Workplan

• Best Information Through Regional Outcomes (BIRO) has been an international collaboration between innovative partners involved in defining and applying common rules for the development of a Shared European Diabetes Information System (SEDIS) between 2005-2008. The project was submitted in 2004 and finally started in September 2005, with a planned duration of 36 months, later extended to 42.

• The workplan started from an accurate review of the best information available from research and policy, to identify target indicators and data required to populate European Diabetes Reports. The development of the system required the application of advanced statistical methods to work effectively even through the exchange of small amounts of aggregated data. This solution allowed partner institutions to contribute to the reporting system through the application of a common data model and the use of standard statistical routines that can be similarly applied to produce the basic components of a European system.  

• The development of the BIRO approach relied on research conducted at the end of the 1990s in the area of clinical epidemiology. It takes advantage from the statistical properties of multivariate models and meta-analytical methods. This means in practical terms that overall results from different areas can be obtained by running separate analyses in parallel and then averaging all partial results using appropriate mathematical formulas.

• In BIRO, a region is not intended as an administrative jurisdiction, but as a geographical area where information on the progression of diabetes and quality of care provided is collected and stored systematically and homogeneously. The workplan does not require altering any routine data collection occurring at the regional level. The core SEDIS works as an overarching decision support system connecting the different levels of the regional health system to favour  informed decision making and clinical governance. 

The implementation of data processing techniques has been based on:

• a population-based approach: epidemiologic and performance indicators taking into account results obtained in a specific catchment area will represent a main target of the project

• great cooperation among primary data collectors: data format will need to be widely applied 

• a knowledge integration design: the different components of the system will be connected and technology widely disseminated among partners

DATA MODEL

• The SEDIS data model is divided into two parts: a static part, related to data collection (hardly changes over time) and a more dynamic part, related to medical concepts (more susceptible to changes in medical knowledge). Isolating the dynamic part will reduce the burden of updating the software.

• The complete SEDIS data cycle is based on the application of two consecutive data processing steps. The fundamental aspect of the system is to ensure its basic functionalities for each register (“local SEDIS”). The model is then generalized through its repeated application across all registers, using an overall step that compiles all “partial” results into a global report. The basic technical characteristics of the model have been described elsewhere; here we present an outline of its realization in the context of diabetes.

• Statistical analysis and epidemiological modelling of a disease register require an in depth understanding of all aspects related to the characteristics stored in the database. The organization of biometric and socio-demographic information needs to rely on solid classification criteria, e.g. normal levels of glycated haemoglobin using different kits, or algorithms for the construction of an index of socio-economic status (SES). In these circumstances it is useful to keep all definitions stored in a data dictionary using a common format . If we include in such a “progressive diary” also more general clinical concepts, such as the list of tests recommended to patients with hypertension, over 65, with a high level of glycated haemoglobin (guidelines) , or a particular “severity score” (comorbidity index ), then the result would be a “concept and data dictionary” (CDD).

• The CDD in the context of a “local SEDIS” can be represented as a chain of steps logically intertwined (Fig. 1). The availability of a CDD is essential to compare different analysis, both geographically and longitudinally. The CDD is the evidence-based and first component in the model chain. It follows the definition of a minimum dataset and requires to be regularly updated. At the opposite end of the chain there is the final output of the system, i.e. a health system report. The content of the report is based on the initial specification of a template that influences the selection of data procedures and statistical methods (“database engine” and “statistical engine”). Engines operate on top of the local databases that are not directly accessible to other partners. The reports will result from the amalgamation of statistical “objects” (tables, parameters, graphs) that will be produced in turn through the sequential application of the engines.

• The definition of an overall model (global SEDIS) directly follows the local implementation (fig.2). Once the statistical objects are available for each register, these can be exchanged across the network using some secure format. The level of aggregation chosen for each object is a combination of: a) formal agreement – the parties will need to support the level of detail; b) legislation – privacy at all levels, including individuals and institutions, need to be preserved; c) practical limits – not too much data can be rapidly transmitted.

• These are conditions that will be tested in practice through the BIRO collaboration. This project will acknowledge independent values and judgment across partners. For instance, a partner may not agree on transmitting some kind of objects across the network, but some minimal requirements can be met and be supported by other partners. For this reason a legislative review will be conducted in close collaboration, so that each partner can reach an informed decision. Eventually, different procedures to summarize different types of objects will be applied. Once objects reach a central location (server), these can be submitted to global database and statistical engines (central engine) and will finally return a global health system report valid for the whole EU collaboration.

WORKPACKAGES

• As a result of the data model specifications, the workplan spreads across different WPs (fig 3) according to the following main categories:
  • coordination and project management
  • building the “Knowledge Repository”
  • developing the database and statistical engines
  • ensuring secure transmission and information exchange
  • dissemination of the results

Coordination and project management

• Activities related to the coordination and project management include WPs that run for the entire duration of the project, and include the WP1: “Coordination”, WP13:“Project Management”, and WP15:“Evaluation”.
  All partners will be involved in these WPs, with evaluation led by the University of Malta, with the supervision of the Scientific Advisory Board.

Building the knowledge repository

• The first and most fundamental step towards the creation of SEDIS is the progressive adoption of a series of concepts and standardized formats that will be included in a shared “knowledge repository”.

• A concept and data dictionary will result from the inclusion of criteria that can be derived from the scientific literature (best practice guidelines, eg. list of examinations required for the diagnosis and follow-up of the diabetes patient, etc). The dictionary will include also details relative to the characteristics included in the minimum dataset and all derived variables.

• WP2: “ Clinical Review” will conduct a semi-systematic review of best practice guidelines that will define target indicators for diabetes care in Europe; WP3: “Common Dataset” , through which a new format will be created to extract data from local registers; and WP4:“Data Dictionary” , through which the newly defined scheme will be shared across centres, in the form of XML meta-data. Identification of best formats for the minimum dataset and the content of the data dictionary will progressively advance through the active collaboration of participants to the WPs.

Developing the database and statistical engines

• Core database and statistical functionalities with provide SEDIS with the powerful data mining ability to screen and compile available data into highly informative reports. These operations aim to provide users with an opportunity to browse a range of pre-formatted epidemiological reports and analyses without having to wait long times to have them delivered by research professionals.

• WP6 will build the “Database Engine”. The database engine will be developed using SQL-compliant standards on top of the meta-data formats identified by the concept dictionary. Advanced statistical routines will be implemented within WP8: “Statistical Engine”. Because of the particular structure of the target databases, it is envisaged that use of mixed multilevel and marginal models as generalised estimating equations (GEE) will be implemented to take into account correlated observations. This feature is very valuable, since such models, although highly relevant for health policy, have never been structurally included in health information systems before. Because of the particular distributed plan of SEDIS, the statistical engine will include two components: the local analyser and the meta-analyser: while the former will run on top of each single register, the WP10: “Central Engine” will focus on developing the same component to produce results valid for the whole universe of participating registers.

• Database and statistical software will be produced using various open source scripting and programming languages, in particular Python, Java, MySQL/PostgreSQL, SAS and R. Software will be available as open source at the end of the project and eventually implemented using other common tools (MS products). Open source model development will be less costly and innovation will be faster. To maximise speed and operability of the system, all the programs will operate in batch mode (i.e. not interactive). However, reports will be highly interactive (browser-based).

Ensuring secure transmission and information exchange

• In parallel with the construction of the software there will be an evaluation of the privacy implications, which will later influence the development of the communication protocols. This will occur with WP5: “Privacy Impact Assessment” (PIA) [35], which will be conducted to identify the impact of SEDIS on privacy, to ensure its adherence to data protection legislation and regulations and to reduce evential adverse effects consequent to the introduction of new technologies.

• There is in fact an enormous concern for privacy in diabetes registers. Currently no specific legislation addresses disease registers, hence “the diabetes community needs guidance and a framework for safe practice, if the perceived benefits of these databases are to be realised”.

• The PIA process will ensure compliance with EU and international data protection laws, conventions, regulations and rules as a minimum acceptable practice. Its goals will be to identify best practice in privacy protection through the definition of guidelines arising from a comparative evaluation of different alternatives under study. The application of this method will guarantee that the deployment of SEDIS aligns with fundamental requirements, thus ensuring its operability. The WP will present the following practical advantages:

• will provide clear responsibility of the use of the proposed tools, involving proponents in the resolution, or at least in the mitigation of possible adverse privacy effects arising from the usage of the system
• will provide a cost-effective solution operating at an early stage of the study design, when changes to meet privacy concerns are still easy to implement, as opposed to resolution of privacy issues when systems are in fact operational and difficult to change.

• Communication software for data transmission and exchange will necessarily follow the specifications advised by the PIA. These steps will be implemented through WP9: “Communication Software”. To run properly, it is important that SEDIS could run on top of completely de-identified information across the network and as far as possible only using aggregated tables. Many statistical analyses are actually possible using this approach, as most epidemiological outputs, with the exception of the more sophisticated longitudinal multivariate models, do not require individual records.

• Some limits can also be imposed to the depth of analysis in highly disaggregated reports (including maps) to avoid that very sparse cells some way could hamper privacy protection of either individuals and/or health care centres. Whenever needed, individual records may only be collected using an encrypted list of fake identifiers, avoiding the use of detailed information for dates and locations.

• For many of these aspects it will be essential that the B.I.R.O. collaboration sets the proper limits, through standards regulated by reciprocal agreement, that will be in large part determined through the above two WPs.

• Dissemination of the results will be ensured through a set of WPs that will pursue the widest access to the outputs of the collaboration.

• A particular attention will be reserved to New Member States and Accessing Countries through WP12: “Technology Transfer”, that will be dedicated to spread the B.I.R.O. knowledge-base across end-usersin the target countries. A series of dedicated activities will involve partners of Romania, Malta and Cyprus, to test use of our products for their needs and the actual applicability of the software under different conditions from those of the major registers. The WP will validate the generalisability of the protocols developed and their effectiveness in real life situations.

• The actual content of health reports to be produced will be decided through WP7: “Report Templates”, which will identify the most important structures needed to deliver meaningful outputs for diabetes management. In doing this, it will be important that those characteristics that have been found more predictive by the specialized literature are adequately represented.

• Dissemination will be sought through a specialised interface realised under WP11: “Web Portal”, which is associated to the present project website organized for the purpose. The publication will include statistical outputs automatically generated on a regular basis, as well as a series of commented reports that will start to appear from the second year of the project.

• A specific plan under WP14: “Dissemination” will focus on the organization of seminars, publications and plenary sessions according to the highest standards for EU projects. A monograph presenting an analytical overview of the results of the project in a European perspective will be published.

REFERENCES

1. Piwernetz K. DIABCARE Quality Network in Europe--a model for quality management in chronic diseases, Int Clin Psychopharmacol 2001 Apr;16 Suppl 3:S5-13.
2. EUDIP Study Group, Establishment of indicators monitoring diabetes mellitus and its morbidity, EUDIP, 2002 Report
3. Boyle D, Cunningham S, Sullivan F and Morris A, on behalf of the Tayside Regional Diabetes Network, Technology Integration for the provision of population-based equitable patient care: the Tayside Regional Diabetes Network: a brief description, Diabetes Nutrition and Metabolism 2001;14(2), 100-103.
4. Bergrem H, Leivestad T, Diabetic nephropathy and end-stage renal failure: the Norwegian story, Adv Ren Replace Ther. 2001 Jan;8(1):4-12
5. Beck, P, Battlogg, K, Gfrerer, R, Trajanoski, Z, Wach, P, and Pieber, T. Internet based client server application for diabetes management, Diabetes, Nutrition & Metabolism, Clinical and Experimental 13(4), 245. 2001.
6. Massi Benedetti M, Orsini Federici M, PROMODR: Progressive Model for Diabetes Register, Diabetes Nutrition and Metabolism 2001;14(2), 96-100.
7. Azzopardi J, Fenech FF, Junoussov Z, Mazovetsky A, Olchanski V., A computerized health screening and follow-up system in diabetes mellitus, Diabet Med. 1995 Mar;12(3):271-6.
8. Pruna S, Georgescu M, Stanciu E, Dixon RM, Harris N, The Black Sea Tele-Diab System: development-implementation-clinical evaluation, Stud Health Technol Inform. 2000;77:656-60.
9. The Diabetes Control and Complications Trial Research Group, The effect of insulin treatment of diabetes on the development and progression of long-term complications in insulin-dependent diabetes mellitus, N Eng J Med, 1993; 329 (14): 970-986.
10. Intensive blood-glucose control with sulphonylureas or insulin compared with conventional treatment and risk of complications in patients with type 2 diabetes (UKPDS 33).UK Prospective Diabetes Study (UKPDS) Group, Lancet, 1998 Sep 12;352(9131):837-53.
11. Effect of intensive blood-glucose control with metformin on complications in overweight patients with type 2 diabetes (UKPDS 34). UK Prospective Diabetes Study (UKPDS) Group, Lancet, 1998 Sep 12;352(9131):854-65.
12. Tight blood pressure control and risk of macro-vascular and microvascular complications in type 2 diabetes: UKPDS UK Prospective Diabetes Study Group, BMJ, 1998 Sep 12;317(7160):703-13.
13. Massi-Benedetti M; CODE-2 Advisory Board, The cost of diabetes Type II in Europe: the CODE-2 Study, Diabetologia. 2002 Jul;45(7):S1-4. Epub 2002
14. Ellrodt G et al., Evidence-based Disease Management, JAMA, 1997; 278:1687-1692.
15. Roos NP, Black C, Roos LL, Frohlich N, DeCoster C, Mustard C, Brownell MD, Shanahan M, Fergusson P, Toll F, Carriere KC, Burchill C, Fransoo R, MacWilliam L, Bogdanovic B, Friesen D., Managing health services: how the Population Health Information System (POPULIS) works for policymakers, Med Care. 1999 Jun;37(6 Suppl):JS27-41
16. Vaughan NJA and Massi Benedetti M, A review of European experience with aggregated diabetes databases in the delivery of quality care to establish a future vision of their structure and their role, Diabetes, Nutrition and Metabolism 2001, 14 (2): 86-87.
17. Davis TM, Cull CA, Holman RR., Relationship between ethnicity and glycemic control, lipid profiles, and blood pressure during the first 9 years of type 2 diabetes: u.k. prospective diabetes study (ukpds 55), Diabetes Care. 2001 Jul;24(7):1167-74.
18. Stratton IM, Adler AI, Neil HA, Matthews DR, Manley SE, Cull CA, Hadden D, Turner RC, Holman RR., Association of glycaemia with macrovascular and microvascular complications of type 2 diabetes (UKPDS 35):prospective observational study, BMJ, 2000 Aug 12;321(7258):405-12.
19. Quality of life in type 2 diabetic patients is affected by complications but not by intensive policies to improve blood glucose or blood pressure control (UKPDS 37). U.K. Prospective Diabetes Study Group. Diabetes Care 1999 22: 1125-113
20. Meigs JB, Singer DE, Sullivan LM, Dukes KA, D'Agostino RB, Nathan DM, Wagner EH, Kaplan SH, Greenfield S., Metabolic control and prevalent cardiovascular disease in non-insulin-dependent diabetes mellitus (NIDDM): The NIDDM Patient Outcome Research Team, Am J Med, 1997;102(1):38-47.
21. Hayward RA, Manning WG, Kaplan SH, Wagner EH, Greenfield S., Starting insulin therapy in patients with type 2 diabetes: effectiveness, complications, and resource utilization, JAMA, 1997 Nov 26;278(20):1663-9.
22. Greenfield S, Rogers W, Mangotich M, Carney MF, Tarlov AR., Outcomes of patients with hypertension and non-insulin dependent diabetes mellitus treated by different systems and specialties. Results from the medical outcomes study, JAMA, 1995, 8;274(18):1436-44.
23. Carinci F, Rose W, Advantages of Distributed Statistical Processing in Health Information Systems: the H+MetaBase project, Submitted, Monash Institute of Health Services Research, Melbourne, Australia, 2003.
24. F.Carinci et al., RISS-H - Progetto Diabete - SDO di Pazienti Diabetici - Analisi stratificata per Ospedali - Regione Veneto 1999,
25. Stier DM, Greenfield S, Lubeck DP, Dukes KA, Flanders SC, Henning JM, Weir J, Kaplan SH., Quantifying comorbidity in a disease-specific cohort: adaptation of the total illness burden index to prostate cancer, Urology. 1999 Sep;54(3):424-9.
26. Sullivan LM, Dukes KA, Harris L, Dittus RS, Greenfield S, Kaplan SH., A comparison of various methods of collecting self-reported health outcomes data among low-income and minority patients, Med Care, 1995 Apr;33(4 Suppl):AS183-94.
27. Greenfield S, Sullivan L, Dukes KA, Silliman R, D'Agostino R, Kaplan SH., Development and testing of a new measure of case mix for use in office practice., Med Care, 1995 Apr;33(4 Suppl):AS47-55.
28. Greenfield S, Kaplan SH, Kahn R, Ninomiya J, Griffith JL., Profiling care provided by different groups of physicians: effects of patient case-mix (bias) and physician-level clustering on quality assessment results, Ann Intern Med 2002 Jan 15;136(2):111-21.
29. Charlson ME, Pompei P, Ales KL, MacKenzie CR, A new method of classifying prognostic comorbidity in longitudinal studies: development and validation, Journal of Chronic Diseases, 1987; 40:373.
30. Elixhauser et al., Comorbidity measures for use with administrative data, Medical Care,36,1,8-27.
31. Currie et al., Epidemiology and Costs of Acute Hospital Care for cerebrovascular disease in diabetic and non diabetic populations, Stroke,1997;28:1142-1146.
32. Krop et al, Predicting expenditures for Medicare beneficiaries with diabetes, Diabetes Care 1999; 22:1660-1666.
33. Hofer TP, Hayward RA, Greenfield S, Wagner EH, Kaplan SH, Manning WG, The unreliability of individual physician "report cards" for assessing the costs and quality of care of a chronic disease, JAMA, 1999 Jun 9;281(22):2098-105.
34. World Health Organisation (Europe) and International Diabetes Federation (Europe). Diabetes Care and research in Europe: the St.Vincent Declaration. Diabet Med 1990; 7:360.
35. Blair Stewart, Privacy Impact Assessment: some approaches, issues and examples, Assistant Commissioners Office of the Privacy Commissioner, New Zealand.
36. Vaughan N, Confidentiality and diabetes registers, Diabetes Nutrition and Metabolism 2001;14(2), 114-117.
37. Diabetes Atlas 2000, International Diabetes Federation, Brussels, Belgium 2000. 
38. Stevens A, Gabbay J, Needs assessment needs assessment, Health trends 1991; 23:20-3.
39. Wright J, Williams R, Wilkinson JR, Development and importance of health needs assessment, BMJ, 1998; 316:1310-3.
40. Feldstein MS, Effects of differences in hospital beds scarcity on type of use, BMJ 1964; ii: 562-5.
41. Kirkup B, Forster D, How will health needs be measured in districts? Implications of variations in hospital use, J Public Health Med 1990; 12:45-50.
42. Wilkin D, Patterns of referral: explaining variation; In: Ronald M, Coulter A eds. Hospital referrals. Oxford: Radcliffe Medical Press, 1994.
43. Roos NP, Black C, Roos LL, Frohlich N, DeCoster C, Mustard C, Brownell MD, Shanahan M, Fergusson, P, Toll F, Carriere KC, Burchill C, Fransoo R, MacWilliam L, Bogdanovic B, Friesen D., Managing health services: how the Population Health Information System (POPULIS) works for policymakers, Med Care. 1999 Jun;37(6 Suppl):JS27-41
44. National Committee on Quality Assurance, The state of managed care quality –1999, Washington DC: National Committee for Quality Assurance, 1999.
45. Mc Cullough DK, Price MJ, Hindmarsch M, Wagner EH, A population-based approach to disease management in a primary care setting: early results and lessons learned, Eff Clin Pract
46. Morris AD, Boyle DI, MacAlpine R, Emslie-Smith A, Jung RT, Newton RW, MacDonald TM, The diabetes audit and research in Tayside Scotland (DARTS) study: electronic record linkage to create a diabetes register, Br Med J, 1997 Aug 30;315(7107):524-8.
47. Kuhlthau K, Walker DK, Perrin JM, Bauman L, Gortmaker SL, Newacheck PW, Stein REK, Assessing managed care for children with chronic conditions, Health Affairs, 1998; 17 (4), 42-52.
48. Keever GW, Disease management on intranet saves $1.2 million, Health Management Technology, Nov 1998, p. 47.
49. New JP, Hollis S, Campbell F, McDowell D, Burns E, Dornan TL, Young RJ., Measuring clinical performance and outcomes from diabetes information systems: an observational study, Diabetologia. 2000, Jul;43(7):836-43.
50. Rice N, Leyland A, Multilevel models: applications to health data, J Health Serv Res Policy, 1996; 1(3), 154-164.
51. Daniels MJ, Gatsonis C, Hierarchical Polytomous Regression Models with Applications to Health Services Research, Statistics in Medicine, 1997; 16, 2311-2325.
52. Carinci F, Pellegrini F, Nicolucci A, Exploring adherence to clinical guidelines: application of regression tree analysis in diabetes, 4th International Conference on the Scientific Basis of Health Services Research, Sydney, 22-25 September 2001.
53. Nicolucci A, Carinci F, Ciampi A., Stratifying patients at risk of diabetic complications: an integrated look at clinical, socioeconomic, and care-related factors. SID-AMD Italian Study Group for the Implementation of the St. Vincent Declaration, Diabetes Care, 1998, Sep;21(9):1439-44.
54. Carinci F, Nicolucci A, Pellegrini F, Regression trees in health services and outcomes research:an application of the RECPAM approach using quality of care as a criterion, Technical Report, Monash Institute of Health Services Research, Monash University, 2001.
55. Carinci F, Corrado D, Dettorre A, Pellegrini F, A multilevel approach to health systems analysis using RISS (Reporting-by-Intranet Stat System), 4th International Conference on the Scientific Basis of Health Services Research, Sydney, 22-25 September 2001.
56. Carinci F, RISS Samples
57. Nicolucci A., Cavaliere D., Scorpiglione N., Carinci F., Capani F., Tognoni G., Massi Benedetti M., the Italian Study Group for the Implementation of the St. Vincent Declaration, A comprehensive assessment of the avoidability of long term complications of diabetes mellitus: a case-control study, Diabetes Care, 19, 9:927-933, 1996.
58. F.Carinci, Southern Population Health Information System (SPHIS).An integrated Population-based Data Warehouse
59. Churches T, Carinci F, Open source at the interface between policy and academia: towards evidence-based information systems, 4th International Conference on the Scientific Basis of Health Services Research, Sydney, 22-25 September 2001.
60. Pruna S., Dixon R., Harris N. Black Sea TeleDiab: Diabetes Computer System with Communication Technology for Black Sea Region. IEEE Transactions on Information Technology in Biomedicine, Vol. 3, No. 3, September 1998, pp. 193-196.
61. Pruna S., Herinean Radu and Pruna. A. Set up an Internet infrastructure, interconnecting clinicians and scientists in countries of the Black Sea area to promote WHO/EURO care quality programs. Fifth Conference of the European Society for Engineering and Medicine, ESEM 99, Barcelona May 30-June 2nd, 1999,293-294.