Statistics from Altmetric.com
McGarvey et al 1 suggest that full investigation of patients with persistent cough improves treatment.
I reviewed 100 such patients seen consecutively. All had normal chest radiographs, two were current smokers, and their mean cough duration was 18.8 months (range one month to 20 years). Initial treatment was given on the basis of history and routine clinical examination with investigations reserved for patients not responding after one month. Thirty four patients failed to return after their initial appointment. Twenty were contacted and all reported complete resolution of their symptoms. Clinical diagnoses in the 14 others were similar and they probably defaulted because of improvement, but have been excluded from analysis. Investigations performed included radiology of the sinuses in 8%, bronchial provocation testing in 16%, and investigation for gastro-oesophageal reflux in 19%. The final diagnoses (table 1) were based on successful response to treatment. Asthma was uncommon (7%) but, as there were few treatment failures, it seems unlikely that asthma was missed. The awareness of asthma by GPs is high in Australia and most had probably been treated by their GPs. Clinical outcomes were excellent with 79 patients (92%) reporting complete or almost complete resolution of cough in a mean of two months.
These results suggest that good outcomes can be achieved in most patients without routine investigation. The poor positive predictive values of symptoms quoted by McGarvey et alreflect poor choice of historical features. These authors confirm that any cause of chronic cough increases the sensitivity of the cough reflex, and the finding that cough precipitated by non-specific stimuli is poorly predictive of asthma is unsurprising. Likewise, most patients with reflux associated cough do not have heartburn.2
Diagnostic protocols advocated by hospital based researchers may be inappropriate for other settings. Such protocols should be subjected to randomised control trial against less interventionist approaches as would be required of a new drug treatment.
authors’ reply We welcome Dr Simpson’s interesting comments. He describes a group of patients which appears to be rather different from the patients reported in our study.1-1 Firstly, our patient group had been troubled with cough for a longer period of time (mean cough duration 67 months (range 2–240) compared with 18.8 months (range 1–240)). Secondly, application to his study of our exclusion criteria—that is, smokers, an abnormal chest radiograph, any preceding viral infections, and patients taking angiotensin converting enzyme inhibitors—would mean that 29 of the 86 patients (33%) he reviewed would not have been included in our study. Dr Simpson relies heavily on patient history in the evaluation of his patients. In our discussion we highlight the limitations of historical features, given the existence of both silent “reflux” and postnasal drip. We do not accept that the poor positive predictive values reflect a bad choice of historical features and believe there are no accurate discriminatory historical features that can be reliably applied to cough patients in general. This is supported by a study which specifically examined features in the clinical history and found that these were unlikely to be useful in diagnosing the cause of cough.1-2
While we agree that a randomised controlled trial may be one way to address the issue of how best to evaluate patients with cough, we suspect that Dr Simpson is describing a very different patient population from those referred to our cough clinic and that a less interventionist approach may not therefore be appropriate. In the meantime we feel a comprehensive protocol which is consistent with the approach of the recent Consensus Panel Report of the American College of Chest Physicians1-3 continues to represent the optimum way to evaluate patients referred with chronic cough.