Hospitalization Among Diabetic Children and Adolescents and the General Population in Germany

Diabetic children and adolescents were consecutively recruited in pediatric departments of university and general hospitals in Germany. These centers are members of the German Working Group for Pediatric Diabetology. For the purpose of quality management, a computer-based program for the continuous documentation of treatment processes and outcomes of the diabetic patients was developed at the Division of Applied Information Technology at Ulm University (14), which is actually used by ∼80 centers. Of these departments, 61 decided to participate in the present analysis, representing ∼15% of all pediatric departments in Germany (13). The number of annually treated diabetic patients per clinic ranged from 20 to >200. Participating centers transmitted anonymous data from all of their diabetic patients, documented from the time of initiating treatment, for central validation and analysis. In 1997, 5,874 patients between 1 and 19 years of age were documented, representing about one-fourth of children and adolescents with diabetes of this age-group in Germany (15).

Information on the German population between 1 and 19 years of age was derived from the national statistic registry (National Office for Statistics). In 1997, 16,850,132 inhabitants in this age-group were registered in Germany.

Sex, date of birth, and date of diabetes diagnosis were available for all subjects. For all patients treated in the participating clinics, hospitalizations were prospectively documented; documentation included the dates of admission and discharge and the reason for admission (coded by the treating diabetologist in five categories: diabetes onset, metabolic control and education, severe hypoglycemia, acute hyperglycemia or diabetic ketoacidosis, and “other reason”). An inpatient hospital stay was considered if there was at least one overnight stay in the clinic. All hospital stays were included, except admission at diabetes onset.

The 1997 hospitalization data from the general population were obtained from the registry of national hospital statistics (Krankenhausdiagnosestatistik). In all German hospitals, each stay is registered with the number of hospital inpatient days, the discharge diagnosis (only one main diagnosis based on the Ninth Revision of the International Classification of Diseases), and data such as sex, date of birth, and residence of the patient. The register was estimated to be 99.7% complete (13,15). All hospital admissions were included.

Because the actual costs of hospital care were not available, cost of hospital care was calculated using the 1997 mean charges per hospital day in German hospitals (15), which was $281 (1 U.S. dollar = 1.98 Deutsche Marks [December 1999]). The charge included medical care, medication, and hospitality services.

Descriptive statistics were performed for all continuous and categorical variables. Differences were tested using t tests or χ² tests, where appropriate. The level of significance was 0.05.

Incidences of hospitalization after diabetes onset were estimated as rates per person-year in the diabetic cohort and the general population, assuming Poisson distribution (16). For the diabetic population, incidence rates were directly standardized, with respect to sex and age (age-groups 1 to <5, 5 to <10, 10 to <15, and 15 to <20 years of age), to the general German population. Furthermore, numbers of hospital days per person-year were estimated (crude and age- and sex-standardized).

To compare hospitalization in the diabetic sample after diabetes onset with the general population, standardized ratios of hospitalization incidences and numbers of hospital days per person-year were calculated using indirect standardization with respect to age and sex (17). Furthermore, age- and sex-specific ratios were estimated. CIs (95%) were determined using normal asymptotics. Incidences and rate ratios were compared using likelihood-ratio statistics.

The cost per person-year in diabetic subjects aged 1–19 years of age after diabetes onset was estimated by multiplying hospital days per person-year and the average cost of a hospital day in 1997 ($281). The total annual cost in the whole diabetic population aged 1–19 years in Germany after diabetes onset (without hospital stay at manifestation) was determined using age-specific model-based estimates of diabetes prevalence. In Germany, ∼0.1% (0.14%) of children and adolescents <20 years of age have been estimated to have diabetes (18). Finally, the total cost in the German diabetic population, including hospitalization at diabetes onset, was calculated, assuming that 3,000 subjects <20 years of age were newly diagnosed in 1997 (18) and that the mean length of hospital stay at diabetes onset was 13.8 days (14).

All analyses were performed using the SAS (SAS for Windows, release 6.12; SAS Institute, Cary, NC) or the STATA analysis systems (Stata Statistical Software, release 5.0; College Station, TX).

Among the 5,874 diabetic patients between 1 and 19 years of age who were treated in 1997, 52% were male. The mean age (± SD) was 12.2 ± 4.3 years, and the mean diabetes duration was 4.6 ± 4.4 years. In total, there were 4,782 person-years under observation in 1997.

In the diabetic cohort, there were 1,330 hospital admissions and 8,781 hospital days in 1997 (without hospitalization at diabetes onset). Mean hospital stay per admission was ∼7 days. Among diabetic subjects, 17.8% were hospitalized at least once in the observation period; 3.4% had more than one admission.

Of total admissions, 57% were due to participation in structured education or metabolic control programs, 12% were due to ketoacidosis, and 5% were due to severe hypoglycemia; 26% were attributed to other reasons, including admissions not related to diabetes.

The hospitalization rate [per person-year (95% CI)] in the diabetic cohort, standardized to the German population with respect to age and sex, was estimated to be 0.27 (0.25–0.29), and the standardized number of hospital days was estimated to be 1.80 (1.75–1.84). In the general population, hospitalization incidence was estimated to be 0.0948 (0.0946–0.0949), and the number of hospital days per person-year was 0.6416 (0.6412–0.6420). The incidences and numbers of hospital days per person-year in the diabetic population were significantly higher in girls than in boys (P = 0.047 and P < 0.001, respectively) and also significantly higher in pubescent children (10 to <15 years) than in older or younger children (P < 0.001)—except the hospital incidence in children aged 1 to <5 years (P = 0.077). Age- and sex-specific incidences and numbers of hospital days are presented in Tables 1 and 2.

The standardized hospitalization incidence ratio for diabetic subjects compared with the general population was estimated to be 3.1 (2.9–3.2), and the standardized ratio of the numbers of hospital days per person-year was estimated to be 2.8 (2.7–2.9). Ratios of hospitalization incidences and hospital days differed with age and sex, with a significantly higher ratio in girls compared with boys (P = 0.027, P < 0.001) and a significantly higher ratio during puberty (10–14 years) compared with older and younger subjects (P < 0.01). Age- and sex-specific incidence rate ratios and ratios of hospital days per person-year are presented in Tables 1 and 2.

The cost per person-year for hospital care in 1997 was estimated to be $506 in diabetic subjects after diabetes onset and $180 in the general population. The total annual cost for hospital care in the German diabetic population 1 to <20 years of age (prevalence 0.14%; ∼25,000 subjects and ∼44,000 hospital days) was $12.4 million, corresponding to 0.4% of the total annual cost for hospital care in the age-matched general population in 1997 (∼10.8 million hospital days).

Including hospitalization at diabetes onset (3,000 manifestations per year, mean hospital stay 13.8 days, resulting in $11.6 million), the annual cost per person-year for hospital care in the German diabetic population between 1 and 20 years of age amounted to $970. For the entire German diabetic population ages 1–19 years, the total cost was $24.0 million, which corresponded to 0.8% of the total costs for hospital care for children and adolescents of the same age-group in Germany in 1997.

The present study is the first to show that children and adolescents between 1 and <20 years of age in Germany had three times more hospital admissions and three times more hospital days per person-year compared with the general population. Estimating relative hospitalization in the diabetic population compared with the general population is useful because relative hospitalization risks are less sensitive to differences in general hospitalization rates between countries and calendar times. Ratios of hospital incidences and days in the present study were significantly higher during puberty (10 to <15 years) and in girls.

There are only a few other studies in Europe and the U.S. with respect to the hospitalization rates of diabetic children compared with the general population. In the present study, lower ratios of hospitalization rates and hospitalization days were found when compared with northern European countries in the 1980s. In Denmark, the number of hospital days was ∼7- and 14-fold increased in boys and girls, respectively, in diabetic versus nondiabetic subjects aged 0–14 years (19). The relative risks for hospitalization were ∼4.0–8.0 in boys and 4.0–5.0 in girls. In Finland, the number of hospital days in diabetic subjects <14 years of age with hospitalization after diabetes onset within 1 year was 10-fold increased compared with the general population (11).

The reasons for the higher relative hospitalization rates in the previous studies in European countries are unknown. A possible explanation is a general trend toward reduction of hospitalization in the past decade because of economic reasons, but also because of changes in diabetes care. For example, in the Netherlands, the mean length of a hospital stay for diabetic children and adolescents decreased from 25 to 11 days between 1980 and 1991 (20). Additionally, differences between health care systems may play a role.

Hospitalization incidences and lengths of hospital stay in the present study were higher than those reported in the U.S. and the U.K. Recently, ∼0.1–0.15 hospital admissions per person-year were reported in diabetic subjects <18 years of age in the U.S. (10). In the U.K. in the early 1990s, hospital incidences in diabetic children and adolescents in the first 4 years after diabetes onset were 0.19 per person-year (21). Nevertheless, in the U.S. and in the U.K., hospitalization in general is lower than that in northern European countries and, in particular, Germany, where much effort has been undertaken to further reduce hospitalization (22). However, the rate of hospitalization in the diabetic population in the present study was lower compared with those reported in Finland and Denmark, where hospital incidences were ∼0.3–0.4 per person-year (11,19).

A high proportion of hospitalization was related to participation in structured inpatient diabetes education program. Incidences of acute complications were low compared with other countries. Thus, the incidence of severe hypoglycemia leading to hospital admission was estimated to be 0.06 per person-year in the U.K. (21) and ∼0.02 in Wisconsin (10). In the present study, where 5% of admissions were due to severe hypoglycemia, it is roughly estimated to be 0.01 per person-year. It is conceivable that structured diabetes education results in fewer acute complications. This effect could be shown in several trials (23–25) and population-based studies (10). Nevertheless, diabetes education does not always require inpatient hospital care; it could also be performed in specialized outpatient settings (26). However, reimbursement for hospitals in Germany actually favors inpatient education programs, although there are increasing efforts to replace these programs with outpatient care.

The estimated cost for hospital care per person-year in the diabetic study population 1–19 years of age after onset was nearly three times higher than that in the general population. By applying this finding to the diabetic German population, the annual cost of hospital care for diabetic children and adolescents aged 1–19 years without hospitalization at diabetes onset would amount to ∼0.5% of the total annual costs of hospital care in the German population in this age-group in 1997; this figure would increase to ∼1% when factoring in hospital admissions at diabetes manifestation. Prevalence of diabetes was only ∼0.1% in this age-group; thus, there were overproportional costs in the diabetic population.

In Sweden, at the beginning of the 1990s, costs for hospital care of subjects with type 1 diabetes were $800 per year higher than those for the general population (2). In a Finnish study, the average annual cost for diabetes care in the first 2 years after onset in diabetic children was $2,800. Of this annual cost, 17% (∼$475) was attributable to hospital care (4,5). Comparisons between studies are difficult, because they differ largely with respect to study design and assessment methods. In general, comparison of costs between different health care systems is limited.

Some limitations of the present analysis have to be considered. We have not performed a population-based study but a multicenter investigation in pediatric departments that offer specialized diabetes care. Therefore, the results of the study may not be representative for all children and adolescents with diabetes in Germany. Unfortunately, there is no nationwide information available about diabetic children and adolescents with diabetes. However, there is good evidence that the study population is representative for the whole German diabetic population: with respect to age distribution, the study population was similar to the whole diabetic population aged 1–19 years in Germany. Furthermore, the relative risks for hospitalization in this study were comparable with the results of a population-based study in a geographically well-defined region in Germany where a cohort was followed for up to 2.5 years after diabetes onset (27,28).

The present study was based on continuous computer-based documentation in pediatric departments that take part in a quality management initiative of the German Working Group for Pediatric Diabetes. All patients treated in these centers were included, and it can be assumed that the hospitalization documentation is considered valid, because clinics also use the documentation for administrative purposes. However, underestimation of hospitalization because of underreporting may exist. Hospitalization data in the general population were derived from official statistics that were described to be almost complete (15).

We did not exclude admissions due to diabetes from the general population data. Thus, the denominator for the relative risk estimation may be overestimated, and the relative risk may be biased toward 1, thereby yielding a more conservative estimation. However, because of the low prevalence of diabetes in this age-group, the bias is assumed to be very small.

The cost of diabetes was only roughly estimated because only the cost of hospital care was included. Data on further direct costs (e.g., those for ambulatory care or insulin) are not currently available. However, the cost for hospital care is assumed to represent a large proportion of costs in pediatric diabetes care. Furthermore, because data about costs in pediatric diabetes care are limited, the results of the present study are expected to be important.

In conclusion, hospitalization in children and adolescents with diabetes in specialized pediatric diabetes care was substantially increased compared with the general population. Expenditures for hospital care in this population were high in relation to the low prevalence of diabetes in children and adolescents. Ongoing efforts are undertaken to reduce the high hospitalization rate among pediatric diabetic patients. A repetition of the present study could evaluate potential successes. Reduction of hospitalization rates and the cost of diabetes care are of rising importance because of increasing health care expenditures and incidence rates in diabetes during childhood in Germany and worldwide (2,29–32).

We are grateful to the following pediatric departments that contributed data to this study: St. Franziskushospital, Ahlen (G. Friedrich); Helios-Klinikum, Aue (K. Praedicow); Kinderklinik Zentralklinikum, Augsburg (M. Hauschild); Kinderklinik Lindenhof (H. Haberland) and Virchow Klinikum Uni, Berlin (O. Kordonouri, T. Danne); Kinderzentrum Gilead, Bielefeld (F. Janssen); Uni-Kinderklinik Bonn, Bonn (N. Albers); Kinderklinik St. Jürgenstrasse, Bremen (W. Marg): Kinderklinik Allgemeines Krankenhaus, Celle (J. Zimmermann); Städtische Kinderklinik Chemnitz, Chemnitz (A. Klinghammer); Kinderklinik St. Vincenz, Coesfeld (E. Lang); Uni-Kinderklinik Dresden, Dresden (A. Naeke); Uni-Kinderklinik Erlangen, Erlangen (C. Renner, I. Knerr); Kinderklinik Erfurt, Erfurt (A. Lemmer); Uni-Kinderklinik Freiburg, Freiburg (O. Schwab); Kinderklinik am Klinikum, Fürth (K. Freundl); Kinderklinik Marienhospital, Gelsenkirchen (M. Papsch); Uni-Kinderklinik Gieβen, Gieβen (A. Dost, R.W. Holl); Städtische Kinderklinik, Görlitz (G. Korth); Kinderklinik Allgemeines Krankenhaus, Hagen (U. Schimmel); Altonaer Kinderkrankenhaus (M. Bentfeld, A. Kohlhase) and Kinderkrankenhaus Wilhelmstift, Hamburg (R. Lepler, K.P. Otto); Märkische Kinderklinik Hamm (A. Otten) and Kinderklinik St. Elisabeth, Hamm (V. Mauck, L. Reinken); Kinderklinik Stadtkrankenhaus, Hanau (W. Krill); Kinderklinik auf der Bult (P. Hürter, K. Lange, W.v. Schütz) and Kinderklinik Medizinische Hochschule, Hannover (A. Thon); Luisenheilanstalt Uni-Kinderklinik Heidelberg, Heidelberg (J. Grulich-Henn); Städtische Kinderklinik, Hildesheim (D. Agena); Kinderklink Herford, Herford (L. Jourdan); Uni-Kinderklinik Saarland, Homburg (A. Seider); Uni-Kinderklinik Köln, Köln (E. Schönau, O. Sewi); Kinderklinik Park Schönfeld (F. Hottenbacher) and Städtische Kinderklinik, Kassel (M. Schebek); Kinderklinik Kemperhof, Koblenz (I. Weis); Uni-Kinderklinik Leipzig, Leipzig (U. Nietzschmann, W. Kiess); Klinik für Kinderheilkunde, Ludwigsburg (H. Haug); Kinderklinik St. Marien, Landshut (R. Herterich); Uni-Kinderklinik, Lübeck (B. Jakisch, V. Wagner); Kinderklinik Kreiskrankenhaus, Lüdenscheid (A. Dimde); Uni-Kinderklinik Magdeburg, Magdeburg (D. Wiemann); Uni-Kinderklinik Mannheim, Mannheim (B. Meyburg, M. Witsch); Uni-Kinderklinik Marburg, Marburg (V. Prietsch); Kinderklinik Minden, Minden (D. Krüger); Dr. von Haunersches Kinderspital, München (K. Raile, H.P. Schwarz); Kinderklinik Neukirchen, Kohlhof (T. Hild-Büttner, I. Staudter); Kinderklinik St. Elisabeth, Neuwied (N. Häffner); Kinderklinik Evangelisches Krankenhaus, Oberhausen (R. Bachran, P. Beyer); St. Vincenz-Kinderklinik, Paderborn (J. Wolf); Städtische Kinderklinik, Pforzheim (R. Schnarz); Kinderklinik St. Nikolaus, Ravenburg (U. Siegel, M. Zerfaβ); Kinderklinik St. Hedwig, Regensburg (H. Segerer); Kinderklinik Winterbergkliniken, Saarbrücken (W. Kirsch); Thüringenklinik Georgius Agricola, Saalfeld (A. Müller); Kinderklinik Stade, Stade (H. Kehrberg); Olgahospital, Stuttgart (W. Hecker); Kinderklinik Krankenanstalt der Borromäerinnen, Trier (K.H. Ludwig); Uni-Kinderklinik Ulm, Ulm (E. Heinze); Uni-Kinderklinik Wien, Wien (A. Schober, B. Rami); Kinderklinik DKD, Wiesbaden (D. Weitzel); and Kinderklinik Wuppertal, Wuppertal (M. Petrikowski, S. Wirth).