Recommendations from external referee (Dr D.E. Grobbee)
2. We cannot really include a mixture of ranges, SDs and 95% confidence limits. So because both the GP editorial committee and the referee are unanimous about 95% confidence intervals we have settled on these for all parameters in both table 1 and 2.
3. We agree with the comments about the Rose Questionnaire and have added some text to cover this point. The questionnaire is, however, still the most widely used method of assessing the likelihood that populations may vary in the frequency of peripheral arterial disease and is, of course, standardised permitting some degree of comparison between studies.
4. The referee is quite correct and we now use the term newly diagnosed cases in place of new cases throughout.
Firstly to make the diagnosis in the absence of a relative with tendon xanthomata, the patient must have tendon xanthomata. There is an enormous literature to the effect that the only causes of tendon xanthomata other than FH are cerebrotendinous xanthomatosis or phytosterolaemia, both of which are so rare that only a handful of cases have ever been described. (Furthermore they are not associated with markedly raised cholesterol levels). Having found patients with tendon xanthomata and raised cholesterol (i.e. FH) one knows that each represents 1 in 500 of the population. Their chances of having a relative with cholesterol >7.5mmol/l from a cause other than FH is only 3 to 5 in 100. This is because a cholesterol of 7.5mmol/l is the 97th percentile for British women in the age range of the relatives in our study and the 95th for British men of similar age (Dong, W., Colhoun, H., Lampe, F. Blood analytes Chapter 11 in Health Survey for England 1994 Colhoun, H. and Prescott-Clarke, P. eds. London: HMSO 1996; 369-419) respectively. Thus potentially 3-5% of the new cases detected in relatives using our method of family screening could be non-FH. Looking at it from a population perspective, if one were to screen the whole population of similar age to the relatives in our study 3-5% would have cholesterol levels >7.5mmol/l whereas only 0.2% (1 in 500) of the population would have FH so a relative of a proband with FH whose own cholesterol is ³ 7.5mmol/l is 15-25 times (3¸ 0.2 and 5¸ 0.2) more likely to have FH than someone encountered randomly whose cholesterol is ³ 7.5mmol/l. This is now discussed on page 11. It should also be noted that on page 5 (last sentence 2nd paragraph) in none of our probands was the diagnosis of FH based on tendon xanthomata in a 2nd degree relative which would increase the likelihood of misdiagnosis.
2. We have deleted tables 1 and 2 and included the findings previously presented in these in the text. We have combined table 3 and 4 (now table 1) and labelled the earlier table 5 as table 2.
For consistency as previously discussed we give 95% confidence intervals for all parameters.
3. The title has been amended as you suggest to Outcome of nurse-led case-finding amongst relatives of patients with known familial hypercholesterolaemia.
4. The number of references has been reduced to 24 as you suggest. However, it proved impossible to reduce the text to 2,000 words without omitting some of the findings, especially in view of the additional material we have been asked to incorporate. We hope you will agree that the new manuscript despite its length ( 3,466 words) is not verbose and crave your indulgence about length. We should be prepared to try to shorten it further if this is a sticking point, but do think it will make it less easy for the reader to see how we arrived at our conclusions and also mean that some important practical points will have to be omitted. Also Dr Groblee, the referee, did comment that the length was appropriate.
ii) What this paper adds (page 20).
iii) A statement that we have no competing interests.