Diabetes and vulnerabilities in people undergoing lower-extremity amputation

In Denmark, nearly 370,000 people are living with diabetes [1], and this figure continues to rise [2]. Moreover, 7.1% of adult Danes have prediabetes [3]. Lower-extremity amputation (LEA) is typically a consequence of peripheral arterial disease, peripheral neuropathy and diabetic foot ulcers, i.e. potential complications of diabetes [4-6]. LEA is among the most feared complications among people with diabetes [7]. LEA affects mobility, social activities and mental well-being [8]. Furthermore, LEA is associated with a high mortality, e.g., an approximately 50% one-year mortality in people with type 1 diabetes (T1D) who have undergone transfemoral amputation [9].

Prevention of LEA depends heavily on early monitoring and timely specialised care, including surgical revascularisation [10]. Knowledge of personal characteristics and healthcare contacts in the period leading up to an LEA is essential for targeted prevention. Few studies have studied these characteristics in a Danish context and in people with diabetes. One study focusing on people who underwent LEA in 2010-2011 found that almost half had diabetes. The majority were men, unmarried and with low educational attainment. Moreover, the prevalence of comorbidity was high [11]. Another study found that compared to an LEA population without diabetes, those with diabetes underwent LEA at a younger age [9]. Regarding healthcare contacts, a study found that 64% had attended a hospital or outpatient clinic within three years prior, and 97% had visited their general practitioner (GP) in the year leading up to their LEA. However, 2% neither had a hospital nor a GP contact in the year preceding LEA [11]. However, we need up-to-date data on the characteristics of people who undergo LEA to develop strategies that may prevent disease progression leading to LEA.

We describe socioeconomic and demographic factors, comorbidity, healthcare contacts and mortality in people who underwent non-traumatic major LEA in 2019-2021. Figures are provided by diabetes status and in a matched study group from the general population.

Methods

Study design

This was a descriptive, observational study using Danish administrative registries.

Data access

The Danish Data Protection Agency approved this project. Data were made available via Statistics Denmark’s online access and were linked via the civil registration number (CPR) provided to all citizens with permanent residence in Denmark.

Major lower-extremity amputation

We identified first-incident major LEA in the National Patient Registry (NPR) from 2014-2021 using NOMESCO procedure codes for trans-tibial (KNGQ09, KNGQ19 and KNGQ99) and trans-femoral (KNFQ09, KNFQ19, and KNFQ99) amputations [12], not excluding people who had undergone minor LEA. We limited our study population to those who underwent their first major LEA in 2019-2021 (the LEA population). We excluded traumatic major LEAs via either concomitant diagnoses indicative of trauma, hospital contacts due to accidents, acts of violence, suicide attempts and other types of self-harm (Supplementary 1). Additionally, we excluded all people ≤ 30 years of age (likely trauma).

Diabetes

We identified people with diabetes using a previously described algorithm [13]. Briefly, it identifies diabetes indicators in the NPR, the National Health Services Register, the National Prescription Registry, the Danish Adult Diabetes Database and the Danish Registry of Diabetic Retinopathy (DiaBase, available until 2018). The Diabase is a Danish clinical quality database that contains data from screenings for diabetic eye disease at private ophthalmologists or hospital eye departments in adults with diabetes [14]. Based on diagnoses and reimbursed medication, we classified people as either T1D or type 2 diabetes (T2D).

Socioeconomic and demographic factors

We gathered data on age, sex, municipality and relationship status from Statistic Denmark’s Population Register. Statistics Denmark defines a relationship from records of either marriage or registered partnership, but their figures are also based on whether people live together and have children. Based on the Population Education Register and the Labour Market Module at Statistics Denmark, we identified educational attainment and occupational status, respectively. Educational attainment was categorised as 1) low (only elementary school), 2) middle level (high school or vocational training) or 3) high (bachelor’s degree and above). We categorised occupational status as 1) employed or student, 2) outside the workforce (unemployed or someone who receives social benefits), 3) early retirement disability pension or 4) age-related retirement. Lastly, we computed income quartiles from the weighted family income registry for the entire Danish adult population separately for those aged 30-64 years and those 65 years and above and categorised our sample accordingly.

HbA_1c, morbidity and all-cause mortality

We identified latest HbA_1c assessment (NPU27300) within one year from the index date from the National Laboratory Database. We retrieved data on comorbidity from the NPR ten years before the LEA index date based on medical diagnoses (primary and secondary). To identify cardiovascular disease and kidney disease, we also retrieved data on treatment procedures. For osteoporosis, chronic obstructive pulmonary disease/asthma and mental illness, two purchases of disease-specific medications two years before the date of LEA, according to the National Prescription Registry, was also suggestive of comorbidity (Supplementary 1). We computed the 30-day, six-month and one-year mortality using Danish Cause of Death Register data.

Healthcare contacts

We summarised the number of GP contacts (including e-contacts) based on the National Health Services Register and the number of hospital inpatient and outpatient contacts from the NPR in the year leading up to the major LEA index date. Ambulatory hospital contact and a hospitalisation were defined as a hospital stay of ≤ 12 hours and > 12 hours, respectively. If any part of a hospital stay was acute, we categorised the entire hospital contact as acute. We also identified those referred to home care in the Eldercare Database at Statistics Denmark.

Statistical analyses

The LEA population was matched (exact matching) 1:4 to people in the Danish Population Register according to age (years), sex (male/female), diabetes status (none/T1D/T2D) and calendar quarter. We computed frequencies, percentages, medians and lower and upper quantiles for descriptive variables stratified by diabetes status. We tested group differences using a t-test (paired or unpaired) or Kruskal-Wallis test (non-normal distributions) for continuous variables. For categorical variables, we conducted either χ² tests (for unpaired data), McNemar’s test or Bowker's test of symmetry (for paired data). We computed incident major LEA per 10,000 citizens by diabetes status, which we modelled using Poisson regression with diabetes status as an independent variable. We deemed α = 0.05 statistically significant.

All statistical procedures were conducted in SAS 9.4.

Trial registration: not relevant.

Results

Sample flow

We identified 3,156 people who underwent a first major LEA in 2019-2021. After exclusion of putative trauma-related major LEA (n = 48), 3,088 people remained in the sample, of whom 1,366 (44.2%) had diabetes (T1D: 5.0% and T2D: 39.3%) at their major LEA index date (Supplementary 2).

The incidence rates of major LEA remained stable during 2019-2021. Compared to those without diabetes, the incidence rates were 14.9 (95% confidence interval (CI): 5.7-34.8; p < 0.001) and 9.0 (95% CI: 3.6-22.6; p < 0.001) times higher for those with T1D and T2D, respectively. Moreover, the incidence rate was 1.6 (95% CI: 1.1-2.3; p = 0.019) higher for those with T1D than for those with T2D (Supplementary 2). There were some numerical differences in rates between Danish regions (Supplementary 2).

Socioeconomic and demographic factors

Compared to the reference, a statistically significantly larger percentage of the LEA population was single, had retired early, had elementary school education, was in the lowest income group and resided in a non-urban municipality (Table 1).

Among the LEA population, those with T1D underwent major LEA approximately ten years earlier than the other groups (Table 1). The LEA population with T1D had the lowest proportion of people with elementary school education and the highest proportion of people who had retired early (one third). The LEA population without diabetes and those with T2D differed in age, sex and relationship status, although they were otherwise not different (p ≥ 0.233 for the remaining sociodemographic variables).

Morbidity

HbA_1c and the proportion with morbidity was higher in the LEA population than in the reference group (p < 0.001 for all variables) (Table 2). The relative differences in percentages were largest for CVD and kidney disease. Notably, around half of the LEA population had a mental illness, whereas this was the case for less than a third in the reference group. The LEA population with T1D had a higher HbA_1c and a longer diabetes duration than the LEA population with T2D (p < 0.001 for both variables) (Table 2).

Healthcare contacts

The LEA population had more contacts with primary and secondary Danish healthcare in the year leading up to their major LEA than the reference group (p < 0.001 for all variables).

Among the LEA population, those with T1D had more elective ambulatory hospital contacts than those with T2D but had fewer GP contacts (Table 3).

A total of 24 people (median age 66.0 years (lower quantile (Q1)-median (Q2): 61.5-78.0 years), 75% male) in the LEA population had no hospital contacts in the year leading up to their LEA (data not shown). Among these, 37.5% had T2D, and the remainder did not have diabetes.

Mortality

The LEA population had a much higher proportion of mortality than the reference group (p < 0.001 for all variables) (Table 4). Approximately one in six patients in the LEA population had died one month after their index date, whereas one in three had died one year after their index date. There were no statistically significant between-group differences in mortality within the LEA population (p ≥ 0.100 for all variables).

Discussion

This national registry-based study identified 3,088 people who underwent their first non-traumatic major LEA in 2019-2021 in Denmark. Among these people, 5.0% had T1D and 39.3% had T2D. The majority were elderly men who were single, had elementary school education, were in age-related retirement and had a low income. We identified marked social inequality in the LEA population compared to a matched reference group, also matched on diabetes status. Moreover, the LEA population had markedly more chronic disease, hospital-, GP- and home care contacts and much higher mortality than the reference group.

As in our study, another Danish study found that among people with major LEA caused by peripheral arterial disease, 51.4% had basic educational attainment and 62.7% were unmarried [6]. In a Canadian study of people with LEA, among those with diabetes and those without diabetes, 75.4% and 64.5% had CVD, respectively [15]. These percentages are lower than those reported in our study. Kidney disease was more prevalent in those who underwent major LEA than in the reference group. Likewise, other studies have reported that among individuals who underwent major LEA, 34.8% of those with diabetes and 13.8% of those without diabetes had renal failure. [15]. Moreover, more than a third of the LEA population died within one year of major LEA and 12.3-15.7% died within 30 days. These figures are slightly lower than the numbers reported elsewhere [16, 17]. We also found that half of the LEA population had mental illness before LEA. Danish national guidelines recommend screening for the psychosocial needs of the major LEA population due to the life-changing circumstances of LEA [18]. Our findings underpin such recommendations. The dual burden of somatic and mental disease adds to the complexity of this group and underscores the need for a targeted, interdisciplinary treatment and care approach. Danish national guidelines further underpin the need for interprofessional treatment and care for people with major LEA [18], where personal characteristics (old age, sex, low educational attainment, and low income) must be considered.

Within the LEA population, we found that a considerable percentage of individuals with T1D had retired early and belonged to the highest education and income groups, unlike those with T2D. Moreover, people with T1D had a higher HbA_1c and a larger proportion of kidney disease. Also, those with T1D had more outpatient contacts, while the LEA population with T2D had more GP contacts. In Denmark, T1D is treated chiefly at hospitals, whereas most people with T2D are treated at their GP. Few people in the LEA population had very few or no healthcare contacts. Similarly, a Danish study found that 18% of people who underwent LEA had only GP contacts in the year prior to their amputation, and 2% had neither a hospital nor a GP contact [11].

Regular foot and footwear inspections, preventive foot and shoe care, multidisciplinary management of foot ulcers and early diagnosis of vascular disease are recommended to prevent LEA among people with diabetes [4]. However, more than half of the LEA population did not have diabetes, while both groups were highly vulnerable. LEA prevention and rehabilitation should target high-risk populations both with and without diabetes.

The strengths of this study include our use of nationwide registries and an algorithm to categorise people by diabetes status. Even so, the algorithm may be subject to under- or misclassification. Despite matching key variables, we acknowledge that many other factors determine the distributions of social and healthcare factors. Moreover, comparisons within the LEA population must be made with extra caution. Lastly, multiple testing increases the risk of type 1 error.

Conclusions

We identified a highly vulnerable LEA population with pronounced social inequality compared to a matched reference group. Two of every five in the LEA population had diabetes. The LEA population had markedly more somatic and mental disease, more healthcare contacts and higher mortality than the reference group. We recommend interdisciplinary preventive initiatives to target this vulnerable population and research investigating predictors of major LEA.