March 1988, Volume 38, Issue 3

Original Article


Mirza Naqi Zafar  ( Department of Pathology, The Aga Khan University Hospital, Karachi. )
Rukhsana Lalani  ( Department of Pathology, The Aga Than University Hospital, Karachi. )
Rubina Kazmi  ( Department of Medicine, The Aga Khan University Hospital, Karachi. )
Mohammed Khursheed  ( Department of Pathology, The Aga Than University Hospital, Karachi. )


The use of reagent strips to monitor blood glucose concentrations via glucose meters in diabetic patients is gaining popularity mainly due to their favourable performance as compared to the autoanalyser methods. As part of a programme of extra laboratory tests, such estimations were performed at a medical nursing station and results compared with simultaneous estimations in the clinical laboratory on autoanalyser. Observations in the first three months of the trial by untrained and unspecified operators were % total difference 16.0 ± 18.6. The % of results more than and less than autoanalyser were 13.2± 16.2 and 18.7 ± 21.4 respectively. During the second three months of the trial results were returned by one specified and trained operator. Results as compared to the autoanalyser were, % total difference 10.09 ± 12.4, more than 8.78 ± 10.9 and less than 14.0 ± 16.0. In this second trial the % total difference in glucose range 100—200 mg/dI was 8.25 ± 6.47, in range 200-300 mg/dl was 9.9 ± 15.0 and range 300-400 mg/dl was 13.9 ±16.3. The results of the second trial conducted by a single trained operator were better than the first trial because majority of the results were within or around 10% accepted deviation. Results show the use of reagent strips and Reflocheck at nursing stations are a viable proposition for monitoring diabetic patients, provided the tests are conducted by trained and specified operators. Our results also confirm observations that results obtained by strip glucose meters are comparable to standardised laboratory estima­tion of glucose. We conclude that strict quality control is necessary for the use of these strips in the hospital, clinic and home environments (JPMA 38: 75 , 1988).


The analyses on patients specimens are traditionally performed by laboratory personnel within their own laboratories. With the develop­ment of disposable dry chemistry systems e.g., urine strips, dextro Stix and portable analysers e.g., glucometers, Reflocheck haemoglobinome­ters, many simple tests are increasingly being per­formed by non-laboratory staff in wards and clinics. Anticipating such trends, a programme of extra laboratory tests was initiated at The Aga Khan Hospital and Medical College. This programme was conducted adhering to the recom­mendations of Association of Clinical Pathologists1 for extra laboratory tests. The use of reagent strips in glucose meters to monitor blood glucose concentrations in diabetic patients is gaining popularity mainly due to their favourable performance compared to standardised autoanalyser methods2 after allowance of differences in chemistry systems and between capillary and venous blood. Their use from self monitoring by diabetics in home environment has extended to clinics and wards3. To initiate our programme of extra laboratory tests, we started glucose estimations of finger prick capillary blood samples of known diabetic patients at a medical nursing station, by non-laboratory staff using a dry strip glucose meter . Simultaneous venous blood samples were sent to the clinical laboratory for glucose estimation by autoanalyser. We report our approach in setting up this extra laboratory procedure and the performance of the glucose meter in terms of usability by nonlaboratory staff and the reliability and accuracy of results obtained as compared to standardised autoanalyser results of the clinical laboratory.


A Reflocheck Glucose meter (Boehringer) was installed at a medical nursing station with reagent strips, disposable lancets and finger pricking device. Instructions on the use of the glucometer , given to resident medical officers and nurses, were: 1. Switch on the Reflocheck by pressing the orange ON/OFF button, 2. apply the blood sample (second drop) to the test patch (do not spread or smear the drop), and 3.then immediately press yellow start button. 4. After 1 minute (when the buzzer sounds and the display starts flashing) wipe off the blood, 5. smoothly and steadily insert the strip with the test patch facing upwards, 6. result is displayed automati­cally, 7. press white button and then withdraw the strip. The users were requested that each time a glucose estimation is done, a simultaneous venous blood sample should be sent to the labora­tory in the yellow fluorinated tubes provided. They were also requested to fill in a performance form (Table I)

each time a test was done and send their results with this form and the venous blood sample. The workability of the glucose meter was checked periodically by the lab staff and the accuracy of the instrument controlled by calibrated reagent strips with known glucose concentrations 100 and 300 rng/dl. Glucose estimations in the clinical laboratory were per­formed on a BECKMAN ASTRA autoanalyser using glucose—oxidase peroxidase technique. These results were controlled internally by com­mercial control sera and externally by Well-come and MEQAS quality control schemes. During the three months of this trial, 41 results were sent to the clinical lab along with perfor­mance forms and blood samples. The results and workability of the glucose meter were discussed with the medical staff. In view of the findings a decision was taken to continue the trial for an­other three months. This second trial, however, was to be conducted by one trained and specified resident medical officer. During this trial 47 results were returned along with performance forms and blood samples. Retrospectively first three months were called the First Trial and the second three months Second Trial.


Comparison of the Reflocheck and autoan­alyser estimations of glucose in the first trial show wide variations in the two systems as grap­hically represented in Figure 1.

The % of difference of glucose values in mg/dl of the two methods are given in

Table II and the results are markedly deviant from the manufacturers’ recom­mended 10% deviation. This 10% deviation is ex­pected and accepted partly due to the differences in chemistry system of the two methods and partly due to the difference in glucose levels in capillary and venous blood, since they tend to be about 10% higher in the capillary blood. Analysis of data from the performance forms is given in Table III.

This shows that the instrument was operational most of the time and most users were able to use it correctly, the main reason for occasionally not getting the results being improper insertion. During this trial 41 results were received by the clinical lab with performance slips and blood samples. The results of the first trial were unaccep­table since they were markedly deviant from the  accepted 10% limits. Operator error was seriously considered and one trained operator was asked to conduct the second trial. Comparison of Reflo­check and autoanalyser results show better cor­relation as graphically represented in Figure 2.

The % of difference in glucose values in mg/cl are given in Table IV

which shows that most results are within 10% deviation. Total deviation % in various glucose ranges is within accepted 10% except 300-400 range (Table V).

Analysis of data from performance slips shows excellent efficiency both by user and instrument (Table VI).

The second trial shows marked improvement in results (Tables II and IV).


The advent of solid state dry chemistry reagent strips and portable analysers have incre­ased the possibilities and accuracy of extra-lab tests. Acceptable results initially obtained with these systems in the home and clinic environments have extended their use to hospital wards. From single estimation systems, e.g., glucose and haemo­globin, portable analysers are now able to perform most routine chemistries with single or multiple reagent strips4. Our Medical department was keen to undertake performance of simple tests at nursing stations. Therefore the clinical laboratory started its programme of extra laboratory tests. Blood glucose measurement was an estimation of choice due to ready availability of glucose meters and their known accuracy. 5 The first trial, un-controlled as far as the selection of users were concerned, returned poor results markedly deviant from the 10% accepted limits, mainly due to the inability to use the glucometer correctly. Too high results (16.0 ± 18.6 % Table II)were attributed to delayed insertion of the strip, since time fraction can adversely effect the results with such strips6. And the accepted 10% high glucose expected in capillary blood samples. 7 Too low results (18.7 ± 21.4% Table II) were attributed to smearing or spreading of a small drop of blood to the surface area of reaction. These results formed the basis of the decision to identify one trained user to conduct the second trial. Consequently, the resuhs improved remarkably in this trial with overall deviation of around 10%. This gave the clinical laboratory and the medical staff confidence to initiate glucose meter blood glucose estimations at nursing stations, our first extra laboratory test. A few well trained operators were requested to use the Reflocheck . The Reflocheck has not replaced the standard glucose estimation in the laboratory, all inordinately high or low results are always confirmed in the clinical lab. Furthermore pro­blem diabetics are usually monitored by the clinical lab. We conclude that Glucose estimation by glucose meters at nursing stations in our set up is a viable proposition, provided strict quality check is maintained on who uses the instrument and that the laboratory controls the workabllity and accuracy of the instrument - Our results further confirm other’s observations that glucose meter results are comparable to standardised laboratory estimations, after allowance of differences in chemistry reagent of the two systems and the difference of glucose level in capillary and venous blood. Finally since the conclusion of this trial the Reflocheck has been in use for several months and we have encountered no problems.


We are grateful to S. Ejazuddin and Com­pany for providing the Reflocheck reagent strips and lancets and (Boehringer) for making this study possible.


1. Whitehead, T.P. and Garvey, K.Quality assess­ment of tests performed outside the laboratory. Association of Clinical Pathologists. Broadshcct. London, British Medical Association. 1985, p.114.
2. King, G., Steggles, D. and Harrop, J.S. Perfor­mance and storage of reagent strips for measuring blood glucose. Br. Med. J., 1982; 285: 1165.
3. Lawson, P.M., Kesson, C. M. and Ireland, J.T. Performance of blood-glucose strips. Lancct, 1979;2 :742.
4. Vincent Marks and Alberti, K.G.M. M. Procee­dings of symposium “Clinical biochemistry nearer the patient”. April 1986. University of Surrey. England, Boehringer Mannheim GMBH Publication, 1986.
5. Webb, DJ., Lovesay, J.M.,Ellis, A. and Knight, A. H. Blood glucose monitors; a laboratory and patient assessment. Br. Med. J., 1980; 280 :362.
6. Worth, R., Johnston, D.G., Anderson, J. and Alberti, K.G. MM. Performance of blood-glucose strips. Lancet, 1979;2 :742.
7. Raphael, 5.5. Basic hematologic techniques, in Lynch’s medical laboratory technology. 3rd ed. Philadelphia, Saunders, 1976; p. 1073.

Journal of the Pakistan Medical Association has agreed to receive and publish manuscripts in accordance with the principles of the following committees: