Within the framework of the European Community Respiratory Health Survey (13), a representative sample of 4,500 subjects (age range: 20 to 44 yr) was drawn from the office of population census of Hamburg. The sample comprised approximately 0.6% of the total population of the given age range. All subjects received a mailed, one-page screening questionnaire (Stage I). When the sample size was corrected for those subjects who had moved (n = 403), died (n = 4), or were outside the established age range (n = 3), the eligible sample comprised 4,090 subjects. All of them were invited for the measurement of airway hyperresponsiveness to methacholine and SO₂ provocation challenges within the hospital (Stage II). The request was accepted by 1,049 subjects, 790 of whom finally underwent the SO₂ inhalation provocation challenges in the study.

The two-step approach of the study has been described in detail elsewhere (13, 14). During each step of subject recruitment, efforts were undertaken to maximize response and participation rates through up to two reminder letters, five telephone calls, and two home visits. All subjects received financial compensation for the time they spent in the hospital for the Stage II measurements.

Besides the SO₂ challenge, Stage II comprised the administration of a detailed questionnaire, spirometric measurements, a methacholine inhalation challenge or bronchodilator inhalation test, and skin-prick testing. The study protocol had been approved by the local ethics committee, and written informed consent was obtained from all participants.

From the detailed questionnaire, which contained 71 items, answers to four selected questions about respiratory symptoms within the previous 12 mo (asthma attack, attack of breathlessness in rest, wheezing or whistling sound in the chest) and current asthma medication were selected in order to evaluate risk factors for SO₂ hyperresponsiveness.

Spirometric measurements were made with pneumotachograph-based electronic spirometers (MasterLab, and PSC-PC, Jaeger, Würzburg, Germany) that met internationally accepted quality criteria (15). In smokers, lung function testing was done at least 1 h after the last cigarette had been smoked. In subjects taking inhaled asthma medication, measurements were not made until at least 4 h had elapsed after the use of any inhaler. Subjects performed five FVC maneuvers, and the maximum FEV₁ and FVC of the two best technically satisfactory maneuvers were recorded as final values. Maneuvers were accepted as technically satisfactory if the variation of the two best FEV₁ values was below 5%, if the back-extrapolated volume was lower than 100 ml or 5% FVC, and if the expiratory time was at least 6 s. If the subjects failed to produce two technically satisfactory maneuvers after five attempts, another four attempts were allowed.

Methacholine challenges were performed in all subjects who were willing to participate, able to perform successful FEV₁ and FVC maneuvers, and did not fulfill the following exclusion criteria. Any subject who had reported a heart attack in the prior 3 mo, or who was taking medication for any heart disease or epilepsy, or who was taking β-blocking drugs was excluded from the challenge, as were pregnant and breast-feeding women. The same was true for subjects showing an FEV₁ below 70% of the mean predicted value (15) or lower than 1.5 L. Subjects who were receiving antiasthmatic treatment were given an appointment when they had taken their inhalers at least 4 h and their oral medication at least 8 h beforehand. Challenges were performed with the Mefar MB3 dosimeter and five individually calibrated dosimeters (Mefar srl, Bovezzo, Italy). After baseline values and the airway response to the isotonic saline diluent were measured, increasing concentrations of standard methacholine concentrations (Provocholine; Hoffmann La Roche, Basel, Switzerland) were given. According to a history of respiratory symptoms and baseline lung function, a short protocol with fourfold increases and a long protocol with twofold increases in methacholine concentrations was used. In either protocol, the subjects took a defined number of breaths (from one to four) at intervals of 6 s, starting from FRC, slowly inhaling to TLC, and holding the breath for 3 s. The FEV₁ maneuver was performed 2 min after inhalation of each dose. Provocation was stopped when FEV₁ had dropped by 20% as compared with postdiluent values, or after a maximum cumulative dose of methacholine of 2.0 mg had been reached. From plots of FEV₁ versus the logarithm of the cumulative dose of methacholine, provocative doses necessary to decrease FEV₁ by 20% (PD₂₀FEV₁) were derived. The details of the inhalation challenge are described elsewhere (16). Any methacholine challenge that did not fulfill the quality criteria for FEV₁ measurement was considered invalid.

Any subject with a baseline FEV₁ below 70% of the mean predicted value (15) received 200 μg of salbutamol from a metered dose inhaler (MDI) and spacer in a standardized fashion. Spirometry was performed 10 min after administration of the bronchodilator. An increase in FEV₁ of more than 15% was considered a positive response.

SO₂ challenges were performed at least 2 h after the end of a negative methacholine provocation test. Subjects with a positive methacholine challenge were given an appointment for SO₂ testing on a different day. For the SO₂ inhalation challenges, the same exclusion criteria were applied as for the methacholine inhalation challenges. Air was passed through filters, humidified with a heated water bath, and filtered once again to avoid aerosol contamination and achieve 40% relative humidity. To obtain isocapnic conditions during hyperventilation, 2.5% carbon dioxide was added under control by a flow meter (GTF; Platon Instrumentierung, Maintal, Germany). The target SO₂ concentrations were achieved by adding an appropriate flow of SO₂ (1,000 ppm; Linde AG, Unterschleißheim, Germany) under regulation with a mass-flow controller (Side-Trak 840, 0 to 100 ml · min⁻¹; Sierra Instruments, Monterey, CA). All materials in contact with SO₂ were made from stainless steel, Teflon, or other inert materials. The concentration of SO₂ was monitored continuously with a pulsed fluorescent SO₂ analyzer (No. 43; Thermo Electron Corporation, Franklin, ME). The calibration of the SO₂ concentration was checked regularly with a calibrator (Model 8550; Monitor Technologies, San Diego, CA) and a gas-mixing system approved by the Environmental Protection Agency of the state of Hamburg. The mixing chamber was connected to a target balloon in order to facilitate visual control of minute ventilation (V˙e). There were no differences in SO₂ concentration between the measurement port in the gas-mixing system and the mouthpiece. During challenges, the subjects were in a standing position and wore a nose clip.

SO₂ challenges comprised consecutive 3-min periods of voluntary isocapnic hyperventilation (40 L · min⁻¹) of filtered air and 0.25, 0.5, 1.0, and 2.0 ppm SO₂. Airway response was spirometrically assessed 2 and 5 min after the end of each inhalation period. When FEV₁ declined over the period between 2 and 5 min after challenge, another measurement, 10 min after inhalation, was added. The challenge was stopped if FEV₁ had fallen by at least 20% from the value obtained after hyperventilation of filtered air or after the maximum concentration of SO₂ had been applied. Concentration–response curves were constructed by plotting FEV₁ against the logarithm of SO₂ concentration (in ppm). The provocative concentration of SO₂ (PC₂₀FEV₁) was computed as the concentration necessary to decrease FEV₁ by 20% relative to the airway response to filtered air.

The subjects' sensitivity to 11 common aeroallergens was assessed by skin testing on the forearms with standardized allergen extracts (Allergopharma, Reinbek, Germany). The allergens applied were trees I (alder, hazel, poplar, elm, willow) and II (birch, beech tree, oak, plane tree); grass; grain; herbs (mugwort, stinging nettle, dandelion, plantain); moulds I (Alternaria, Botrytis, Cladosporium, Curvularia, Fusarium, Helminthosporium) and II (Aspergillus, Mucor, Penicillium, Pullularia, Rhizopus, Serpula); cat; house dust mites (Dermatophagoides pteronyssinus and D. farinae); storage mites (Acarus siro and Tyrophagus putrescentiae); and cockroach. Results were read after 20 min by marking the perimeter of the wheal with a ballpoint pen. A skin test was considered positive if the mean wheal diameter was at least 3 mm.

Arithmetic means and standard deviations were calculated for spirometric indices. Regression equations for responsiveness to SO₂ were calculated for each gender, taking into account age and height. For the analysis of risk factors for SO₂ hyperresponsiveness, an “impaired FEV₁” was attributed to subjects whose residuals of FEV₁ were below the 10th percentile of the regression equations.

In a logistic regression analysis, age, gender, and seven additional binary factors were entered to predict SO₂ hyperresponsiveness: (1) an affirmative answer to at least one of the four selected questions given in the detailed questionnaire; (2) a history of occupational exposure to vapors, gas, dust, or fumes; (3) a history of job change because a job affected breathing; (4) an impaired FEV₁ with residuals of the respective regression equation that fell below the 10th percentile; (5) an increased airway responsiveness to methacholine, with a PD₂₀-FEV₁ below 2.0 mg; (6) a positive skin prick test; and (7) a positive smoking history. After being entered into the model, the six additional variables were then eliminated in a stepwise manner, starting with those with the lowest predictive value. Cutoff values of 0.05 and 0.15 were used as the stepwise conditions for dropping variables out of the logistic regression model.

Potential effects of subject selection were analyzed with chi-square tests. The prevalence of positive methacholine challenge tests and the relationship between airway responsiveness to methacholine and to SO₂ were used to estimate the prevalence of increased responsiveness to SO₂ in the eligible population sample. Statistical significance was assumed for p < 0.05.