- Email: [email protected]
Costs of cardiovascular events of diabetic patients in the French hospitals
&
Diabetes Metabolism
Diabetes & Metabolism 33 (2007) 310–313
http://france.elsevier.com/direct/diabet
Short Report
Costs of cardiovascular events of diabetic patients in the French hospitals X. Colina,*, A. Lafumaa, B. Gueronb b
a Cemka-Eval, 43, boulevard du Maréchal Joffre, 92340 Bourg-la-reine, France Pfizer, division scientifique et médicale, 23-25, avenue du Docteur-Lannelongue, F-75668 Paris cedex 14, France
Received 11 July 2006; accepted 19 December 2006 Available online 28 March 2007
Abstract Aims. – To test the assumption that hospital management of macro-vascular complications of Diabetes is more resource consuming in diabetic than in non-diabetic patients and to estimate, if relevant, the extra costs for diabetic patients. Methods. – The French national DRG database (PMSI- 2003) was analysed for the following cardio-vascular events (CVE): Stroke, Myocardial Infarction, Unstable Angina, Cardiac Arrest and Coronary Revascularization. Diabetic patients (Type 1 and 2) were identified using comorbidity diagnosis. Differences in various indicators of resource consumption were tested between diabetic and non-diabetic patients. Using length of stay (LOS) as a cost driver, the extra hospital costs of each CVE were then estimated by reference to mean costs measured on the whole database. Results. – Average LOS of patients with diabetes were significantly longer than of non diabetic patients. (stroke: +2.5 days, myocardial infarction: +1.5 days, unstable angina: +1.3 days, revascularisation: +2.8 days; P < 0.001). The mean numbers of medical procedures by stay were also higher in the diabetic group. Extra hospital costs of CVE for diabetic patients as compared with mean costs were the following: +23.9% (non fatal stroke), +10.4% (non fatal myocardial infarction), +6.1% (unstable angina), +9.1% (coronary revascularization). Conclusion. – The hospital costs of CVE in diabetic patients are higher than average. Specific costs for these complications should be used to improve the relevance of economic studies of Diabetes. © 2007 Elsevier Masson SAS. All rights reserved. Résumé Coûts hospitaliers des événements cardiovasculaires chez les patients diabétiques en France Objectifs. – Vérifier l’hypothèse selon laquelle la prise en charge hospitalière des complications macrovasculaires chez un patient diabétique est plus consommatrice de ressources que chez un patient non diabétique et estimer en cas de réponse positive les surcoûts correspondants. Méthodes. – Dans la base nationale PMSI 2003, nous avons analysé les séjours qui avaient respectivement pour motifs principaux les accidents vasculaires cérébraux (AVC), l’infarctus du myocarde, l’angor instable (IDM), et les revascularisations cardiaques. Les patients diabétiques ont été repérés lorsqu’un diabète (de type 1 ou 2) était notifié en diagnostic associé. Différents indicateurs de ressources consommées ont été comparés dans chaque groupe de patients et leurs différences testées. Les différences de durée de séjour par GHM (Groupes homogènes de malades) ont servi ensuite à valoriser les surcoûts correspondants pour chaque type de complications cardiovasculaires. Résultats. – Les durées moyennes de séjour des patients diabétiques étaient significativement plus élevées que chez les non-diabétiques (AVC : +2,5 jours, IDM : +1,5 jours, angor instable : +1,3 jours, revascularisation : +2,8 jours ; P < 0,001). Le nombre moyen d’actes par séjour était également plus élevé chez les diabétiques. Les surcoûts associés des évènements survenus chez les patients diabétiques par rapport aux coûts moyens étaient les suivants : +23,9 % pour un AVC non fatal, +10,4 % pour un IDM non fatal, +6,1 % pour un angor instable et +9,1 % pour une revascularisation. Conclusion. – Les coûts de prise en charge hospitalière des complications macrovasculaires chez les patients diabétiques sont plus élevés qu’en population générale. Des coûts spécifiques devraient être utilisés pour rendre plus pertinentes les évaluations économiques dans le champ du diabète. © 2007 Elsevier Masson SAS. All rights reserved.
* Corresponding
author. E-mail address: [email protected] (X. Colin).
1262-3636/$ - see front matter © 2007 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.diabet.2006.12.004
X. Colin et al. / Diabetes & Metabolism 33 (2007) 310–313
311
Keywords: Costs; Cardiovascular events; Hospital; Diabetes; France Mots clés : Coûts ; Événements cardiovasculaires ; Hôpital ; Diabète ; France
1. Introduction and objectives
2.2. Comparison of diabetic versus non diabetic patients
In the domain of diabetes care, public health policy analysis and cost-effectiveness evaluation require integrating the costs of major macro-vascular and micro-vascular complications. Most of the time, and in absence of specific data about diabetic patients, mean cost based upon general population estimates are used. The objectives of this study were, in a first step, to examine the relevance of the assumption that diabetic patients required more human and material resources than average in the medical management of the acute phase of these complications. In case of a positive response to this first question, a second objective was then to estimate the corresponding extra costs. The analysis was actually restricted to inpatient care in the French setting for 5 cardiovascular events (CVE): Stroke, Myocardial Infarction, Unstable Angina, Cardiac Arrest, Cardiac Revascularization.
A descriptive analysis of the total set of hospital stays extracted from the DRG database for the 5 types of complications was carried out comparing diabetic and non diabetic patients on these variables. Statistical significance of differences in these variables was tested.
2. Methods Two independent general sources of information are currently available about hospital activity and costs in France: ● the DRG hospital database (PMSI database [1]), covering both public and private sectors; ● the “National Cost Study” (“Etude Nationale de Coûts” [2]) which integrates the results of detailed accounting data on a national sample of French public hospitals. Individual data from this database are not accessible. Detailed items of costs per DRG are the only available data.
2.3. Validation of the results The consistency of our results with data from external sources (French registries) was examined. 2.4. Economic assessment of extra cost of diabetic patients The length of stay (LOS) was chosen as the most appropriate cost driver to allow a linkage between the two databases. For each CVE, a case mix of DRGs was identified (excluding stays less than 24 hours) for diabetic patients. The differences of these case mix with the corresponding mean values obtained on the total population were estimated. A daily cost (restricted to hostel costs) was then applied to these differences to get the extra cost of hospitalisation of diabetic patients. The final cost of a specific event was then calculated by weighting all the corrected costs previously mentioned. Our approach was conservative insofar as the daily cost used to calculate the extra cost of diabetic stays was excluding all direct medical consumptions of the patient. For private clinics, as the financial and accounting systems differed from the public sector, we used a DRG cost scale [3] built on a similar method.
2.1. Identification of cases in the DRG database
3. Results
The PMSI database contained in 2003 around 18 million records of hospital stays and provided detailed clinical information such as main diagnosis, co-morbidities and main procedures performed (i.e. surgery, imaging, etc). Each record was associated with a specific DRG but didn’t contain any information on costs. In this database, we first extracted the hospital stays for the selected CVE identified by their main diagnosis (ICD-10): Stroke (I60-I64 codes), Myocardial Infarction (I21 and I22 codes), Unstable Angina (I200 code), and Cardiac Arrest (I460, I461 code). In addition, Coronary Revascularizations were identified using specific DRG codes (DRG# 154,155,158,159). Patients with Diabetes Mellitus (Type 1 or 2) were then identified by using associated diagnosis. Two main items in the database were identified as cost drivers and/or indicators of intensity of care during hospital stay. These indicators were the total length of stay and the number of procedures performed.
Out of the 281,320 selected stays, 15.9% had diabetesrelated diagnosis. This percentage was ranging from 10.8% to 19% according to the type of CVE (Table 1). The mean age for all CVE was 68.9 years (± 14.1) and was generally higher for diabetic patients than for non diabetics. (+1.4 years for stroke, +2.2 years for myocardial infarction, Table 1 Number of stays according to the type of event and diabetic status
Stroke Myocardial infarction Unstable angina Non fatal cardiac arrest Fatal cardiac arrest Revascularization Total
Number of stays
Diabetic patients
93 754 75 274 59 487 1 834 825 50 146 281 320
11 808 12 306 10 823 237 89 9 531 44 794
Non diabetic patients 81 946 62 968 48 664 1 597 736 40 615 236 526
Diabetic patients rates 12.6% 16.3% 18.2% 12.9% 10.8% 19.0% 15.9%
312
X. Colin et al. / Diabetes & Metabolism 33 (2007) 310–313
Table 2 Length of stay (LOS) of diabetic patients versus non diabetic patients
Stroke Myocardial infarction Unstable angina Non fatal cardiac arrest Fatal cardiac arrest Revascularization
LOS of diabetic patients 15.2 8.2 6.0 13.5 6.6 16.3
LOS of non diabetic patients 12.8 6.7 4.7 11.8 5.5 13.5
Extra days
P (Student)
2.5 1.5 1.3 1.6 1.1 2.8
< 0.001 < 0.001 < 0.001 0.3630 0.3319 < 0.001
Table 3 Costs of events for diabetic patients Hospitalization costs for diabetic patients Non fatal Stroke € 5 703 Fatal Stroke € 6 749 Non fatal MI € 4 721 Fatal MI € 4 434 Unstable angina € 4 147 Non fatal cardiac arrest € 9 313 Revascularization € 11 679 (*) Difference in LOS not significant.
Mean hospitalization costs (all patients) € 4 603 € 5 808 € 4 276 € 4 094 € 3 909 € 7 061 € 10 705
% Variation
+23.9% +16.2% +10.4% +8.3% +6.1% +31.9% (*) +9.1%
+0.8 year for unstable angina, +5.2 for non fatal cardiac arrest, and +1.8 years for revascularization, P < 0.001 for each event). The sex ratio (men/women) for all CVE was 1.78 and the analysis of diabetes rates by sex showed that for all hospitalized patients, except in case of stroke, diabetic patients were more often women (18.8% of diabetes for women versus 15.2% for men in case of myocardial infarction; 23.1% versus 18.0% in case of revascularization; 20.7% versus 17.1% in case of angina; P < 0.001 for each event). Concerning the main indicators, the mean number of procedures notified during the hospital stay was higher for diabetic patients and additional numbers of procedures were ranging from +0.5 to +3.34 according the type of event. (stroke: +0.5; myocardial infarction: +0.8, unstable angina: +0.9, revascularization: +1.9, cardiac arrest: +3.3); P < 0.001 for each, except for cardiac arrest). The length of stay was also longer for diabetic patients. (+2.8 days for revascularizations, +2.5 days for stroke, and +1.5 days for myocardial infarction) (Table 2). The costs of hospitalizations calculated for diabetic patients after correction of DRG length of stay were ranging from € 4,147 for unstable angina to € 11,679 for revascularization. Compared with the mean costs of CVE in the whole population, the estimated extra costs for diabetic patients were ranging from +6.1% to +23.9% according to the event (Table 3). 4. Discussion The choice of length of stay to correct the DRG costs was based on the high correlation of this variable with the mean cost of hospitalization which implicitly supposes a relative proportionality between costs and length of stay. If this relation has been clearly established for some cost components (hostel cost for example), it is not true for all direct medical consump-
tions which are not evenly distributed over time, during the stay. This is especially true for procedures, which are often performed in the first days after admission. In this respect, we kept a conservative approach by excluding all direct consumptions in our correction of DRG. Conversely, our approach might probably underestimate our extra cost estimates. We did not identify any publications addressing the question of the exhaustiveness of diabetes record as co-morbidity in the DRG database. Therefore, we compared diabetes rates in our sample with results issued from studies based on existing French Registries (stroke and myocardial infarction). In the case of myocardial infarction, diabetes rates observed in the French studies [4–7] were ranging between 17% to 21% versus 16.3% in our database. The rates concerning strokes were derived from the Dijon register [8] and the Bongard et al. study [7]. They were respectively 14.0% and 16.8% versus 12.6% in the DRG database. These comparisons are only indicative and indirect due to the differences in the design of the studies. They might suggest a moderate under-reporting of diabetes, but did not modify significantly the results of our analysis. Our results provide mean extra hospital costs of CVE in the diabetic population. It has to be emphasised that these extra cost are not necessarily attributable to the presence of diabetes alone. Our approach was purely descriptive and our objective did not suppose to disaggregate the respective effect of age, sex, or any other confounding factors. Due to the limitations of our data, it was not possible to perform a full cost approach of the CVE, including for example the linkage with further events occurring in the same patients. For similar reasons, we could not envisage to integrate all ambulatory care costs associated with the CVE although such costs should be taken into account in a cost of illness approach as their share in the total cost of diabetes cannot be neglected [9]. 5. Conclusion According to the main indicators available in the French PMSI database, the management of a cardiovascular event in diabetic patients requires a larger amount of resources compared with non diabetic patients. The hospital costs of CVE for diabetic patients were estimated between €4,434 and €11,679 according to the event, which were substantially higher than the mean values issued from the whole set of cases. It should be important to use such modified values in cost-effectiveness studies to improve the relevance of cost effectiveness ratios especially in case of comparison of diabetic versus non-diabetic target groups [10,11]. References [1] Base de données PMSI MCO 2003 [CD-ROM]. Agence Technique de l’Information Hospitalière 2004. [2] Échelle nationale des coûts relatifs par GHM, année 2003 [Online]. Agence Technique de l’Information Hospitalière 2004 (www.atih.sante. fr).
X. Colin et al. / Diabetes & Metabolism 33 (2007) 310–313 [3]
[4]
[5]
[6]
[7]
Échelle de coûts du secteur privé. Bureau de l’Assurance Qualité et de l’Information Médico-Economique en Hospitalisation Privée (FHP), previously ADIMEHP 2000 (www.baqimehp.com). Cambou JP, Danchin N, Boutalbi Y, et al. Evolution of the management and outcomes of patients admitted for acute myocardial infarction in France from 1995 to 2000: data from the USIK 1995 and USIC 2000 nationwide registries. Ann Cardiol Angeiol (Paris) 2004;53:12–7. Vaur L, Danchin N, Hanania G, et al. Management and short-term outcome of diabetic patients hospitalized for acute myocardial infarction: results of a nationwide French survey. Diabetes Metab 2003;29:241–9. Grenier O, Cambou JP, Ferrieres J, et al. Baseline characteristics and management of patients less than 45 years of age hospitalized for acute coronary syndromes: results from the nationwide French PREVENIR 1 and PREVENIR 2 studies. Ann Cardiol Angeiol (Paris) 2002;51:15–9. Bongard V, Cambou JP, Coppe G, Jullien G, Vahanian A, Mas JL. Prevalence and management of patients with a prior history of atherothrom-
[8]
[9]
[10]
[11]
313
botic disease in primary care in France. Results of the ECLAT1 survey. Arch Mal Coeur Vaiss 2003;96:833–40. Lemesle M, Madinier G, Menassa M, Billiar T, Becker F, Giroud M. Incidence of transient ischemic attacks in Dijon, France. A 5-year community-based study. Neuroepidemiology 1998;17:74–9. Weill A, Chinaud F, Vallier N, et al. Fréquence et coûts des trente affections de longue durée de la région Midi-Pyrénées, année 2003. Revue Médicale de l’Assurance Maladie 2005;4:273–87. Jonsson B, Hansson L, Stalhammar NO. Health economics in the Hypertension Optimal Treatment (HOT) study: costs and cost-effectiveness of intensive blood pressure lowering and low-dose aspirin in patients with hypertension. J Intern Med 2003;253:472–80. Zwart-van Rijkom JE, van Hout BA. Cost-efficacy in interventional cardiology; results from the EPISTENT study. Evaluation of Platelet IIb/IIIa Inhibitor For Stenting Trial. Eur Heart J 2001;22:1476–84.
Diabetes Metabolism
Diabetes & Metabolism 33 (2007) 310–313
http://france.elsevier.com/direct/diabet
Short Report
Costs of cardiovascular events of diabetic patients in the French hospitals X. Colina,*, A. Lafumaa, B. Gueronb b
a Cemka-Eval, 43, boulevard du Maréchal Joffre, 92340 Bourg-la-reine, France Pfizer, division scientifique et médicale, 23-25, avenue du Docteur-Lannelongue, F-75668 Paris cedex 14, France
Received 11 July 2006; accepted 19 December 2006 Available online 28 March 2007
Abstract Aims. – To test the assumption that hospital management of macro-vascular complications of Diabetes is more resource consuming in diabetic than in non-diabetic patients and to estimate, if relevant, the extra costs for diabetic patients. Methods. – The French national DRG database (PMSI- 2003) was analysed for the following cardio-vascular events (CVE): Stroke, Myocardial Infarction, Unstable Angina, Cardiac Arrest and Coronary Revascularization. Diabetic patients (Type 1 and 2) were identified using comorbidity diagnosis. Differences in various indicators of resource consumption were tested between diabetic and non-diabetic patients. Using length of stay (LOS) as a cost driver, the extra hospital costs of each CVE were then estimated by reference to mean costs measured on the whole database. Results. – Average LOS of patients with diabetes were significantly longer than of non diabetic patients. (stroke: +2.5 days, myocardial infarction: +1.5 days, unstable angina: +1.3 days, revascularisation: +2.8 days; P < 0.001). The mean numbers of medical procedures by stay were also higher in the diabetic group. Extra hospital costs of CVE for diabetic patients as compared with mean costs were the following: +23.9% (non fatal stroke), +10.4% (non fatal myocardial infarction), +6.1% (unstable angina), +9.1% (coronary revascularization). Conclusion. – The hospital costs of CVE in diabetic patients are higher than average. Specific costs for these complications should be used to improve the relevance of economic studies of Diabetes. © 2007 Elsevier Masson SAS. All rights reserved. Résumé Coûts hospitaliers des événements cardiovasculaires chez les patients diabétiques en France Objectifs. – Vérifier l’hypothèse selon laquelle la prise en charge hospitalière des complications macrovasculaires chez un patient diabétique est plus consommatrice de ressources que chez un patient non diabétique et estimer en cas de réponse positive les surcoûts correspondants. Méthodes. – Dans la base nationale PMSI 2003, nous avons analysé les séjours qui avaient respectivement pour motifs principaux les accidents vasculaires cérébraux (AVC), l’infarctus du myocarde, l’angor instable (IDM), et les revascularisations cardiaques. Les patients diabétiques ont été repérés lorsqu’un diabète (de type 1 ou 2) était notifié en diagnostic associé. Différents indicateurs de ressources consommées ont été comparés dans chaque groupe de patients et leurs différences testées. Les différences de durée de séjour par GHM (Groupes homogènes de malades) ont servi ensuite à valoriser les surcoûts correspondants pour chaque type de complications cardiovasculaires. Résultats. – Les durées moyennes de séjour des patients diabétiques étaient significativement plus élevées que chez les non-diabétiques (AVC : +2,5 jours, IDM : +1,5 jours, angor instable : +1,3 jours, revascularisation : +2,8 jours ; P < 0,001). Le nombre moyen d’actes par séjour était également plus élevé chez les diabétiques. Les surcoûts associés des évènements survenus chez les patients diabétiques par rapport aux coûts moyens étaient les suivants : +23,9 % pour un AVC non fatal, +10,4 % pour un IDM non fatal, +6,1 % pour un angor instable et +9,1 % pour une revascularisation. Conclusion. – Les coûts de prise en charge hospitalière des complications macrovasculaires chez les patients diabétiques sont plus élevés qu’en population générale. Des coûts spécifiques devraient être utilisés pour rendre plus pertinentes les évaluations économiques dans le champ du diabète. © 2007 Elsevier Masson SAS. All rights reserved.
* Corresponding
author. E-mail address: [email protected] (X. Colin).
1262-3636/$ - see front matter © 2007 Elsevier Masson SAS. All rights reserved. doi:10.1016/j.diabet.2006.12.004
X. Colin et al. / Diabetes & Metabolism 33 (2007) 310–313
311
Keywords: Costs; Cardiovascular events; Hospital; Diabetes; France Mots clés : Coûts ; Événements cardiovasculaires ; Hôpital ; Diabète ; France
1. Introduction and objectives
2.2. Comparison of diabetic versus non diabetic patients
In the domain of diabetes care, public health policy analysis and cost-effectiveness evaluation require integrating the costs of major macro-vascular and micro-vascular complications. Most of the time, and in absence of specific data about diabetic patients, mean cost based upon general population estimates are used. The objectives of this study were, in a first step, to examine the relevance of the assumption that diabetic patients required more human and material resources than average in the medical management of the acute phase of these complications. In case of a positive response to this first question, a second objective was then to estimate the corresponding extra costs. The analysis was actually restricted to inpatient care in the French setting for 5 cardiovascular events (CVE): Stroke, Myocardial Infarction, Unstable Angina, Cardiac Arrest, Cardiac Revascularization.
A descriptive analysis of the total set of hospital stays extracted from the DRG database for the 5 types of complications was carried out comparing diabetic and non diabetic patients on these variables. Statistical significance of differences in these variables was tested.
2. Methods Two independent general sources of information are currently available about hospital activity and costs in France: ● the DRG hospital database (PMSI database [1]), covering both public and private sectors; ● the “National Cost Study” (“Etude Nationale de Coûts” [2]) which integrates the results of detailed accounting data on a national sample of French public hospitals. Individual data from this database are not accessible. Detailed items of costs per DRG are the only available data.
2.3. Validation of the results The consistency of our results with data from external sources (French registries) was examined. 2.4. Economic assessment of extra cost of diabetic patients The length of stay (LOS) was chosen as the most appropriate cost driver to allow a linkage between the two databases. For each CVE, a case mix of DRGs was identified (excluding stays less than 24 hours) for diabetic patients. The differences of these case mix with the corresponding mean values obtained on the total population were estimated. A daily cost (restricted to hostel costs) was then applied to these differences to get the extra cost of hospitalisation of diabetic patients. The final cost of a specific event was then calculated by weighting all the corrected costs previously mentioned. Our approach was conservative insofar as the daily cost used to calculate the extra cost of diabetic stays was excluding all direct medical consumptions of the patient. For private clinics, as the financial and accounting systems differed from the public sector, we used a DRG cost scale [3] built on a similar method.
2.1. Identification of cases in the DRG database
3. Results
The PMSI database contained in 2003 around 18 million records of hospital stays and provided detailed clinical information such as main diagnosis, co-morbidities and main procedures performed (i.e. surgery, imaging, etc). Each record was associated with a specific DRG but didn’t contain any information on costs. In this database, we first extracted the hospital stays for the selected CVE identified by their main diagnosis (ICD-10): Stroke (I60-I64 codes), Myocardial Infarction (I21 and I22 codes), Unstable Angina (I200 code), and Cardiac Arrest (I460, I461 code). In addition, Coronary Revascularizations were identified using specific DRG codes (DRG# 154,155,158,159). Patients with Diabetes Mellitus (Type 1 or 2) were then identified by using associated diagnosis. Two main items in the database were identified as cost drivers and/or indicators of intensity of care during hospital stay. These indicators were the total length of stay and the number of procedures performed.
Out of the 281,320 selected stays, 15.9% had diabetesrelated diagnosis. This percentage was ranging from 10.8% to 19% according to the type of CVE (Table 1). The mean age for all CVE was 68.9 years (± 14.1) and was generally higher for diabetic patients than for non diabetics. (+1.4 years for stroke, +2.2 years for myocardial infarction, Table 1 Number of stays according to the type of event and diabetic status
Stroke Myocardial infarction Unstable angina Non fatal cardiac arrest Fatal cardiac arrest Revascularization Total
Number of stays
Diabetic patients
93 754 75 274 59 487 1 834 825 50 146 281 320
11 808 12 306 10 823 237 89 9 531 44 794
Non diabetic patients 81 946 62 968 48 664 1 597 736 40 615 236 526
Diabetic patients rates 12.6% 16.3% 18.2% 12.9% 10.8% 19.0% 15.9%
312
X. Colin et al. / Diabetes & Metabolism 33 (2007) 310–313
Table 2 Length of stay (LOS) of diabetic patients versus non diabetic patients
Stroke Myocardial infarction Unstable angina Non fatal cardiac arrest Fatal cardiac arrest Revascularization
LOS of diabetic patients 15.2 8.2 6.0 13.5 6.6 16.3
LOS of non diabetic patients 12.8 6.7 4.7 11.8 5.5 13.5
Extra days
P (Student)
2.5 1.5 1.3 1.6 1.1 2.8
< 0.001 < 0.001 < 0.001 0.3630 0.3319 < 0.001
Table 3 Costs of events for diabetic patients Hospitalization costs for diabetic patients Non fatal Stroke € 5 703 Fatal Stroke € 6 749 Non fatal MI € 4 721 Fatal MI € 4 434 Unstable angina € 4 147 Non fatal cardiac arrest € 9 313 Revascularization € 11 679 (*) Difference in LOS not significant.
Mean hospitalization costs (all patients) € 4 603 € 5 808 € 4 276 € 4 094 € 3 909 € 7 061 € 10 705
% Variation
+23.9% +16.2% +10.4% +8.3% +6.1% +31.9% (*) +9.1%
+0.8 year for unstable angina, +5.2 for non fatal cardiac arrest, and +1.8 years for revascularization, P < 0.001 for each event). The sex ratio (men/women) for all CVE was 1.78 and the analysis of diabetes rates by sex showed that for all hospitalized patients, except in case of stroke, diabetic patients were more often women (18.8% of diabetes for women versus 15.2% for men in case of myocardial infarction; 23.1% versus 18.0% in case of revascularization; 20.7% versus 17.1% in case of angina; P < 0.001 for each event). Concerning the main indicators, the mean number of procedures notified during the hospital stay was higher for diabetic patients and additional numbers of procedures were ranging from +0.5 to +3.34 according the type of event. (stroke: +0.5; myocardial infarction: +0.8, unstable angina: +0.9, revascularization: +1.9, cardiac arrest: +3.3); P < 0.001 for each, except for cardiac arrest). The length of stay was also longer for diabetic patients. (+2.8 days for revascularizations, +2.5 days for stroke, and +1.5 days for myocardial infarction) (Table 2). The costs of hospitalizations calculated for diabetic patients after correction of DRG length of stay were ranging from € 4,147 for unstable angina to € 11,679 for revascularization. Compared with the mean costs of CVE in the whole population, the estimated extra costs for diabetic patients were ranging from +6.1% to +23.9% according to the event (Table 3). 4. Discussion The choice of length of stay to correct the DRG costs was based on the high correlation of this variable with the mean cost of hospitalization which implicitly supposes a relative proportionality between costs and length of stay. If this relation has been clearly established for some cost components (hostel cost for example), it is not true for all direct medical consump-
tions which are not evenly distributed over time, during the stay. This is especially true for procedures, which are often performed in the first days after admission. In this respect, we kept a conservative approach by excluding all direct consumptions in our correction of DRG. Conversely, our approach might probably underestimate our extra cost estimates. We did not identify any publications addressing the question of the exhaustiveness of diabetes record as co-morbidity in the DRG database. Therefore, we compared diabetes rates in our sample with results issued from studies based on existing French Registries (stroke and myocardial infarction). In the case of myocardial infarction, diabetes rates observed in the French studies [4–7] were ranging between 17% to 21% versus 16.3% in our database. The rates concerning strokes were derived from the Dijon register [8] and the Bongard et al. study [7]. They were respectively 14.0% and 16.8% versus 12.6% in the DRG database. These comparisons are only indicative and indirect due to the differences in the design of the studies. They might suggest a moderate under-reporting of diabetes, but did not modify significantly the results of our analysis. Our results provide mean extra hospital costs of CVE in the diabetic population. It has to be emphasised that these extra cost are not necessarily attributable to the presence of diabetes alone. Our approach was purely descriptive and our objective did not suppose to disaggregate the respective effect of age, sex, or any other confounding factors. Due to the limitations of our data, it was not possible to perform a full cost approach of the CVE, including for example the linkage with further events occurring in the same patients. For similar reasons, we could not envisage to integrate all ambulatory care costs associated with the CVE although such costs should be taken into account in a cost of illness approach as their share in the total cost of diabetes cannot be neglected [9]. 5. Conclusion According to the main indicators available in the French PMSI database, the management of a cardiovascular event in diabetic patients requires a larger amount of resources compared with non diabetic patients. The hospital costs of CVE for diabetic patients were estimated between €4,434 and €11,679 according to the event, which were substantially higher than the mean values issued from the whole set of cases. It should be important to use such modified values in cost-effectiveness studies to improve the relevance of cost effectiveness ratios especially in case of comparison of diabetic versus non-diabetic target groups [10,11]. References [1] Base de données PMSI MCO 2003 [CD-ROM]. Agence Technique de l’Information Hospitalière 2004. [2] Échelle nationale des coûts relatifs par GHM, année 2003 [Online]. Agence Technique de l’Information Hospitalière 2004 (www.atih.sante. fr).
X. Colin et al. / Diabetes & Metabolism 33 (2007) 310–313 [3]
[4]
[5]
[6]
[7]
Échelle de coûts du secteur privé. Bureau de l’Assurance Qualité et de l’Information Médico-Economique en Hospitalisation Privée (FHP), previously ADIMEHP 2000 (www.baqimehp.com). Cambou JP, Danchin N, Boutalbi Y, et al. Evolution of the management and outcomes of patients admitted for acute myocardial infarction in France from 1995 to 2000: data from the USIK 1995 and USIC 2000 nationwide registries. Ann Cardiol Angeiol (Paris) 2004;53:12–7. Vaur L, Danchin N, Hanania G, et al. Management and short-term outcome of diabetic patients hospitalized for acute myocardial infarction: results of a nationwide French survey. Diabetes Metab 2003;29:241–9. Grenier O, Cambou JP, Ferrieres J, et al. Baseline characteristics and management of patients less than 45 years of age hospitalized for acute coronary syndromes: results from the nationwide French PREVENIR 1 and PREVENIR 2 studies. Ann Cardiol Angeiol (Paris) 2002;51:15–9. Bongard V, Cambou JP, Coppe G, Jullien G, Vahanian A, Mas JL. Prevalence and management of patients with a prior history of atherothrom-
[8]
[9]
[10]
[11]
313
botic disease in primary care in France. Results of the ECLAT1 survey. Arch Mal Coeur Vaiss 2003;96:833–40. Lemesle M, Madinier G, Menassa M, Billiar T, Becker F, Giroud M. Incidence of transient ischemic attacks in Dijon, France. A 5-year community-based study. Neuroepidemiology 1998;17:74–9. Weill A, Chinaud F, Vallier N, et al. Fréquence et coûts des trente affections de longue durée de la région Midi-Pyrénées, année 2003. Revue Médicale de l’Assurance Maladie 2005;4:273–87. Jonsson B, Hansson L, Stalhammar NO. Health economics in the Hypertension Optimal Treatment (HOT) study: costs and cost-effectiveness of intensive blood pressure lowering and low-dose aspirin in patients with hypertension. J Intern Med 2003;253:472–80. Zwart-van Rijkom JE, van Hout BA. Cost-efficacy in interventional cardiology; results from the EPISTENT study. Evaluation of Platelet IIb/IIIa Inhibitor For Stenting Trial. Eur Heart J 2001;22:1476–84.