An Algorithmic Approach to Chronic Cough

Methods

Patient Selection

All patients who came to our university-based pulmonary practice with a chief complaint of chronic cough were considered for inclusion in the study. Cough was considered to be chronic if it had been present for 3 weeks or longer [7, 8]. The following were exclusion criteria: 1) immunocompromise, including known lung cancer or other active malignancy, the acquired immunodeficiency syndrome, or current treatment with corticosteroids or other immunosuppressive agents; 2) cigarette use within 12 months; 3) use of an angiotensin-converting-enzyme inhibitor within 4 weeks; and 4) contraindication to the use of an antihistamine-decongestant or to bronchoprovocation challenge, including pregnancy and an FEV₁ less than 70% of predicted value.

Algorithm

The approach involved a standardized initial evaluation, weekly follow-up, a series of sequential diagnostic and therapeutic steps, and a predefined end point.

Initial Evaluation

The initial evaluation included a history; physical examination; review of previous diagnostic studies; and a questionnaire on the duration, frequency, and severity of cough, postnasal drip symptoms, dyspnea, wheeze, and symptoms of gastroesophageal reflux. Spirometry was done next, followed by a methacholine bronchoprovocation challenge using a modification of the method of Hargreave and colleagues [10]. Bronchial hyper-responsiveness was defined as a 20% decrease in the FEV₁ (PC ₂₀) at a methacholine concentration of 8 mg/mL or more [11]. Because of its low yield in previous studies [7-9], a chest roentgenogram was not included in the initial evaluation unless there was clinical suspicion of infection or neoplasm (such as fever, weight loss, or hemoptysis).

Weekly Follow-up

Patients were contacted by phone each week and seen in person whenever clinically necessary. We created a standardized scale for the patient to use in rating weekly cough severity and a second scale for rating side effects of therapy. The cough severity scale ranged from 0 (cough gone) to 7 (cough markedly worse). The descriptors for the scale are shown in Figure 1. Side effects were rated as none, mild, moderate, or severe (intolerable).

End Point

The end point for treatment was cough resolution, defined as a patient report of complete absence of cough for 2 consecutive weeks or of diminution to the point that the patient considered the cough insignificant.

Step 1

An antihistamine-decongestant preparation containing 1 mg of azatadine maleate plus 120 mg of sustained-release pseudoephedrine sulfate (Trinalin, Key Pharmaceuticals Inc.; Kenilworth, New Jersey) was prescribed to be taken twice daily as empiric therapy for possible postnasal drip syndrome caused by rhinitis (postviral, allergic, or vasomotor). If the cough did not improve at the end of 1 week of antihistamine-decongestant therapy, step 2 was begun immediately. If the cough improved after 1 week of therapy, antihistamine-decongestant alone was continued either until it resolved or until no further improvement occurred. At that point, patients with persistent symptoms of the postnasal drip syndrome (sensation of postnasal drip, throat clearing, nasal congestion, or a tickle in the back of the throat) were given nasal corticosteroids in addition to the antihistamine-decongestant.

Sinus imaging was obtained before nasal corticosteroids were prescribed if chronic sinusitis was suspected or subsequently if postnasal drip persisted despite the combination of antihistamine-decongestant and nasal corticosteroid therapy. Patients with sinus roentgenograms consistent with sinusitis (sinus opacification, air-fluid levels, or mucosal thickening) were treated with twice daily oxymetazoline hydrochloride nasal spray (Afrin, Schering-Plough HealthCare Products; Memphis, Tennessee) for 3 days, nasal corticosteroids twice daily, antihistamine-decongestant twice daily, and antibiotics for as long as 6 weeks. If cough still persisted, a computed tomographic study of the sinuses was obtained. If substantial abnormalities were present, the patient was referred for an otolaryngologic evaluation and possible sinus surgery.

Substitutions for Trinalin could be made at any time if severe side effects developed. For severe drowsiness, astemizole (Hismanal, Janssen Pharmaceutica, Piscataway, New Jersey), 10 mg once daily, plus pseudoephedrine, 60 mg twice daily, were substituted. For severe insomnia, jitteriness, or urinary obstruction, astemizole once daily plus nasal corticosteroids twice daily were substituted.

Step 2

Patients who were still coughing after step 1 were next evaluated for asthma. Because the treatment was the same, no attempt was made to distinguish between asthma and postviral bronchial hyper-responsiveness [12]. Patients who had bronchial hyper-responsiveness on bronchoprovocation challenge were treated for 1 week with an inhaled ₂-agonist, albuterol, two puffs four times a day via metered-dose inhaler. If cough persisted, prednisone therapy (1 mg/kg body weight per day [maximum, 60 mg/d]) was added to the albuterol for a week. An oral ₂-agonist or theophylline was substituted for the inhaled albuterol if a patient reported marked cough in response to inhaled albuterol.

Step 3

Patients who continued to have persistent cough next had chest and sinus roentgenograms if they had not already been obtained. Any abnormalities considered to be clinically significant were evaluated and treated. If no significant abnormalities were found or the patient continued to cough despite appropriate therapy, the patient was advanced to step 4.

Step 4

Patients were next evaluated for gastroesophageal reflux. Patients with symptoms consistent with gastroesophageal reflux first had a 2-week trial of therapy with ranitidine, 150 mg twice daily, along with antireflux measures (no eating or drinking for at least 2 hours before going to bed or lying down, 20-cm elevation of the head of the bed using blocks, and avoidance of caffeine, alcohol, chocolate, and other foods known to exacerbate gastroesophageal reflux). Lack of a response to the 2-week trial of ranitidine led to a 24-hour esophageal pH-probe study. Patients without symptoms of gastroesophageal reflux had 24-hour esophageal pH-probe monitoring before therapy for gastroesophageal reflux was begun. pH-probe monitoring was performed and interpreted according to the criteria of Demeester and colleagues [13]. All patients whose pH-probe studies showed pathologic gastroesophageal reflux were treated for at least 8 weeks with omeprazole, 20 mg daily, in addition to antireflux measures. An upper gastrointestinal barium swallow or a gastroenterology consultation or both were obtained if the patient had any persistent gastrointestinal complaints.

Step 5

Patients who continued to cough next had bronchoscopy. If the bronchoscopy was nondiagnostic, the patient was treated (or retreated) with asthma therapy even if the bronchoprovocation challenge had been negative or if ₂-agonist and prednisone therapy was previously ineffective. If cough persisted, uncommon causes were considered (Table 1) [6-9]. Psychogenic cough was considered a diagnosis of exclusion.

Three-Month Follow-up

All patients were reevaluated using a standardized questionnaire 3 months after cough had resolved to determine whether cough had recurred and whether they were still taking cough-specific medications.

Final Diagnostic Criteria

No cause of cough was considered definitive until treatment for that cause had been effective (that is, was associated with marked improvement or resolution). When therapy for only one diagnosis resolved the cough, that diagnosis was considered to be the sole cause of cough. When therapy for a diagnosis effected marked improvement short of resolution, that diagnosis was considered to be one of the causes of cough and treatment was continued while additional diagnoses were investigated.

Statistical Analysis

Recurring data were analyzed by paired t-tests. The extent of association between two variables in a cross-tabulation was analyzed using chi-square analysis. Group comparisons were performed using the Fisher two-tailed exact test. Multiple regression analysis was done to determine the relationships between various baseline factors and time to resolution of cough and to determine which, if any, factors were of value in predicting asthma to be a cause of cough. A P value <0.05 was considered significant for all statistical analyses.

Results

Between 6 February 1990 and 26 November 1991 we evaluated 61 patients with a primary complaint of chronic cough. Sixteen failed to meet criteria for the study and were excluded: Three patients had known cancer; three were current smokers; one patient was positive for human immunodeficiency virus and had active pulmonary infection with Mycobacterium avium-intracellulare complex; four patients could not undergo methacholine bronchoprovocation challenge (three had an initial FEV₁ <70% of predicted value and responded to treatment for asthma with resolution of cough, and the fourth was pregnant; her cough responded to nasal corticosteroids for the postnasal drip syndrome due to rhinitis). Finally, five patients were unwilling to comply with the proposed work-up and treatment.

The study group consisted of 45 patients (17 men, 28 women). The mean age was 47 years (range, 18 to 75 years). The mean duration of cough was 140 weeks (range, 3 to 2080 weeks, SD 336 weeks). Thirty-four patients had already consulted a physician (mean, 2.1 physicians) for cough; 24 came to us by referral, and 10 sought us because their initial physicians had been unable to cure their cough. We were the first physicians consulted by 11 patients.

The initial mean severity of cough using a rating scale from 1 to 4 was 3.0 1.0 (1 = mild, 2 = moderate, 3 = severe, 4 = very severe). Before consulting us, patients had taken a mean of 4.4 medications. Twenty-two of 45 (49%) patients had been treated with antibiotics, 18 of 45 (40%) with a codeine-based cough suppressant, and 10 of 45 (22%) with oral corticosteroids.

Efficacy of Initial Antihistamine-Decongestant Therapy

After 1 week of antihistamine-decongestant therapy, cough resolved in 6 patients (13%), improved markedly in 19 (42%), improved moderately in 6 (13%), improved mildly in 7 (16%), was unchanged in 4 (9%), was mildly worse in 2 (4%), and was moderately worse in 1 (2%) (Figure 1). Ultimately, 16 patients (36%) required only antihistamine-decongestant to resolve the cough. The mean time to resolution for these 16 patients was 2.3 weeks (SD 1.3 weeks).

After 1 week of therapy, 28 patients (62%) had no side effects; 10 (22%), mild side effects; 10 (22%), moderate side effects; and 2 (4%), side effects that they rated as intolerable (insomnia in one case and drowsiness in the other). For the duration of treatment with antihistamine-decongestant, 18 of 45 (40%) patients had no side effects; 10 (22%), mild; 10 (22%), moderate; and 7 (16%), severe (requiring a change in therapy). Severe side effects included drowsiness in 1 patient (week 1), insomnia in 3 (weeks 1, 5, and 5), urinary obstruction symptoms in 2 (weeks 2 and 4), and jitteriness in 1 (week 6). The therapy for the patient with drowsiness was changed to astemizole and pseudoephedrine and nasal corticosteroids, whereas patients with insomnia or urinary retention were treated with astemizole and nasal corticosteroids.

Overall Efficacy

Twenty-five of 45 (55%) patients reported at least a marked improvement in cough by week 1 (Figure 2). By week 8, 41 of 45 (90%) patients and by week 20, 45 of 45 (100%) patients (including those with resolution) had reported marked improvement. The mean time to marked improvement for all 45 patients was 3.1 weeks (SD 3.8 weeks).

For 23 of 45 (51%) patients, cough resolved by week 5. Forty-one of 45 (90%) patients had resolution by week 16. Ultimately 43 of 45 (96%) patients had resolution of cough, (mean time to resolution, 7.1 weeks [SD 3.8 weeks]).

We used multiple linear regression analysis with binary independent variables to determine whether cough duration, cough severity, or a history of preceding upper respiratory tract infection was associated with time to resolution. For this analysis, a cough duration of less than 8 weeks before study entry was considered shorter and 8 weeks or more was considered longer. Initial cough severities of 1 (mild) and 2 (moderate) were combined and designated less severe and severities of 3 (severe) and 4 (very severe) were combined and designated as more severe. There was no significant relationship between time to resolution of cough and duration of cough, severity of cough, or a history of preceding upper respiratory infection (all P values > 0.2).

Testing

We did a total of 83 diagnostic tests on our 45 patients, for a mean of 1.8 tests per patient. These included 45 methacholine bronchoprovocation challenge studies, 15 sinus imagining studies, 9 chest roentgenograms, 8 pH-probe studies (2 in 1 patient), 3 bronchoscopies, and 2 barium swallows.

Tests obtained by other physicians before we saw the patients included 9 chest and 3 sinus roentgenograms, 3 barium swallows, 1 upper gastrointestinal endoscopy, and 2 bronchoscopies. Using our algorithm, we would have ordered only four of these tests (two chest and two sinus roentgenograms) had we seen the patients de novo.

Diagnoses

We determined diagnoses in all patients. The diagnosis was considered definitive in 43 patients based on cough resolution in response to specific therapy [7, 8]. The diagnosis was considered probable in two patients because of continued complaint of cough. One patient with postnasal drip had severe chronic pansinusitis refractory even to endoscopic sinus surgery. The patient with psychogenic cough had an extensive, nondiagnostic work up, failed multiple therapeutic trials, and had never been heard to cough. (She nevertheless stated that she had marked improvement in cough by the end of the study.)

Thirty-nine of 45 (87%) patients had the postnasal drip syndrome (36 postnasal drip syndrome secondary to rhinitis, 3 postnasal drip syndrome secondary to sinusitis), 13 of 45 (29%) had asthma, 5 of 45 (11%) had gastroesophageal reflux, 1 (2%) had Hodgkin disease, and 1 (2%) had psychogenic cough (Figure 3). (Because patients could have more than one cause of cough, the total given exceeds 100%.) The patient with Hodgkin disease did not have a known malignancy at study entry. After initial evaluation, he had a diagnostic biopsy of an anterior mediastinal mass. His cough resolved with treatment of the Hodgkin disease [14].

Thirty-two patients (71%) had a single diagnosis (Figure 3). Twenty-seven had the postnasal drip syndrome, 3 had asthma, 1 had Hodgkin disease, and 1 had psychogenic cough. Eleven patients (24%) had two diagnoses (8, postnasal drip syndrome plus asthma; 2, postnasal drip syndrome plus gastroesophageal reflux; 1, asthma plus gastroesophageal reflux). Two patients (4.4%) had three diagnoses (the postnasal drip syndrome plus asthma plus gastroesophageal reflux).

The Postnasal Drip Syndrome

Thirty-nine of 45 patients were diagnosed as having cough caused, at least in part, by the postnasal drip syndrome. In 38 of these 39 patients, cough resolved with specific therapy. In 27, the postnasal drip syndrome was the sole cause of cough (caused by rhinitis in 25 and sinusitis in 2). Sixteen of 27 (59%) responded to antihistamine-decongestant therapy alone whereas 9 of 27 (33%) responded to antihistamine-decongestant plus nasal corticosteroids (Figure 4). One of the two patients with the postnasal drip syndrome caused by chronic sinusitis failed to respond to antihistamine-decongestant plus nasal corticosteroids and antibiotics but did respond to endoscopic sinus surgery. The second continued to have refractory postnasal drip and cough despite treatment including sinus surgery.

Twelve of 39 patients had the postnasal drip syndrome (caused by rhinitis in 11 and sinusitis in 1) in combination with another diagnosis or diagnoses. For treatment of the postnasal drip component, 3 responded to antihistamine-decongestant alone, 8 responded to antihistamine-decongestant plus nasal corticosteroids, and 1 responded to antihistamine-decongestant plus nasal corticosteroids and antibiotics (Figure 4).

Of the 39 patients with a final diagnosis of the postnasal drip syndrome, 8 (20%) had denied any throat clearing or sensation of postnasal drip. In 23 (59%), the initial physical examination had been negative for both nasal secretions and cobblestoning of the posterior pharynx. Six patients whose cough resolved in response to antihistamine-decongestant had neither a history nor physical examination findings usually associated with postnasal drip [8].

Asthma

Fourteen of 45 (29%) patients were diagnosed as having cough caused by asthma. Three had asthma alone, 11 had asthma in combination with other diagnoses (8, asthma plus the postnasal drip syndrome; 1, asthma plus gastroesophageal reflux; and 2, asthma plus the postnasal drip syndrome plus gastroesophageal reflux). Cough resolved with inhaled albuterol alone in only 4 of the 14 patients; the remaining 10 responded to prednisone (Figure 4) and were subsequently maintained on inhaled corticosteroids plus bronchodilator therapy.

Nineteen of 45 (42%) patients had had bronchial hyper-responsiveness on bronchoprovocation challenge. Thirteen of 19 (68%) patients with bronchial hyper-responsiveness required asthma therapy for cough resolution. (One patient who required asthma therapy for cough resolution had a technically inadequate bronchoprovocation challenge, which was eliminated from further analysis of bronchoprovocation.)

Multiple logistic regression analysis was done to identify covariates that predicted the need for asthma therapy to achieve cough resolution. A history of chronic dyspnea, episodic dyspnea, wheezing, a previous diagnosis of asthma, and cough starting with a respiratory tract infection were among the hypothesized predictors. Only a history of dyspnea, however, was significantly associated with the need for asthma therapy (P = 0.02 for chronic and P = 0.04 for episodic dyspnea). The odds ratios were 13.5 (CI, 1.4 to 128.3) and 6.5 (CI, 1.1 to 40.4), respectively, for chronic and episodic dyspnea. A history of wheezing, of cough starting with a respiratory tract infection, and a previous diagnosis of asthma were not associated with the need for asthma therapy. For a diagnosis of asthma-induced cough, the positive predictive value of methacholine bronchoprovocation challenge was 74%. The negative predictive value was 100%.

Gastroesophageal Reflux

Five of 45 (11%) patients were diagnosed as having cough caused by gastroesophageal reflux, always in combination with other diagnoses. Only 2 of 13 patients who initially had symptoms consistent with gastroesophageal reflux had a final diagnosis of reflux-induced cough. Both responded to H₂-blocker therapy (Figure 4). Three of 32 patients without symptoms of gastroesophageal reflux were found to have silent reflux on 24-hour pH-probe monitoring. In all three, cough responded to antireflux therapy.

Three-Month Follow-up

All 43 patients whose cough resolved were contacted 3 months after resolution. Thirty-five were cough-free, whereas cough had recurred in 8. Three of 8 had the postnasal drip syndrome plus asthma, 2 had the postnasal drip syndrome plus gastroesophageal reflux, 1 had asthma alone, and 2 had the postnasal drip syndrome alone. Seven of 8 had experienced cough recurrence while on therapy (1 patient with the postnasal drip syndrome was off therapy). Only one patient had cough recur with its original severity. At the time of the 3-month follow-up, only 5 of 26 patients with a sole diagnosis of the postnasal drip syndrome were still on therapy compared with 13 of 17 patients with other diagnoses (P < 0.001). In addition, patients with a diagnosis of the postnasal drip syndrome alone were significantly less likely than patients with other diagnoses to have a cough recurrence at 3 months (P = 0.04).

Discussion

Our algorithm emphasizing the preliminary use of an antihistamine-decongestant as part of a stepwise approach to the patient with chronic cough proved effective. The reasons for the up-front use of an antihistamine-decongestant were that the postnasal drip syndrome is reported to be the most common cause of chronic cough [2, 7-9], that antihistamine-decongestant is an effective treatment for the postnasal drip syndrome cause by rhinitis [7], and that there is no way to predict who will respond to treatment for the postnasal drip syndrome. Thirty-eight of 45 patients reported some improvement in cough at 1 week. Moreover, 36% of our patients (16 of 45) required only antihistamine/decongestant therapy to achieve cough resolution (mean time to resolution, 2.2 weeks [SD 1.3 weeks]). The prompt response to antihistamine-decongestant in this group of patients eliminated the need for any further diagnostic testing.

Largely because of the efficacy of initial antihistamine-decongestant therapy, a mean of only 1.8 tests per patient was required. Cough resolution was achieved in a mean of 7.1 weeks. The number of tests required and the time to resolution both represent improvements over published reports [8] without loss in diagnostic or therapeutic success rates [7, 8]. In concordance with previous studies [7, 8], cough resolution was achieved in 96% of patients and a diagnosis in 100%.

We do not believe that the more rapid resolution of cough and need for fewer diagnostic tests in our study were caused by less severe cough or more obvious causes of cough in our patients than in those in other studies [7-9]. Our patients' coughs were severe and interfered seriously with their lives. The mean duration of cough was nearly 3 years. Three quarters of the patients had seen at least one other physician. Patients had been treated with a mean of 4.4 medications including antibiotics, codeine-based antitussives, and corticosteroids. Further, our data were not skewed by a subset of patients with less severe, less prolonged cough that resolved especially rapidly and therefore lowered mean time to resolution.

Like other investigators [7-9], we found that the postnasal drip syndrome was the single most common cause of chronic cough, but our finding of a prevalence of 87% was higher than previously reported. This may reflect the fact that all of our patients were treated for the postnasal drip syndrome irrespective of clinical findings. The importance of this approach was borne out by the fact that eight of our patients who responded to treatment for the postnasal drip syndrome had denied any awareness of throat clearing or sensation of drip, suggesting that a subset of patients exists whose sole manifestation of the postnasal drip syndrome is cough (the silent postnasal drip syndrome). This finding is analogous to cough-variant asthma [15, 16] and to cough caused by silent gastroesophageal reflux [17].

It is important to acknowledge that determination of the cause of cough is necessarily inferential. As in all previous studies on chronic cough [7-9, 15, 17], we used response to treatment as part of the final diagnostic criteria. Although the diagnostic criteria for sinusitis (abnormal roentgenograms), asthma (documented bronchial hyper-responsiveness) and gastroesophageal reflux (pathologic reflux by 24-hour pH-probe criteria) at least include objective findings, no objective test and certainly no gold or criterion standard for the diagnosis of the postnasal drip syndrome exists. Until such an objective test for the postnasal drip syndrome can be developed, cough response to specific treatment provides the best test for clinical purposes. Further complicating the issue is the fact that the presence of a condition capable of causing cough (such as bronchial hyper-responsiveness) is not proof in any individual patient that the condition is actually causing cough. Therefore, although the diagnostic labels used in this and other studies [7-9] help to organize the clinical approach to chronic cough, they should not be over-emphasized. This study has shown that in most patients with chronic cough, a stepwise approach starting with antihistamine-decongestant therapy resolves cough.

Like Irwin and colleagues [7, 8], we found asthma to be the second most common cause of chronic cough. Our data reconfirm the value of methacholine bronchoprovocation challenge; no patient without bronchial hyper-responsiveness required asthma therapy for cough resolution (negative predictive value, 100%). In contrast, the presence of bronchial hyper-responsiveness was only moderately specific; only 13 of 19 patients with bronchial hyper-responsiveness required asthma therapy for resolution of cough (positive predictive value, 74%). This finding does not necessarily mean that the other six patients with bronchial hyper-responsiveness did not have asthma. Postnasal drip may trigger cough in some asthmatic patients; when postnasal drip is treated, cough may resolve even if underlying asthma is present.

Although we believe that pure cough-variant asthma exists [16, 18, 19], we had no cases. All 14 patients who responded to asthma therapy reported chronic or episodic dyspnea, with 11 complaining of both. Our data suggest that cough-predominant asthma is more common than pure cough-variant asthma.

Given the fact that many patients with chronic cough will respond to therapy for the postnasal drip syndrome alone, we conclude that bronchoprovocation challenge need not be done routinely in all patients with chronic cough. In our study group, had we delayed bronchoprovocation challenge and tested only those patients whose cough had failed to resolve in response to therapy for the postnasal drip syndrome, less than 50% of the patients would have had bronchoprovocation challenge. Therefore, we believe that bronchoprovocation challenge is not necessary as part of the initial evaluation of patients with chronic cough unless other symptoms suggestive of asthma are present.

Based on the low yield previously reported [7, 8], we delayed chest roentgenography until step 4 of our algorithm. We obtained only nine chest roentgenograms and found only one to be useful. This finding suggests that chest roentgenography is not automatically indicated in the initial evaluation of all patients with chronic cough unless other findings are present (for example, smoking history, immunocompromise, hemoptysis, fever, weight loss, or dyspnea).

Similar to the findings of recent reports [7, 8, 17], we found that gastroesophageal reflux was the third most common cause of chronic cough. For those patients whose cough persisted after the postnasal drip syndrome and asthma had been evaluated, the likelihood of gastroesophageal reflux as a cause of cough was high regardless of whether reflux symptoms were present. Five of seven patients not diagnosed after work-up for the postnasal drip syndrome and asthma had reflux-induced cough. Three of 5 had no symptoms of heartburn. In all 3 of these patients, 24-hour pH-probe monitoring showed silent pathologic gastroesophageal reflux.

In our study, 18% of patients had a recurrence of cough in the relatively short follow-up period of 3 months. We do not find this surprising: It highlights the fact that in many patients, cough is the manifestation of a chronic or reoccurring process. Many patients are therefore likely to require chronic or episodic therapy. This appears to be particularly true for patients with asthma or gastroesophageal reflux. In our study, six of eight of the patients with recurrent cough had asthma, gastroesophageal reflux, or both.

Conclusions

Our algorithm for chronic cough, which emphasizes a stepwise approach and weekly follow-up, is effective. It begins with empiric antihistamine-decongestant therapy for possible postnasal drip syndrome and then progresses from simpler and less invasive tests (sinus and chest imaging and methacholine bronchoprovocation challenge) to more specialized and invasive tests (pH-probe monitoring and bronchoscopy). The initial steps can be performed by primary care physicians (especially if bronchoprovocation challenge is available in the community) [20, 21]. It should be possible to diagnose and treat most patients without referral. Using our approach, resolution of cough can be expected in more than 95% of patients. The use of nonspecific antitussives is avoided and diagnostic tests are kept to a minimum.