Common procedures and conditions leading to inpatient hospital admissions in adults with and without diabetes from 2015 to 2019 in Germany

Approximately 8 million people with documented diabetes mellitus were living in Germany in 2020, implying a type 2 diabetes (T2D) prevalence of about 9% [1]. While the estimated number of undiagnosed cases decreased from around 2 million in 1997–1999 to 1.3 million in 2008–2011 [2], it has been estimated that the population with diagnosed diabetes will to rise to 12 million in 2040 according to data from statutory health insurances [1].

This development is not restricted to Germany but is predicted for several upper income and middle income countries, estimating a prevalence of up to 25% for diabetes for some of these countries in 2030 [3]. The increase in diabetes prevalence around the world [4] together with stable or slightly increasing per capita healthcare costs in individuals with diabetes in Germany (1.7 times higher than in individuals without diabetes) [5] could lead to challenging nationwide and global healthcare costs in the upcoming decades [6, 7].

Higher healthcare costs in people with diabetes are mainly the consequence of prescribed medication from pharmacies and inpatient treatment [5] but also of outpatient treatment and indirect costs, e.g., due to reduced productivity in the work place [8]. Higher inpatient costs in patients with diabetes may be due to more frequent hospitalizations, longer hospital stays or more complications compared to people without diabetes. There are few publications reporting a high prevalence of diabetes among hospitalized cases [9] as well as frequent readmissions in people with diabetes [10] but data on admission rates (accounting for the respective reference population with and without diabetes) are scarce. Furthermore, it is less clear which procedures and diagnoses are mostly related to more frequent hospital admissions, longer hospital stays as well as higher rates of complications and mortality among individuals with type 1 diabetes (T1D) or T2D compared to those without diabetes.

The aim of this study was to compare the frequency and outcomes of inpatient hospital admissions for several high-volume procedures and diagnoses between all cases with or without diabetes from 2015 to 2019 on a nationwide basis using mandatorily documented data in Germany.

This is the largest evaluation in Germany of data on frequency of hospitalization, length of hospital stays and complications in more than 14 million inpatient cases with diabetes compared to more than 54 million inpatient cases without diabetes between the years 2015 and 2019. Every fifth inpatient case aged over 40 years in Germany had a diagnosis of type 1 or type 2 diabetes. We could detect higher hospitalization rates in the population with diabetes compared to without diabetes. Higher proportion of complications as well as longer hospital stay were observed in inpatient cases with vs. without diabetes for nearly all procedures and diagnoses over all age groups. Mortality was generally higher in inpatient cases with vs. without diabetes.

Data on the frequency of initial inpatient admission rates referring to the respective population are scarce; however, higher hospital readmission rates in people with diabetes have been reported previously [10]. The risks for revision of hip endoprosthesis [15] as well as readmission because of cardiac diagnoses [16, 17] have particularly been mentioned in the literature. More frequent hospital admissions in individuals with diabetes have been reported for appendectomy [18] but not for cholecystectomy [19]. We found a higher hospitalization rate for nearly all procedures and diagnoses in females aged 40 up to 70 years (magnitude depending on the procedure). In males these findings were restricted to myocardial infarction and stroke. We are not aware of previous studies reporting such gender differences regarding hospitalization for surgery; however, it is known that male individuals are diagnosed with diabetes earlier than females [20] leading to a higher prevalence of diabetes in middle-aged men than women, but the impact of diabetes on mortality is stronger in females [21]. As we analyzed case-related data, we could not ascertain whether the differences in hospitalization rates between females with and without diabetes were mainly because of a higher risk for inpatient admission in general or due to readmissions of some individuals, except for hip and knee replacements where revisions are separately encoded (hip: OPS 5‑821, knee: OPS 5‑823). It is assumed that readmissions are one of the reasons for higher hospitalization rates of people with diabetes concerning myocardial infarction and stroke. Diabetes was considered as risk factor for readmission 30 days after myocardial infarction [17] and as a possible risk factor for readmissions after stroke according to a systematic review but data were too heterogeneous to provide clear evidence [16]. We assume that both risk for initial hospital admissions and readmissions are responsible for the higher number of inpatient cases/100,000 individuals in the population with diabetes, especially in females. Additionally, there seems to be a shift towards earlier orthopedic surgery in the population with diabetes. The higher frequency of orthopedic procedures in individuals without diabetes in higher age groups could therefore be due to the fact that the surgery was already conducted in earlier ages in people with diabetes. Furthermore, the admission rate for individuals with diabetes might be still underestimated, because of undiagnosed inpatient cases of diabetes. A survey to estimate the prevalence of T2D among patients aged ≥ 55 years in German hospitals found a proportion of 9.5% of individuals with unrecognized T2D at admission [22]. Another study reported a rate of 4% with undiagnosed diabetes in hospitalized patients aged 50 years or older [23]. It must be kept in mind that in many hospitalizations for diabetes, the diagnosis of diabetes might not be documented, especially because in the German payment system (DRG), diabetes yields no high return compared to other diagnoses. Furthermore, we used billing data of panel doctors, which excludes about 14% of the population (especially from private health insurances) to estimate the proportion with T2D of the entire population, but the DRG statistics cover the whole German population. Consequently, the real prevalent population with type 2 diabetes may differ slightly, which, in turn may have led to a misjudgement of the proportion of inpatient cases among patients with diabetes. Furthermore, it must be assumed that the prevalence of T2D has increased since 2011 which would have led to an underestimation of the actual T2D population in the years 2015–2019. Considering all these limitations, the total admission rate must be interpreted with caution, but the shift towards younger ages and the gender differences in individuals with diabetes remains as all these limitations should be equipollent in all subgroups. Especially the lower rate of surgery in men with diabetes needs to be observed in future as it might depict a possible undertreatment among this group.

Data from the United Kingdom (UK) Arthroplasty Pain Experience (APEX) trials were similar to our results on length of hospital stay in knee and hip replacements with about 1 day longer stays in individuals with diabetes; however, the overall length of hospital stay was nearly doubled in our cohort and the differences in hospital days in the APEX study vanished after adjustment for further comorbidities [24]. This might suggest that longer hospital stays in inpatient cases with diabetes might be a consequence of the combination of diabetes itself and diabetes-associated long-term complications and comorbidity. In addition, hospital acquired complications were markedly higher in diabetes patients, which also accounts for longer hospital stays. It is further known that in Germany hospital stays are generally longer than in most other European countries [25], which can be partly explained by the German healthcare system that provides three times more hospital beds than the UK [26] (7.9 vs. 2.4 per 1000 inhabitants). In addition, diabetes treatment initiated and adjusted during a hospital stay for a surgical procedure may add to the excess length of hospitalization in cases with diabetes. Spine surgery was mentioned in a previous review to be associated with longer hospital stays, more complications, higher mortality and higher risk for readmissions in individuals with diabetes [27], which is in line with our results. Another study found that the length of stay is highly dependent on glycemic control. The authors reported a difference of up to 5 days in hospital stay between people with uncontrolled diabetes and individuals without diabetes, but only 1 day difference in patients with controlled diabetes [28]. These are important results suggesting that the risk for longer hospital stays in people with diabetes might be markedly reducible by improving glycemic control. We found especially high differences in the length of hospital stays for appendectomy comparing cases with and without diabetes aged 40–< 60 years. Longer hospital stay for appendectomy was previously mentioned for individuals with diabetes [29, 30] and in people with preoperative fasting blood glucose levels of ≥ 123 mg/dl [31]. Length of hospital stay and healthcare costs were lower in patients with diabetes for laparoscopic appendectomy compared to open appendectomy according to a study from Taiwan. The authors proposed that laparoscopic appendectomy should be used particularly in individuals with diabetes to reduce the risk for longer hospital stay and healthcare costs [32]. Our results depict that this might be especially important in middle-aged people (40–< 50 years) with diabetes, where length of stay and the proportion of complications were nearly doubled compared to cases without diabetes.

It must be kept in mind that especially in the younger age groups other factors besides the diabetes itself might contribute to the higher number of orthopedic surgeries and the longer hospital stay. Obesity and reduced physical activity could be both underlying reasons for premature type 2 diabetes and risk factors for orthopedic surgery as well as for longer hospital stay, but the extent is still discussed in the literature [33‐35].

We detected higher mortality rates in cases with diabetes compared to without diabetes for most procedures. Appendectomy, cholecystectomy and spine surgery were procedures with increased mortality in patients with diabetes over all age groups. In terms of appendectomy and cholecystectomy, this is in line with previous studies [36, 37], while data on mortality in cases with diabetes undergoing spine surgeries are controversial [38, 39]. Our results indicate that spine surgery in individuals with diabetes should get high attention in hospitals, because despite the overall low mortality rate in this procedure, inpatient cases with diabetes showed an up to 4‑fold higher risk for fatal outcome compared to inpatient cases without diabetes. For hip and knee endoprostheses as well as shoulder surgery, our findings are quite congruent to the literature concerning complications in diabetes patients. Infections, the risk for revision of endoprosthesis or tendon re-tearing at the shoulder joint, are often reported [40‐43]. Publications on mortality are scarce for these procedures, which might be due to the overall low mortality associated with these surgeries. We found higher mortality in cases with diabetes for hip and knee replacements and for shoulder refixation but only above the age of 70 years.

The strength of this study was the coverage of all inpatient cases between 2015 and 2019 in Germany irrespective of their insurance status, providing a representative picture of the actual hospitalizations and complications for inpatient cases with and without diabetes. Limitations were that the DRG data are only case-related and therefore no information on the patient level was obtainable. Additionally, it must be mentioned that the cases are based on billing data which could have influenced the coding of diagnoses and the classification of diabetes types to some extent. For this reason, we decided to combine type 1 and type 2 diabetes and exclude other rarer diabetes types. In addition, we cannot exclude that some cases may have had unrecognized diabetes. The results for hospital admissions/100,000 individuals must further be interpreted with caution as the prevalence of T1D and T2D is only an estimation based on data from previous years, while data on the whole German population are assumed to be relatively precise. Therefore, we concentrated on remarkable differences in hospitalization rates and age distribution. Furthermore, due to the structure of case-related billing data with restricted variable content, additional information on the diabetes disease, such as duration of disease, glycemic control, diabetes-related comorbidities and medication could not be analyzed. The same applies to relevant patient characteristics, namely social background, educational status and ethnicity.

The longer hospital stays as well as higher proportion of complications and mortality in patients with compared to those without diabetes clearly indicate that inpatient care of these patients must be intensified, especially in departments not specialized for the treatment of diabetes. Therefore, trained consultant diabetes specialists are desirable in hospitals. Furthermore, higher hospitalization rates in younger patients with diabetes especially for orthopedic surgery point at an increased disease burden in patients with diabetes. Lower hospitalization rates for orthopedic procedures in patients with diabetes in the high age groups might indicated that surgical treatment in these high-risk patients is avoided.

As the prevalence of diabetes is likely to increase in the next decades and individuals with diabetes need surgery earlier, have longer hospital stays, more complications and higher mortality than the general population, these results depict a challenging future for the German healthcare system. People living with diabetes requiring surgery represent a vulnerable group, and sufficient personnel trained in diabetes care is required in all hospitals. In addition, it is important to lower the rate of unrecognized diabetes by screening prior to hospital admission, to enable adequate medical care. Despite the fact that hospitalization rates and length of stay differ between healthcare systems, our results should be transferable to many middle income and high income countries.