N A L A R T I C L E Attending the Diabetes Center Is Associated With Increased 5-Year Survival Probability of Diabetic Patients The Verona Diabetes Study GIUSEPPE VERLATO, MD MlCHELE MUGGEO, MD EN/.O BONORA, MD MAUR1ZIO CORBELLINL, MD FRANCO BRESSAN ROBERTO DE MARCO to verify whether, in a well-defined area, survival is significantly different in dia- betic patients attending diabetes centers with respect to diabetic patients exclu- sively consulting family physicians. OBJECTIVE — The aim of the present study is to compare the survival of patients attending diabetes centers with that of patients exclusively consulting family physicians. RESEARCH DESIGN AND METHODS— The study was carried out in the frame of the Verona Diabetes Study, a population-based survey of known diabetes prevalence with a subsequent 5-year mortality follow-up. A cohort of 7,488 diabetic patients were identified on 31 December 1986 from three different sources: a drug consumption database, family physicians, and diabetes centers (one for children and one for adults). As of 31 December 1986, 3,288 patients in the entire cohort exclusively consulted their own family physicians, while 4,200 patients also had periodic examinations at the diabetes centers. The life status of the diabetic cohort was ascertained on 31 December 1991. RESULTS — Compared with the nondiabetic population, diabetic patients seen only by fam- ily physicians had a standardized mortality ratio (SMR) of 1.62 (95% Cl 1.51-1.74), while patients attending both family physicians and diabetes centers showed an SMR of 1.44 (1.34- 1.54), the difference being statistically significant (P = 0.017). The 5-year survival probability, estimated by the Kaplan-Meier method, was 0.76 (0.75-0.78) in patients seen only by family physicians and 0.81 (0.80-0.82) in patients attending the diabetes centers. Multivariate analysis by Cox regression model showed that attending the diabetes centers was an independent pre- dictor of survival even after adjusting for sex, age, and therapy of diabetes. The relative risk of '5-year all-cause mortality amounted to 0.83 (0.75-0.92) in patients also attending the diabetes centers with respect to patients consulting only family physicians (P < 0.001). CONCLUSIONS — These data on patients' survival indicate that diabetes centers play a crucial role in diabetes care. As a corollary, an integration between primary-care physicians and diabetes centers is strongly recommended. D iabetes, like other chronic diseases, is particularly resource-consuming (1) and more than doubles costs for health care (2). To reduce the burden of medical care, primary care is currently fa- vored with respect to higher levels of care (3). For instance, in the U.K. (4) and The Netherlands (5), diabetes is mostly man- aged by general practitioners, and in the former East Germany, since the reunifica- tion, centralized diabetes care by hospi- tals has been markedly reduced (6). How- ever, little attention has been given to the effectiveness of managing diabetes in gen- eral practice compared with specialized centers (7). In particular, it has not yet been assessed whether diabetes centers are more effective than primary care phy- sicians in terms of patients 1 survival. The aim of the present study was From the Division of Metabolic Diseases (M.M., E.B., M.C.), the Division of Medical Statistics (G.V., R.D.M.), and the Institute of Statistics (F.B.), University of Verona, Verona, Italy. Address correspondence and reprint requests to Michele Muggeo, MD, Malattie del Ricambio, Ospedale Civile Maggiore, Piazzale Stefani, t, 37126 Verona, Italy. Received for publication 23 June 1995 and accepted in revised form 12 October 1995. SMR, standardized mortality ratio. RESEARCH DESIGN AND METHODS — The present study was conducted in the frame of the Verona Di- abetes Study, the design and methods of which have been described in detail else where (8,9). In this survey, a total of 7,488 diabetic patients residing in Verona, Italy, were identified on 31 December 198b. All-cause mortality in the prevalent co- hort was assessed in the subsequent ') years (1987-1991), during which \, ! Wi patients died (172 patients, correspond- ing to 2.3% of the initial cohort, were noi traced and for statistical purposes were considered alive at the end of follow-up). In Verona there are two hospital based diabetic centers, one for adults and one for people younger than 18 years old. As of 31 December 1986, 3,288 diabetic patients exclusively consulted family physicians, while 4,200 patients also reg- ularly attended one of the diabetes centers after primary-care physician referral. First referral to the diabetes centers was de- cided by family physicians according to their personal opinions, while subse- quent visits were scheduled by the physi- cians at the diabetes centers. In the period 1980-1991, the average number of visits at the diabetes centers per patient per year was 2.85. Drug-treated patients followed by family physicians had to see them at least ever)' 3 months to get prescriptions for diabetes medications. Statistical analysis included the unpaired (test, x l test, standardized mor- tality ratios (SMRs) (10) computed using the nondiabetic population of Verona as reference, univariate survival analysis by the Kaplan-Meier method, and multivari- ate survival analysis by Cox regression model (11), which included sex, age, treatment of diabetes, and attendance at DIABETES CARE, VOLUME 19, NUMBER 3, MARCH 1996 211