Defining six asthma subtypes may promote personalized treatment

Six subtypes of asthma that may facilitate personalized treatment were identified and confirmed in a large database review of approximately 50,000 patients, according to a recent study.

Previous studies of asthma subtypes have involved age of disease onset, the presence of allergies, and level of eosinophilic inflammation, and have been limited by factors including small sample size and lack of formal validation, Elsie M.F. Horne, MD, of the Asthma UK Centre for Applied Research, Edinburgh, and colleagues wrote.

In a study published in the International Journal of Medical Informatics, the researchers used data from two databases in the United Kingdom: the Optimum Patient Care Research Database (OPCRD) and the Secure Anonymised Information Linkage Database (SAIL). Each dataset included 50,000 randomly selected nonoverlapping adult asthma patients.

The researchers identified 45 categorical features from primary care electronic health records. The features included those directly linked to asthma, such as medications; and features indirectly linked to asthma, such as comorbidities.

The subtypes were defined by the clinically applicable features of level of inhaled corticosteroid use, level of health care use, and the presence of comorbidities, using multiple correspondence analysis and k-means cluster analysis.

The six asthma subtypes were identified in the OPCRD study population as follows: low inhaled corticosteroid use and low health care utilization (30%); low to medium ICS use (36%); low to medium ICS use and comorbidities (12%); varied ICS use and comorbid chronic obstructive pulmonary disease (4%); high ICS use (10%); and very high ICS use (7%).

The researchers replicated the subtypes with 91%-92% accuracy in an internal dataset and 84%-86% accuracy in an external dataset. “These subtypes generalized well at two future time points, and in an additional EHR database from a different U.K. nation (the SAIL Databank),” they wrote in their discussion.

The findings were limited by several factors including the retrospective design, the possible inclusion of people without asthma because of the cohort selection criteria, and the possible biases associated with the use of EHRs; however, the results were strengthened by the large dataset and the additional validations, the researchers noted.

“Using these subtypes to summarize asthma populations could help with management and resource planning at the practice level, and could be useful for understanding regional differences in the asthma population,” they noted. For example, key clinical implications for individuals in a low health care utilization subtype could include being flagged for barriers to care and misdiagnoses, while those in a high health care utilization subtype could be considered for reassessment of medication and other options.

The study received no outside funding. Lead author Dr. Horne had no financial conflicts to disclose.

Six subtypes of asthma that may facilitate personalized treatment were identified and confirmed in a large database review of approximately 50,000 patients, according to a recent study.

Previous studies of asthma subtypes have involved age of disease onset, the presence of allergies, and level of eosinophilic inflammation, and have been limited by factors including small sample size and lack of formal validation, Elsie M.F. Horne, MD, of the Asthma UK Centre for Applied Research, Edinburgh, and colleagues wrote.

In a study published in the International Journal of Medical Informatics, the researchers used data from two databases in the United Kingdom: the Optimum Patient Care Research Database (OPCRD) and the Secure Anonymised Information Linkage Database (SAIL). Each dataset included 50,000 randomly selected nonoverlapping adult asthma patients.

The researchers identified 45 categorical features from primary care electronic health records. The features included those directly linked to asthma, such as medications; and features indirectly linked to asthma, such as comorbidities.

The subtypes were defined by the clinically applicable features of level of inhaled corticosteroid use, level of health care use, and the presence of comorbidities, using multiple correspondence analysis and k-means cluster analysis.

The six asthma subtypes were identified in the OPCRD study population as follows: low inhaled corticosteroid use and low health care utilization (30%); low to medium ICS use (36%); low to medium ICS use and comorbidities (12%); varied ICS use and comorbid chronic obstructive pulmonary disease (4%); high ICS use (10%); and very high ICS use (7%).

The researchers replicated the subtypes with 91%-92% accuracy in an internal dataset and 84%-86% accuracy in an external dataset. “These subtypes generalized well at two future time points, and in an additional EHR database from a different U.K. nation (the SAIL Databank),” they wrote in their discussion.

The findings were limited by several factors including the retrospective design, the possible inclusion of people without asthma because of the cohort selection criteria, and the possible biases associated with the use of EHRs; however, the results were strengthened by the large dataset and the additional validations, the researchers noted.

“Using these subtypes to summarize asthma populations could help with management and resource planning at the practice level, and could be useful for understanding regional differences in the asthma population,” they noted. For example, key clinical implications for individuals in a low health care utilization subtype could include being flagged for barriers to care and misdiagnoses, while those in a high health care utilization subtype could be considered for reassessment of medication and other options.

The study received no outside funding. Lead author Dr. Horne had no financial conflicts to disclose.

Six subtypes of asthma that may facilitate personalized treatment were identified and confirmed in a large database review of approximately 50,000 patients, according to a recent study.

Previous studies of asthma subtypes have involved age of disease onset, the presence of allergies, and level of eosinophilic inflammation, and have been limited by factors including small sample size and lack of formal validation, Elsie M.F. Horne, MD, of the Asthma UK Centre for Applied Research, Edinburgh, and colleagues wrote.

In a study published in the International Journal of Medical Informatics, the researchers used data from two databases in the United Kingdom: the Optimum Patient Care Research Database (OPCRD) and the Secure Anonymised Information Linkage Database (SAIL). Each dataset included 50,000 randomly selected nonoverlapping adult asthma patients.

The researchers identified 45 categorical features from primary care electronic health records. The features included those directly linked to asthma, such as medications; and features indirectly linked to asthma, such as comorbidities.

The subtypes were defined by the clinically applicable features of level of inhaled corticosteroid use, level of health care use, and the presence of comorbidities, using multiple correspondence analysis and k-means cluster analysis.

The six asthma subtypes were identified in the OPCRD study population as follows: low inhaled corticosteroid use and low health care utilization (30%); low to medium ICS use (36%); low to medium ICS use and comorbidities (12%); varied ICS use and comorbid chronic obstructive pulmonary disease (4%); high ICS use (10%); and very high ICS use (7%).

The researchers replicated the subtypes with 91%-92% accuracy in an internal dataset and 84%-86% accuracy in an external dataset. “These subtypes generalized well at two future time points, and in an additional EHR database from a different U.K. nation (the SAIL Databank),” they wrote in their discussion.

The findings were limited by several factors including the retrospective design, the possible inclusion of people without asthma because of the cohort selection criteria, and the possible biases associated with the use of EHRs; however, the results were strengthened by the large dataset and the additional validations, the researchers noted.

“Using these subtypes to summarize asthma populations could help with management and resource planning at the practice level, and could be useful for understanding regional differences in the asthma population,” they noted. For example, key clinical implications for individuals in a low health care utilization subtype could include being flagged for barriers to care and misdiagnoses, while those in a high health care utilization subtype could be considered for reassessment of medication and other options.

The study received no outside funding. Lead author Dr. Horne had no financial conflicts to disclose.