Validation and reproducibility of the lung function questionnaire (LFQ) for the diagnosis of COPD in Colombia

Chronic obstructive pulmonary disease (COPD) is a highly frequent condition that requires early preventive and diagnostic measures so that proper management and treatment can be implemented (1). However, despite the high prevalence of COPD, ranging from 4% to 12.16% worldwide (2,3,4), and the fact that the initial diagnosis of the disease can be done based on a simple clinical approach, the under-diagnosis of this condition is high and can vary from 15% to 81% in different populations (5,6,7), affecting the proper management of patients and increasing the economic costs of the disease (5). To address this problem, in recent years, different strategies have emerged for trying to reduce the misdiagnosis: among them are the search for cases through conventional spirometry, the use of portable spirometers, peak expiratory flow measurement (PEF), use of standardised questionnaires and a combination of them. These strategies have shown an increase in the diagnosis of COPD but with variable validity and cost-effectiveness results (8,9,10).

The use of targeted questionnaires remains a promising strategy, since they can be implemented in different scenarios and populations, without requiring specialised equipment or personnel to use them (10,11). Among the different questionnaires currently used for the diagnosis of COPD are the Lung Function Questionnaire (LFQ), a self-administered questionnaire based on five main questions derived from the Third National Health and Nutrition Examination Survey (NHANES III) (12), and the Likert scale questionnaire, which has shown, in initial validation studies, an area under the receiver operating characteristic (ROC) curve ranging from 0.652 to 0.720, sensitivity from 73.2% to 82.6% and specificity from 47.8% to 58.2% in a primary care population in the United States, with validation studies recommended in different populations (13,14).

The early recognition of subjects with COPD through questionnaires and, in this case, through the LFQ questionnaire can optimise the use of health resources, optimise the use of spirometry and implement early management measures, such as strategies to stop smoking and exposure to other risk factors (15). On the other hand, improving the follow-up and management of underdiagnosed patients could decrease exacerbations and thus improve exercise capacity and prognosis in these patients (16). However, in our environment, the reproducibility and validity of this questionnaire for fixed airflow obstruction is unknown. This study aims to determine the diagnostic performance value of this questionnaire in our environment.

A cross-sectional study was performed, with analysis of reproducibility and diagnostic validity of the use of LFQ in a population of subjects over 40 years of age who underwent a spirometry pre- and post-beta-2-agonist administration in the Pulmonary Function Laboratory of the University of La Sabana Clinic in Chia, Colombia. The period of recruitment was between January 2015 and June 2019. The data collection was carried out in two visits. During the first visit, the data obtained included identification variables, sex, height, weight, race, education level, presence of respiratory symptoms, age at onset of respiratory symptoms, history of smoking, year package index, exposure to wood smoke, passive smoking and prior history of COPD and asthma. The LFQ questionnaire was administered, and the pre- and post-beta-2-agonist spirometry was performed, measuring the forced vital capacity (FVC), forced expiratory volume in the first second (FEV1) and the FEV1/FVC ratio, as well as the percentage change of FVC and FEV1. In the second visit, which occurred 24 h to 15 days after the first one, the LFQ questionnaire was again completed for the analysis of reproducibility.

The LFQ questionnaire in the Spanish version consists of the following five questions:

How often do you have a cough with mucus production?

Each question has several response options that score from the highest to the lowest on a 1–5 Likert scale; the cutoff point for discrimination is 18 points, i.e. if a score ≤18 is obtained, the greater is the risk of presenting the disease.

The inclusion criteria were as follows: age over 40 years, scheduled to perform spirometry regardless of medical indication, informed consent signature for voluntary participation in the study, native Spanish language speaker and availability of time to complete the questionnaire. Subjects whose spirometry did not meet the quality criteria according to the guidelines of the American Thoracic Association, subjects submitting incomplete questionnaires, and subjects with some type of disability or limitation for communication and/or understanding of the questionnaire or spirometry technique were excluded (mental retardation, dementia, stroke sequelae, hearing loss, deafness and blindness). To carry out the spirometry tests, the same trained and qualified personnel of the pulmonary function laboratory were used, with prior calibration of the equipment and measurement of the pulmonary function tests according to the validity characteristics and reproducibility. COPD was defined as the presence of fixed airflow obstruction as defined by the American Thoracic Society (ATS) as FEV1/FVC ratio <0.7 after bronchodilator administration.

Convenience sampling was performed by sequentially entering the subjects in the study. The sample size was calculated using the Epidat 4.0 programme using the diagnostic test formula, with calculation of confidence interval with unknown patient condition and taking the proportion of disease (26.7%) in the PUMA study, where the prevalence of COPD in subjects attending medical centres in the country was evaluated (17,18); the expected sensitivity value of 73.2% and expected specificity of 58.2% of the study by Hanania et al (13) were taken, with a minimum of 1129 subjects being required for a precision of 5% and a 95% confidence interval (95% CI). The sample size for the reproducibility analysis was calculated using the formula to determine the intra-class correlation coefficient (ICC). An anticipated ICC = 0.85, precision = 5, confidence level = 95% and two observers were required, requiring a minimum of 475 subjects. The p Editions were replaced with a new subject the entrance of the study.

The data was obtained automatically through an electronic collection form, which automatically entered them into an Excel spread sheet for later verification of their values by the research group for the search for transcription and correction errors. Subsequently, the database was analysed using the statistical programme SPSS version 25 (licenced), where an initial description of the study variables was made using the measures of summary of frequencies and percentages for the qualitative and average variables, respectively, and standard deviation for the quantitative variables if their distribution was normal; or medium and interquartile range for quantitative variables with non-normal distribution. Bivariate analysis was performed between the study variables and the presence or absence of COPD using the chi-square test for qualitative variables and Student's t-test or Mann–Whitney U-test for quantitative variables according to their distribution; 2 × 2 tables were constructed for diagnostic test analysis and calculation of sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio, negative likelihood ratio with the dichotomous outcome of the questionnaire ≤18 and the presence or absence of non-reversible airflow obstruction by spirometry. The construction of the ROC curve with the quantitative result of the questionnaire and the presence or absence of COPD by spirometry, the number needed to diagnose (NND) and the number needed for misdiagnosis (NNMD) were also calculated for the general population and divided by groups of respiratory symptoms, history of smoking and exposure to wood smoke. The result was considered a significant if the p-value was <0.05.

The research protocol follows the international ethical guidelines of the Declaration of Helsinki and the national ethical considerations of Resolution 8430 of 1993 and the Data Protection Law 1581 of 2012. It was presented and approved by the research committee of the University of La Sabana and by the ethics committee of the Clinic of the University of La Sabana.

A total of 1996 potentially eligible patients were admitted during the study period. A total of 397 subjects did not meet inclusion criteria, and thus, 1599 subjects were taken to the final analysis. Figure 1 shows the flowchart for inclusion of the subjects in the study. Regarding the general characteristics of the population, the average age was 65 years (SD: 11.97 years), male sex comprised 44%, mixed race comprised 86% and the prevalence of COPD was 21.3%. The general characteristics of the studied population and their lung function are shown in Tables 1 and 2.

This is the first study where the use of a clinical questionnaire for the diagnosis of COPD in our region has been evaluated. Taking into account the high frequency of patients with misdiagnosed COPD worldwide, it is important to evaluate potential tools easy to implement in the search for subjects with the disease. The evaluation of the LFQ questionnaire showed excellent ICC reproducibility: 0.884 (95% CI: 0.863–0.901) and good validity for the recognition of patients with COPD, defined as a FEV1/FVC post-bronchodilator ratio <0.7. The values obtained are similar to those obtained in American populations, where the AUC was 0.718 (95% CI: 0.673–0.763) for the sum of the total score, and in our study, we obtained a value of 0.715 (95% CI: 0.685–0.745). For the cutoff point of ≤18 for the diagnosis of obstruction, we found sensitivity in our population to be greater (91.2% vs 82.6%) and lower specificity (32.4% vs 47.8%), compared to the original validation study of Hanania et al (13,19).

The prevalence of COPD (21.3%) is higher in our study than that reported in the community, namely 14.3% in Latin America (20) and 8.9% in Colombia (21). This increase in prevalence is explained by the sampling of subjects among people who attended care centres, not among an unselected population. However, our prevalence is similar to that reported in the PUMA study (20.6%), which was carried out at the first-level care centres in Latin America (18). This prevalence increases as the positive predictive value is known. However, the use of the questionnaire increases the possibility of detecting cases of the disease regardless of the study population; the sub-diagnosis of the disease occurs both in the community and in hospitals. In our study, only 47.9% of the subjects with the disease had a previous diagnosis of COPD, showing that the recognition percentage can increase significantly with the use of this questionnaire. On the other hand, characteristics in our population regarding male sex, low educational level and history of exposure to wood smoke could also be taken into account for the early suspicion and diagnosis of the disease (22).

In the group analysis, the highest performance of AUC was found in subjects without prior evidence of respiratory symptoms, with an area of 0.798 (95% CI: 0.706–0.889) and where the NND = 2 in relation to the NNDM = 3. However, the best relationship between the NND and the NNDM is 4–14, found in subjects in whom there is fixed obstruction to the air flow and who have not been exposed to cigarette or wood smoke. Conversely, in subjects with chronic respiratory symptoms and in whom there is a clear risk factor, the use of this questionnaire has regular performance. The use of questionnaires in search of COPD in subjects with risk factors or attending medical centres is still valuable (23) and increases the diagnosis of the disease. However, its performance may be greater as a screening tool in the general population (24). In this manner, the spirometry resource could be optimised; subjects with a negative questionnaire result would not require additional evaluations, and in those with the positive questionnaire result, the conventional spirometry evaluation would be directed towards complete evaluation and thus the cost-effectiveness of spirometry in these scenarios is improved (12).

The correlation of the clinical history and respiratory symptoms to the limitation of air flow is variable. Two different observers had good agreement in determining the history of smoking with kappa of 0.95; however, it decreased with the evaluation of other clinical findings. Reproducibility for the determination of wheezing had a kappa of 0.61, dyspnoea had kappa of 0.44–0.48 and cough showed a kappa of 0.46 (25). In this sense, the reproducibility of the LFQ questionnaire is very good, with an ICC of 0.884 for the quantitative score and 0.777 for the dichotomous outcome. The medical history remains a useful tool in the recognition of subjects with COPD or at risk for it (26); however, isolated information regarding symptoms has low sensitivity, history of cigarette consumption has sensitivity of 40%, expectoration 20%, wheezing 51%, dyspnoea 33% and coughing 51%. The use of a questionnaire that integrates the symptomatic characteristics increases the instrument's sensitivity, an important characteristic in an instrument to search for cases without increasing costs or additional resources.

Nowadays, there are questionnaires such as the CDQ (27), COPD-PS (28) and IT BE COPD, among others (29), which have also been evaluated for the diagnosis of COPD, whose performance is similar to the LFQ questionnaire; however, it is not known exactly whether one is better than the others or whether they can be used interchangeably with each other, which could still be the subject of future research. However, it seems that the use of such tools could be one of the alternatives to address the problem of misdiagnosis of the disease by focusing adequately on the use of pulmonary function tests so that a balance in cost-effectiveness is achieved in the procedures for early diagnosis of the disease (30). On the other hand, studies should be conducted on the use of additional alternatives, such as portable devices or cost-efficiency biomarkers, for easy implementation as additional alternatives that may be available in the future.

In summary, this questionnaire, which evaluates five clinical symptomatology items along with additional risk factors, has good reproducibility and validity for the detection of patients with COPD. Its validity is similar to previous validation studies in other populations, ir meets the characteristics of a screening method (high sensitivity) and it can be used in our region easily and economically. Among the possible weaknesses of the study we can consider the type of population evaluated, namely that it assesses a hospital population and thus may increase the risk of disease bias. However, patients without previous symptoms were evaluated, in whom the questionnaire showed greater capacity for discrimination, a situation similar to that found in other questionnaire validation studies, where, based on a specific score, the spirometry study is recommended for air flow confirmation.

The LFQ questionnaire has good reproducibility and validity for the detection of subjects with COPD, and its use can optimise the use of conventional spirometry to confirm the diagnosis of airflow obstruction.