Trends in diabetes mellitus and related costs among hospital admissions in Switzerland, 2012–2020

Abstract

Background

In Switzerland, the prevalence of diabetes mellitus (DM) has increased in the general population, but little is known regarding DM-related hospitalizations and their impact on mortality and health costs. Hence, our objectives were to assess (1) the evolution of hospitalizations for DM as first diagnosis and as comorbidity; (2) the association of DM with ICU admission, length of stay (LOS), in-hospital mortality and costs.

Methods

Swiss hospital discharge data for period 2012–2020. Type 1 (T1DM), type 2 (T2DM) and other types (OTDM) of DM were considered.

Results

Between 2012 and 2020, the number of hospitalizations (% total) increased from 4204 (0.27) to 4980 (0.45) for T2DM, 539 (0.05) to 854 (0.08) for T1DM and 221 (0.02) to 381 (0.03) or OTDM. Hospitalizations with DM as comorbidity increased from 89,752 (8.6) to 128,700 (11.7) for T2DM, 2934 (0.29) to 3536 (0.32) or T1DM and 5774 (0.58) to 9143 (0.83) for OTDM. Compared to non-DM hospitalizations, all types of DM had a higher likelihood of lower limb amputation; hospitalizations for T1DM had a higher likelihood of ICU admission: odds ratio and 95% CI: 3.38 (3.19–3.59), while T2DM had higher LOS: 5.5 ± 1.0 vs. 5.1 ± 1.0 days, and all DM types had a lower odds of in-hospital mortality. Patients with any type of DM as comorbidity had a longer LOS than patients without. Total cost of DM rose from CHF 42 million in 2012 to almost 100 million in 2019 and decreased afterwards.

Conclusion

DM represents an increasing health and economic burden in Switzerland.

Highlights

We assessed 2012-20 trends and costs of hospitalized diabetes mellitus in Switzerland.

Hospitalizations due to diabetes mellitus increased by 25%.

Hospitalizations with diabetes mellitus as a comorbidity increased by 50%.

Costs for hospitalisations rose from 42 to almost 100 million CHF in 2019.

Background

Non-communicable diseases (NCDs) have received increased attention in recent years, and diabetes mellitus (DM) has emerged as a major global health issue [1]. The most prevalent form of diabetes, Type 2 Diabetes Mellitus, accounts for about 90% of all cases [2].

DM is one of the fastest-growing diseases worldwide, projected to affect 693 million adults by 2045 [3]. In Switzerland, the prevalence of reported DM in the general population increased from 3.3% in 1997 to 4.8% in 2007 [4] and to a further 5.6% in 2017 [1]. Still, the prevalence of clinically assessed DM is higher (6.3%), as one-third of DM patients are unaware of their status. Indeed, the prevalence of clinically assessed DM was estimated to be 6.6% in the canton of Vaud, a Swiss canton of about 720,000 residents [5].

In parallel, between 2003 and 2008, the number of hospital admissions of patients with DM increased by 38%, and the number of DM-related amputations increased by 34% [6], putting considerable pressure on the Swiss health system. A study conducted on patients aged over 64 years reported that the annual costs associated with DM increased from 5500 Swiss Francs (CHF) per patient in 2006 to 5830 CHF in 2011 [8] (1 CHF = 1.11 US$ or 1.03 € as of 01/07/2024). The same study also reported that the most frequent DM- and non-DM-related comorbidities were cardiovascular diseases (91%), rheumatologic conditions (55%), and hyperlipidemia (53%) [7]. Further, patients with DM and more than two comorbidities incurred 10,280 CHF higher total costs than patients without comorbidity [8]. Still, the data was based on reimbursement claims, and comorbidities were based on medications, not on clinical diagnoses. As reimbursements do not cover the entire costs of hospitalizations [9] the economic impact of DM is likely underestimated. Also, the analysis was conducted for a single health insurer and not for the entire Swiss population. Still, most information dates back from 2011 and no recent data has been published on the burden of DM on hospital admissions and cost in Switzerland regarding adults.

Thus, this study aimed to assess (1) the evolution of the prevalence of DM as first diagnosis and as comorbidity among adult hospitalized patients and (2) the association of DM with length of stay, ICU stay, costs, and in-hospital mortality. Data from 2012 to 2020 covering all Switzerland was used.

Methods

Data

Swiss hospital discharge data for the period 2012–2020 was obtained from the Swiss Federal Office of Statistics, contract CHUV220456 (https://www.bfs.admin.ch/bfs/en/home.html). Data provision to the Federal Office of Statistics is compulsory and covers over 98% of public and private hospitals in Switzerland. The information collected includes gender, age, length of stay, discharge status (main and secondary diagnoses, vital status), and procedures. As it was not possible to identify patients, the results relate to the number of discharges and not to the number of patients. Hospitalizations related to obstetrical conditions (ICD-10 codes beginning with “O”) were excluded. Also, age group [0–20] was not considered as it included infants and adolescents.

Diabetes

Main and secondary diagnoses at discharge were coded using the International Classification of Diseases 10^th revision (ICD-10) of the World Health Organization. Type 1 DM (T1DM) was defined as an ICD-10 code E10X; type 2 DM (T2DM) was defined as an ICD-10 code E11X; other types of DM (OTM) was defined as ICD-10 codes E12X, E13X and E14X, where X = any number. The details of the codes are provided in the supplementary Table 1. DM types were further categorized as main cause of hospitalization or as comorbidity. In almost all Swiss hospitals, ICD-10 coding is performed by specialized staff, based on the content of the discharge letter; if necessary, consultation of the electronic medical record is performed.

Outcomes

Overall length of stay (LOS) was indicated in days. Intensive care unit (ICU) and in-hospital mortality were categorized as yes/no variables. Procedures were coded using the systematic inventory of the Swiss Classification of Operations (CHOP) [10]. CHOP codes 84.10 to 84.17, corresponding to lower limb amputation, were considered. Hospital-related costs included total costs and costs related to medicines, doctors, laboratory, ICU, emergency, and surgery. They were expressed in CHF.

Covariates

Age was categorized into four categories as suggested elsewhere [11]: active (< 65 years), young-old (65–74), old-old (75–84) and oldest-old (85 + years). Gender was categorized as men or women. Seven administrative regions were considered (supplementary Table 2), as previous studies revealed that management of disease differs between them [12]. Nationality was defined as Swiss / non-Swiss as it has been shown that foreign nationals use health care differently from Swiss nationals [13]. No other information such as socio-economic status or income level was available. The severity of the case was defined using two metrics: the number of comorbidities reported and the Swiss version of the Charlson’s index [14], which was categorized into five categories: [0–1], [2–3], [4–5], [6–7] and [8+].

Statistics

Statistical analyses were conducted using Stata version 18.0 (Stata corp, College Station, TX, USA). Descriptive values were expressed as number of hospitalizations (percentage) for categorical variables and as average ± standard deviation or median [interquartile range] for continuous variables. Trends in hospitalizations were assessed using logistic regression. Bivariate comparisons were performed using chi-square for categorical variables and analysis of variance (ANOVA) or Kruskal-Wallis test for continuous variables. Multivariable association between DM and outcomes was performed using logistic regression, and results were expressed as odds ratio (OR) and (95% confidence interval). Multivariable comparison of LOS and costs was performed using ANOVA. As LOS and hospital costs had a skewed distribution, analysis was conducted on log-transformed data. For multivariable analyses, adjustment for year, gender, age categories, administrative region, and severity of disease was performed. Two models were used, one using the number of comorbidities and another using the Charlson’s index. Finally, a sensitivity analysis stratified by administrative region was conducted. Due to the large sample size, statistical significance was considered for a two-sided test with p < 0.001.

Results

Trends

Data from almost 10 million hospitalizations was used. The characteristics of the hospitalizations and the prevalence of the different types of DM as cause for hospitalization or comorbidity are summarized in Table 1. Men represented half of the hospitalizations. Patients aged less than 50 tended to decrease while patients aged over 70 increased in proportion. Patients with Swiss nationality slightly decreased, while the distribution of hospitalizations among administrative regions remained relatively stable.

Table 1 Characteristics of hospital admissions by year, Switzerland, 2012–2020

Regarding the main cause for hospitalization, T1DM remained stable, while T2DM increased and OTDM remained stable. Regarding DM as comorbidity, all types of DM increased, the increase in T2DM being stronger than the other types. In both cases (cause of hospitalization and comorbidity), the absolute number of hospitalizations declined between 2019 and 2020.

Outcomes

The hospital outcomes for each type of DM are summarized in Table 2. Regarding DM as the main cause for hospitalization, all DM types were associated with a higher likelihood of limb amputation; T1DM and OTDM were associated with a higher likelihood of ICU admission, while T2DM was associated with a lower risk of ICU admission than hospitalizations for other diseases. All DM types were associated with a lower in-hospital mortality than hospitalizations for other diseases. All DM types were associated with a higher LOS than hospitalizations for other diseases; this association was confirmed after multivariable adjustment including Charlson’s index, but after multivariable adjustment including total number of comorbidities, the LOS for all DM types became lower than for the other diseases. Similar results were found after stratifying on administrative region (supplementary Table 3).

Table 2 Hospital outcomes for each type of diabetes in Switzerland, 2012–2020, grouped by first diagnosis and comorbidity

Regarding DM as comorbidity, T1DM and T2DM were associated with a higher likelihood of limb amputation; all types of DM were associated with a higher frequency of ICU admission than hospitalizations for other diseases, and this association was confirmed on multivariable analysis adjusting for Charlson’s index; conversely, when the multivariable analysis was adjusted for number of comorbidities, admissions for DM had a lower likelihood of being admitted to ICU than other diseases. All DM types were associated with a higher LOS than hospitalizations for other diseases; this association was confirmed after multivariable adjustment including Charlson’s index, but after multivariable adjustment including total number of comorbidities, the LOS for all DM types became lower than for the other diseases.

The differences in LOS between the different types of DM and the admissions for other diseases are provided in supplementary Table 4.

Costs

The costs of hospitalization as per DRG are indicated in Table 3. Regarding DM as the main cause for hospitalization, the total cost for all types of DM was higher than for non-diabetics. This result was confirmed for T1DM after multivariable adjustment, while the costs of OTDM became lower and an economically non-significant difference (~ 60 CHF) was found for T2DM. Similar results were obtained after stratifying on administrative region (supplementary Table 5).

Table 3 Yearly hospital costs related to the different types of diabetes in Switzerland, 2012–2022, stratified by first diagnosis and comorbidity

Regarding DM as comorbidity, the total cost for all types of DM was higher than for non-diabetics. This association was confirmed after multivariable adjustment including Charlson’s index, but after multivariable adjustment including total number of comorbidities, the total cost for all types of DM was lower than for the other diseases. The differences in costs between the different types of DM and the admissions for other diseases are provided in supplementary Table 4.

Total yearly costs increased from almost 42 million CHF in 2012 to slightly less than 100 million CHF in 2019, to decrease afterwards. T2DM represented the largest part of the costs, and this percentage increased (81.5% in 2012 to 82.5% in 2020), while the part of T1DM decreased from 13.4 to 12.5% during the corresponding period and OTDM remained relatively stable (5.0% and 5.1%).

Discussion

Our results indicate that hospitalizations for T2DM as the main cause for admission increased by 25%, while the hospitalizations for T2DM or as comorbidity increased by almost 50%. Our results also show that hospitalizations for T1DM have a high probability of amputation, ICU admission, and that the costs related to DM hospitalizations are increasing faster than the number of hospitalizations.

Trends

The prevalence of diabetes as the cause for hospitalization increased, this increase being stronger for T2DM than for the others. Similar findings were obtained when DM was considered as a comorbidity. This increase in T2DM-related hospitalizations is likely due to the increase in obesity rates in the Swiss population [15, 16], associated with increasing ageing, as reported by Bellary et al. [15]. Overall, our results indicate that hospitalizations for or with T2DM are increasing in Switzerland, and that this increase could partly be prevented by lifestyle changes aimed at reducing obesity [17].

Outcomes

All types of DM had a higher likelihood of lower limb amputation, a finding also reported elsewhere [18, 19]. The main reasons are peripheral artery disease due to progression or inadequate management of the disease, as over half of people treated for diabetes in Switzerland do not achieve adequate control [20]. Although no information regarding diabetes control was available, our results stress the need for a tighter control of diabetes, to reduce amputation levels. Hospitalizations for T1DM and OTDM had a higher rate of ICU admission, while ICU admission rates for T2DM were lower. A study conducted in Saudi Arabia reported that patients with DM had a higher rate of ICU admission but did not differentiate between types of DM [21]. A possible explanation for the higher ICU admission rate among T1DM is diabetic ketosis [22]. Conversely, hospitalization for DM was associated with a lower in-hospital mortality rate, contrary to a Saudi Arabian study, where in-hospital mortality among patients with DM was higher [21]. A possible explanation is that management of DM-related complications might be simpler than for other causes of admission, or to the fact that many hospitalizations for DM are unnecessary [23]. Hospitalizations for DM also had a longer LOS than hospitalizations for other diseases, and this was mainly the case for T2DM. Those findings replicate those of a study conducted in England, where hospitalizations for T2DM had a longer LOS than for T1DM, but no information regarding non-DM hospitalizations was provided [24].

As comorbidity, both T1DM and T2DM were associated with a higher likelihood of lower limb amputation, indicating that patients hospitalized with those conditions might be afflicted with worse peripheral vessel status than patients devoid of diabetes. All types of DM were associated with a higher frequency of ICU admission and with higher LOS. A study conducted in the UK reported that hospitalized patients with diabetes had a longer LOS, a higher ICU admission rate and a higher mortality rate than patients without diabetes [25]. Another study conducted in Germany also reported that hospitalized patients with DM had a longer LOS than non-DM patients [26]. In both studies, no differentiation between types of diabetes was provided.

Overall, our results suggest that hospitalizations for T1DM or ODM carry a higher likelihood of amputation, IUC admission and a longer LOS.

Costs

The total cost of all types of DM increased until 2019, to decrease afterwards. A likely explanation is the COVID-19 Pandemic, which deterred patients to come to hospital [23]. The cost related to hospitalizations for T1DM decreased, while costs related to hospitalizations for T2DM increased in absolute and relative values. Our findings replicate those of a Polish study that reported a doubling of the direct cost of diabetes between 2005 and 2009. A French study reported a 906 € million increase in DM-related expenditures between 2015 and 2019, the highest absolute increase among all diseases [27]. Importantly, our results likely underestimate the total cost of DM, as it has been reported that outpatient care represents 57.1% of all DM-related costs in Switzerland [28].

Interestingly, costs related to DM evolved at a slower pace than admissions for the disease, namely T2DM. Several hypotheses can be put forward to explain this diverging trend. First, patients with T2DM might be better managed in the ambulatory setting, thus making their disease less severe when being admitted to hospital [29]. Second, improvement in in-hospital management of DM cases could decrease LOS [29] and consequently costs.

Overall, our results indicate that costs related to hospitalizations for DM are escalating in Switzerland but might correspond only to a small fraction of all DM-related costs.

Implications for public health

T2DM can partly be prevented if strong, adequate lifestyle interventions are applied, although the implementation of such measures at the population level is hard to achieve. Still, our results suggest that if 1 out of 10 hospitalizations for T2DM could be avoided, a yearly saving of 7 million CHF regarding hospitalisations would be attained, and likely much more considering ambulatory and other (i.e., sick leave) costs. This value might appear small compared to the total expenditures for prevention and health promotion in Switzerland (1.8 billion CHF), but relatively important if only the expenditures for addiction and non-communicable diseases are considered (193 million CHF in 2019) [30]. Hence, any intervention aimed at preventing T2DM would likely be cost-effective.

Strengths and limitations

Our study as several strengths: first, it included almost all hospitalizations in Switzerland, spanned several years, and used data collected using the same methodology.

Our study also has some limitations: firstly, it was not possible to identify patients with multiple hospitalizations; hence, it is likely that the number of patients hospitalized for DM might be lower. Secondly, due to the Swiss health system, which is characterized by the highest out-of-pocket share of OEDC countries (representing 28% of total health spending) [31], generalizability of the findings to other countries and other health systems might not be possible. Still, our results provide important information for health managers and public health professionals. Thirdly, coding of DM might have been either over or underestimated due to coding guidelines or aspects regarding whether it generates billable effort for a hospital [32]. Still, a previous study concluded that coding of DM had a high degree of concordance between different health databases in Switzerland [33].

Conclusion

DM represents an increasing health and economic burden in Switzerland. Most increase was in T2DM, which could be limited via lifestyle changes.