Subjects were considered evaluable if they made at least two study visits and were followed for at least 1 mo. Written informed consent was obtained from each subject before study enrollment. The study was approved by the Baylor College of Medicine Institutional Review Board.

Initial evaluation included a medical history, focused physical examination, and spirometry. A baseline serum specimen was collected. Follow-up visits were scheduled in September, December, and April of each year. At these visits, a brief medical history was obtained, interim physical examination and spirometric studies were performed, and blood was drawn for future serologic studies. Each subject was contacted by telephone every 2 wk by a study nurse (M.A.) who asked about recent new respiratory symptoms. Each subject was asked to notify the study nurse when signs or symptoms of a respiratory tract illness began, and a visit was scheduled to obtain a more complete history, conduct a physical examination, and collect an acute serum specimen and culture specimens for respiratory viruses. Each subject was offered commercial influenza vaccine every fall throughout the study, and vaccine use was recorded.

Illness Definitions

Upper respiratory tract illness (URTI) was defined by the presence of symptoms of rhinitis and/or pharyngitis (i.e., sneezing, rhinorrhea, or sore throat). Lower respiratory tract illness was defined by increased cough, shortness of breath, or sputum production and/or a change in sputum color. Each illness was defined as febrile (temperature > 37.7° C orally) or afebrile.

An illness was considered to be virus-associated if its onset occurred within 7 d of virus isolation or if the illness was the only illness occurring in the interval between two consecutive serum samples of which the second showed a significant rise in antibody titer. When two or more illnesses occurred in an interval in which an antibody titer rose, virus infection (n = 8) was considered to be the first illness occurring during that interval. The results and conclusions would not have been changed substantially if the other illness had been selected as the first illness.

Virus Cultures and Serology

Respiratory tract specimens for virus culture were obtained at each follow-up visit as well as during acute illness. Serum for antibody studies was collected at each follow-up visit and at the time of acute respiratory illness (acutely and 2 to 4 wk later, during convalescence) and was stored at 20° C until used for serologic assays. Specimens for virus cultures consisted of nasal secretions and throat swabs.

Viral Cultures

Specimens were inoculated (usually within 12 h) onto the following four different cell culture lines: (1) primary human diploid lung fibroblasts (WI-38); (2) Madin-Darby canine kidney (MDCK) cells; (3) human epidermoid carcinoma cells (HEp-2); and (4) monkey kidney (LLC-MK2) cells (BioWhittaker, Walkersville, MD). Standard detection and identification methods for viruses were used (20). Rhinoviruses were distinguished from enteroviruses through acid lability or reverse transcription-polymerase chain reaction (RT-PCR) (21), and some influenza A viruses were subtyped through RT-PCR (22). Herpes simplex virus isolates were not considered pathogens in this study, and are not reported in the data analysis.

Viral Serology

Microneutralization tests were used to measure antibody levels to influenza virus types A and B, parainfluenza virus types 1, 2, and 3, coronavirus type 229E, and respiratory syncytial virus (RSV), using previously described techniques (23). An enzyme-linked immunosorbent assay (ELISA) for coronavirus OC43 antibody was used as described previously (26). Because of the large number of tests performed and the performance characteristics of the antibody assay, a sixfold rise in antibody titer was used to identify parainfluenza virus infections. In the case of other viral antigens, infections were identified as described previously (25).

Statistical Methods

Discrete variables were compared using the chi-square test or Fisher's exact test. Parametric data were analyzed with Student's t test. The Mann-Whitney U test was used for nonparametric data. The protective efficacy of influenza vaccine was calculated as follows: vaccine efficacy = (1  odds ratio [OR] of having been vaccinated) × 100, with logistic regression analysis used to determine the OR and its 95% confidence interval (CI). Statistical analysis was done with SPSS for Windows software (SPSS, Inc., Chicago, IL).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Demographics

There were 55 evaluable control subjects and 62 evaluable COPD subjects (Table 1). Thirteen additional control and 16 additional COPD subjects were enrolled but were not evaluable and were not used in further analysis. The mean duration of follow-up was 35.1 mo for the control subjects and 26.3 mo for the COPD subjects (range: 1 to 43.7 mo). The numbers of control and COPD subjects followed during each month of the study were 44 ± 8 and 37 ± 6 (mean ± SD), respectively.

Three COPD subjects were followed for less than one complete fall/winter season. Significantly more control subjects than COPD subjects were followed for two or more winter seasons (89% versus 64%, respectively). The mean age, percent males and percent non-white subjects were similar for the two groups. Other demographic variables were similar in the two groups, although the mean educational level was higher for the control subjects (14.3 ± 2.2 yr versus 12.9 ± 3.3 yr). More COPD subjects than controls gave a history of cigarette smoking as well as current cigarette use. Approximately half of the subjects in each group gave a history of coronary artery disease, hypertension, or diabetes mellitus. A diagnosis of congestive heart failure was reported in no control subjects and in 5% of the COPD subjects. There was no significant difference in the COPD and control groups in the mean number of household members living with the subject (1.9 versus 2.2) or in the number of households with at least one child. Less than half of the COPD subjects had received pneumococcal vaccine at study enrollment. There were no significant differences in demographics of the COPD subjects with mild obstruction (30 subjects) and those with moderate/severe obstruction (32 subjects).

More subjects in the COPD group than in the control group discontinued the study prematurely (29 of 62 versus 10 of 55, p = 0.002, chi-square analysis). However, reasons for early study termination were similar in the two groups. The most common reason for discontinuation was subject request (COPD group: n = 14; control group: n = 5), usually because of intensity of study procedures, number of study visits, or problems with travel. Other reasons for study discontinuation included death of the subject (COPD group: n = 8; control group: n = 4), moving out of state (COPD group: n = 4; control group: n = 1), and loss to follow-up (COPD group: n = 3).

Documented Viral Infections

Documented viral infections were common, with picornaviruses, parainfluenza viruses, and coronaviruses being the most frequently identified agents (Table 2). Approximately 75% of the recovered picornaviruses were rhinoviruses; the other 25% were either unclassified or enteroviruses. All but three of the coronavirus infections were due to the OC43 virus. Rhinoviruses and coronaviruses were responsible for 35% of the virus infections identified in the control subjects and 43% of the virus infections in the COPD subjects. Of the parainfluenza virus, influenza virus, and RSV infections identified, 12% were demonstrated by both culture and serologic methods, 24% by culture only, and 64% by serologic methods alone. Adenovirus and cytomegalovirus infections were identified infrequently. There were no significant differences in frequency of specific viruses identified in the COPD subjects with mild and those with moderate/severe obstruction.

Virus-Associated Illnesses

There were 221 episodes of respiratory illness (1.4 acute respiratory illnesses/yr) identified in the control subjects and 323 episodes of respiratory illness (2.4 acute respiratory illnesses/ yr) in the COPD subjects (Table 3). Respiratory illnesses occurred more frequently in the subjects with moderate/severe COPD than in those with mild COPD (3.0 versus 1.8 acute respiratory illnesses/yr, respectively). RTVIs occurred with a similar frequency in the control and COPD groups (0.54 and 0.45 RTVIs/yr, respectively) and in the mild and moderate/ severe COPD groups (0.38 and 0.52 RTVIs/yr, respectively) (Table 4).

Respiratory illnesses were associated with a virus infection in 39% (87 of 221) of the control subjects, as compared with only 19% (61 of 323) of the COPD subjects (p < 0.001). Virus infection was documented more commonly when subjects were evaluated during their acute illness than for nonevaluated, self-reported illnesses in both the control (44% versus 28%, respectively, p < 0.05) and COPD groups (27% versus 8%, respectively, p < 0.001), although the frequency of documentation of virus infection was consistently higher for the control group (Table 3). All illnesses evaluated at a study visit in members of the control group were associated with upper respiratory symptoms, whereas 45 illnesses evaluated at visits by members of the COPD group were associated with lower respiratory symptoms only. Virus infections were identified with similar frequencies in the control and COPD groups when only upper respiratory symptoms were present, but when lower respiratory symptoms were present, virus infections were identified more commonly in the control group than in the COPD group (Table 3).

The median time from the onset of illness to a visit for an acute illness was the same (3 d) for both the control and COPD groups, whether or not a virus infection was identified. This also was true for those subjects for whom the respiratory virus infection was documented by cell culture. Cell culture assays were positive for virus twice as frequently for control subjects as for COPD subjects (50 of 157 [32%] versus 30 of 185 [16%], respectively, p = 0.001). The median interval between acute and convalescent sera used to identify virus-associated illnesses was 1.7 mo; the interval was less than 2 mo for 59% of such paired sera, less than 3 mo for 80%, and less than 4 mo for 88%.

Asymptomatic Infections

Virus cultures were obtained at each routine visit when the subjects appeared to be well. Among these "well"-visit viral cultures, 17 of 926 (1.8%) specimens were positive for viruses. Picornaviruses over half of these virus-positive cultures. The other viruses isolated included adenovirus (2), influenza virus type A (2), and parainfluenza virus (4). Increases in antibody titers in paired sera were detected in 33 of 7,704 (0.4%) of serologic tests performed at a time when no respiratory illness was detected. Over half of the responsible viruses were found to be parainfluenza virus. Other serologic increases in antibody titer were detected for influenza virus type A (6), coronavirus (5), and RSV (2). Herpes simplex viruses were identified in 16 cultures, including 10 from asymptomatic subjects and six from subjects with respiratory illness. These data justify our decision not to include these viruses as potential causes of respiratory illness in the study population.

Duration of Illness

When an illness was reported, the subject was contacted every third day until he or she felt well or in a state identical to his or her baseline. The duration of the respiratory illness was calculated from the time the subject stated the symptoms began until the subjects was well or had returned at baseline. There were no significant differences in the mean (or median) durations of respiratory illnesses in the control and COPD subjects (12.8 ± 6.9 versus 13.7 ± 7.8 d, respectively). There were no significant differences in the duration of illness according to the type of virus identified or in the control versus COPD groups, or in subjects with upper versus lower respiratory tract symptoms.

The use of steroids was determined by the patient's physician. The number of elderly controls who received steroids during an illness episode was approximately 5%. Of COPD patients with FEV₁  50%, approximately 20% received or were taking steroids during acute episodes, and of those COPD patients with baseline FEV₁ < 50%, approximately 85% received steroids. No difference in the duration of respiratory symptoms was seen in those who received or did not receive steroids, and the likelihood of identifying an RTVI in steroid recipients and nonrecipients was similar (22% versus 29%, respectively; p = 0.25).

Influenza Vaccination

On an annual basis, influenza vaccination rates in our study cohort varied from 89.2% to 93.9%. Symptomatic influenza virus infections were more common in the nonvaccinated subjects than in the vaccinated subjects (13.3% versus 5.2%, respectively; p = 0.086, Fisher's exact test). The overall protection rate against symptomatic influenza virus infection was calculated to be 65% (95% CI: 13% to 89%). One emergency center visit associated with an influenza virus infection was made by a COPD subject who had been vaccinated 6 mo earlier.

Seasonal Variation in RTVIs

RTVIs were detected in every month of the year. However, a definite increase in both total respiratory illnesses and virus-associated illnesses was noted during the fall and winter months. A peak in both total respiratory illnesses and virus-associated illnesses was documented in the control subjects during December, January, and February. During these months, over half of all illnesses were associated with a respiratory virus infection. In the COPD subjects, a rise in reported respiratory illnesses began in September and remained through May of the year. A smaller percentage of virus-associated illnesses was detected during each month in the COPD subjects than in the control subjects.

Even when influenza virus was circulating in the community, the majority of virus-associated illnesses in both the control and COPD subjects were not due to influenza viruses (Figure 1). Rhinoviruses were detected in every month except July. Coronavirus infections were detected throughout the year, although the majority were identified during winter months.

View larger version (25K): [in this window] [in a new window]   Figure 1.   Seasonal variation in respiratory virus infections. Bar graphs represent cases of proven viral respiratory illnesses by month and etiology in control (hatched squares) and COPD (open squares) subjects. A = adenovirus, C = coronavirus, CMV = cytomegalovirus; E = enterovirus; F = influenza virus A or B; P = parainfluenza virus type 1, 2, or 3; R = rhinovirus; RS = respiratory syncytial virus. Dual viral infections are designated with backslash (/). Circles = hospitalizations.

Utilization of Medical Services

Utilization of medical services was determined by documentation of physician calls, physician visits, emergency-center visits, and hospitalization for respiratory illness (Table 4). Although a significantly higher number of control than of COPD subjects had at least one RTVI, the annual rate of RTVIs was not different in the two study groups. Thirty-one percent of the control subjects made at least one physician visit for RTVIs, but no RTVI-associated emergency-center visits or hospitalizations were reported.

The COPD subjects had significantly more physician visits, emergency-center visits, and hospitalizations for respiratory tract illnesses than did the control group subjects. The increased utilization of medical services was accounted for by COPD subjects with moderate/severe obstruction (FEV₁ < 50% predicted).

Twelve (35%) of the COPD subjects with moderate/severe airway obstruction had 52 total hospitalizations for respiratory tract illnesses. Most (82%) were admitted with a diagnosis of "acute exacerbation," and 12 (22%) were diagnosed with acute "pneumonia." In contrast to the emergency-center visits, the majority of hospitalizations (37 of 52 = 71%) were evaluated during the acute phase of illness, and virus cultures and acute serum samples were obtained. Of the 37 hospitalizations that were evaluated acutely, a virus-associated illness was found in 30%. Only 7% (one of 15) of subjects not seen during hospitalization had an associated respiratory viral infection identified by serologic studies. The average length of hospital stay for virus-associated illness was 16.4 d. There were no deaths during these hospitalizations. Of the 12 virus-associated hospitalizations, pneumonia was present in two, the patient was febrile in six, both upper and lower respiratory signs and symptoms were present in five, and lower respiratory tract signs and symptoms only were present in nine. The virus infections associated with hospitalizations included coronavirus (five cases); influenza virus, parainfluenza virus, and RSV (two cases each); and rhinovirus and CMV (one case each) infections. Forty-five percent (nine of 20) of the hospitalizations between December and March were associated with a respiratory virus infection. Three of the 10 virus-associated illnesses in the hospitalized patients were dual virus infections.

There were 12 deaths during the course of the study: four in the control group and eight in the COPD group. None of the deaths in the control group was related to respiratory illness, whereas six of the eight in the COPD group were attributed to the subject's underlying lung disease. Five of these six subjects were in the moderate/severe COPD group. Two of the six subjects had virologic studies performed during the illness preceding their death, and one of the illnesses was associated with an influenza A virus infection. The other four subjects had no virologic studies during their final illness.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

This is the first longitudinal, cohort study of older COPD subjects stratified by enrollment FEV₁ (mild versus moderate/ severe obstruction) to define the etiology, frequency, severity and medical-care impact of RTVIs as compared with RTVIs in a control group of comparable age. Approximately 90% of the control and COPD subjects received influenza virus vaccinations annually during the study, so that illness patterns and epidemiology reflect what could be expected in a similar, heavily vaccinated population.

Picornaviruses (primarily rhinoviruses), parainfluenza viruses, and coronaviruses were the most commonly identified viruses associated with respiratory illness in the population of COPD and control subjects who were heavily vaccinated against influenza viruses. Influenza virus infections were identified with a frequency similar to or slightly greater than RSV infections. Adenovirus and cytomegalovirus infections were uncommon. The relative importance of these viruses as causes of respiratory illness is similar to that found in previously reported studies (17, 27, 28). Smith and colleagues (17) evaluated the occurrence of respiratory virus infections over an 8-yr period in persons with varying degrees of obstructive lung disease. Respiratory virus infections were documented in 16% of episodes of acute respiratory illness. Rhinoviruses and influenza viruses were the most common infecting agents, and parainfluenza and coronaviruses were the next most common. Influenza vaccine was used for approximately 40% of the observation time. Nicholson and associates (27) evaluated upper respiratory illnesses in a geriatric population, and identified a respiratory pathogen in 43% of episodes, a rate similar to that seen in our control population. No evaluation of pulmonary function was reported. Rhinoviruses and coronaviruses were the most common virus infections seen in Nicholson and colleagues' study followed by influenza viruses and RSV. Parainfluenza virus infections were seen in only 3% of the episodes in which a respiratory pathogen was identified. The extent of use of influenza vaccine was not noted. Walsh and coworkers (28) studied a heavily influenza vaccinated (90%) population with COPD and congestive heart failure over two winter seasons, and found that influenza A virus infections were the most commonly identified respiratory virus infections (28). RSV and rhinovirus infections were the next most commonly identified infections.

In our study, annual symptomatic respiratory virus infection rates were similar for the control and COPD populations (0.54 and 0.45 RTVIs/yr, respectively). These rates also closely parallel those reported for an ambulatory geriatric population (0.53 RTVIs/yr) and for a population with varying degrees of obstructive lung disease (0.44 RTVIs/yr) (17, 27). Additionally, the incidence of all acute respiratory illnesses in the control population in our study (1.3 acute respiratory illnesses/yr) was similar to that reported for a similar ambulatory geriatric population (1.2 acute respiratory illnesses/yr) (27). The mild COPD group in the current study had a slightly higher incidence of acute respiratory illness (1.8 episodes/yr), whereas subjects with moderate/severe obstruction had an incidence of acute respiratory illness (3.0 episodes/yr) that was more than twice that of the control population. These findings contrast with the observations of Smith and colleagues (17), who found that the occurrence of acute respiratory illness was similar in subjects with no disease to mild obstructive lung disease (2.2 episodes/yr) and those with more advanced airway obstruction (2.2 episodes/yr).

The observation that acute respiratory illness occurred twice as frequently in our COPD subjects as in the control subjects, whereas the RTVI incidence was similar has several possible explanations. First, it is possible that illnesses not identified as RTVIs were caused by another pathogen, such as Chlamydia pneumoniae, Mycoplasma pneumoniae, or other bacteria, and that these infections occurred more commonly in COPD subjects (29). We did not perform any bacteriologic tests in this study, although Smith and colleagues documented M. pneumoniae infections only infrequently in their study population (17). A second possibility is that the detection methods used in the current study were less sensitive in identifying virus infection in the COPD group than in the control group. Atmar and associates (32) recently demonstrated the improved sensitivity of RT-PCR in detecting picornavirus and coronavirus infections in adult asthmatic subjects with acute respiratory illnesses. These assays were not performed in our study. Time from symptom onset to clinical evaluation does not explain the differences seen, because this time was similar for both the COPD and control groups for illnesses for which a study visit was made. The receipt of systemic corticosteroids could theoretically have diminished the likelihood of detecting a serologic response associated with infection, and could be an explanation for the differences seen for illnesses for which an acute study visit was not made. However, RTVIs were as likely to be identified in recipients as in nonrecipients of corticosteroids among COPD subjects who were seen acutely. In addition, virus isolation frequencies were lower in the COPD group for illnesses evaluated through cell culture, a finding that would not be expected as a result of corticosteroid administration. The third possibility for the greater frequency of acute respiratory illness but not of RTVI in our COPD subjects is that the occurrence of upper and lower respiratory symptoms in COPD subjects was less specific an indicator of RTVI than it was for control subjects. Some of the illnesses seen in COPD subjects may have represented variability in the baseline symptomatology of these individuals.

Almost 90% of our study subjects received influenza virus vaccinations each fall. An apparent protection rate can be calculated by comparing the frequency of symptomatic influenza virus infection in vaccinated and unvaccinated individuals, and the calculated rate of 65% in our study is similar to previously reported rates for influenza vaccination (33). Although influenza infection and an associated respiratory illness were seen in a few vaccinated subjects in our study, there was only one documented influenza infection in a vaccinated subject that was associated with an emergency-center visit (and eventual mortality). These observations are in accord with previous studies supporting the role of influenza vaccination for decreasing serious influenza-related illnesses in elderly persons and patients with COPD (34).

Although our COPD subjects did not appear to be more susceptible to RTVIs than did age-matched control subjects, they showed significantly more utilization of medical care facilities, including physician visits, emergency-center visits and hospitalizations. Within the COPD cohort, there was a striking increase in medical care use by subjects with moderate/severe obstruction. Among COPD patients with baseline FEV₁ < 50% predicted, approximately 25% of hospitalizations for acute respiratory illnesses were associated with RTVIs. Eighty percent of these hospitalizations occurred from December through March. Only 20% (two of 10) of these virus-associated illnesses were due to influenza virus. Coronaviruses were the most frequently identified viruses associated with the need for hospitalization. Dual virus infections were associated with three of the 10 illnesses requiring hospitalization.

Utilization of medical services for RTVIs was not different for the COPD subjects with mild airway obstruction (FEV₁,  50%) than for the control subjects. No short-term increase in morbidity was noted in these mildly obstructed COPD subjects who were ill. This group did not utilize the emergency center or become hospitalized in significantly greater numbers than the control subjects. In addition, illness duration was similar for the two groups.

A recent study of Medicare beneficiaries with COPD documented significant increases in expenditures per capita as compared with those for all aged Medicare beneficiaries (35). The study found that COPD accompanied by a diagnosis of URTI increased per capita expenditures by 10%. With a diagnosis of COPD plus pneumonia, the cost was twice that for all Medicare recipients. Inpatient hospital costs were 2.7 times higher for Medicare recipients with COPD than for the total group. The study concluded that the most costly 10% of Medicare beneficiaries with COPD accounted for nearly one-half of total expenditures. These results, combined with our observations of the role of respiratory viral infections in COPD patients and the utilization of medical services, has implications for future interventions in this population. Wide use of currently approved influenza vaccine will aid but not eliminate the need for other strategies to control respiratory viral infections in this vulnerable population. Newer prophylactic antiviral agents are warranted for those respiratory viruses for which vaccines are unlikely to be developed. Seasonal prophylactic measures should be a priority in COPD patients with moderate/severe obstruction.