Is it time to rethink the urea criterion in CURB-65?

To the Editors:

We read the article by Capelastegui et al. 1 on the validation of the severity assessment rules for the management of community-acquired pneumonia (CAP) with interest. The authors concluded that CURB-65 (confusion, urea >7 mmol·L⁻¹, respiratory rate ≥30·min⁻¹, low systolic (<90 mmHg) or diastolic (≤60 mmHg) blood pressure, and aged ≥65 yrs) and CRB-65 indices recommended by the British Thoracic Society are indeed useful and easy to apply in day-to-day clinical practice, thereby helping in risk stratification and management as demonstrated by Lim et al. 2.

With the current demographic trends, the proportion of the older population is set to rise. Globally, the projected number of people aged ≥65 yrs will double between 2000 and 2030 3. Moreover, CAP is common in older people and is associated with high mortality 4. We note that the mean ages of patients reported in Capelastegui et al. 1 and Lim et al. 2 were 64.1 and 61.8 yrs, respectively. In a prospective study of elderly patients aged ≥65 yrs with CAP (mean±sd age 81.1±7.9 yrs), we reported that the sensitivity and specificity of CURB in predicting death were 81 and 52%, respectively 5. The specificity figure of this elderly cohort was much lower than in other studies of younger patients 2, 6, 7. Due to this low specificity, the CURB-65 criteria in their current form are not ideal for assessing older CAP patients.

In the same study, we found that the urea criterion was only of borderline significance and cautioned that the urea cut-off point of >7 mmol·L⁻¹ in the CURB criteria may not be a sufficiently specific parameter 5. This may be due to the fact that chronic renal impairment is more prevalent in the older population, and dehydration, which is common in acutely unwell older patients, may also contribute to a higher urea level 8. With this background, we set out to assess whether raising the cut-off point of urea, while keeping the other criteria in CURB-65 as they are, would improve the specificity and positive predictive value (PPV) of the criteria in predicting mortality.

Analyses were conducted based on data from two prospective cohort studies 5, 7, which were carried out in a single institute in UK using similar inclusion and exclusion criteria. Although all ages were included in the first study 7, in the second study only those who were aged ≥65 yrs were included 5. Therefore, the compilation cohort consisted of a higher number of older patients with CAP and a higher mean age (mean±sd age 70.3±18.2 yrs) compared with previously reported studies.

A total of 193 patients were included in the currently reported analysis. It was hypothesised that increasing the urea cut-off level in older people would improve the specificity of CURB-65. This hypothesis was tested by increasing the urea cut-off point by 2 mmol·L⁻¹ each time from the currently recommended urea cut-off level of >7 mmol·L⁻¹ (i.e. CU₉RB-65 with urea cut-off >9 mmol·L⁻¹, CU₁₁RB-65 using urea cut-off >11 mmol·L⁻¹ and CU₁₃RB-65 using urea cut-off >13 mmol·L⁻¹), and the sensitivities, specificities, PPVs and negative predictive values (NPVs) for these indices in predicting 6-week mortality outcome were examined.

It was found that the specificity and PPV do increase with a rising cut-off level of urea, but there was a corresponding drop in sensitivity. A linear increase in PPVs was observed along with a slight decrease in NPVs. The resulting sensitivities, specificities, PPVs and NPVs are presented in table 1⇓ in comparison with corresponding values for predicting death by CU₇RB-65 (urea cut-off point >7 mmol·L⁻¹) in the same cohort.

A large prospective study to verify these findings is certainly desirable, particularly in view of the demographic trend towards an ageing population. Whilst we agree with the editorial comment by Ewig et al. 9 that, at present, CURB-65 (confusion, urea >7 mmol·L⁻¹, respiratory rate ≥30·min⁻¹, low blood pressure, and aged ≥65 yrs) and CRB-65 should be the tools of choice for severity assessment in community-acquired pneumonia, it is possible to improve the specificity of identifying true severe cases in older adults by raising the urea cut-off level to a more discriminative value.