Burden and treatment of chronic obstructive pulmonary disease among people using illicit opioids: matched cohort study in England

Objective To understand the burden of chronic obstructive pulmonary disease among people who use illicit opioids such as heroin, and evaluate inequalities in treatment. Design Cohort study. Setting Patients registered at primary care practices in England. Participants 106 789 patients in the Clinical Practice Research Datalink with illicit opioid use recorded between 2001 and 2018, and a subcohort of 3903 patients with a diagnosis of chronic obstructive pulmonary disease. For both cohorts, the study sampled a comparison group with no history of illicit opioids that was matched by age, sex, and general practice. Main outcome measures In the base cohort: diagnosis of chronic obstructive pulmonary disease and death due to the disease. In the subcohort: five treatments (influenza vaccine, pneumococcal vaccine, pulmonary rehabilitation, bronchodilators or corticosteroids, and smoking cessation support) and exacerbations requiring hospital admission. Results 680 of 106 789 participants died due to chronic obstructive pulmonary disease, representing 5.1% of all cause deaths. Illicit opioid use was associated with 14.59 times (95% confidence interval 12.28 to 17.33) the risk of death related to chronic obstructive pulmonary disease, and 5.89 times (5.62 to 6.18) the risk of a diagnosis of the disease. Among patients with a new diagnosis, comorbid illicit opioid use was associated with current smoking, underweight, worse lung function, and more severe breathlessness. After adjusting for these differences, illicit opioids were associated with 1.96 times (1.82 to 2.12) times the risk of exacerbations requiring hospital admission, but not associated with a substantially different probability of the five treatments. Conclusions Death due to chronic obstructive pulmonary disease is about 15 times more common among people who use illicit opioids than the general population. This inequality does not appear to be explained by differences in treatment, but late diagnosis of the disease among people who use illicit opioids might contribute.


Changes from analysis protocol
This study followed a published protocol. We made the following changes to the protocol: a) Approach to analysis of missing data. In our protocol, [1] we planned to use a 'missing indicator' method in which missing categorical observations are included in analysis with a separate category of 'missing', and missing numerical observations are included as zero with a second missing indicator category. We chose to use multiple imputation instead (as described in the methods section) because simulation studies have shown that the 'missing indicator' method is likely to be biased. [2,3] b) Sensitivity analysis excluding COPD cases who did not have records of current smoking at diagnosis. For analyses of outcomes after COPD diagnosis, we did an unplanned sensitivity analysis excluding participants who did not have records of current smoking (i.e. we classified as never-smokers, ex-smokers, or had no records of smoking status). We did this analysis because never-smokers with COPD may have unusual exposures and disease types, and were also more common among cases without a history of using illicit opioids.
c) A matching ratio of 1:3 rather than 1:5 in the 'base cohort'. In our protocol, we planned to match each participant with a history of using illicit opioids with five patients without a history of using illicit opioids. We reduced this ratio to 1:3 to meet data sharing requirements of the Clinical Practice Research Datalink. As the base cohort was large (106,789 participants with a history of using illicit opioids), this did not affect the power of our analyses. We retained 1:5 matching for the smaller sub-cohort of patients with a COPD diagnosis.

Definition of comparison groups
This study used two comparison groups: (1) CPRD patients with no records of illicit opioid use; and (2) COPD cases with no records of illicit opioid use at diagnosis. Both comparison groups are drawn from the entire CRPD database (i.e. the second comparison group is not a subset of the first). This is shown in Figure 1 of the main article.
Comparison groups were sampled from CPRD patients who were unexposed (i.e. had no prior records of illicit opioid use) at the date of cohort entry of the corresponding exposed participant. This process is called 'exposure density sampling', [4] and is designed to minimise biases related to the definition of cohort entry. The exposure density sampling procedures are shown in Figure 1 and  In this figure, participant A joins the cohort when they first use illicit opioids. Participant B has a record of illicit opioid use prior to joining CPRD. The cohort is designed to capture people with a history of illicit opioid use (rather than new opioid use) and therefore participant B is included. They enter the cohort after the washout period, and are matched with patients of the same age and sex who are unexposed on that day. Participant 1 may be matched to both participants A and B, and may therefore be duplicated in the comparison group. Participant 5 may be matched to participant B because they are unexposed at the time when participant B joins the cohort, but will be censored or change exposure status at their first record of illicit opioid use and therefore is not available to be matched with participant A.  Figure 2: Exposure density sampling to create a comparison group of people with a COPD diagnosis but no history of illicit opioid use In this example, patient A has a new diagnosis after cohort entry, while patient B has prevalent COPD at cohort entry and is excluded. Ticks represent potential matches from which the unexposed group for patient A is sampled.

SNOMED concepts used to identify COPD cases in CPRD Aurum
For patients in CPRD Gold, we used a validated definition of COPD. [5] We used this codelist to create a similar codelist for CPRD Aurum, as described in the methods. The SNOMED codes are shown in Table 1 and are also available at: https://github.com/danlewer/hupio/blob/main/codelists/aurum_copd.csv/. Note that numeric codes are prefixed with an 'x' to prevent errors due to class conversion. Once the codes have been read as strings, this 'x' should be removed. Giant bullous emphysema x105519017 Chronic bronchitis x123588010 Mucopurulent chronic bronchitis x216596014 End stage chronic obstructive airways disease x301456017 Mixed simple and mucopurulent chronic bronchitis x301460019 Chronic bullous emphysema x301463017 Segmental bullous emphysema x301470017 Acute vesicular emphysema x457168017 Mild chronic obstructive pulmonary disease x457169013 Moderate chronic obstructive pulmonary disease x475431013 Chronic obstructive pulmonary disease x506053014 Purulent chronic bronchitis x508561017 Simple chronic bronchitis x1484924013 Chronic obstructive pulmonary disease monitoring x299001000000116 Chronic obstructive pulmonary disease disturbs sleep x885281000006118 Mucopurulent chr. bronchitis x909721000006115 [RFC] Emphysema x1771201000006116 Chronic obstructive pulmonary disease multidisciplinary review x1856491000006119 Chronic obstructive pulmonary disease monitoring in primary care x1856501000006110 Chronic obstructive pulmonary disease monitoring secondary care x1856571000006116 Chronic obstructive pulmonary disease severity x1856591000006115 Chronic obstructive pulmonary disease follow-up assessment x1882421000006113 1 COPD exacerbation in past year x1882431000006111 3+ COPD exacerbations in past year x1882441000006118 2 COPD exacerbations in past year x2219861000000114 Chronic obstructive pulmonary disease rescue pack declined x2423731000000115 Telehealth chronic obstructive pulmonary disease monitoring x2010061000006113 Acute non-infective exacerbation of chronic obstructive pulmonary disease x19421011 Interstitial emphysema x139979010 Fetid chronic bronchitis x301444018 Simple chronic bronchitis NOS x301455018 Obstructive chronic bronchitis NOS x301458016 Other chronic bronchitis NOS x301468014 Chronic bullous emphysema NOS x301477019 Emphysema NOS x301545019 Chronic obstructive airways disease NOS x1222334016 Other specified chronic obstructive pulmonary disease x1222335015 Chronic obstructive pulmonary disease NOS x553211000006119 Chron obstruct pulmonary dis wth acute exacerbation, unspec x555461000006119 Chronic obstruct pulmonary dis with acute lower resp infectn x977891000006112 COPD accident and emergency attendance since last visit x998281000006115 Multiple COPD emergency hospital admissions x1704531000006115 COPD patient unsuitable for pulmonary rehab -enh serv admin x1704541000006113 COPD patient unsuitable for pulmonary rehabilitation x1783891000006115 Chronic obstructiv pulmonary disease medication optimisation x1784071000006113 Preferred place of care for next exacerbation of COPD x1811661000006110 COPD self-management plan agreed x1885981000006112 Chronic obstructive pulmon dis wr self managem plan declined x301469018 Other emphysema x11932351000006110 COPD GOLD group C x11932361000006112 COPD GOLD group D x2160051010 Admit COPD emergency x8312921000006111 COPD (chronic obstructive pulmonary disease) management plan declined x4733031000006114 Chronic obstructive bronchitis x2716351000006113 COLD -Chronic obstructive lung disease x7484341000006112 Optimization of medication for chronic obstructive lung disease x4781461000006117 Emphysematous bulla x4733011000006115 COB -Chronic obstructive bronchitis BMJ Publishing Group Limited (BMJ) disclaims all liability and responsibility arising from any reliance Supplemental material placed on this supplemental material which has been supplied by the author(s) BMJMED doi: 10.1136/bmjmed-2022-000215 :e000215.
Term x4781421000006111 Acute exacerbation of COPD x8260631000006112 COPD (chronic obstructive pulmonary disease) self-management plan review x8294671000006116 COPD (chronic obstructive pulmonary disease) written self management plan declined x9317331000006113 Asthma-COPD overlap syndrome (ACOS) x4733021000006111 Obstructive chronic bronchitis x4732991000006119 Chronic bronchitis with emphysema x4781431000006114 Acute exacerbation of chronic obstructive pulmonary disease x3764021000006117 Interstitial emphysema of lung x8316481000006116 Shared care COPD (chronic obstructive pulmonary disease) monitoring x8125721000006119 At risk of COPD (chronic obstructive pulmonary disease) exacerbation x12489801000006116 Admit COPD emergency x2716321000006116 COPD -Chronic obstructive pulmonary disease x6763231000006119 Emergency hospital admission for chronic obstructive pulmonary disease x3921361000006112 Emphysema of lung x12704691000006117 Interstitial emphysema x12759361000006112 Chronic obstructive pulmonary disease NOS x285104011 Obstructive chronic bronchitis x301539010 Other specified chronic obstructive airways disease x1880061000006110 COPD management plan declined x1765681000000110 History of chronic obstructive pulmonary disease x1573111000000115 Issue of chronic obstructive pulmonary disease rescue pack x516801000000112 Very severe chronic obstructive pulmonary disease x1813871000006117 On chronic obstructive pulmonary disease supprtv cre pathway x11932341000006113 COPD GOLD group B x977911000006114 Number of COPD exacerbations in past year x7484351000006114 Optimisation of medication for chronic obstructive lung disease x457581000006111 Acute interstitial emphysema x4781471000006112 Bullous emphysema x5516841000006119 History of chronic obstructive airway disease x2716231000006114 Chronic obstructive lung disease x6763251000006114 Emergency hospital admission for COPD (chronic obstructive pulmonary disease) x7484331000006119 Optimization of medication for chronic obstructive pulmonary disease x8235101000006114 COPD (chronic obstructive pulmonary disease) 3 monthly review x12728161000006116 Chronic obstructive airways disease NOS x8235131000006118 COPD (chronic obstructive pulmonary disease) 6 monthly review x1656601000006119 COPD patient unsuitable for pulmonary rehabilitation x909711000006111 [RFC] Chronic obstructive pulmonary disease (COPD) x851261000006116 Chronic bronchitis, acute exac x1780380013 Chronic obstructive pulmonary disease monitoring by nurse x1780381012 Chronic obstructive pulmonary disease monitoring by doctor x1488424013 Chronic obstructive pulmonary disease annual review x457171013 Severe chronic obstructive pulmonary disease x301450011 Chronic asthmatic bronchitis x301451010 Chronic wheezy bronchitis x301464011 Zonal bullous emphysema x977901000006111 Emergency COPD admission since last appointment x8120981000006112 COPD (Chronic obstructive pulmonary disease) patient unsuitable for pulmonary rehabilitation x8058301000006110 COPD (chronic obstructive pulmonary disease) disturbs sleep x9317341000006115 ACOS -asthma-chronic obstructive pulmonary disease overlap syndrome x301448015 Mucopurulent chronic bronchitis NOS x3011135010 Step down change in chronic obstructive pulmonary disease management plan x424365019 Acute infective exacerbation of chronic obstructive airways disease x2009451000006113 Chronic obstructive pulmonary disease post discharge review x1882371000006118 Shared care chronic obstructive pulmonary disease monitoring x1683181000000112 Chronic obstructive pulmonary disease 3 monthly review x508562012 Chronic catarrhal bronchitis x1488423019 Chronic obstructive pulmonary disease follow-up x396108018 Bullous emphysema with collapse x301453013 Acute exacerbation of chronic obstructive airways disease x1813881000006119 On COPD (chr obstruc pulmonary disease) supportv cre pathway x1948051000006112 Asthma-chronic obstructive pulmonary disease overlap syndrom x301835010 [X]Other emphysema x396110016 Other emphysema NOS x11932331000006115 COPD GOLD group A x8294651000006114 COPD (chronic obstructive pulmonary disease) rescue pack declined x4510801000006114 End stage chronic obstructive pulmonary disease x8219501000006114 Issue of COPD (chronic obstructive pulmonary disease) rescue pack x3764031000006119 Interstitial pulmonary emphysema x1683221000000119 Chronic obstructive pulmonary disease 6 monthly review x2152091000000112 Has chronic obstructive pulmonary disease care plan x2308511000000113 Shared care chronic obstructive pulmonary disease monitoring x1823851000006119 Chronic obstructive pulmonary disease confirmed x8287171000006115 Acute non-infective exacerbation of COPD (chronic obstructive pulmonary disease) x301572010 Chronic emphysema due to chemical fumes x87480013 Chronic tracheobronchitis x1904861000006117 Chronic obstruct pulmonary disease management plan declined x301457014 Other chronic bronchitis x301459012 Chronic bronchitis NOS x301836011 [X]Other specified chronic obstructive pulmonary disease x555471000006114 Chronic obstructive airways disease x640491000006111 Emphysema

Causal model for selecting confounding variables in analysis of treatment and outcomes after diagnosis of COPD
We used a causal model to select confounding variables in our analysis of treatment and outcomes after diagnosis of COPD. This model is shown in Figure 3.

Characteristics of participants in the base cohort
The main article includes a table showing the characteristics of participants with a new diagnosis of COPD. Table 2 shows the characteristics of the 'base cohort'.

Definitions and exclusions for outcomes after diagnosis of COPD
Our analysis of outcomes after diagnosis of COPD (using the 'sub-cohort' 'of patients diagnosed with COPD, as described in the methods section) includes 5 treatment-related outcomes and 4 adverse outcomes. Definitions of each outcome and exclusion criteria (patients who are considered ineligible for the treatment and therefore excluded from analysis) are shown in Table  3. More detailed codelists are included in the study protocol. [1]

Sensitivity analysis of outcomes after COPD diagnosis, restricted to current smokers
We did a sensitivity analysis restricted to COPD cases with records of current smoking at diagnosis, because never-smokers may have different exposures (for example may have more genetic risk factors for COPD) and disease phenotypes, and may be more common among the comparison group. The results of this sensitivity analysis are shown in Figure 4.

Detailed results of survival analysis
We used survival analysis to compare time-to-COPD-related death and time-to-COPD-diagnosis by opioid use status. The Kaplan-Meier curves are shown in Figure 5 and Figure 6, and the hazard ratios including covariates are shown in Table 4 and Table 5. We did not present the covariate coefficients in the main article because they are not the focus of this study, and the adjustment strategy may not be appropriate (ie., they may be subject to the "table 2 fallacy").