Epidemiology and resource use in Spanish type 2 diabetes patients without previous cardiorenal disease: CaReMe Spain study summary.

AIMS
To determine the first manifestation of cardiovascular or kidney disease (CVKD) and associated resource use in type 2 diabetes mellitus (T2DM) patients during seven years of follow-up.


METHODS
Observational-retrospective secondary data study using medical records of patients aged ≥18 years with T2DM and without prior CVKD between 2013 and 2019. The index date was 01/01/2013 (fixed date). The manifestation of CVKD was defined by the first diagnosis of heart-failure (HF), chronic-kidney disease (CKD), myocardial-infarction (MI), stroke or peripheral-artery disease (PAD). The main variables were baseline characteristics, manifestation of CVKD, mortality, resource use and costs. Descriptive analyses and Cox model were applied to the data.


RESULTS
26,542 patients were selected (mean age: 66.6 years, women: 47.8%, mean duration of T2DM: 17.1 years). 18.7% (N=4974) developed a first CVKD manifestation during the seven years [distribution: HF (22.4%), CKD (36.6%), MI (14.5%), stroke (15.3%) and PAD (11.3%)]. Overall mortality was 8.3% (N=2214). The mortality risk of the group that developed HF or CKD as the first manifestation compared to the CVKD-free cohort was higher [HR: 2.5 (95% CI: 1.8-3.4) and 1.8 (95% CI: 1.4-2.3)], respectively. The cumulative costs per patient of HF (€50,942.80) and CKD (€48,979.20) were higher than MI (€47,343.20) and stroke (€47,070.30) and similar to PAD (€51,240.00) vs. €13,098.90 in patients who did not develop CVKD, p<0.001.


CONCLUSIONS
In T2DM patients, HF and CKD were the first most common manifestations and had higher mortality and re-hospitalisation rates. HF and CKD were associated with the highest resource use and costs for the Spanish National-Health-System.


Page 3/17
Background Type 2 diabetes mellitus (T2DM) is a disease with a large social and health impact due to its high prevalence, the chronic macro and microvascular complications, and, mainly, due to cardiovascular manifestations [1][2]. The prevalence is about 6-10% in the general population [3] and is estimated to be 13.8% in Spain [4]. Around 11.3% of all cause-deaths are associated with diabetes, and half these deaths occur in persons aged < 60 years [5].
A diagnosis of T2DM is often accompanied by heart failure (HF,13-47%), chronic kidney disease (CKD, 30-40%) or both, in a high proportion of patients [2]. In fact, HF is more frequent, earlier, more recurrent and has a worse prognosis in people with diabetes than ischemic heart disease [6][7]. It has recently been reported that a diagnosis of HF at any time after the diagnosis of diabetes is associated with the highest risk of relative and absolute mortality at 5 years, and with a decrease in life expectancy at 5 years, when compared with any other cardiovascular or renal diagnosis [8].
In T2DM, myocardial involvement presents very early, and it is not easily recognized by usual tests. Up to two-thirds of patients present systolic and/or diastolic dysfunction within a few years of diagnosis [9] and even when cardiovascular risk factors (blood glucose, blood pressure, cholesterol, smoking, albuminuria) are controlled, the incidence of HF is not reduced, unlike that of ischemic heart disease or stroke [10].
The onset of CKD is a determining factor in the increased risk of HF [10]. Through a series of complex mechanisms (con uence of hemodynamic, neurohormonal and in ammatory interactions), the heart and kidney are intimately connected and contribute to the onset of HF. When presented together, they create a disorder known as cardiorenal syndrome [11][12][13]. Cardiovascular and renal disease (CVKD) in T2DM is a signi cant source of morbidity and contribute to continuing increases in health expenditure [6]. Real-life data show that, in patients with T2DM, cardiorenal events (HF or CKD) are the most common rst manifestation of cardiovascular disease (four times more frequent than stroke and myocardial infarction (MI) and six times more frequent than peripheral artery disease (PAD) and are associated with a high risk of mortality and other cardiovascular risks.
In Spain, there is little evidence on the type and time to the rst manifestation of CVKD after diagnosis of T2DM and subsequent impact on health resource use in usual clinical practice [14]. The aim of this study was to determine the rst manifestation requiring hospital admission due to a CVKD manifestation in initially CVKD-free T2DM patients during a 7-year follow up and quantify the use of health and non-health related resources and costs in these patients.

Design and study population
This was a retrospective, observational study based on electronic medical records from BIG-PAC [15] administrative database (secondary data source; owner: Atrys Health-RLD) which includes 1.8 million patients (http://www.encepp.eu/encepp/viewResource.htm?id=29236#). Primary data came from the computerized medical records of seven integrated Spanish public healthcare areas (primary care centers and hospitals) from seven Spanish Autonomous Communities. Before exporting to the BIG-PAC database [15] all electronic records were rigorously anonymized by the centers/hospitals of origin, in compliance

Inclusion and exclusion criteria
Inclusion criteria were: (a) age ≥ 18 years, b) patients active in the database for ≥ 12 months before study inclusion, c) inclusion in the prescription program (≥ 2 prescriptions during follow up), d) regular monitoring of patients (≥ 2 health records in the computer system) and e) without CVKD on or before the index date. Exclusion criteria were a) patients displaced or out of area, b) permanently institutionalized (geriatric residences), and (c) patients with severe mental illness or terminal illness.

Other variables of interest
Demographic variables, comorbidities, biochemical/anthropometric parameters, and baseline treatment were obtained (Table 1). Medication was obtained according to records from the pharmacological dispensing of medicinal products according to the Anatomical Therapeutic Chemical Classi cation System (ATC) [16]. The choice of drug in a speci c patient was at the discretion of the physician (clinical practice). Resource use and costs during follow up (2013-2019) We analyzed: (a) direct healthcare costs, de ned as costs related to care (primary care medical visits, specialist visits, days of hospitalization (hospital admissions), hospital emergency room admissions, diagnostic or therapeutic requests and medication; and b) indirect costs, de ned as days of productivity lost due to disability. Costs were expressed as the mean cost per patient and absolute cumulative cost. Rates were obtained from hospital accounting, except for the costs of medication and indirect costs (Table S1). Medical prescriptions were quanti ed according to the retail price per package at the time of dispensing (according to the General Council of O cial Pharmacists' Colleges of Spain (https://botplusweb.portalfarma.com). Indirect costs were considered as days of lost productivity due to disability according to the mean interprofessional wage (source: NEI) [17]. Use of resources and costs were calculated only for cardiovascular or renal disease. Total costs were obtained for the rst manifestation of CVKD and for each hospitalization due to CVKD.
Con dentiality of information/ ethical aspects Con dentiality of the anonymous records was respected in accordance with the Organic Law on the Protection of Personal Data. The study was classi ed and approved by local regulatory authorities and a Research Ethics Committee (ERC: Consorci Sanitari de Terrassa). In the participating centers, the informed consent of the patients is obtained. The study methods were carried out in accordance with the Declaration of Helsinki.

Statistical analysis
The presented data were initially validated to ensure the quality of the results. A descriptive, univariate analysis of the variables of interest was made. Qualitative data were expressed as absolute and relative frequencies and quantitative data as means and standard deviation (SD). The 95% ratios and con dence intervals (CI) were analyzed according to the total number of subjects with non-missing values. A bivariate analysis was made using the Chi-square test and the ANOVA test for independent groups. A Cox proportional risk model was constructed to estimate the time-to-rst manifestation and the mortality risk according to the different CVKD. The dependent variable was time (manifestation or death) and the covariates (confounding factors) were age, sex, and time from diagnosis (procedure: enter; method: maximum likelihood 2). A value of p < 0.05 was considered statistically signi cant. The analysis was made using SPSSWIN version 23 statistical program.

Baseline characteristics
Of an initial population of 731,759 patients aged ≥ 18 years on index date 01/01/2013, 43,596 had a diagnosis of T2DM (prevalence in the population attended: 6%; 95%CI: 5.8-6.2%). Of the 38,691 T2DM patients who met the inclusion/exclusion criteria on the index date, 31.4% (N = 12,149) had prior CVKD and were excluded from the study population, while 68.6% (N = 26,542) were free from CVKD and were selected for the study (Fig. 1).   (Fig. 2B).
The cumulative incidence rates of HF, CKD, MI, stroke, and PAD were 51. 6 Figure S1 shows the percentage of patients who had additional CVKD events/diagnoses different from the rst event/diagnosis. Patients with HF or CKD as the rst event had the greatest number of events due to other CVKD. In fact, in these two groups, 10% of the patients had three or more additional events/diagnoses different from the rst one (HF, CKD, MI, stroke and/or PAD; Fig. S1). HF and CKD were associated with a higher frequency of hospitalizations than classic CV complications (MI, stroke, or PAD). Days of occupational disability were low, as might be expected, since the mean patient age was ≥ 65 years (Table S2).
There were important baseline differences between patients who remained without CVKD compared with those who developed any CVKD manifestation during the follow up (

Use of resources in CVKD
Total healthcare and indirect related costs for the population studied during the 7-year follow were € 527 million, of which 96.0% were for direct healthcare costs and 4.0% for indirect costs. The main cost components were hospital admissions (39%), cardiovascular and antidiabetic medication (19%), specialist visits (14%) and primary care visits (13%) ( Table 2). According to the rst manifestation of CVKD the cost was HF: € 50,943, CKD: € 48, 979, MI: € 47,343, stroke: € 47,070 and PAD: € 51,240. The mean cost in patients without CVKD was € 13,099 (Table 2).  Fig. 3 and Table S3. The total hospital cost for the population over the 7 years of the study studied was €188.5 million, equivalent to a mean per patient of € 37,905. The total hospital cost due to HF (€ 58,258,479) and CKD (€ 51,471,084) as the rst complications of T2DM was signi cantly higher than that for MI (€ 28,092,277), stroke (€ 32,051,591) and PAD (€18,665,338); p < 0.001. Moreover, a signi cantly higher expenditure was observed when comparing the mean cost per patient/year before the rst event versus that after (Table S4) for the groups presenting HF (N = 1,112) or CKD (N = 1,820) as rst manifestation: a) HF (€ 1,509 vs. € 10,063; p < 0.001) and b) CKD € 2.162 vs. € 9,508; p < 0.001).

Discussion
The results of this population-based study show that two-thirds of T2DM patients were free of CVDK on the index date. During the follow up (2013-2019), HF and CKD were the most frequent rst manifestations compared with MI, stroke, and PAD, and also had higher mortality rates. In addition, both were associated with a higher incidence of rehospitalization when compared with MI, stroke, and PAD, increasing the use of healthcare and indirect resources, and increasing Spanish National Health System expenditure. These two diseases also resulted in a worse prognosis because, as described in our study, both HF and CKD are accompanied by a higher incidence of subsequent diagnoses that worsen the evolution. It should be noted that the 6% prevalence that we found corresponds approximately to the 50% of known diabetes that is described in the usual prevalence studies [4].
Winell [18], in a 1996-2012 study in Finland, found that diabetic patients have a higher incidence rate of HF, with a worse prognosis, than non-diabetic patients. Sukkar [19] studied 9,313 diabetic patients and found that 22.6% developed CKD during a mean follow up of 5.7 years. Advanced age and cardiovascular comorbidity, among other factors, were associated with an increased incidence. Our results are in line with these studies, with HF and CKD being the two most common manifestations of T2DM, and these two complications presented earlier than other CVD analyzed [20].
Albuminuria increases the cardiovascular risk, and its association with reduced glomerular ltration increases mortality [21]. Based on the results of the study, which show the importance of CKD and HF as serious and early complications in T2DM, strategies should be established to prevent and treat these patients early. Among them, the use of albuminuria and the albumin / creatinine ratio as early markers are of special importance. Koye [22] found an annual incidence rate of microalbuminuria of 7-8% in T2DM patients. Until recently, the treatments used for preventing kidney disease (renin angiotensin system inhibitors) were of very limited e cacy and there was no optimal tool to prevent HF development or its progression in any way. Our data are not surprising and show one of the therapeutic areas with unmet needs. Therefore, current data on SGLT-2 inhibitors are encouraging since they have demonstrated a capacity to both prevent and slow the progression of HF and CKD in diabetic patients and reduce hospitalization rate due to HF which is the most important cause of increase healthcare expenditure [23][24][25][26][27][28][29][30][31]. HF of ischemic etiology is associated with a higher risk of death compared with non-ischemic HF among patients with T2DM [32], however, this has not been con rmed in more recent data [33]. We had no access to the etiology of HF in our sample and therefore cannot analyze this aspect, which is a limitation of the study.
The cumulative hospital costs per patient of HF (€ 50,942.8) and CKD (€ 48,979.2) were higher than those for MI (€ 47,343.2) and stroke (€ 47,070.3) and similar to those for PAD (€ 51,240.0), compared with € 13,098.9 in patients without CVKD. Rehospitalization is common in this type of patients (HF and CKD), and results in high health resource use and costs. The most striking cost components were hospital admissions (39%) and medication (19%). A review by Einarson [34] described high comorbidity in T2DM patients with a mean annual cost per patient according to the absence or presence of CVD of $3,418 and $9,705, respectively. Wan [35], in a large cohort of patients (1.6 million patient-years) and with an 8.5 year follow up, found that the effect of CVD, cerebrovascular accidents, CKD and the combination of these factors has an additive impact to health costs, with special emphasis on CKD on T2DM patients. A Spanish study also highlighted a higher cost in patients with CKD + HF (€ 14,868) compared with those with HF (€ 9,365). The comorbidity associated with HF was high [14]. Goncalves [36] described the associated cost of HF and CKD attributable to diabetes in 2010-2016 in Brazil, which was $ 180 million per year for HF, with an upward trend. The presence of CKD increased the cost ($ 475 million). The authors emphasized that the economic burden of CKD will gradually increase in coming years, with serious implications for the nancial sustainability of the Brazilian public health system. McQueen [37] found that costs increase as kidney function decreases in T2DM patients (phase 1: $1,732 vs. phase 5: $6,949). Other authors highlighted the effect of prevention and self-care in the early stages of HF, reviewing the medical record and symptoms with the aim of reducing the economic burden of HF + T2DM on hospital admissions [38]. All these studies conclude that the cost of hospitalization and the presence of HF and/or CKD increases health costs in T2DM patients.
The study had some limitations: (a) the main limitation was the bias regarding the time of evolution of T2DM, as a single xed index date was used for patient selection; (b) the inherent limitations of retrospective, observational studies using databases, such as disease underreporting or possible variations in the recording of information by health professionals; the database was constructed in 2012, and therefore, for some patients it was not possible to determine with certainty whether patients were CVKD-free on or before the index date, even though the obtention of records from both primary and

Conclusions
In T2DM patients, HF and CKD were the rst and most common manifestations during a 7-year follow up, with a signi cantly higher impact on mortality and rehospitalization rates. This resulted in increased healthcare resource use and related costs for HF and CKD, respectively, in the Spanish National Health System.

DATA AVAILABILITY
The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

CONSENT TO PUBLICATION
Not applicable.

ETHICAL ASPECTS
The study was classi ed and approved by local regulatory authorities and a Research Ethics Committee (ERC: Consorci Sanitari de Terrassa). In the participating centers, the informed consent of the patients is obtained. The study methods were carried out in accordance with the Declaration of Helsinki.
The study authors declare that they have no competing con icts of interest for the remaining authors.

AUTHOR CONTRIBUTIONS
The conception and design of the manuscript were made by all authors. The collection of data and statistical analysis by ASN, and the interpretation of the data, drafting, revision and approval of the manuscript submitted, by all authors.