Alteration of the faecal flora is attractive as a low risk means of preventing or, possibly, treating C difficile and associated infections.² Induced disequilibrium of the colonic flora can be achieved either through supplemental probiotic loading or by therapeutic manipulation. Faecal flora can be manipulated through the use of lactulose, a synthetic disaccharide that is predominantly used as an osmotic laxative. It is neither absorbed nor metabolised in the upper gastrointestinal tract but is degraded by the bacterial flora of the proximal colon to organic acids. These acidify the proximal colon and result in a dose dependent catharsis.

Lactulose has several additional properties, including an antiendotoxin effect and alteration of faecal floral patterns.³ This quantitative alteration in faecal floral patterns, with an increase in faecal Lactobacillus acidophilus and a reduction in both coliforms and bacteroides, has been confirmed in our own unpublished work. A qualitative alteration in bacterial pathogenicity may result with lactulose as an alternative substrate.

Additional historical evidence shows effective treatment of shigellosis and salmonellosis with lactulose.^4,5 Several risk and host factors have been recognised, and identification of people at high risk of antibiotic associated diarrhoea could facilitate targeted pretreatment with lactulose or conjoint treatment with lactulose and antibiotics. An effective means of gaining rapid benefit might be through combining lactulose with a probiotic such as L acidophilus.

Only four of the studies analysed included elderly patients (Gotz, Surawicz, McFarland, and Lewis), and in only four studies were any analyses for C difficile done (Orrhage (normal volunteers, no cases seen), Surawicz, McFarland, Lewis). Either on their own or combining the results of these studies shows no benefit from probiotics (odds ratio 0.71, 95% confidence interval 0.29 to 1.76) (figure).

There are several rigorous systematic reviews on the subject of probiotics and diarrhoea, but the temptation to perform a meta-analysis has been resisted because of the lack of comparability of different probiotic organisms. D'Souza et al recognise this limitation when postulating the different mechanisms of action for the yeast and bacterial probiotics.

The author's principal inclusion criterion was the occurrence of diarrhoea, defined as two or more loose motions per day for at least two days. Of the studies analysed only three met this criterion (Surawicz, McFarland, and Vanderhoof).

There was an error in table 1, instead of no cases of diarrhoea (Gotz et al) three cases (8.3%) occurred in the active group. In the study by Orrhage et al only half the volunteers who received active treatment were included, those not included had a higher incidence of diarrhoea. It is not clear why the prospective double blind placebo controlled study by Borgia was excluded^1-2; yet a trial by Orrhage was included that was principally a bacteriological study in healthy adults.

Furthermore, a trial by Colombel et al was excluded for being single blinded, although it was a double blind placebo controlled trial.^1-3 Five studies were not analysed on an intention to treat basis (Surawicz, Tankanow, Vanderhoof, Gotz, Adam). The study by Adam et al had the biggest impact on the analysis yet is the most suspect of all the trials, with only 19.4% of recruited cases being analysed, the remainder being excluded because of protocol violations.

Furthermore, the use of lactulose is limited by lack of knowledge on the optimum dose at which it can have an effect on gut flora while not causing diarrhoea.

We agree that Clostridium difficile accounts for only 10-25% of cases of antibiotic associated diarrhoea. It accounts, however, for the most cases of colitis that occur and remains treatable once diagnosed.^2-2 Our recommendations on the use of probiotics were not specific to elderly people—we did point out that a trial was needed in this age group.

The supposed error in table 1 relates to the percentage free of diarrhoea in the active and placebo groups in the study by Gotz et al.^2-3 Our analysis was performed by using the numbers as mentioned (8.3% in the active group v 21% in the placebo group); the table highlighted the numbers in the groups when other treatments likely to induce diarrhoea were excluded.

The study by Borgia et al was excluded because it primarily studied effects on body weight.^2-4 The study by Colombel et al did not provide enough information on the number of cases of diarrhoea alone (diarrhoea was grouped as a side effect along with nausea, vomiting and abdominal pain).^2-5 We are criticised for these exclusions and yet for including studies analysed on a “per protocol” basis, an acceptable scientifically rigorous procedure.

We acknowledge the risks posed by probiotics, but most complications occurred in immunosuppressed individuals and people with other life threatening illnesses. This was mentioned in our paper, and we would leave it to clinicians to decide the individual suitability of patients to receive probiotics.