This cross sectional diagnostic study was performed between October 2003 and May 2004 in six general practices. The results from spirometry (FEV₁/VC = Tiffeneau-Quotient) and the variability of PEF-measurement within two weeks were used as a reference standard.

Sixty adult patients coming to the general practitioner for the first-time with complaints leading to the assumption of some kind of OAD were included consecutively. They presented symptoms like dyspnea, coughing, expectoration or self-reported wheezing. The patients have not been diagnosed earlier for OAD and they have not received any anti-obstructive medicine before. Other exclusion criteria related to well known side effects of inhaled sympathomimetics, namely untreated hyperthyreosis, unstable coronary artery disease or cardiac arrhythmia. Pregnancy also leads to exclusion. The study was approved by the Medical Ethics Committee of the University Heidelberg. Patients gave written informed consent.

Medical history was taken with a structured interview similar to the asthma screening questionnaire of the European Community Respiratory Health Survey (ECRHS) 13. This questionnaire was designed to assess the prevalence of asthma (= reversible OAD) in several European areas. The interview was expanded to symptoms of COPD (= irreversible OAD) (table 1). Patients were asked for the severity of their dyspnea, categorized as dyspnea at rest (severe); dyspnea while walking straight ahead (moderate); dyspnea while taking the stairs of two floors (mild) and no dyspnea. Patients were questioned about their smoking habits, allowing calculation of the number of pack years (= [years of smoking: 20] × number of daily cigarettes). The physician performed the physical examination, always including auscultation of the patients' lungs. All findings were documented and checked for completeness.

Spirometry was performed in the general practices by two research assistants, who were initially trained at the pulmonary outpatient clinic of the University Hospital of Heidelberg. In case of an unacceptable flow-volume curve, the spirometric software gave an error message. The best of three consecutive spirometry recordings was used, according to the guidelines of the American Thoracic Society (ATS) 14. The maximal inspiratory and expiratory flow volume curves were generated by forced deep inspiration and expiration with short intervening periods of tidal breathing. The forced inspiratory manoeuvres were always done after forced expiratory manoeuvres.

Every patient received a bronchodilation test with an additional performance of spirometry 20 minutes after inhaling Salbutamol through a spacer. An obstructive airway disease was diagnosed if FEV₁/VC ≤ 70%. The obstruction was classified as not reversible on Salbutamol, if the bronchodilation response Δ FEV₁ was below 12% of the baseline and below 200 ml 15. Blood was taken after the spirometry manoeuvre for determining CRP, which was assessed with an automated chemiluminometric assay kit.

Patients with no signs of airway obstruction after the results of the spirometry (i.e. FEV₁/VC > 70%) received a PEF-meter (Vitalograph^®) in order to perform PEF-measurements twice daily for 14 days. The patients were trained in performing this maneuver and documented the results in a diary. If the variability ((highest value - lowest value)/highest value) within these two weeks was above 20 %, reversible airway obstruction (= asthma) was diagnosed according to the international asthma guidelines 15. The final diagnosis (reference standard) was confirmed by the main investigator (A.S.) and the pulmonologist (J.M.), who were kept unaware of the measured inspiratory parameters and CRP.

As the sensitivities and specificities of clinical signs and symptoms and spirometric parameters are yet unknown for OAD diagnosed in primary care, a proper power calculation could not be performed. The data were analyzed with SPSS 11.0 for Windows.

The CRP values were log-transformed and entered the ROC-analyses to assess the best cut-off point. Sensitivity, specificity, and predictive values were calculated from 2 × 2 tables at the best cut-off-point. The best cut-off-point was chosen at the highest sum of positive and negative predictive value (PPV and NPV). CRP and clinical signs and symptoms were compared with the reference outcomes of spirometry and PEF-protocol. 95% confidence intervals were calculated using Wilson's method 16. Binary logistic regression was done to estimate the diagnostic odds ratios (OR) of CRP and clinical signs and symptoms (as independent variables) for the presence of OAD (as dependent variable). Positive likelihood ratios (LR+) were calculated to receive the ratio of abnormal findings in ill and healthy patients. Negative likelihood ratios (LR-) were calculated for the ratio of normal findings in ill and healthy patients. 95% confidence intervals were derived from the log method 17.

In subgroup analyses, diagnostic values of CRP were calculated in combination with dyspnea, smoking history, coughing, expectoration, wheezing and abnormal finding in auscultation. In patients with the established diagnosis of OAD, the relationships between severity of dyspnea, CRP and spirometric parameters respectively were investigated using Spearman's correlation coefficient (for ordinal variables) and Pearson's correlation coefficient (for continuous variables).

Two-thirds of the 60 patients were female; the average age was 46 years. 27 (45%) had a reversible OAD, of which 8 were identified by spirometry and 19 through the PEF-protocol (figure 1).

Ten showed an airway obstruction irreversible after inhalation of Salbutamol; one patient had a restrictive pattern. Fourteen (23%) patients with PEF-variability below 20 % within two weeks did not show abnormal signs in spirometry. Eight (13%) patients with inconspicuous spirometry did not bring back the PEF-protocol, assuring that they had no problems anymore. Thus, in total 37 (62%) did have OAD, whereas 23 (38%) patients were diagnosed as not having OAD.

The patients without OAD were four years younger than those with reversible OAD and the latter ones were eight years younger than those with not reversible OAD (table 1). The patients complained most frequently about wheezing (65%) and dyspnea (58%), coughing (43%) and expectoration (37%). Dyspnea and expectoration were mainly associated with OAD, whereas coughing was related with not having OAD. Wheezing was more pronounced for those having OAD. Waking up by an attack of coughing seems not to distinguish healthy from ill patients, as there is a similar distribution in all patients. Patients with non-reversible OAD seem to have a common cold (60%) more often, taking more than 10 days for recovering (70%). These patients also had the highest tobacco use (21 pack years). Those with reversible OAD had the lowest use (9 pack years).

The best cut-off point for CRP was at 2 mg/l, showing a diagnostic OR of 4.4 (table 2). The five symptoms with the best test characteristics are listed in table 2. Only self reported wheezing was significantly associated with having OAD (OR 3.4), showing a sensitivity of 75.7% and specificity of 52.2%. The sensitivity of "often coughing" was only 29.7% and the specificity was 34.8%. Thus, coughing was negatively related with OAD (OR 0.2). Moreover, auscultation was not significantly associated with the diagnosis.

The diagnostic odds ratio of CRP increased when it was combined with the clinical symptom dyspnea (OR 8.5; 95% CI 1.7–42.3) or a positive smoking history (OR 8.4; 95% CI 1.5–48.9) (not in table). However, confidence intervals were wide because of small sample sizes (n = 35 in both subgroups). There was no significant improvement when CRP was combined with the other clinical signs and symptoms.

There was a significant association between CRP and the severity of dyspnea in patients with established diagnosis (correlation coefficient_Spearman [co_sp] = 0.46) (table 3). There was also a significant relationship between severity of dyspnea and decrease of FEV₁ (co_sp = -0.52) and FIV₁ (co_sp = -0.41). CRP increased with the number of pack years (correlation coefficient_Pearson = 0.648) (table 4).