Acute viral infections of upper respiratory tract in elderly people living in the community: comparative, prospective, population based study of disease burdenBMJ 1997; 315 doi: https://doi.org/10.1136/bmj.315.7115.1060 (Published 25 October 1997) Cite this as: BMJ 1997;315:1060
- Karl G Nicholson, senior lecturer in infectious diseasesa,
- Julie Kent, research assistanta,
- Victoria Hammersley, research assistanta,
- Esperanza Cancio, postdoctoral research fellowa
- a Leicester University School of Medicine, Department of Microbiology and Immunology, Leicester LE1 9HN
- Correspondence to: Dr Nicholson
- Accepted 19 June 1997
Objective: To evaluate the disease burden of upper respiratory infections in elderly people living at home.
Design: Prospective surveillance of elderly people.
Setting: Leicestershire, England
Subjects: 533 subjects 60 to 90 years of age.
Main outcome measures: Pathogens, symptoms, restriction of activity, duration of illness, medical consultations, interval between onset of illness and medical consultation, antibiotic use, admission to hospital, and death.
Results: 231 pathogens were identified for 211 (43%) of 497 episodes for which diagnostic specimens were available: 121 (52%) were rhinoviruses, 59 (26%) were coronaviruses, 22 (9.5%) were influenza A or B, 17 (7%) were respiratory syncytial virus, 7 (3%) were parainfluenza viruses, and 3 (1%) were Chlamydia species; an adenovirus and Mycoplasma pneumoniae caused one infection each. Infections occurred at a rate of 1.2 episodes per person per annum (95% confidence interval 1.0 to 1.7; range 0-10) and were clinically indistinguishable. Lower respiratory tract symptoms complicated 65% of upper respiratory infections and increased the medical consultation rate 2.4-fold (χ2 test P<0.001). The median interval between onset of illness and medical consultation was 3 days for influenza and 5 days for other infections. Rhinoviruses caused the greatest disease burden overall followed by episodes of unknown aetiology, coronaviruses, influenza A and B, and respiratory syncytial virus.
Conclusions: Respiratory viruses cause substantial morbidity in elderly people. Although respiratory syncytial virus and influenza cause considerable individual morbidity, the burden of disease from rhinovirus infections and infections of unknown aetiology seems greater overall. The interval between onset of illness and consultation together with diagnostic difficulties raises concern regarding the role of antiviral drugs in treating influenza.
There are few data on the morbidity associated with respiratory viruses other than influenza in elderly people
Respiratory virus infections in elderly people are clinically indistinguishable, and patients with influenza will be difficult to target for antiviral treatment without a near patient diagnostic test
Overall, two thirds of elderly people with colds and four fifths of those with influenza and respiratory syncytial virus can be expected to develop lower respiratory illness
Although influenza and respiratory syncytial virus cause substantial morbidity in elderly people, the disease burden from rhinovirus infections and colds of unknown aetiology is greater overall
Most elderly patients seek medical attention beyond 48 hours when the benefits of antiviral treatment of influenza remain unproved
Excess deaths have consistently been shown in elderly people during the winter and have largely been attributed to influenza and low temperature.1 Until recently the possible contribution of respiratory viruses other than influenza has attracted little attention. During winter 1988-9 we observed the cocirculation of various respiratory viruses, including influenza, in homes for elderly people in Leicestershire.2 The illnesses were indistinguishable and were associated with lower respiratory complications and deaths. We speculated that the burden of respiratory viruses other than influenza was considerably underestimated. As remarkably little is known about respiratory viral infections in elderly people living at home, we prospectively evaluated upper respiratory infections in such people in Leicestershire over two winters.3
Subjects and methods
Population and study
The study was conducted among people aged 60 years and older during the winters of 1992-3 and 1993-4 in Leicestershire.3 During April to June 1992 we sent letters to 800 of the 129 000 people aged 60 years and older who lived in Leicestershire inviting them or their spouses, or both, to participate in the study; the sample was randomly selected by the family health services authority computer. We received 617 responses including 52 that were returned unanswered because of incorrect address, death, or disinterest. A total of 441 subjects were recruited when the study began in 1992. Ninety four of the 441 subsequently died, deteriorated, or declined to take part during 1993-4, and an additional 92 subjects were recruited in 1993 from the original respondents. Patients living in residential care were excluded. Basic demographic data, medical and drug history, and nose and throat swabs were collected at recruitment. During surveillance periods each subject was contacted weekly by telephone at a prearranged time. By using a questionnaire, volunteers were asked whether an upper respiratory infection had occurred during the previous week. When illness was reported, a record was made of date of onset, symptoms,3 incapacitation, medical consultations, drug prescriptions, admission to hospital, and death.
Subjects were seen at home as soon as possible after onset of illness. Diagnostic specimens were collected as described previously,3 and symptoms were converted into syndromes.3 4 The illness was considered lower respiratory if productive cough, wheezy breathing, or pain on respiration were present, irrespective of other respiratory symptoms. It was considered to be an upper respiratory tract infection if coryza was present without lower respiratory symptoms. If sore throat or hoarseness was present without any of the above symptoms the illness was identified as laryngopharyngeal. Illnesses without any of the above symptoms but with only non-productive cough, earache, nasal stuffiness, or other symptoms were classified as other.
We studied 533 volunteers, 441 during the first winter and 439 during the second.3 The 257 men and 276 women were aged 63-90 (mean (SD) 72.6 (5.7)) years and 60-90 (71.8 (6.1)) years, respectively, on recruitment. More men than women (207 (81%) v 129 (47%), χ2 test P<0.001) were either current or past smokers, but men and women were comparable with respect to indications for influenza vaccine5; vaccine uptake; admission to hospital during the preceding 5 years; attendance at a hospital outpatient department during the preceding 12 months; and proportions consulting their medical practitioner during the preceding 12 months. The project was approved by the Leicestershire ethics committee and signed informed consent was obtained from all volunteers.
Nasal swabs were placed high in the anterior nares and throat swabs were passed firmly over the tonsils and pharynx. Swabs were immediately placed in medium containing nutrient broth, transported on dry ice, and stored at −70°C. Serum samples taken during the acute and convalescent phase were stored at −20°C and tested later by complement fixation tests for antibodies to adenovirus; influenza A and B; respiratory syncytial virus; parainfluenza viruses types 1, 2, and 3; Mycoplasma pneumoniae; and Chlamydia psittaci. Haemagglutination inhibition tests were also carried out for the identification of infections by influenza type A. A fourfold rise in antibody titre was taken as indicating infection. Enzyme linked immunosorbent assay was used to detect rises in antibodies to coronaviruses 229E and OC43.6 Rhinoviruses in nose and throat swabs were identified with a seminested reverse transcriptase polymerase chain reaction.6 7 Rhinovirus serotypes 14 and 1B were used as positive controls; additional controls included baseline samples, water, and transport medium. The appearance of a 202 base pair amplification was taken to indicate rhinovirus infection.
Estimates of disease burden
We compared the disease burden of episodes for which diagnostic specimens were provided with the method used by the US Institute of Medicine's committee on issues and priorities for new vaccine development for diseases of importance in the United States.8 The remit of the committee was to develop a comprehensive approach to setting priorities for accelerated vaccine development. In the decision making framework information on morbidity and mortality are combined into a single numerical score, which permits quantitive comparison of the burdens of morbidity and mortality rising from different pathogens. Given that individual pathogens may cause a spectrum of acute and chronic illness the committee estimated the number of cases of different infections occurring in different morbidity categories, namely: A—causing moderate localised pain, mild systemic reaction, or impairment requiring minor change in normal activities; B—causing moderate pain or moderate impairment requiring moderate change in normal activities (for example, housebound or in bed); C—requiring admission to hospital; D and E—relating to chronic disability; F—relating to total impairment; G—relating to reproductive impairment resulting in infertility; and H—relating to death. The unit of comparison between categories was designated as the “infant mortality equivalent.”
In the present study illnesses not affecting the lower respiratory tract or not causing impairment resulting in a change in normal activities (confinement to bed or inability to cope with household activities) were categorised “low morbidity” (category A). Episodes affecting the lower respiratory tract or confining subjects to bed or affecting their ability to cope with shopping, cooking, or washing were considered “moderate” (category B); those resulting in admission to hospital were category C, and deaths were category H. Episodes with identification of more than one pathogen and pathogens causing fewer than 10 episodes were excluded from the comparative analysis. Disease burden values for categories A, B, and C were calculated from the product of the number of cases and the median duration of illness for that category divided by infant mortality equivalence values for each category: 2 000 000 for category A, 100 000 for category B, and 80 000 for category C.8 The disease burden from category H was calculated from the number of deaths divided by an infant mortality equivalence value of 3.8 These values represent a median of the perspectives of original committee members. The total score for each pathogen is the sum of category subtotals.
Baseline variables in men and women and people with coronavirus, rhinovirus, influenza, and respiratory syncytial virus infections and episodes caused by unknown agents were compared by χ2 tests for discrete variables and Kruskall-Wallis tests for continuous variables. Differences in the distributions of variables between different infections were assessed by χ2 tests for discrete variables and Kruskall-Wallis tests for continuous variables. The Mann-Whitney U test was used to compare the intervals between onset of illness and medical consultation for people with influenza and other infections and duration of illness in those with and without lower respiratory illness.
Volunteers completed 24 700 patient weeks of observation. We identified 706 episodes, occurring at a median rate of 1.2 episodes per person per annum (95% confidence interval 1.0 to 1.7; range 0-10) in 384 (72%) subjects. Symptoms were documented for 691 episodes. Laboratory specimens were collected a median of four days after onset of symptoms (range 1-21 days) for 497 (72%) of the 691 classified episodes. Missing specimens occurred when there were delays in reporting illness—notably, during Christmas, New Year, and Easter and periods of travel.
Infection with 231 pathogens was identified for 211 (43%) of the 497 episodes (table 1). Of the 231, 121 were rhinoviruses (52%), 59 (26%) were coronaviruses, 22 were influenza A or B (9.5%), 17 were respiratory syncytial viruses (7%), 7 (3%) were parainfluenza viruses, and 3 (1%) were Chlamydia; an adenovirus and Mycoplasma pneumoniae caused one infection each.
Characteristics of respiratory viral illness
To avoid over-representation of symptoms of subjects with more than one infection, we focused on infections in different subjects. Table 2 shows the manifestations of 291 single infections occurring in 291 people with rhinovirus, coronavirus, influenza A and B, or respiratory syncytial virus infection and infections of unknown aetiology; and demographic features associated with episodes. Coinfections and infections due to parainfluenza viruses, adenoviruses, Mycoplasma pneumoniae, and Chlamydia species are excluded because of their small number. Most subjects (284; 98%) had upper respiratory symptoms; 189 (65%) had lower respiratory syndromes, and more than half (170; 58%) had systemic features. Table 2 shows that age, sex, and current smoking status of the groups were comparable; though the prevalence of chronic medical conditions that are indications for influenza vaccine differed among the groups (χ2 11.09; 4df; P<0.05).
There were no pathognomonic features for any pathogen (table 2). The median duration of the 291 episodes was 15 days (range 2-79). It was longer in those with lower respiratory symptoms (median duration 16 days v 12 days (Mann-Whitney test, P<0.0001)), which occurred in 18/42 (43%; 95% confidence interval 28% to 58%) coronavirus infections, 54/85 (64%; 54% to 74%) rhinovirus infections, 93/134 (69%; 61% to 77%) unknown infections, 15/19 (79%; 61% to 97%) influenza infections, and 9/11 (82%; 59% to 100%) respiratory syncytial virus infections (χ2 13.26; 4df; P<0.02) (table 2). The incidence of sweats, myalgia, rigors, earache, confinement to bed, capacity to carry out shopping, cooking or washing, and domiciliary medical consultations also differed when infections in table 2 were compared. Patients with influenza had high rates of myalgia, sweats, and rigors; 63% (12/19) were confined to bed, almost three quarters were unable to carry out shopping, cooking, or washing, and lower respiratory symptoms were common (15/19; 79%). Similarly most (9/11; 82%) of those with respiratory syncytial virus had lower respiratory symptoms.
During the influenza A epidemic in 1993-4, 41% (7/17) of patients with influenza confirmed by laboratory tests had myalgia with one or more respiratory symptoms (sensitivity), and the percentage of all episodes during the epidemic with myalgia and respiratory symptoms that were confirmed as influenza A (the positive predictive value) was 28% (7/25). The sensitivity was 29% (5/17) with the symptom complex of myalgia, respiratory symptoms, and feverishness or sweats, and the positive predictive value was 33% (5/15). The sensitivities remained identical during non-epidemic periods, but the positive predictive values fell to 7% (7/99) and 9% (5/58), respectively, with the above symptoms.
General practitioner review
Overall 117 (40%) of the 291 episodes were reviewed medically and 100 (34%) were treated with antibiotics. Consultation rates were higher for lower respiratory episodes than the remainder (96/189 (51%) v 21/102 (21%); χ2 25.14; P<0.001). Comparison of the different infections revealed a difference in domiciliary consultation rates (χ2 10.07; 4df; P<0.05), though neither the combined practice and domiciliary consultation rates nor the rates of antibiotic prescription differed (table 2). Of 19 people with influenza A as sole pathogen, nine were reviewed by a medical practitioner a median of 3 days after onset of symptoms (range 1-14). Similarly, 108/272 (40%) coronavirus, rhinovirus, respiratory syncytial virus and unidentified infections were reviewed after a median of 5 days (range 1-29 days) (Mann-Whitney U test, z=0.538; P=0.59) (table 2).
Deaths and admissions to hospital
Altogether three of the 497 infections led to admission to hospital and another was fatal. One woman died from chronic obstructive airways disease exacerbated by a rhinovirus. A second woman with chronic airways disease developed wheeze with sputum production and was in hospital for 6 days with influenza A. Two patients with chronic respiratory disease were admitted for 12 days and 4 weeks with exacerbations after upper respiratory infections of unknown aetiology.
Disease burden values
Table 3 shows the disease burden values and the proportion of cases in category A. Influenza had the smallest proportion of cases with low morbidity (category A), but it ranked fourth overall after rhinovirus infections, episodes of unknown aetiology, and coronavirus infections.
Our subjects suffered a median of 1.2 acute “upper” respiratory tract infections per annum, which is virtually identical with rates reported for frail elderly people attending day care units9 and people aged 60 years and over living in the community in Tecumseh.10 In contrast with the results of Falsey et al, who found respiratory syncytial virus to be the most common cause of acute respiratory illness in elderly people attending day care units,9 we used the polymerase chain reaction instead of viral cultures to identify rhinoviruses and an enzyme immunoassay to identify infections with coronavirus OC43. We used a less sensitive technique to identify respiratory syncytial virus, but the same techniques to identify infections with influenza A and B and coronavirus 229E.
Although we invited a randomly selected population of elderly people in Leicestershire to take part in the study, participants may have been more health conscious than non-participants and report symptoms and consult their general practitioner for minor respiratory complaints more readily, thus introducing bias. Indeed, comparatively few were current smokers, and the overall immunisation rate in relation to the prevalence of chronic medical conditions was high.3 The timing of the study could also introduce bias as there may have been unusually low attack rates for influenza in elderly people during 1992-3 and 1993-4 or variants of influenza causing little morbidity. Similarly failure to obtain diagnostic specimens due to delays in reporting illness could introduce bias relating to seasonal infections.
Scope for antiviral treatment
As in other studies of colds we found rhinoviruses followed by coronaviruses to be the most common pathogens,6 10 11 12 and as in children,13 14 healthy adults,15 16 and frail elderly people9 10 17 we identified no pathognomonic features for any pathogen. As amantadine and rimantadine are effective when given within 24 to 48 hours after onset of influenza A18 19 20 21 we evaluated the sensitivity and positive predictive value of influenzal symptoms. One or more respiratory symptoms and myalgia occurred in only 41% of patients with influenza, and only 28% of episodes with these features were confirmed as influenza. The inclusion of feverishness or sweats as diagnostic criteria reduced the sensitivity to 29% and increased the positive predictive value to 33%. Temperature was not measured in our study, but raised temperature occurs in similar proportions of influenza, respiratory syncytial virus, rhinovirus, and coronavirus infections in elderly people.9 17 We conclude that patients with influenza will be difficult to target for antiviral chemotherapy without a rapid, near-patient diagnostic test. Moreover patients with influenza A in our study consulted their practitioner a median of 3 days after onset, suggesting that many elderly people with influenza may seek medical attention too late for successful treatment.
Burden of illness
To compare burden of illness caused by respiratory viruses we used a method developed to set priorities for accelerated vaccine development. The method was originally applied by using estimates of disease incidence in the United States, but even with this and other difficulties the system has been a useful tool. In our study we applied the method to a small cohort, but disease incidence and morbidity were monitored closely over two winters.
An intriguing observation in this study is the high incidence of lower airways being affected during colds. Respiratory syncytial virus and influenza were often complicated by lower respiratory illness, and respiratory syncytial virus closely resembled influenza in terms of domiciliary medical consultations. Interestingly, Falsey et al noted similar clinical manifestations during 159 respiratory syncytial virus and 221 influenza illnesses among elderly people living in the community who were admitted to hospital with acute cardiopulmonary conditions; they observed mortality of 10% and 6% for respiratory syncytial virus and influenza, respectively, and concluded that respiratory syncytial virus causes serious disease in these older people.22 The overall burden of respiratory syncytial virus in our study was lower than that for influenza but is probably underestimated because of the use of the complement fixation test to diagnose respiratory syncytial virus. None the less, our observations and those of other investigators1 9 17 22 provide strong support for an assessment of candidate respiratory syncytial virus vaccines in elderly people.
Unlike respiratory syncytial virus and influenza, coronaviruses caused respiratory illness throughout the study. They were associated with lower respiratory illness in more than 40% of patients and a quarter consulted a medical practitioner and received antibiotics. Coronaviruses represent the second most common cause of colds in adults and, in our cohort, gave a higher disease burden value than influenza or respiratory syncytial virus.
In our first report we speculated whether the burden of rhinovirus infections in elderly people might approach that of influenza.3 In this study we found that a greater burden came from rhinoviruses, pathogens that we were unable to identify, and coronaviruses. Mortality increases considerably during the winter months, when consultations for upper respiratory syndromes increase.1 23 It is therefore highly plausible that considerable morbidity and mortality from regular seasonal infections with rhinoviruses, coronaviruses, and respiratory syncytial virus have been overshadowed by less regular, readily recognisable epidemics of influenza.
We gratefully acknowledge the support of volunteers who participated in this study.
Funding: The study was supported by a grant from the British Lung Foundation.
Conflict of interest: None.