Diabetes registries in Asia*

Bulletin of the World Health Organization, 65 (6): 897-903 (1987) © World Health Organization 1987 Diabetes registries in Asia* DIABETES EPIDEMIOLOGY RESEARCH INTERNATIONAL REGISTRY GROUP1 The importance of diabetes registries and the current status of registries in Asia are described. Such registries are needed in order to understand the etiology and natural history ofthe disease. Moreover, registries provide important public health information to the community. Guidelines for registry development are presented, and the existing, as well as planned, registries in Asia are reviewed. The importance of diabetes registries has begun to be recognized by researchers and WHO. As a result, a day-long workshop that overviewed the importance of diabetes registries was held at the International Symposium on Epidemiology of Diabetes Mellitus, Bangkok, 27-29 November 1986. Topics discussed included standards for registries, difficulties of developing registries, diabetes registries in Asia, and the current status of diabetes registries worldwide. The workshop was sponsored by the Endocrine Society of Thailand, the International Diabetes Federation, and WHO. IMPORTANCE OF DIABETES REGISTRIES A critical problem in diabetes research, as well as in non-communicable disease research in general, has been the lack of population-based data. Hospital- based series provide only limited and perhaps biased data and use of these in diabetes research makes it difficult to determine the factors related to the devel- opment and natural history ofthe disease. Population- based data are therefore needed. The best method of obtaining population-based data is to develop registries of community health infor- mation. Such registries permit the evaluation of existing or proposed health care measures, and registry information can rapidly be communicated to the local area health authorities to assist in altering patterns of care. Disease registries form an important tool for assessing the clinical course of diseases and can lead to an understanding of their etiology and patho- genesis. Moreover, by facilitating identification of cases, disease registries can serve as a population source for genetic and immunological testing, the * Based on discussions that took place during a workshop held at the International Symposium on Epidemiology of Diabetes Melli- tus, Bangkok, November 1986. ' A list of members of the Registry Group is given on p. 902. results of which can be directly related to absolute risk. Evaluation of factors associated with a disease can suggest methods for reducing its incidence and even its eradication. Development of registries should perhaps be the first step in understanding the impact of a disease within populations over a period of time and also between populations. Registries are therefore important for understand- ing the etiology and complications of chronic diseases as well as for evaluation of medical care in popu- lations. For diabetes, in particular, insulin-dependent diabetes mellitus (IDDM), registries are perhaps more important than for other diseases, and a prime reason for developing them is that the disease is very costly both for patients and society. It is therefore essential to determine how frequently IDDM occurs and to monitor its societal cost, with the goal of primary prevention of the disease and secondary prevention of diabetic complications. Because of the extraordinary differences in the incidence of IDDM worldwide (Table 1), develop- ment of diabetes registries to study the geographical variability of the disease could lead to an under- standing of its etiology. For example, a child in Finland is over 20-times more likely to develop diabetes than a child in Japan, a difference that is greater than that for coronary heart disease or cancer. Moreover, these geographical variations indicate the existence of major risk factor(s) for developing diabetes, where the relative risk of this in Finland compared with that in Japan is greater than that for other IDDM risk factors, including HLA-DR3/4, HLA-DR3 /X, HLA-DR4/X, HLA-identicality or coxsackievirus infection. In terms of absolute risk, IDDM is rare in Japan with only 250 patients per year diagnosed under the age of 15 years. If children in Japan had the same risk of developing IDDM as those in Finland, there would be over 5500 new cases diagnosed annually, which would be a major burden on the health care system. 4846 -897- DIABETES EPIDEMIOLOGY RESEARCH INTERNATIONAL REGISTRY GROUP Table 1. Distribution of the incidence of insulin- dependent diabetes mellitus among children in various countries Incidence Age group Country (per 100 000) (years) Finland 28.6 0-14 Sweden 23.6 0-14 Scotland 21.7 0-18 Norway 17.6 0-14 USA 14.7 0-19 Denmark 13.7 0-14 Netherlands 11.0 0-19 New Zealand 10.4 0-19 Canada 9.7 0-14 England 7.7 0-15 France 4.7 0-15 Poland 4.2 0-16 Israel 4.2 0-20 Kuwait 4.0 0-14 Japan 1.3 0-15 In addition to monitoring its geographical vari- ability, it is essential to gauge the incidence of diabetes over time. Epidemics of IDDM have been identified. For example, in Poland the incidence has doubled in certain groups in less than 5 years (2), while in Finland a doubling in 10 years has been reported (3). These changes are greater than those for any other chronic disease. It is therefore important to determine epidemic periods since they may facilitate identification of the environmental agent responsible, and this could potentially lead to the prevention of such rapid increases in incidence. Maintenance of a dynamic registry would permit the study of the occurrence and associated risk factors as the inci- dence changes. IDDM registries are relatively easy to establish, standardize, and maintain because accurate case definition is much more straightforward than for other chronic complaints such as coronary heart disease and cancer. The major gains with regard to health care and scientific information as well as the small development cost therefore make registries ideally suited for IDDM. STANDARDS FOR REGISTRY DEVELOPMENT There are certain inherent principles for worldwide standardization of IDDM registries. Of critical importance is the need for registries to be based on an established, well-defined target population. Hospital- based series cannot be considered to be a registry without proper validation of selection criteria and coverage. Also, a common definition of IDDM must be used in different countries and over time. The currently established criteria for IDDM cases are: -diagnosis of diabetes; -treatment with insulin; -age at onset of diabetes at least 0-14 years; and -resident of the defined area/population at the time of diagnosis (1, 4, 12). Approaches to case ascertainment may vary de- pending on the health care system and can involve review of hospital or school records, and prospective monitoring of hospitals and insulin prescriptions or other sources. It is important to recognize that different approaches can be used to identify cases; however, a secondary source of ascertainment is also essential to validate data in the registry. In this way, by matching the cases in the secondary source with those in the primary, the proportion of cases can be determined and this provides relatively exact infor- mation on the degree of ascertainment. There is also a need to standardize the information to be collected for IDDM registries, and for this purpose a common core of information has been agreed upon (1). This information, which can be collected in all study populations, includes the patient's name, date of birth, sex, race, and place of residence at the time of diagnosis. Both the date of diagnosis and the date of first insulin administration should be established. It is likely that the international comparisons of registry data will primarily be of the date of first insulin administration, since this can readily be ascertained in all countries. If the registry is designed to examine incidence of IDDM, it is important that the core information should be kept simple. Additional information of relevance to a specific aspect under investigation can subsequently be obtained from medical records or individuals (4). However, it is essential not to sacrifice case ascertainment for the quantity of information. PROBLEMS OF REGISTRY DEVELOPMENT Several of the difficulties that are likely to be encountered in setting up an IDDM registry were dis- cussed at the meeting and are outlined below. Lack of cooperation by hospitals When a registry is started, it is important to obtain the cooperation of all the hospitals within a defined 898 DIABETES REGISTRIES IN ASIA geographical area. Failure to obtain the cooperation of even one hospital jeopardizes the complete ascer- tainment of cases. Following are some of the reasons, stated and unstated, for lack of cooperation: -fear of "stealing" patients; -checking of medical care; -cost; inconvenience; fear ofcriticism of diagnostic or treatment practice; and scientific "competition". Several approaches can be used to overcome such resistance by hospitals, depending on the country. The first is to discuss the project with the medical staff involved. Assurances should be made that the research is of a purely scientific nature and that its outcomes are likely to be beneficial. Estimating the number of records that need to be reviewed in each hospital lessens the concern about cost and in- convenience. If there are research workers at the hospital, they should be informed that they would also have access to the data on cases registered. Making them collaborators reduces jealousy or resistance due to fear that cases will be "stolen". Should this course of action be unsuccessful, an independent letter of support from outside the country (from either WHO or research workers in other countries) can be helpful. It is difficult for physicians or hospital administrators to refuse a reasonable request from abroad asking for their participation in an international effort. However, should this approach be ineffective, it may be possible to gain access to the medical records through the intervention of government officials. Correspondence from the Ministry of Health to the hospital concerned is sometimes successful in this respect, although the feasibility of this method varies from country to country. Cases can also be identified through contact with physicians; however, sometimes these physicians are not cooperative. An interesting approach has been used by Tuchinda et al. in Thailand (6): when all else failed they contacted the wives of the physicians, a tactic that proved successful. Often researchers have developed their registry through cooperation with hospitals or physicians, whereas other sources would have been potentially better and more easily accessible. For example, an easy and accurate way to identify all IDDM patients in a population would be to discover who is using insulin by examining prescription records. Another method, which is being used in Denmark (6), Israel (M. Karp, personal communication, 1987), and possibly in the Republic of Korea (K. W. Ko, per- sonal communication, 1987) is to review military conscription records. All IDDM patients are rejected at the time of draft registration; therefore, by reviewing the records of those rejected and scrutinizing death certificates for diabetes, all IDDM cases within a population can be accurately identified. IDDM reduces combat readiness, but to an unknown extent. Since quantification of the degree of this reduction would clearly be of interest to the military, it may be possible to obtain their permission to examine the medical records ofmembers ofthe armed services who develop the disease; such information would also provide data on the incidence of IDDM. Prospective dynamic registries eliminate some of the difficulties associated with examining past medical records. Such registries can be developed by having monthly contact with hospital or individual physicians who see IDDM cases. The introduction of ""postcard systems", whereby hospitals return monthly reports of any new cases to researchers, produces accurate surveillance data. In contrast, it is less efficient to have physicians or nurses in hospitals give notification of new cases as they are identified; however, the compilers of the registry should contact hospitals that fail to return a postcard. Once a new case is identified, the medical record can be reviewed. In this way, accurate incidence rates for IDDM can be obtained within a year of setting up the registry. Should problems occur in developing a registry, it is critical to discuss them with workers who have previously set up registries and who may already have experienced similar difficulties. Table 2. Status of registries in Asia insulin-dependent diabetes mellitus Country Area Period China Beijing Planned Israel Whole country 1965-80 Japan Hokkaido 1973 onwards Whole country 1975 onwards (central registry) Tokyo 1974 onwards Kuwait Whole country 1980-81 Philippines - Planned Republic of Korea Seoul Planned Thailand Whole country 1985 Brunei Darussalam' - Planned India' Madras Planned New Delhi Planned Sri Lanka' - Planned China (Province of Taiwan) Taipei Planned ' Registries planned as the result of the workshop. 899 DIABETES EPIDEMIOLOGY RESEARCH INTERNATIONAL REGISTRY GROUP CURRENT STATUS OF DIABETES REGISTRIES Table 2 gives information about the status ofIDDM registries in Asia. Following the workshop, new registries are being planned. Once they are developed, they should provide valuable information on the distribution of diabetes in the region. The participants at the workshop also surveyed the status of IDDM registries in their respective countries, as outlined below. Australia The Tasmania Diabetes Registry, which contains data on all insulin-using diabetics in Tasmania, is nearing completion. Cases have been identified through examination of hospital and physicians' records. The completeness of ascertainment has been determined by comparison with the total number of insulin prescriptions in Tasmania, and a complete description of how the registry was developed has been reported previously (7). India In India there are no population-based IDDM registries; however, there are several very large diabetes research centres that could provide data for population-based studies. It is intended to develop IDDM registries in Madras and New Delhi. As it may not be feasible in either of these areas to retro- spectively identify cases, a prospective surveillance of local hospitals and clinics is being planned. The main difficulty in such an approach will be to deter- mine the number of patients who die at the onset of IDDM but who are never diagnosed. Japan There are three primary IDDM registries in Japan. The Central Registry, consisting of data on young diabetics who are reimbursed health care costs, was set up in 1975 and has been continuing since. Although it covers all of Japan, physicians were slow to contribute data to the system. It is likely that information on less than 60% of IDDM cases in Japan is contained in the register. Special permission is needed to access the system. Also, a second population-based registry has been developed that examines a cohort of schoolchildren who have been continuously screened by urine testing (8). The third registry, which has been operating since 1973 on the island of Hokkaido, has been developed by identify- ing IDDM cases from hospitals and clinics (9). Republic ofKorea The Seoul IDDM registry is just being set up. It is based on data obtained by retrospective review of medical records for 1985 and 1986 in approximately 80 hospitals in Seoul as well as on prospective surveillance. Identification of IDDM cases among individuals rejected during conscription is also being planned. Information from the records of the national health care system, which currently covers about 60% of residents in Seoul, will be used as a secondary source of validation. New Zealand An IDDM registry has recently been developed in Christchurch. Cases were prospectively identified through physicians and hospitals from 1982 to 1986 (10). The average annual incidence of IDDM was 11 per 100 000, which is similar to that reported from Europe and North America. Data have also been produced from a small series of cases in Auckland; the reported incidence of 10.1 per 100 000 is very similar to that for Christchurch (11). Philippines Attempts have been made to develop a hospital- based registry in Quezon City, where there are 42 hospitals. However, difficulty has been experienced in obtaining access to medical records. Thailand Two registry projects have been developed in Thailand. In one, all the major diabetes centres in the country were contacted to identify newly onset IDDM cases. The provisional results indicate that the risk of IDDM in Thailand is apparently lower than that for any other country; however, the data in the registry need to be validated externally for completeness. In a second project, the prevalence of diabetes is being examined in very low and moderately low socioeconomic areas of Bangkok. Cases are being identified from a slum area and from an apartment house in a higher income area. Over 1500 individuals in each population are being evaluated. Other planned registries It is intended to develop registries in Brunei Darussalam, China (Province of Taiwan), and Sri Lanka. Current status of registries in other countries The current and planned IDDM registries in other parts of the world are shown in Table 3. If most of them are elaborated over the next few years we should be able to determine the true risk of developing malnutrition-related diabetes mellitus, and non- 900 DIABETES REGISTRIES IN ASIA 901 insulin-dependent diabetes mellitus for greater than 3% of the world's population.a In particular, infor- mation about the incidence in Asia is essential to understand the global patterns of the disease. FUTURE RESEARCH Once registries are established, the incidence of IDDM in different parts of the world can be compared. The first step in this process will be to compare the age-specific and age-adjusted rates in different countries and among racial groups. It will also be essential to monitor trends in the incidence of diabetes over time to identify pockets where epidemics of IDDM are occurring. International cooperation in the sharing of data is critical in this respect. Once the global pattern of diabetes has been defined, it is essential to analyse the information obtained. This could lead directly to an understanding of the etiology and prevention of the disease. For example, is the difference in incidence of IDDM between Finnish and Japanese children due to genetic or environmental factors, or both? How do these two factors interact to produce the disease in different populations? The first step in the identification of risk factors for diabetes will be to compare the epidemiology of the condition worldwide. Comparison of the differences in age at onset, seasonality, and sex ratio in high- and low-risk populations and different races should facilitate identification of the factors responsible for the enormous disparities in incidence rates of diabetes between countries. Subsequently, common protocols can be established for registries to assess the contribution of genetic and environmental factors in causing diabetes. Protocols should include the collection of blood samples for genetic marker testing, determination of viral titres, as well as questionnaires for assessing family histories of disease, diet, stress, etc. If carried out using a standardized method in more than one country, case- control studies designed to evaluate the risk factors a Malnutrition-related diabetes mellitus (MRDM) occurs in some tropical areas in the world (Latin America, Africa, and south- east Asia) where it seems to be associated with malnutrition. Two major forms are now recognized: fibro-calculus pancreatic diabetes (FCPD) associated with florid pancreatic calcification and pro- tein-deficient pancreatic diabetes (PDPD) without pancreatic calcification. MRDM appears as a major clinical subclass of diabetes ranking with IDDM and non-insulin-dependent diabetes mellitus. It has an heterogeneous clinical spectrum and affects young individuals (15-35 years). The most favoured theories about the etiology of MRDM implicate protein-calorie deficiency alone or combined with cassava consumption. Other toxins (in food?) are also thought to play a role in places where cassava is not consumed and MRDM is very frequently encountered. Studies on genetic and immuno- logical factors are badly needed. Table 3. Current and planned insulin-dependent diabetes mellitus registries in countries in Africa, Europe, the Americas, and Australasia Country Area Period Africa Kenya Nairobi Planned South Africa Durban Planned Europe Denmark Copenhagen 1970-74 Jutland 1970-76 Finland Whole country 1970 onwards France Whole country 1975 German Democratic Whole country 1960 onwards Republic Italy Lombardy 1981 -82 Netherlands Whole country 1978-80 Norway Whole country 1973 Poland Warsaw 1984 Wielkopolska 1970 onwards Scotland Whole country 1968-76 Tayside region 1980-83 Sweden Whole country 1970-75 Whole country 1977 onwards United Kingdom BDA Registrya 1973-81 Oxford 1985 onwards North America Canada Montreal 1971 onwards Toronto 1976-78 Prince Edward 1964-84 Island USA Colorado 1978 onwards Hawaii 1985 North Dakota 1967 onwards Pittsburgh 1965 onwards Rochester, MN 1945-80 Rhode Island 1979 onwards San Diego 1978-81 Wisconsin Planned Central and South America Brazil Sao Paulo Planned Cuba Whole country 1979-80 Ecuador - Planned Jamaica - Planned Mexico Mexico City 1985 Australasia Australia Tasmania 1984 onwards Sydney Planned New Zealand Whole country 1968-72 Auckland 1977-84 Canterbury 1982-86 BDA - British Diabetic Association. DIABETES EPIDEMIOLOGY RESEARCH INTERNATIONAL REGISTRY GROUP for diabetes within a population could also assess the contribution of these factors in causing the observed differences in incidence between countries. * * LIST OF PARTICIPANTS M. Ahuja, Department of Endocrinology and Met- abolism, All India Institute of Medical Sciences, New Delhi, India S. Bunnag, Department of Medicine, Chulalongkau University Hospital, Bangkok, Thailand N. Cho, Department of Epidemiology, University of Pittsburgh, Pittsburgh, PA, USA J. Dorman, Department of Epidemiology, University of Pittsburgh, Pittsburgh, PA, USA J. M. Ekoe, Epidemiology Unit, University of Mon- treal, Montreal, Quebec, Canada H. M. 0. King, Senior Research Officer (Polar Medicine), Antarctic Division, Department of Science, Kingston, Tasmania, Australia T. Kitagawa, Department of Pediatrics, Nihon Uni- versity School of Medicine, Chiyoda-Ku, Tokyo, Japan K. Ko, Seoul National University, Children's Hos- pital, Chongno-ku, Seoul, Korea R. LaPorte, Department of Epidemiology, Univer- sity of Pittsburgh, Pittsburgh, PA, USA (Rap- porteur) H. Lee, Seoul National University, School of Medi- cine, Chongno-ku, Seoul, Korea N. Matsuura, Hokkaido University School of Medi- cine, Department of Pediatrics, Kitaku, Sapporo, Japan V. Mohan, Diabetes Research Centre and M.V. Hospital for Diabetes, Royaparam, Madras, India T. Orchard, Department of Epidemiology, Univer- sity of Pittsburgh, Pittsburgh, PA, USA A. Panelo, University of the East, Ramon Magsay- say Memorial Medical Center, Quezon City, The Philippines S. Ramachandran, Diabetes Research Centre and M.V. Hospital for Diabetes, Royaparam, Madras, India M. Rewers, Department of Pediatrics, Medical Academy, Poznan, Poland R. Scott, Diabetes Lipid Disorders, Department of Medicine, Christchurch Clinical School of Medi- cine, Christchurch, New Zealand A. S. Sinnatamby, R.I.P.A.S. Hospital, Bandar Seri Begawan, Negara Brunei Darussalam C. Sitthi-Amorn, Department of Medicine, Chula- longkorn University, Bangkok, Thailand T. Tai, Department of Internal Medicine, National Taiwan University Hospital, Taipei, China (Pro- vince of Taiwan) N. Tajima, JIKEI University School of Medicine, Third Department of Internal Medicine, Minatoku, Tokyo, Japan S. Tsai, Metabolism Division, Veterans General Hospital, Taipei, China, (Province of Taiwan) C. Tuchinda, Faculty of Medicine Siriraj Hospital, Mahidol University, Bangkoknol, Bangkok, Thailand J. Tuomilehto, MONICA Project Data Center, Department of Epidemiology, National Public Health Institute, Helsinki, Finland N. Verma, Landy Hardinge Medical College & Associated Hospitals, New Delhi, India M. Wijesuriya, 102 Barnes Place, Colombo, Sri Lanka P. Zimmett, WHO Collaborating Centre for Epi- demiology of Diabetes Mellitus, Royal Southern Memorial Hospital, Caulfield South, Victoria, Australia ACKNOWLEDGEMENTS This research was supported, in part, by a grant from the National Institutes of Health (ROl DK35905). We thank the Endocrine Society of Thailand, the International Diabetes Federation and WHO for their help in organizing the meeting. RESUME REGISTRES DU DIABtTE EN ASIE Cet article decrit les rdsultats d'un atelier qui s'est tenu dans le cadre du symposium international sur l'epide- miologie du diabete sucr6, en novembre 1986 A Bangkok. Les participants a l'atelier ont examine l'importance des 902 DIABETES REGISTRIES IN ASIA 903 registres du diabOte et de 1'etablissement de normes pour ces registres, et se sont pench6s sur les problemes que pose 1'6laboration de tels registres et sur la situation mondiale en matiere de registres du diabete. Les registres ont et6 juges indispensables pour connaitre l'impact du diabete sur la sante. De plus, ils jouent un role important dans la determi- nation des facteurs lies a 1'etiologie et a 1'histoire naturelle de la maladie. Des normes ont recemment ete etablies afin de faciliter la comparaison des donn&es sur le diabete sucr6 dans differentes populations. Actuellement, le risque d'appari- tion d'un diabXte, notamment d'un diabete sucre insulino- d6pendant, n'est connu que pour moins de 3% de la popula- tion mondiale, avec peu de donn6es pour l'Asie. Au cours de l'atelier, six nouveaux registres ont et prevus pour l'Asie, et il sera possible, dans quelques annees, d'evaluer les diffe- rences g6ographiques du risque de diabete sur ce continent. REFERENCES 1. LAPORTE, R. E. ET AL. Geographic differences in the risk of insulin-dependent diabetes mellitus-the importance of registries. Diabetes care, 8 (Suppl. 1): 101-107 (1985). 2. REwERs, M. ET AL. An apparent "epidemic" of youth- onset insulin-dependent diabetes mellitus in mid- western Poland. Diabetes, 36: 106-113 (1987). 3. AKERBLOM, H. K. & REUNANEN, A. The epidemio- logy of insulin-dependent diabetes mellitus (IDDM) in Finland and northern Europe. Diabetes care, 8 (Suppl. 1): 10-16 (1985). 4. GREEN, A. ET AL. Workshop on diabetes registries: The role of IDDM registries in diabetes research and care. In: Serrano-Rios, M. & Lefebvre, P. J., ed. Diabetes 1985. Amsterdam, Elsevier, 1986, pp. 443-448. 5. TUCHNDA, C. ET AL. The epidemiology of diabetes mellitus in Thai children in 1984. In: Epidemiology of diabetes mellitus. Proceedings of the International Symposium on Epidemiology of Diabetes Mellitus. Bangkok, Crystal House Press, 1987, pp. 36-41. 6. GREEN, A. ET AL. Epidemiological studies of diabetes mellitus in Denmark: I. A case-finding method based on the National Service Conscript Registry. Dia- betologia, 19: 355-358 (1980). 7. KING, H. ET AL. The Tasmanian Insulin-Treated Diabetes Register: inception and progress in the first twelve months. Medical journal of Australia, 144: 414-416 (1986). 8. KITAGAWA, T. ET AL. The epidemiology of childhood diabetes mellitus in Japan. Clinico-genetic genesis of diabetes mellitus. In: Mimura, G. et al., ed. Inter- national Congress Series 597. Amsterdam, Excerpta Medica, 1982, pp. 51-57. 9. MATSUURA, N. ET AL. Epidemiologic survey of juvenile-onset insulin-dependent diabetes mellitus (IDDM) in Hokkaido, Japan: 1973-1981. Tohuku journal of experimental medicine, 141 (Suppl.): 181-189 (1983). 10. MASON, D. R. ET AL. Epidemiology of insulin-depen- dent diabetes mellitus in Canterbury, New Zealand. Diabetes research and clinical practice, 3: 4-29 (1987). 11. ELLIOT, R. B. & PILCHER, C. C. Child diabetes in the Auckland area. New Zealand medical journal, 98: 922-923 (1985). 12. WHO Technical Report Series, No. 727, 1985 (Dia- betes mellitus: report of a WHO Study Group).