1 BACKGROUND

Chronic leg ulceration, defined as a wound on leg or foot that does not progress through the healing process in a timely manner,1 is both a major challenge in healthcare and burdensome to affected individuals.2 This is especially true when wound treatment continues for extended periods of time and still ends in relapse, amputation, or death. Reliable epidemiological data on incidence and outcomes in patients with chronic leg ulceration is underreported,3, 4 which hampers efficient resource allocation, planning, and improvement of wound care for the most affected patients.

Globally, it has been estimated that approximately 1.5 per 1000 people in the general population have a chronic leg wound (>3 weeks, all aetiologies).4 It has also been suggested that 15% to 25% of the diabetic population will have a foot wound at some point5, 6 with an annual incidence of 2%.7 Although diabetic foot ulcers without complications generally heal within 6 weeks, this is not a realistic outcome for the majority of diabetic patients as only 45% of patients presenting with an initial diabetic foot ulcer will become both ulcer free without experiencing amputation and alive at 12 months.8 There are several reasons why leg wounds do not heal spontaneously with the most common being infection, venous insufficiency, and ischaemia. These conditions are often found in combination.1, 9 Critical limb threatening ischaemia has been suggested as a contributing factor for the absence of wound healing in 30% to 40% of cases10 and is a strong predictor of amputation and early death.10 In the western world, most amputations (>90%) are necessitated by complications associated with vascular disease. The majority of these patients have a history of chronic wounds.11

The aim of this study was to investigate incidence and predictors of wound healing, relapse, major amputations, and/or death among patients referred to specialist treatment at hospital for chronic leg wounds.

2 METHODS

The study was designed as an observational register-based cohort study based on nationwide individual-level linked data from several Danish national registries with 5 years of follow-up.

The study is reported in concordance with the Reporting of Studies Conducted and uses the Observational Routinely-Collected Data (RECORD) statement (an extension of the existing STROBE guidelines).12

2.1 Study population

The Danish healthcare system is government funded, which ensures free access to health care at hospitals, general practitioners, and homecare for all residents.13 Patients with complicated leg wounds are referred to specialist treatment at hospital wound clinics while less complicated cases are treated in primary care. For this study, patients aged ≥18 years were included on their first referral to hospital for leg wounds within a 5-year period (wounds on leg or foot, all aetiologies). Patients were included if a wound required more than 6 weeks of treatment (admissions and/or outpatient visits) during the 6-year period from 1 January 2007 to 31 December 2012 (n = 8705). The index date was the date of admission/first outpatient visit plus 6 weeks. Disease-related risk factors were identified from ICD-10 codes on primary or secondary diagnoses 5 years prior to index date. Follow-up was 5 years after date of admission or until death.

A national group of wound specialists selected a comprehensive list of ICD-10 codes (supplementary appendix), which identified eligible cases of leg wounds. Other wound specialist in seven hospitals then confirmed this list. Comprehensiveness of the list was further validated by comparing codes used at two hospitals for patients with known leg wounds without more codes being identified. Pressure ulcers and cancer-related wounds were excluded as the ICD-10 codes could not definitively define whether these wounds were situated on a foot or leg. Patients were excluded if they had had major amputation within 5 years before study start (n = 243), died within 6 weeks of 1 January 2007 (n = 38) or emigrated during the study period (n = 30). The final sample for analysis consisted of 8394 patients.

2.2 Data sources In Denmark, all residents are provided with a unique personal identification number,13 which is used as the key identifier in all Danish health and social care registers. Data from the following national registers were used to identify the study population and to extract data on baseline characteristics, risk factors, and outcomes: The Danish Civil Registration System (all persons living in Denmark).13 The Danish National Patient Register (all somatic inpatient admissions and outpatient visits including diagnosis codes).14 The Danish National Prescription Registry (individual data on all dispensed prescription pharmaceuticals sold in Danish community pharmacies. Data consist of information on ATC codes, price, reimbursement, and so on, at patient level).15 The Danish Population Education Register (information on completed education).16 Different socio-economic registers from Statistics Denmark (population-level demographic data). 2.3 Variables 2.3.1 Outcomes

Four outcome variables were chosen for analysis: wound healing, relapse of wound, amputation, and death. Wound healing was defined as end of contact with hospital as patients are normally monitored at hospital until wound healing is achieved (wound is completely epithelised). Patients with major amputation who continued hospital contact with a wound diagnosis were not considered healed until end of contact. Relapse wound was defined as new contact with one of the abovementioned diagnoses more than 2 weeks after end of first hospital contact. Index date was last day of first hospital contact. Major amputation was defined as amputation above the ankle. Death was defined as death by all causes; see supplementary appendix for more details.

2.3.2 Explanatory variables The following explanatory variables were chosen based on the potential influence on extended wound healing and risk of amputation1, 3, 9, 17: Age (20-year intervals [eg, 40–59]) Sex (male/female) Cohabitation (cohabiting [married or residing with partner]/living alone). Educational level basic (primary school)/vocational (vocational or lower secondary)/higher. For each patient, the highest completed level at day of admission was selected. Place of residence (five regions). This variable controls for local variability in health care. Diabetes (yes/no). Individuals with diabetes were identified with the ICD-10 code for diabetes type 1 or 2 (E10-11) and/or having picked up prescribed antidiabetic medication from pharmacies more than once in 5 years before inclusion. To identify patients diagnosed with diabetes when they presented with a wound, individuals who had the diagnosis or had picked up prescribed antidiabetic medication from pharmacies at least once from day of admission until index date were classified as having diabetes. Peripheral arteria disease (PAD) (yes/no). Individuals with at least one ICD-10 code for PAD or arteriosclerosis within 5 years prior to inclusion were considered. Comorbidity (categorised in Charlson comorbidity index 0, 1-2, and 3+ where higher score predicted increased risk of death18). In the hazard ratio (HR) models, diagnoses of diabetes and PAD were not included in the Charlson comorbidity calculation. This was to avoid controlling for the same diagnosis multiple times as diabetes and PAD were included as separated variables in the models. All variables were used at date of admission unless otherwise stated. More details, including diagnostic codes, are available in the supplementary appendix. 2.3.3 Statistics

Incidence of wound healing, relapse, major amputation, and mortality were calculated as 5 years cumulated incidence among patients with chronic leg wounds (patients referred to specialist treatment at hospital for >6 weeks) and as incidence rate per 1000 person-years (PY) for both each outcome and among patients with diabetes or PAD. Characteristics of individuals are presented for all outcomes using numbers and percentages.

Cox proportional hazards regression analysis was used to estimate the effects (HRs) of covariates on the four outcomes. Results are presented as crude values and HR adjusted according to the distribution of the other covariates, 95% confidence intervals (CIs) and P values. The proportional hazard assumption was tested using Schoenfeld residuals.19 For variables that did not meet the proportional hazard assumption (cut-off value P < .01), time splitting was performed, and HR was calculated the first year and from start of years 2 to 5. All analyses were performed using Stata version 16 (StataCorp, College Station, Texas).

3 RESULTS

The characteristics of patients referred to specialist treatment at hospital for chronic leg wounds in Denmark from 2007 to 2012 are presented in Table 1. Most patients (76%) had basic or vocational levels of education. Within the study population, 17% had diabetes and 62% had a Charlson comorbidity index score18 of 0.

TABLE 1. Characteristics of patients referred to specialist treatment at a hospital for chronic leg wounds in Denmark from 2007 to 2012 n (%) Total 8394 (100) Mean age (SD) 68.2 (17.7) 18 to 39 612 (7) 40 to 59 1669 (20) 60 to 79 3593 (43) 80+ 2520 (30) Sex Male 4002 (48) Female 4392 (52) Cohabitation Cohabiting 3839 (46) Living alone 4527 (54) Missing 28 (0) Educational level Basic 3525 (42) Vocational 2820 (34) Higher 1204 (14) Missing 845 (10) Place of residence (region) Capital 2928 (35) Zealand 1148 (14) Southern Denmark 1881 (22) Central Denmark 1743 (21) North Denmark 666 (8) Missing 28 (0) Diabetes 1436 (17) Peripheral arteria disease (PAD) 1902 (23) Comorbidity (Charlson index) 0 5227 (62) 1 to 2 2395 (29) ≥3 772 (9) Comorbidity (Charlson modified)a 0 5774 (69) 1 to 2 1958 (23) ≥3 662 (8)

Of patients with chronic leg wounds, n = 7659 (91%) fulfilled the criteria for having a healed wound, and n = 91 (1%) were still in treatment for leg wounds after 5 years (Table 2). Almost one fourth of those who healed had a relapse wound. In total, n = 514 (7%) had a major amputation with 310 of these amputations happening after primary healing. Thirty-nine percent (n = 3240) died during follow-up and 630 of these died with the wound. The largest proportions of deaths occurred among the oldest and those with the most comorbidities. It is worth noting that 29% of those without registered comorbidities also died within 5 years of follow-up.

TABLE 2. Five years cumulated incidence of wound healing, relapse, major amputation, and death among patients referred to specialist treatment at hospital for chronic leg wounds in Denmark from 2007 to 2012 Total n (%)a Wound healing n (%)b Relapse n (%)b Major amputation n (%)b Death n (%)b Total 8394 (100) 7659 (91)c 2012 (24) 514 (7) 3240 (39) Age 18 to 59 2281 (27) 2195 (96) 473 (21) 48 (2) 218 (10) 60 to 79 3593 (43) 3298 (92) 923 (26) 254 (7) 1291 (36) 80+ 2520 (30) 2166 (86) 616 (24) 212 (8) 1731 (69) Sex Male 4002 (48) 3647 (91) 979 (24) 270 (7) 1427 (36) Female 4392 (52) 4012 (91) 1033 (24) 244 (6) 1813 (41) Cohabitation Cohabiting 3839 (46) 3554 (93) 870 (23) 209 (5) 1126 (29) Living alone 4527 (54) 4077 (90) 1135 (25) 304 (7) 2111 (47) Missing 28 28 7 1 3 Educational level Basic 3525 (42) 3216 (91) 916 (26) 257 (7) 1447 (41) Vocational 2820 (34) 2605 (92) 661 (23) 159 (6) 883 (31) Higher 1204 (14) 1129 (94) 268 (22) 43 (4) 317 (26) Missing 845 709 167 55 593 Place of residence (region) Capital 2928 (35) 2654 (91) 647 (22) 150 (5) 1159 (40) Zealand 1148 (14) 1013 (88) 245 (21) 74 (6) 414 (36) Southern Denmark 1881 (22) 1724 (92) 498 (26) 94 (5) 758 (40) Central Denmark 1743 (21) 1603 (92) 414 (24) 139 (8) 622 (36) North Denmark 666 (8) 637 (96) 201 (31) 56 (8) 284 (45) Missing 28 28 7 1 3 Diabetes Yes 1436 (17) 1244 (87) 463 (32) 139 (10) 576 (40) Peripheral arteria disease (PAD) Yes 1902 (23) 1625 (85) 513 (27) 352 (19) 1114 (59) Charlson index 0 5227 (62) 4959 (95) 1152 (22) 249 (5) 1496 (29) 1 to 2 2395 (29) 2093 (87) 685 (29) 195 (8) 1199 (50) ≥3 772 (9) 607 (79) 175 (23) 70 (9) 545 (71) Charlson index modified)c,d, c,d 0 5774 (69) 5432 (94) 1349 (23) 298 (5) 1663 (29) 1 to 2 1958 (23) 1706 (87) 516 (26) 162 (8) 1094 (56) ≥3 662 (8) 521 (79) 147 (22) 54 (8) 483 (73)

Incidence rates of a healed wound were 1812 per 1000 PY, decreasing to 1100 and 1485 PY among patients with diabetes or PAD, respectively. For relapse, the incidence rate was 75 per 1000 PY, which increased to 112 and 107 per 1000 PY among patients with diabetes or PAD, respectively. The incidence rate of amputation was 16 per 1000 PY, which increased to 27 and 66 per 1000 PY among patients with diabetes or PAD, respectively. The mortality rate was 100 per 1000 PY in the total population increasing to 104 and 184 per 1000 PY among patients with diabetes and PAD, respectively.

All four variables were found to be negatively associated with wound healing (place of residence, PAD, diabetes, and Charlson index >0) (Table 3). Place of residence and PAD did not fit the model conditions for Cox regression analysis, and time splitting was performed. It was found that place of residence (living in the Capital or Zealand regions) and PAD decreased the chance of wound healing the first year. Sex, cohabitation, and educational level did not influence chance of wound healing. A sensitivity test was performed with amputation before wound healing being put in the model as competing risk factors with only one significant change in result: older age became a predictor of lower chance of wound healing (age 80+, HR adjusted 0.89 CI 0.81–0.98, P .01).

TABLE 3. Cox regression analysis for predicting wound healing among patients referred to specialist treatment at a hospital for chronic leg wounds in Denmark from 2007 to 2012 with 5 years of follow-up Wound healing Variable Person-years HR crude (95% CI) P HR adjust (95% CI)a P Number of subjects 7431 Age .001 >.10 18 to 39 282 1 (Ref.) 1 (Ref.) 40 to 59 950 0.88 (0.80–0.97) 1.00 (0.91–1.11) 60 to 79 1940 0.84 (0.77–0.92) 1.01 (0.92–1.10) 80+ 1054 0.90 (0.82–0.99) 1.05 (0.95–1.16) Sex .003 >.10 Male 2152 1 (Ref.) 1 (Ref.) Female 2074 1.07 (1.02–1.12) 0.99 (0.94–1.04) Cohabitation >.10 >.10 Cohabiting 2014 1.01 (0.97–1.06) 1.00 (0.95–1.05) Living alone 2194 1 (Ref.) 1 (Ref.) Educational level .01 >.10 Basic 1684 1.08 (1.01–1.16) 1.04 (0.97–1.11) Vocational 1523 1.01 (0.94–1.08) 1.01 (0.94–1.09) Higher 668 1 (Ref.) 1 (Ref.) Place of residence (region) Year 1c <.001 Year 1c <.001 Capital 1828 1 (Ref.) 1 (Ref.) Zealand 769 1.02 (0.94–1.10) 1.05 (0.96–1.14) Southern Denmark 669 1.49 (1.39–1.59) 1.47 (1.37–1.58) Central Denmark 737 1.41 (1.32–1.50) 1.40 (1.30–1.50) North Denmark 206 1.90 (1.73–2.08) 1.88 (1.70–2.07) Place of residence (region) Years 2 to 5c <.001 Years 2 to 5c .02 Capital 1 (Ref.) 1 (Ref.) Zealand 0.75 (0.60–0.89) 0.81 (0.65–0.97)