A community-based, time-matched, case-control study of respiratory viruses and exacerbations of COPD

ARTICLE IN PRESS

Respiratory Medicine (2007) 101, 2472–2481

0954-6111/$ - see frdoi:10.1016/j.rmed.

�Corresponding aufax: +61 03 9342 849

E-mail address: A

A community-based, time-matched, case-controlstudy of respiratory viruses and exacerbationsof COPD

Anastasia F. Hutchinsona,b,c,�, Anil K. Ghimirea, Michelle A. Thompsona,d,Jim F. Blackc, Caroline A. Brandb, Adrian J. Loweb, David M. Smallwooda,Ross Vlahosd, Steven Bozinovskid, Graham V. Brownc,Gary P. Andersond, Louis B. Irvinga

aDepartment of Respiratory Medicine, Melbourne Health, Victoria 3050, AustraliabDepartment of Clinical Epidemiology and Health Services Evaluation Unit, Melbourne Health, Victoria 3050, AustraliacVictorian Infectious Diseases Service, Melbourne Health, & NHMRC Centre for Clinical Research Excellence in InfectiousDiseases, University of Melbourne, Victoria 3010, AustraliadDepartments of Medicine and Pharmacology Cooperative Research Centre for Chronic Inflammatory Diseases,University of Melbourne, Parkville, Victoria 3010, Australia

Received 26 April 2007; accepted 16 July 2007Available online 5 September 2007

KEYWORDSLung diseases:obstructive;Respiratory tractinfections;Respiratory viruses

ont matter & 20072007.07.015

thor. Department3.

nastasia.Hutchins

SummaryRespiratory viruses are associated with severe acute exacerbations of chronic obstructivepulmonary disease (COPD) in hospitalized patients. However, exacerbations are increas-ingly managed in the community, where the role of viruses is unclear. In communityexacerbations, the causal association between viruses and exacerbation maybe con-founded by random fluctuations in the prevalence of circulating respiratory viruses.Therefore, to determine whether viral respiratory tract infections are causally associatedwith community exacerbations, a time-matched case-control study was performed.Ninety-two subjects (mean age 72 yrs), with moderate to severe COPD, (mean FEV1 40%predicted), were enrolled. Nasopharyngeal swabs for viral multiplex polymerase chainreaction and atypical pneumonia serology were obtained at exacerbation onset. Controlsamples were collected in synchrony, from a randomly selected stable patient drawn fromthe same cohort.In 99 weeks of surveillance, there were 148 exacerbations. Odds of viral isolation were 11times higher in cases, than their time-matched controls (34 discordant case-control pairs;in 31 pairs only the case had virus and in three pairs only control). Picornavirus (26),

Elsevier Ltd. All rights reserved.

of Respiratory Medicine, Melbourne Health, Victoria 3050, Australia. Tel.: +61 03 9342 8772;

[email protected] (A.F. Hutchinson).

ARTICLE IN PRESS

A case-control study of viruses in COPD 2473

influenza A (3), parainfluenza 1,2,3 (2), respiratory syncytial virus (1), and adenovirus (1)were detected in cases while adenovirus (1) and picornavirus (2) were detected incontrols. In patients with moderate or severe COPD the presence of a virus in upper airwaysecretions is strongly associated with the development of COPD exacerbations. These datasupport the causative role of viruses in triggering COPD exacerbations in the community.& 2007 Elsevier Ltd. All rights reserved.

Introduction

Patients with chronic obstructive pulmonary disease (COPD),particularly those with moderate or severe disease (GOLDstage 2+illness), often experience frequent debilitatingexacerbations during the course of their disease.1 Fromthe patient’s perspective, acute exacerbations of COPD(AECOPD) result in acute ill health and reduced quality oflife, especially the ability to perform normal activities2 andcan contribute to accelerated decline in lung function andloss of lean muscle mass.3,4 AECOPD are also a major burdenon the health care system, especially during the winterseason.5 As much of the cost of AECOPD stems fromhospitalization,6,7 and hospitalization increases the risk ofcomplications, there is increasing interest in the communitymanagement of this disease to optimize outcomes andminimize costly admissions.

It is believed that the majority of AECOPD are caused byrespiratory infections8 although exposure to respirableparticulates in polluted air,9 pulmonary thromboembolism10

and other, unknown causes are also implicated. Viralrespiratory infection is strongly suspected to cause AECOPD.Previous longitudinal, community-based and hospitalizationstudies have found that viral infections are common(39–56%) in patients with AECOPD11,12 and are most usuallyassociated with severe presentations. Most recently, Papiand colleagues have demonstrated the association of provenviral infection with worsening of cellular and biochemicalindices of inflammation.13 While these studies do notformally exclude bystander effects, they provide very strongevidence that viral respiratory infection is an importantcause of severe AECOPD requiring hospitalization andsuggest that viral infection is a cause of AECOPD in thecommunity. In contrast, Hurst and colleagues recentlystudied common cold, assessed by coryzal symptoms, andAECOPD frequencies in the community, by comparing diarycards with spirometry.14 They concluded that exacerbationfrequency is associated with an increased symptomatic coldfrequency. This study does not, however, control for othercauses of coryzal symptoms such as allergic rhinitis that areprevalent in the Australian context. As there may beregional differences in the type of viruses circulatingthrough a community, it was relevant to explore thespectrum of respiratory viruses associated with AECOPD inan Australian community cohort.

To address this limitation in our understanding, we haveestablished a long-term longitudinal cohort in Melbourne,that allows rigorous case control methodology, to addresswhether respiratory viruses are important causes of com-munity AECOPD. This paper addresses the research question;is isolation of a respiratory virus more strongly associatedwith AECOPD than stable disease, in community-dwelling

patients chosen at the same time, from the same COPDcohort. A particular advantage of the time-matched case-control method, where cases (AECOPD) are matched withcontrols at the time of exposure, is that it can accommodatefluctuations in the type, virulence and prevalence ofrespiratory viruses passing through the community.

Our data demonstrate that viruses are a major cause ofAECOPD, but also that virally mediated AECOPD are notinvariably severe enough to warrant hospitalization. Ourdata may therefore contribute to understanding the naturalhistory of COPD and optimizing patient care in communityhealth networks.

Methods

The study, designated the Melbourne Longitudinal COPDCohort study, has been established as an open cohort ofparticipants with COPD to investigate the aetiology andpathophysiology of COPD and its exacerbations.

Participants

Figure 1 summarizes recruitment to the cohort in2003–2005. Inclusion criteria were diagnosis of COPDaccording to GOLD criteria stage II–IV, smoking history ofgreater than or equal to 10 pack-years, at least oneinpatient admission with an exacerbation of COPD in thepast 24 months, age less than 85 years and willingness togive informed consent. Patient characteristics are shown inTable 1. Exclusion criteria were living more than 50 km fromthe hospital, predominant asthma (FEV1 reversibility X15%on spirometry),15,16 lung cancer, a primary diagnosis ofbronchiectasis or idiopathic pulmonary fibrosis, chronicsystemic inflammatory conditions such as rheumatoidarthritis, renal failure requiring dialysis and patientsidentified as requiring palliative care. The Human ResearchEthics Committee (HREC) of Melbourne Health approved thestudy and written informed consent was obtained from allsubjects.

Study design

The study design was a time-matched case control study inwhich an individual with an exacerbation was matched witha control randomly selected from within the cohort who wasnot exacerbating at that time. The study was powered toassess the primary outcome of association of viral respira-tory infection with AECOPD.

COPD exacerbations were defined by the Anthonisencriteria17: Type I as an increase in dyspnoea, sputum volumeand sputum purulence for more than 24 h, Type II as any two

ARTICLE IN PRESS

Patients Enrolled in COPD Cohort N =92 (2003 to 2005)

Total AECOPD over the follow-up period from 2003-2005N = 159 AECOPD

Number of AECOPD per patient0 AECOPD = 32 (35%)1 to 2 AECOPD = 36 (40%)3 to 4 AECOPD = 15 (16%)5 to 6 AECOPD = 6 (6%)7 to 8 AECOPD = 3 (3%)

AECOPDIncluded in Case-Control Analysis N = 148

Excluded AECOPD (N = 11)Aspiration Pneumonia = 1Missed = 7No control obtained = 3

Person Weeks of Follow-upMedian = 47 weeks per patient(Range 1-99 weeks per patient)

Reasons for withdrawal during follow-up. (N= 35)Participated in 2003 only = 15Died = 7Deterioration in co morbid disease = 5Moved > 50 kms = 2Clinical Trial Participation = 2

Figure 1 Melbourne Longitudinal COPD Cohort: patients and COPD exacerbations included in this study.

A.F. Hutchinson et al.2474

of the above symptoms and Type III as one of the abovesymptoms accompanied by sore throat and nasal dischargewithin 5 days, fever without other cause, increased coughand an increase in respiratory rate or heart rate 20% abovebaseline values.18 Severity of exacerbations was definedaccording to the Burge and Wedzicha18 consensus report(2003). At resolution, the exacerbation severity was codedas the maximum severity of each episode according to theconsensus criteria.18 Stable COPD was defined as norequirement for increased treatment above maintenancetherapy (other than bronchodilators) for 30 days.

Identification of exacerbations

Identification of exacerbations at an early stage wasachieved by use of individualised patient action plans thatincluded information about symptoms and instructions tocontact the study team when key symptoms developed(increased dyspnoea, increased wheeze, decreased exercisetolerance, increased cough or change in sputum colour,rhinorrhoea, nasal congestion, sore throat, myalgia orheadaches, fever and or chills). This was further reinforcedby fortnightly phone contact. Resolution of an exacerbationwas defined as completion of treatment with antibiotics andincreased steroids, and return of symptoms to baselinelevels for 48 h. If symptoms had not fully resolved within 30days of exacerbation onset, symptoms needed to be stablefor 48 h and not require further acute treatment. A missed

exacerbation was defined as symptoms not reported within 7days of onset.

Data and sample collection

At baseline, demographic data, spirometry, carbon mon-oxide diffusing capacity (DLCO), 6-min walk test,19 MRCDyspnoea Scale,20 St. George Respiratory Questionnaire andclinical examination data were obtained. The BODE Index21

at recruitment was calculated. Baseline pathology samplesconsisted of nose and throat swabs for respiratory virusmultiplex PCR, atypical pneumonia serology, sputum (ifavailable) for bacterial culture and serum for measurementof inflammatory indices (differential white cell count(WCC), erythrocyte sedimentation rate (ESR), and C-reactive protein (CRP)).

Cases

During exacerbations nose and throat swabs were collectedon the day of identification and again 5–7 days later.Respiratory multiplex PCR was repeated at day 30–60 oncethe exacerbation had resolved.

Controls

Identical respiratory PCR samples were obtained fromcontrols within 5 days of the identification of the case.

ARTICLE IN PRESS

Table 1 Baseline characteristics of the patients.

N. 92Age� 72 yrs (49–85 yrs)Sex Male: Female M 63% F 37%FEV1 (L)� 0.93 (L) (0.35–1.91L)

% Pred. 40% (12–69%)FVC (L)� 2.07 (L) (0.78–3.90L)

% Pred. 66% (21–100%)FEV1/FVC Ratio� 45% (23–70%)DLCO* (N ¼ 65) 7.8 (3.4–28.1),

% Pred. 31% (1–105%)KCO 1.85 (0.4–4.8),

% Pred. 44% (9–112%)White cell count� 8.8 109/L (5–18.3)ESR� 19.1mm/1 h (1–77)C-reactive protein� 6.4mg/L (o2–42)Hypercapnoea 17%Pack years smoking� 53 (10–160)Current smoker 22%Chronic bronchitisy 82%Domiciliary Oxygen 29%BODE Index21/GOLD GOLD 2 N ¼ 19 BODE 3.89 (1–8)

GOLD 3 N ¼ 33 BODE 6.39 (4–10)GOLD 4 N ¼ 40 BODE 7.93 (4–10)

Influenza vaccination 87%Living situation Alone 29%

With spouse 47%With family 21%Other 3%

Abbreviations: (FEV1) forced expiratory volume in 1 s, (FVC(L) forced viral capacity in litres, (DLCO) diffusion lung carbonmonoxide, KCO ¼ (DLCO/unit alveolar volume), % Pred.,percent of normal age adjusted predicted values; ESR,erythrocyte sedimentation rate.�Data expressed as mean (range). Rest as number and

percentage of the cohort.yClinical definition of productive sputum for more than

three months at the time of enrolment.

A case-control study of viruses in COPD 2475

The controls were identified from within the cohort usingcomputer generated random number selection from amongthe currently stable COPD patients, who did not carry thesame virus as the case at recruitment.

Pathogen detection

Respiratory virus multiplex PCR and atypical pneumoniaserology were performed at the Victorian Infectious DiseaseReference Laboratory (VIDRL, a WHO virology referencelaboratory).22 The following viruses were screened; influen-za A and B, picornavirus, respiratory syncytial virus (RSV),parainfluenza and adenovirus. Previous studies conducted byVIDRL have established that the multiplex PCR has highsensitivity and specificity for detection of these viruses.22

Blood for atypical pneumonia serology was obtained onday one and thirty of each exacerbation and screened forInfluenza A and B, Chlamydia pneumoniae and C. psittaci,Legionella pneumophila and Mycoplasma pneumoniae. An

exacerbation associated with an atypical pathogen wasdefined as at least a four-fold rise in serum antibody titrebetween the day one and day thirty specimens. Atypicalserology was performed by VIDRL Serology Laboratory, usingthe following assays; Legionella spp. using an in-houseindirect fluorescence antibody (IFA) assay developed accord-ing to standard procedures,23 Chlamydia spp. specific serumIgG antibodies were detected using SeroELISATM Chlamydia-IgG Enzyme-linked Immunosorbent Assay (ELISA), andC. pneumoniae, C. trachomatis and C. psittaci wereidentified using Chlamydia IgG SeroFIATM’ IFA test kits(Savyon Diagnostics Ltd, Ashdod, Israel). M. pneumoniaewas detected using SERODIAs-MYCO 11 particle agglutina-tion test (Fujirebio Inc. Tokyo Japan), and IgM antibodieswere detected by ELISA using SeroMPTM (Savyon DiagnosticsLtd). Influenza A and B antibodies were detected using an in-house complement fixation antibody test developed byVIDRL Serology Laboratory, in accordance with standarddiagnostic methods.24

Spontaneously expectorated sputum samples were ob-tained on day 1–5 after onset of AECOPD. Respiratorybacteria were identified by sputum microscopy and cultureperformed by Melbourne Pathology according to standardclinical laboratory procedures. Bacterial load was deter-mined using semi-quantitative methods.

Measurement of inflammatory serum markers

Measurement of the WCC, ESR and CRP were performed bythe clinical pathology service. Serum CRP was performed byMelbourne Pathology, using an automated spectrometer(Olympus AU2700) and Olympus immuno-turbidimetric test,manufactured by Olympus Diagnostica GmbH (Irish Branch).Limits of detection were 2–300mg/L, and values4300mg/Lwere obtained by automated onboard dilution.

Measurement of symptoms

Severity of respiratory symptoms was scored using the MRCdyspnoea scale,20 the Borg dyspnoea score when performingroutine daily activities25 and the Symptom Severity Index.26

AECOPD were confirmed by calculating the change insymptom severity from patients’ own stable baseline values.A score for the presence of viral symptoms was calculated onday 1 and 5 of each AECOPD. This measured the followingsymptoms; rhinorrhoea, nasal congestion, sore throat,myalgia or headaches, and subjective fever, chills or rigors.Each symptom was recorded on a scale of zero (absentsymptom) to three (severe).27–33

Statistical analysis

Viral infection was defined as a respiratory virus identifiedon day one and/or day five by respiratory PCR of nose andthroat swabs. To determine whether a virus was more likelyto be associated with an exacerbation (Case) or stable COPD(Control), a matched pair analysis of case-control pairs wasperformed and an odds ratio with 95% confidence intervalwas calculated (Table 2). McNamara’s chi square test forpaired samples was used to measure the significance of theassociation.

ARTICLE IN PRESS

Table 2 Time-matched case-control analysis for infec-tion with respiratory viruses.

Controls

PCR

Positive Negative

CasesPCR positive 0 33PCR negative 3 112

Cases were patients suffering from AECOPD. Controls weretime matched clinically stable patients drawn randomly fromthe same cohort. Positive means a respiratory virus wasdetected and negative means no virus was detected. Theodds ratio for detection of a virus during AECOPD was 11 [95%CI; 3.45–56.07], (po0.001).

A.F. Hutchinson et al.2476

Paired conditional logistic regression was performed totest whether the odds of virus detection was equivalent incase and controls. Potential predictors tested using uni-variate analysis were; age, sex, smoking status, diseaseseverity, COPD sub-type and functional status (measured byBODE Index).21 Cluster analysis was used to adjust forrepeated AECOPD from the same individual—this techniqueadjusts the standard error around the estimate of associa-tion between predictor and the outcome in question. Thesensitivity and specificity of viral symptom scores aspredictors of PCR positivity were calculated using standardmethodology.34

To account for varying periods of follow-up, person-weeksof observation were calculated for the surveillance periodsin 2003 and 2004–2005. The incidence of atypical pneumoniainfections was calculated as incidence per time at risk(incidence density). The total person-weeks of observation,minus the time in which participants had exacerbationsassociated with atypical organisms, was used as thedenominator. The incidence density of atypical organismsis reported as incidence per 1000 person-weeks of observa-tion.

Measurements of inflammatory indices were right skewedand were log-transformed prior to further analysis, (chisquare tests were used to test the distribution was normal).To test whether there was an acute change in levels ofinflammatory indices (CRP, ESR) at AECOPD onset, inflam-matory indices were compared between stable baseline andAECOPD onset using paired t-tests. Results are presented asthe geometric mean ratio of AECOPD onset to stablebaseline values.

All analyses were performed using STATA Version 8.2.35

Results

Patients

Characteristics of the study participants are shown inTable 1, (additional baseline data Supplementary Table 3).Twenty-nine percent were on long-term home oxygen. Meanpack years of smoking were 45 (Range 10–160) and 22% ofpatients were current smokers. Eighty-seven percent had

received annual influenza vaccination for several years andwere current for the multivalent pneumococcal vaccine.Twenty-eight (30%) patients who were colonized withpotentially pathogenic bacteria and had frequent AECOPD(greater than 2.5 per year) were referred for diagnostic HighResolution Computerized Tomography Scan (HRCT), changesconsistent with bronchiectasis were identified in 13.

Exacerbations

Patients were monitored over 3 years; from July toDecember 2003 (Winter–Spring) and from August 2004 toDecember 2005. The total number of person-weeks ofobservation was 4289. The median number of weeks ofmonitoring was 47 weeks per patient (range 1–99 weeks).There were 148 exacerbations; 63% patients had at least oneexacerbation and 44% had more than one exacerbation(range 2–6 exacerbations). The mean number of exacerba-tions per patient per month was 0.14 and exacerbations perpatient per year 1.68. Time from symptom onset to samplingwas short (mean 2.4 days). Sixty-four percent of patientswith exacerbations contacted the study staff, while the restwere identified on routine fortnightly phone call. Thirty-eight percent of AECOPD were mild, 43% moderate and 19%severe.18 Eighty percent of AECOPD were treated in thecommunity with oral antibiotics and/or oral corticosteroids.Table 3 summarises details of exacerbation type andseverity.

Viruses at baseline

Adenovirus and RSV were detected by respiratory PCR fromtwo patients at baseline. They did not report any symptomsof viral upper respiratory tract infection. All subsequenttests on these patients were negative.

Viruses during AECOPD

There was a total of thirty-three viruses detected byrespiratory PCR at the onset of AECOPD; influenza A (3),picornavirus (26), parainfluenza 1, 2, or 3 (2), RSV (1) andadenovirus (1). Twenty-eight (84%) of viruses were detectedon day 1 after symptom onset and an additional fivepicornaviruses were detected at day 5 after onset inpatients who remained symptomatic. At the day 30 follow-up one patient remained colonised with picornavirus whenstable.

Three time-matched controls had viruses detected;picornavirus (2) and adenovirus (1) although these patientshad no symptoms other than slight nasal congestion orrhinorrhoea. Respiratory viruses detected on day 1–5 ofexacerbations in both cases and controls were included inthe case-control analysis. There were 34 discordant pairsfrom 148 case-control pairs. The odds ratio was 11 (95%CI;3.45–56.07, po0.001). In 33 pairs the case was positive forvirus and the control was negative and in three pairs thecontrol had virus while the case had no virus. Results for thecase control analysis are summarized in Table 2.

Paired conditional logistic regression demonstrated thatnone of the following were predictors of viral detection;age (p ¼ 0.287), gender (p ¼ 0.578), living situation

ARTICLE IN PRESS

Table 3 Characteristics of exacerbations.

Total exacerbations 148Mean time to sampling� 2.4 days (1–12 days)Exacerbations reported bypatients

64%

Pathogens detectedOnly virus 21Only bacteria 30Co-infection 12Atypical organism 3

Virus detected in casesInfluenza A 3Rhinovirus 26Adenovirus 1Parainfluenza 1,2,3 2RSV 1

Bacteria detected in cases at onsetHemophilus influenzae 16Pseudomonas spp. 9Streptococcus pneumoniae 7Moraxella catarrhalis 3Staphylococcus aureus 3Stentrophomonas maltophilia 3Other Gram negatives 8

Atypical organisms on serologyChlamydia pneumoniae 1Chlamydia psittaci 1Mycoplasma pneumoniae 1

Severity of exacerbations�

Mild 38%Moderate 43%Severe 17%Life-threatening 2%

Anthonisen type17

Type I 33%Type II 16%Type III 51%

�Burge and Wedzicha Consensus Report.18

A case-control study of viruses in COPD 2477

(p ¼ 0.328), current smoker or not (p ¼ 0.165), diagnosis ofchronic bronchitis (p ¼ 0.544) or childhood asthma(p ¼ 0.277). On the other hand, less severe disease(measured by FEV1% predicted) was predictive of viralisolation. The odds of having a positive viral PCR increasedby 6% for each 1-unit improvement in FEV1% predicted (OR1.06, [95%CI 1.03–1.09], po0.001). Functional status wasalso predictive PCR positivity with a 34% decrease in theodds of having a positive viral PCR per 1-unit increase in theBODE Index (OR 0.66, [95% CI 0.54–0.80], po0.001). All theabove predictors were analysed using univariate regressionmodels.

Contact patterns with children

Twenty-eight patients (30%) had AECOPD of viral aetiologyconfirmed by respiratory PCR. Eight patients (32%) of thecohort had regular contact (3–7 days per week) with infants

or children less than 10 years of age. Seven (88%) of thesepatients had at least one PCR positive AECOPD. There werefour patients who had multiple viral AECOPD confirmed byPCR, three of the four (75%) had regular contact withchildren.

Association between viral symptoms and ratesof viral detection

A virus was identified on respiratory PCR in 21.0% ofexacerbations overall and 31% of cases with three or moreviral symptoms and an overall score of X4. The individualsymptoms with the highest predictive values were thepresence of rhinorrhoea (Odds ratio [OR] 4.86, [95%CI:1.82–14.31], p ¼ 0.004), and sore throat (OR 2.24, [95%CI:0.93–5.55], p ¼ 0.05). Other symptoms associated with viralinfection were not statistically significant predictors of PCRpositivity in this study; nasal congestion (OR 1.46, [95%CI0.58–3.57], p ¼ 0.37), fever or chills (OR 1.92, [95%CI0.80–4.63], p ¼ 0.11) and headaches or myalgia (OR 1.36,[95%CI 0.57–3.26], p ¼ 0.45).

Bacteria at AECOPD onset

Patients were able to produce spontaneous sputum samplesat the onset of 88 (60%) of AECOPD (Refer SupplementaryTable 1). Bacteria cultured from spontaneous sputumsamples obtained at AECOPD onset of 42 AECOPD (Day 1)are listed in Table 3. The most common organisms wereHemophilus influenzae (16) Streptococcus pneumoniae (7)and Pseudomonas spp. (9). Additional bacterial isolatesobtained at Day 5–7 after AECOPD onset were H. Influenzae(7), S. pneumoniae (1), Pseudomonas spp. (5) Staphylococ-cus aureus (2), and other gram negatives (4).

Viral–bacterial co-infection

Eight (19%) of the 42 AECOPD that had a bacterial isolate onDay 1 also had a virus isolated on respiratory PCR and anadditional four had both viruses and bacteria isolated byDay 5. The viral–bacterial coinfections were Influenza A andS. pneumoniae and H. influenzae (1), Picornavirus withStreptococcus pnuemoniae (3), H. influenzae (4), Pseudo-monas sp. (2) and Enterobacter sp. (1) and adenovirus withS. pneumoniae and Acinetobacter sp. (1). Thirty-six percentof exacerbations in which a virus was detected at onsetdeveloped secondary bacterial infection over the following 7days. In 30 (71%) of AECOPD in which bacteria were isolatedat onset, the patients had also reported some viralsymptoms at AECOPD onset. Overall 78% of AECOPD in whichH. influenzae was isolated in the first 5 days after onsetwere preceded by viral symptoms.

Atypical pneumonia organisms and AECOPD

There were three AECOPD associated with at least a four-fold rise in atypical pneumonia serology titres; C. pneumo-niae (1), C. psittaci (1) and Mycoplasma sp. (1) Theseexacerbations were associated with a positive viral symptomscore, including symptoms of sore throat, rhinorrhea, fever

ARTICLE IN PRESS

A.F. Hutchinson et al.2478

and chills and myalgia. The incidence density of newinfections with atypical organisms was low, at 0.70 per1000 person-weeks. However, AECOPD associated withatypical organisms were severe or life threatening and hada prolonged recovery time.

Community acquired pneumonia (CAP)

This study did not exclude AECOPD complicate by CAP. ChestX-rays were clinically indicated in 26% of AECOPD, (criteriawere; clinical signs of pneumonia, patients attending theemergency department or requiring inpatient admission).Review of these chest X-rays showed that six AECOPD wereassociated with acute collapse and consolidation. Organismsidentified were Influenza A (1), Moraxella catarrhalis andPseudomonas aeruginosa coinfection (1), H. influenzae (1).No organisms were identified in the other three cases.

Clinical severity and inflammatory indices

AECOPD associated with rhinovirus were nine (35%) mild, 15(58%) moderate and two (7%) severe. AECOPD associatedwith RSV and adenovirus were mild, while one parainfluenzaAECOPD was mild, and the other severe complicated byhypercapnic respiratory failure. The three influenza AAECOPD were of moderate severity, with the initial infectionresolving within 2 weeks. At 3 weeks, two of these patientsdeveloped mixed viral and bacterial exacerbations thatrequired inpatient admission.

Serum inflammatory indices were obtained at the onset of121 (82%) of exacerbations. There was an acute rise in serumCRP levels at AECOPD onset compared to baseline. Table 4summarises inflammatory indices according to the labora-tory confirmed pathogens detected at AECOPD onset. As thedistribution of CRP and ESR values was right skewed, datawas log-transformed prior to further analysis, chi squaretests confirmed that the natural log of these markers wasnormally distributed. At AECOPD onset there was an acuteincrease in serum CRP levels whether or not a virus wasdetected. In PCR positive AECOPD there was a 4.5 foldincrease (geometric mean ratio 4.491 [95%CI: 2.256–8.939],po0.001) and in PCR negative a 2.7 fold increase (2.709[95%CI: 1.958–3.747], po0.001).

Table 4 Blood inflammatory indices at onset of AECOPD by pa

Viral PCR positive Bacterial

WCC (109/L) 10.9 (4.9–30.1) 12.36 (7.9–19.2)Neutrophils (109/L) 8.39 (2.1–25.6) 9.7 (3–16.4)Lymphocytes (109/L) 1.74 (0.5–2.8) 1.3 (0.3–3.3)Eosinophils (109/L) 0.14 (0–0.6) 0.12 (0–7)ESR (mm/1 h) 27.9 (9–71) 36.46 (o1–97)CRP (mg/L) 37.6 (o2–160) 65 (o2–341)

Mean and range of values for each test grouped according to pathogWCC, white cell count; ESR, erythrocyte sedimentation rate; CRP, C

Discussion

In this study, discordant case-control pairs were 11 timesmore likely to have the case positive for viral infectionthan the control positive. Only three time-matchedcontrols were positive for virus. The present study demon-strates that infection with respiratory viruses, specificallypicornavirus, parainfluenza and influenza, is strongly asso-ciated with the development of COPD exacerbations(AECOPD).

The need for rigorous infectious disease epidemiologicalstudy designs that can establish whether or not a novel virusis pathogenic in humans has been identified as an importantmethodological issue.36–40 The strength of this community-based, case-control study design is time-matching con-trolled for seasonal fluctuations in viral prevalence. Thispresents a solution to the problem of establishing causationwhen exposure patterns are random, seasonal or unknown.The estimated strength of the association between AECOPDand viral detection will probably change with the preva-lence, type and virulence of virus circulating in a particularcommunity. In the present study, the lower limit of the 95%confidence interval around the odds ratio is greater thanthree, establishing that exposure to respiratory virusesincreases the risk of an AECOPD. The use of controls fromwithin the cohort controlled for both susceptibility to viralinfection and susceptibility to symptoms indicative of anAECOPD.

To establish that respiratory viruses have a causative rolein triggering exacerbations it is essential to determinewhether viruses are present at the onset of AECOPD.Previous studies of AECOPD requiring hospitalizationhave reported an association, but are not able to establishwhether viral infection precedes the development ofAECOPD.41 This is important because finding a virus in anestablished exacerbation (e.g. more than a week old)may reflect a secondary rather than a primary event.Basing the study in the community facilitated earlyidentification of an exacerbation and measurement ofpotential triggers shortly after onset. We found that themajority of viruses were detected in samples obtainedwithin two days of symptom onset and only a smallnumber of viruses (five picornaviruses), could still bedetected in exacerbating patients, 5 days after onset ofviral symptoms.

thogen identified.

Coinfection No pathogenidentified

Total

10.13 (4.9–16.7) 9.9 (5.9–18) 10.7 (4.9–30.1)7.48 (2.5–15.6) 7.3 (3–15.2) 8.2 (2.1–25.6)1.32 (0.8–3.1) 1.5 (0.2–2.9) 1.4 (0.2–3.3)0.28 (0–1) 0.15 (0–0.5) 0.15 (0–1)

32.11 (o1–65) 33.19 (o1–111) 33.02 (o1–111)63 (o2–171) 26 (o2–226) 41 (o2–341)

en detection on microbiological tests.-reactive protein.

ARTICLE IN PRESS

A case-control study of viruses in COPD 2479

Respiratory viruses were detected at the onset of 21% ofAECOPD, and all but one exacerbating patient had clearedthe virus from the upper respiratory tract by day 30. Theoverall detection rates in this study were lower than thosereported in other studies, this may relate to the timebetween symptom onset and patient report of symptoms orthe detection technique used or the number of virusescirculating through our community at that time. Thesensitivity and specificity of respiratory PCR for thedetection of respiratory virus infection is better thantraditional methods using viral culture and serology,particularly for viruses such as picornavirus.22,42 Thereremains some controversy over whether to use nasophar-yngeal swabs,43 nasal lavage samples44 or induced sputumsamples, for viral detection using PCR techniques. We choseto use nasopharyngeal swabs, as these are the standardspecimens used by our diagnostic laboratory, are easilyobtained in the community setting, and have the advantageof capturing virus in the upper respiratory tract shortly afterexposure. In other studies viral detection in nasopharyngealswabs is well correlated with infection.22,45 The strength ofthe case control methodology is that while some viralinfections may have gone undetected using this technique,we were still able to establish a strong association betweenviral detection and the onset of AECOPD.

In our study picornavirus, RSV and adenovirus weredetected in low numbers in stable COPD. Reported rates ofvirus detection in stable COPD vary, with chronic carriage ofRSV, picornavirus and adenovirus identified in differentpatient cohorts.11,12,42 The London COPD cohort detectedvirus in 39% of stable COPD patients, with RSV accounting for23% of these.12 In contrast, a recent study found that mostRSV infections were associated with respiratory symptomsand measurable immune responses and did not find evidencefor persistent RSV infection in stable COPD.45 As it is possiblethat viral mRNA may be retained for lengthy periods ininduced sputum, in the absence of active infection this mayalso influence the higher detection rate reported in otherstudies.

Consistent with recent studies using PCR techniques,picornavirus was the most common virus associated withAECOPD in our cohort and was detected throughout theyear.43,46 There were very low levels of influenza infectiondetected in the cohort, reflecting low to normal seasonalinfluenza activity and that 98% of the cohort was vaccinatedagainst influenza. Influenza surveillance reports indicatethat in 2003–2005, there was good vaccine coverage forcirculating influenza strains in Australia.47,48 Detection ratesof both picornavirus and parainfluenza increased in syn-chrony with increased detection of these viruses in thegeneral community, as reported by influenza surveillance.47

Upper respiratory symptoms of viral infection (rhinorrhoeaand sore throat) were predictive of PCR positivity whilesystemic symptoms were not, probably because infectionwith viruses such as influenza, that typically cause severesystemic illness were uncommon.

There was a trend for cohort patients with less severedisease and better functional capacity to have virusisolated. Possibly patients with better functional statuswere more active in the community, had higher rates ofsocial contact, resulting in greater exposure to circulatingrespiratory viruses. It appears virus exposure rates, rather

than severe lung function impairment, determined thelikelihood of a viral AECOPD. This finding is supported byHurst’s study that found that frequent exacerbations wereassociated with more frequent colds, rather than anincreased susceptibility to exacerbate once infected witha cold.14 It is noteworthy that patients in our study withrepeated viral infections, tended to be those providingchildcare to infants or young children, indicating thatchildren were one source of repeated exposure to circulat-ing respiratory viruses. This is consistent with studies thathave found a correlation between the rates of admission foracute respiratory illness in children and hospital admissionfor AECOPD,49,50 and that older adults and children areusually infected with the same circulating virus strain.51

Laboratory confirmed viral infection preceded the isola-tion of bacteria in 29% of AECOPD in which bacteria wereisolated in the first five days after exacerbation onset. Theassociation between influenza virus and bacterial infectionis well documented.52–54 New evidence is emerging thatother respiratory viruses interact with bacteria in the lowerairway in a virus specific, cell-type specific manner.55 Recentclinical studies indicate that Picornaviridae may interactwith new or colonising bacteria in the lower airway,resulting in increased bacterial load and heightenedinflammatory responses.14,56 Measuring viral–bacterial in-teractions was not the primary aim of this study, however wedid observe that confirmed viral infection or probable viralinfection (based on symptoms) preceded 71% of laboratoryconfirmed bacterial infections in the first week afterAECOPD onset and 78% of H. influenzae infections. Thishighlights the importance of studying the onset of AECOPD inthe community setting, where initial exposures and infec-tive triggers can be accurately measured. It is clear thatstudying hospitalised AECOPD alone will underestimate theimportance of viral infection in triggering AECOPD and theeffect of viral infection on host susceptibility to bacterialinfection. Further research should address the impact ofviral infections on the delicate balance between colonisingbacteria and host immunity

Given the feasibility of reaching patients rapidly after theonset of AECOPD and the suggestive evidence that that earlytherapeutic intervention may reduce burden of disease,57 itwould be highly desirable to accurately and rapidlydifferentiate between viral and other causes of AECOPD.The capacity to identify exacerbations early and differenti-ate between viral pathogens by PCR, suggests the possibilityof prospective trials of newer antiviral therapies in themanagement of AECOPD.58 Community-based studies areneeded to evaluate whether effective early management ofviral exacerbations may decrease the incidence of second-ary bacterial infection and the overall severity and durationof the AECOPD episode.

This study demonstrates that the presence of virus inupper airway secretions is strongly associated with thedevelopment of AECOPD. The strength of the methodologi-cal approach used in this study is that it controlled forrandom and seasonal fluctuations in the levels of viruscirculating in the community. By conducting this study in thecommunity, we have shown that respiratory viruses cantrigger the onset of AECOPD and that picornaviruses are themost common viral trigger for AECOPD. These data supportthe causative role of viruses in AECOPD.

ARTICLE IN PRESS

A.F. Hutchinson et al.2480

Appendix A. Supplementary materials

Supplementary data associated with this article can befound in the online version at doi:10.1016/j.rmed.2007.07.015.

Conflict of interest and sources of funding

Funding for Staff salaries

Research staff received support as follows: Anil Ghimire:Master of Medicine Student, Melbourne University, Austra-lian Aid ‘‘AusAid’’ Scholarship in 2003.

Anastasia Hutchinson: PhD Student, Melbourne University.NHMRC Centre for Clinical Research Excellence in InfectiousDiseases (CCREID) funded scholarship in 2004 and NationalHealth and Medical Research Council (NHMRC Australia),Dora Lush PhD Scholarship 2005 onwards.

Michelle Thompson: Research Scientist/Study Coordina-tor, employed by University of Melbourne PharmacologyDepartment, Collaborative Research Centre for ChronicInflammatory Diseases (CRC-CID), Parkville, AUSTRALIA.

Steve Bozinovski and Ross Vlahos are supported by theNHMRC (Australia) and CRC-CID.

Funding for Diagnostic Tests was received from Tubercu-losis Association; Edgar Tatnall Respiratory Research Fundand the Victor Hurley Scholarship Fund.

The authors declare that they had no conflict of interest;preparation of this paper was independent from fundingbodies.

References

1. Pauwels RA, Rabe KF. Burden and clinical features of chronicobstructive pulmonary disease (COPD). Lancet 2004;364:613–9.

2. Seemungal TA, Donaldson GC, Paul EA, Bestall JC, Jeffries DJ,Wedzicha JA. Effect of exacerbation on quality of life inpatients with chronic obstructive pulmonary disease. Am JRespir Crit Care Med 1998;157:1418–22.

3. Donaldson GC, Seemungal TA, Bhowmik A, Wedzicha JA.Relationship between exacerbation frequency and lung functiondecline in chronic obstructive pulmonary disease. Thorax2002;57:847–52.

4. Glezen WP, Greenberg SB, Atmar RL, Piedra PA, Couch RB.Impact of respiratory virus infections on persons with chronicunderlying conditions. J Am Med Assoc 2000;283:499–505.

5. Miravitilles M, Murio C, Guerrero T, Gisbert R. DAFNE StudyGroup Pharmaco-economic evaluation of acute exacerbations ofchronic bronchitis and COPD. Chest 2002;121:1449–55.

6. Hilleman DE, Dewan N, Malesker M, Friedman M. Pharmaco-economic evaluation of COPD. Chest 2000;118:1278–85.

7. Strassels S. The costs of treating COPD in the United States.Chest 2001;119:344–352,.

8. Sethi S, Murphy TF. Bacterial infection in chronic obstructivepulmonary disease in 2000: a state-of-the-art review. ClinMicrobiol Rev 2001;14:336–63.

9. Pope CA. What do epidemiologic findings tell us about healtheffects of environmental aerosols. J Aeorsol Med 2000;13:335–54.

10. Tillie-Leblond I, Marquette CH, et al. Pulmonary embolism inpatients with unexplained exacerbation of chronic obstructivepulmonary disease: prevalence and risk factors. Ann Intern Med2006;144:390–6.

11. Rohde G, Wiethege A, Borg I, Kauth M, Bauer TT, Gillissen A,et al. Respiratory viruses in exacerbations of chronic obstruc-tive pulmonary disease requiring hospitalisation: a case-controlstudy. Thorax 2003;58:37–42.

12. Seemungal T, Harper-Owen R, Bhowmik A, Moric I, Sanderson G,Message S, et al. Respiratory viruses, symptoms, and inflam-matory markers in acute exacerbations and stable chronicobstructive pulmonary disease. Am J Respir Crit Care Med2001;164:1618–23.

13. Papi A, Bellettato M, Braccioni F, Romagnoli M, Casolari P,Caramori G, et al. Infections and airway inflammation in chronicobstructive pulmonary disease severe exacerbations. Am JRespir Crit Care Med 2006;173:1114–21.

14. Hurst JR, Wilkinson TMA, Perera WR, Wedzicha JA. Epidemio-logical relationships between the common cold and exacerba-tion frequency in COPD. Eur Respir J 2005;26:846–52.

15. Brand PL, Quanjer PhH, Postma DS. Interpretation of broncho-dilator response in patients with chronic obstructive airwaysdisease. Thorax 1992;47:429–36.

16. American Thoracic Society. COPD guidelines: /www.thoracic.orgS. In: ATS, 2005.

17. Anthonisen NR, Manfreda J, Warren CPW, Hershfield ES, HardingGKM, Nelson NA. Antibiotic therapy in exacerbations of chronicobstructive pulmonary Disease. Ann Intern Med 1987;106:196–204.

18. Burge S, Wedzicha JA. COPD exacerbations: definitions andclassifications. Eur Respir J Suppl 2003;41:46s–53s.

19. Enright PL. The six-minute walk test. Respir Care 2003;48:783–5.

20. Bestall JC, Paul EA, Garrod R, Garnham R, Jones PW, WedzichaJA. Usefulness of the Medical Research Council (MRC) dyspnoeascale as a measure of disability in patients with chronicobstructive pulmonary disease. Thorax 1999;54:581–6.

21. Celli BR, Cote CG, Marin JM, Casanova C, Montes de Oca M,Mendez RA, et al. The body-mass index, airflow obstruction,dyspnea, and exercise capacity index in chronic obstructivepulmonary disease. N Eng J Med 2004;350:1005–12.

22. Druce J, Tran T, Kelly H, Kaye M, Chibo D, Kostecki R, et al.Laboratory diagnosis and surveillance of human respiratoryviruses by PCR in Victoria, Australia, 2002–2003. J Med Virol2005;75:122–9.

23. Rose NR, De MacArio EC, Folds JD, Lane HC, Nakanura RM,editors. Manual of clinical laboratory immunology. Washington,DC: ASM Press; 1997.

24. Grist NR, Ross CAC, Bell EJ, Stott EJ. In: Diagnostic methods inclinical virology. Oxford: Blackwell distributed by Lippioncott,J.B.; 1966. p. 54–71 [Chapter 6].

25. Ambrosino N, Scano G. Dyspnoea and its measurement. Breathe2004;1:101–7.

26. Selim AJ, Ren XS, Fincke G, Rogers W, Lee A, Kazis L. Asymptom-based measure of the severity of chronic lungdisease*. Results from the Veterans Health Study. Chest1997;111:1607–14.

27. Boivin G, Hardy I, Tellier G, Maziade J. Predicting influenzainfections during epidemics with use of a clinical casedefinition. Clin Infect Dis 2000;31:1166–9.

28. Gwaltney Jr JM, Hendley JO, Patrie JT. Symptom severitypatterns in experimental common colds and their usefulness intiming onset of illness in natural colds. Clin Infect Dis2003;36:714–23.

29. Monto AS, Bramley TJ, Sarnes M. Development of a predictiveindex for picornavirus infections. Clin Infect Dis 2003;36:253–8.

30. Monto AS, Gravenstein S, Elliott M, Colopy M, Schweinle J.Clinical signs and symptoms predicting influenza infection. ArchIntern Med 2000;160:3243–7.

31. Neuzil KM, O’Connor TZ, Gorse GJ, Nichol KL. Recognizinginfluenza in older patients with chronic obstructive pulmonary

ARTICLE IN PRESS

A case-control study of viruses in COPD 2481

disease who have received influenza vaccine. Clin Infect Dis2003;36:169–74.

32. Norman DC. Fever in the elderly. Clin Infect Dis 2000;31:148–51.

33. Roghmann MC, Warner J, Mackowiak PA. The relationship betweenage and fever magnitude. Am J Med Sci 2001;322:68–70.

34. Sackett DL, Haynes RB, Guyatt GH, Tugwell P. Clinicalepidemiology. A basic science for clinical medicine, 2nd ed.Boston/Toronto/London: Little, Brown and Company; 1991.

35. Statacorp. Stata statistical software release 8.2. CollegeStation Texas: Stata Corporation; 2003.

36. Fouchier RAM, Rimmelzwaan GF, Kuiken T, Osterhaus ADE.Newer respiratory virus infections: human metapneumovirus,avian influenza virus and human coronaviruses. Curr Opin InfectDis 2005;18:141–6.

37. Bastein N, Brandt K, Dust K, Ward D, Li Y. Human Bocavirusinfection, Canada. Emerging Infect Dis 2006;12:848–50.

38. Fredricks DN, Relman DA. Sequence-based identification ofmicrobial pathogens: a reconsideration of Koch’s postulates.Clin Microbiol Rev 1996;9:18–33.

39. Martinello RA, Esper F, Weibel C, Ferguson D, Landry ML, KahnJS. Human metapneumovirus and exacerbations of chronicobstructive pulmonary disease. J Infection 2005;53:248–54.

40. Allander T, Tammi MT, Eriksson M, Bjerkner A, Tiveljung-LindellA, Andersson B. Cloning of a human parvovirus by molecularscreening of respiratory samples. PNAS 2005;102:12891–6.

41. Tan WC, Xiang X, Qiu D, Ng TP, Lam SF, Hegele RG. Epidemiologyof respiratory viruses in patients hospitalised with near-fatalasthma, acute exacerbations of asthma, or chronic obstructivepulmonary disease. Am J Med 2003;115:272–7.

42. Beckham J, Cadena A, Lin J, Piedra PA, Glezen WP, GreenbergSB, et al. Respiratory viral infections in patients with chronic,obstructive pulmonary disease. J Infect 2005;50:322–30.

43. Seemungal TA, Harper-Owen R, Bhowmik A, Jeffries DJ,Wedzicha JA. Detection of rhinovirus in induced sputum atexacerbation of chronic obstructive pulmonary disease. EurRespir J 2000;16:677–83.

44. Rohde G. Relevance of human metapneumovirus in exacerba-tions of COPD. Respir Res 2005;6:150.

45. Falsey AR, Formica MA, Hennessey PA, Criddle MM, SullenderWM, Walsh EE. Detection of respiratory syncytial virus in adults

with chronic obstructive pulmonary disease. Am J Respir CritCare Med 2006;173:639–43.

46. Greenberg SB. Respiratory viral infections in high-risk patients.Am J Respir Crit Care Med 2004;170:1142–3.

47. Victorian Infectious Diseases Laboratory (VIDRL). Fortnightlysurveillance reports /S, 2005.

48. Turner JL, Fielding JE, Clothier HJ, Kelly HA. Influenzasurveillance in Victoria, 2005. Commun Dis Intell 2005;30:137–43.

49. Kidney J, McManus T, Coyle PV. Exacerbations of chronicobstructive pulmonary disease. Thorax 2002;57:753–4.

50. McManus TE, Coyle PV, Kidney JC. Childhood respiratoryinfections and hospital admissions for COPD. Respir Med2006;3:512–8.

51. Zambon M, Stockton J, Clewley J, Fleming D. Contribution ofinfluenza and respiratory syncytial virus to community cases ofinfluenza-like illness: an observational study. Lancet 2001;358(9291):1410–6.

52. Peltola VT, Mccullers J. Respiratory viruses predisposing tobacterial infections: role of neuraminidase. Pediatr Infect Dis J2004;23:S87–97.

53. Engelich G, White M, Hartshorn KL. Role of the respiratory burstin so-operative reduction in neutrophil survival by influenza Avirus and Escherichia coli. J Med Microbiol 2002;51:484–90.

54. Smith CB, Golden C, Klauber MR, Kanner R, Renzetti A.Interactions between viruses and bacteria in patients withchronic bronchitis. J Infect Dis 1976;134:552–61.

55. Avadhanula V, Rodriguez CA, De Vincenzo JP, Wang Y, Webby RJ,Ulett GC, et al. Respiratory viruses augment adhesion ofbacterial pathogens to respiratory epithelium in a viral-speciesand cell type dependent manner. J Virol 2005;80:1629–36.

56. Wilkinson TM, Hurst JR, Perera WR, Wilks M, Donaldson G,Wedzicha JA. Effect of interactions between lower airwaybacterial and rhinovirus infection in exacerbations of COPD.Chest 2006;129:317–24.

57. Wilkinson TMA, Donaldson GC, Hurst JR, Seemungal TAR,Wedzicha JA. Early therapy improves outcomes of exacerba-tions of chronic obstructive pulmonary disease. Am J Respir CritCare Med 2004;169:1298–303.

58. Remero JR. Pleconaril: a novel anti-picornaviral drug. ExpertOpin Investigat Drugs 2001;10:369–79.