Numbers-Needed-to-Treat

Bandolier 11 introduced the concept of numbers-needed-to-treat (NNT) as one way to define effectiveness. Concepts like this are always better understood with a worked example, and so in this edition we are bringing NNTs together with another Bandolier theme, that of Helicobacter pylori eradication as a cure for peptic ulceration.

Helicobacter pylori eradication and peptic ulcer

There is a growing body of evidence concerning the way in which antibiotics can be used to eradicate H pylori in the stomach and duodenum, and the way in which that eradication affects the course of peptic ulcer disease. The early literature in the 1980s contained few randomised controlled trials (RCTs), but the evidence is now much stronger.

A systematic overview of trials looking at H pylori eradication between 1983 and 1992 has recently been published [1]. This well conducted overview used solid methodology and identified eight trials involving duodenal ulcer, of which five were considered high quality. For the purposes of the present review, trials from 1992 were sought by use of a MEDLINE search, together with an approach to a number of pharmaceutical companies known or thought to be conducting clinical research in this area with a request for published studies which involved peptic ulcer and H pylori eradication.

There are three main issues to be addressed:

whether antibiotic regimens are successful in eradicating H pylori infections
whether antibiotic regimens affect immediate ulcer healing
whether eradication has any effects on the subsequent course of the disease

Meta-analysis

Only randomised controlled trials published in peer-reviewed journals were accepted; abstracts and meeting proceedings, or preliminary findings were excluded. Most studies treated duodenal and gastric ulcers together, though some did not; no differentiation was made here. Treatments found involved triple therapy (combinations of metronidazole, tetracycline and bismuth, together with an H2-antagonist, usually ranitidine) and amoxycillin plus omeprazole; antibiotics were usually given for two weeks, and acid suppressing drugs for four weeks. Treatments were combined under these two headings. Active control treatments consisted of cimetidine, ranitidine or omeprazole given for 4-6 weeks to promote ulcer healing.

Single arm treatments were combined; for ulcer healing at about 6 weeks determined by endoscopy; for H pylori eradication determined from samples taken at endoscopy; and ulcers remaining cured at one year determined by endoscopy (Table 1: note that the references in this table can be obtained from the original report which is referenced at the end of the article, or from Bandolier ).

Combined data from single arms for different treatments is shown in Table 2 and the Figure. Numbers-needed-to-treat [2] were calculated from combined single arms (Table 3).

Results

Use of histamine receptor antagonists (cimetidine, ranitidine) or the proton pump inhibitor omeprazole as a single therapy resulted in high levels of ulcer healing of about 78% at 10 weeks after therapy commenced, but eradication of H pylori was low, at 5% or less.

Studies of triple therapy (in combination with ranitidine or cimetidine) covering up to 768 patients demonstrated eradication rates of about 96%, with ulcer healing in 93% of patients at 10 weeks. A recent study [3] examined the need for an acid suppressing medicine (omeprazole) in combination with triple therapy, and found comparable eradication rates and ulcer healing; they concluded that addition of omeprazole was not needed, and that one week of treatment was sufficient. Most other studies continued acid suppressing medicine for some weeks longer than eradication therapy.

Use of amoxycillin with omeprazole, the other major treatment regimen, produced eradication rates of 70-86%, with ulcer healing of 89-95%. The numbers of patients in these studies is smaller than those with triple therapy. Omeprazole was usually used for some weeks longer than amoxycillin.

Numbers-needed-to-treat

The NNT analysis (Table 3) shows the following:-

For ulcer healing at about 6 weeks as judged by endoscopy, treating five patients with eradication therapy compared with acid-suppression alone resulted in one extra ulcer healed. The NNT for ulcer healing was about 5.
For H pylori eradication as judged by endoscopy and testing, treating one patient with eradication therapy compared with acid suppression alone resulted in one cure of H pylori infection. The NNT for H pylori eradication was 1.1.
For ulcers remaining cured at one year as judged by endoscopy, treating two patients with eradication therapy compared with acid suppression alone resulted in one extra ulcer remaining cured. The NNT for ulcers cured at one year was 1.8.

Although NNTs are a useful way of presenting the evidence, numbers with low values being good and high values worse they are not easily understood by all. Using the form of 100/NNT carries the same message in a slightly different and perhaps more intuitive way. Restating the results in this form gives the following message:-

"Using antibiotics to cure 100 patients with ulcers caused by H pylori infections, compared to using a short course of conventional acid suppressing medicines, results in 90 patients cured of the H pylori infection, 20 extra ulcers healed at six weeks and 56 more patients whose ulcers remain cured at one year."

previous or next story in this issue