- Email: [email protected]
Translation of the recommendations for the diagnosis of diabetes mellitus into daily clinical practice in a primary health care setting
Diabetes Research and Clinical Practice 62 (2003) 123 /129 www.elsevier.com/locate/diabres
Translation of the recommendations for the diagnosis of diabetes mellitus into daily clinical practice in a primary health care setting J.M. Gonza´lez-Clemente a,b,*, G. Galdon c, J. Mitjavila c, A. Min˜arro c, G. Gime´nez-Pe´rez a, D. Mauricio a,d a
Servei de Diabetis, Endocrinologia i Nutricio´, Hospital de Sabadell, Corporacio´ Parc Taulı´, s/n. 08208 Sabadell (Barcelona), Spain b Departament de Ciencies Experimentals i de la Salut, Universitat Pompeu Fabra, Barcelona, Spain c CAP Les Corts, Corporacio´ Sanita`ria Clı´nic, Barcelona, Spain d Institut Universitari Parc Tauli, Universitat Auto`noma de Barcelona, Sabadell, Spain Received 17 March 2003; received in revised form 7 July 2003; accepted 11 July 2003
Abstract Aims: to assess the implementation of 1985 recommendations for the diagnosis of diabetes (World Health Organization [WHO]) in a primary care setting, and the physician’s attitude toward the diagnosis of diabetes mellitus (DM). Material and methods: subjects with a fasting plasma glucose (FPG) /6.1 mmol/l (]/110 mg/dl) and with previously unknown glucose tolerance status were identified retrospectively in a primary health care center during a 45month period. The following variables were evaluated: anthropometric parameters, fasting plasma glucose and oral glucose tolerance test (OGTT) values, registration of a diagnosis of diabetes in clinical records, smoking status, lipid profile and blood pressure. Results: 1181 subjects with a FPG /6.1 mmol/l were identified (target population): 171 with a FPG /7.8 mmol/l and 1010 with a FPG between 6.1 and 7.7 mmol/l. In the latter group, an OGTT was performed in 553 subjects (54.8%) (173 yielded a diagnosis of diabetes). During the study period, diabetes was diagnosed in 29.1% (n /344) of the target population. Following the 1985 WHO recommendations, a confirmatory diagnostic test was repeated in 92 (69.7%) subjects with a FPG between 7.8 and 11.0 mmol/l, and in 132 subjects (23.87%) who had already received an initial OGTT. The analysis of the diagnostic process followed by the different physicians revealed a high interindividual variability in terms of: proportion of cases diagnosed as diabetes by an OGTT (from 35.7 to 65.2), percentage of subjects with a FPG 6.1 /7.7 mmol/l without an OGTT (7.33 /70.27%), proportion of confirmatory OGTTs (0 /57.89%), and percentage of misdiagnosed cases (1.16 /6.34%). The percentage of subjects misdiagnosed was negatively correlated with the proportion of OGTT repetitions. Conclusions: 1985 WHO recommendations for the diagnosis of diabetes are only partially followed at a primary health care level. There is a high interindividual variability among physicians in the implementation of these recommendations that is associated with the misdiagnosis of diabetes. # 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Diabetes diagnosis; Oral glucose tolerance test; Impaired fasting glucose; Impaired glucose tolerance; Primary care
* Corresponding author. Tel.: /34-93-745-8412; fax: /34-93-716-0646. E-mail address: [email protected] (J.M. Gonza´lez-Clemente). 0168-8227/03/$ - see front matter # 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0168-8227(03)00171-2
124
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
1. Introduction Early detection of diabetes mellitus is recommended by the World Health Organization (WHO) and the American Diabetes Association. Therefore, early diagnosis of diabetes should be performed in those subjects considered to be at risk for the disease. Screening is usually recommended in subjects with the following risk factors: age /45 years, obesity, family history of diabetes mellitus, members of high risk ethnic populations, history of gestational diabetes, hypertension or dyslipidaemia (HDL cholesterol B/35 mg/dl and/or triglyceride /250 mg/dl), previous impaired glucose tolerance (IGT) or impaired fasting glucose (IFG) [1,2]. In clinical practice, it is recommended that the diagnosis of diabetes in assymptomatic subjects should always be confirmed by repeating the diagnostic test on a different day [1,2]. Opportunistic screening during routine check-ups in the primary health care setting is probably the best strategy in order to identify and diagnose the disease in high risk individuals [3], since health care is delivered to the majority of subjects with diabetes at this level [4]. However, it is estimated that up to 50% of the population with diabetes remain undiagnosed [5,6]. Furthermore, diabetes may remain undiagnosed and without treatment since the information given to some subjects leads them to admit that they have ‘high sugar’ or ‘a little touch of sugar’ but not diabetes [7 /9]. Most of the studies which have evaluated the attitude of primary care physicians towards diabetes focused mainly on the assessment of process and outcome measures of the disease [10]. The initial diagnosis of diabetes is a crucial step for the subsequent care and prognosis of diabetes, leading to decisions about treatment and the use of health resources [11]. However, little is known about the translation of the recommended diagnostic procedures and strategies in a primary care setting. Also, few data are available about the quality and accuracy of the diagnostic procedures of diabetes applied in real daily practice. In this study, we aimed to evaluate the implementation of the diagnostic recommendations of the WHO issued in 1985 in a primary health care
setting. We also evaluated the interindividual variation in physician’s diagnostic practice.
2. Material and methods In Spain, the public health care authority of each region covers the primary health care needs of the population. Each citizen has a reference primary health care team (a family physician and a nurse) who work in a Primary Health Care Center. This study was carried out at the primary health care center of Les Corts, in one of the nine health care districts of the city of Barcelona (population /1 500 000). The center opened in December 1994. By 31st October 1998, 20 878 subjects (age /15 years) had been registered (all living in the area, and almost all of them Caucasian). In the center, there are eight primary health care teams. Subjects are distributed among the health care teams regardless of their clinical characteristics. No formal scheme existed for sharing diabetic care between the reference hospital and the primary care center. Subjects were identified through the computerized database of clinical records. This database was combined with other databases in order to identify those patients with known diabetes by January 1995: disease database, pharmacy database and another one which included data on subjects using insulin and self-monitoring systems. Those subjects with known diabetes as recorded in any of the databases by 1st January 1995 were excluded from further analysis. The clinical records of these subjects with diabetes were checked in order to confirm the diagnosis. Type 1 diabetes was diagnosed according to previously described criteria [4]. After exclusion of women with gestational diabetes, the remaining subjects with known diabetes were classified as having type 2 diabetes. The computerized laboratory database was searched in order to identify the target population, consisting of those subjects with a fasting plasma glucose (FPG) value of /6.1 mmol/l (/110 mg/ dl) and previously unknown glucose tolerance status. In most cases, blood glucose was routinely determined when subjects were first assessed by the
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
primary health care team. After identification of subjects belonging to the target population, all clinical records were retrospectively reviewed in detail. For each subject the following data were obtained: age; sex; height; weight; type of DM; smoking status; physician in charge; FPG value; glucose values of the OGTT, whenever performed; presence and results of a confirmatory test (FPG or OGTT) in the first 6 months after the first diagnostic test; diagnosis of glucose tolerance status as recorded by the physician in charge. History of hypertension was also registered and was defined as values of systolic blood pressure and diastolic blood pressure /140 and 90 mmHg, respectively. Inclusion of these data in the clinical records were among the internal quality control issues predefined before the primary health care center opened. Every OGTT was performed with patient preparation and result interpretation according to the 1985 recommendations of the WHO [2]. Therefore, subjects were classified as having diabetes, impaired glucose tolerance or normal glucose tolerance (NGT) [2]. Misdiagnosis by the physician, as recorded in the clinical records, was defined whenever any of the following occurred: assignment of a given subject to one of the diagnostic categories without fulfilling the WHO criteria for that category (false positive cases); or, fulfillment of criteria for any category of glucose tolerance status, without registration of the diagnosis in the clinical record (false negative cases). Plasma glucose was measured by the glucose oxidase method (Hitachi 704 analyzer). Data are presented as means9/S.D. The statistical analysis was performed using the SPSS software (SPSS, Chicago). Group frequencies were compared using the x2 or Fisher’s exact test where appropriate. Differences between group means were tested using the t test. All tests were twosided. To evaluate partial correlations, the nonparametric Spearman rank correlation test was used. A P value B/0.05 was considered statistically significant and subsequently the 95% CI were calculated.
125
3. Results By the combined search of the different databases, 736 subjects with known diabetes mellitus were identified (691 with type 2 and 45 with type 1 diabetes). An additional number of 1212 subjects belonging to the target population were identified (Fig. 1), i.e. subjects with a FPG /6.1 mmol/l and previously unknown glucose tolerance status. Data were missing from nine subjects, who were thus not included in the analysis. During the study period, four newly-diagnosed subjects with type 1 diabetes and 27 women with gestational diabetes were detected. After exclusion of these 31 subjects, a total of 1181 subjects with a FPG /6.1 mmol/l were finally evaluated (age: 62.619/13.7 years, 588 women and 593 men). Of these subjects, 171 had a FPG /7.8 mmol/l (]/140 mg/dl), and 1010 between 6.1 and 7.7 mmol/l. Table 1 shows the differences between subjects with a FPG /7.8 mmol/l and those with a FPG between 6.1 and 7.7 mmol/l. As shown, in the group of subjects with a FPG /7.8 mmol/l, there was a higher percentage of men and smokers, and they had higher values of BMI and FPG. Analysis of the implementation of the 1985 WHO recommendations revealed that there were 344 subjects with type 2 diabetes (29.13%) (Fig. 1). Of these subjects, 173 (50.29%) were diagnosed by OGTT. Forty-four subjects (3.73%) were misdiagnosed. Thus, in the group with a FPG between 7.8 and 11.0 mmol/l (n /132), 13 subjects (9.85%) were not recorded as having diabetes (false negative cases). In the group of subjects without an OGTT (n /457), 14 (3.06%) were classified in any of the 1985 WHO diagnostic categories without fulfilling the criteria (false positive cases). In the group of subjects who underwent an OGTT (n / 553), there were 12 false positive cases of diabetes, three false negative cases of IGT, and two false positive cases of IGT. This yielded an overall prevalence of misdiagnosis of 3.07%. The investigation of the presence of a confirmatory diagnostic test showed that in 132 subjects with a FPG between 7.8 and 11.0 mmol/l, FPG was reassessed in 92 cases (69.70%) in the first 6 months after the first FPG. Of the 553 subjects who had received an OGTT, 132 (23.87%) under-
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
126
Fig. 1. Implementation of the WHO criteria in 1181 subjects with a FPG /6.1 mmol/l, attending a primary care unit.
went a new test in the 6 months that followed the first OGTT. However, 457 subjects did not receive the recommended OGTT after an abnormal FPG. This latter group of subjects were older and had lower FPG levels as compared to the group of subjects who received an OGTT (Table 2). Each physician saw on average 800 patients/ month, except for one physician (number 5) who consistently had only 60% of this workload. Following the WHO criteria, the total number of subjects diagnosed with diabetes ranged from 29 to 69 and the proportion of subjects diagnosed with diabetes by an OGTT from 35.71 to 65.22% (Table 3). Similarly, we found a high variability in the proportion of subjects with a FPG between 6.1 and 7.7 mmol/l but without a subsequent OGTT (from 7.33 to 70.27%), in the percentage of
subjects with a FPG between 7.8 and 11.0 mmol/ l who had an FPG repeated (from 52.38 to 85.72%) and in the proportion of subjects who had an OGTT repeated (from 0.00 to 56.83%) (Table 3). Percentages of subjects misdiagnosed ranged from 1.16 to 6.34%. The percentage of subjects misdiagnosed was negatively correlated with the proportion of confirmatory OGTT repetitions (r / /0.73, P /0.04).
4. Discussion In relation to the 1985 WHO recommendations, several problems have been detected. First, this study confirms that physicians do not perform an adequate investigation of subjects with an FPG /
Table 1 Clinical characteristics of the 1181 subjects with an FPG/6.1 mmol/l, according to their FPG values
Age (years) Sex (n , % women) BMI (kg/m2) FPG (mmol/l) Smokers (n , %) Hypertension (n , %)
Subjects with a FPG/7.8 (n /171)
Subjects with a FPGB/7.8 (n /1010)
P
61.29/12.3 72 (42.1) 30.89/6.4 10.49/3.5 59 (35.5) 102 (59.7)
62.89/13.9 516 (51.1) 29.29/4.6 6.69/0.4 204 (20.2) 529 (52.4)
0.32 0.03 0.00 0.00 0.00 0.08
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
127
Table 2 Clinical characteristics of the 1010 subjects with an FPG between 6.1 and 7.7 mmol/l, with and without OGTT
Age (years) Sex (% women) BMI (kg/m2) FPG (mmol/l) Smokers (%) Hypertension (n , %)
Subjects with OGTT (n /553)
Subjects without OGTT (n /457)
P
61.49/13.1 286 (51.7) 29.19/4.6 6.79/0.4 111 (20.1) 296 (53.5)
64.49/14.6 230 (50.3) 29.49/4.6 6.59/6.4 93 (20.4) 233 (51.0)
0.00 0.66 0.40 0.00 0.91 0.42
6.1 mmol/l. Although 50% of the subjects were diagnosed with diabetes by an OGTT, a figure that compares favorably with previous data [12], almost 50% of subjects with a FPG value between 6.1 and 7.7 mmol/l who should have undergone an OGTT did not receive the test. These subjects had slightly lower FPG values and were older in relation to subjects who received the OGTT. This is a relevant finding since it has been shown that in older people a higher proportion of cases of diabetes are detected through an OGTT [13,14]. Our results show that phsysician’s attitudes towards the assessment of subjects with a FPG between 6.1 and 7.7 mmol/l are inadequate to detect diabetes and impaired glucose tolerance. These results are especially relevant, if we take into account that life-style and pharmacological interventions could prevent the development of diabetes in subjects with impaired glucose tolerance [15,16]. Few published studies have evaluated the physician’s attitudes towards screening and diagnosis of diabetes [17,18]. In a previous study evaluating screening for diabetes, it was shown that screening rates improved with an increasing number of risk factors for diabetes [17]. However, this study did not evaluate the physician’s attitudes towards screening, the data being obtained through a questionnaire administered to subjects included in the study. A more recent investigation concluded that factors associated with screening diabetes in a general population aged /45 years are: the number of visits to any health care provider and the presence of high cholesterol levels, or family history of diabetes [18]. In the present study, the glucose tolerance status of 3.73% of subjects was not properly diagnosed.
Although data on errors made in the diagnosis of diabetes are scarce, it has been previously described that about 10% of the subjects registered as having diabetes in a general practice do not have the disease [10]. In concordance with this finding, it should be noted that a remarkable proportion of subjects (about 10%) with a FPG between 7.8 and 11.0 mmol/l were not diagnosed as being diabetes by the physicians in our study. Additionally, the proportion of subjects misdiagnosed was similar in subjects with an FPG between 6.1 and 7.7 mmol/l regardless of the performance of an OGTT. It could seem likely that all these subjects are not explicitly informed of the diagnosis of diabetes. To some extent, this finding may help to explain why some subjects only know that they have ‘sugar’ [8] or a ‘little touch of sugar’ [9]. In a clinical setting, it is recommended that an asymptomatic individual with an abnormal OGTT should undergo a second confirmatory test before a final diagnosis of diabetes is made [1,2]. This study confirms previous research showing that in everyday practice the OGTT is not only performed infrequently, but it is commonly not even used to confirm suspected cases [19]. Thus, we found that additional confirmatory testing is only performed in one out of four initial OGTTs, and seven out of ten high FPG values (]/7.8 mmol/l). We are not aware of studies investigating strategies to improve the adherence to the recommendations on the diagnosis of diabetes. Information technologies could be useful tools for improving the diagnosis process, as recently suggested [20]. As subjects are randomly distributed among health care teams, it is assumed that the wide variation observed in practice patterns reflects different physicians’ attitudes towards the impor-
44 9 (6.25) 3 (2.05) 9 (6.34) 5 (5.95) 2 (1.16) 4 (2.48) 7 (3.98)
5 (3.23)
132 0 (0.00) 2 (1.56) 0 (0.00) 0 (0.00) 79 (56.83) 21 (19.63) 12 (15.19) 18 (29.51)
176 161 155 173 42 53 57 69 4 8 3 3 16 19 22 21 22 (52.38) 26 (49.06) 32 (56.14) 45 (65.22) 155 136 131 150 94 (60.65) 57 (41.91) 24 (18.32) 11 (7.33) 12 (75.00) 15 (78.95) 12 (54.55) 18 (85.72)
Subjects evaluated (n ) Subjects with newly-diagnosed DM (n ) With FPG/11.1 mmol/l (n ) With FPG 7.8 /11.0 mmol/l (n ) Through OGTT (n , %) Subjects with FPG 6.1 /7.7 mmol/l (n ) Without OGTT (n , %) Subjects with FPG 7.8 /11.0 mmol/l and FPG repetition (n , %) Subjects with FPG 6.1 /7.7 mmol/l, OGTT and OGTT repetition (n , %) Misdiagnosis (n , %)
8 7 6 5 4 3 2 1
Physician
Table 3 Data referring to the diagnosis of diabetes performed by the eight physicians evaluated during a 45-month period
84 142 146 144 1181 14 41 39 29 344 4 10 2 5 39 5 16 21 12 132 5 (35.71) 15 (36.59) 16 (41.03) 12 (41.38) 173 74 111 128 125 1010 52 (70.27) 52 (46.85) 85 (66.41) 82 (65.60) 457 3 (60.00) 13 (81.25) 11 (52.38) 8 (66.67) 92
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129 Group
128
tance of the diagnosis of diabetes. There is wide discordance among physicians in relation to the way they diagnose type 2 diabetes (by OGTT or FPG), in the rate of performance of additional confirmatory tests, and in the proportion of subjects misdiagnosed. In addition, the proportion of misdiagnosed subjects was negatively correlated to the percentage of subjects with a confirmatory OGTT. Thus, physicians who do not routinely repeat the OGTT show a poorer performance in the diagnosis of diabetes, and have a higher rate of misdiagnosed cases. Therefore, it is suggested that the repetition of a confirmatory OGTT may be regarded as an index of the quality of the diagnostic process of diabetes. Finally, our study has some obvious limitations. First, the analysis is limited to a relatively small area where a relatively high proportion of subjects (50.29%) were diagnosed through an OGTT. For instance, in the US most diagnoses of diabetes are made in symptomatic subjects [21], and the OGTT was previously used in the diagnosis of about 38.0% of the subjects [12]. Thus, some of the results obtained would be of limited value in areas where subjects with diabetes are mainly diagnosed through a FPG. Furthermore, it could also be argued that the data about physicians’ attitudes are difficult to interpret since part of the data were collected following the recommendations of the American Diabetes Association in 1997. However, the provisional revision of the WHO recommendations was issued after the collection of data for the present study [22]. Furthermore, the local family practice recommendations on the diagnosis of diabetes were only modified in January 2000. However, in Spain, there are no studies evaluating the adherence to the 1985 WHO recommendations or the 1997 American Diabetes Association recommendations. In conclusion, subjects with a FPG /6.1 mmol/l evaluated in a primary care setting were not accurately diagnosed according to the current WHO recommendations. A high proportion of these subjects did not receive an OGTT and were older with a more detrimental lipid profile than subjects who received an OGTT. The high between-physician variability in the implementation of the current diagnostic recommendations of
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
diabetes should be considered in order to significantly improve the detection of the disease.
Acknowledgements Without the cooperation of the CAP Les Corts and all its members (especially A. Anguita, J. Benavent, M. Borrell, F. Cano, C. Forcada, G. Martı´nez, J. Ortiz, M. Pinyol, J. Ricart) this study would have been impossible. Many thanks to them. We are also deeply indebted to M. P. Delgado (Hospital Casa Maternitat, Corporacio´ Sanita`ria Clı´nic). This work was partially supported by a grant from the ‘Fondo de Investigaciones Sanitarias (FIS)’: RGDM 03/212 (Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo, Spain).
References [1] The expert committee on the diagnosis and classification of diabetes mellitus, Report of the expert committee on the diagnosis and classification of diabetes mellitus, Diabetes Care 20 (1997) 1183 /1197. [2] World Health Organization. Diabetes Mellitus: report of a WHO Study Group. Technical Report Series No. 727. Geneva: WHO, 1985. [3] M.M. Engelgau, R.E. Aubert, T.J. Thompson, W.H. Herman, Screening for NIDDM in non-pregnant adults, Diabetes Care 18 (1995) 1606 /1618. [4] M.I. Harris, Medical care for patients with diabetes, Ann. Intern. Med. 124 (1996) 117 /122. [5] M.I. Harris, Undiagnosed NIDDM: clinical an public health issues, Diabetes Care 16 (1993) 642 /652. [6] N.J. Wareham, S. O’Rahilly, The changing classification and diagnosis of diabetes, Br. Med. J. 317 (1998) 359 /360. [7] M.I. Harris, R.E.: Klein, T.A. Welborn, M.W. Knuiman, Onset of NIDDM occurs at least 4 /7 yr before clinical diagnosis, Diabetes Care 15 (1992) 815 /819. [8] J. Schorling, J.T. Saunders, Is ‘‘sugar’’ the same as diabetes?, Diabetes Care 23 (2000) 330 /334. [9] R.G. Hiss, The concept of diabetes translation. Addressing barriers to widespread adoption of new science into clinical care, Diabetes Care 24 (2001) 1293 /1296.
129
[10] P. Patchett, D. Roberts, Diabetic patients who do not have diabetes: investigation of register of diabetic patients in general practice, Br. Med. J. 308 (1994) 1225 /1226. [11] F. Vinicor, When is diabetes diabetes?, J. Am. Med. Assoc. 281 (1999) 1222 /1223. [12] M.I. Harris, Frequency of oral glucose tolerance testing in the U.S, Diabetes Care 18 (1995) 143 /145. [13] The Decode Study Group on behalf or the European Diabetes Epidemiology Group, Consequences of the new diagnostic criteria for diabetes in older men and women. DECODE Study (Diabetes Epidemiology: Collaborative Analysis of Diagnostic Criteria in Europe), Diabetes Care 22 (1999) 1667 /71. [14] The Decode Study Group on behalf or the European Diabetes Epidemiology Group, Will new diagnostic criteria for diabetes mellitus change phenotype of patients with diabetes? Reanalysis of European epidemiological data, Br Med J 317 (1998) 371 /37. [15] J. Tuomilehto, J. Lindstro¨m, J.G. Eriksson, et al., Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance, N. Engl. J. Med. 344 (2001) 1343 /1350. [16] Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes mellitus with lifestyle intervention or metformin. N Engl J Med 2002; 346: 393 / 403. [17] C.C. Cowie, M.I. Harris, M.S. Eberhardt, Frequency and determinants of screening for diabetes in the U.S, Diabetes Care 17 (1994) 1158 /1163. [18] T.S. Harwell, J.G. Smilie, J.M. McDowall, S.D. Helderson, D. Gohdes, Diabetes screening practices among individuals aged 45 years and older, Diabetes Care 23 (2000) 125 /126. [19] R.P. Solk, T.J. Orchard, D.E. Grobbee, Why use the oral glucose tolerance test?, Diabetes Care 18 (1995) 1045 / 1049. [20] D.W. Bates, A.A. Gawande, Improving safety with information technology, N. Engl. J. Med. 348 (2003) 2526 /2534. [21] M.I. Harris, Definition and classifications of diabetes mellitus and the new criteria for diagnosis, in: D. LeRoith, S.I. Taylor, J.M. Olefsky (Eds.), Diabetes Mellitus. A Fundamental and Clinical Text, 2nd edition, Lippincott, Williams and Wilkins, Philadelphia, PA, 2000, pp. 326 / 334. [22] KGMM Alberti, PZ Zimmet for the WHO Consultation, Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus. Provisional report of a WHO consultation, Diab Med 15 (1998) 539 /553.
Translation of the recommendations for the diagnosis of diabetes mellitus into daily clinical practice in a primary health care setting J.M. Gonza´lez-Clemente a,b,*, G. Galdon c, J. Mitjavila c, A. Min˜arro c, G. Gime´nez-Pe´rez a, D. Mauricio a,d a
Servei de Diabetis, Endocrinologia i Nutricio´, Hospital de Sabadell, Corporacio´ Parc Taulı´, s/n. 08208 Sabadell (Barcelona), Spain b Departament de Ciencies Experimentals i de la Salut, Universitat Pompeu Fabra, Barcelona, Spain c CAP Les Corts, Corporacio´ Sanita`ria Clı´nic, Barcelona, Spain d Institut Universitari Parc Tauli, Universitat Auto`noma de Barcelona, Sabadell, Spain Received 17 March 2003; received in revised form 7 July 2003; accepted 11 July 2003
Abstract Aims: to assess the implementation of 1985 recommendations for the diagnosis of diabetes (World Health Organization [WHO]) in a primary care setting, and the physician’s attitude toward the diagnosis of diabetes mellitus (DM). Material and methods: subjects with a fasting plasma glucose (FPG) /6.1 mmol/l (]/110 mg/dl) and with previously unknown glucose tolerance status were identified retrospectively in a primary health care center during a 45month period. The following variables were evaluated: anthropometric parameters, fasting plasma glucose and oral glucose tolerance test (OGTT) values, registration of a diagnosis of diabetes in clinical records, smoking status, lipid profile and blood pressure. Results: 1181 subjects with a FPG /6.1 mmol/l were identified (target population): 171 with a FPG /7.8 mmol/l and 1010 with a FPG between 6.1 and 7.7 mmol/l. In the latter group, an OGTT was performed in 553 subjects (54.8%) (173 yielded a diagnosis of diabetes). During the study period, diabetes was diagnosed in 29.1% (n /344) of the target population. Following the 1985 WHO recommendations, a confirmatory diagnostic test was repeated in 92 (69.7%) subjects with a FPG between 7.8 and 11.0 mmol/l, and in 132 subjects (23.87%) who had already received an initial OGTT. The analysis of the diagnostic process followed by the different physicians revealed a high interindividual variability in terms of: proportion of cases diagnosed as diabetes by an OGTT (from 35.7 to 65.2), percentage of subjects with a FPG 6.1 /7.7 mmol/l without an OGTT (7.33 /70.27%), proportion of confirmatory OGTTs (0 /57.89%), and percentage of misdiagnosed cases (1.16 /6.34%). The percentage of subjects misdiagnosed was negatively correlated with the proportion of OGTT repetitions. Conclusions: 1985 WHO recommendations for the diagnosis of diabetes are only partially followed at a primary health care level. There is a high interindividual variability among physicians in the implementation of these recommendations that is associated with the misdiagnosis of diabetes. # 2003 Elsevier Ireland Ltd. All rights reserved. Keywords: Diabetes diagnosis; Oral glucose tolerance test; Impaired fasting glucose; Impaired glucose tolerance; Primary care
* Corresponding author. Tel.: /34-93-745-8412; fax: /34-93-716-0646. E-mail address: [email protected] (J.M. Gonza´lez-Clemente). 0168-8227/03/$ - see front matter # 2003 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/S0168-8227(03)00171-2
124
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
1. Introduction Early detection of diabetes mellitus is recommended by the World Health Organization (WHO) and the American Diabetes Association. Therefore, early diagnosis of diabetes should be performed in those subjects considered to be at risk for the disease. Screening is usually recommended in subjects with the following risk factors: age /45 years, obesity, family history of diabetes mellitus, members of high risk ethnic populations, history of gestational diabetes, hypertension or dyslipidaemia (HDL cholesterol B/35 mg/dl and/or triglyceride /250 mg/dl), previous impaired glucose tolerance (IGT) or impaired fasting glucose (IFG) [1,2]. In clinical practice, it is recommended that the diagnosis of diabetes in assymptomatic subjects should always be confirmed by repeating the diagnostic test on a different day [1,2]. Opportunistic screening during routine check-ups in the primary health care setting is probably the best strategy in order to identify and diagnose the disease in high risk individuals [3], since health care is delivered to the majority of subjects with diabetes at this level [4]. However, it is estimated that up to 50% of the population with diabetes remain undiagnosed [5,6]. Furthermore, diabetes may remain undiagnosed and without treatment since the information given to some subjects leads them to admit that they have ‘high sugar’ or ‘a little touch of sugar’ but not diabetes [7 /9]. Most of the studies which have evaluated the attitude of primary care physicians towards diabetes focused mainly on the assessment of process and outcome measures of the disease [10]. The initial diagnosis of diabetes is a crucial step for the subsequent care and prognosis of diabetes, leading to decisions about treatment and the use of health resources [11]. However, little is known about the translation of the recommended diagnostic procedures and strategies in a primary care setting. Also, few data are available about the quality and accuracy of the diagnostic procedures of diabetes applied in real daily practice. In this study, we aimed to evaluate the implementation of the diagnostic recommendations of the WHO issued in 1985 in a primary health care
setting. We also evaluated the interindividual variation in physician’s diagnostic practice.
2. Material and methods In Spain, the public health care authority of each region covers the primary health care needs of the population. Each citizen has a reference primary health care team (a family physician and a nurse) who work in a Primary Health Care Center. This study was carried out at the primary health care center of Les Corts, in one of the nine health care districts of the city of Barcelona (population /1 500 000). The center opened in December 1994. By 31st October 1998, 20 878 subjects (age /15 years) had been registered (all living in the area, and almost all of them Caucasian). In the center, there are eight primary health care teams. Subjects are distributed among the health care teams regardless of their clinical characteristics. No formal scheme existed for sharing diabetic care between the reference hospital and the primary care center. Subjects were identified through the computerized database of clinical records. This database was combined with other databases in order to identify those patients with known diabetes by January 1995: disease database, pharmacy database and another one which included data on subjects using insulin and self-monitoring systems. Those subjects with known diabetes as recorded in any of the databases by 1st January 1995 were excluded from further analysis. The clinical records of these subjects with diabetes were checked in order to confirm the diagnosis. Type 1 diabetes was diagnosed according to previously described criteria [4]. After exclusion of women with gestational diabetes, the remaining subjects with known diabetes were classified as having type 2 diabetes. The computerized laboratory database was searched in order to identify the target population, consisting of those subjects with a fasting plasma glucose (FPG) value of /6.1 mmol/l (/110 mg/ dl) and previously unknown glucose tolerance status. In most cases, blood glucose was routinely determined when subjects were first assessed by the
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
primary health care team. After identification of subjects belonging to the target population, all clinical records were retrospectively reviewed in detail. For each subject the following data were obtained: age; sex; height; weight; type of DM; smoking status; physician in charge; FPG value; glucose values of the OGTT, whenever performed; presence and results of a confirmatory test (FPG or OGTT) in the first 6 months after the first diagnostic test; diagnosis of glucose tolerance status as recorded by the physician in charge. History of hypertension was also registered and was defined as values of systolic blood pressure and diastolic blood pressure /140 and 90 mmHg, respectively. Inclusion of these data in the clinical records were among the internal quality control issues predefined before the primary health care center opened. Every OGTT was performed with patient preparation and result interpretation according to the 1985 recommendations of the WHO [2]. Therefore, subjects were classified as having diabetes, impaired glucose tolerance or normal glucose tolerance (NGT) [2]. Misdiagnosis by the physician, as recorded in the clinical records, was defined whenever any of the following occurred: assignment of a given subject to one of the diagnostic categories without fulfilling the WHO criteria for that category (false positive cases); or, fulfillment of criteria for any category of glucose tolerance status, without registration of the diagnosis in the clinical record (false negative cases). Plasma glucose was measured by the glucose oxidase method (Hitachi 704 analyzer). Data are presented as means9/S.D. The statistical analysis was performed using the SPSS software (SPSS, Chicago). Group frequencies were compared using the x2 or Fisher’s exact test where appropriate. Differences between group means were tested using the t test. All tests were twosided. To evaluate partial correlations, the nonparametric Spearman rank correlation test was used. A P value B/0.05 was considered statistically significant and subsequently the 95% CI were calculated.
125
3. Results By the combined search of the different databases, 736 subjects with known diabetes mellitus were identified (691 with type 2 and 45 with type 1 diabetes). An additional number of 1212 subjects belonging to the target population were identified (Fig. 1), i.e. subjects with a FPG /6.1 mmol/l and previously unknown glucose tolerance status. Data were missing from nine subjects, who were thus not included in the analysis. During the study period, four newly-diagnosed subjects with type 1 diabetes and 27 women with gestational diabetes were detected. After exclusion of these 31 subjects, a total of 1181 subjects with a FPG /6.1 mmol/l were finally evaluated (age: 62.619/13.7 years, 588 women and 593 men). Of these subjects, 171 had a FPG /7.8 mmol/l (]/140 mg/dl), and 1010 between 6.1 and 7.7 mmol/l. Table 1 shows the differences between subjects with a FPG /7.8 mmol/l and those with a FPG between 6.1 and 7.7 mmol/l. As shown, in the group of subjects with a FPG /7.8 mmol/l, there was a higher percentage of men and smokers, and they had higher values of BMI and FPG. Analysis of the implementation of the 1985 WHO recommendations revealed that there were 344 subjects with type 2 diabetes (29.13%) (Fig. 1). Of these subjects, 173 (50.29%) were diagnosed by OGTT. Forty-four subjects (3.73%) were misdiagnosed. Thus, in the group with a FPG between 7.8 and 11.0 mmol/l (n /132), 13 subjects (9.85%) were not recorded as having diabetes (false negative cases). In the group of subjects without an OGTT (n /457), 14 (3.06%) were classified in any of the 1985 WHO diagnostic categories without fulfilling the criteria (false positive cases). In the group of subjects who underwent an OGTT (n / 553), there were 12 false positive cases of diabetes, three false negative cases of IGT, and two false positive cases of IGT. This yielded an overall prevalence of misdiagnosis of 3.07%. The investigation of the presence of a confirmatory diagnostic test showed that in 132 subjects with a FPG between 7.8 and 11.0 mmol/l, FPG was reassessed in 92 cases (69.70%) in the first 6 months after the first FPG. Of the 553 subjects who had received an OGTT, 132 (23.87%) under-
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
126
Fig. 1. Implementation of the WHO criteria in 1181 subjects with a FPG /6.1 mmol/l, attending a primary care unit.
went a new test in the 6 months that followed the first OGTT. However, 457 subjects did not receive the recommended OGTT after an abnormal FPG. This latter group of subjects were older and had lower FPG levels as compared to the group of subjects who received an OGTT (Table 2). Each physician saw on average 800 patients/ month, except for one physician (number 5) who consistently had only 60% of this workload. Following the WHO criteria, the total number of subjects diagnosed with diabetes ranged from 29 to 69 and the proportion of subjects diagnosed with diabetes by an OGTT from 35.71 to 65.22% (Table 3). Similarly, we found a high variability in the proportion of subjects with a FPG between 6.1 and 7.7 mmol/l but without a subsequent OGTT (from 7.33 to 70.27%), in the percentage of
subjects with a FPG between 7.8 and 11.0 mmol/ l who had an FPG repeated (from 52.38 to 85.72%) and in the proportion of subjects who had an OGTT repeated (from 0.00 to 56.83%) (Table 3). Percentages of subjects misdiagnosed ranged from 1.16 to 6.34%. The percentage of subjects misdiagnosed was negatively correlated with the proportion of confirmatory OGTT repetitions (r / /0.73, P /0.04).
4. Discussion In relation to the 1985 WHO recommendations, several problems have been detected. First, this study confirms that physicians do not perform an adequate investigation of subjects with an FPG /
Table 1 Clinical characteristics of the 1181 subjects with an FPG/6.1 mmol/l, according to their FPG values
Age (years) Sex (n , % women) BMI (kg/m2) FPG (mmol/l) Smokers (n , %) Hypertension (n , %)
Subjects with a FPG/7.8 (n /171)
Subjects with a FPGB/7.8 (n /1010)
P
61.29/12.3 72 (42.1) 30.89/6.4 10.49/3.5 59 (35.5) 102 (59.7)
62.89/13.9 516 (51.1) 29.29/4.6 6.69/0.4 204 (20.2) 529 (52.4)
0.32 0.03 0.00 0.00 0.00 0.08
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
127
Table 2 Clinical characteristics of the 1010 subjects with an FPG between 6.1 and 7.7 mmol/l, with and without OGTT
Age (years) Sex (% women) BMI (kg/m2) FPG (mmol/l) Smokers (%) Hypertension (n , %)
Subjects with OGTT (n /553)
Subjects without OGTT (n /457)
P
61.49/13.1 286 (51.7) 29.19/4.6 6.79/0.4 111 (20.1) 296 (53.5)
64.49/14.6 230 (50.3) 29.49/4.6 6.59/6.4 93 (20.4) 233 (51.0)
0.00 0.66 0.40 0.00 0.91 0.42
6.1 mmol/l. Although 50% of the subjects were diagnosed with diabetes by an OGTT, a figure that compares favorably with previous data [12], almost 50% of subjects with a FPG value between 6.1 and 7.7 mmol/l who should have undergone an OGTT did not receive the test. These subjects had slightly lower FPG values and were older in relation to subjects who received the OGTT. This is a relevant finding since it has been shown that in older people a higher proportion of cases of diabetes are detected through an OGTT [13,14]. Our results show that phsysician’s attitudes towards the assessment of subjects with a FPG between 6.1 and 7.7 mmol/l are inadequate to detect diabetes and impaired glucose tolerance. These results are especially relevant, if we take into account that life-style and pharmacological interventions could prevent the development of diabetes in subjects with impaired glucose tolerance [15,16]. Few published studies have evaluated the physician’s attitudes towards screening and diagnosis of diabetes [17,18]. In a previous study evaluating screening for diabetes, it was shown that screening rates improved with an increasing number of risk factors for diabetes [17]. However, this study did not evaluate the physician’s attitudes towards screening, the data being obtained through a questionnaire administered to subjects included in the study. A more recent investigation concluded that factors associated with screening diabetes in a general population aged /45 years are: the number of visits to any health care provider and the presence of high cholesterol levels, or family history of diabetes [18]. In the present study, the glucose tolerance status of 3.73% of subjects was not properly diagnosed.
Although data on errors made in the diagnosis of diabetes are scarce, it has been previously described that about 10% of the subjects registered as having diabetes in a general practice do not have the disease [10]. In concordance with this finding, it should be noted that a remarkable proportion of subjects (about 10%) with a FPG between 7.8 and 11.0 mmol/l were not diagnosed as being diabetes by the physicians in our study. Additionally, the proportion of subjects misdiagnosed was similar in subjects with an FPG between 6.1 and 7.7 mmol/l regardless of the performance of an OGTT. It could seem likely that all these subjects are not explicitly informed of the diagnosis of diabetes. To some extent, this finding may help to explain why some subjects only know that they have ‘sugar’ [8] or a ‘little touch of sugar’ [9]. In a clinical setting, it is recommended that an asymptomatic individual with an abnormal OGTT should undergo a second confirmatory test before a final diagnosis of diabetes is made [1,2]. This study confirms previous research showing that in everyday practice the OGTT is not only performed infrequently, but it is commonly not even used to confirm suspected cases [19]. Thus, we found that additional confirmatory testing is only performed in one out of four initial OGTTs, and seven out of ten high FPG values (]/7.8 mmol/l). We are not aware of studies investigating strategies to improve the adherence to the recommendations on the diagnosis of diabetes. Information technologies could be useful tools for improving the diagnosis process, as recently suggested [20]. As subjects are randomly distributed among health care teams, it is assumed that the wide variation observed in practice patterns reflects different physicians’ attitudes towards the impor-
44 9 (6.25) 3 (2.05) 9 (6.34) 5 (5.95) 2 (1.16) 4 (2.48) 7 (3.98)
5 (3.23)
132 0 (0.00) 2 (1.56) 0 (0.00) 0 (0.00) 79 (56.83) 21 (19.63) 12 (15.19) 18 (29.51)
176 161 155 173 42 53 57 69 4 8 3 3 16 19 22 21 22 (52.38) 26 (49.06) 32 (56.14) 45 (65.22) 155 136 131 150 94 (60.65) 57 (41.91) 24 (18.32) 11 (7.33) 12 (75.00) 15 (78.95) 12 (54.55) 18 (85.72)
Subjects evaluated (n ) Subjects with newly-diagnosed DM (n ) With FPG/11.1 mmol/l (n ) With FPG 7.8 /11.0 mmol/l (n ) Through OGTT (n , %) Subjects with FPG 6.1 /7.7 mmol/l (n ) Without OGTT (n , %) Subjects with FPG 7.8 /11.0 mmol/l and FPG repetition (n , %) Subjects with FPG 6.1 /7.7 mmol/l, OGTT and OGTT repetition (n , %) Misdiagnosis (n , %)
8 7 6 5 4 3 2 1
Physician
Table 3 Data referring to the diagnosis of diabetes performed by the eight physicians evaluated during a 45-month period
84 142 146 144 1181 14 41 39 29 344 4 10 2 5 39 5 16 21 12 132 5 (35.71) 15 (36.59) 16 (41.03) 12 (41.38) 173 74 111 128 125 1010 52 (70.27) 52 (46.85) 85 (66.41) 82 (65.60) 457 3 (60.00) 13 (81.25) 11 (52.38) 8 (66.67) 92
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129 Group
128
tance of the diagnosis of diabetes. There is wide discordance among physicians in relation to the way they diagnose type 2 diabetes (by OGTT or FPG), in the rate of performance of additional confirmatory tests, and in the proportion of subjects misdiagnosed. In addition, the proportion of misdiagnosed subjects was negatively correlated to the percentage of subjects with a confirmatory OGTT. Thus, physicians who do not routinely repeat the OGTT show a poorer performance in the diagnosis of diabetes, and have a higher rate of misdiagnosed cases. Therefore, it is suggested that the repetition of a confirmatory OGTT may be regarded as an index of the quality of the diagnostic process of diabetes. Finally, our study has some obvious limitations. First, the analysis is limited to a relatively small area where a relatively high proportion of subjects (50.29%) were diagnosed through an OGTT. For instance, in the US most diagnoses of diabetes are made in symptomatic subjects [21], and the OGTT was previously used in the diagnosis of about 38.0% of the subjects [12]. Thus, some of the results obtained would be of limited value in areas where subjects with diabetes are mainly diagnosed through a FPG. Furthermore, it could also be argued that the data about physicians’ attitudes are difficult to interpret since part of the data were collected following the recommendations of the American Diabetes Association in 1997. However, the provisional revision of the WHO recommendations was issued after the collection of data for the present study [22]. Furthermore, the local family practice recommendations on the diagnosis of diabetes were only modified in January 2000. However, in Spain, there are no studies evaluating the adherence to the 1985 WHO recommendations or the 1997 American Diabetes Association recommendations. In conclusion, subjects with a FPG /6.1 mmol/l evaluated in a primary care setting were not accurately diagnosed according to the current WHO recommendations. A high proportion of these subjects did not receive an OGTT and were older with a more detrimental lipid profile than subjects who received an OGTT. The high between-physician variability in the implementation of the current diagnostic recommendations of
J.M. Gonza´lez-Clemente et al. / Diabetes Research and Clinical Practice 62 (2003) 123 /129
diabetes should be considered in order to significantly improve the detection of the disease.
Acknowledgements Without the cooperation of the CAP Les Corts and all its members (especially A. Anguita, J. Benavent, M. Borrell, F. Cano, C. Forcada, G. Martı´nez, J. Ortiz, M. Pinyol, J. Ricart) this study would have been impossible. Many thanks to them. We are also deeply indebted to M. P. Delgado (Hospital Casa Maternitat, Corporacio´ Sanita`ria Clı´nic). This work was partially supported by a grant from the ‘Fondo de Investigaciones Sanitarias (FIS)’: RGDM 03/212 (Instituto de Salud Carlos III, Ministerio de Sanidad y Consumo, Spain).
References [1] The expert committee on the diagnosis and classification of diabetes mellitus, Report of the expert committee on the diagnosis and classification of diabetes mellitus, Diabetes Care 20 (1997) 1183 /1197. [2] World Health Organization. Diabetes Mellitus: report of a WHO Study Group. Technical Report Series No. 727. Geneva: WHO, 1985. [3] M.M. Engelgau, R.E. Aubert, T.J. Thompson, W.H. Herman, Screening for NIDDM in non-pregnant adults, Diabetes Care 18 (1995) 1606 /1618. [4] M.I. Harris, Medical care for patients with diabetes, Ann. Intern. Med. 124 (1996) 117 /122. [5] M.I. Harris, Undiagnosed NIDDM: clinical an public health issues, Diabetes Care 16 (1993) 642 /652. [6] N.J. Wareham, S. O’Rahilly, The changing classification and diagnosis of diabetes, Br. Med. J. 317 (1998) 359 /360. [7] M.I. Harris, R.E.: Klein, T.A. Welborn, M.W. Knuiman, Onset of NIDDM occurs at least 4 /7 yr before clinical diagnosis, Diabetes Care 15 (1992) 815 /819. [8] J. Schorling, J.T. Saunders, Is ‘‘sugar’’ the same as diabetes?, Diabetes Care 23 (2000) 330 /334. [9] R.G. Hiss, The concept of diabetes translation. Addressing barriers to widespread adoption of new science into clinical care, Diabetes Care 24 (2001) 1293 /1296.
129
[10] P. Patchett, D. Roberts, Diabetic patients who do not have diabetes: investigation of register of diabetic patients in general practice, Br. Med. J. 308 (1994) 1225 /1226. [11] F. Vinicor, When is diabetes diabetes?, J. Am. Med. Assoc. 281 (1999) 1222 /1223. [12] M.I. Harris, Frequency of oral glucose tolerance testing in the U.S, Diabetes Care 18 (1995) 143 /145. [13] The Decode Study Group on behalf or the European Diabetes Epidemiology Group, Consequences of the new diagnostic criteria for diabetes in older men and women. DECODE Study (Diabetes Epidemiology: Collaborative Analysis of Diagnostic Criteria in Europe), Diabetes Care 22 (1999) 1667 /71. [14] The Decode Study Group on behalf or the European Diabetes Epidemiology Group, Will new diagnostic criteria for diabetes mellitus change phenotype of patients with diabetes? Reanalysis of European epidemiological data, Br Med J 317 (1998) 371 /37. [15] J. Tuomilehto, J. Lindstro¨m, J.G. Eriksson, et al., Prevention of type 2 diabetes mellitus by changes in lifestyle among subjects with impaired glucose tolerance, N. Engl. J. Med. 344 (2001) 1343 /1350. [16] Diabetes Prevention Program Research Group. Reduction in the incidence of type 2 diabetes mellitus with lifestyle intervention or metformin. N Engl J Med 2002; 346: 393 / 403. [17] C.C. Cowie, M.I. Harris, M.S. Eberhardt, Frequency and determinants of screening for diabetes in the U.S, Diabetes Care 17 (1994) 1158 /1163. [18] T.S. Harwell, J.G. Smilie, J.M. McDowall, S.D. Helderson, D. Gohdes, Diabetes screening practices among individuals aged 45 years and older, Diabetes Care 23 (2000) 125 /126. [19] R.P. Solk, T.J. Orchard, D.E. Grobbee, Why use the oral glucose tolerance test?, Diabetes Care 18 (1995) 1045 / 1049. [20] D.W. Bates, A.A. Gawande, Improving safety with information technology, N. Engl. J. Med. 348 (2003) 2526 /2534. [21] M.I. Harris, Definition and classifications of diabetes mellitus and the new criteria for diagnosis, in: D. LeRoith, S.I. Taylor, J.M. Olefsky (Eds.), Diabetes Mellitus. A Fundamental and Clinical Text, 2nd edition, Lippincott, Williams and Wilkins, Philadelphia, PA, 2000, pp. 326 / 334. [22] KGMM Alberti, PZ Zimmet for the WHO Consultation, Definition, diagnosis and classification of diabetes mellitus and its complications. Part 1: diagnosis and classification of diabetes mellitus. Provisional report of a WHO consultation, Diab Med 15 (1998) 539 /553.