Potential prognostic factors for delayed healing of common, non‐traumatic skin ulcers: A scoping review

Healing of non‐traumatic skin ulcers is often suboptimal. Prognostic tools that identify people at high risk of delayed healing within the context of routine ulcer assessments may improve this, but robust evidence on which factors to include is lacking. Therefore, we scoped the literature to identify which potentially prognostic factors may warrant future systematic reviews and meta‐analyses. We conducted electronic searches in MEDLINE and Embase to identify studies in English published between 1997 and 2017 that tested the association between healing of the three most common non‐traumatic skin ulcers encountered by health care professionals (venous leg, diabetic foot, and pressure ulcers) and patient characteristics, ulcer characteristics, and results from clinical investigations. We included 42 studies that investigated factors which may be associated with the healing of venous leg ulcers (n = 17), diabetic foot ulcers (n = 15), and pressure ulcers (n = 10). Across ulcer types, ulcer characteristics were most commonly reported as potential prognostic factors for healing (n = 37), including the size of the ulcer area (n = 29) and ulcer duration at first assessment (n = 16). A total of 35 studies investigated the prognostic value of patient characteristics (n = 35), including age (n = 31), gender (n = 30), diabetes (n = 22), smoking status (n = 15), and history of deep vein thrombosis (DVT) (n = 13). Of these studies, 23 reported results from clinical investigations as potential prognostic factors, with the majority regarding vessel quality. Age, gender, diabetes, smoking status, history of DVT, ulcer area, and ulcer duration at time of first assessment warrant a systematic review and meta‐analysis to quantify their prognostic value for delayed ulcer healing.

services, largely because of the volume of nursing time required. For example, in 2016, the annual cost for treating and managing pressure ulcers in the United Kingdom and the United States was estimated at approximately £2.6 billion 5 and $9.1 to $11.6 billion, 6 respectively; for foot ulcer in people with diabetes, estimates were £650 million 5,7 and $9 to $13 billion, respectively. 3 Whilst many venous leg ulcers, pressure ulcers, and foot ulcers will heal, for some people, this will be protracted, with some never fully healing. A recent study of 247 people with venous leg ulcers reported that 62% of ulcers had healed within 24 weeks, with the other 38% remaining unhealed at the time of follow up. 8 Another study 9 on venous leg ulcers reported a median time to healing in three treatment arms of 84, 77, and 91 days. Studies in diabetic foot 10 and pressure ulcers 11 have reported a median healing time of 10 weeks.
Prognostic tools are used in several disease areas to identify patients at risk of a certain outcome and to aid clinical decisions or manage resources. 12 There is scope to use such tools to predict slow healing risk in those with common, non-traumatic skin ulcers. Yet, there is no overarching intelligence about who is likely to heal and who is not, who may benefit from targeted healing-oriented intervention and who will not, and whether we can use resources more efficiently by targeting those at highest risk.
Prognosis research is invaluable in providing ways to answer these questions. However, available tools only use ulcer characteristics (eg, ulcer size, tissue type) to predict healing time, 13 whereas others require variables that may be difficult to measure as part of regular ulcer assessments in practice (eg, ankle-brachial pressure index). 14 To further improve these tools, we need robust evidence on what factors may have prognostic value for assessing the risk of delayed ulcer healing. Parker et al reviewed the literature up to 2013 in order to identify risk factors for delayed venous leg ulcer healing, 15 informing the subsequent development of a prognostic tool. 16 However, they applied a limited set of search terms related to prognostic factors, which was not in line with Cochrane guidance (http:// methods.cochrane.org/prognosis). Furthermore, it is unclear to what extent Parker's findings generalise to other ulcer types.
Therefore, we conducted a scoping review of the literature to gain insight into which factors may have potential prognostic value for delayed healing of several common nontraumatic skin ulcer types, with a focus on factors that can be collected as part of routine ulcer assessments. We expect our findings to guide future decisions about what potential prognostic factors should be prioritised for further investigations through systematic reviews and meta-analyses.

| METHODS
We designed and reported our scoping review guided by Arksey and O'Malley's 17 framework, further recommendations by Levac et al 18 and the Preferred Reporting Items for Systematic Reviews and Meta-Analyses statement. 19

| Search strategy
We searched Medline and Embase via Ovid for publications in English between 1997 and 2017. The search syntax consisted of terms related to the three ulcer types (informed by input from the Cochrane Wounds group, http://wounds.cochrane.org), combined with terms for prognosis, prognostic factors, and prediction models, as recommended by the Cochrane Prognosis Methods Group (http://methods.cochrane.org/prognosis); the full search syntax is available in Appendix S1.

| Study selection
We assessed the eligibility of studies through a two-stage screening process. Studies were eligible if they adhered to the following inclusion criteria: • Conducted in an adult population with venous leg ulcers, diabetic foot ulcers, or pressure ulcers. These three nontraumatic skin ulcers types were selected because of their high prevalence. Other less common ulcer types, such as vasculitic ulcers, were therefore excluded. Traumatic ulcers, such as those arising from surgery or damage to the skin caused by thermal or chemical means, were excluded because of the different aetiology and management of these ulcer types. • Investigating ulcer healing as an outcome. We accepted authors' definitions of ulcer healing and excluded studies that focused on other outcomes, such as ulcer infection or recurrence. • Performed a statistical test on the relationship between individual, potential prognostic factors and ulcer healing as an outcome. We included results of clinical investigations as factors (eg, common blood and urine tests, ankle-brachial pressure index, transcutaneous oxygenation saturation measurements) but excluded those for which measurement would be impractical in the context of routine ulcer assessments (eg, genetic factors). We also excluded papers that did not report significance levels for individual factors when evaluating a score or model combining multiple factors because this would make them ineligible for inclusion in future meta-analyses. • Designed as an observational cohort study. We excluded randomised controlled trials (RCTs) because they aim to determine the effect of treatment on ulcer healing while minimising the impact of other influencing factors, whereas these factors are the primary focus of our review. We did include studies that retrospectively used data from a single arm of a trial. • Original studies, thereby excluding review articles and contributions to conference proceedings.
We first screened the titles and abstracts of all articles. Two authors (D.J. and S.M.) independently screened 50% of articles each and a random sample of 10% in duplicate. Any disagreement was solved through discussion. For all studies deemed relevant, the full text was reviewed using the same screening procedure as in the first stage.

| Data extraction and synthesis
We developed a structured form to aid extraction of items related to: general study characteristics (year of publication, country, study setting), study population (type of ulcer, sample size), length of follow up, if data collection was retrospective or prospective, and association between potential prognostic factors and the outcome (outcome definition, factors for which an association with the outcome had been tested [ie, potential prognostic factors], statistical method used). Data for all papers were extracted by one author (D.J.); data from a random sample of 20% of included studies were extracted independently by a second author (S.M.), with discrepancies solved through discussion.
Two researchers (S.V. and J.T., who is a clinician with experience of managing skin ulcers) independently reviewed all factors and categorised them as patient characteristics (eg, age, comorbidities, history of ulcers), ulcer characteristics and treatment (eg, size, depth, duration), or results from clinical investigations (eg, ankle-brachial pressure index, serum albumin). Ulcer characteristics and treatment were defined as any observation, measurement, or treatment specific to the current ulcer site and immediate surrounding area. Clinical investigations included clinical measurements, clinical imaging, biochemical analysis of blood and urine, and microbiology results. Where needed, the same two researchers independently grouped similar factors into further subcategories to aid the synthesis of results, with discrepancies solved through discussion. For example, the "Patient characteristics" category included subcategories, such as "Socio-economic status" (consisting of, eg, marital status, income, and educational attainment) and "Cardiovascular disease" (covering conditions such as congestive heart failure, peripheral arterial disease, and angina).
In order to identify potential prognostic factors that may warrant a systematic review, we selected factors that had been investigated by at least 10 studies and, for these factors, assessed if they had been defined in ways that were comparable across studies.

| RESULTS
The search yielded 6798 unique studies. We reviewed the full text of 72 studies, of which 42 were included. The most common reason for exclusion was that studies did not report ulcer healing as an outcome. Figure 1 shows the flow chart for study selection.  32 Studies in community settings (2/42) and nursing homes (2/42) were less common.

| Characteristics of included studies
Of the three ulcer types, venous leg ulcers were considered in 17 of 42 of the studies, foot ulcers in 15 of 42, and pressure ulcers in 10 of 42. Sample sizes (ie, participants recruited) ranged from 25 19 to 19 280, 36 with a median of 155 across studies. The majority of studies included fewer than 1000 patients (38/42). The maximum follow-up period was reported in 35 studies and ranged from 4 weeks 19 to greater than 10 years, 20 with a median follow up of 24 weeks. Most studies (38/42) had a follow up of a year or less, with only one study 20 reporting a follow up longer than 2 years. In two studies, follow up depended on length of hospital stay 26 or time to the next clinic visit. 49 Most studies (24/42) defined ulcer healing as a dichotomous outcome indicating whether the ulcer had healed (yes/no); another 10 used time to complete healing. Other definitions were: percentage of ulcer area reduction 39,47 ; healing rate 19 ; and the pressure ulcer scale for healing score, 49 which comprises of sub-scores for ulcer size, tissue type, and exudate. Four studies considered more than one definition. 10,45,53,59 Overall, almost half of the studies (20/42) were prospective, but only three investigated pressure ulcer healing. 38,53,56 The most common statistical method used to analyse the data was regression analysis (30/42), and this was Cox regression in nine cases. t tests were the second most used statistical analysis performed and was used in 15 studies, some of which also used regression. Table 2 shows the result of the categorisation of potential prognostic factors for which studies investigated the association with ulcer healing; see Appendix S2 for the factors as originally reported by the authors of the studies. The number of factors investigated in individual studies ranged from 2 to 45 (median, 15). Factors regarded patient characteristics, ulcer characteristics, or results of clinical investigation in 35 of 42, 37 of 42 and 23 of 42 studies, respectively. Six studies investigated the prognostic value of factors related to only one category, whereas 14 investigated an association with ulcer healing for factors pertaining to two categories, and 22 investigated potential factors related to all three categories.

| Patient characteristics
In the 36 studies investigating patient characteristics, age was investigated as a possible prognostic factor in 31 studies (11 in diabetic foot ulcers, 14 in venous leg ulcers, and 9 in pressure ulcers). This was the most commonly investigated factor, followed by gender, which was considered in 30 studies. Most studies considered age a continuous variable, whereas one used categories. 20 Socio-economic status (eg, marital status, educational level, home ownership) was considered in 10 studies. The three comorbidities most commonly investigated as a potential prognostic factor were diabetes (23/42: 7 diabetic foot, 10 venous leg, and 5 pressure ulcer studies); cardiovascular disease (17/42); and musculoskeletal disease (16/42). The subcategory of diabetes was defined homogeneously as most studies discussed diabetes as a dichotomous variable, labelling patients as diabetic or not. However, cardiovascular disease contained angina (n = 2), peripheral arterial disease (n = 6), and several other conditions grouped under this subcategory. Musculoskeletal disease was also very heterogeneous, with rheumatoid arthritis (n = 6) being the most common factor in this subcategory. Other patient characteristics investigated as possible prognostic factors included: smoking status (15/42), history of deep vein thrombosis (DVT; 13/42), body mass index (BMI; 13/42), and immobility (12/42). While smoking status and history of DVT were defined relatively homogeneously across studies (mostly as current smoker yes/no and prior DVT yes/no), immobility contained many definitions, from walking aid use to paralysis; BMI was considered either a continuous or categorical variable with different categories across studies.

| Ulcer characteristics
Ulcer characteristics were the most commonly investigated potential prognostic factor (39/42). Overall, 14, 15, and 10 studies investigated the association of at least one ulcer characteristic with the healing of diabetic foot ulcers, venous leg ulcers, and pressure ulcers, respectively. Across ulcer types, the size of the ulcer area was most frequently reported (33/42). Most studies (26/42) defined ulcer areas as the complete area of the ulcer in centimetres squared at baseline (ie, time of the first assessment), with only some studies using a different definition (eg, 90% area reduction in area at 4 weeks 48 or area change in first 2 weeks 33 ). Duration of

| Summary of findings
We conducted a scoping review and identified a body of literature investigating potential prognostic factors for healing of venous leg ulcers, diabetic foot ulcers, and pressure ulcers that could be assessed in routine care settings. We included 42 papers, from which we identified age, gender, smoking status, diabetes, musculoskeletal disease, cardiovascular disease, immobility, ulcer area, ulcer location, ulcer duration at time of first assessment, and vessel quality as the factors that were most often investigated as a potential prognostic factor for ulcer healing. Of these, we selected those that were defined sufficiently homogeneously across studies in order to be investigated in a future systematic review and metaanalysis; all selected factors are listed under "Recommendations for future research" below. cochrane.org/) and broadening the scope to include two additional prevalent ulcer types: diabetic foot and pressure ulcers. There were several studies in Parker's review that were captured in our search but did not meet our   Kapoor et al 34 11 Age, gender, history of previous ulcers, immobility [3], incontinence [2], terminal illness   Empty cells mean that no potential factors were identified.

| Relation to other studies
Medications negatively affecting ulcer healing include steroids, immunomodulating drugs, and anti-coagulants known to prolong bleeding time.
Local physical manifestations of poor vessel quality include clinical findings such as lipodermatosclerosis, varicose eczema, hyperpigmentation, etc. PUSH score which uses ulcer characteristics to provide a score between 0 and 10.
inclusion criteria. The main reasons for excluding these papers were: ineligible population (ie, also including people with arterial ulcers), ineligible study design (randomised control trial), ineligible publication type (literature review), or analysing effectiveness of an intervention. Parker's review identified ulcer area, ulcer duration, DVT, and history of previous ulcers as consistently reported risk factors for healing. The former three factors were also identified by our review, which suggests that they might have prognostic value across ulcer types.

| Recommendations for future research
Age, gender, diabetes, smoking status, history of DVT, ulcer area, and ulcer duration at time of first assessment were investigated in a substantial number of studies, with definitions being sufficiently homogeneous. These potential prognostic factors therefore each warrant a dedicated systematic review and meta-analysis. Effect sizes of each factor may vary between ulcer types; applying meta-regression 60 would adjust for this. Another option would be to address each ulcer type in a separate meta-analysis, but this would reduce statistical power. However, there is probably sufficient similarity between ulcer types to warrant a meta-analysis that pools the results across types, thus increasing statistical power. Individual patient data meta-analyses 61 are recommended as the gold standard for reducing heterogeneity and standardisation of definitions, but this will require access to individual patient data of all included studies, which is known to be difficult. Access to individual patient data would also allow access to raw data and not rely on categories used in the studies, which would open up possibilities to investigate additional factors in a meta-analysis, such as BMI. Many studies included in our review had short follow-up times and small sample sizes, which will negatively affect the precision of the individual study effect sizes for the potential prognostic factors. Future meta-analyses will increase this precision by combining evidence across studies into a single combined estimate of prognostic effect size, which will contribute to a better understanding of prognostic factors for ulcer healing. However, the robustness of findings from meta-analyses investigating prognostic factors depends on the quality and risk bias of included studies, 17 and future systematic reviews and meta-analyses would need to assess this, for example, using the Quality in Prognostic studies (QUIPS) tool. 62 Results from clinical investigations may have prognostic value for ulcer healing, but the current evidence in observational studies is too scarce to warrant systematic reviews in this area. Therefore, future research may focus on conducting larger cohort studies investigating the association between results from clinical investigation and ulcer healing. While we excluded comparative effectiveness studies, a recent Cochrane review 63 included RCTs to investigate the prognostic value of protease for the healing of venous leg ulcers, resulting in the inclusion of 11 studies for meta-analysis. A hierarchical approach or down-weighting of observational data through the use of a power prior would be advised 64 in this circumstance.

| Strengths and limitations of this review
To our knowledge, this scoping review is the first to investigate potential prognostic factors across the most common ulcer types. We used rigorous and transparent methods, including a reproducible search strategy; duplicate assessment of articles for relevance independently by two authors, and use of a pre-determined data extraction template. A recent study 65 assessing the methodological quality of 494 scoping reviews found that only 22% reported a search strategy, 36% had used two reviewers, and 43% had a data extraction template.
Our scoping review also has limitations. We did not consult stakeholders as the last phase of the review process. 66 This may have resulted in additional relevant studies being missed. In addition, we excluded studies on the potential prognostic value of genetic factors because they are currently difficult to include in routine ulcer assessments. Future scoping reviews should investigate this further as soon as simple genetic tests become available.

| CONCLUSION
Age, gender, diabetes, smoking status, ulcer area, ulcer location, and ulcer duration at time of first assessment are potential prognostic factors that warrant a systematic review and meta-analysis to quantify their value for predicting delayed healing of common non-traumatic skin ulcers during routine assessment. This will contribute to optimising our understanding of ulcer healing and provide valuable information for clinical practice and guidelines.