Albumin excretion rate, albumin concentration, and albumin/creatinine ratio compared for screening diabetics for slight albuminuria.

Slight albuminuria, an overnight albumin excretion rate (AER) greater than 30 micrograms/min in an "Albustix"-negative sample, predicts development of diabetic nephropathy. This study compares the AERs for 261 timed overnight urine collections with the albumin concentrations and albumin/creatinine ratios for the same specimens (equivalent to first morning specimens). Thirty-one specimens (11.9%) had AERs greater than 30 micrograms/min. Use of an albumin/creatinine ratio greater than 3.0 mg/mmol to predict an AER greater than 30 micrograms/min gave a sensitivity of 96.8%, a specificity of 93.9%, and a predictive value of 68.2%, with a correlation coefficient of 0.921. Use of an albumin concentration greater than 17 mg/L gave a sensitivity of 96.8%, a specificity of 90.9%, a predictive value of 58.8%, and a slightly poorer correlation (r = 0.904). Evidently either method is acceptable as an initial screening procedure, but determination of albumin concentration alone would be preferable because of lesser cost.

Sampling errors in pH and blood gas analysis--an evaluation of three new arterial blood samplers.

We have tested the accuracy, acceptability and general performance of three recently-marketed samplers for arterial blood gas measurement (the Corning Arterial Blood Sampler, the Concord 'Pulsator' and the Sarstedt 'Monovette'). All three greatly reduce or eliminate the error of venous sampling, and the Corning and Sarstedt samplers eliminate the risk of dilution of the sample by excess heparin solution. A positive bias in pO2 measurement, more marked at higher levels, was demonstrated with the Concord and Sarstedt samplers, and the latter carry a slightly increased risk of cross-infection. None of the samplers completely overcame potential sampling errors.

Evaluation of Visidex II--a new strip for blood glucose measurement.

We have evaluated a new blood glucose test-strip (Visidex II, Ames), which does not require washing with water after contact with blood. Visidex II showed a significant negative bias when compared with a similar diagnostic strip (BM 20-800, Boehringer) and with the Beckman II glucose analyser (p less than 0.001 in both cases), and a poorer correlation with either method than BM strips with the Beckman II. A significant number (6.3%) of Visidex strips in this study were unreadable due to removal of the colour pads on blotting, and some difficulty was reported by patients in reading values around 6-8 mmol/l. Visidex II can safely be stored in the refrigerator and developed strips may be stable longer than BM strips. A significant error in reading was noted if the incubation time was varied by +/- 5 seconds. Further experience and perhaps some modifications may be necessary before Visidex II can be recommended for widespread use.

Are insulin dependent diabetics in the West of Scotland prone to nutritional deficiencies?

In a cross-sectional survey, the weight, vitamin and mineral status of 57 insulin dependent diabetics attending a clinic in Glasgow were studied. There were no significant differences in any of the measured parameters between poorly controlled and better controlled diabetics (as assessed by HbA1 status). No gross evidence of nutritional deficiency was found although serum magnesium levels tended to be low. This may be clinically relevant as hypomagnesaemia is associated with cardiac arrhythmias and ischaemic heart disease.

BM strips: how accurate are they in general wards?

In order to test the accuracy of BM Test Glycemie 20-800 strips when used in general wards, simultaneous BM strip results were obtained for 183 emergency out-of-hours plasma glucose requests. The results were considered accurate if the value obtained by the Emergency Biochemistry Laboratory (using a Beckman II glucose analyser) was within one strip-range on either side of the stated BM strip value. Fifty-nine (32.2%) of the BM strip results were considered inaccurate by this definition, and in 22 of these inaccurate results the differences were large enough to have led to inappropriate management if this was based on the BM strip result alone. The reasons for these inaccuracies are not clear, but it is suggested that improved training of ward staff and formal quality control schemes may help to improve results.

A fifteen-month double-blind cross-over study of the efficacy and antigenicity of human and pork insulins.

The antigenicity and efficacy of semi-synthetic human and pancreatic pork insulins have been compared in a double-blind double cross-over study in 96 insulin-treated diabetic patients. Transfer from pork to human insulin was associated with a 1.2 +/- 0.5 mmol/l deterioration (p less than 0.05) in the fasting blood glucose level, while the opposite change caused a 1.1 +/- 0.5 mmol/l improvement (p less than 0.05). After 4 months treatment, glycosylated haemoglobin levels were lower on pork (11.1 +/- 0.3%) than on human (11.7 +/- 0.3% p less than 0.01) insulin. The incidence of hypoglycaemia was similar with the two insulins. IgG insulin antibody levels were identical after human insulin treatment (5.7 +/- 0.4 micrograms/l) compared to pork insulin treatment (5.9 +/- 0.5 micrograms/l). Patients with high levels of antibodies (greater than 10 micrograms/l) showed a similar reduction in level when switched to either species of highly purified insulin. The deterioration in fasting blood glucose control is consistent with similar reports for biosynthetic human insulin and suggests, in the absence of changes in insulin antibody levels, a small but clinically significant pharmacokinetic difference between human and pork insulin.