- Top of page
- Material and methods
Background: Epidemiologic data on asthma and allergies among adults are mainly based on questionnaires: this study validates the questions on asthma, allergic rhinitis, and conjunctivitis of a new Finnish questionnaire.
Methods: To validate questions used in a country-wide study among university students aged 18–25 years, we examined 150 subjects who had ever reported asthma or wheezing, and 140 without asthma symptoms. Questions were validated in relation to current diseases including
1) symptoms detected during the preceding year at the physician's interview
2) objective measurements, such as methacholine challenge, skin prick tests, and specific IgE.
Data were adjusted for original proportions of “asthmatics” and “nonasthmatics” in the questionnaire study.
Results: Questions on “reported asthma” and “doctor-diagnosed asthma” had good positive predictive value (PPV) and specificity in diagnosing current asthma. The question on “attacks of shortness of breath with wheezing”, and especially the question on “cough with wheezing” were most sensitive. Questions on “allergic nasal symptoms” and “allergic eye symptoms” that were “related to pollen or animals” were sensitive, but a further question on doctor's diagnosis yielded higher specificity and PPV.
Conclusions: Diagnosis-based questions were found suitable for risk-factor studies, because of their good specificity and PPV, and symptom-based questions for screening, because they were highest in sensitivity.
Our present knowledge of prevalence and risk factors for asthma and atopic disorders among adults is based mainly on questionnaire data. Questions on asthma rely on either self-reported asthma or doctor-diagnosed asthma, or on a combination of typical symptoms, such as wheezing or episodic shortness of breath. Written questionnaires on asthma and asthma-like symptoms have been available since the 1960s (1–5); more recently, in the 1990s, a video-assisted questionnaire was developed (6) and used widely among children in the International Survey of Asthma and Allergies in Childhood (ISAAC) (7).
The first questions on allergic rhinitis relied on self-report of “hay fever” (8, 9); later, questions on triggering factors, such as pollen, dust, and animals, were developed (4, 10). A symptom question on “sneezing or a runny or blocked nose” apart from “a cold or flu” was introduced in a community survey of rhinitis in London (UK) (11) and further used in the ISAAC study (7, 11). Questions on “itchy, watery eyes” have usually been included in the nasal symptom questions (7, 11, 12). In a screening questionnaire for atopy, subjects were asked about allergies to specific allergens, such as house-dust mite and pets (13).
The Tuohilampi questionnaire (14) was developed by Finnish experts for population studies of respiratory, eye, and skin allergies, and also for screening allergic patients. It includes 12 sections: asthma, cough and chronic bronchitis, allergic alveolitis, respiratory infections, rhinitis, pharyngeal and laryngeal symptoms, eye symptoms, contact dermatitis, atopy, smoking, and sick building syndrome.
The aim of the present study was to validate a new questionnaire used in a study among 10 667 first-year university students aged 18–25 years. Similar criteria to those used in clinical practice to confirm the diagnosis were used. The present work provides data on the seldom reported validation of allergic rhinitis, and especially allergic conjunctivitis.
- Top of page
- Material and methods
Subjects who had ever had asthma or shortness of breath with wheezing (“asthmatics”), and those without these symptoms (“nonasthmatics”) were equal as regards age, sex, and smoking status (Table 1). The groups differed significantly with respect to skin prick test result, a positive serum specific IgE for common allergens, and bronchial hyperreactivity (Table 1).
Table 1. Characteristics of subjects in clinical study. Group “asthmatics” based on positive questionnaire response on asthma or wheezing during lifetime, and group “nonasthmatics” without these symptoms
|Characteristic||Reported asthma and/or wheezing during lifetime n=150||No reported asthma or wheezing* during lifetime n=140|
| ||Mean (range) years 21.5 (19.3–26.5)||Mean (range) years 21.5 (18.7–26.6)|
| Current occasional smoker||14.0||21/150||9.4||13/139||NS|
| Current daily smoker||13.3||20/150||13.7||19/139||NS|
|Parental atopic disorder||51.0||76/149||36.4||51/140||0.013|
|Skin prick test >3 mm||68.7||103/150||42.1||59/140||0.001|
|Positive specific IgE‡||73.0||108/148||47.9||67/140||0.001|
Among lifetime “asthmatics”, 48.7% of the subjects fulfilled criteria of current asthma. Among “nonasthmatics”, the criteria of current asthma were fulfilled in 4.3% (6/140) (Table 2). Current allergic rhinitis and conjunctivitis were more prevalent in the group of asthmatics.
Table 2. Current asthma, allergic rhinitis, and allergic conjunctivitis in clinical examination in groups of “asthmatics” and “nonasthmatics” defined according to questionnaire responses
|Clinical examination||Questionnaire response|
|Reported asthma and/or wheezing* during lifetime n=150||No reported asthma or wheezing* during lifetime n=140|
|Current asthma†||48.7||73 ||4.3||6|
|Current allergic rhinitis†||60.8||90#||25.7||36|
|Current allergic conjunctivitis§||58.1||86#||22.9||32|
Validity of questions on asthma, attacks of shortness of breath with wheezing, and cough with wheezing
The questions on asthma and asthma-like symptoms are shown in Appendix 1. The specificity of questions on reported asthma diagnosis (question 3a), as well as on doctor-diagnosed asthma during the lifetime (3a and b), was high, 0.98 and 0.99, respectively, and these questions also had the highest positive predictive values (PPV); 0.76 and 0.82, respectively (Table 3).
Table 3. Validity of questions on lifetime asthma and asthma-like symptoms in relation to confirmed current asthma* in young adults. Validity measures adjusted for proportions of asthma and nonasthma groups in original questionnaire study. Detailed questions in Appendix 1
|Questionnaire item||Positive predictive value†||Specificity||Sensitivity|
|Cough with wheezing apart from respiratory infection||0.56||0.96||0.44|
|Attacks of shortness of breath with wheezing, breathing normal between attacks, apart from respiratory infection||0.48||0.95||0.39|
|Attacks of shortness of breath with wheezing, breathing normal between attacks||0.45||0.94||0.43|
|Attacks of shortness of breath with wheezing||0.42||0.93||0.45|
|Cough with wheezing||0.25||0.72||0.76|
The symptom question “attacks of shortness of breath with wheezing” (question 2a) had good specificity but lower PPV than the diagnosis-based questions. By the addition of the further definitions “breathing normal between the attacks” (question 2b) and “not only in relation to respiratory infection” (question 4), better specificity and PPV were achieved (Table 3).
On the contrary, the symptom question on “cough with wheezing” (question 1a) was rather unspecific (specificity 0.72), and only 25% of positive respondents fulfilled current asthma criteria (PPV 0.25). This question, however, was the most sensitive of all questions (sensitivity 0.76). When a further question on “symptoms also at times when you did not have a cold” (question 1b) was included, specificity and PPV increased, but sensitivity was lost (Table 3). The sensitivity of this question combination was comparable to the question on “shortness of breath with wheezing” (sensitivity 0.45) (Table 3), and the combination question with the further addition “breathing normal between the attacks” (sensitivity 0.43) (Table 3).
Validity of questions on allergic rhinitis and conjunctivitis
Diagnostic questions on rhinitis and conjunctivitis are shown in Appendix 2. The question on “hay fever” or “other allergic nasal symptoms (sneezing, itchy, runny nose) from, e.g., pollen and animals” (question 2a) yielded the best sensitivity, and high PPV and specificity (Table 4). Comparable results were achieved by the question on “allergic eye symptoms (watering, itching, redness of the eyes) from, e.g., pollen and animals” (question 4a) (Table 4). The addition of further questions on doctor's diagnosis (questions 2b or 4b) (Table 4) improved PPV and specificity, but sensitivity was clearly lower. On the other hand, the question on noninfectious rhinitis, “recurrent or chronic nasal symptoms (sneezing, runny or stuffy nose) not related to a cold” (question 1), was neither specific nor sensitive (Table 4) in diagnosing current allergic rhinitis. The corresponding figures were found for “irritation of the eyes not related to a cold” (question 3) (Table 4).
Table 4. Validity of questionnaire items on lifetime allergic rhinitis/conjunctivitis in relation to confirmed current allergic rhinitis/conjunctivitis*. Validity measures adjusted for proportions of asthma and nonasthma groups in original questionnaire study. Detailed questions in Appendix 2
|Questionnaire item||Positive predictive value†||Specificity||Sensitivity|
|Hay fever or nasal allergy diagnosed by doctor||0.76||0.93||0.52|
|Hay fever or allergic nasal symptoms as from pollen or animals||0.75||0.87||0.87|
|Recurrent or chronic nasal symptoms apart from respiratory infection||0.41||0.69||0.48|
|Allergic eye condition diagnosed by a doctor||0.85||0.98||0.30|
|Allergic eye symptoms as from pollen or animals||0.82||0.93||0.79|
|Eye irritation apart from respiratory infection||0.31||0.63||0.43|
- Top of page
- Material and methods
Sensitive questions are useful during the early stages of a diagnostic work-up for screening of the diseased subjects. In population studies of the risk factors of a disease, a high positive predictive value (proportion of truly diseased among positive questionnaire respondents), as well as high specificity (small number of false positive answers), is important to confirm the diagnosis.
Validation measures depend on the prevalence of the disease and the characteristics of the population. In the original postal survey, the prevalence of reported asthma and “shortness of breath with wheezing apart form respiratory infection” during lifetime was low; 5.5% and 9.1%, respectively. To obtain enough asthmatics for clinical examination, we selected two random samples based on having had or not having had symptoms suggestive of asthma. Using weights for proportions of “asthmatics” and “nonasthmatics” in the questionnaire study was considered to reflect more accurately the young adult population. With this method we could not exclude the fact that subjects with symptoms were somewhat eager to participate, although the weighting regarded subjects without any atopic symptoms as well. The weighting emphasizing examined “nonasthmatics” decreased sensitivity and increased not only the specificity of questions on asthma, but also that of questions on allergic rhinitis and conjunctivitis (data not shown). This means that among symptomatic subjects and among populations with a higher prevalence of the diseases better sensitivity and somewhat poorer specificity than shown in the present study could be assumed.
Furthermore, validity measures among the study population, young adults starting university studies, may not differ much from Finnish young adults in general, but caution is needed when generalizing the results to other age groups.
Validity of questions on asthma and asthma-like symptoms
Respiratory questionnaires have been validated mainly in relation to another questionnaire (6, 21–23), nonspecific bronchial hyperreactivity (BHR) (5, 6, 22–27), clinician's interview (28, 29), and a combination of these methods (30). In the European Community Respiratory Health Survey (ECRHS), among subjects with self-reported asthma, only 48% had BHR (31). However, BHR is fairly common among people with no asthma symptoms, in particular atopic subjects (32), smokers (33), and women (31, 34). Thus, at least among these subjects, BHR alone as the reference standard of asthma biases estimation of the validity of the questions.
We adopted typical symptoms of asthma at the clinician's interview with objective measurements, including BHR, bronchial obstruction, or reversibility of obstruction, as criteria. These criteria were chosen to match the confirmation of asthma diagnosis in clinical practice. A comparable algorithm, including asthma symptoms, diagnostic history, and lung function, was used in genetic studies of asthma, in which defining the phenotype is important (29).
For use in population studies of risk factors for asthma, questions on “asthma” and “doctor-diagnosed asthma” in the Tuohilampi questionnaire will be recommended due to their high PPV and specificity (Table 3). However, the quite low sensitivity of diagnosis-based questions still suggests underdiagnosis, due possibly to mild symptoms not requiring a doctor's consultation. Recently, among the Swedish population, self-reported asthma in the ECRHS questionnaire yielded somewhat higher sensitivity (0.48), but lower specificity (0.90) and PPV (0.28) when validated against BHR (31).
Symptom-based questions have been recommended to avoid underestimation of the prevalence, to achieve comparable results in different populations, and to screen even patients with mild symptoms. Surprisingly, in the present study, the question on “attacks of shortness of breath with wheezing” detected only 45% of asthmatics, but did not attract many false positive cases. Of these wheezers, 42% fulfilled the criteria of current asthma, in comparison to 76% and 82% among those with “asthma” and “doctor-diagnosed asthma”. When the International Union against Tuberculosis and Lung Disease (IUATLD) questionnaire in Finnish was validated for BHR among young men, the sen-sitivity of the question on “wheeze” during the preceding 12 months was much higher (0.95) and the specificity lower (0.74) (25). The higher sensitivity in that study is partly explained by characteristics of the subjects, of whom 50% were smokers and 50% were asthmatics. In the Tuohilampi questionnaire, the question on “attacks of shortness of breath” is combined “with wheezing”, leading to higher specificity and lower sensitivity than separate questions. By the addition of two further questions on “breathing normal between the attacks” and “not only at times when having a cold”, a fairly specific “asthma” question was formulated. However, the combination of questions was clearly less sensitive, a finding reported also in other studies validating multiple symptom-based questions (6, 23).
The question on “cough with wheezing” was most sensitive, detecting that 76% of asthmatics, but only 25% of positive respondents fulfilled the criteria of current asthma. When “not only at times when having a cold” was included, a higher specificity with a clear loss in sensitivity was noticed, and still only 56% fulfilled the criteria of asthma. Apparently, the validity of the question on “cough with wheezing” in detecting asthmatics will be lower in the general population, which includes a larger number of smokers and chronic bronchitis patients. A comparable question on “cough with wheezing” has not been used in earlier widely distributed questionnaires. The question on “dry cough at night” and “not associated with cold” in the ISAAC written questionnaire, was neither sensitive (sensitivity 0.38) nor specific (specificity 0.65) in predicting BHR among adolescents (23). In the IUATLD questionnaire, subjects responding to only cough questions (without wheezing) were not different from normal subjects in terms of BHR and lung function (30), suggesting that cough questions may be inappropriate in defining asthma in adults.
Validity of questions on allergic and noninfectious rhinitis and conjunctivitis
In the Tuohilampi questionnaire, the questions on nasal and eye allergies are formulated separately to assess conjunctival diseases more exactly. This approach is different from earlier questionnaires: for example, the American Thoracic Society (ATS) questionnaire asks about “rhinitis and/or conjunctivitis”, and in the ISAAC questionnaire nasal problems are followed by a question on increasing specificity: “accompanied by itchy, watery eyes”. A recent study among pregnant women evaluated a screening questionnaire for atopy that included self-reported or physician-diagnosed allergy to allergens such as those of pets and house-dust mite, and a combination of these questions (13).
Skin prick testing (11, 34) or specific IgE measurements (13, 36, 37) as criteria of allergic rhinoconjunctivitis tend to underestimate the sensitivity of the questions, because not all those sensitized have symptoms. Triggers and seasonal occurrence of symptoms are commonly used by clinicians to exclude allergy; thus, physician's diagnosis seems to agree well with specific questions (38). To avoid validation bias, we added symptoms at interview to measures of sensitization.
The prevalence of allergic rhinitis is known to be highest, and the onset of perennial rhinitis most common, in the young adult population (39). In the original questionnaire study, the lifetime prevalence of doctor-diagnosed allergic rhinitis was 19.2% and that of allergic conjunctivitis 9.9%. Students with a mean age of 22 years seemed to be aware of allergic diseases and also had had good access to allergy tests (30% were tested for allergy earlier). Thus, not only the specificity but also the sensitivity of questions referring to “allergic nasal symptoms (sneezing etc.), from pollen or animals” and “allergic eye symptoms (watering, etc.) related to pollen or animals” was reasonably high; 0.87 and 0.79, respectively. Two-thirds of these subjects also had current nasal or eye allergy in the clinical examination. The question seemed to work better in screening than the combination of questions on specific allergies studied among Dutch pregnant women, in which the sensitivity was 0.55 at the maximum (13).
In the southwest London study, the question on “sneezing/runny nose/blocked nose, when you do not have a cold or the flu” was highly sensitive and specific (sensitivity 0.96 and specificity 0.91) in identifying subjects with rhinitis among a population including a large number of symptomatic subjects (11). In the present study, the question on noninfectious rhinitis: “recurrent nasal symptoms (sneezing, runny or stuffy nose) not related to a cold or a flu” had low validity in diagnosing allergic rhinitis (Table 4), a result which is in agreement with the quite similar ISAAC rhinitis question validated among Swiss schoolchildren (35). The question, however, may work in diagnosing chronic rhinitis whether allergic or not.
In summary, among the young adult population, the questions on reported “asthma” and “doctor-diagnosed asthma” throughout life had the highest positive predictive value, thus detecting “real” asthmatics among positive questionnaire respondents, and were most specific, not detecting many asthmatics among negative respondents. On that basis, these questions are suitable for use in risk-factor studies. The combination of the most sensitive questions useful in screening, such as those on “cough with wheezing” and “attacks of shortness of breath with wheezing”, with a further item, “also apart from a cold”, considerably increased the specificity almost to that of the diagnosis-based questions, but an equally high PPV was not achieved. Questions referring to “allergic” nasal and eye symptoms throughout life worked well in screening and diagnosing, but it was by a further question, “diagnosed by doctor”, that the most exact diagnosis was obtained.