Search tips
Search criteria 


Logo of gutGutView this articleSubmit a manuscriptReceive email alertsContact usBMJ
Gut. 2007 August; 56(8): 1166.
PMCID: PMC1955523

Authors' reply

We appreciate the interest shown in our work by Zein and McCullough. In their series of 241 patients with primary biliary cirrhosis (PBC), they found that the presence of self‐reported fatigue at clinical presentation was associated with an increased risk of death during follow‐up on univariate analysis, but that, in contrast with our earlier report (Gut 2006;55:536–41), this was not independent of other previously identified prognostic factors in PBC, such as age and bilirubin. There are a number of important differences between the Zein and McCullough study and our original study. We think that two of these are likely to be of importance in explaining the differences in the conclusions reached by these two studies.

The first important difference is that Zein and McCullough used self‐reporting of the presence of fatigue, as opposed to formal fatigue impact assessment, as the criterion for the presence of fatigue. This reflects the fact that their cohort dates from the early 1990s, an era that pre‐dated the validation and application of formal fatigue impact scoring systems in PBC. In our experience, self‐reporting of fatigue is very unreliable in predicting the presence of actual fatigue‐associated lifestyle change. In the Newcastle population, self‐reporting of the presence of fatigue tends to overestimate the true prevalence and impact of fatigue (>60% of our patients self‐report the presence of fatigue, but significant increase in fatigue scores is seen in only ca 35–40% of patients). If also true of the Cleveland population included in the Zein and McCullough study, the implication is that their population actually has quite a low degree of fatigue, a potentially interesting observation in its own right.

The second important difference between the two studies is that the death rate in the Zein and McCullough patient cohort is significantly lower than that in our cohort (a crude death rate of 1% of patients per year of follow‐up compared with 5.1% per year of follow‐up). The most probable explanation for this difference would seem to be the ages of the cohorts at the outset of the follow‐up period described in each of the studies (65 years in the Newcastle cohort compared with 53 years in the Cleveland cohort). This difference is likely to reflect the fact that the Cleveland cohort was an incident cohort of patients, whereas the Newcastle cohort was a geographically defined prevalent cohort (ie, the Cleveland cohort were followed up from the point of disease diagnosis, whereas the Newcastle patients were followed up from a fixed time point and were therefore typically several years into their disease time course). Other potential differences could result from either different referral patterns in the two centres or differences in the natural history of the disease in different continents.

We found, however, the low death rate, and apparent absence of an independent association between fatigue and mortality in the younger Cleveland population to be of great interest and reanalysed our data to explore further the impact of age on fatigue‐associated mortality. Strikingly, when we split our patient cohort into those aged <65 and >65 years at the point of study, we found that those <65 years (mean age 53 years) had a death rate comparable to the Cleveland cohort (1.4% per year of subsequent follow‐up), with no significant association between mortality and fatigue at the study outset (fig 1A1A).). In marked contrast, the patients aged >65 years at the point of study (mean age 74 years) had a significant increase in their mortality (9.3% per year of follow‐up). It was in this cohort that there was a significant association between fatigue at the study outset (defined as Fatigue Impact Scale >40) and subsequent mortality (fig 1B1B).). Furthermore, when looking at the individual ages of the patients who died during follow‐up, increase in the death rate was present at all ages from 65 years upwards, with age 65 years seeming to act as a natural cut‐off point for increased mortality (eg, mortality in patients aged 60–64 years at study was 1.3% per year of follow‐up compared with 5.5% per year for the 65–69‐year group, 5% per year for the 70–74‐year group and so on).

figure gt113209.f1
Figure 1 Survival analysis of patients with high (Fatigue Impact Scale (FIS) >40, solid line) and low (FIS<40, broken line) baseline fatigue scores who were (A) aged <65 years at the outset of our previously reported ...

In conclusion, we value the contribution of Zein and McCullough to this important discussion. We support their view that further long‐term studies are required and suggest that particular attention should be focused, in these studies, on the cohort aged 65–75 years where the effects of fatigue‐associated mortality seem to have a disproportionate impact.


Competing interests: None.

Articles from Gut are provided here courtesy of BMJ Publishing Group