Workstream 1: cross-sectional survey and cohort study of SWHSIs

The dearth of national and international clinical and research information regarding the prevalence, patient characteristics and treatments for SWHSIs meant that we did not know the extent and nature of these wounds, how and where they were managed or the associated outcomes. To enable accurate assessment of time to healing and treatment use, we planned an inception cohort study to address key uncertainties. An inception cohort approach was essential to allow accurate assessment of SWHSI duration. Planning the cohort study was difficult, however, as there were no data regarding where most SWHSIs occurred. For example, were most SWHSIs left open in theatre or did they subsequently break down in hospital or in the community? We needed basic information regarding the numbers of patients with SWHSIs and their location in order to put adequate resources in place for the inception cohort study. We therefore preceded the cohort study with a cross-sectional survey. This approach facilitated the gathering of initial intelligence to support subsequent work and, in addition, the use of a cross-sectional design allowed us to estimate a point prevalence estimate for SWHSIs. The cross-sectional survey has been published (see Appendix 1).16 The cohort study has also been published (see Appendix 2).17

Cross-sectional survey

Objectives

The cross-sectional survey aimed to capture information to aid the design of the cohort study. This included the:

• number of patients with a SWHSI (point prevalence), their characteristics and the NHS settings where they were receiving treatment
• proportions of SWHSIs planned prior to surgery and those that occurred as a result of wound breakdown (dehisced) after surgery.

Information was also collected on typical durations of SWHSIs, the types of surgery preceding the SWHSI and the treatments received by patients.

Methods

The cross-sectional survey was conducted over a 2-week period (spring 2012) in primary, secondary and community care settings in Hull and the East Riding of Yorkshire, UK. We asked health-care professionals to identify patients on their case load who had at least one SWHSI (i.e. a wound deliberately left open owing to infection, swelling, contamination or empty tissue space, a wound initially closed but subsequently dehisced or a wound arising from surgical debridement). When patients had multiple SWHSIs, the largest wound was deemed to be the ‘reference SWHSI’. Detailed data were collected, including patient and wound characteristics, health-care provider, clinical and surgery details, and treatment events (see Appendix 3). As this survey was limited to secondary use of routinely collected information, Research Ethics Committee approval was not required.18 Approval was, however, obtained from associated NHS trusts prior to commencement of data collection.

Data analysis was conducted using IBM SPSS Statistics, version 21 (IBM Corporation, Armonk, NY, USA). Categorical variables were summarised as frequencies and proportions, whereas continuous variables were summarised as medians and associated 95% confidence intervals (CIs) and ranges. The usual resident regional population aged ≥ 20 years, as taken from the 2011 census, was used as the denominator when calculating point prevalence.19

Results

Following removal of duplicate cases (identified through age, ethnicity and wound characteristics), data were analysed for 187 patients with a SWHSI. In total, 62% of patients were male, 95.2% were of white British ethnicity and the median age was 58.0 years (range 19–90 years), as detailed in Table 1.

We estimated the point prevalence of treated SWHSIs as 4.1 per 10,000 population (95% CI 3.5 to 4.7 per 10,000 population), using the resident regional population value as the denominator.19 The observed maximum number of SWHSIs per patient was four (1/187, 0.5%), with the majority of patients (164/187, 87.7%) having only one SWHSI. Patients were more frequently treated in community than in secondary care settings (109/187, 58.3% vs. 56/187, 29.9%, respectively), with the remaining patients treated within primary care (8/187, 4.3%) or other care settings (2/187, 1.1%). There were 12 patients (12/187, 6.4%) for whom treatment location was not recorded.

We found that almost half of the SWHSIs were planned to heal by secondary intention (n = 89, 47.6%) and 77 (41.2%) were primarily closed wounds that had subsequently dehisced. A further six wounds (3.2%) were surgically opened, six (3.2%) arose for other reasons [surgical debridement (n = 2, 1.1%), sutures removed and left open to heal (n = 1, 0.5%), skin graft failure (n = 1, 0.5%), non-healing wound (n = 1, 0.5%) and necrotising fasciitis requiring debridement (n = 1, 0.5%)] and for nine patients (4.8%) the information was not known or was missing. The median time from surgery to wound breakdown was 9 (95% CI 7 to 10) days.

In addition, we found that the median duration of wounds at the point of survey was 28 days (95% CI 21 to 35 days). Fully dehisced SWHSIs had the longest median duration (35 days, 95% CI 15 to 150 days), and those with long wound durations were most frequently treated within primary care (median duration 112 days, 95% CI 21 to 469 days) and community settings (median duration 35 days, 95% CI 28 to 56 days) (Table 2).

In total, we identified 43 different surgical procedures that preceded the development of a SWHSI. The most common surgical procedures were for pilonidal sinuses/abscesses (28/187, 15.0%), lower-limb amputations (19/187, 10.2%) and laparotomy with bowel resection (19/187, 10.2%). As detailed in Table 3, the most common surgical specialties associated with SWHSIs were colorectal (80/187, 42.8%), plastic (24/187, 12.8%) and vascular (22/187, 11.8%) surgery.

At the time of the survey, most patients (184/187, 98.4%) were receiving active treatment for their wound. Dressings were the most common single treatment, being received by 169 (93.4%) patients. Eleven (6.1%) patients were receiving NPWT (10 patients in secondary care and one patient in the community setting). One patient (0.5%) was receiving larval therapy.

Conclusions

This cross-sectional study is the first to characterise SWHSI patients and their wound origin, duration and treatment.

This study provided initial data on wound and treatment characteristics, for which there were limited data when the research programme was conceived. Identifying common surgical specialties (within secondary care) that lead to SWHSI, that almost 50% of SWHSIs were planned and that the most common treatment location was the community helped us to plan recruitment strategies (i.e. identification of planned SWHSIs in advance of surgery) and time frames for both the cohort study and the subsequent pilot, feasibility RCT (workstream 4), and will also help to inform research studies in the future.

A total of 43 surgical procedures preceding SWHSIs were identified, with common associated surgical specialties being colorectal, plastic and vascular surgeries. SWHSIs occur in a wide range of specialties; however, there are specific populations in whom SWHSIs more frequently occur and so these may be appropriate as the focus of further research.

Data collected in relation to wound duration at the point of the survey have enabled us to estimate the duration of follow-up required to capture healing events. Again, this information helped to inform our design of the cohort study and pilot, feasibility RCT (workstream 4).

When this research programme commenced, we had identified the need for a rigorous population-based estimate of SWHSI prevalence. This study has provided a population-based estimate of treated SWHSI prevalence, which has suggested that these wounds are relatively common. Given the prevalence observed, this therefore supports the need for further research into SWHSI epidemiology and treatments.

Cohort study

Objectives

Following the cross-sectional survey, we conducted a prospective, inception cohort study with the following objectives:

• to investigate the clinical characteristics of patients with SWHSIs from the point of inception of the wound
• to clearly delineate the clinical outcomes of patients with SWHSIs over the lifetime of their wound, with a particular focus on time to wound healing and associated determinants
• to describe the types of treatments received by those with a SWHSI
• to assess the measurable impact SWHSIs have on patients’ quality of life.

Methods

The cohort study was conducted over a 21-month period (18 February 2013 to 30 November 2014) in primary, secondary and community care settings at eight study sites across Yorkshire and the Humber, UK. Prior to commencement of the study, approval was obtained from Yorkshire and Humber – Humber Bridge Research Ethics Committee (reference 12/YH/0350) and from the associated NHS trusts.

Patients were eligible for inclusion if they had a SWHSI, defined as ‘an acute (< 3 weeks’ duration) open wound, resulting from surgery and requiring treatment, which was healing from the bottom up by the formation of granulation tissue’. When patients had multiple SWHSIs, the largest wound was deemed to be the ‘reference SWHSI’ and detailed data were collected for this one wound. Health-care professionals in primary, secondary and community care settings initially identified and screened patients for eligibility. Clinical areas identified as having a high prevalence of SWHSIs were targeted for recruitment, including colorectal and vascular surgical wards. Patients meeting the eligibility criteria were approached for informed consent by a research nurse before any data collection. A formal sample size calculation was not required for this study, although it was anticipated that 443 participants could be recruited over a 6-month period.

Data, including patient, surgery and wound information, quality of life and a wound photograph (used to assess wound features), were collected at baseline. Participants were followed up every 1–2 weeks for a minimum of 12 months and a maximum of 21 months to collect key clinical outcome data [including healing through either participant self-reporting or clinician observation, defined as ‘complete epithelial cover in the absence of a scab (eschar)’]; key clinical events such as surgical site infection (in accordance with Public Health England guidance on surgical site infection surveillance) or hospital admission;20 SWHSI treatments received; and any changes to a participant’s study involvement, including death. To ensure accurate data collection, healing status was further verified through a home visit to a sample of patients (10–20%). Participants also completed quality-of-life [Short Form questionnaire-12 item (SF-12)21 and EuroQol-5 Dimensions, three-level version (EQ-5D-3L)22] and pain [Brief Pain Inventory (BPI)23] assessments via postal questionnaires at 3-monthly intervals for a minimum of 12 months and a maximum of 21 months.

Data analysis was conducted using Stata^® version 13 (StataCorp LP, College Station, TX, USA). Summary statistics for all variables were generated. Association with healing was examined for prespecified variables and multivariate logistic regression analyses conducted. Kaplan–Meier curves and Cox proportional hazards model analyses were used to examine time to event data. SF-12,21 EQ-5D-3L22 and BPI23 data were summarised, with SF-1221 data subsequently analysed using linear mixed models.

Results

A total of 396 patients were recruited to the cohort study from primary and acute care settings in Yorkshire, UK, with 393 participants included in data analysis (three patients were found to be ineligible following recruitment and so were excluded from the analysis). Just over half of participants were male (n = 222, 56.5%), participants had a median age of 55 (range 19–95) years and 69.5% of the study population were classed as overweight or obese (31.3% and 38.17%, respectively). Current smokers made up 28.5% of participants (i.e. were currently smoking or had quit within the last year).

The most common comorbidities we observed were cardiovascular disease (n = 151, 38.4%), followed by diabetes mellitus (n = 103, 26.2%) and peripheral vascular disease (n = 57, 14.5%). The majority of participants had no previous history of SWHSIs (n = 93, 72.0%) and had only one SWHSI (n = 358, 91.0%), and their surgical wound had been planned to heal by secondary intention (n = 236, 60.1%). The median area of the reference SWHSI at baseline was 6 cm² (range 0.01–1200 cm²). SWHSI location varied; however, the most common sites were the abdomen (n = 132, 33.6%), foot (n = 59, 15.0%) and leg (n = 58, 14.8%), linking the commonly represented surgical specialities of colorectal (n = 136, 39.7%) and vascular (n = 82, 20.9%). At baseline, 164 patients (41.7%) were receiving hydrofibre dressings and 89 patients (22.7%) were receiving NPWT. Further details of baseline demographics (patient, wound and surgery) and treatments received are presented in Tables 4–7.

The mean length of participant follow-up was 499 days [standard deviation (SD) 127.4 days], with a range of 13–651 days (median 528 days). Sixty-six participants (16.8%) were not followed up for the entire study duration (31 withdrew, 29 died and six were lost to follow-up). When the reference SWHSI was situated on a limb, this was amputated for 3.3% of participants (n = 13), of whom 61.5% had experienced an infection during follow-up (n = 8) and 76.9% had diabetes mellitus (n = 10). Infections were experienced by 32.1% of study participants (n = 126). Hospital admissions were reported for 97 participants (24.7%), of which 66 participants (68.0%) returned to the operating theatre. Thirty-six cases (38.1%) of hospital admissions were related to SWHSIs.

Data indicated that the reference SWHSI healed for 81.4% of participants (n = 320) during the study. Agreement between patient self-report and nurse assessment of healing status verification was relatively high (77.8% at 6 months and 74.2% at 12 months). Disagreements were present for four participants: three patients had healing confirmed during a visit but paperwork was not completed to this effect and one participant was not healed at the verification visit but healed shortly afterwards. The median time to healing for all cohort participants was 86 (95% CI 75 to 103) days, as presented in Figure 2. When we assessed time to healing against baseline participant and wound characteristics in a Cox proportional hazards model, baseline wound area above median (p < 0.01) and surgical wound contamination level as determined at the point of surgery (p = 0.04) were significant predictors of prolonged time to healing (see Appendix 4). Infection at any point was also found to be a significant predictor of prolonged time to healing (p < 0.01).

FIGURE 2 Cohort study: Kaplan–Meier curve of time to healing in days. (Figure not reproduced here; see the original article.)

The most commonly received treatment throughout the study was hydrofibre dressings, with 65.9% (n = 259) of participants receiving this treatment at least once. Nearly one-third (n = 115, 29.3%) of participants received NPWT at some point during the study.

Health utility scores captured by the EuroQol-5 Dimensions (EQ-5D) measure22 (in which 0 = death and 1 = perfect health) were a mean of 0.5 (SD 0.4) at baseline, increasing to a mean of 0.6 (SD 0.4) at 6 months and 0.6 (SD 0.4) at 12 months. Further details are provided in Table 8.

Pain, measured as BPI scores for pain severity (in which 0 = no pain and 10 = pain as bad as you can imagine) and pain interference (in which 0 = does not interfere and 10 = completely interferes)23 reduced slightly over time. Mean pain severity reduced from 3.7 (SD 2.54) at baseline to 3.0 (SD 3.0) by 15 months; similarly, mean pain interference scores reduced from 4.2 (SD 2.54) at baseline to 3.1 (SD 3.17) at 15 months. Further details are provided in Table 9.

Quality of life was measured using the SF-12 and presented as SF-12 physical component scores (PCSs) and mental component scores (MCSs).21 At baseline, the mean PCS was 33.1 (SD 10.7) and the mean MCS was 42.2 (SD 12.35). At 15 months’ follow-up, the mean PCS was 40.8 (SD 13.46) and the mean MCS was 48.3 (SD 10.73). Further details are provided in Table 10. A linear mixed model, adjusting for duration of SWHSI, baseline SWHSI area, location, participant age and baseline SF-12 subscale score, found follow-up time point, baseline score, age and SWHSI duration to be significant predictors of PCS (p < 0.001 in all cases) and MCS (p = 0.03 for time point and p < 0.01 for age, baseline score and SWHSI duration).

Conclusions

This prospective inception cohort study in patients with SWHSIs is the first we have identified. Broad patient groups (patients undergoing abdominal colorectal surgery, lower-limb vascular surgery and a mixed group of other procedures and specialities) were identified, which were similar to those identified in our cross-sectional survey. This information flags populations of clinical relevance, as well as those likely to be the focus of further research.

The median time to healing for these wounds, of approximately 3 months, is new information, which highlights the extended period of time that patients might expect to cope with open surgical wounds. There was high variability, with time to healing ranging from 8 days to 1 year and 4 months (prior to censoring) and from 5 days to 1 year and 7 months (following censoring). The chronicity of these wounds is crucial for patients and carers to appreciate, in order for them to have realistic expectations regarding wound healing. The cohort also highlights the large number of adverse events that occur in SWHSI patients, with wound infection and readmission to hospital (with the associated care costs) being particularly common. This cohort study data clearly demonstrated that SWHSIs pose a distinctive management challenge that was previously unsupported by high-level evidence.

Adjusted analyses provide further initial insights into possible patient and wound ‘risk’ factors that might have a detrimental impact on healing, including wound infection at any point, baseline wound area and higher levels of surgical wound contamination, as determined at the point of surgery. The predictive baseline factors may be important for the stratification of randomisation or adjustment for prognostic factors in future trials; however, it should be noted that there are many other covariates that are associated and correlate with these predictors. The SF-12 data indicate that, at baseline, SWHSI patients have quality-of-life limitations comparable to patients with congestive cardiac failure; however, they were significantly younger.24 This therefore demonstrates the potential impact that SWHSIs have on patient quality of life. However, as the improvements in SF-12 scores during the study demonstrate, quality of life may be improved by time or wound healing. The research to date is therefore building a picture of a relatively common wound type that takes a prolonged time to heal, negatively impacts quality of life and puts patients at high risk of infection and other adverse events.

The common treatments for SWHSIs, in the centres involved in this cohort that were treating SWHSI patients across Yorkshire and the Humber, UK, were dressings and NPWT. There is currently limited and generally low-quality evidence on how effective dressings are in terms of wound management and promoting optimal clinical outcomes in a cost-effective way.14 A recent Cochrane systematic review found only two RCTs, with a total of 69 participants, that had investigated the relative effectiveness of NPWT on SWHSIs, both reporting limited outcome data, which prevents any firm conclusions to be made.14 Given that all SWHSI patients in this cohort were receiving dressings and/or NPWT, further research to explore clinical effectiveness and cost-effectiveness of these treatments in one or more of the common surgical groups identified seems warranted, and so was considered appropriate for further assessment in workstream 2.

The prospective and observational design of this study allowed us to follow patients over time and to collect the important natural history, treatment and outcome data presented here. Although we tried to maintain a representative sample, the need for consent meant that the subset recruited might differ in some systematic way from the wider SWHSI patient population. Comparison of the cohort and the cross-sectional survey data may be useful here, as the survey data were captured away from the patient’s bedside, did not require consent and thus may provide a more epidemiologically complete overview of the SWHSI population. In terms of patient, epidemiological, wound and surgery data, the populations in the survey and cohort appear similar; however, the findings from the survey may have influenced specific areas for focused recruitment into the cohort study (e.g. subspecialty-specific hospital wards). Some patient groups for whom wound problems are recognised nationally seem under-represented in the cohort study (e.g. post-caesarean wounds, for which dehiscence rates as high as 6% have been reported).25

The cohort study was conducted in a single geographical area (two large centres in Yorkshire, UK), and therefore the patient groups and the treatment availability observed may not be reflective of national or international patient populations or treatment availability. However, we have no reason to suspect that there are huge differences between these surgical patients and those elsewhere, particularly given the breadth of our inclusion criteria.

We also note that our analyses exploring the impact of factors on time to healing are based on our observational data set and thus these cannot be assumed to be causal relationships. Thus, firm conclusions regarding factors that may affect healing cannot be drawn. However, we are able to report the associations found between observed covariates and healing, and reflect on them. From the outset of the study, there was careful consideration regarding measured study variables, with the aim of capturing important data, including factors that had the potential to be prognostic in relation to healing. We consulted widely with clinical experts and extensively scoped the literature, with the aim of providing the most comprehensive analysis possible.

The further research conducted as part of this research programme (workstreams 2–4) focused on further investigation of interventions that may offer clinical effectiveness and cost-effective treatment options for SWHSIs (workstream 2), as well as more detailed exploration of patients’ experiences of living with SWHSIs and health professionals’ views of delivering care to patients with these challenging wounds.