Diabetes Research and Clinical Practice, 10 (1990) 177-182 Elsevier 177 DIABET 000424 Quantifying the extent to which random plasma glucose underestimates diabetes prevalence in the Nauruan population Epidemiological note C.F. Finch, G.K. Dowse, V.R. Collins and P.Z. Zimmet WHO Collaborating Centre for the Epidemiology o/Diabetes Mellitus and Health Promotion for Non-Communicable Disease Control, Lions-International Diabetes Institute, Melbourne, Australia (Received 20 March 1990) (Accepted 19 April 1990) Summary The extent to which random plasma glucose levels underestimate the true prevalence of diabetes has been determined in Micronesian Nauruans. In 337 individuals who were screened on the basis of their random plasma glucose levels, the age-standardised prevalence based on a cut-off of 11.1 mmol/l underestimated the population prevalence based on a complete oral glucose tolerance test by 42% in males and 63% in females. At a cut-off level of 7.8 mmol/l the true age-standardised prevalence was underestimated by 16 and 38 %, in males and females, respectively. The use of random plasma glucose concentrations to determine the prevalence of diabetes, as currently defined, seems inappropriate. Performing oral glucose tolerance tests on smaller representative population samples should provide more accurate data at less expense than through large-scale screening utilizing random glucose levels. Key words: Random plasma glucose screening; Diagnostic cut-off, Prevalence Introduction Diabetes mellitus is currently defined on the basis of a 75 g oral glucose tolerance test (OGTT) although, for clinical purposes, this is rarely neces- sary [ 1,2]. In population surveys which aim to identify individuals with non-insulin-dependent diabetes mellitus (NIDDM), however, OGTTs are mandatory, since a large proportion of the Address for correspondence: Caroline F. Finch, Lions- International Diabetes Institute, PO Box 185, Caulfield South, Victoria, 3162, Australia. total number of cases are asymptomatic, or at least undiagnosed [4-61. In some instances it is difficult, or even impossible, to perform an OGTT and many researchers utilise fasting [4-61 or random [ 7-91 blood glucose screening tests, using various cut-offs, to either directly estimate the prevalence of diabetes or to select a sub-group in whom OGTTs may be more economically per- formed. In either instance, the prevalence will be an under-estimate of that which would be derived using a complete OGTT [3,10-131. Nauru is a small isolated island nation situated in the Central Pacific Ocean whose Micronesian 0168~8227/90/$03.50 0 1990 Elsevier Science Publishers B.V. (Biomedical Division)
Transcript
Diabetes Research and Clinical Practice, 10 (1990) 177-182
Elsevier 177
DIABET 000424
Quantifying the extent to which random plasma glucose underestimates diabetes prevalence in the Nauruan population
Epidemiological note
C.F. Finch, G.K. Dowse, V.R. Collins and P.Z. Zimmet
WHO Collaborating Centre for the Epidemiology o/Diabetes Mellitus and Health Promotion for Non-Communicable Disease Control,
Lions-International Diabetes Institute, Melbourne, Australia
(Received 20 March 1990) (Accepted 19 April 1990)
Summary
The extent to which random plasma glucose levels underestimate the true prevalence of diabetes has been determined in Micronesian Nauruans. In 337 individuals who were screened on the basis of their random plasma glucose levels, the age-standardised prevalence based on a cut-off of 11.1 mmol/l underestimated the population prevalence based on a complete oral glucose tolerance test by 42% in males and 63% in females. At a cut-off level of 7.8 mmol/l the true age-standardised prevalence was underestimated by 16 and 38 %, in males and females, respectively. The use of random plasma glucose concentrations to determine the prevalence of diabetes, as currently defined, seems inappropriate. Performing oral glucose tolerance tests on smaller representative population samples should provide more accurate data at less expense than through large-scale screening utilizing random glucose levels.
Key words: Random plasma glucose screening; Diagnostic cut-off, Prevalence
Introduction
Diabetes mellitus is currently defined on the basis of a 75 g oral glucose tolerance test (OGTT) although, for clinical purposes, this is rarely neces- sary [ 1,2]. In population surveys which aim to identify individuals with non-insulin-dependent diabetes mellitus (NIDDM), however, OGTTs are mandatory, since a large proportion of the
Address for correspondence: Caroline F. Finch, Lions- International Diabetes Institute, PO Box 185, Caulfield South, Victoria, 3162, Australia.
total number of cases are asymptomatic, or at least undiagnosed [4-61. In some instances it is difficult, or even impossible, to perform an OGTT and many researchers utilise fasting [4-61 or random [ 7-91 blood glucose screening tests, using various cut-offs, to either directly estimate the prevalence of diabetes or to select a sub-group in whom OGTTs may be more economically per- formed. In either instance, the prevalence will be an under-estimate of that which would be derived using a complete OGTT [3,10-131.
Nauru is a small isolated island nation situated in the Central Pacific Ocean whose Micronesian
inhabitants have become wealthy from phosphate mining. Two population-based surveys performed in Nauru (1975/76 and 1982) identified an unu- sually high prevalence of NIDDM [ 14-161. In February 1987, a further survey was undertaken with the aim of describing the current epi- demiology of diabetes in Nauru. This most recent study was primarily designed to be a longitudinal follow-up of individuals who had attended at least one of the earlier surveys in Nauru, i.e. in 1975/76 or 1982. Additionally, random plasma glucose concentrations were measured on all Nauruans, aged at least 25 years, who had not attended a previous diabetes survey. This report concerns the likely prevalence of diabetes in this latter group of individuals.
Materials and Methods
Survey procedure In January 1987, a household census of all
ethnic Nauruans, then resident in Nauru, was performed by community health nurses and individuals were identified as to whether or not they had attended a previous diabetes survey. Based on the completed census lists, all Nauruans were invited to attend a central site on specified days between 08.00 and 10.00 hours, during which venesection and other procedures were completed. Full details of the survey procedures are given elsewhere [ 171.
Individuals who had previously attended a diabetes survey, and those who were not known to have diabetes underwent a 2-h OGTT using 75 g of glucose monohydrate dissolved in 250 ml of water. A fasting plasma glucose determination only was made on individuals with a known and validated history of diabetes. Non-responders at the two earlier surveys were invited to undergo random blood glucose screening for diabetes. The WHO Study Group on Diabetes Mellitus [l]
suggest that details of the time and nature of the last meal eaten before a random blood glucose test have little effect on improving the interpre-
tation of the results. For this reason, such infor- mation was not obtained from the screenees, and so the random samples collected probably include post-breakfast as well as fasting values.
Blood for glucose measurement was collected into heparinized fluoridated tubes and centrifuged immediately. Plasma glucose was measured on- site within 3 h using a Yellow Springs Instrument 23AM glucose analyser (YSI Co., OH, U.S.A.).
Response rates Overall, the response rate in the 402 individuals
eligible for random glucose screening (22 y0 of the total Nauruan resident population aged at least 25 years) was 8 1% in males and 88 % in females. The total number of responders was 337. The age and sex structure of the study population was similar to that of the total Nauruan population.
Diagnostic criteria Random plasma glucose values were cate-
gorised as suggested by the WHO Study Group [ 11. Individuals with a random plasma glucose > 11.1 mmol/l were placed into the diabetes likely category whereas those with a value < 5.5 mmol/l were categorised as diabetes unlikely. Measure- ments falling between these two limits (i.e., 5.5-l 1.0 mmol/l, inclusive) were classified as diabetes uncertain.
For previous survey attendees who underwent an OGTT in 1987, diabetes was diagnosed if there was a confirmed previous history or a 2-h plasma glucose (2hPG) value 2 11.1 mmol/l. In the ab- sence of a 2hPG value, a fasting level > 7.8 mmol/l was considered diagnostic.
Statistical methods Unadjusted estimates of prevalence were based
on the numbers of individuals within the diabetes likely category. Directly age-standardised rates and the corresponding 95% confidence intervals (95% CI) were computed according to the method outlined in Armitage and Berry [ 181; the total screened population was used as the stan- dard.
Results
Prevalence of ‘likely’ diabetes Classification of diabetes status according to
random plasma glucose levels is shown in Table 1 by sex and age group. Overall, the majority of individuals were categorised as diabetes unlikelv, but 38.9% of all males and 38.5% of all females had glucose concentrations in the uncertain range (5.5-11.0 mmol/l).
The crude prevalence of diabetes in individuals aged at least 25 years (i.e., the proportion with likely diabetes) was 15.4% in males and 9.6% in females. The age-standardised prevalence of likely diabetes was higher in males 15.0% (95% CI: 9.8-20.2) than in females 9.8% (95% CI: 5.1-14.5). This sex differential was also reflected in the proportion of cases with unlikely diabetes status. The proportion of individuals in the unlikely category was highest in the 25-34 year olds.
TABLE 1
179
Age-adjusted diabetes prevalence figures based on different random plasma glucose cut-offs are given in Table 2. The true prevalence of diabetes in Nauru, based on the OGTT results is also quoted for each age-group. Prevalences based on random glucose levels are closest to the true values when a cut-off of 7.8 mmol/l is used except in the young males. On the other hand, a cut-off of 5.5 mmol/l (i.e., the combined uncertain and likely ranges) yields prevalence estimates which grossly over-estimate the true values in all but males aged 55 + years.
Agreement between the prevalences based on the various cut-offs and the true prevalence in Nauru varies with age. For both sexes, preva- lences based on the 5.5 cut-off are closest to the true values for individuals aged at least 55 years. For individuals aged 35-34 years, agreement is best when a cut-off of 7.8 mmol/l is used. The results are consistent with an increasing hypergly- caemia with age.
Classification of diabetes status according to random plasma glucose (RPG) levels (mmol/l): age- and sex-specific prevalence
* Age-standardised to the total screened population.
180
TABLE 2
Comparison of estimates of the prevalence of diabetes in Nauru (1987) calculated on the basis of random plasma glucose (RPG) levels with the true prevalence based on OGTT results
a Prevalence of diabetes (new and known cases) in previous survey attenders who had blood glucose levels measured in 1987. b Age-standardised to the total screened population.
Discussion
It is well-recognised that both random and fasting glucose perform poorly as diagnostic criteria for diabetes in either population screening programmes or epidemiologic studies [ 3,8-131. This is by no means surprising given that diabetes is defined on the basis of fasting hyperglycaemia and/or glucose intolerance as determined in a standardised OGTT. In the clinical situation, however, either an elevated random or fasting glucose measurement is usually sufficient to make the diagnosis in a patient in whom diabetes is suspected [ 1,191.
Few studies have quantified the magnitude of the under-estimation of diabetes prevalence through use of random plasma glucose levels relative to the 75 g OGTT. In this report we have found that the random plasma glucose diagnostic criterion, set at the level considered by WHO to strongly suggest diabetes [ 11, led to age-standar- dised prevalence estimates which were only 57.6% and 36.8% of the true Nauruan prevalence in males and females, respectively.
Combining the uncertain and likely categories of random plasma glucose levels (a 5.5 mmol/l) gave prevalence estimates in excess of those derived by the OGTT in all age groups except for those aged at least 55 years for whom the preva- lence approximated the true level (98.6% agree- ment in males, 89.0% in females). This confirms that a random plasma glucose cut-off around 5.5 mmol/l cannot be used to obtain reasonable population prevalence estimates. A similar con- clusion is drawn when one considers a diagnostic cut-off of 7.8 mmol/l. In this instance, prevalence estimates are generally less than the tnre figures but only marginally so for males aged 35-54 years; the only exceptions to this are the males aged 25-34 years.
Our observations suggest that, in order to ade- quately estimate diabetes prevalence, a more suit- able diagnostic cut-off for random plasma glucose levels in this population may lie in the range 5.5-7.8 mmol/l, although the optimal cut-off cannot be accurately assessed with the present data. The further indication that the most favour- able cut-off level is likely to be age dependent is
probably a reflection of the well documented age- related deterioration of glucose tolerance [20-231.
In certain situations, particularly in under- developed countries where resources are limited, investigators may still be tempted to use random glucose levels in estimating diabetes prevalence for health planning purposes [ 7-91. Certainly any such use should not be extended to aetiologic hypothesis testing, nor as a sole means of clinical diagnosis. We would recommend that if random measurements are to be used to estimate the pre- valence of diabetes in a population, a sub-sample of individuals should have OGTTs performed as well. In this way, the usual performance charac- teristics of the test (e.g., sensitivity, specificity etc.) could be determined and a more precise assessment of the usefulness of random glucose levels for diagnosis and prevalence estimation could be made. However, if this is to be done, it may be more prudent to dispense with the random plasma glucose determination altogether. Per- forming OGTTs on smaller, representative popu- lation samples may well provide more accurate data at similar or less expense than large-scale random plasma glucose screening.
Acknowledgements
This study was supported by NIH Grant DK 25446. We gratefully acknowledge the co- operation of the Government and people of Nauru. In particular, we would like to thank Dr. Kiki Thoma, Director of Medical Services, Nauru General Hospital for his support and assistance. Field work was carried out by stti from the Nauru General Hospital and from the Lions- International Diabetes Institute and Fairfield Hospital, Melbourne, Australia. We also thank Sue Foumel for preparation of the manuscript.
1
2
3
4
5
6
I
8
9
10
11
12
13
Report of a WHO Study Group. (1985) WHO Tech. Rep. Ser. 727. 10-16. National Diabetes Data Group. (1979) Classification and diagnosis of diabetes mellitus and other categories of glucose intolerance. Diabetes 28, 1039-1057. Dowse, G., Zimmet, P. and Alberti, G. (1988) Screening for diabetes and other categories of abnormal glucose tolerance. In: Keen H, DeFronzo R, Albert K, Zimmet P (eds). International Textbook of Diabetes Mellitus. John Wiley, London (in press). Ah&, B. and Corrigan, C.B. (1984) Prevalence of diabetes mellitus in north-western Tanzania. Diabeto- logia 26, 333-336. Fisch, A., Pichard, E., Prazuck, T., Leblanc, H., Sidibe, Y. and Briicker, G. (1987) Prevalence and risk factors of diabetes mellitus in the rural region of Mali (West Africa): a practical approach. Diabetologia 30, 859-862. Eason, R.J., Pada, J., Wallace, R., Henry, A. and Thornton, R. (1987) Changing patterns of hypertension, obesity and diet among Melanesians and Micronesians in the Solomon Islands. Med. J. Aust. 146, 465-473. Teuscher, T., Baillod, P., Rosman, J.B. and Teuscher, A. (1987) Absence of diabetes in a rural West African popu- lation with a high carbohydrate/cassava diet. Lancet i, 765-768. Moses. R., Colagiuri, S. and Shannon A. (1985) Effec- tiveness of mass screening for diabetes mellitus using random capillary blood glucose measurements. Med. J. Aust. 143, 544-546. Ohwovoriole, A., Kuti, J. and Kabiawu, S. (1988) Casual blood glucose levels and prevalence of undiscovered diabetes mellitus in Lagos Metropolis Nigerians. Diabet. Res. Clin. Pratt. 4, 153-158. Taylor, R. and Zimmet, P. (1978) Limitation of fasting plasma glucose for the diagnosis of diabetes mellitus. Diabet. Care 4, 556-558. Simon, D., Coignet, MC., Thibult, N., Senan, C. and Eschwege E. (1985) Comparison of glycosylated hemo- globin and fasting glucose with two-hour postload plasma glucose in the detection of diabetes mellitus. Am. J. Epi- demiol. 122, 589-593. Modan, M., Halkin, H., Karasik, A. and Lusky, A. (1984) Effectiveness of glycosylated hemoglobin, fasting plasma glucose and a single post-load glucose level in population screening for glucose intolerance. Am. J. Epidemiol. 119, 43 l-444. Ferrell, R., Hanis, G., Aguilar, L., Tulloch, B., Garcia, C. and Schull, W. (1984) Glycosylated haemoglobin deter- mination from capillary blood samples. Am. J. Epidemiol. 119. 159-166.
181
References
182
14 Zimmet, P.,Taft, P., Guinea, A., Guthrie, W. and Thoma, K. (1977) The high prevalence of diabetes mellitus on a central Pacific island. Diabetologia 13, 11 l-l IS.
15 Zimmet, P., Arblaster, M. and Thoma, K. (1978) The effect ofwesternization on native populations. Studies on a Micronesian community with a high diabetes pre- valence. Aust. N.Z.J. Med. 8, 141-146.
16 Zimmet, P., King, H., Taylor, et al. (1984) The high pre- valence of diabetes mellitus, impaired glucose tolerance and diabetic retinopathy in Nauru; the 1982 survey. Diabet. Res. 1, 13-18.
17 Collins, V.R., Dowse, G.K., Finch, C.F., Zimmet, P.Z. and Linnane, A.W. (1989) Prevalence and risk factors for micro- and macroalbuminuria in diabetic subjects and the total population of Nauru. Diabetes 38, 1602-1610.
18 Armitage, P. and Berry, G. (1987) Statistical methods in medical research, 2nd Edn., pp. 399-403, Blackwell Scientific Publications, Oxford.
19 Alberti, K.G.M.M. (1986) Tools for the diagnosis of diabetes. In: LP Krall (ed). World Book of Diabetes in Practice, Vol. 2, pp. 16-20. Elsevier Science Publishers, London.
20 Harris, M.I., Hadden, W.C., Knowler, WC. and Bennett, P.H. (1987) Prevalence of diabetes and impaired glucose tolerance and plasma glucose levels in US population aged 20-74 years. Diabetes 36, 523-534.
21 Mahns, J.M. (1974) Glucose tolerance and the diabetic population. Postgrad. Med. J. 50 (Suppl. 3) 529-537.
22 Defronzo, R.A. (1981) Glucose intolerance and aging. Diabet. Care 4, 493-501.
23 Zimmet, P. and Whitehouse, S. (1979) The effect of age on glucose tolerance. Studies in a Micronesian popula- tion with a high prevalence of diabetes. Diabetes 28, 617-623.
