Familial insulin-dependent diabetes mellitus (IDDM) epidemiology: standardization of data for the DIAMOND Project. The WHO Multinational Project for Childhood Diabetes Group.

Familial insulin-dependent diabetes mellitus (IDDM) epidemiology: standardization of data for the DIAMOND Project* The WHO Multinational Project for Childhood Diabetes Group1 The WHO Multinational Project for Childhood Diabetes, known as the DIAMOND Project, has been responsible for establishing insulin-dependent diabetes mellitus (IDDM) registries and for carrying out diabetes incidence studies, descriptive epidemiological research, and analytical investigations which are being used to test specific hypotheses regarding the etiology of the disease. Standardized epidemiological data are being collected from countries around the world, permitting international comparisons between registries. Multinational studies are also beginning to investigate the potential genetic determinants of the disease, and are contributing to the development of familial IDDM epidemiology worldwide. The develop- ment of standards for data collection of IDDM family histories is important for multinational studies of IDDM recurrence risks in families, descriptive analyses of patterns of familial aggregation, and compara- tive analytical investigations of specific etiological determinants of IDDM in relatives. These activities are being implemented through the DIAMOND Project. Introduction Research by IDDM registries The establishment of standardized population-based registries has produced data which show that the incidence of insulin-dependent diabetes mellitus (IDDM) varies widely between racial groups and countries (1-3). Children living in Finland, for example, have more than a 30-fold increase in their risk of developing the disease compared with child- ren living in the Republic of Korea or Japan (Fig. 1). Racial variations in risk within a population have also been observed, with a higher IDDM incidence rate among Whites than Blacks or Hispanics living in the same geographically-defined area (Table 1). Besides reporting the global patterns of inci- dence, these IDDM registries have been utilized for evaluating the epidemiology of the disease (2-4). A comparative study in the USA (Allegheny County, PA) and Japan shows that despite a 20-fold differ- ence in IDDM incidence, the epidemiological pat- * This article was prepared by Dr J. S. Dorman, WHO Collabo- rating Center for Diabetes Registries and Training, Department of Epidemiology, Graduate School of Public Health, University of Pittsburgh, Pittsburgh, PA 15261, USA. Requests for reprints should be sent to this address. 1 The participating countries and members of the group were: Algeria (Oran), K. Bessaoud; Argentina (Buenos Aires), M. de Sereday, M. Marti; Austria (Vienna), E. Schober; Brazil (Sao Paulo), L. Franco, C. Negrato, E. Russo, M. Schmidt, M. Vivolo; Canada (Montreal), E. Colle, A. Schiffrin, J. Siemiatycki; and (Nova Scotia), M.H. Tan, C. Wornell; Chile (Santiago), E. Car- rasco, G. Lopez, M. Garcia de los Rios; Cuba (Havana), 0. Mateo de Acosta, 0. Diaz-Diaz, M. Vera, T. Norot, A. Uriate; Dominican Republic (Santo Domingo), A. Selman-Geara; Egypt (Cairo), N. Hashem, R. Shawki, S. Eid, S. Taha; Finland (Helsinki), A. Reunanen, J. Tuomilehto, E. Tuomilehto-Wolf; Germany (Frankfurt), B. Boehm, C. Rosak, K. Schmidt; Hungary (Pecs), G. Soltesz and the Hungarian Childhood Diabetes Epidemiology Study Group; India (Madras), A. Ramachandran, V. Mohan, C. Snehalatha, M. Viswanathan; and (New Delhi), N.P.S. Verma, A. Goel; Italy (Cagliari), S. Muntoni, P. Masile, M. Silvetti, M. Songini, P. Tronci; (Chieti), F. Chiarelli; (Milan), G. Chiumello, F. Meschi, E. Bognetti; (Pavia), M. Tenconi, G. DeVoti, M. Martinetti, R. Lorini, F. Severi; (Rome), P. Pozzilli, M. Boccuni, R. Buzzetti, Reprint No. 5235 L. Sebastiani, N. Visalli; and (Torino), G. Pagano, G. Bruno, F. Merletti, E. Pisu; Japan (Hokkaido), N. Matsuura, K. Fujieda, A. Okuno, K. Ooyanagi, K. Yano; and (Okinawa), G. Mimura, H. Futenma, S. Higa, K. Murakami, M. Nagayoshi; Kuwait (Safat), M. Khogali, N. Abdella, K. Gumaa, A. Shaltout; New Zealand (Auckland), R. Elliott; and (Christchurch), R. Scott; Norway (Oslo), G. Joner, 0. Sovik; Paraguay (Asuncion), J. Jimenez, C. Palacios, F. Canete, J. Vera, R. Almiron; Poland (Poznan), M. Rewers, P. Fichna, M. Walczak, M. Jozwiak, D. Woznicka; Romania (Bucharest), C. lonescu-Tirgoviste, D. Cheta; SpAin (Barcelona), A. Goday, C. Castell, R. Gomis, G. Lloveras; and (Madrid), M. Serrano-Rios; Sudan (Khartoum), A. Elamin, K. Eltayeb, M. Omer, K. Zein; Sweden (Stockholm), G. Dahlqvist; United Kingdom (Leicester, England), A. Burden, M. Boddington, M. Burden, S. Sheera; and (Belfast, Northern Ireland), C. Pat- terson, D. Hadden; USA (Alabama, Birmingham), J. Roseman, R. Acton, R. Go, L. Wagenknecht; (Colorado, Denver), R. Hamman, J. Kostraba; (Florida, Tampa), J. Malone, P. Leaverton, D. Schocken; (Georgia, Atlanta), T. Hodge; (Illinois, Chicago), R. Lipton; (Minnesota, Rochester), D. Ballard, P. Palumbo; (North Dakota, Grand Forks), J. Brosseau, D. Sumbureru; (Pennsylvania, Pittsburgh), J. Dorman, R. LaPorte, C. Moy, M. Trucco; and (Puerto Rico), T. Frazer de Llado; USSR (Estonia), T. Podar, B. Adojaan, 1. Kalits; (Lithuania), V. Grabauskas, A. Norkus, Z. Padaiga, R. Preiksa, B. Urbonaite; and (Novosibirsk), E. Shubnikov, L. Kalashnikova; and Yugoslavia (Ljubljana), C. Krzisnik. WHO Secretariat (Geneva, Switzerland), H. King. Bulletin of the World Health Organization, 69 (6): 767-777 (1991) C, World Health Organization 1991 767 WHO Multinational Project for Childhood Diabetes Group Fig. 1. Age-adjusted IDDM Incidence rates (per 100000 population). Source: reference 3. Sweden Norway USA: N. Dakota USA: Alabama. Whites New Zealand Netherlands I Poland USA: Alabama, Blacks Japan Incidence rate terns for the two populations are virtually identical (4), which suggests that the marked difference in IDDM incidence is due to variation in the preva- lence of common etiological risk factors, either genetic or environmental, in the population. Compa- rative descriptive epidemiological studies have led to hypotheses on the causes of the marked geographical and racial differences in IDDM incidence. It has recently been proposed that population variation in the prevalence of IDDM susceptibility genes, rather than differences in environmental factors, is the primary determinant of the worldwide patterns of disease (5). Table 1: Age-adjusted (0-14 years) Incidence of IDDM (per 100000 population) by ethnic group, USA" Ethnic group Registry White Black Oriental Hispanic Colorado 16.4 9.7 (1978-83) (15.-17.8)b (7.4-12.4) Pennsylvania, 16.2 11.8 Allegheny (14.1-18.4) (7.9-17.2) County (1978-83) Alabama, 16.9 4.4 Jefferson (13.4-21.4) (2.3-7.5) County (1979-83) California, 13.8 3.3 6.4 4.1 San Diego (9.8-18.9) (0.4-11.9) (1.3-18.7) (1.3-9.6) (1978-80) a Source: reference3. b Figures in parentheses are the 95%-confidence intervals for the incidence rates. The next step is to test these etiological hypoth- eses (6). This is a primary objective of the WHO Multinational Project for Childhood Diabetes (7), also known as the DIAMOND Project. The initi- ation of analytic epidemiological studies requires an understanding of the factors that contribute to the occurrence of disease within specific populations. Although the environmental determinants of IDDM remain unclear, much is known about genetic sus- ceptibilities to the disease. Genetic factors and IDDM Independent of the search for specific IDDM suscep- tibility markers, descriptive studies of the relatives of individuals with the disease demonstrated familial aggregation of IDDM (8-15), but attempts to define a specific mode of inheritance have been unsuc- cessful. It was concluded that an underlying suscepti- bility to the disease was inherited, and that in genetically predisposed individuals, the interaction between environmental and immunological factors led to the development ofIDDM (16-18). HLA (human leukocyte antigen) association studies (19, 20), as well as linkage studies in families (21-23), confirm that IDDM susceptibility genes are located within the HLA region of chromosome 6. With advances in molecular biology, it has become apparent that HLA-DQ locus polymorphisms, par- ticularly those at position 57 of the beta chain, play a functional role in determining IDDM susceptibility and are strongly associated with the disease (24, 25). Recent data suggest that the proposed hypothesis regarding the contribution of the DQ beta polymor- phisms to the worldwide patterns of IDDM inci- dence is correct, with a direct relationship between the frequency of this genetic marker and disease risk (5). In addition to the presence of IDDM suscepti- bility genes, potential environmental risk factors are likely to cluster in the families of affected individuals. Familial IDDM epidemiology is currently expanding beyond descriptive evaluations to include molecular assessments of potential genetic, environmental, and other epidemiological risk factors in first-degree rela- tives. These analytical studies permit the testing of specific hypotheses regarding the causes of IDDM clustering within families. When conducted in inci- dence cohorts using epidemiological methods, such investigations provide estimates of the absolute risk of the disease for relatives with specific host- environmental exposures (26). This illustrates the importance of a population-based approach for further research in familial IDDM epidemiology. The establishment of IDDM incidence registries in countries across the world, through the DIAMOND Project, provides an excellent founda- WHO Bulletin OMS. Vol 69 1991 I~~~~~~~~~~~~~~~~~~~~~=I Finland 3010 20 768 Familial Insulin-dependent diabetes mellitus epidemiology tion for the development of such research worldwide. Standardization is an essential component of inter- national epidemiological studies (1-3, 27), for without standards, accurate geographical and racial comparisons cannot be achieved. This paper dis- cusses the applications of IDDM family history information on a national and multinational level, describes the importance of standard family history data collection and the specific areas which require standardization, and presents recommendations for the implementation of these standards through the DIAMOND Project. Epidemiology and data collection Development of familial IDDM epidemiology Within countries, family history information will permit (1) estimation of IDDM recurrence risks in families, (2) assessments of the familial aggregation of the disease, (3) evaluations of the epidemiology of IDDM in family members, and (4) identification of probands and families for genetic marker studies. When standard family history data are obtained across populations, comparative descriptive and analytical studies of familial IDDM epidemiology can be conducted. This information will greatly con- tribute to our knowledge of host-environmental interactions in the etiology of IDDM, and will lead to a better understanding of the determinants of the worldwide distribution of the disease. IDDM recurrence risks in families. Table 2 shows that siblings, parents and children of a diabetic proband have higher IDDM risks than unrelated individuals in the general population, approximating 3% to 10% by the age of 30 years (10-13, 28, 29). The data pre- sented are not directly comparable, primarily Table 2: Risk of developing IDDM for relatives of an IDDM proband Relationship to Cumulative proband risk Reference Parents: Denmark 1.1% to age 35 yrs 10 USA 2.6% to age 40 yrs 12 Germany 2.2% to age 80 yrs 29 Siblings: Denmark 5.1% to age 20 yrs 10 USA 3.3% to age 20 yrs 12 USA 5.5% to age 40 yrs 11 Germany 6.9% to age 80 yrs 29 India 1.5% to age 20 yrs 13 Children: Denmark 2.8% to age 20 yrs 10 Germany 5.6% to age 80 yrs 29 because of methodological differences. Most of the study populations consisted of patients identified from hospitals or diabetes clinics (10-14, 28, 29). Owing to a potential referral bias, multiple-case fam- ilies were likely to be overrepresented in these cohorts. This may have increased the IDDM risk estimates for family members above those obtained from more representative groups. Furthermore, defi- nitions of diabetes among family members varied considerably. Several investigations did not dis- tinguish insulin-dependent from noninsulin- dependent diabetes mellitus (NIDDM), and others defined diabetes in terms of current therapy, but did not consider the age at onset of the disease. There were also inconsistencies in the methods employed for data collection. Sources included hospital and clinic records, personal interviews with the probands and/or family members, and self-report question- naires. Because standard data were not obtained in a uniform manner from representative populations, accurate between-population comparisons of IDDM risks to family members could not be made. Most of the studies in Table 2 represent Cauca- sian populations from areas with similar incidence rates. There is a paucity of information for other racial groups and from populations with a very high or very low disease incidence. Despite geographical differences in the prevalence of IDDM susceptibility genes (5), the worldwide variation in IDDM risk to relatives may be less dramatic than that observed for the general population. First-degree relatives from various nationalities are likely to be genetically more homogeneous and to share exposure to environ- mental risk factors more frequently than unrelated individuals in the general population. Thus, multi- national comparisons of IDDM risks to family members may reveal less variability across countries than the reported 50-fold difference in population incidence rates. It is only through the collection of standard family history data that accurate interna- tional comparisons in IDDM recurrence risks can be made. Familal aggregation of IDDM. One of the most intrigu- ing patterns of familial aggregation of IDDM is the sex difference in the prevalence of parental diabetes. Diabetic children from families with an IDDM parent (i.e., parent-offspring families) are signifi- cantly more likely to have an affected father than an IDDM mother (28, 29). Although this observation has been reported by a number of independent studies (Table 3), the mechanism by which paternal IDDM is more frequently transmitted remains poorly understood. Several possible explanations have been proposed, including an increase in sponta- neous abortion of susceptible fetuses by IDDM WHO Bulletin OMS. Vol 69 1991 769 WHO Multinational Project for Childhood Diabetes Group Table 3: Risk of developing IDDM for children of IDDM mothers and fathers Children of Cumulative IDDM parents risk Reference Father: Denmark 4.2% to age 20 yrs 28 USA 6.1% to age 20 yrs 28 Germany 33.4% to age 80 yrs 29 Mother: Denmark 1.8% to age 20 yrs 28 USA 1.3% to age 20 yrs 28 Germany 4.4% to age 80 yrs 29 mothers (28), maternal immunological tolerance of fetuses to autoantigens of the beta cells (30), an increase in the paternal transmission of HLA suscep- tibility genes (31, 32), and genomic imprinting (33- 35). Most of the information regarding sex differ- ences in parental IDDM has been retrospectively obtained from studies of newly diagnosed Caucasian children. Comparable data are needed to determine whether similar findings would be observed for non- Caucasian populations or in countries which vary in overall disease incidence. There have been few pro- spective studies of children born to individuals with IDDM (28, 30, 36). Such evaluations are required for precise risk estimates and accurate comparisons of the incidence of IDDM among children of diabetic fathers versus mothers. Hypotheses regarding the sex difference in parental IDDM can be tested by ana- lysing standardized family history data collected from IDDM registries extending back to the 1950s and 1960s. Epidemiology of IDDM In famiiy members. The epidemi- ology of IDDM in family members has been investi- gated in several populations. Characteristics including a higher birth order (37, 38) and older maternal age (38) have been shown to be associated with an increased IDDM risk among siblings. These descriptive studies have contributed to current hypotheses regarding possible host-environmental interactions in familial IDDM. To directly test these hypotheses, analytical family studies are now includ- ing determinations of specific genetic markers (i.e., HLA-DQ beta polymorphisms and/or HLA hap- lotype sharing), as well as exposure to potential environmental risk factors (i.e., breast-feeding, viruses, etc.) in first-degree relatives. When the inci- dence of IDDM among family members is also known, it is possible to estimate the absolute risk of disease for relatives with specific risk factors (26). With standardized family history data, analytical studies can be conducted across populations, provid- ing assessments of potential determinants of geo- graphical differences in IDDM recurrence risks. Temporal increases in incidence and 'epidemics' of IDDM have been reported in several European countries (39-41). However, it is not known whether similar temporal increases in risk occurred among family members. Temporal trends in the risk to rela- tives could be due to a greater clustering of environ- mental diabetogenic factors within families or possible host-environmental interactions. One may also predict that during periods of increasing popu- lation incidence, the interval between the onset of disease in the proband and other family members may be shorter than that observed during periods of more stable incidence rates. An evaluation of the temporal changes in familial IDDM incidence would provide important insights as to the nature of the environmental determinants of the disease and their interaction with host susceptibility. Identifying families for genetic marker studies. The col- lection of family history data is also important for identifying probands and informative simplex and multiplex families for serological and molecular genetic studies of the etiology of IDDM. The ration- ale and methods for standardizing international case-control and family studies of host susceptibility are important issues that will require careful con- sideration and planning to assure comparability across populations. Collection of standardized family history Information The development of standards for the collection and analyses of family history data must be similar to those developed for the standardized IDDM inci- dence registries (27). The core information should be minimal and easy to ascertain so that the standards are applicable in developed and developing countries across the world. To achieve these goals, standards are needed for: (1) definition of the family structure, (2) definition of core variables to determine the pre- sence of diabetes in all family members, (3) methods of family history data collection, and (4) data storage and analyses. Through the DIAMOND Project, it will be possible to implement such standards and develop familial IDDM epidemiology on a multi- national level. Determining the family structure. As with the evolution of population-based registry research, the develop- ment of familial IDDM epidemiology has its founda- tion in determining the number of family members with the disease (numerator) and evaluating the population of first-degree relatives at risk WHO Bulletin OMS. Vol 69 1991770 Familial Insulin-dependent diabetes mellitus epidemiology (denominator). Thus, it is necessary to obtain a full family pedigree to determine the structure of the nuclear family. This information is critical for accu- rate estimations of IDDM recurrence risks or descriptive analyses of familial IDDM aggregation. In some of the developing countries, it may not be possible to identify all individuals who are biologically-related to the proband. Geographical differences in cultural practices and family roles may lead to biased evaluations of family structure. Situ- ations such as divorce and death, including the occurrence of miscarriages and stillbirths, may also contribute to inaccuracies in defining the population of first-degree relatives at risk. Obtaining accurate family history data in developed countries may also be difficult and requires special skills on the part of the interviewer. Given the personal nature of these data, the respondent must feel relaxed and assured that all family history information will be completely confidential. This will facilitate an accurate assess- ment of the family structure. Such issues must be addressed in each population prior to beginning data collection. Definition of core varlables. One of the most impor- tant core variables for the assessment of diabetes in family members is the definition of the disease. A practical and accurate definition of IDDM was developed for standardized population-based inci- dence registries (1-3). For purposes of registering childhood IDDM cases (age, 0-14 years), it was agreed that the disease definition should be based on a confirmed diagnosis by a physician, and the patient should be: (i) aged less than 15 years at the disease's onset, (ii) a resident of the defined target population for the registry at the time of diagnosis, and (iii) on insulin at the time of hospital discharge (1-3, 27). This information could be accurately obtained from the medical records of patients, easily verified, and standardized in countries around the world. A similar approach is required for international studies of familial diabetes. Since laboratory or clini- cal facilities may not be available in all populations, the only feasible way of assessing the occurrence of diabetes in family members is to utilize a set of core variables, to be obtained by survey, from cases iden- tified from population-based registries. The core data required for evaluating familial IDDM are outlined in Table 4. These items must be obtained for each first-degree relative (i.e., parents, siblings and children) in the family and include: (1) sex, (2) living status, (3) date of birth, (4) if deceased, information concerning cause, date and place of death, and (5) diabetes status; in addition, (6) the month and year (or age) at diagnosis, and (7) the Table 4: Core family history data required for compara- tive genetic epidemiological research For all first-degree relatives: -Sex -Whether alive or dead -Date of birth -If deceased, the cause, date and place of death -Diabetes status For diabetic first-degree relatives: -Month/year (or age) at diagnosis -Month/year (or age) when continuous insulin treatment was started month and year (or age) when continuous insulin treatment was started should be obtained for all individuals with diabetes. By utilizing core variables, such as age at diagnosis and type of diabetes therapy, it is possible to assess the occurrence of familial IDDM by questionnaire. Although the age-at-onset for registering all childhood IDDM cases in a community was fixed at less than 15 years, we can consider a less conserva- tive definition to distinguish parental IDDM from NIDDM. For example, individuals who were under 35 years old at disease onset and were placed on insulin therapy at diagnosis are most likely to be insulin-dependent. To ensure the accuracy of this definition, these data will be verified with hospital and physician records, and then standardized across populations. The DIAMOND Project's international com- parisons of IDDM risks to family members focus exclusively on the first-degree relatives of probands. However, extended family history data on the occurrence of diabetes in maternal and paternal grandparents, aunts, uncles and cousins may also be obtained if more in-depth pedigree analyses are desired by individual centres. Methods of data collectlon. Once the standard core variables have been defined, family history informa- tion can be collected by survey, which is the only practical approach to make international compari- sons. By standardizing the core variables, as described above, it is possible to assess the occurrence of familial IDDM using questionnaires. A standardized form, which can be easily adapted, will facilitate the collection of family history core data for the DIAMOND Project. An example of such forms is given in the Annex. This instrument can be modified so that it is convenient for data col- lection by individual centres and for comparison of data between populations. There are several methods of administering the questionnaire to probands and/or family members WHO Bulletin OMS. Vol 69 1991 771 WHO Multinational Project for Childhood Diabetes Group Table 5: WHO Multinational Project for Childhood Diabetes: assessment of family history Information from 42 registries In 26 countries Method of data Information Method of data Information Registry collection obtained Registry collection obtained Argentina, Buenos Survey Aires Austria, Vienna Interview/medical records Brazil, Sao Paulo Survey Canada, Montreal Interview Chile, Santiago Interview/medical records Finland, Helsinki Survey/interview Egypt, Cairo Interview/medical records Germany, Frankfurt Interview/medical records Hungary, Pecs Survey India: Madras Interview New Delhi Interview Italy: Cagliari Medical records Milan Interview Pavia Interview Rome Medical records Japan: Hokkaido Survey Okinawa Interview/medical records Kuwait, Safat Interview New Zealand: Interview/medical Auckland records Christchurch Survey/interview medical records Norway, Oslo Medical records Most core data: sibs, parents Core data: sibs, parents Core data: sibs, parents, children Diabetes status: sibs, parents Core data: sibs, parents, children Core data: sibs, parents, children Core data: sibs, parents, children Most core data: sibs, parents Core data: sibs, parents, children Most core data: sibs, parents Core data: sibs, parents Most core data: sibs, parents Core data: sibs, parents, children Most core data: sibs, parents Core data: sibs, parents Core data: sibs, parents, children Core diabetes data: sibs, parents Core data: sibs, children Most core data: sibs, parents Core data: sibs, parents, children Core data: sibs, parents, children Diabetes status: sibs, parents Paraguay, Asuncion Poland, Poznan Romania, Bucharest Spain: Barcelona Madrid Sudan, Khartoum Sweden, Stockholm United Kingdom: Leicester, England Belfast, N. Ireland USA: Alabama, Birmingham Colorado, Denver Florida, Tampa Minnesota, Olmsted County North Dakota, Grand Forks Pennsylvania, Allegheny County Puerto Rico USSR: Estonia Lithuania Interview/medical records Interview/medical records Survey Medical records Survey/interview Interview Survey Survey/interview Medical records Interview Survey/interview Survey/medical records Survey/interview/ medical records Survey/interview Survey/interview Interview/medical records Survey Interview Novosibirsk Survey Yugoslavia, Ljubljana Survey Most core data: sibs, parents Core data: sibs, parents Most core data: sibs, parents Most core data: sibs, parents Core data: sibs, parents Core diabetes data: sibs, parents Diabetes status: sibs, parents Core data: sibs, parents, children Diabetes status: sibs, parents Diabetes status: sibs, parents Core data: sibs, parents Core data: sibs, parents, children Core data: sibs, parents, children Core diabetes data: sibs, parents Core data: sibs, parents, children Core data: sibs, parents Core diabetes data: sibs, parents Core data: sibs, parents Core data: sibs, parents, children Core data: sibs, parents, children which warrant consideration. The form shown in the Annex was designed so that it could be completed by IDDM cases who were aged 16 years and over, or by their parents or guardians if the proband was under 16 years. The form can be mailed to each registered case, completed at home and returned in a stamped envelope. This approach provides the participant and his/her family an opportunity and time to accu- rately recall or look for information regarding dates of birth, dates of diabetes onset, etc. concerning other family members. A disadvantage of this method of data collection is that many individuals will not respond or return the survey form without a reminder by the research centre. Thus, to obtain family history information from more than 80% of the registered cases, several reminders are generally required for approximately one-half of the target population. To improve the response rate, family history information can also be obtained by telephone inter- views with the proband or his/her first-degree rela- tive. This approach generally provides a high response rate, but the data may be less accurate owing to memory biases. A combination of these methods will facilitate complete data collection in developed countries with good mail and phone WHO Bulletin OMS. Vol 69 1991 Algeria, Oran Survey n72 Familial Insulin-dependent diabetes mellitus epidemiology systems. These approaches are also valuable for registries with retrospective case ascertainment. However, they may not be useful in developing countries or for prospective registries. Family history data may also be obtained from short personal interviews at the time of first hospital admission or during diabetes clinic appointments. This approach may be most useful in populations employing prospective case registration. Interviews are also an ideal method for data collection in developing countries where many individuals are not accessible by mail or phone. Using these approaches, family history informa- tion has already been successfully obtained from a number of the population-based IDDM incidence registries which are part of the DIAMOND Project (Table 5). The survey instruments varied across populations and the core information was not stan- dardized. Thus, it was not possible to accurately compare the existing data. However, the similarities in the approaches employed by various countries indicate that the use of standards for family history data collection is feasible for international collabo- rative research. Data storage and analyses. Irrespective of the method of obtaining family history information, the collected data for comparative analyses must be computerized in a standardized manner. In particular, the core data for each first-degree relative must be entered as a separate record. Identification numbers must be defined to distinguish between families (family identi- fication numbers), diabetic probands (case identifica- tion numbers), and individuals within families (member identification numbers) and must be included with the core data for that individual. An example of a numerical coding system which has been employed to identify specific relatives in family data sets is also given in the Annex. Standards for data storage will also facilitate the sharing of computer analyses programs designed to estimate the recurrence risks to relatives, etc. Such programs are being developed for the DIAMOND Project using the computer software dBASE IV. Standardized dBASE database management systems for family history data and analyses will be distrib- uted to all DIAMOND participating centres. These programs will permit the identification of single versus multiple IDDM case families, and distinguish between those with affected siblings and parents. Secondly, programs for the analyses of familial IDDM aggregation will be included. This will permit a description of unaffected and affected relatives, characterizing their age at onset of diabetes, sex, birth order, living status of relatives, etc. Thirdly, life-table analyses programs will estimate IDDM recurrence risks in siblings, parents and children. Finally, these programs will be linked to those being established for the DIAMOND incidence studies, providing comparisons between the IDDM inci- dence rates among relatives with the risk estimates for the general population. The distribution of these programs through the DIAMOND Project will ensure that family history data can be analysed in the same manner across populations. Conclusions The next generation of IDDM research is now beginning with the development of familial IDDM epidemiology on a national and multinational level. This article has focused on standardization of methods for data collection concerning IDDM family histories. The development of these standards will greatly facilitate international collaborative research in areas where IDDM incidence registries are established. Population-based epidemiological studies are essential to our understanding of the con- tribution of familial and genetic factors to the global patterns of IDDM incidence. In addition, they will permit investigators to develop hypotheses regarding possible environmental factors and even preventive strategies, which is the ultimate objective of the WHO Multinational Project for Childhood Dia- betes. Annex Standardization of core Information on diabetes family histories Although core variables (Table 4) from studies of the familial risk of diabetes across populations can be collected in any format, it is important for compara- tive studies that they accurately represent the status of all first-degree relatives in the family. Without complete data for the entire family, it is not possible to identify the number of non-diabetic relatives, which prohibits the calculation of the overall and age-specific familial risk of IDDM. The two exam- ples of survey forms for use in the DIAMOND Project to assess the occurrence of IDDM in parents, siblings, and children (see p. 774) may be translated or modified according to the specific needs of each registry. The data to be collected as core information for the entire nuclear family of the diabetic index case must include the following: sex; whether alive or deceased; date of birth; if deceased, the cause, date (or age) and place of death; diabetes status (diabetic by WHO criteria or non-diabetic); age diabetes diag- nosed; use of insulin for treatment of diabetes (yes or WHO Bulletin OMS. Vol 69 1991 773 INFORMATION ON FAMILY OF DIABETIC I.D. #: DATE: THIS FORM PERTAINS TO THE DIABETIC'S IMMEDIATE FAMILY. PLEASE COMPLETE ALL INFORMATION FOR THE MOTHER, FATHER, BROTHERS, AND SISTERS OF (BEGINNING WITH THE OLDEST AND ENDING WITH THE YOUNGEST). PLEASE INCLUDE IN THE LIST OF SIBLINGS, AS WELL AS HALF-BROTHERS AND SISTERS, ADOPTED BROTHERS AND SISTERS, AND BROTHERS AND SISTERS WHO ARE DECEASED. IF THERE ARE ANY ADDITIONAL SIBLINGS, PLEASE LIST THEM ON A SEPARATE SHEET. MONTH/YEAR CURRENT IF DECEASED, MONTH/YEAR (OR AGE) NAME AGE INDICATE YEAR (OR AGE) CONTINUOUS DATE OF (OR AGE AND CAUSE OF USE DIABETES INSULIN (PLEASE INCLUDE MARRIED NAMES) SEX BIRTH ALIVE? DECEASED) DEATH DIABETIC? INSULIN? DIAGNOSED TREATMENT MOTHER FATHER OLDEST SIBLING 2ND SIBLING 3RD SIBLING IF ANY OF THE INDIVIDUALS LISTED ABOVE ARE HALF- OR STEP- BROTHERS AND SISTERS, OR ADOPTED, PLEASE LIST THEIR NAMES AND SPECIFY THEIR RELATIONSHIP TO THE DIABETIC. FOR HALF- BROTHERS AND SISTERS, INDICATE WHETHER THE MOTHER OR THE FATHER IS THE COMMON PARENT. INFORMANT: SOURCE OF INFORMATION: PERSONAL INTERVIEW _ TELEPHONE INTERVIEW MAILED SURVEY INFORMATION ON FAMILY OF DIABETIC I.D. #: DATE: THIS FORM PERTAINS TO THE DIABETIC'S IMMEDIATE FAMILY. PLEASE COMPLETE ALL INFORMATION FOR THE SPOUSE AND CHILDREN OF _ _ (BEGINNING WITH THE OLDEST CHILD AND ENDING WITH THE YOUNGEST). PLEASE INCLUDE ANY ADOPTED CHILDREN, MISCARRIAGES, STILLBIRTHS, ABORTIONS, AND CHILDREN WHO MAY BE DECEASED. IF THERE ARE ANY ADDITIONAL CHILDREN, PLEASE LIST THEM ON A SEPARATE SHEET. MONTH/YEAR CURRENT IF DECEASED, MONTH/YEAR (OR AGE) NAME AGE INDICATE YEAR (OR AGE) CONTINUOUS DATE OF (OR AGE AND CAUSE OF USE DIABETES INSULIN (PLEASE INCLUDE MARRIED NAMES) SEX BIRTH ALIVE? DECEASED) DEATH DIABETIC? INSULIN? DIAGNOSEED TREATMENT SPOUSE OLDEST CHILD 2ND CHILD 3RD CHILD 4-- 4 - 4- 4 4- I4 ± 1 + - 4 4 4 I-I +i + + 4 4- 4 . IF ANY OF THE INDIVIDUALS LISTED ABOVE ARE ADOPTED, OR ARE STEP- CHILDREN, PLEASE LIST THEIR NAMES. INFORMANT: SOURCE OF INFORMATION: PERSONAL INTERVIEW _ TELEPHONE INTERVIEW MAILED SURVEY Familial insulln-dependent diabetes mellitus epidemiology no); and age when insulin was first used (for insulin- using diabetics). With the survey forms, the data are collected for the diabetic index case, his/her natural parents, and each sibling (beginning with the oldest, ending with the youngest and including deceased siblings). Whether the diabetic was an adopted child or if there are half-siblings or other adopted children in the family should also be noted. If information on a relative is not known, it is coded as missing. It is also helpful if the name of the person who provided the information (the informant) and the source of data (i.e., mailed survey, interview, etc.) are recorded. Similar data can also be collected for the child- ren of the diabetic index case (see box), the diabetic's spouse, or extended family members. Forms for this type of data collection may also need modification for use in more in-depth genetic or pedigree studies. Identification numbers for family history data. Depending on the relationship (parents, sibs, spouses, children), the following identification numbers are recommended: Relationship to Identification IDDM proband number Parents: Natural 0101 = mother, 0102 = father Step 0201 = mother, 0202 = father Adopting 0131 = mother, 0132 = father Sibs: Natural 0103 = oldest, then 0104, 0105, etc. including IDDM proband in sibship Half 0203 = oldest, then 0204, 0205, etc. Adopted 0133 = oldest, then 0134, 0135, etc. Spouses: For proband and 2101 = wife of 0103 natural siblings 2102 = husband of 0103 2201 = wife of 0104 2202 = husband of 0104, etc. Children: For probands and 2103, etc. = natural children of 0103 natural siblings 2133, etc. = adopted children of 0103. Acknowledgements This work was supported by National Institutes of Health Grants R01 DK24021 and R01 DK39125. Resume Epidemlologle du dlab6te insulino-dependant (DID) familial: Standardlsation des donnees pour le projet DIAMOND Le Projet multinational conduit par l'OMS pour l'etude du diabete juvenile et connu sous le nom de Projet DIAMOND est a l'origine de la creation de registres du diabete insulino-dependant (DID), de la realisation d'etudes sur l'incidence du diabete, de recherches en epid6miologie descrip- tive et d'investigations analytiques destinees a tester les hypotheses sur l'etiologie de la maladie. Des donnees epid'emiologiques standardisees sont recueillies un peu partout dans le monde, autorisant des comparaisons internationales des differents registres. Des etudes multinationales ont commence a rechercher les determinants genetiques eventuels de la maladie et contribuent elles aussi au developpement mondial de l'epidemiologie du diabete insulino-dependant familial. L'epidemiologie du DID familial repose tout d'abord sur l'identification des cas de DID dans les families et sur l'evaluation du nombre de per- sonnes a risque pour la survenue de la maladie dans ces familles. Apres recueil des antecedents familiaux de base pour un registre, il est possible de determiner le risque de DID pour les membres des familles habitant une region donnee et de le comparer a l'incidence dans la population genera- le. L'etude descriptive d'agregats familiaux et l'investigation analytique de l'epidemiologie du DID chez les membres d'une famille peuvent ega- lement etre realisees. Grace au projet DIAMOND, le risque de DID chez les membres d'une mime famille et le profil de l'agregation de la pathologie dans les families pourront etre compares d'un groupe ethnique et d'un pays a l'autre. Une etude multinationale des determinants eventuels de l'incidence du DID familial sera en outre possible. L'exactitude des comparaisons internationales ne peut etre obtenue que si le recueil et I'analyse des donnees de base sur les antec6dents familiaux sont standardises. La codification du recueil des antecedents familiaux de DID a une importance double: pour l1'tude multinationale des risques de recurrence du DID dans les familles et pour I'analyse com- paree de certains determinants etiologiques du DID familial. Ces activites sont mises en oeuvre dans le cadre du Projet DIAMOND. References 1. LaPorte, R.E. et al. Geographic differences in the risk of insulin-dependent diabetes mellitus: the impor- tance of registries. Diabetes care, 8 (suppl. 1): 101-107 (1985). 2. Rowers, M. et al. Trends in the prevalence and inci- dence of diabetes: insulin-dependent diabetes mel- litus in childhood. World health statistics quarterly, 41: 179-190 (1988). WHO Bulletin OMS. Vol 69 1991 775 WHO Multinational Project for Childhood Diabetes Group 3. Diabetes Epidemiology Research International Group. Geographic patterns of childhood insulin- dependent diabetes mellitus. Diabetes, 37: 1113-1119 (1988). 4. Tallma, N. et al. A comparison of the epidemiology of youth-onset insulin-dependent diabetes mellitus between Japan and the United States (Allegheny County, PA). Diabetes care, 8 (suppl. 1): 17-23 (1985). 5. Dorman, J.S. et al. Worldwide differences in the inci- dence of type 1 diabetes are associated with amino acid variation at position 57 of the HLA-DO beta chain. Proceedings of the National Academy of Sci- ences, USA, 87: 7370-7374 (1990). 6. Dorman, J.S. & LaPorte, R.E. IDDM epidemiology: next generation of research (Letter). Diabetes care, 13: 184-185 (1990). 7. WHO DIAMOND Project Group. The WHO Multi- national Project for Childhood Diabetes, DIAMOND: DIABETES MONDIALE. Diabetes, 13: 1062-1068 (1990). 8. Gamble, D.R. An epidemiologic study of childhood diabetes affecting two or more siblings. Diabetologia< 19: 341-344 (1980). 9. Jaworskl, M.A. et al. Insulin-dependent diabetes: a comparison of families with single and multiple affected siblings. Diabetologia, 19: 97-100 (1980). 10. Degnbol, B. & Green, A. Diabetes mellitus among first- and second-degree relatives of early onset dia- betics. Annals of human genetics, 42: 25-47 (1978). 11. Chern, M.M. et al. Empirical risk for insulin- dependent diabetes (IDD) in sibs: further definition of genetic heterogeneity. Diabetes, 31: 1115-1118 (1982). 12. Wagener, D.K. et al. The Pittsburgh study of insulin- dependent diabetes mellitus: risk for diabetes among relatives of IDDM. Diabetes, 31: 136-144 (1982). 13. Ramachandran, A. et al. Familial aggregation in type 1 (insulin-dependent) diabetes mellitus: a study from South India. Diabetic medicine, 7: 876-879 (1990). 14. Bessaoud, K. et al. Epidemiology of juvenile insulin- dependent diabetes mellitus in Algeria, Department of Oran. Revue d'epid6miologie et sante publique, 38: 91-99 (1990). 15. Cheta, D. et al. A study on the types of diabetes mel- litus in first-degree relatives of IDDM patients. Diabete et metabolisme, 16: 1 1-15 (1990). 16. Fleld, L.L. Insulin-dependent diabetes mellitus: a model for the study of multifactorial disorders. Amer- ican journal of human genetics, 43: 793-798 (1988). 17. Thomson, G. et al. Genetic heterogeneity, modes of inheritance and risk estimates for a joint study of Caucasians with insulin-dependent diabetes mellitus. American journal of human genetics, 43: 799-816 (1988). 18. Rotter, J.l. et al. Genetics of diabetes mellitus. In: Rifkin, H. & Porte, D., ed. Diabetes mellitus: theory and practice. New York, Elsevier, 1990, pp. 378-413. 19. Bertram, J. & Baur, M. Insulin-dependent diabetes mellitus. In: Albert, E.D. et al., ed. Histocompatibility testing 1984. Heidelberg, Springer-Verlag, 1984, pp. 348-368. 20. Wolf, E. et al. The genetic susceptibility to type I (insulin-dependent) diabetes: analyses of the HLA-DR associations. Diabetologia, 23: 224-230 (1983). 21. Green, J.R. & Woodrow, J.C. Sibling method for detecting HLA-linked genes in disease. Tissue anti- gens, 9: 31-35 (1977). 22. Barbosa, J. et al. Linkage analysis between the major histocompatibility system and insulin- dependent diabetes in families with patients in two consecutive generations. Journal of clinical investi- gation, 65: 592-601 (1980). 23. Green, A. et al. The genetic susceptibility to insulin- dependent diabetes mellitus: combined segregation and linkage analysis. Genetic epidemiology, 2: 1-15 (1985). 24. Todd, J.A. et al. HLA-DQ beta gene contributes to susceptibility and resistance to insulin-dependent diabetes mellitus. Nature, 329: 559-604 (1987). 25. Morel, P.A. et al. Aspartic acid at position 57 of the HLA-DQ beta chain protects against type 1 diabetes: a family study. Proceedings of the National Academy of Sciences, USA, 85: 8111-8115 (1988). 26. Dorman, J.S. et al. Familial insulin-dependent dia- betes mellitus (IDDM): an investigation of host- environmental interactions. American journal of human genetics, 47 (suppl.): A132 (1990). 27. Diabetes Epidemiology Research International Registry Group. Diabetes registries in Asia. Bulletin of the World Health Organization, 65: 897-903 (1987). 28. Warram, J.H. et al. Differences in risk of insulin- dependent diabetes in offspring of diabetic mothers and diabetic fathers. New England journal of medi- cine, 311: 149-152 (1984). 29. Tilill, H. et al. Age-corrected empirical genetic risk estimates for first-degree relatives of IDDM patients. Diabetes, 36: 93-99 (1987). 30. Warram, J.H. et al. Determinants of IDDM and peri- natal mortality in children of diabetic mothers. Dia- betes, 37: 1328-1334 (1988). 31. Kay, P.H. et al. Preferential paternal transmission of the diabetogenic supratype marked by HLA B18 BfF1 DR3. Journal of immunology, 12: 327-329 (1985). 32. Vadhelm, C.M. et al. Preferential transmission of dia- betic alleles within the HLA gene complex, New England journal of medicine, 315: 1314-1318 (1986). 33. Swain, J.L. et al. Parental legacy determines methyl- ation and expression of an autosomal transgene: a molecular mechanism for parental imprinting. Cell, 50: 719-727 (1986). 34. Hall, J.G. Genomic imprinting: review and relevance to human diseases. American journal of human genetics, 46: 857-873 (1990). 35. McCarthy, B.J. et al. Investigating genomic imprint- ing and susceptibility to insulin-dependent diabetes mellitus (IDDM): an epidemiologic approach. Genetic epidemiology, (in press). 36. Kobberllng, J. & Bruggeboes, B. Prevalence of dia- betes among children of insulin-dependent diabetic mothers. Diabetologia, 18: 459-462 (1980). 37. Deschamps, I. et al. Evaluation of recurrence risk in siblings of diabetic children: importance of age and 776 WHO Bulletin OMS. Vol 69 1991 Familial Insulin-dependent diabetes mellitus epidemiology birth order in relation to HLA genotypes. Diabetes research, 1: 125-130 (1984). 38. Cavender, D.E. et al. Multivariate analyses of the risk of insulin-dependent diabetes mellitus for siblings of insulin-dependent diabetic patients. American journal of epidemiology, 120: 315-327 (1984). 39. Blngley, P.J. & Gale, E.A.M. Rising incidence of IDDM in Europe. Diabetes care, 12: 289-295 (1989). 40. Akerblom, H.K. & Reunanen, A. The epidemiology of insulin-dependent diabetes mellitus (IDDM) in Finland and in northern Europe. Diabetes care, 8 (suppl. 1): 10-16(1985). 41. Rowers, M. et al. Apparent epidemic of insulin- dependent diabetes mellitus in midwestern Poland. Diabetes, 36: 106-113 (1987). WHO Bulletin OMS. Vol 69 1991 777