Advertisement

Estimating the contribution of respiratory pathogens to acute exacerbations of COPD using routine data

Open AccessPublished:January 24, 2023DOI:https://doi.org/10.1016/j.jinf.2023.01.012

      Highlights

      • Linkage of microbiology data to national admissions data over a two year period
      • Respiratory viruses are detected in a substantial proportion of COPD exacerbations
      • Increased respiratory viral testing could aid antimicrobial stewardship efforts
      • ICD codes have low sensitivity in identifying respiratory pathogens in routine data
      • Large-scale data linkage delineates the burden of specific pathogens with greater accuracy

      Abstract

      Objectives

      To characterise microbiology testing and results associated with emergency admissions for acute exacerbation of COPD (AECOPD), and determine the accuracy of ICD-10 codes in retrospectively identifying laboratory-confirmed respiratory pathogens in this setting.

      Methods

      Using person-level data from the Secure Anonymised Information Linkage Databank in Wales, we extracted emergency admissions for COPD from 1/12/2016 to 30/11/2018 and undertook linkage of admissions data to microbiology data to identify laboratory-confirmed infection. We further used these data to assess the accuracy of pathogen-specific ICD-10 codes.

      Results

      We analysed data from 15,950 people who had 25,715 emergency admissions for COPD over the two-year period. 99.5% of admissions could be linked to a laboratory test within 7 days of admission date. Sputum was collected in 5,013 (19.5%) of admissions, and respiratory virus testing in 1,219 (4.7%). Where respiratory virus testing was undertaken, 46.7% returned any positive result. Influenza was the virus most frequently detected, in 21.5% of admissions where testing was conducted. ICD-10 codes exhibited low sensitivity in detecting laboratory-confirmed respiratory pathogens.

      Conclusions

      In people admitted to hospital with AECOPD, increased testing for respiratory viruses could enable more effective antibiotic stewardship and isolation of cases. Linkage with microbiology data achieves more accurate and reliable case definitions.

      Keywords

      Introduction

      Chronic obstructive pulmonary disease (COPD) is characterised by reduced lung function, resulting in progressive dyspnoea and acute exacerbations which necessitate hospital admission if severe. The majority of acute exacerbations of COPD (AECOPDs) are considered to have an infective aetiology,[
      • Sapey E
      • Stockley RA.
      COPD exacerbations·2: Aetiology.
      ] and antibiotic use for this indication is frequent, even though a significant proportion are likely to have a viral trigger.
      To develop, implement and evaluate interventions to reduce the burden of respiratory tract infections, pathogen-specific data establishing the frequency and consequences of infection in different populations is imperative. Thus far, the focus of many studies examining the burden of specific respiratory pathogens has been on children.[
      • Wahl B
      • O'Brien KL
      • Greenbaum A
      • et al.
      Burden of Streptococcus pneumoniae and Haemophilus influenzae type b disease in children in the era of conjugate vaccines: global, regional, and national estimates for 2000–15.
      ,
      • Nair H
      • Nokes DJ
      • Gessner BD
      • et al.
      Global burden of acute lower respiratory infections due to respiratory syncytial virus in young children: a systematic review and meta-analysis.
      ,
      • Troeger C
      • Blacker B
      • Khalil IA
      • et al.
      Estimates of the global, regional, and national morbidity, mortality, and aetiologies of lower respiratory infections in 195 countries, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016.
      ] Impact on other risk groups, such as those with COPD, requires further study, especially since acute exacerbations of COPD are one of the commonest reasons for emergency hospital admission in the UK,[
      National Institute for Health and Care Excellence.
      ] resulting in substantial morbidity, mortality and healthcare cost each year.
      International Statistical Classification of Diseases and Related Health Problems (ICD) codes relating to respiratory pathogens are readily available in databases of routinely collected electronic health data and have therefore often been used for research purposes.[
      • Hauge SH
      • Bakken IJ
      • de Blasio BF
      • Håberg SE.
      Risk conditions in children hospitalized with influenza in Norway.
      ,
      • Hungerford D
      • Ibarz-Pavon A
      • Cleary P
      • French N.
      Influenza-associated hospitalisation, vaccine uptake and socioeconomic deprivation in an English city region: An ecological study.
      ,
      • Moss JWE
      • Davidson C
      • Mattock R
      • Gibbons I
      • Mealing S
      • Carroll S.
      Quantifying the direct secondary health care cost of seasonal influenza in England.
      ,
      • San-Román-Montero JM
      • Gil Prieto R
      • Gallardo Pino C
      • Hinojosa Mena J
      • Zapatero Gaviria A
      • Gil De Miguel A.
      Inpatient hospital fatality related to coding (ICD-9-CM) of the influenza diagnosis in Spain (2009-2015).
      ,
      • Tong S
      • Amand C
      • Kieffer A
      • Kyaw MH.
      Incidence of Respiratory Syncytial Virus Related Health Care Utilization in The United States.
      ,
      • Huh K
      • Kim YE
      • Ji W
      • et al.
      Decrease in hospital admissions for respiratory diseases during the COVID-19 pandemic: a nationwide claims study.
      ] However, the accuracy of diagnostic coding is likely to vary depending on local practices, and not all pathogens map to a specific ICD code.
      Large-scale linkage of microbiology data to other routinely collected health data, where possible, is likely to be the more accurate data source in establishing the burden of particular pathogens. Researchers who have had access to microbiology data report that they would have grossly underestimated respiratory pathogen-related hospital admissions had they relied on diagnosis codes alone.[
      • Hardelid P
      • Verfuerden M
      • McMenamin J
      • Gilbert R.
      Risk factors for admission to hospital with laboratory-confirmed influenza in young children: birth cohort study.
      ,
      • Nicholson KG
      • McNally T
      • Silverman M
      • Simons P
      • Stockton JD
      • Zambon MC.
      Rates of hospitalisation for influenza, respiratory syncytial virus and human metapneumovirus among infants and young children.
      ,
      • Lim FJ
      • Blyth CC
      • Fathima P
      • de Klerk N
      • Moore HC.
      Record linkage study of the pathogen-specific burden of respiratory viruses in children.
      ,
      • Alchikh M
      • Conrad T
      • Hoppe C
      • et al.
      Are we missing respiratory viral infections in infants and children? Comparison of a hospital-based quality management system with standard of care.
      ] However, such linkage is currently not possible in many regions, rendering person-level analyses using microbiology data unattainable. Further, even in regions where such linkage is possible, the likelihood of successful linkage may vary depending on age, duration of hospitalisation, ethnic group and rurality.[
      • Moore HC
      • de Klerk N
      • Keil AD
      • et al.
      Use of data linkage to investigate the aetiology of acute lower respiratory infection hospitalisations in children.
      ]
      In this national study, we utilise linkable, routinely collected microbiology and healthcare utilisation data covering all admissions to National Health Service (NHS) hospitals in Wales over a 2 year period.
      Our aims were: (i) to illustrate the utility of a national microbiology dataset that can be linked to other routinely collected electronic health data; (ii) to quantify the proportion of emergency admissions for COPD undergoing testing for respiratory pathogens, and the proportion tested where viral or bacterial pathogens were identified; and (iii) to determine the accuracy of using ICD-10 diagnosis codes to identify laboratory-confirmed respiratory pathogens associated with COPD exacerbations.

      Methods

      We accessed complete coverage, person-level, anonymised datasets from the Secure Anonymised Information Linkage (SAIL) Databank,[
      • Jones KH
      • Ford D v
      • Thompson S
      • Lyons RA.
      A profile of the SAIL databank on the UK secure research platform.
      ] which receives routinely collected data from all NHS hospitals in Wales, to examine all emergency hospital admissions due to COPD over a two-year period (1/12/2016 to 30/11/2018). This report of our findings was prepared according to the REporting of studies Conducted using Observational Routinely-collected Data (RECORD) statement.[
      • Benchimol EI
      • Smeeth L
      • Guttmann A
      • et al.
      The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement.
      ] The study was approved by the SAIL Information Governance Review Panel (project number 0996).

      Data sources, population, and case definitions

      Hospital admissions data were extracted from the Patient Episodes Database for Wales (PEDW), a dataset which includes diagnosis codes for all NHS hospital admissions in Wales. Emergency admissions for COPD were defined as those with an ICD-10 COPD code (J43, J44) recorded in the first or second position in any Finished Consultant Episode (FCE) in the population aged ≥35 at date of admission.
      We defined two methods of classifying emergency COPD admissions as being associated with specific respiratory pathogens: (1) using pathogen-specific ICD-10 diagnosis codes (the full code set used is available in supplementary material S1), and (2) using microbiology test results. Microbiology data were extracted from the Welsh Results Reporting Service (WRRS), which receives results from all NHS hospital laboratories in Wales. Person-level linkage with hospital admissions data was possible using the Anonymised Linkage Field, which has been described previously.[
      • Lyons RA
      • Jones KH
      • John G
      • et al.
      The SAIL databank: Linking multiple health and social care datasets.
      ] We included individuals with high quality linkage (where there was either exact matching of NHS number or matching probability ≥ 90% (using name, sex, date of birth, postcode)). Laboratory diagnosis of respiratory viruses was all based on reverse transcription polymerase chain reaction (RT-PCR) and PCR testing, but it could not be established whether nasal or throat swabs were collected. Bacteria were detected by culture of sputum samples.
      Pathogens of interest were influenza (A and B), respiratory syncytial virus (RSV), parainfluenza, rhinovirus, human metapneumovirus (hMPV), adenovirus, enterovirus, Haemophilus influenzae, Streptococcus pneumoniae, Moraxella catarrhalis and Pseudomonas aeruginosa. The choice of respiratory viruses was dictated by those included on multiplex panels over the time period studied.
      We used laboratory data to determine the proportion of COPD admissions where testing for bacterial and/or viral respiratory pathogens was conducted and estimated the proportion of COPD admissions with laboratory-confirmed infection (with the total number tested as our denominator). We defined admissions associated with laboratory-confirmed infection as those where the collection date of the specimen with a positive result was within (+/-) 7 days of hospital admission date. We assessed the seasonality of virus-associated admissions using methods adapted from Li et al,[
      • Li Y
      • Hodgson D
      • Wang X
      • Atkins KE
      • Feikin DR
      • Nair H.
      Respiratory syncytial virus seasonality and prevention strategy planning for passive immunisation of infants in low-income and middle-income countries: a modelling study.
      ] where seasonality exists if at least 75% of the annual total of positive cases occur in ≤ 5 consecutive months in a year.

      Accuracy of ICD-10 coding

      Using COPD admissions associated with laboratory-confirmed infection as our reference standard, we assessed the accuracy of ICD-10 diagnosis codes in identifying respiratory pathogens associated with COPD admissions. Measures of accuracy were positive predictive value (PPV), negative predictive value (NPV), sensitivity, and specificity, each with a 95% confidence interval. The sampling frame for these analyses was all COPD admissions where testing for the pathogen in question was conducted within (+/-) 7 days of admission. PPV was calculated as the proportion of COPD admissions with a pathogen-specific ICD-10 code that had a laboratory-confirmed respiratory pathogen on testing. NPV was calculated as the proportion of COPD admissions without a pathogen-specific ICD-10 code when there was a negative respiratory pathogen test. Sensitivity was calculated as the proportion of COPD admissions with a laboratory-confirmed respiratory pathogen that had a pathogen-specific ICD-10 code. Specificity was calculated as the proportion of COPD admissions with a negative respiratory pathogen test and without a pathogen-specific ICD-10 code.

      Results

      We analysed data from 15,950 people who had 25,715 emergency admissions for COPD over the two year period 1/12/2016 to 30/11/2018. Median age at admission was 73 (interquartile range 66 – 80). Emergency admissions for COPD were fewer over the summer months (figure 1), and peaked at a mean of 1,480 admissions in the month of January.
      Figure 1
      Figure 1Mean count of monthly emergency admissions for COPD in Wales, by age group
      99.5% (25,596/25,715) of emergency COPD admissions could be linked to a laboratory record with a specimen collection date within (+/-) 7 days of the admission date. The proportion of admissions successfully linked to laboratory data was >99% in the majority of NHS Providers in Wales (supplementary material S2).

      Respiratory pathogens associated with emergency COPD admissions: laboratory data

      A specimen for sputum culture was collected in 5,013 (19.5%) out of 25,715 COPD admissions. Of those admissions where a sputum sample was collected, 1,232 (24.6%) were associated with growth of one of the aforementioned bacteria. The most frequently observed bacteria was H.influenzae, which was cultured in 16.7% of admissions where a sputum sample was collected. The other bacteria under investigation all occurred at an overall percent positivity of <5% (table 1).
      Table 1Respiratory pathogen testing and results for specimens associated with emergency admission for COPD
      Respiratory pathogenCount of COPD admissions testedCount of COPD admissions positive (percent positive
      Percent positive = detected on laboratory test within 7 days of admission as a percentage of the total tested
      )
      H.Influenzae5,013837 (16.7%)
      S.Pneumoniae5,013154 (3.1%)
      M.Catarrhalis5,013131 (2.6%)
      P.Aeruginosa5,013200 (4.0%)
      Influenza1,219262 (21.5%)
      Rhinovirus1,208150 (12.4%)
      RSV1,21483 (6.8%)
      Parainfluenza1,20856 (4.6%)
      hMPV1,20863 (5.2%)
      Adenovirus1,00710 (1.0%)
      Enterovirus1,00710 (1.0%)
      low asterisk Percent positive = detected on laboratory test within 7 days of admission as a percentage of the total tested
      A specimen for respiratory virus testing was collected in 1,219 (4.7%) of COPD admissions, and of those admissions associated with any respiratory virus testing, 569 (46.7%) returned any positive result. The most frequently detected virus was influenza, identified in 21.5% of admissions where testing was carried out, followed by rhinovirus (12.4%), RSV (6.8%) and hMPV (5.2%).
      Of the 491 admissions where both sputum collection and viral testing was carried out, 41 (8%) admissions were associated with both bacterial growth in sputum and a positive respiratory virus test.
      Testing frequency and percent positive rate varied by pathogen (table 1), with sputum collection more frequent than viral testing, and the highest percent positive levels seen for influenza, rhinovirus and H.influenzae. There was also marked variation in testing frequency and percent positivity by month of admission for the viral pathogens, with influenza percent positivity ranging from 0% in the UK summer months, to > 30% in January and February (table 2). Over 75% of the total influenza-associated COPD admissions occurred over the UK winter months (December, January, February). RSV and hMPV-associated COPD admissions also showed seasonality (>75% of total admissions occurring over November to January for RSV, and December to March for hMPV), in contrast to admissions associated with rhinovirus, parainfluenza, and the bacterial pathogens. Counts of positive adenovirus and enterovirus tests were small, and thus not analysed by month.
      Table 2Respiratory pathogen testing, results and percent positive rate for specimens associated with emergency admission for COPD, by month of admission (1/12/2016 to 30/11/2018 data combined)
      Month of admission with COPDInfluenza (percent positive
      Percent positive = detected on laboratory test within 7 days of admission as a percentage of the total tested
      )
      RhinovirusRSVParainfluenzahMPVHINFSPNEMCATPAER
      January109/338 (32.2%)30/333 (9.0%)29/333 (8.7%)6/333 (1.8%)21/333 (6.3%)119/578 (20.6%)19/578 (3.3%)16/578 (2.8%)16/578 (2.8%)
      February70/202 (34.7%)14/202 (6.9%)5/202 (2.5%)<5/202 (0.5 – 1.0%)8/202 (3.9%)82/491 (16.7%)15/491 (3.1%)7/491 (1.4%)15/491 (3.1%)
      March32/122 (26.2%)11/122 (9.0%)<5/122 (0.8 – 3.3%)<5/122 (0.8 - 3.3%)5/122 (4.1%)77/440 (17.5%)15/440 (3.4%)11/440 (2.5%)15/440 (3.4%)
      April11/72 (15.2%)14/72 (19.4%)<5/72 (1.4 – 5.6%)14/72 (19.4%)<5/72 (1.4 - 5.6%)73/405 (18.0%)10/405 (2.5%)13/405 (3.2%)13/405 (3.2%)
      May<5/49 (2.0 – 8.2%)5/49 (10.2%)0/49<5/49 (2.0 – 8.1%)<5/49 (2.0 – 8.1%)61/376 (16.2%)10/376 (2.7%)11/376 (2.9%)13/376 (3.5%)
      June0/427/42 (16.7%)0/425/42 (11.9%)<5/42 (2.4 – 9.5%)50/358 (14.0%)10/358 (2.8%)5/358 (1.4%)20/358 (5.6%)
      July0/29<5/29 (3.4 – 13.8%)0/29<5/29 (3.4 – 13.8%)<5/29 (3.4 – 13.8%)66/334 (19.8%)5/334 (1.5%)7/334 (2.1%)14/334 (4.2%)
      August0/235/23 (21.7%)0/23<5/23 (4.3 – 17.4%)<5/23 (4.3 – 17.4%)50/307 (16.3%)6/307 (2.0%)7/307 (1.0%)19/307 (5.8%)
      September0/3110/31 (32.3%)<5/31 (3.2 – 12.9%)<5/31 (3.2 – 12.9%)<5/31 (3.2 – 12.9%)53/344 (15.4%)8/344 (2.3%)6/344 (1.7%)22/344 (6.4%)
      October<5/71 (1.4 – 5.6%)22/69 (31.9%)<5/71 (1.4 – 5.6%)5/69 (7.2%)<5/69 (1.4 – 5.8%)66/394 (16.8%)8/394 (2.0%)9/394 (2.3%)17/394 (4.3%)
      November<5/86 (1.2 – 4.7%)8/82 (9.8%)9/86 (10.5%)<5/82 (1.2 – 4.9%)<5/82 (1.2 – 4.9%)63/471 (13.4%)21/471 (4.5%)19/471 (4.0%)22/471 (4.7%)
      December30/154 (19.5%)20/154 (13.0%)32/154 (20.8%)7/154 (4.5%)14/154 (9.1%)87/515 (16.9%)27/515 (5.2%)20/515 (3.9%)14/515 (2.7%)
      Small counts have been masked as <5.
      HINF = H.Influenzae, SPNE = S.Pneumoniae, MCAT = M.Catarrhalis, PAER = P.Aeruginosa
      low asterisk Percent positive = detected on laboratory test within 7 days of admission as a percentage of the total tested

      Accuracy of ICD-10 diagnosis codes in detecting respiratory pathogens associated with emergency COPD admissions

      The ICD-10 diagnosis codes used to identify influenza-associated COPD admissions performed best in our data, compared to those used for other pathogens, with a sensitivity of 59.2% (95% confidence interval 52.9 – 65.1%), PPV of 86.1% (80.6 – 90.2%), NPV of 89.7% (88.3 – 91.0%) and specificity of 97.4% (96.2 – 98.3%) (table 3).
      Table 3Contingency table showing accuracy of influenza ICD-10 codes associated with emergency admission for COPD, as compared to laboratory test result
      Laboratory test
      DiagnosisPositiveNegativeTotal
      ICD-10 code present15525180
      ICD-10 code absent1079321039
      Total2629571,219
      Sensitivity = 59.2% (52.9 – 65.1%); Specificity = 97.4% (96.2 – 98.3%); PPV = 86.1% (80.6 – 90.2%); NPV = 89.7% (88.3 – 91.0%)
      Sensitivities of the ICD-10 codes related to the other respiratory pathogens were substantially lower, although those of the bacteria were generally higher than the non-influenza viruses (supplementary material S3). P.aeruginosa ICD-10 codes had the highest sensitivity of all the non-influenza pathogens (42.5% (95% CI 35.6 – 49.7%), but also had a lower PPV (48.0% (41.6 – 54.5%)), compared to a PPV >80% for all other pathogens where PPV can be reported; table 4). Specificities were high (>95%) for all pathogens where this can be reported.
      Table 4Contingency table showing accuracy of P.aeruginosa ICD-10 codes associated with emergency admission for COPD, as compared to laboratory test result
      Laboratory test
      DiagnosisPositiveNegativeTotal
      ICD-10 code present8592177
      ICD-10 code absent1154,7214836
      Total2004,8135,013
      Sensitivity = 42.5% (35.6 – 49.7%); Specificity = 98.1% (97.7 – 98.5%); PPV = 48.0% (41.6 – 54.5%); NPV = 97.6% (97.3 – 97.9%)
      Due to low counts in the contingency tables for parainfluenza and hMPV, all accuracy measures cannot be reported for these pathogens, but sensitivities were <8.9% (parainfluenza) and <7.9% (hMPV). M.catarrhalis is not associated with any pathogen-specific ICD-10 codes, and so could not be included in these analyses.

      Discussion

      Our national level analyses of microbiology testing related to emergency COPD admissions in Wales reveals widespread under-utilisation of diagnostic microbiology for severe AECOPD, and a high percent positivity for respiratory viruses, with 46.7% of admissions in which testing was carried out returning a positive result. Given the frequency with which AECOPD admissions occur, the prevalent use of antibiotics for this presentation, and the growing problem of antimicrobial resistance, our data suggests that more widespread use of viral testing could be beneficial in advancing antibiotic stewardship. In addition, increased viral testing would enable timely isolation of cases in healthcare settings in order to reduce transmission to other, often vulnerable, hospital inpatients.
      We demonstrate how using ICD-10 diagnosis codes to detect respiratory pathogens associated with COPD admissions in our study would have considerably underestimated the burden of these infections. Researchers using ICD-10 codes as substitutes for raw microbiology data should be aware that coding deficiencies may lead to a substantial loss in accuracy.
      Furthermore, our study illustrates the feasibility and utility of linking national microbiology data with other routinely collected electronic health data, which has innumerable potential applications in future research.
      The strengths of our study include the complete population coverage, with a base dataset covering all admissions to hospitals in Wales, and the high level (99.5%) of successful linkage between hospital admissions and microbiology data. This was possible due to the recent development of a national system in Wales, the Wales Results Reporting System, which has brought together the results of all laboratory investigations from all providers, whether instituted in community or hospital settings. Whilst there are linkages in individual hospital based systems in the UK, there is no other system that links laboratory results to other routinely collected health data at a national population level. This system has allowed us to gain insight into real-world patterns of microbiology testing in emergency admissions with COPD, estimate the contribution of bacterial and viral pathogens to severe AECOPD, and report the accuracy of pathogen-specific ICD-10 diagnosis codes.
      Our study has some limitations. People undergoing microbiology tests may be those whom clinicians consider most likely to have an infective aetiology of their AECOPD, therefore extrapolating the proportion of positive tests to the total number of AECOPD hospitalisations would likely be an overestimate. Nevertheless, the majority of AECOPD are thought to be infective in nature,[
      • Sapey E
      • Stockley RA.
      COPD exacerbations·2: Aetiology.
      ] the proportion of our admissions associated with respiratory viruses was consistent with previous studies,[
      • Clark TW
      • jo Medina M
      • Batham S
      • Curran MD
      • Parmar S
      • Nicholson KG
      Adults hospitalised with acute respiratory illness rarely have detectable bacteria in the absence of COPD or pneumonia; viral infection predominates in a large prospective UK sample.
      ,
      • Mohan A
      • Chandra S
      • Agarwal D
      • et al.
      Prevalence of viral infection detected by PCR and RT-PCR in patients with acute exacerbation of COPD: A systematic review.
      ] and the high percent positivity rate suggests a large number who were not tested would also test positive. Patterns of respiratory virus transmission change year to year, and the winter of 2017/2018 saw a high level of influenza in the UK. We were confined to analysing a 2 year period due to data availability, so how representative our data is of other years is uncertain, though general patterns are likely to remain consistent. Co-infection between respiratory viruses and bacteria may be underestimated in our data, since some pathogens were not tested for or not included in our study, such as the seasonal coronaviruses. We have not reported subtypes of the respiratory viruses such as influenza or parainfluenza due to paucity of data in this regard. Lastly, the role of microorganisms in the aetiology of AECOPD has previously been unclear due to potential contamination of sputum samples by the upper airway, chronic bacterial colonisation that can occur in the lower airways of those with COPD, a lack of definitive improvement in AECOPDs treated with antibiotics in randomised controlled trials,[
      • Vollenweider DJ
      • Frei A
      • Steurer-Stey CA
      • Garcia-Aymerich J
      • Puhan MA.
      Antibiotics for exacerbations of chronic obstructive pulmonary disease.
      ] and studies reporting asymptomatic detection of rhinovirus.[
      • Falsey AR
      • Walsh EE
      • Esser MT
      • Shoemaker K
      • Yu L
      • Griffin MP.
      Respiratory syncytial virus–associated illness in adults with advanced chronic obstructive pulmonary disease and/or congestive heart failure.
      ] However, studies demonstrating the development of specific immune responses to bacteria after exacerbations support the causative role of bacteria in the exacerbation,[
      • Murphy TF
      • Brauer AL
      • Grant BJB
      • Sethi S.
      Moraxella catarrhalis in chronic obstructive pulmonary disease: Burden of disease and immune response.
      ,
      • Sethi S
      • Wrona C
      • Grant BJB
      • Murphy TF.
      Strain-specific Immune Response to Haemophilus influenzae in Chronic Obstructive Pulmonary Disease.
      ] trials may have been weakened due to inclusion of AECOPD with non-bacterial aetiology, and respiratory viruses aside from rhinovirus are not commonly identified in stable disease.[
      • Falsey AR
      • Walsh EE
      • Esser MT
      • Shoemaker K
      • Yu L
      • Griffin MP.
      Respiratory syncytial virus–associated illness in adults with advanced chronic obstructive pulmonary disease and/or congestive heart failure.
      ] Nevertheless, we are unable to infer from our data that any microorganisms detected directly contributed to development of the COPD exacerbation.
      In the context of other research, prospective studies[
      • Wilkinson TMA
      • Aris E
      • Bourne S
      • et al.
      A prospective, observational cohort study of the seasonal dynamics of airway pathogens in the aetiology of exacerbations in COPD.
      ,
      • Greenberg SB
      • Allen M
      • Wilson J
      • Atmar RL.
      Respiratory viral infections in adults with and without chronic obstructive pulmonary disease.
      ,
      • Rohde G
      • Wiethege A
      • Borg I
      • et al.
      Respiratory viruses in exacerbations of chronic obstructive pulmonary disease requiring hospitalisation: A case-control study.
      ,
      • Seemungal T
      • Harper-Owen R
      • Bhowmik A
      • et al.
      Respiratory viruses, symptoms, and inflammatory markers in acute exacerbations and stable chronic obstructive pulmonary disease.
      ,
      • Beckham JD
      • Cadena A
      • Lin J
      • et al.
      Respiratory viral infections in patients with chronic, obstructive pulmonary disease.
      ,
      • Glezen WP
      • Greenberg SB
      • Atmar RL
      • Piedra PA
      • Couch RB.
      Impact of respiratory virus infections on persons with chronic underlying conditions.
      ,
      • Sethi S
      • Murphy TF.
      Bacterial infection in chronic obstructive pulmonary disease in 2000: A state-of-the-art review.
      ] have provided useful estimates of pathogen distribution in AECOPD due to avoidance of selective testing strategies. However, due to the expense involved with such study designs, they are mostly limited to small, selected populations or regions. Our results broadly reflect those previously published. For example, in Brendish et al's prospective study,[
      • Brendish NJ
      • Mills S
      • Ewings S
      • Clark TW.
      Impact of point-of-care testing for respiratory viruses on antibiotic use in adults with exacerbation of airways disease.
      ] 42% of patients presenting to hospital with an exacerbation of their airways disease tested positive for respiratory viruses, compared to 46.7% in our data. Others have also demonstrated increased percent positivity of respiratory viruses in COPD admissions during winter months.[
      • Greenberg SB
      • Allen M
      • Wilson J
      • Atmar RL.
      Respiratory viral infections in adults with and without chronic obstructive pulmonary disease.
      ]
      Data from a randomised controlled trial has shown that point-of-care testing (POCT) for influenza can allow timely treatment with antivirals and swift isolation in healthcare settings, minimising nosocomial spread to frail populations.[
      • Clark TW
      • Beard KR
      • Brendish NJ
      • et al.
      Clinical impact of a routine, molecular, point-of-care, test-and-treat strategy for influenza in adults admitted to hospital (FluPOC): a multicentre, open-label, randomised controlled trial.
      ] Another study demonstrates increased early discontinuation of antibiotics in those with underlying airways disease testing positive for a respiratory virus on POCT.[
      • Brendish NJ
      • Mills S
      • Ewings S
      • Clark TW.
      Impact of point-of-care testing for respiratory viruses on antibiotic use in adults with exacerbation of airways disease.
      ]
      Evidence of the importance of the non-influenza respiratory viruses on the adult population has grown in recent years. RSV has been shown to be associated with 11% of COPD hospitalisations in a prospective US-based study,[
      • Falsey AR
      • Hennessey PA
      • Formica MA
      • Cox C
      • Walsh EE.
      Respiratory Syncytial Virus Infection in Elderly and High-Risk Adults.
      ] and >900,000 inappropriate antibiotic prescriptions each season in a UK-based modelling study.[
      • Fleming DM
      • Taylor RJ
      • Lustig RL
      • et al.
      Modelling estimates of the burden of Respiratory Syncytial virus infection in adults and the elderly in the United Kingdom.
      ] With progress in the development of vaccination against RSV,[
      • Sadoff J
      • de Paepe E
      • DeVincenzo J
      • et al.
      Prevention of Respiratory Syncytial Virus Infection in Healthy Adults by a Single Immunization of Ad26.RSV.preF in a Human Challenge Study.
      ] further data regarding the burden of RSV-associated disease by age, geography, and comorbidity is key to guide future vaccination strategy,[
      • Campbell H
      • Bont L
      • Nair H.
      Respiratory syncytial virus (RSV) disease - new data needed to guide future policy.
      ] and our data contributes to this. With regard to previous research examining routinely collected microbiology data, other studies have also identified underutilisation of diagnostic microbiology testing[
      • Bathoorn E
      • Groenhof F
      • Hendrix R
      • et al.
      Real-life data on antibiotic prescription and sputum culture diagnostics in acute exacerbations of COPD in primary care.
      ] and low sensitivity of ICD codes to identify laboratory-confirmed non-influenza respiratory infection.[
      • Higgins TL
      • Deshpande A
      • Zilberberg MD
      • et al.
      Assessment of the Accuracy of Using ICD-9 Diagnosis Codes to Identify Pneumonia Etiology in Patients Hospitalized With Pneumonia.
      ,
      • Moore HC
      • Lehmann D
      • de Klerk N
      • et al.
      How accurate are international classification of diseases-10 diagnosis codes in detecting influenza and pertussis hospitalizations in Children?.
      ]
      To summarise, our findings of a high percent positivity of respiratory viruses amongst those hospitalised with AECOPD, low utilisation of microbiology testing, and limitations of ICD-10 codes in identifying the burden of respiratory infections generate several key targets for further work. These include increasing the use of diagnostic testing in AECOPD, along with translation of this into reduced antimicrobial prescribing and timely infection control measures, and increasing the ability to perform linkage of existing electronic health databases to routinely collected microbiology data, which would have almost countless future applications.

      Data availability

      The anonymised person-level data used in this study are held by the SAIL Databank and not publicly available. All proposals to use the SAIL Databank are carefully reviewed by an independent Information Governance Review Panel to ensure proper and appropriate use of data (https://www.saildatabank.com/application-process). When approved, access is then provided through the SAIL Gateway, a privacy-protecting safe haven and a secure remote access system.

      Contributors

      SS conceived the study in collaboration with all authors, conducted the analysis with advice from MA, and drafted the manuscript. All authors critically reviewed and approved the final version of the manuscript.

      Ethics

      We were granted permissions from SAIL's independent Information Governance Review Panel to conduct this study. Ethical review was not required as only anonymised data were used.

      Funding

      This study was funded by Swansea University Medical School with the support of BREATHE – The Health Data Research Hub for Respiratory Health [MC_PC_19004], which is funded through the UK Research and Innovation Industrial Strategy Challenge Fund and delivered through Health Data Research UK.
      The funding source had no role in study design; in the collection, analysis and interpretation of data; in the writing of the report; or in the decision to submit the article for publication.

      Uncited References

      [
      • Hurst JR
      • Cumella A
      • Niklewicz CN
      • Philip KEJ
      • Singh V
      • Hopkinson NS.
      Long-Term Acceptability of Hygiene, Face Covering, and Social Distancing Interventions to Prevent Exacerbations in people living with Airways Diseases.
      ]

      Declaration of Competing Interest

      The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

      Acknowledgements

      This study made use of anonymised data held in the SAIL Databank. We would like to acknowledge all the data providers who make anonymised data available for research.

      Appendix. Supplementary materials

      References

        • Sapey E
        • Stockley RA.
        COPD exacerbations·2: Aetiology.
        Thorax. 2006; 61: 250-258https://doi.org/10.1136/thx.2005.041822
        • Wahl B
        • O'Brien KL
        • Greenbaum A
        • et al.
        Burden of Streptococcus pneumoniae and Haemophilus influenzae type b disease in children in the era of conjugate vaccines: global, regional, and national estimates for 2000–15.
        Lancet Glob Health. 2018; 6: e744-e757https://doi.org/10.1016/S2214-109X(18)30247-X
        • Nair H
        • Nokes DJ
        • Gessner BD
        • et al.
        Global burden of acute lower respiratory infections due to respiratory syncytial virus in young children: a systematic review and meta-analysis.
        The Lancet. 2010; 375: 1545-1555https://doi.org/10.1016/S0140-6736(10)60206-1
        • Troeger C
        • Blacker B
        • Khalil IA
        • et al.
        Estimates of the global, regional, and national morbidity, mortality, and aetiologies of lower respiratory infections in 195 countries, 1990–2016: a systematic analysis for the Global Burden of Disease Study 2016.
        Lancet Infect Dis. 2018; 18: 1191-1210https://doi.org/10.1016/S1473-3099(18)30310-4
      1. National Institute for Health and Care Excellence.
        Quality standards and indicators briefing paper: Chronic obstructive pulmonary disease. 2015; (PublishedAccessed March 22, 2021)
        • Hauge SH
        • Bakken IJ
        • de Blasio BF
        • Håberg SE.
        Risk conditions in children hospitalized with influenza in Norway.
        BMC Infect Dis. 2020; 20: 2017-2019https://doi.org/10.1186/s12879-020-05486-6
        • Hungerford D
        • Ibarz-Pavon A
        • Cleary P
        • French N.
        Influenza-associated hospitalisation, vaccine uptake and socioeconomic deprivation in an English city region: An ecological study.
        BMJ Open. 2018; 8: 23275https://doi.org/10.1136/bmjopen-2018-023275
        • Moss JWE
        • Davidson C
        • Mattock R
        • Gibbons I
        • Mealing S
        • Carroll S.
        Quantifying the direct secondary health care cost of seasonal influenza in England.
        BMC Public Health. 2020; 20https://doi.org/10.1186/s12889-020-09553-0
        • San-Román-Montero JM
        • Gil Prieto R
        • Gallardo Pino C
        • Hinojosa Mena J
        • Zapatero Gaviria A
        • Gil De Miguel A.
        Inpatient hospital fatality related to coding (ICD-9-CM) of the influenza diagnosis in Spain (2009-2015).
        BMC Infect Dis. 2019; 19https://doi.org/10.1186/s12879-019-4308-5
        • Tong S
        • Amand C
        • Kieffer A
        • Kyaw MH.
        Incidence of Respiratory Syncytial Virus Related Health Care Utilization in The United States.
        J Glob Health. 2020; 10: 1-10https://doi.org/10.7189/jogh.10.020422
        • Huh K
        • Kim YE
        • Ji W
        • et al.
        Decrease in hospital admissions for respiratory diseases during the COVID-19 pandemic: a nationwide claims study.
        Thorax. 2021; 29 (Published online Marchthoraxjnl-2020-216526)https://doi.org/10.1136/thoraxjnl-2020-216526
        • Hardelid P
        • Verfuerden M
        • McMenamin J
        • Gilbert R.
        Risk factors for admission to hospital with laboratory-confirmed influenza in young children: birth cohort study.
        Eur Respir J. 2017; 501700489https://doi.org/10.1183/13993003.00489-2017
        • Nicholson KG
        • McNally T
        • Silverman M
        • Simons P
        • Stockton JD
        • Zambon MC.
        Rates of hospitalisation for influenza, respiratory syncytial virus and human metapneumovirus among infants and young children.
        Vaccine. 2006; 24: 102-108https://doi.org/10.1016/j.vaccine.2005.02.004
        • Lim FJ
        • Blyth CC
        • Fathima P
        • de Klerk N
        • Moore HC.
        Record linkage study of the pathogen-specific burden of respiratory viruses in children.
        Influenza Other Respir Viruses. 2017; 11: 502-510https://doi.org/10.1111/irv.12508
        • Alchikh M
        • Conrad T
        • Hoppe C
        • et al.
        Are we missing respiratory viral infections in infants and children? Comparison of a hospital-based quality management system with standard of care.
        Clinical Microbiology and Infection. 2019; 25 (.e9-380.e16): 380https://doi.org/10.1016/j.cmi.2018.05.023
        • Moore HC
        • de Klerk N
        • Keil AD
        • et al.
        Use of data linkage to investigate the aetiology of acute lower respiratory infection hospitalisations in children.
        J Paediatr Child Health. 2012; 48: 520-528https://doi.org/10.1111/j.1440-1754.2011.02229.x
        • Jones KH
        • Ford D v
        • Thompson S
        • Lyons RA.
        A profile of the SAIL databank on the UK secure research platform.
        Int J Popul Data Sci. 2019; 4https://doi.org/10.23889/ijpds.v4i2.1134
        • Benchimol EI
        • Smeeth L
        • Guttmann A
        • et al.
        The REporting of studies Conducted using Observational Routinely-collected health Data (RECORD) Statement.
        PLoS Med. 2015; 12e1001885https://doi.org/10.1371/journal.pmed.1001885
        • Lyons RA
        • Jones KH
        • John G
        • et al.
        The SAIL databank: Linking multiple health and social care datasets.
        BMC Med Inform Decis Mak. 2009; 9 (-3): 3https://doi.org/10.1186/1472-6947-9-3
        • Li Y
        • Hodgson D
        • Wang X
        • Atkins KE
        • Feikin DR
        • Nair H.
        Respiratory syncytial virus seasonality and prevention strategy planning for passive immunisation of infants in low-income and middle-income countries: a modelling study.
        Lancet Infect Dis. 2021; (0(0))https://doi.org/10.1016/s1473-3099(20)30703-9
        • Hurst JR
        • Cumella A
        • Niklewicz CN
        • Philip KEJ
        • Singh V
        • Hopkinson NS.
        Long-Term Acceptability of Hygiene, Face Covering, and Social Distancing Interventions to Prevent Exacerbations in people living with Airways Diseases.
        medRxiv. 2021; 04 (Published online January 1, 2021)21255189https://doi.org/10.1101/2021.04.09.21255189
        • Clark TW
        • jo Medina M
        • Batham S
        • Curran MD
        • Parmar S
        • Nicholson KG
        Adults hospitalised with acute respiratory illness rarely have detectable bacteria in the absence of COPD or pneumonia; viral infection predominates in a large prospective UK sample.
        Journal of Infection. 2014; 69: 507-515https://doi.org/10.1016/j.jinf.2014.07.023
        • Mohan A
        • Chandra S
        • Agarwal D
        • et al.
        Prevalence of viral infection detected by PCR and RT-PCR in patients with acute exacerbation of COPD: A systematic review.
        Respirology. 2010; 15: 536-542https://doi.org/10.1111/j.1440-1843.2010.01722.x
        • Vollenweider DJ
        • Frei A
        • Steurer-Stey CA
        • Garcia-Aymerich J
        • Puhan MA.
        Antibiotics for exacerbations of chronic obstructive pulmonary disease.
        Cochrane Database of Systematic Reviews. 2018; 2018https://doi.org/10.1002/14651858.CD010257.pub2
        • Falsey AR
        • Walsh EE
        • Esser MT
        • Shoemaker K
        • Yu L
        • Griffin MP.
        Respiratory syncytial virus–associated illness in adults with advanced chronic obstructive pulmonary disease and/or congestive heart failure.
        J Med Virol. 2019; 91: 65-71https://doi.org/10.1002/jmv.25285
        • Murphy TF
        • Brauer AL
        • Grant BJB
        • Sethi S.
        Moraxella catarrhalis in chronic obstructive pulmonary disease: Burden of disease and immune response.
        Am J Respir Crit Care Med. 2005; 172: 195-199https://doi.org/10.1164/rccm.200412-1747OC
        • Sethi S
        • Wrona C
        • Grant BJB
        • Murphy TF.
        Strain-specific Immune Response to Haemophilus influenzae in Chronic Obstructive Pulmonary Disease.
        Am J Respir Crit Care Med. 2004; 169: 448-453https://doi.org/10.1164/rccm.200308-1181oc
        • Wilkinson TMA
        • Aris E
        • Bourne S
        • et al.
        A prospective, observational cohort study of the seasonal dynamics of airway pathogens in the aetiology of exacerbations in COPD.
        Thorax. 2017; 72: 919-927https://doi.org/10.1136/thoraxjnl-2016-209023
        • Greenberg SB
        • Allen M
        • Wilson J
        • Atmar RL.
        Respiratory viral infections in adults with and without chronic obstructive pulmonary disease.
        Am J Respir Crit Care Med. 2000; 162: 167-173https://doi.org/10.1164/ajrccm.162.1.9911019
        • Rohde G
        • Wiethege A
        • Borg I
        • et al.
        Respiratory viruses in exacerbations of chronic obstructive pulmonary disease requiring hospitalisation: A case-control study.
        Thorax. 2003; 58: 37-42https://doi.org/10.1136/thorax.58.1.37
        • Seemungal T
        • Harper-Owen R
        • Bhowmik A
        • et al.
        Respiratory viruses, symptoms, and inflammatory markers in acute exacerbations and stable chronic obstructive pulmonary disease.
        Am J Respir Crit Care Med. 2001; 164: 1618-1623https://doi.org/10.1164/ajrccm.164.9.2105011
        • Beckham JD
        • Cadena A
        • Lin J
        • et al.
        Respiratory viral infections in patients with chronic, obstructive pulmonary disease.
        Journal of Infection. 2005; 50: 322-330https://doi.org/10.1016/j.jinf.2004.07.011
        • Glezen WP
        • Greenberg SB
        • Atmar RL
        • Piedra PA
        • Couch RB.
        Impact of respiratory virus infections on persons with chronic underlying conditions.
        J Am Med Assoc. 2000; 283: 499-505https://doi.org/10.1001/jama.283.4.499
        • Sethi S
        • Murphy TF.
        Bacterial infection in chronic obstructive pulmonary disease in 2000: A state-of-the-art review.
        Clin Microbiol Rev. 2001; 14: 336-363https://doi.org/10.1128/CMR.14.2.336-363.2001
        • Brendish NJ
        • Mills S
        • Ewings S
        • Clark TW.
        Impact of point-of-care testing for respiratory viruses on antibiotic use in adults with exacerbation of airways disease.
        Journal of Infection. 2019; 79: 357-362https://doi.org/10.1016/j.jinf.2019.06.010
        • Clark TW
        • Beard KR
        • Brendish NJ
        • et al.
        Clinical impact of a routine, molecular, point-of-care, test-and-treat strategy for influenza in adults admitted to hospital (FluPOC): a multicentre, open-label, randomised controlled trial.
        Lancet Respir Med. 2020; (Published online)https://doi.org/10.1016/S2213-2600(20)30469-0
        • Falsey AR
        • Hennessey PA
        • Formica MA
        • Cox C
        • Walsh EE.
        Respiratory Syncytial Virus Infection in Elderly and High-Risk Adults.
        New England Journal of Medicine. 2005; 352: 1749-1759https://doi.org/10.1056/NEJMoa043951
        • Fleming DM
        • Taylor RJ
        • Lustig RL
        • et al.
        Modelling estimates of the burden of Respiratory Syncytial virus infection in adults and the elderly in the United Kingdom.
        BMC Infect Dis. 2015; 15: 443https://doi.org/10.1186/s12879-015-1218-z
        • Sadoff J
        • de Paepe E
        • DeVincenzo J
        • et al.
        Prevention of Respiratory Syncytial Virus Infection in Healthy Adults by a Single Immunization of Ad26.RSV.preF in a Human Challenge Study.
        J Infect Dis. 2021; (Published online January 5)https://doi.org/10.1093/infdis/jiab003
        • Campbell H
        • Bont L
        • Nair H.
        Respiratory syncytial virus (RSV) disease - new data needed to guide future policy.
        J Glob Health. 2015; 5https://doi.org/10.7189/jogh.05.020101
        • Bathoorn E
        • Groenhof F
        • Hendrix R
        • et al.
        Real-life data on antibiotic prescription and sputum culture diagnostics in acute exacerbations of COPD in primary care.
        International Journal of COPD. 2017; 12: 285-290https://doi.org/10.2147/COPD.S120510
        • Higgins TL
        • Deshpande A
        • Zilberberg MD
        • et al.
        Assessment of the Accuracy of Using ICD-9 Diagnosis Codes to Identify Pneumonia Etiology in Patients Hospitalized With Pneumonia.
        JAMA Netw Open. 2020; 3e207750https://doi.org/10.1001/jamanetworkopen.2020.7750
        • Moore HC
        • Lehmann D
        • de Klerk N
        • et al.
        How accurate are international classification of diseases-10 diagnosis codes in detecting influenza and pertussis hospitalizations in Children?.
        J Pediatric Infect Dis Soc. 2014; 3: 255-260https://doi.org/10.1093/jpids/pit036