Your survey ('Nervous tick', pages 13-14, news focus, 7 September) asked trust chief executives to speculate about the Commission for Health Improvement's clinical governance review process before they have been reviewed.

I would contend the survey was statistically unreliable as the response rate was 17 per cent. While it claims some organisations are 'fearful' of the CHI process, this is not based on the experiences of the four pilot trusts which have been reviewed. In fact, David Moss, chief executive of Southampton University Hospitals trust, one of CHI's four pilot review sites, is the only source quoted who has experienced a review and he says the 'whole exercise has been very developmental'.

In clinical governance reviews, CHI aims to hold up a mirror to the NHS to allow it to see how it appears to others, especially patients. A key aspect is the pre-assessment phase: a CHI team will meet staff from the NHS organisation, outline the process and explain the data required, giving time for preparation.

The reviews are based on evidence, not opinion, and look at the effectiveness and quality of the organisation's procedures and working practices, identifying areas for improvement and best practice, which will be shared within the NHS.

The people carrying out the CHI reviews are drawn from the NHS and the public. NHS staff are closely involved with CHI, helping improve performance and quality of care.

CHI is committed to working with the NHS to help it continuously improve. We will do this by working with people and organisations, not by intimidating them.

Dr Peter Homa Director Commission for Health Improvement

Editor's note: Many opinion polls use samples far smaller than ours from 'universes' much larger than the one we surveyed. We sent the survey to trust chief executives and quality leads throughout England, Scotland and Wales, and the response rate - 130 out of 761 (17 per cent) - is robust in research terms. The confidence levels for our survey are as follows:

a figure of 50 per cent is accurate to + or - 8. 6 per cent;

a figure of 20 per cent or 80 per cent is accurate to + or - 6. 8 per cent;

a figure of 10 per cent or 90 per cent is accurate to + or - 5. 2 per cent.

These figures show confidence levels at a 95 per cent significance level: if we repeated the survey, 19 times out of 20 the results would again show within these margins, though readers may wish to bear in mind that those responding chose to participate and so may not reflect a completely random sample.