Significance of low serum ferritin levels in the elderly.

Significance of low serum ferritin levels in the elderly Sir, Fortuitously, the letter by Joosten et al.' describing the receiver operating characteristic curve for serum ferritin coincided with the publication of the study showing the impact of ventilation/perfusion scanning on diagnostic precision in pulmonary embolism.2 In the PIOPED study the use of high probability lung scan as the decisive diagnostic parameter yielded a specificity of 97% corresponding to a sensitivity of 41%,3 the latter statistic perhaps providing comfort for the perception prevailing amongst 47% of Scottish consultants, that pulmonary scintigraphy was of no value in the diagnosis of pulmonary embolism.4 The sensitivity of 98% which accompanied the use of the low probability scan was accompanied by a fall in specificity to 10%,3 exemplifying the truism that all diagnostic cutoff levels are a trade-off between sensitivity and specificity. For common, and highly treatable, conditions such as iron deficiency and pulmonary embolism, most clinicians are prepared to maximize sensitivity at the expense of specificity, especially if supportive circumstantial evidence can be obtained from tests such as the haematological response to iron supplements in the case of suspected sideropaenia,5 and lower limb venography in suspected thrombo-embolic disease.6 Sensitivity becomes subser-vient to specificity, however, when maximum diagnostic accuracy is a precondition for testing a hypothesis, hence the use of a serum ferritin < 10 ng/ml as the diagnostic parameter for iron deficiency in the study comparing mean corpuscular volume and mean corpuscular Prolonged fever and rigors in disseminated tuberculosis Sir, We read with great interest the report by Harvey and colleagues on rigors in tuberculosis.' We would like to report a similar case whereby rigors along with fever were also prominent presenting features but, unlike those cases reported by Harvey, the response to anti-tuberculous treatment was much slower. A 30 year old Indian baker was admitted with fever, chills and sweating of 18 days duration. He had cervical lymphadenopathy and mild tender hepatomegaly without jaundice. His temperature was 39-40C, intermittent and associated with frequent rigors. Investigations showed raised liver enzymes, but normal bilirubin, albumin and prothrombin time. ESR was 93 mm in the first hour and tuberculin test of 20 mm induration. Full blood count, electrolytes, urea, urine examination and chest X-ray were normal. Blood cultures, blood films for malaria parasites and brucella antibodies were also negative. Ultrasound showed mild hepatomegaly with a normal gallbladder. Biopsies from lymph node and liver were identical and …