WK 8 Annotated Bibliography Assignment

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 1/14

Find a copy

document 1 of 1 Full Text | Scholarly Journal

Identifying Risk of Future Asthma Attacks Using UK Medical Record Data: A Respiratory Effectiveness Group Initiative

Blakey, John D; Price, David B; Pizzichini, Emilio; Popov, Todor A; Dimitrov, Borislav D; et al. Journal of Allergy and Clinical Immunology. In Practice; Amsterdam Vol. 5, Iss. 4, (Jul 1, 2017): 1015-1024. DOI:10.1016/j.jaip.2016.11.007

https://resolver.ebscohost.com/openurl?ctx_ver=Z39.88-2004&ctx_enc=info:ofi/enc:UTF- 8&rfr_id=info:sid/ProQ%3Ahealthcompleteshell&rft_val_fmt=info:ofi/fmt:kev:mtx:journal&rft.genre=article&rft.jtitle=Journal+of+Allergy+and+Clinical+Immunology.+In+Practice& 07- 01&rft.volume=5&rft.issue=4&rft.spage=1015&rft.isbn=&rft.btitle=&rft.title=Journal+of+Allergy+and+Clinical+Immunology.+In+Practice&rft.issn=22132198&rft_id=info:doi/10.1

Abstract

Background

Asthma attacks are common, serious, and costly. Individual factors associated with attacks, such as poor symptom control, are not robust predictors.

Objective

We investigated whether the rich data available in UK electronic medical records could identify patients at risk of recurrent attacks.

Methods

We analyzed anonymized, longitudinal medical records of 118,981 patients with actively treated asthma (ages 12-80 years) and 3 or more years of data. Potential risk factors during 1 baseline year were evaluated using univariable (simple) logistic regression for outcomes of 2 or more and 4 or more attacks during the following 2-year period. Predictors with significant univariable association (P< .05) were entered into multiple logistic regression analysis with backward stepwise selection of the model including all significant independent predictors. The predictive accuracy of the multivariable models was assessed.

Results

Independent predictors associated with future attacks included baseline-year markers of attacks (acute oral corticosteroid courses, emergency visits), more frequent reliever use and health care utilization, worse lung function, current smoking, blood eosinophilia, rhinitis, nasal polyps, eczema, gastroesophageal reflux disease, obesity, older age, and being female. The number of oral corticosteroid courses had the strongest association. The final cross-validated models incorporated 19 and 16 risk factors for 2 or more and 4 or more attacks over 2 years, respectively, with areas under the curve of 0.785 (95% CI, 0.780-0.789) and 0.867 (95% CI, 0.860-0.873), respectively.

Conclusions

Routinely collected data could be used proactively via automated searches to identify individuals at risk of recurrent asthma attacks. Further research is needed to assess the impact of such knowledge on clinical prognosis.

Full Text

What is already known about this topic? Asthma attacks are common, serious, and costly. Individual factors associated with attacks, such as poor symptom control, are not robust predictors. Adequately powered studies are required to progress toward a multivariable predictor.

What does this article add to our knowledge? This large study shows that a combination of risk factors from routine medical record data can identify individuals at high risk of subsequent recurrent asthma attacks.

How does this study impact current management guidelines? Routine data from electronic medical records could be used to assess individuals' risks of recurrent asthma attacks, and to guide targeted management of modifiable risk factors.

Asthma is a common and heterogeneous disease with a wide variety of presentations and clinical courses.1 However, in all subtypes there is the potential for abrupt clinical and

lung function deteriorations termed asthma attacks (or severe exacerbations).2 A common cause of unscheduled health care utilization,3 asthma attacks are associated with

substantial physical4 and psychological morbidity,5 and major direct and indirect health care costs.6

Asthma management strategies and action plans have focused largely on symptom control, with less attention to risk stratification schemes and prevention. This focus on symptom management may have contributed to the incidence of asthma attacks and deaths remaining relatively constant, whereas there have been substantial improvements

in other disease areas (eg, cardiovascular disease) for which risk-centered strategies using objective measures have been developed.3,7

Although poor control of asthma symptoms is associated with risk of future attacks, it is not a robust predictor in isolation.8,9 Moreover, there may be a pronounced discordance

between daily symptoms and the risk of attack in a substantial proportion of individuals.1,10 Asthma treatments may be selected by some clinicians for their effect on symptoms

but not on future risk of exacerbations (eg, theophylline), whereas other treatments may be chosen for the opposite profile (eg, mepolizumab).11 Assessing risk could therefore

reduce the potential for inappropriate undertreatment or overtreatment, as well as have the positive effect of facilitating shared decision making.12

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 2/14

Available guidelines do discuss future risk,13,14 and there are a large number of publications that report single or grouped risk factors for asthma attacks.13-15 A simple risk

questionnaire based on such published risk factors16 has generated substantial public interest. This risk assessment tool has been intended primarily as a conduit to health promotion opportunities but also highlights a range of risk factors-from smoking status and extent of reliever use to hospitalization history-that need to be evaluated in a single study alongside biomarkers. The relative effect size of these risk factors and their interaction is currently not well characterized, but establishing these elements is an essential step toward the production of a validated risk assessment tool for use in routine practice.

One study suggested that the implementation of practice-based asthma risk registries is feasible in routine clinical care, but a validated risk assessment tool was not used.17

More recently, a risk score for asthma attacks has been developed from a large clinical trial data set.18 However, enrolled patients were preselected to have uncontrolled asthma symptoms and at least 1 attack the previous year; thus, the external validity of the risk score is uncertain when applied to the wider population of patients treated for asthma in

routine clinical practice, both because most of these patients would not meet typical trial eligibility criteria19 and because the ecology of care in clinical trials is difficult to replicate in general practice.

All individuals in the United Kingdom have their electronic medical records centralized at their primary care practice, where information from secondary care and hospitalizations is also aggregated. Our objective was to identify routinely collected characteristics from electronic medical records to develop a multivariable prediction model for multiple asthma attacks over a 2-year outcome period. We hypothesized that the rich data available in longitudinal medical records of UK patients (including previously identified risk factors) could reliably identify patients who subsequently experienced recurrent attacks. We aimed to produce estimates of effect size for risk factors when considered in combination.

Methods

Data source and study population

The Optimum Patient Care Research Database (OPCRD) is a quality-controlled, respiratory-focused database containing anonymous data from general practices throughout the

United Kingdom and approved for clinical research by the Health Research Authority of the UK National Health Service (REC reference no. 15/EM/0150).20 At the time of the study, the OPCRD contained longitudinal medical record data of more than 1.7 million patients from more than 400 UK general practices. The anonymized point-of-care records for each patient include demographic information, disease diagnoses as Read codes, prescriptions issued during consultations or as renewals, test results, and information transcribed from secondary care visits and hospitalizations.

This study was an initiative of the Respiratory Effectiveness Group, an investigator-led, not-for-profit, real-life respiratory research and advocacy initiative.21 The study was conducted in line with recommendations for observational research, including an a priori research plan, study registration, commitment to publish, and an independent steering committee not remunerated for participation (please see this article's Online Repository at www.jaci-inpractice.org). Written informed consent was not necessary because data were anonymous; however, patients had been given the option to prohibit use of their anonymized data for research use.

Twelve- to 80-year-old patients with an asthma diagnostic Read code recorded before study start, active asthma, and at least 3 years of continuous data were included in the study population. Active asthma was defined as 2 or more prescriptions for asthma drugs during study year 1 (short-acting β2 agonist, inhaled corticosteroids [ICSs], long- acting β2 agonist [LABA], fixed-dose ICS/LABA combination, leukotriene receptor antagonist, and/or theophylline), as well as no Read code for resolved asthma during the 3- year study period. Those with a concurrent diagnosis of chronic obstructive pulmonary disease (chronic obstructive pulmonary disease Read code) recorded at any time in the database (ever-recorded) were excluded from the analyses.

Study design

This was a historical, follow-up cohort study of patients with asthma, using longitudinal OPCRD data from February 2005 through September 2014. The study period thus began

after the 2004 institution of the UK Quality and Outcomes Framework,22 an initiative that provides financial incentives for annual review of patients with asthma in primary care and promotes regular coding of symptoms, peak flow, and smoking status.

We examined the most recent 3 years of continuous data for each patient, including 1 year of data for baseline characterization and 2 years of outcome data. Anonymized individual patient data, including patients' demographic characteristics, comorbidities, attack history, and current therapy were extracted from routine electronic clinical patient records in primary care practice management systems.

Candidate predictors were selected on the basis of literature review and expert opinion (Table I).23,24

Model building

The primary end point was the occurrence of an asthma attack (severe exacerbation), as defined by the European Respiratory Society/American Thoracic Society,25 namely, an asthma-related hospitalization, emergency department attendance, or an acute respiratory presentation resulting in a course of oral corticosteroids (OCSs). Multiple events occurring within a 2-week window were considered as a single attack.

Univariable logistic regression analysis was used to identify individual characteristics that were predictive of 2 different binary outcomes: (1) 2 or more (yes/no) asthma attacks during the 2-year outcome period and (2) 4 or more (yes/no) asthma attacks during the 2-year outcome period. Collinear associations between potentially related predictors were assessed using Spearman rank-order correlation coefficients. The values of variables were rank-ordered for calculating these correlation coefficients, and relationships with rank correlation coefficients greater than 0.30 were defined as being collinear.

All predictors with a significant univariable association (P < .05) were entered into a multiple logistic regression analysis with backward selection of the model, performed manually on the basis of significant P values. For the variables that were found to be collinear, we repeated the multiple regression analyses, substituting the second variable of the pair for the first (eg, number of acute OCS courses for number of asthma attacks) and selected the variable leading to the lowest Akaike information criterion of the model.

Because not all patients had recorded values for all predictors, we categorized predictors and included a separate category to indicate absence of available data for the following variables: body mass index, smoking status, percent predicted peak expiratory flow, and blood eosinophil count.

Model performance and internal validation

The ability of the model to distinguish patients with multiple asthma attacks from other patients with asthma was assessed by its discrimination performance calculating the C statistic (area under the receiver operating characteristic curve). The C statistic CIs were generated by bootstrapping with 1000 resamples. Other performance measures, including sensitivity, specificity, and positive and negative predictive values, were plotted for different cutoff points of the estimated risk of multiple asthma attacks as calculated by the models in plots generated using R package ROCR version 1.0-5.

Potential optimism in estimated discrimination performance and overfitting of the models was evaluated using bootstrapping with 100 resamples and by cross-validation with a random split of the data as 70% for model development (sample set) and 30% for performance testing (test set).

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 3/14

Calibration analysis was performed and results were presented by plots showing the correlation of the mean observed risk with mean predicted risk among 500 groups encompassing all patients in the study (n = 118,981).

Results

Of 338,482 patients in the OPCRD with an asthma diagnosis and 3 consecutive years of data, 132,717 (39%) patients aged 12 to 80 years had active asthma (see Figure E1 in this article's Online Repository at www.jaci-inpractice.org). We excluded patients with an ever-recorded chronic obstructive pulmonary disease diagnosis (n = 13,736; 10%), leaving 118,981 patients in the total study population.

Key patient characteristics are summarized in Table II. The mean age at start of the study was 45 ± 18 years, 67,534 (57%) patients were women, 35,544 (30%) were obese, and 19,022 (16%) were current smokers. Most patients (n = 104,345; 88%) were prescribed ICS, either as monotherapy (n = 61,358; 52%) or in combination with a LABA (n = 42,987; 36%); 40% (n = 47,652) were prescribed high-dose ICS at their last prescription (>=400 μg/d fluticasone-equivalent). Seventeen percent of patients (n = 20,711)

had at least 1 OCS course prescribed in the baseline year. (Table E1 in this article's Online Repository at www.jaci-inpractice.org depicts distributions of all other candidate predictors at baseline.)

During the subsequent 2-year outcome period, one-quarter of patients (n = 30,234; 25%) experienced 1 or more, 12,736 (11%) experienced 2 or more, and 3,198 (3%)

experienced 4 or more asthma attacks (Table III).

Model building

All candidate predictors recorded in the baseline period, with the exception of beta-blocker prescriptions, were significantly associated with the risk of frequent asthma attacks

(2 or more or 4 or more) during the outcome period (see Table E2 in this article's Online Repository at www.jaci-inpractice.org). Descriptions of collinear associations among risk factors are given in this article's Online Repository at www.jaci-inpractice.org.

The final multivariable (multifactor) models contained 19 independent predictors for 2 or more attacks (Table IV) and 16 predictors for 4 or more attacks (Table V), of which the number of acute OCS courses in the baseline year had the strongest association.

Older age, female sex, current smoking, and obesity were significant risk predictors for both outcomes, as were blood eosinophilia, higher mean daily short-acting β2 agonist dose, and leukotriene receptor antagonist or LABA prescriptions in the baseline year. Comorbidities significantly contributing to risk prediction of both outcomes were active rhinitis and a history of nasal polyps or anaphylaxis. The odds of frequent attacks were increased for patients with more frequent primary care consultations and for those with

baseline-year markers of asthma attacks, such as acute OCS courses or emergency department attendance (Tables IV and V). The odds of 2 or more or 4 or more attacks were significantly lower for patients with lower medication possession ratio.

Model performance and internal validation

The overall C statistic was 0.785 (95% CI, 0.780-0.789) for the ability of the model to distinguish patients who experienced 2 or more asthma attacks in the 2-year outcome

period (see Figure E2 in this article's Online Repository at www.jaci-inpractice.org). The model performed better in predicting 4 or more attacks with a C statistic of 0.867 (0.860-

0.873) (see Figure E3 in this article's Online Repository at www.jaci-inpractice.org). We found no indication of relevant optimism in estimated model performance or overfitting of the model in this large data set (data not shown).

Calibration plots showed good correlation between the probabilities of having multiple asthma attacks in the outcome period as estimated by the models and the observed

outcome frequencies, although higher predicted risks, observed in relatively small proportions of the population, were slightly overestimated (Figure 1).

As forecasted by the multivariable model, 3% (n = 3497) of the population had a 50% or more predicted risk of experiencing 2 or more asthma attacks in the next 2 years; and 58% (n = 2019) of these individuals actually experienced 2 or more attacks in the outcome period (positive predictive value at the cutoff point). The negative predictive value was 91% at that cutoff point.

Only 246 (0.2%) patients had a 50% or more predicted risk of experiencing 4 or more asthma attacks and 54% (n = 133) experienced 4 or more attacks in the outcome period. Only 3% (n = 3065) of the patients with a lower predicted risk experienced 4 or more attacks (negative predictive value 97%).

Table VI illustrates the predicted risk calculation for 4 hypothetical patients with asthma.

Discussion

A combination of risk factors from longitudinal medical records of UK patients was effective in predicting which individuals subsequently experienced recurrent attacks, and in particular in predicting the high-risk patients who experienced 4 or more attacks over a 2-year period. This large database study has confirmed that asthma attacks are common in an unselected UK population, with 25% of patients experiencing 1 or more attacks during the 2-year outcome period. The risk factors we identified are largely consistent with previous findings.

This study has strengths in its large sample size and the range of factors considered concurrently (for post hoc power calculations, see this article's Online Repository at www.jaci-inpractice.org Online Repository). Asthma is a common and important disease with a variety of presentations and underlying mechanisms; therefore, multiple factors should be included in any risk prediction model. Previous studies have evaluated individual risk factors or limited numbers of risk factors to predict asthma attacks, for example,

those representing subacute lack of asthma control.26 Questionnaire-based methods of predicting risk have been studied as well.27 Instead, the risk factors we identified are all collected from routine electronic patient data, suggesting that an informatics-based approach to risk stratification is possible, with lists of high-risk patients being automatically generated for the attention of the clinical team, for example, by alerts placed on the clinical records. Moreover, the present study also formally describes the potential predictive ability of the risk model developed and lends itself to the development of an individualized Web-based assessment tool as used in other disease areas, such as for cardiovascular

risk assessment.28

The risk factors included in our model have been identified in previous studies including the recent UK National Review of Asthma Deaths29; these include previous asthma

attacks, asthma severity as described by level of treatment, current symptom control, nasal disease, and generally hazardous comorbidities (smoking, obesity).13,30 Obesity may predispose to asthma attacks through the effect of extrathoracic restriction from adipose tissue and from the effect of adipokines on overall immune function and airway

inflammation.31 In addition, there may be a common genetic predisposition to both asthma and obesity.32,33

For those individuals with available blood cell counts, blood eosinophil counts (>0.4 x 109/L) were also associated with frequent asthma attacks. This finding is consistent with a

recent large database study investigating the dose-response relationship between blood eosinophils and exacerbation risk.34 Furthermore, this work expands on and

complements a study published earlier this year.35 Although of a similar design, that study investigated a narrower range of risk factors over a shorter follow-up period (1 year) for the subpopulation of patients who had a blood eosinophil count; the findings therefore may not be representative of the wider population of individuals with asthma.

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 4/14

In this general population of people treated for asthma, 51% filled less than 60% of their prescription refills during the baseline year, and the odds of multiple attacks were lower among those with lower medication possession ratios than among patients with medication possession ratios of 80% to 100%. We can speculate that perhaps individuals

with milder asthma took their treatment less regularly (eg, over a pollen season) and this was an effective strategy for them.36 In their systematic review of medication

adherence and risk of asthma attacks, Engelkes et al37 reported that some studies found an association between low adherence (expressed as medication possession ratio) and low risk of attack, perhaps because of self-titration according to level of control or of heterogeneity in treatment response. Others have reported variations in adherence over

time.38 Up to a third of people treated for asthma do not have objective supportive evidence of asthma when tested for airway dysfunction and inflammation.39 Therefore, it may be that some individuals in this study were not regularly collecting medication because they did not have active asthma symptoms, and they were also at very low risk for asthma attacks. Conversely, individuals who have experienced a recent attack and have less stable asthma may be concordant with inhaled therapy but still remain at a higher risk of attack.

Given the population we studied and the method of data collection, these real-life findings are directly applicable to patients treated for asthma in the United Kingdom. This is in contrast to the limited inclusion criteria of most randomized controlled trials, which often exclude up to 95% of typical patients seen in general practice, such as smokers and

those with comorbidities.19 The generalizable nature of these findings has the potential to inform future changes in practice and thus have an early clinical impact.

As with any observational study, these findings do not provide mechanistic insight into how the identified factors increase future risk. Moreover, several other potential risk factors would have been of interest to consider, including allergen exposure, inhaler technique assessment, and socioeconomic status, but these were not readily available from the database. Although the study population is dispersed across the country, it is unclear whether the findings would be applicable outside the UK National Health Service framework and its largely white population in terms of relative magnitude of effects. In addition, this type of data carry the potential for underrecording of secondary care attendances: asthma attacks that require emergency department attendance are not invariably recorded in primary care notes because recording requires a manual step. This potential for missing outcomes could result in underestimating the attack rate or biasing the predictors toward those associated with more moderate exacerbations that do not require hospitalization.

Our study period (February 2005 to September 2014) began after the 2004 institution of the UK Quality and Outcomes Framework, which has improved data recording in

electronic patient records through financial incentives.22,40,41 Within that period, we analyzed the most recent 3-year interval of data for eligible patients to include their most current available data. The prescription data used in this study were drawn from the electronic record of prescriptions issued at the time of a consultation (eg, for acute illness or change in regular medication) or as renewals that continued existing chronic prescriptions. Although there is currently no UK-wide system that links prescribing and dispensing data for primary care, several sources cite the reliability of prescribing data in another similar UK primary care database, the General Practice Research Database (now the Clinical Practice Research Datalink), noting that there is good agreement between General Practice Research Database prescribing data and national dispensing

data.42,43 Moreover, in the United Kingdom, pharmacists must dispense medications as prescribed.

We are developing a simple risk scoring tool as an example of the type of individualized information that could be available to people with asthma and their health care providers in the near future, or that could be automatically applied to routine electronic medical records where computer-based clinical record-keeping is used. During the development of

the model, the extent of missing data varied from 6% for smoking status to 34% for blood eosinophil count, as recorded in Table II. For those variables with missing data, we were able to include a “missing data” category in the risk model, thereby enabling clinicians to use the risk calculator even when some data are missing, a common situation in real life.

This study provides clinically relevant measures of the relative importance of risk factors for recurrent asthma attacks. Additional work will be required to validate the model in other data sets, and prospectively for patients in different settings, and to develop these findings into questions or data queries to create a reliable tool for clinical practice. Further analyses will be required to explore potential time-to-event measures and also to ascertain which are the most important predictors in the models. Prospective trials will be required to assess the implementation of such models in clinical practice and the effect on asthma-related outcomes of risk-based decision making, at both individual and group levels.

Acknowledgments

We thank Ian D. Pavord, Hilary Pinnock, Gene Colice, Alexandra Dima, Janet Holbrook, Cindy Rand, Iain Small, and Sam Walker for their valuable contributions to discussions about the study design. We thank Anne Burden, Vasilis Nikolaou, Victoria Thomas, and Maria Batsiou for contributions to the data elaboration and statistical analyses.

Appendix

Methods

The study was conducted in line with recommendations for observational research, including an a priori research plan, study registration, commitment to publish, and an

independent steering committee not remunerated for participation.E1,E2 The study protocol was approved by the Anonymised Data Ethics Protocols and Transparency Committee, the independent scientific advisory committee for the OPCRD, and was registered with the European Network of Centres for Pharmacoepidemiology and

Pharmacovigilance (http://www.encepp.eu/encepp/viewResource.htm?id=6303).E3

The Charlson comorbidity index scoreE4 in the baseline year was categorized as 0, 1 to 4, 5 to 9, and 10 or more, with comorbidity weights taken from the Hospital

Standardised Mortality Ratios.E5

Post hoc power calculations showed that the large study population of 118,981 patients provided sufficient statistical power (>=80%; α = 0.05) to detect an association with an odds ratio of 1.10 for the risk of 2 or more asthma attacks, assuming a risk of 11% in patients without the characteristic of the predictor and a prevalence of the characteristic of at least 8%. For the risk of 4 or more asthma attacks, the study population size would allow detecting an odds ratio of 1.17, assuming a risk of 3.0% in patients without the characteristic for predictors with a prevalence of at least 9%.

Results

Additional patient demographic and clinical characteristics are presented in Table E1.E6

Univariable analyses

All the potential baseline risk factors tested in univariable analyses with the exception of beta-blocker prescriptions (yes/no) were significantly associated with the presence of

asthma attacks (>=2 or >=4 attacks) in the follow-up period (study years 2 and 3; Table E2).E6-E9

Multivariable analyses

Age was collinear with gastroesophageal reflux disease (GERD) diagnosis (active/ever) and/or GERD drugs, cardiovascular disease diagnosis, and prescriptions for statins.

Acute OCS courses were collinear with acute OCS courses with evidence of lower respiratory consultation, antibiotic courses (with evidence of lower respiratory consultation), acute respiratory events, and severe exacerbations (baseline year).

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 5/14

ICS adherence was collinear with number of ICS prescriptions and inhalers, ICS average daily dose, ICS prescribed and actual duration, and number of SABA prescriptions and inhalers.

ICS prescribed dose was collinear with ICS device type.

Rhinitis diagnosis (active) was collinear with rhinitis diagnosis (ever), rhinitis diagnosis (active)/drugs, and rhinitis diagnosis (ever)/drugs.

Eczema diagnosis (active) was collinear with eczema (ever).

GERD diagnosis (active) was collinear with GERD diagnosis (ever).

Primary care consultations were collinear with diabetes diagnosis, asthma consultations, Charlson comorbidity index score, paracetamol prescriptions, antibiotic courses, and asthma control status.

Nonsteroidal anti-inflammatory drug prescriptions were collinear with paracetamol prescriptions.

During the stepwise backward logistic regression, heart failure and anxiety/depression were dropped from the final model for 2 or more attacks; and diagnoses of GERD (active), heart failure, eczema (active), and anxiety/depression, as well as prescriptions for nonsteroidal anti-inflammatory drugs were dropped from the final model for 4 or more attacks.

Table ICandidate predictors assessed for inclusion in the modelsBAI, Breath-actuated inhaler; BMI, body mass index; BTS, British Thoracic Society; DPI, dry powder inhaler; ED, emergency department; FP, fluticasone propionate; GERD, gastroesophageal reflux disease; LRTI, lower respiratory tract infection; LTRA, leukotriene receptor antagonist; MDI, metered-dose inhaler; NSAIDs, nonsteroidal anti-inflammatory drugs; PEF, peak expiratory flow; SABA, short-acting β2 agonist; Theo, theophylline.

Variable Description

Sex Male or female

Age In years at the start of the 3-y study period

BMI Last recorded, in kg/m2; categorized as underweight (<18.5), normal (18.5-24.9), overweight (25-29.9), or obese (>=30)

Smoking status Last recorded, categorized as never smoker, current smoker, or ex-smoker

Charlson comorbidity index Score in the baseline year, categorized as 0, 1-4, 5-9, >=10 (comorbidity weights taken from Hospital Standardised Mortality Ratios, version 9)22,23

Comorbidities∗ Recorded ever or active: eczema, allergic and nonallergic rhinitis, nasal polyps, anaphylaxis diagnosis, anxiety/depression diagnosis, diabetes (type 1 or 2), GERD, cardiovascular disease, ischemic heart disease, heart failure, psoriasis

Comedications In baseline year, prescription (yes/no) for paracetamol, NSAIDs, beta-blockers, statins

% predicted PEF Recorded ever, expressed as percentage of predicted normal, categorized as unknown, <60%, 61%-79%, and >=80%

Blood eosinophil count Last recorded, in 109cell/L, categorized as <=0.4 or >0.4

BTS step†

Step 1 Inhaled SABA as needed

Step 2 ICS or LTRA

Step 3 Add LABA to ICS or use high-dose ICS (>=400 μg/d FP equivalent)

Step 4 Add LTRA/Theo to [ICS + LABA] or add LABA/LTRA/Theo to high-dose ICS

Step 5 Add OCS

Average daily dose of SABA/ICS

Cumulative dose of SABA/ICS prescribed in baseline year, expressed in μg/d albuterol or FP equivalent and divided by 365.25

Prescribed daily ICS dose Dose of ICS prescribed at last prescription of baseline year in μg/d, FP equivalents

ICS medication possession ratio

ICS refill rate during the baseline year: sum of number of days per pack (number of actuations per pack/number of actuations per day)/365.25

ICS device type In baseline year: categorized as no ICS, MDI, BAI, or DPI

Spacer use with ICS pMDI Recorded in baseline year (yes/no)

Oral corticosteroid use Any maintenance prescription for corticosteroids in baseline year (yes/no)

Prior asthma education Recorded ever (yes/no)

Primary care consults Number of primary care consultations, categorized as 0, 1-5, 6-12, >=13

Primary care consults for asthma

Number of primary care consultations with an asthma-related Read code

Antibiotics with lower respiratory consult

Number of consultations that resulted in antibiotic prescription (included to capture asthma events that may have been misclassified as LRTI)

Acute respiratory events Number of events in the baseline year, defined as asthma-related hospitalization or ED attendance or an acute course of OCS or antibiotics prescription with lower respiratory consultation

Acute OCS courses Number of acute courses of OCS in baseline year, categorized as 0, 1, >=2

Acute OCS courses with lower respiratory consult

Number of OCS courses with Read code for lower respiratory consultation in baseline year, categorized as 0, 1, >=2

Antibiotics courses Number of antibiotics prescriptions with Read code for lower respiratory consultation in baseline year, categorized as 0, 1, >=2

Hospital attendance/admission

Number of asthma-related‡ ED, inpatient, and outpatient attendance/admission in baseline year

Asthma attacks Number of asthma-related‡ hospital ED attendance, inpatient admission, or acute OCS course

Table IIPatients' demographic and clinical characteristics during the baseline yearED, Emergency department; GERD, gastroesophageal reflux disease; LTRA, leukotriene receptor antagonist; NSAIDs, nonsteroidal anti-inflammatory drugs; PEF, peak expiratory flow; SABA, short-acting β2 agonist.Data are n (%) unless otherwise noted.

Variable All patients (n = 118,981)

Male sex∗ 51,447 (43)

Age at study start (y), mean ± SD∗ 45 ± 18

12-18 13,452 (11)

19-34 21,381 (18)

35-54 44,375 (37)

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 6/14

55-80 39,773 (33)

Body mass index∗

Underweight 3,480 (3)

Normal 35,400 (30)

Overweight 36,608 (31)

Obese 35,544 (30)

Unknown 7,949 (7)

Smoking status∗

Current smokers 19,022 (16)

Ex-smokers 26,758 (22)

Nonsmokers 65,489 (55)

Unknown smoking status 7,712 (6)

Recorded comorbidity†

Rhinitis diagnosis, active∗ 3,567 (3)

Rhinitis diagnosis/therapy, active 36,312 (31)

Nasal polyps, ever∗ 3,933 (3)

Eczema diagnosis, active∗ 4321 (4)

Anaphylaxis diagnosis, ever∗ 512 (0.4)

GERD diagnosis, active∗ 1,444 (1)

Anxiety or depression diagnosis, ever 5,812 (5)

>=1 prescription during baseline

NSAIDs∗ 27,862 (23)

%predicted PEF, median (IQR)∗ 80 (68-91)

<=60% 13,808 (12)

61%-79% 33,850 (28)

>=80% 47,780 (40)

Unknown 23,543 (20)

Blood eosinophil count∗

<=0.4 x 109/L 64,803 (55)

>0.4 x 109/L 13,184 (11)

Missing 40,994 (34)

Mean daily SABA dose (μg/d)∗‡

0 11,992 (10)

1-200 50,467 (42)

201-400 29,866 (26)

>400 26,656 (22)

Last ICS dose prescribed in baseline year (μg/d)‡

0 14,636 (12)

<400 56,693 (48)

>=400 47,652 (40)

ICS medication possession ratio∗§

>0%-39.9% 37,723 (32)

40%-59.9% 23,374 (20)

60%-79.9% 9,385 (8)

80%-100% 15,493 (13)

>100% 18,370 (15)

No ICS prescribed 14,636 (12)

>=1 prescription during baseline

LTRA∗ 6,995 (6)

LABA (standalone)∗ 8,253 (7)

Acute OCS courses∗

0 98,270 (83)

1 14,554 (12)

>=2 6,157 (5)

Primary care consultation∗

0 5,618 (5)

1-5 56,023 (47)

6-12 40,074 (34)

>=13 17,266 (14)

>=1 Asthma-related ED admission∗ 696 (0.6)

Asthma attacks¶

0 97,583 (82)

1 15,058 (13)

2 4,202 (4)

>=3 2,138 (2)

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 7/14

Table IIINumber of asthma attacks (severe exacerbations) in the baseline and outcome years for 118,981 patients with asthmaThe category “Years 2 & 3 combined” includes those patients who had a single exacerbation in year 2 and/or in year 3.

Asthma attacks Year 1 Year 2 Year 3 Years 2 & 3 combined

>=1, n (%) 21,398 (18.0) 20,132 (16.9) 17,984 (15.1) 30,234 (25.4)

>=2, n (%) 6,340 (5.3) 6,169 (5.2) 5,517 (4.6) 12,736 (10.7)

>=4, n (%) 770 (0.6) 732 (0.6) 681 (0.6) 3,198 (2.7)

Table IVIndependent baseline predictors (year 1) of 2 or more asthma attacks during the 2-y follow-up period as identified in the final multivariable modelED, Emergency department; GERD, gastroesophageal reflux disease; LTRA, leukotriene receptor antagonist; MPR, medication possession ratio; NSAID, nonsteroidal anti-inflammatory drug; OR, odds ratio; PEF, peak expiratory flow; ref, reference category; SABA, short-acting β2 agonist.Collinearity of variables is described in this article's Online Repository at www.jaci- inpractice.org.

Year 1 predictors Adjusted OR (95% CI) P value∗

Age (y) 12-18 (ref) 1.00 <.001

19-34 1.27 (1.14-1.40)

35-54 1.43 (1.29-1.57)

55-80 1.47 (1.33-1.62)

Sex, female 1.35 (1.29-1.41) <.001

Body mass index, normal (ref) 1.00 <.001

Underweight 1.10 (0.95-1.27)

Overweight 1.16 (1.09-1.22)

Obese 1.27 (1.21-1.34)

Unknown 0.96 (0.86-1.08)

Smoking status, nonsmoker (ref) 1.00 <.001

Current smoker 1.17 (1.11-1.24)

Ex-smoker 1.01 (0.96-1.06)

Unknown 1.02 (0.93-1.11)

Rhinitis diagnosis, active† 1.14 (1.03-1.27) .015

Eczema diagnosis, active 1.13 (1.02-1.25) .017

GERD diagnosis, active 1.29 (1.11-1.50) .017

Nasal polyps, ever 1.60 (1.46-1.76) <.001

Anaphylaxis diagnosis, ever 1.66 (1.29-2.13) <.001

NSAID prescription, >=1 1.13 (1.08-1.18) <.001

PEF % predicted, >=80% (ref) 1.00 <.001

<=60% 1.62 (1.52-1.27)

61%-79% 1.21 (1.15-1.27)

Unknown 1.25 (1.17-1.33)

Blood eosinophil count, <=0.4 x 109/L (ref) 1.00 <.001

>0.4 x109/L 1.21 (1.14-1.29)

Missing 0.88 (0.83-0.93)

Mean SABA dose (μg/d),‡ 0 (ref) 1.00 <.001

1-200 1.05 (0.97-1.14)

201-400 1.28 (1.16-1.39)

>400 1.63 (1.45-1.77)

LTRA prescription, >=1 2.05 (1.92-2.18) <.001

LABA prescription (stand alone), >=1 1.21 (1.13-1.30) <.001

ICS MPR (%), 80%-100% (ref) 1.00 <.001

>0%-39.9% 0.88 (0.82-0.94)

40%-59.9% 0.88 (0.82-0.95)

60%-79.9% 0.94 (0.86-1.02)

>=100% 0.92 (0.86-0.98)

No ICS prescribed 0.65 (0.59-0.71)

Acute OCS courses, 0 (ref) 1.00 <.001

1 3.34 (3.37-3.71)

>=2 9.50 (8.94-10.08)

Asthma-related ED admission, >=1 1.76 (1.45-2.13) <.001

Primary care consultations, 0 (ref) 1.00 <.001

1-5 1.29 (1.13-1.48)

6-12 1.66 (1.45-1.90)

>=13 2.05 (1.78-2.36)

Table VIndependent baseline predictors (year 1) of 4 or more asthma attacks during the 2-y follow-up period as identified in the final multivariable modelED, Emergency department; LTRA, leukotriene receptor antagonist; MPR, medication possession ratio; OR, odds ratio; PEF, peak expiratory flow; ref, reference category; SABA, short-acting β2 agonist.Collinearity of variables is described in this article's Online Repository at www.jaci-inpractice.org.

Year 1 predictors Adjusted OR (95% CI) P value∗

Age (y), 12-18 (ref) 1.0 <.001

19-34 1.13 (0.91-1.40)

35-54 1.45 (1.19-1.77)

55-80 1.61 (1.31-1.97)

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 8/14

Sex, female 1.31 (1.20-1.43) <.001

Body mass index, normal (ref) 1.0 <.001

Underweight 0.89 (0.65-1.22)

Overweight 1.18 (1.06-1.31)

Obese 1.27 (1.15-1.41)

Unknown 0.95 (0.76-1.20)

Smoking status, nonsmoker (ref) 1.0 <.001

Current smoker 1.29 (1.16-1.43)

Ex-smoker 1.02 (0.93-1.12)

Unknown 1.19 (1.01-1.39)

Rhinitis diagnosis, active† 1.24 (1.03-1.49) .023

Nasal polyps, ever 1.65 (1.42-1.93) <.001

Anaphylaxis diagnosis, ever 1.77 (1.17-2.68) .007

PEF % predicted, >=80% (ref) 1.0 <.001

<=60% 1.67 (1.50-1.86)

61%-79% 1.29 (1.17-1.43)

Unknown 1.26 (1.10-1.43)

Blood eosinophil count, <=0.4 x 109/L (ref) 1.0 <.001

>0.4 x 109/L 1.37 (1.24-1.53)

Missing 0.95 (0.86-1.05)

Mean SABA dose (μg/d),‡ 0 (ref) 1.0 <.001

1-200 0.89 (0.76-1.05)

201-400 1.13 (0.96-1.33)

>400 1.68 (1.43-1.97)

LTRA prescription, >=1 2.22 (2.01-2.45) <.001

LABA prescription (standalone), >=1 1.15 (1.03-1.30) .018

ICS MPR (%), 80%-100% (ref) 1.00 <.001

>0%-39.9% 0.81 (0.71-0.92)

40%-59.9% 0.90 (0.79-1.02)

60%-79.9% 1.01 (0.87-1.17)

>=100% 0.95 (0.84-1.07)

No ICS prescribed 0.71 (0.59-0.84)

Acute OCS courses, 0 (ref) 1.0 <.001

1 4.34 (3.94-4.79)

>=2 15.49 (14.09-17.04)

Asthma-related ED admissions, >=1 2.01 (1.55-2.62) <.001

Primary care consultations, 0 (ref) 1.0 <.001

1-5 0.94 (0.71-1.23)

6-12 1.39 (1.06-1.82)

>=13 1.81 (1.38-2.39)

Table VIPredicted risk (over 2 y) as calculated for 4 hypothetical patients with asthmaED, Emergency department; LTRA, leukotriene receptor antagonist; MPR, medication possession ratio; NSAIDs, nonsteroidal anti-inflammatory drugs; PEFR, peak expiratory flow rate; SABA, short-acting β2 agonist.

Patient description Risk of >=2 attacks

Risk of >=4 attacks

A 35-y-old woman who is obese, takes NSAIDs, and uses a lot of her SABA (mean, >400 μg/d)[list][list_item]Nonsmoker, PEFR >=80%, no comorbidities, no OCS courses the prior year, 80%-100% MPR, 1-5 primary care consultations, no blood eosinophilia[/list_item][/list]

8.9% 1.1%

A 56-y-old man at step 4 who has a PEFR of 65% predicted and an incident finding of a high blood eosinophil count[list][list_item]Nonsmoker, normal weight, no comorbidities, no OCS courses the previous year, 80%-100% MPR, 1-5 primary care consultations, SABA mean dose 1-200 μg/d[/list_item] [/list]

4.7% 0.7%

An 18-y-old woman with rhinitis and eczema who has had 2 attacks in the last year and is on LTRA[list][list_item]Nonsmoker, PEFR >=80%, normal weight, no other comorbidities, 80%-100% MPR, 6-12 primary care consultations, SABA mean dose 1-200 μg/d, no blood eosinophilia[/list_item][/list]

49.7% 17.1%

A 23-y-old man who smokes, has had a couple of ED attendances in the last year, and takes 25% of his ICS[list][list_item]PEFR >=80%, normal weight, no comorbidities, >=2 OCS courses, 6-12 primary care consultations, SABA mean dose 1-200 μg/d, no blood eosinophilia[/list_item][/list]

38.8% 12.0%

Table E1Additional patient demographic and clinical characteristics during the baseline yearBAI, Breath-actuated inhaler; BTS, British Thoracic Society; DPI, dry powder inhaler; FDC, fixed-dose combination; FP, fluticasone propionate; LTRA, leukotriene receptor antagonist; NSAIDs, nonsteroidal anti-inflammatory drugs; PEF, peak expiratory flow; pMDI, pressurized metered-dose inhaler; SABA, short-acting β2 agonist.Data are n (%) unless otherwise noted.

Variable All patients (n = 118,981)

Charlson comorbidity index score

0 54,974 (46)

1-4 58,034 (49)

5-9 3,351 (3)

>=10 2,622 (2)

Recorded comorbidity∗

Rhinitis diagnosis, active† 3,567 (3)

Rhinitis diagnosis/therapy, active 36,312 (31)

Rhinitis diagnosis, ever 30,644 (26)

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 9/14

Rhinitis diagnosis/therapy, ever 81,991 (69)

Nasal polyps, ever† 3,933 (3)

Eczema diagnosis, active† 4,321 (4)

Eczema diagnosis, ever 32,213 (27)

GERD diagnosis, active† 1,444 (1)

GERD diagnosis/therapy, active 23,861 (20)

GERD diagnosis, ever 9,640 (8)

GERD diagnosis/therapy, ever 40,593 (34)

Diabetes (type 1 or 2), ever 15,105 (13)

Cardiovascular disease, ever 29,688 (25)

Ischemic heart disease, ever 6,208 (5)

Heart failure, ever 873 (0.7)

Asthma education, ever 47,356 (40)

Mean daily ICS dose (μg/d)‡

0 14,636 (12)

<400 87,543 (74)

>=400 16,802 (14)

>=1 prescription during baseline

Paracetamol 28,166 (24)

Beta-blockers 3,334 (3)

Statins 18,159 (15)

BTS step§

No therapy 0 (0)

Step 1 13,761 (12)

Step 2 39,222 (33)

Step 3 27,837 (23)

Step 4 36,004 (30)

Step 5 2,144 (2)

Not assignable 13 (0.01)

ICS or FDC inhaler device type, last prescription

pMDI 69,604 (59)

DPI 28,920 (24)

BAI 5,821 (5)

No ICS 14,636 (12)

Spacer device prescribed with ICS pMDI 6,212 (9)

Acute OCS courses with lower respiratory consultation

0 115,117 (97)

1 3,436 (3)

>=2 428 (0.4)

Primary care consultation for asthma

0 37,367 (31)

1 51,115 (43)

>=2 30,499 (26)

Acute respiratory events

0 81,387 (68)

1 24,538 (21)

>=2 13,056 (11)

Antibiotics with lower respiratory consult

0 90,247 (76)

1 19,692 (17)

>=2 9,042 (7)

Asthma limiting daily activities, n with data 35,526

Yes 7,784 (22)

Asthma limiting night-time activities, n with data 36,250

Yes 6,261 (17)

Asthma is causing daytime symptoms, n with data 43,762

Yes 27,690 (63)

Table E2Results of univariable logistic regression analyses of asthma attack frequency (n = 118,981)BAI, Breath-actuated inhaler; BMI, body mass index; DPI, dry powder inhaler; ED, emergency department; FP, fluticasone propionate; GINA, Global Initiative for Asthma; LRTI, lower respiratory tract infection; LTRA, leukotriene receptor antagonist; MDI, metered-dose inhaler; OR, odds ratio; PEF, peak expiratory flow; SABA, short-acting β2 agonist.

Year 1 predictors Asthma attacks within the 2-y outcome period

<2, n (%) >=2, n (%) OR (95% CI)

<4, n (%) >=4, n (%) OR (95% CI)

Sex, female 58,816 (55) 8718 (68) 1.75 (1.69-1.82) 65,282 (56) 2252 (70) 1.85 (1.72-2.00)

Age (y)

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 10/14

12-18 12,753 (12) 699 (6) 1.0 13,312 (12) 140 (4) 1.0

19-34 19,520 (18) 1861 (14) 1.74 (1.59-1.90) 20,984 (18) 397 (12) 1.80 (1.48-2.18)

35-54 39,459 (37) 4916 (39) 2.27 (2.09-2.47) 43,134 (37) 1241 (39) 2.74 (2.29-3.26)

55-80 34,513 (32) 5260 (41) 2.78 (2.56-3.02) 38,353 (33) 1420 (44) 3.52 (2.96-4.19)

Body mass index

Normal 32,339 (30) 3061 (24) 1.0 34,685 (30) 715 (22) 1.0

Underweight 3,221 (3) 259 (2) 0.85 (0.75-0.97) 3,430 (3) 50 (2) 0.71 (0.53-0.94)

Overweight 32,700 (31) 3908 (31) 1.26 (1.20-1.33) 35,629 (31) 979 (31) 1.33 (1.21-1.47)

Obese 30,536 (29) 5008 (39) 1.73 (1.65-1.82) 34,200 (29) 1344 (42) 1.91 (1.74-2.09)

Unknown 7,449 (7) 500 (4) 0.71 (0.64-0.78) 7,839 (7) 110 (3) 0.68 (0.56-0.83)

Smoking status

Nonsmoker 59,015 (56) 6474 (51) 1.0 63,941 (55) 1548 (48) 1.0

Current smoker 16,655 (16) 2367 (19) 1.30 (1.23-1.36) 18,394 (16) 628 (20) 1.41 (1.28-1.55)

Ex-smoker 23,631 (22) 3127 (24) 1.21 (1.15-1.26) 25,951 (22) 807 (25) 1.28 (1.18-1.40)

Unknown 6,944 (6) 768 (6) 1.01 (0.93-1.09) 7,497 (6) 215 (7) 1.19 (1.03-1.37)

Charlson comorbidity index score

0 50,250 (47) 4724 (37) 1.0 53,925 (47) 1049 (33) 1.0

1-4 50,853 (48) 7181 (56) 1.50 (1.44-1.56) 56,116 (48) 1918 (60) 1.76 (1.63-1.89)

5-9 2,876 (3) 475 (4) 1.76 (1.59-1.95) 3,218 (3) 133 (4) 2.12 (1.77-2.56)

>=10 2,266 (2) 356 (3) 1.67 (1.49-1.88) 2,524 (2) 98 (3) 1.99 (1.62-2.46)

Asthma education 42,009 (40) 5347 (42) 1.11 (1.07-1.15) 45,900 (40) 1456 (46) 1.27 (1.18-1.37)

PEF % predicted

<=60 11,045 (11) 2763 (22) 2.87 (2.72-3.02) 12,931 (11) 877 (27) 3.90 (3.54-4.30)

61-79 29,804 (28) 4046 (32) 1.56 (1.48-1.63) 32,797 (28) 1053 (33) 1.85 (1.69-2.03)

>=80 43,945 (41) 3835 (30) 1.0 46,964 (41) 816 (26) 1.0

Unknown 21,451 (20) 2092 (16) 1.12 (1.06-1.18) 23,091 (20) 452 (14) 1.13 (1.00-1.26)

Blood eosinophil count (x 109/L)

<=0.4 56,856 (54) 7947 (62) 1.0 62,834 (54) 1969 (62) 1.0

>0.4 11,271 (10) 1913 (15) 1.21 (1.15-1.28) 12,608 (11) 576 (18) 1.46 (1.33-1.60)

Unknown 38,118 (36) 2876 (23) 0.54 (0.52-0.56) 40,341 (35) 653 (20) 0.52 (0.47-0.57)

Rhinitis diagnosis, active∗ 3,060 (3) 507 (4) 1.39 (1.27-1.54) 3,415 (3) 152 (5) 1.64 (1.39-1.94)

Rhinitis diagnosis/drugs, active 31,073 (29) 5239 (41) 1.69 (1.63-1.76) 34,792 (30) 1520 (48) 2.11 (1.96-2.26)

Rhinitis diagnosis, ever∗ 26,921 (25) 3723 (29) 1.22 (1.17-1.27) 29,633 (26) 1011 (32) 1.34 (1.25-1.45)

Rhinitis diagnosis/drugs, ever 72,082 (68) 9909 (78) 1.66 (1.59-1.74) 79,389 (69) 2602 (81) 2.0 (1.83-2.19)

Eczema diagnosis, active 3,741 (4) 580 (5) 1.31 (1.19-1.43) 4,159 (4) 162 (5) 1.43 (1.22-1.68)

Eczema diagnosis, ever 28,570 (27) 3643 (29) 1.09 (1.05-1.14) 31,281 (27) 932 (29) 1.11 (1.03-1.20)

GERD diagnosis, active 1,188 (1) 256 (2) 1.82 (1.58-2.08) 1,373 (1.2) 71 (2.2) 1.89 (1.49-2.41)

GERD diagnosis/drugs, active 19,890 (19) 3,971 (32) 1.97 (1.89-2.05) 22,667 (20) 1194 (37) 2.45 (2.28-2.63)

GERD diagnosis, ever 8,130 (8) 1510 (12) 1.62 (1.53-1.72) 9,217 (8) 423 (13) 1.76 (1.59-1.96)

GERD diagnosis/drugs, ever 34,421 (32) 6172 (48) 1.96 (1.89-2.04) 38,853 (34) 1740 (54) 2.36 (2.20-2.54)

Cardiovascular disease, ever 25,616 (24) 4072 (32) 1.48 (1.42-1.54) 28,590 (25) 1098 (34) 1.59 (1.48-1.72)

Ischemic heart disease diagnosis, ever 5,297 (5) 911 (7) 1.47 (1.36-1.58) 5,948 (5) 260 (8) 1.63 (1.44-1.86)

Diabetes diagnosis, ever 12,983 (12) 2122 (17) 1.44 (1.37-1.51) 14,496 (13) 611 (19) 1.65 (1.51-1.81)

Heart failure diagnosis, ever 731 (0.7) 142 (1.1) 1.63 (1.36-1.95) 835 (0.7) 38 (1.2) 1.66 (1.19-2.29)

Anxiety or depression diagnosis, ever 4,909 (5) 903 (7) 1.57 (1.46-1.69) 5,558 (5) 254 (8) 1.71 (1.50-1.95)

Nasal polyps, ever 3,159 (3) 774 (6) 2.11 (1.95-2.29) 3,672 (3) 261 (8) 2.71 (2.38-3.09)

Anaphylaxis diagnosis, ever 414 (0.4) 98 (0.8) 1.98 (1.59-2.47) 482 (0.4) 30 (0.9) 2.27 (1.57-3.29)

Beta-blockers 2,964 (3) 370 (3) 1.04 (0.93-1.16) 3,248 (2.8) 86 (2.7) 0.96 (0.77-1.19)

Nonsteroidal anti-inflammatory drugs 23,930 (23) 3932 (31) 1.54 (1.47-1.60) 26,859 (23) 1003 (31) 1.51 (1.40-1.63)

Paracetamol 23,519 (22) 4647 (36) 2.02 (1.94-2.10) 26,841 (23) 1325 (41) 2.34 (2.18-2.52)

Statins 15,709 (15) 2450 (19) 1.37 (1.31-1.44) 17,531 (15) 628 (20) 1.37 (1.25-1.49)

Preventer device

No ICS 13,697 (13) 939 (7) 0.59 (0.55-0.63) 14,434 (12) 202 (6) 0.55 (0.47-0.63)

MDI 62,370 (59) 7234 (57) 1.0 67,861 (59) 1743 (55) 1.0

BAI 5,385 (5) 436 (3) 0.70 (0.63-0.77) 5,745 (5) 76 (2) 0.52 (0.41-0.65)

DPI 24,793 (23) 4127 (33) 1.44 (1.38-1.50) 27,743 (24) 1177 (37) 1.65 (1.53-1.78)

% ICS medication possession ratio

>0-39.9 34,519 (32) 3204 (25) 0.57 (0.54-0.61) 37,098 (32) 625 (19) 0.42 (0.38-0.47)

40-59.9 20,930 (20) 2444 (19) 0.72 (0.68-0.77) 22,775 (20) 599 (19) 0.66 (0.59-0.74)

60-79.9 8,144 (8) 1241 (10) 0.94 (0.87-1.01) 9,029 (8) 356 (11) 0.98 (0.86-1.13)

80-100 13,328 (12) 2165 (17) 1.0 14,896 (13) 597 (19) 1.0

>=100 15,627 (15) 2743 (22) 1.08 (1.02-1.15) 17,551 (15) 819 (26) 1.16 (1.05-1.30)

No ICS prescribed 13,697 (13) 939 (7) 0.42 (0.39-0.46) 14,434 (12) 202 (6) 0.35 (0.30-0.41)

ICS prescriptions

0 13,697 (13) 939 (8) 1.0 14,434 (13) 202 (6) 1.0

1-3 46,896 (44) 3999 (31) 1.24 (1.16-1.34) 50,142 (43) 753 (24) 1.07 (0.92-1.26)

>=4 45,652 (43) 7798 (61) 2.49 (2.32-2.67) 51,207 (44) 2243 (70) 3.13 (2.71-3.62)

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 11/14

ICS inhalers

0 13,697 (13) 939 (8) 1.0 14,434 (13) 202 (6) 1.0

1-3 38,220 (36) 3100 (24) 1.18 (1.10-1.28) 40,761 (35) 559 (18) 0.98 (0.84-1.16)

>=4 54,328 (51) 8697 (68) 2.34 (2.18-2.52) 60,588 (52) 2437 (76) 2.87 (2.49-3.34)

ICS prescribed dose†

0 13,697 (13) 939 (7) 1.0 14,434 (13) 202 (6) 1.0

<400 52,506 (49) 4187 (33) 1.16 (1.08-1.25) 55,906 (48) 787 (25) 1.01 (0.86-1.18)

>=400 40,042 (38) 7610 (60) 2.77 (2.58-2.98) 45,443 (39) 2209 (69) 3.47 (3.00-4.02)

ICS average daily dose†

0 13,697 (13) 939 (7) 1.0 14,434 (13) 202 (6) 1.0

<400 79,400 (75) 8143 (64) 1.49 (1.39-1.60) 85,778 (74) 1765 (55) 1.47 (1.27-1.71)

>=400 13,148 (12) 3654 (29) 4.05 (3.76-4.37) 15,571 (13) 1231 (38) 5.65 (4.86-6.56)

ICS actual duration (d)

<=100 43,117 (41) 3374 (26) 1.0 45,837 (40) 654 (21) 1.0

101-219 34,177 (32) 4454 (35) 1.67 (1.59-1.75) 37,503 (32) 1128 (35) 2.11 (1.92-2.33)

>=220 28,951 (27) 4908 (39) 2.17 (2.07-2.27) 32,443 (28) 1416 (44) 3.06 (2.79-3.36)

ICS prescribed duration (d)

<=200 37,484 (35) 3116 (24) 1.0 39,966 (35) 634 (20) 1.0

201-319 33,676 (32) 4025 (32) 1.44 (1.37-1.51) 36,725 (32) 976 (30) 1.67 (1.52-1.86)

>=320 35,085 (33) 5595 (44) 1.92 (1.84-2.01) 39,092 (34) 1588 (50) 2.56 (2.34-2.82)

SABA prescriptions

0 11,051 (10) 941 (7) 1.0 11,783 (10) 209 (6) 1.0

1-3 55,143 (52) 4897 (39) 1.04 (0.97-1.12) 59,060 (51) 980 (31) 0.94 (0.81-1.09)

>=4 40,051 (38) 6898 (54) 2.03 (1.88-2.17) 44,940 (39) 2009 (63) 2.52 (2.19-2.92)

SABA inhalers

0 11,051 (11) 941 (7) 1.0 11,783 (10) 209 (7) 1.0

1-3 45,978 (43) 3925 (31) 1.00 (0.93-1.08) 49,151 (43) 752 (24) 0.86 (0.74-1.01)

>=4 49,216 (46) 7870 (62) 1.88 (1.75-2.02) 54,849 (47) 2237 (70) 2.29 (1.99-2.66)

SABA dose†

0 11,051 (10) 941 (7) 1.0 11,783 (10) 209 (7) 1.0

1-200 46,452 (44) 4015 (32) 1.02 (0.94-1.09) 49,692 (43) 775 (24) 0.88 (0.76-1.03)

201-400 26,490 (25) 3376 (27) 1.50 (1.39-1.62) 29,057 (25) 809 (25) 1.57 (1.35-1.84)

>400 22,252 (21) 4404 (35) 2.33 (2.16-2.51) 25,251 (22) 1405 (44) 3.14 (2.71-3.63)

LTRA prescriptions

0 101,223 (95) 10,763 (85) 1.0 109,546 (95) 2440 (76) 1.0

>=1 5,022 (5) 1973 (15) 3.69 (3.49-3.91) 6,237 (5) 758 (24) 5.46 (5.01-5.95)

LABA prescriptions

0 99,401 (94) 11,327 (89) 1.0 107,950 (93) 2778 (87) 1.0

>=1 6,844 (6) 1409 (11) 1.81 (1.70-1.92) 7,833 (7) 420 (13) 2.08 (1.87-2.32)

Spacer use

No 100,138 (94) 11,523 (91) 1.0 108,812 (94) 2849 (89) 1.0

Yes 6,107 (6) 1213 (10) 1.73 (1.62-1.84) 6,971 (6) 349 (11) 1.91 (1.71-2.14)

BTS step therapy (missing n = 13)§

1 12,983 (12) 778 (6) 1.0 13,615 (12) 146 (5) 1.0

2 36,863 (35) 2359 (19) 9.91 (8.85-11.10) 38,869 (34) 353 (11) 21.32 (17.5- 25.9)

3 25,354 (24) 2483 (20) 9.28 (8.42-10.23) 27,335 (24) 502 (16) 25.18 (21.7- 29.3)

4 29,689 (28) 6315 (50) 6.07 (5.51-6.68) 34,207 (30) 1797 (56) 12.45 (10.8- 14.3)

5 1,345 (1) 799 (7) 2.79 (2.55-3.06) 1,745 (2) 399 (12) 4.35 (3.97-4.9)

Asthma attacks

<2 103,116 (97) 9525 (75) 1.0 110,875 (96) 1766 (55) 1.0

>=2 3,129 (3) 3211 (25) 11.11 (10.5-11.7) 4,908 (4) 1432 (45) 18.32 (17.0- 19.8)

<4 106,055 (99.8) 12 156 (95) 1.0 115 409 (99.7)

2802 (88) 1.0

>=4 190 (0.2) 580 (5) 26.6 (22.6-31.4) 374 (0.3) 396 (12) 43.6 (37.7-50.5)

Acute respiratory events

0 77,098 (73) 4289 (34) 1.0 80,779 (70) 608 (19) 1.0

1 20,828 (20) 3710 (29) 3.20 (3.06-3.36) 23,791 (21) 747 (23) 4.17 (3.74-4.65)

>=2 8,319 (8) 4737 (37) 10.20 (9.77- 10.74)

11,213 (10) 1843 (58) 21.84 (19.9- 23.9)

Acute OCS courses

0 92,120 (87) 6150 (48) 1.0 97,310 (84) 960 (30) 1.0

1 11,106 (10) 3448 (27) 4.65 (4.44-4.87) 13,717 (12) 837 (26) 6.18 (5.63-6.80)

>=2 3,019 (3) 3138 (25) 15.57 (14.7-16.5) 4,756 (4) 1401 (44) 29.86 (27.4- 32.6)

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 12/14

Acute OCS courses (with lower respiratory consultation)

0 103,833 (98) 11284 (89) 1.0 112,499 (97) 2618 (82) 1.0

1 2,252 (2.1) 1184 (9) 4.84 (4.50-5.21) 3,004 (2.6) 432 (13) 6.18 (5.55-6.88)

>=2 160 (0.2) 268 (2) 15.41 (12.7-18.8) 280 (0.2) 148 (5) 22.7 (18.5-27.8)

Antibiotics (with lower respiratory consultation)

0 83,644 (79) 6603 (52) 1.0 88,966 (77) 1281 (40) 1.0

1 16,441 (15) 3251 (26) 2.51 (2.39-2.62) 18,886 (16) 806 (25) 2.96 (2.71-3.24)

>=2 6,160 (6) 2882 (23) 5.93 (5.64-6.24) 7,931 (7) 1111 (35) 9.73 (8.95-10.6)

Asthma-related ED admissions‡

0 105,760 (99) 12525 (98) 1.0 115,177 (99) 3108 (97) 1.0

>=1 485 (1) 211 (2) 3.67 (3.12-4.33) 606 (1) 90 (3) 5.51 (4.40-6.89)

Asthma consultations

0 34,564 (32) 2803 (22) 1.0 36,776 (32) 591 (18) 1.0

1 46,704 (44) 4411 (35) 1.16 (1.11-1.22) 50,219 (43) 896 (28) 1.11 (1.00-1.23)

>=2 24,977 (24) 5522 (43) 2.73 (2.59-2.86) 28,788 (25) 1711 (54) 3.70 (3.36-4.07)

Primary care consultations

0 5,363 (5) 255 (2) 1.0 5,557 (5) 61 (2) 1.0

1-5 52,274 (49) 3749 (29) 1.51 (1.32-1.72) 55,348 (48) 675 (21) 1.11 (0.85-1.44)

6-12 34,909 (33) 5165 (41) 3.11 (2.73-3.54) 38,760 (33) 1314 (41) 3.09 (2.39-3.99)

>=13 13,699 (13) 3567 (28) 5.47 (4.80-6.24) 16,118 (14) 1148 (36) 6.49 (5.01-8.41)

Asthma is limiting daily activities¶ 6,388 (6) 1396 (11) 2.14 (1.99-2.29) 7,327 (6) 457 (14) 2.88 (2.55-3.27)

Asthma is limiting night activities¶ 5,199 (5) 1062 (8) 1.83 (1.69-1.98) 5,905 (5) 356 (11) 2.46 (2.16-2.80)

Asthma is causing daytime symptoms¶ 24,270 (23) 3420 (27) 1.63 (1.52-1.74) 26,754 (63) 936 (29) 2.26 (1.96-2.60)

GINA controlǁ

Not available 30,739 (29) 3520 (28) 33,363 (29) 896 (28)

Controlled 12,833 (17) 881 (10) 1.0 13,562 (16) 152 (7) 1.0

Partly controlled 53,889 (71) 6376 (69) 1.72 (1.60-1.85) 58,765 (71) 1500 (65) 2.27 (1.93-2.69)

Uncontrolled 8,784 (12) 1959 (21) 3.25 (2.98-3.53) 10,093 (12) 650 (28) 5.75 (4.81-6.87)

References 1 P. Haldar, I.D. Pavord, D.E. Shaw, M.A. Berry, M. Thomas, C.E. Brightling, Cluster analysis and clinical asthma phenotypes, Am J Respir Crit Care Med, Vol. 178, 2008, 218-224

2 J. Bousquet, E. Mantzouranis, A.A. Cruz, N. Ait-Khaled, C.E. Baena-Cagnani, E.R. Bleecker, Uniform definition of asthma severity, control, and exacerbations: document presented for the World Health Organization Consultation on Severe Asthma, J Allergy Clin Immunol, Vol. 126, 2010, 926-938

3 H.R. Anderson, R. Gupta, D.P. Strachan, E.S. Limb, 50 years of asthma: UK trends from 1955 to 2004, Thorax, Vol. 62, 2007, 85-90

4 T.R. Bai, J.M. Vonk, D.S. Postma, H.M. Boezen, Severe exacerbations predict excess lung function decline in asthma, Eur Respir J, Vol. 30, 2007, 452-456

5 M. Thomas, A. Bruton, M. Moffat, J. Cleland, Asthma and psychological dysfunction, Prim Care Respir J, Vol. 20, 2011, 250-256

6 K. Bahadori, M.M. Doyle-Waters, C. Marra, L. Lynd, K. Alasaly, J. Swiston, Economic burden of asthma: a systematic review, BMC Pulm Med, Vol. 9, 2009, 24

7 M. Schmidt, J.B. Jacobsen, T.L. Lash, H.E. Botker, H.T. Sorensen, 25 year trends in first time hospitalisation for acute myocardial infarction, subsequent short and long term mortality, and the prognostic impact of sex and comorbidity: a Danish nationwide cohort study, BMJ, Vol. 344, 2012, e356

8 M. Romagnoli, G. Caramori, F. Braccioni, F. Ravenna, E. Barreiro, N.M. Siafakas, Near-fatal asthma phenotype in the ENFUMOSA cohort, Clin Exp Allergy, Vol. 37, 2007, 552- 557

9 M.K. Miller, J.H. Lee, P.D. Blanc, D.J. Pasta, S. Gujrathi, H. Barron, TENOR risk score predicts healthcare in adults with severe or difficult-to-treat asthma, Eur Respir J, Vol. 28, 2006, 1145-1155

10 F. Wang, X.Y. He, K.J. Baines, L.P. Gunawardhana, J.L. Simpson, F. Li, Different inflammatory phenotypes in adults and children with acute asthma, Eur Respir J, Vol. 38, 2011, 567-574

11 I.D. Pavord, S. Korn, P. Howarth, E.R. Bleecker, R. Buhl, O.N. Keene, Mepolizumab for severe eosinophilic asthma (DREAM): a multicentre, double-blind, placebo-controlled trial, Lancet, Vol. 380, 2012, 651-659

12 T. Krones, H. Keller, A. Sonnichsen, E.M. Sadowski, E. Baum, K. Wegscheider, Absolute cardiovascular disease risk and shared decision making in primary care: a randomized controlled trial, Ann Fam Med, Vol. 6, 2008, 218-227

13 Global Initiative for Asthma. GINA report, Global Strategy for Asthma Management and Prevention. 2015. Available from: http://www.ginasthma.org/, . Accessed May 4, 2016.

14 British Thoracic Society, Scottish Intercollegiate Guidelines Network. British guideline on the management of asthma: a national clinical guideline (SIGN 141). October 2014. Available from: http://www.sign.ac.uk/guidelines/fulltext/141/, . Accessed May 4, 2016.

15 J.D. Blakey, K. Woulnough, A.C. James, J. Fellows, M. Obeidat, V. Navaratnam, A systematic review of factors associated with future asthma attacks to inform a risk assessment questionnaire, Thorax, Vol. 67, 2012, A31-A32

16 The Triple A Test: Avoid Asthma Attacks (Asthma UK Risk Test). Asthma UK. Available from: https://www.asthma.org.uk/advice/manage-your-asthma/risk-test/, . Accessed February 8, 2016.

17 J.R. Smith, M.J. Noble, S. Musgrave, J. Murdoch, G.M. Price, G.R. Barton, The At-Risk Registers in Severe Asthma (ARRISA) study: a cluster-randomised controlled trial examining effectiveness and costs in primary care, Thorax, Vol. 67, 2012, 1052-1060

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 13/14

Details

Subject Patients; Medical records; Electronic health records; Asthma; Smoking; Clinical medicine; Studies;

18 E.D. Bateman, R. Buhl, M. Humbert, H.K. Reddel, M.R. Sears, Development and validation of a novel risk score for asthma exacerbations: the risk score for exacerbations, J Allergy Clin Immunol, Vol. 135, 2015, 1457-1464.e4

19 J. Travers, S. Marsh, M. Williams, M. Weatherall, B. Caldwell, P. Shirtcliffe, Thorax, Vol. 62, 2007, 219-223

20 Optimum Patient Care Research Database (OPCRD). Available from: http://www.optimumpatientcare.org/Html_Docs/OPCRD.html, . Accessed June 4, 2016.

21 Respiratory Effectiveness Group (REG). Available from: http://www.effectivenessevaluation.org/, . Accessed June 4, 2016.

22 NHS Employers. Quality and Outcomes Framework. Available from: http://www.nhsemployers.org/your-workforce/primary-care-contacts/general-medical-services/quality- and-outcomes-framework, . Accessed June 4, 2016.

23 M.E. Charlson, P. Pompei, K.L. Ales, C.R. MacKenzie, A new method of classifying prognostic comorbidity in longitudinal studies: development and validation, J Chronic Dis, Vol. 40, 1987, 373-383

24 Understanding HSMRs: a toolkit on hospital standardised mortality ratios, version 9. Available from: http://www.drfoster.com/dr-foster-learning-labs-modules/, . Accessed June 4, 2016.

25 H.K. Reddel, D.R. Taylor, E.D. Bateman, L.P. Boulet, H.A. Boushey, W.W. Busse, Am J Respir Crit Care Med, Vol. 180, 2009, 59-99

26 E.R. Bleecker, A. Long, D. Tashkin, S. Peters, D. Klingman, Subacute lack of asthma control and acute asthma exacerbation history as predictors of subsequent acute asthma exacerbations: evidence from managed care data, J Asthma, Vol. 47, 2010, 422-428

27 M.L. Osborne, K.L. Pedula, K.M. Ettinger, T. Stibolt, A.S. Buist, Assessing future need for acute care in adult asthmatics: the Profile of Asthma Risk Study: a prospective health maintenance organization-based study, Chest, Vol. 132, 2007, 1151-1161

28 Joint British Societies (JBS) for the prevention of cardiovascular disease. JBS3 cardiovascular risk assessment calculator. Available from: http://www.jbs3risk.com/JBS3Risk.swf, . Accessed June 4, 2016.

29 Royal College of Physicians. Why asthma still kills: the National Review of Asthma Deaths (NRAD) Confidential Enquiry Report. May 2014. Available from: https://www.rcplondon.ac.uk/sites/default/files/why-asthma-still-kills-full-report.pdf, . Accessed June 4, 2016.

30 J.D. Blakey, S. Zaidi, D.E. Shaw, Defining and managing risk in asthma, Clin Exp Allergy, Vol. 44, 2014, 1023-1032

31 E. Melen, B.E. Himes, J.M. Brehm, N. Boutaoui, B.J. Klanderman, J.S. Sylvia, Analyses of shared genetic factors between asthma and obesity in children, J Allergy Clin Immunol, Vol. 126, 2010, 631-637.e1-8

32 O. Sideleva, B.T. Suratt, K.E. Black, W.G. Tharp, R.E. Pratley, P. Forgione, Obesity and asthma: an inflammatory disease of adipose tissue not the airway, Am J Respir Crit Care Med, Vol. 186, 2012, 598-605

33 B. Pattnaik, M. Bodas, N.K. Bhatraju, T. Ahmad, R. Pant, R. Guleria, IL-4 promotes asymmetric dimethylarginine accumulation, oxo-nitrative stress, and hypoxic response- induced mitochondrial loss in airway epithelial cells, J Allergy Clin Immunol, Vol. 138, 2016, 130-141.e9

34 D.B. Price, A. Rigazio, J.D. Campbell, E.R. Bleecker, C.J. Corrigan, M. Thomas, Blood eosinophil count and prospective annual asthma disease burden: a UK cohort study, Lancet Respir Med, Vol. 3, 2015, 849-858

35 D. Price, A.M. Wilson, A. Chisholm, A. Rigazio, A. Burden, M. Thomas, Predicting frequent asthma exacerbations using blood eosinophil count and other patient data routinely available in clinical practice, J Asthma Allergy, Vol. 9, 2016, 1-12

36 J. Chong, C. Haran, B.F. Chauhan, I. Asher, Intermittent inhaled corticosteroid therapy versus placebo for persistent asthma in children and adults, Cochrane Database Syst Rev, Vol. 7, 2015, CD011032

37 M. Engelkes, H.M. Janssens, J.C. de Jongste, M.C. Sturkenboom, K.M. Verhamme, Medication adherence and the risk of severe asthma exacerbations: a systematic review, Eur Respir J, Vol. 45, 2015, 396-407

38 L.K. Williams, E.L. Peterson, K. Wells, B.K. Ahmedani, R. Kumar, E.G. Burchard, Quantifying the proportion of severe asthma exacerbations attributable to inhaled corticosteroid nonadherence, J Allergy Clin Immunol, Vol. 128, 2011, 1185-1191.e2

39 D. Shaw, R. Green, M. Berry, S. Mellor, B. Hargadon, M. Shelley, A cross-sectional study of patterns of airway dysfunction, symptoms and morbidity in primary care asthma, Prim Care Respir J, Vol. 21, 2012, 283-287

40 J.S. Taggar, T. Coleman, S. Lewis, L. Szatkowski, The impact of the Quality and Outcomes Framework (QOF) on the recording of smoking targets in primary care medical records: cross-sectional analyses from The Health Improvement Network (THIN) database, BMC Public Health, Vol. 12, 2012, 329

41 J.K. Quint, H. Mullerova, R.L. DiSantostefano, H. Forbes, S. Eaton, J.R. Hurst, Validation of chronic obstructive pulmonary disease recording in the Clinical Practice Research Datalink (CPRD-GOLD), BMJ Open, Vol. 4, 2014, e005540

42 T. Walley, A. Mantgani, The UK General Practice Research Database, Lancet, Vol. 350, 1997, 1097-1099

43 R.L. Tannen, M.G. Weiner, D. Xie, Use of primary care electronic medical record database in drug efficacy research on cardiovascular outcomes: comparison of database and randomised controlled trial findings, BMJ, Vol. 338, 2009, b81

Copyright Elsevier Limited Jul 1, 2017

1/5/22, 9:01 PM https://www.proquest.com/printviewfile?accountid=14872

https://www.proquest.com/printviewfile?accountid=14872 14/14

Rhinitis; Health risk assessment; Cardiovascular disease; Risk factors; Primary care; Gastroesophageal reflux; Eosinophilia; Eczema; Chronic obstructive pulmonary disease; Polyps; Electronic medical records; Respiratory function; Databases

Location United Kingdom--UK Identifier / keyword Asthma; Attack; Control; Medical record; Observational; Risk factor Title Identifying Risk of Future Asthma Attacks Using UK Medical Record Data: A Respiratory Effectiveness Group Initiative Author Blakey, John D; Price, David B; Pizzichini, Emilio; Popov, Todor A; Dimitrov, Borislav D; Postma, Dirkje S; Josephs, Lynn K; Kaplan, Alan;

Papi, Alberto; Kerkhof, Marjan; Hillyer, Elizabeth V; Chisholm, Alison; Thomas, Mike Publication title Journal of Allergy and Clinical Immunology. In Practice; Amsterdam Volume 5 Issue 4 Pages 1015-1024 Publication year 2017 Publication date Jul 1, 2017 Section Original Article Publisher Elsevier Limited Place of publication Amsterdam Country of publication United Kingdom, Amsterdam Publication subject Medical Sciences--Allergology And Immunology ISSN 22132198 e-ISSN 22132201 Source type Scholarly Journal Language of publication English Document type Journal Article DOI http://dx.doi.org/10.1016/j.jaip.2016.11.007 ProQuest document ID 1917932140 Document URL https://www.proquest.com/scholarly-journals/identifying-risk-future-asthma-attacks-using-uk/docview/1917932140/se-2 Copyright Copyright Elsevier Limited Jul 1, 2017 Last updated 2020-03-30 Database ProQuest One Academic

Database copyright © 2022 ProQuest LLC. All rights reserved. Terms and Conditions