Introduction

Top Abstract Introduction Methods Results Discussion References

General practitioners and other health professionals are encouraged to recommend over the counter cough medicines as a first line treatment for acute cough,¹ but evidence regarding their effectiveness is inconclusive. The NHS direct healthcare guide also recommends simple cough medicines for dry cough.²

Acute cough is a common symptom and cough medicines are used by many patients. In the United Kingdom retail sales of over the counter cough medicines rose by 3.0% to £94m between 1998 and 1999. ^{3 4} We conducted this systematic review to determine whether these medicines are effective for acute cough due to upper respiratory tract infections in adults. This review is based on a Cochrane systematic review of over the counter treatments in adults and children.⁵

	Methods

Top Abstract Introduction Methods Results Discussion References

Searching
We searched the Cochrane Acute Respiratory Infections Group specialised register (database of studies of acute respiratory infections based on regular database searches, personal contributions from Cochrane review group members, and hand searching of journals), the Cochrane Controlled Trials Register (issue 2, 2000, which includes randomised controlled trials published in Medline and Embase up to 1998), Medline (January 1998 to December 1999), Embase (January 1998 to December 1999), the UK Department of Health National Research Register (December 2000), personal collections of references, and reference lists of all retrieved articles for original randomised controlled trials. We wrote to study authors, the Proprietary Association of Great Britain, and pharmaceutical companies for information on unpublished studies. Details of the search strategy are available on bmj.com.

Study selection and validity assessment
We selected studies for review if (a) the participants were adults (aged 16 years or older) with acute cough (less than three weeks' duration) due to upper respiratory tract infection (presumed to be viral in origin with no auscultatory chest signs or signs on chest radiography) in an ambulatory setting; (b) the interventions were over the counter cough preparations; (c) a reported outcome was cough (frequency or duration assessed with any assessment tool); and (d) studies were randomised controlled trials with a contemporaneous control group receiving placebo or no intervention. We excluded studies if participants had chronic cough (more than three weeks' duration or due to a chronic underlying disease such as asthma, tuberculosis, or bronchial malignancy); cough was artificially induced in healthy volunteers; or they used non-conventional (herbal or homoeopathic) or non-oral preparations. A study had to meet all our inclusion criteria to be included. Drugs were divided into six categories according to their mode of action (table).

	Results

Top Abstract Introduction Methods Results Discussion References

After evaluating 328 citations and abstracts from all sources, we included 15 trials involving 2166 participants.^9-23 The main characteristics of the included randomised controlled trials are available on bmj.com. The number of studies for each type of drug was small, ranging from one to five. Outcomes included frequency and severity of cough and were measured in many different waysfor example, self report, physician assessment, cough sound pressure levels, and tape recordings. Ten studies reported data on adverse effects.

The methodological quality of included studies in terms of randomisation, blinding, and reports of losses to follow up was variable and generally not high. The randomisation process was adequate in three trials. Only two studies reported blinding of outcome assessors. It was unclear for three trials whether participants or treatment providers were blinded. Loss to follow up was well documented in 12 studies, with differential loss to follow up in both treatment arms reported in four studies. One trial reported a power calculation, and only one study fulfilled all the quality criteria. Many trials were too small to detect clinically important differences.

Quantitative data synthesis
We could not pool the results because there was clear clinical heterogeneity between trials in terms of participants, interventions, and outcome measurements. Furthermore, the number of trials in each category was small and the amount of quantitative data available limited.

Antitussives
Five trials tested antitussives versus placebo. Two studies tested codeine and found it no more effective than placebo. One of two studies of dextromethorphan favoured active treatment over placebo (differences in mean changes of cough counts 19% to 36% in three substudies, P<0.05), whereas the other found no significant effect. Moguisteine (one trial) led to mean differences in cough scores of about 0.5 in groups with severe cough on days 2 and 3 (P<0.05), but there were no differences between groups at final follow up. Only two trials reported adverse effects. ^{10 13} Nausea, vomiting, and abdominal pain were more common in participants treated with moguisteine than placebo (22% v 8%),¹⁰ and in one trial participants did not report any adverse effects from dextromethorphan.¹³

Expectorants
Participants in one study found guaifenesin more helpful than placebo (75% v 31%, P<0.01).¹⁴ However, a second trial found no significant differences between the groups.¹⁵ Guaifenesin led to a low incidence of nausea and urticaria in the active treatment group in one trial¹⁴; the other did not report on adverse effects.¹⁵

Mucolytics
In the only study of mucolytics, frequent cough was less prevalent in the Bisolvon linctus group than the placebo group (8.6% v 15.2%, P<0.02).¹⁶ This study did not report on adverse effects.

Antihistamine-decongestant combinations
One of the two trials of antihistamine-decongestant combinations showed a lower mean severity cough score in the active treatment group on days 3-5 (1.4 in active group v 2.0 in placebo group, P<0.05).¹⁷ The other trial found no significant differences between the two treatments.¹⁸ Antihistamine-decongestant combinations seemed to have a slightly higher incidence of adverse effects than placebo. These included dry mouth, dizziness, headache, and insomnia.

Other drug combinations
We included three studies of medicines containing fixed drug combinations.^19-21 These studies were very heterogeneous and used different drug preparations, limiting their comparability. In a study of EM-Vier, more participants in the treatment group improved within the first three days than in the placebo group (26/58 v 15/55, P=0.05).¹⁹ In a crossover trial of Vicks Medinite syrup, 58% of participants rated active treatment good or better in relieving cough symptoms compared with 32% for placebo.²⁰ Dextromethorphan plus salbutamol was better than placebo or dextromethorphan alone in relieving cough at night but there were no significant differences for cough symptoms during the day.²¹ Adverse effects for all preparations were rare and usually mild.

Antihistamines
Based on two trials, terfenadine was no more effective than placebo in relieving cough symptoms. ^{22 23} The incidence of adverse effects, which included excess fatigue and headache, was low with no significant differences between the groups.

	Discussion

Top Abstract Introduction Methods Results Discussion References

We found only a small number of randomised controlled trials investigating each category of cough medicine, so evidence on effectiveness is limited. In nine out of 15 trials, active treatment was no better than placebo. The positive results in the other six studies were of questionable clinical relevance. Most over the counter cough preparations were generally well tolerated and did not lead to serious adverse effects.

Study limitations and potential sources of bias
The included studies varied with respect to settings, populations, interventions (drugs, doses, and frequency), and outcome measures, which makes comparison difficult. Our results should therefore be interpreted with caution. Potential sources of bias such as randomisation procedure, blinding of outcome assessment, and losses to follow up were inadequately reported in several studies, suggesting poor methodological quality. The effect sizes of active treatment over placebo were often reported as differences between cough scores, which are difficult to interpret in a clinically meaningful way. Several studies were supported by the pharmaceutical industry, and others did not report their sources of funding or conflicts of interest.

We tried to obtain information on unpublished studies from study authors and pharmaceutical companies but obtained a limited response. If studies with negative results were less likely to be submitted for publication, this could have led to publication bias.

Implications
It remains unclear whether over the counter cough preparations are helpful in acute cough. We therefore cannot yet recommend these medicines as first line treatment for cough associated with upper respiratory tract infection. The advice given by NHS direct to use over the counter cough medicines should therefore be restricted until more evidence becomes available on their effectiveness. Future studies should use outcome measures that can be easily assessed in a primary care setting and that produce clinically meaningful results, such as patient satisfaction, disturbance at night, side effects, or time to return to normal daily activities.