What do we now know about the SWHSI patient population?
This research programme emerged from clinical uncertainty about the best way of managing SWHSIs. To the best of our knowledge, there has been very little previous research on open surgical wounds, and what small amount there has been has focused only on dehisced surgical wounds;64,65 we have now shown that these constitute approximately half of all SWHSIs. We have studied both dehisced surgical wounds and those intentionally left open after surgery because they present similar challenges for clinical management. We planned this research at a time (2009) when the use of a costly intervention, NPWT, was increasing; however, access to it varied hugely, based on clinical setting, local policies and resources. Prior to this research, little was known about the frequency of SWHSIs, the characteristics of the patients who experience them and longer-term outcomes. Without basic information about the frequency and prognosis of a health condition and how patients are affected by it, it is impossible to plan research on treatments.66 Initially, we undertook a cross-sectional survey of the prevalence of treated SWHSIs, which had the dual purpose of deriving the first population-based estimate of prevalence, as well as identifying where and when SWHSIs occurred, to aid the planning of an inception cohort. The cross-sectional survey (conducted in 2012) was anonymised, completed away from the patient and did not require consent in order to optimise data capture. To the best of our knowledge, this survey provides the first estimate of the prevalence of treated SWHSIs, which we estimate at 4.1 per 10,000 of the population (95% CI 3.5 to 4.7).16 There are several things worth noting about this estimate: the prevalence is similar to best estimates of the UK prevalence of leg ulcers (4.4 per 10,000),27 almost half of SWHSIs (48%) were planned (before surgery) to be left open, the wounds had been present for a median of 28 days, the patients were mainly receiving care in primary and community care settings and 93% of patients were receiving wound dressings (6% had NPWT). Colorectal surgery was the most frequent type of surgery to precede the SWHSI, although plastic surgery and vascular surgery were also common.
This survey, although rather basic in design (lacking independent case ascertainment, for example, due to resourcing), provided crucial intelligence for planning the cohort study. Our median time from surgery to wound breakdown (in those whose wounds were not intentionally left open) of 9 days accords with previous studies of dehiscence.64,67
Our inception cohort collected data from 393 patients with SWHSIs and, to the best of our knowledge, is the first detailed analysis of this patient population, their wound treatments and healing trajectories. A notable difference between this and the cross-sectional survey (apart from the longitudinal follow-up) was the requirement for informed consent, which may have influenced participation. In total, 57% of the cohort were men, the overall median age was 55 years, 70% of the cohort were overweight or obese (higher than the 2015 national average of 63%)68 and 29% were smokers or had quit in the last 12 months (compared with 19% of the adult population in Great Britain).69 Common comorbidities were cardiovascular disease (in 38%), diabetes mellitus (in 26%) and peripheral vascular disease (in 14.5%). The wounds of 60% of the cohort participants were planned to be left open before surgery, higher than for the cross-sectional survey in which 48% were planned. This probably reflects the relative ease of recruitment of participants to the cohort, when it was known in advance that their wound would be open after surgery.
Our research illustrates, for the first time, the heterogeneous nature of the population of people with SWHSIs and of their wounds. Although it was anticipated that SWHSIs would stem from a range of surgical specialties, the distribution of SWHSI sites is noteworthy (34% abdominal, 15% feet, 15% leg) and anatomical site is very likely to influence the impact of the wound on quality of life (if it influences mobility or ability to drive, for example). The clinical specialty under whose auspices the preceding surgery took place was most commonly colorectal (40%), but also commonly vascular (21%). This information is vital to the planning of future research to ensure that the studies are carried out in the populations most affected. Moreover, these subpopulations of people with SWHSIs are likely to demonstrate very different healing trajectories from one another (an open wound on the foot in somebody with peripheral vascular disease and/or diabetes mellitus is likely to heal differently from a partially dehisced abdominal wound post appendectomy). Those conducting future trials in this area will need to make decisions about trading off ease of recruitment of a heterogeneous population (with consequent ‘noisy’ healing data) with recruitment from smaller, focused patient populations. This is illustrated by our pilot trial, to which we recruited 40 eligible participants, 60% of whom had a SWHSI on the foot (compared with 15% in the cohort study). The nature of those recruited to the pilot trial meant that the time to healing was much longer than for the cohort study and far fewer healing events were observed (only 25% healed), despite the 90-day follow-up period. The reasons for the excess of patients with foot wounds in the pilot trial are unclear, but it is likely that they were more readily recruited for some reason. It is possible that the current enthusiasm for NPWT as a treatment for SWHSIs meant that some health professionals were not in equipoise and were unwilling to randomise, thus making it difficult to recruit specific patient subgroups.
Future trials will therefore need to be carefully planned, with a desired population in mind and appropriate targeted recruitment. The cohort study time to healing data will be invaluable to the planning of new research, particularly in relation to appropriate follow-up time frames.
Exploratory analysis of predictors of healing within the cohort identified larger-area surgical wound contamination classification (as judged at the point of surgery)70 and infection at any time point as predictors of slower healing. The median time to healing of 86 (95% CI 75 to 103) days is very similar to that for venous leg ulcers (84 days in typical venous ulcer patients in a recent trial).71 The typical healing trajectory for people with a SWHSI is not smooth, with episodes of infection and hospital readmission commonplace. The relatively young age of SWHSI patients compared with other patient groups with chronic wounds (e.g. leg and pressure ulcers) is particularly noteworthy. The wide-ranging impact on quality of life was detectable using the generic EQ-5D and SF-12 instruments, and was both physical and mental.21,22 We heard directly from the patients themselves of the often devastating impact on their lives. Patients often felt isolated and confined to the home with disruption of working and caring responsibilities, causing feelings of dependency and burden. There were frequently financial impacts of the SWHSIs on the person and their family. To the best of our knowledge, the cost to the NHS of a SWHSI had not previously been estimated. We estimated that the mean cost of a SWHSI to health services was £1060 before the inclusion of treatment costs. Use of NPWT was estimated to cost, on average, an additional £1323 per month and use of dressings was estimated to cost, on average, an additional £441 per month.
What do we now know about treating SWHSIs?
Prior to this research, we had no information about treatments that were being used for SWHSIs. From our research we have learned that SWHSI treatments fall into three main categories: dressings, NPWT and further surgery. We observed a plethora of treatments being used for SWHSIs; however, hydrofibre dressings were the most commonly used (66% of patients received these at some time), alongside a high use of NPWT. Although the cross-sectional survey identified NPWT use in only 6% of patients, the cohort study indicated that a much larger proportion (29%) were treated with NPWT at some time.
Dressings are the cornerstone of treatment for complex wounds and there are many categories available, ranging from simple, basic wound contact dressings to ‘advanced’ dressings containing ingredients such as silver.29 However, the evidence for the effectiveness of specific dressings and NPWT for SWHSIs is sparse and of low quality.12 The most recent systematic review of the effects of dressings for open surgical wounds highlighted the paucity of evidence and a review of the effects of NPWT on open surgical wounds (search date June 2015) similarly concluded that there is currently no rigorous evidence available and therefore the benefits and harms are unclear.14 There is an increasing amount of research regarding NPWT for the primary prevention of surgical site infections in closed surgical wounds; however, its effects in this context also remain unclear.72,73 There is some evidence of potential benefit of NPWT for foot wounds in diabetes mellitus; however, this evidence is of low certainty due to its low quality.31
We planned to construct a cost-effectiveness model for NPWT compared with dressings for open surgical wounds using a combination of published and cohort study data. We were unable to populate this model with published data as there were no data reporting time to healing as an outcome in RCTs of NPWT in this patient population. Using the cohort data alone to compare the effects of NPWT and dressings on time to healing effectiveness, and after using different approaches to adjust for confounding, we concluded that NPWT is less effective and more costly than wound dressings for SWHSIs. We had originally intended to conduct a value of information analysis to determine the potential impact of uncertainty on the decision regarding NPWT to treat SWHSIs. This analysis would assess the health and cost implications of the wrong decision being made and thus inform investments in future research to mitigate these uncertainties. However, our analyses showed that uncertainty around the decision not to use NPWT was very small. Even with the use of advanced adjustment techniques, such as those applied here, observational data may still affect the conclusions. Meta-analyses comparing the findings of RCTs and cohorts designed to answer the same clinical question give an OR of 1.04 (95% CI 0.89 to 1.21), highlighting the potential for agreement between the designs and the associated uncertainty.74 The current economic evaluation uses all available evidence on the clinical effectiveness and cost-effectiveness of NPWT; thus, we argue that it should be seriously considered. However, the potential for unresolved confounding may limit confidence in the associated conclusions and therefore this may lead to justified calls for definitive RCT evidence to be obtained.
Interviews with patients and health professionals provided further insights into the use of, and views about, treatments for SWHSIs. The wide range of dressings used and inconsistency of use was noted by patients. Sharp, sometimes painful, debridement of SWHSIs was also reported by patients, despite the lack of high-quality evidence to support its use.75 However, debridement was not measured explicitly as a treatment in the cohort study, so further data are needed on how widespread sharp debridement is as a treatment for SWHSIs. Finally, although our study participants generally viewed NPWT as an effective treatment, they reported various drawbacks, including pain during dressing changes. Some study participants reported feeling too embarrassed to go outside the home with NPWT in situ, due to the bulky appearance of the equipment and their fears that the exudate and odour might be detected by other people. Nurses’ varying levels of expertise in use of NPWT appeared to provoke anxiety in patients. These findings mirror those from other small qualitative interview studies with patients receiving NPWT.76–78
What do we now know about living with a SWHSI?
The interviews with patients, together with the cohort HRQoL data, show the devastating effect that open surgical wounds can have on those affected. For those who did not expect to have an open wound after surgery, there were feelings of shock, fear and anxiety; these are feelings akin to those reported by people who have undergone life-changing trauma.79 Significantly, the presence of an open wound resulted in the withdrawal of many people from normal employment, leisure and social activities. During interviews, participants indicated that their physical and mental resources had become depleted through the dominance of the wound in their lives.
We note the low baseline mean SF-12 PCSs and MCSs for SWHSI patients [mean PCS 33.1 (SD 10.17); mean MCS 42.2 (SD 12.35)]. These values, for people who have recently undergone surgery and who have an open wound, are lower than the mean PCS reported for an, on average, older venous leg ulcer population [mean PCS 38.4 (SD 11.2); mean MCS 48.6 (SD 12.0)],71 and are also lower than mean scores for other patient groups, such as those with coronary heart disease and arthritis.80,81 However, the SWHSI cohort mean scores are similar to postoperative SF-12 data from 85 participants with mainly closed wounds [mean PCS 35.1 (SD 7.8); mean MCS 43.5 (SD 12.4) 30 days post surgery].82 Thus, it is difficult to separate the impact of surgery from the impact of an open wound on HRQoL. In the SWHSI cohort, we noted a significant improvement over time in both PCS and MCS, suggesting that the SF-12 is responsive to changes in HRQoL in this population. Improved quality of life was likely to be the result of recovery from surgery, changes in the underlying condition the surgery aimed to ameliorate or resolve and wound healing. The baseline EQ-5D index score was found to be, on average, 0.43 (SD 0.45). Modelling results indicated that wound healing was associated with a small (but statistically significant) increase in EQ-5D index score (mean 0.055, SE 0.02). This level of improvement is smaller than what has been detected for the healing of other wound types,71 but this may be due to different estimation methods. This value is also within the range of minimal clinically important differences evaluated across a wide range of conditions.83,84
Although baseline HRQoL was relatively poor in this SWHSI cohort, pain severity and interference scores of 4 (SD 3) (measured using the BPI)23 were indicative of relatively mild pain when compared with, for example, the mean pain severity score of 7.0 (SD 1.8) and pain interference score of 7.6 (SD 2.0) reported in 440 patients with known chronic, non-malignant pain.85 In addition, for the SWHSI cohort, mean pain scores reduced, on average, by only 1 point, to 3 points, by 15 months, follow-up. In interviews, some patients discussed pain and its impact, but it did not seem to be a key topic of concern and was not an obvious source for the overall poor HRQoL observed in those with a SWHSI.
Our findings, from exploring living with a SWHSI with patients, illuminate the significance of psychosocial effects, due to diminished sense of self, inability to fulfil social roles and obligations, and curtailment of normal social activities, especially when patients’ hopes for healing are not met. Our study also highlights the impact of financial pressures on younger patients whose wound rendered them unfit for work, and their concerns about consequences for their family and implications for employment in the longer term.
From a patient perspective, the aspiration for complete wound healing was a major focus of all those interviewed. Healing was largely linked to return to some semblance of life prior to surgery, and was considered a sign that the body was returning to ‘normality’. This focus on healing as a key outcome agrees with previous findings from interviews with 33 patients with a range of complex wounds, including leg ulcers, foot ulcers and surgical wounds.27 Here, most people, again, said that being healed or ‘cured’ was their main goal, as did eight participants with complex wounds who, when responding to the question ‘what did you most want from treatment?’, reported that they most wanted the wound to heal.46
Health professionals’ management of patient expectations is crucial to the patient experience. We know from our cohort study that, although the median time to healing was around 3 months, some people would live with a SWHSI for significantly longer than this. We heard how comments from health-care professionals regarding healing and the condition of their wound had profound impacts on patients. Positive feedback was a boost for morale and sustained the patient’s hope for healing, whereas negative remarks adversely affected the patient’s general outlook. Conflicting information led to feelings of confusion and consternation in patients. Even quite casual remarks from health-care professionals could have a profound impact, which has been noted in other recent research on patients with complex wounds.86 Patients’ expectations for wound healing may often appear unrealistic, yet clinicians need to manage these expectations within a framework that allows patients some measure of hope for healing, while avoiding the arousal of ‘false hope’, which may result in feelings of disappointment, disillusionment and distrust, which undermine professional–patient relationships.46 Open, non-healing wounds that endure, such as SWHSIs, are associated with the ‘forever healing’ process, characterised by a constant need for treatment and care and a diminished quality of life, as described in patients with venous, diabetic and pressure ulceration.47–51,53,87 This is challenging for clinicians, who must communicate uncertainty regarding wound progress and likely treatment outcomes without raising false hope or inducing a ‘spiral of hopelessness’ in patients.51
Given the severity of impact that these wounds have on patients’ well-being, it is not clear whether or not psychological interventions may be useful in helping patients with hard-to-heal wounds to accept and adapt to their non-healing wound. Our findings also point to the need for improvements in communication with health-care professionals concerning the slow healing processes associated with SWHSIs; further research is required to investigate how, when and by whom information should be conveyed to patients with these types of wounds. Study (nurse) participants also highlighted the need for patients to receive written, as well as verbal, information, including information about signs and symptoms of infection and what action to take should these occur.
Strengths of the research
A key strength of this research is the new information that has been captured and reported on this SWHSI patient population. Prior to this research, we did not know the prevalence of SWHSIs, the surgeries involved, how many SWHSIs were planned compared with how many formed spontaneously, how costly these wounds are to the health service, how best to treat these wounds and how best we might direct future research on SWHSIs. There was also an absence of rigorous data regarding SWHSI management in the UK, how long they took to heal and what the impact of these wounds is on patients and their families. This research has provided this information for one geographical area of the UK (Yorkshire and the Humber).
We present data and conclusions that address many of the uncertainties that were faced at the start of the work, which we aimed to address using the methodologies described here. This work should provide a solid foundation for the design and planning of further research in the field.
We have employed robust epidemiological, analytical and qualitative methodologies, as required. By conducting an inception cohort study (as opposed to a prevalence cohort study), we were able to obtain accurate estimates of time to event data on outcomes for this SWHSI patient group; we also obtained a full picture of treatments received over the natural history of the wound. Although a prevalence cohort would have allowed higher recruitment figures, it would have provided far less clarity on crucial outcomes, such as time to healing. We were unable to rely on retrospective extraction of routinely collected data on wound treatments and outcomes, as systematically collected data, especially in the community, are limited. Thus, obtaining robust data, such as we have, requires data collection to be embedded in an appropriately designed primary study from the inception of the wound.
Limitations of the research
Our cohort study was conducted in only two centres in the UK and we therefore acknowledge that practice may vary across different centres and that the prevalence rates observed may not reflect those observed across the UK.
We have noted previously that the generalisability of findings of the cohort study may have been reduced by the need for consent, which removes some of the patient group from the study group. Our cross-sectional survey findings reassured us, however, as there were no major differences in key demographic features, although more wounds in cohort participants were planned to be left open before surgery. Those patients treated with NPWT in the cohort study were found to differ from those who were not, therefore indicating potential treatment selection. Interviews with clinicians helped to explore how NPWT may be allocated; however, this did not provide a definitive explanation for treatment decisions.
The nature of the cross-sectional survey may have resulted in some cases of SWHSIs not being reported, resulting in an underestimation of treated SWHSI prevalence. The number of responses received was as we expected, and so we are confident that the number of missed wounds was low and not systematic, probably due to publicity of the survey and support provided for nurses to complete study activity.
Within the cohort study, 73 wounds (18.6%) did not heal by the end of the study follow-up period. Time to healing was used as the primary outcome, given that this is an important and recommended outcome in wounds research. However, when using this outcome, it is acknowledged that there will inevitably be some participants who do not heal by the end of the follow-up time point. We acknowledge that the lack of healing data for these participants may have had an impact on the robustness of the findings; however, we expect this to be minimal, given the small number of participants this affected.
To meet the objective of our research programme, we opted to take a broad approach and collected data on a wide range of SWHSIs. This means that we have captured data on the heterogeneity of the wound type, but this also results in more limited data with which to investigate subgroups of SWHSIs in more detail, and this may be the focus of future work.
We present findings on the relative treatment effects of NPWT using observational data, given that there was a lack of RCT data available on relative effects of NPWT on SWHSIs. It is widely acknowledged that experimental data with groups formed by randomisation provide the least biased estimates of treatment effects. We implemented use of advanced adjustment approaches to try and reduce bias as much as possible, for both known (e.g. factors identified as prognostic for wound healing such as wound size, previous wound history or level of tissue involvement) and unknown confounders. However, even with adjustment, there is still the potential for unresolved confounding to affect the results and reduces confidence in them, leading to calls for definitive evidence from a RCT.
The relative treatment effects of NPWT were evaluated only on the healing of SWHSIs and so could not account for other direct effects on HRQoL (e.g. wound management). Again, these limitations may benefit from further assessment in a RCT.
The qualitative interviews conducted as part of this research programme identified a diverse range of views from both patients and clinicians. It is, however, noted that only a limited number of interviews were conducted with both clinicians (12 clinicians comprising seven nurses and five surgeons) and patients, which may limit the conclusions drawn. Interviews with both patients and clinicians continued until data saturation was achieved, thus ensuring comprehensive data. Further qualitative research may, however, be warranted to build on these findings.
Implications for practice
Findings from this research indicate that SWHSIs are relatively common, at least within the Yorkshire and the Humber region, UK (4.1 per 10,000 population), and such wounds frequently arise following colorectal, plastic or vascular surgery.
Data suggest that wounds will demonstrate different healing trajectories. Wounds with a large surface area, contaminated wounds (at the point of surgery) and infection at any time were found to predict slow healing. Furthermore, wound location and patient comorbidity may have an impact on wound healing trajectory.
This research has identified the vast range of treatments in use for SWHSIs and has identified increasing use of NPWT as a treatment during the wound healing pathway (rising from 6% to 29% over the course of approximately 3 years). Despite increasing use of this treatment within the NHS, RCT evidence supporting the use of this treatment for SWHSIs is sparse and of low quality. Our results indicate that NPWT has a negative impact on healing and is not cost-effective. However, as our results are based solely on observational data and, even after thorough adjustment, there may be unresolved confounding, the potential for which may limit confidence in the research findings.
Implications for research
This research signals the importance of, scope of, feasibility of and opportunity for further research on SWHSIs. As is the case across the wound care field, there are few good data on the prognostic factors for healing and on which factors may be amenable to treatment to promote better outcomes. NPWT, as with other medical devices in wound care, has been adopted widely, with relatively little evidence to support its use. If decision-makers and funders call for definitive evidence derived from a RCT, the design of such a RCT would need to be carefully considered, based on data presented here. The overall population of those with SWHSIs is too heterogeneous for inclusion into a single trial and we would suggest that future research should focus on specific groups. Although people with open surgical wounds after colorectal surgery formed our largest subpopulation, careful engagement with colorectal surgeons and patients would be required for a full trial in this patient population given the difficulties and barriers outlined in the pilot trial.
Careful consideration also needs to be given to the outcomes measured in a full trial. We focused in this research on healing, and maintain, based on the findings of patient interviews, that this is a key outcome for future research. However, it is important to note that if NPWT is not cost-effective for healing, it may still offer worthwhile advantages for patients and health services, which need to be balanced against disbenefits and costs. Impacts of NPWT on outcomes such as infection and reoperation should also be considered, as should patients’ view of the treatment. The type of patient group recruited and the outcomes of interest will all influence the duration of follow-up of any planned study.
In terms of how NPWT is evaluated, investigators must carefully consider the level of pragmatism that will be employed in any future trial.88 There is a range of NPWT machines available for use, including those that use gauze dressings, foam dressings, single-use machines and repeat-use devices. There is no good evidence of any differences in the clinical effectiveness or cost-effectiveness of different machines, thus the trial design choice of whether to mandate use of a specific machine or allow sites to use their current NPWT provision will be down to the investigator. Although use of a single device will reduce the ‘noise’ in an evaluation, it is likely to limit the appeal of involvement to potential sites, as well as the potential generalisability of data. Thus, a pragmatic approach will arguably reflect effectiveness findings as they might be seen in the NHS and also make a large study more feasible. The same is true in terms of NPWT usage protocols (e.g. levels of pressure and cycle length), as well as how NPWT is used as part of the treatment pathway. The more prescriptive a future study is, the less pragmatic and feasible it may become but there would be less ‘noise’ in the findings.
Similar intervention issues also need to be considered in terms of the comparator in any future study. We know that SWHSI patients receive multiple dressings and there is no high-quality evidence that one dressing type is better than another for SWHSIs or any other complex wound type.89 Use of a specific dressing type, such as hydrofibre, could be explored in a trial, but large deviations from use of a single dressing over the duration of a patient’s SWHSI would be expected.
Blinded outcome assessment is crucial for studies with subjective outcomes, such as healing and infection.90 We have demonstrated here that blinded outcome assessment in a NPWT trial in people with SWHSIs is possible, but protocols and equipment that ensure high-quality images are required. We also note the importance of taking photographs beyond the point at which the unblinded investigator considers a wound healed, so that a later healing date can be recorded by a blinded assessor should they disagree with the unblinded assessment.
Careful consideration should be given to use of data collection via text message in future trials investigating treatments for SWHSIs. Although this method has been identified as a simple, inexpensive, valid and acceptable method of data collection in previous studies,91–93 our pilot, feasibility RCT suggested that data collection via text messaging was more successful in younger patients (respondents were, on average, almost 10 years younger than non-respondents). Other studies, which have successfully used text messaging for data collection have also included younger populations ranging from 22.0 (SD 1.47) years93 to 44.0 (SD 13.4) years of age.91 Future trials of SWHSI treatments should therefore consider the demographics of the target patient population prior to inclusion of such methods, to ensure that if utilised, data collection is likely to be successful and so provide sufficient data for analysis.
Questions for future research
We consider key research questions raised by this programme of research to be:
- Which treatments are clinically effective and cost-effective for SWHSIs for all patients or for particular patient subgroups?
- Can particular prognostic factors predict time to healing of SWHSIs?
- Do psychosocial interventions have the potential to improve quality of life in people with hard-to-heal SWHSIs?
- What are the current care pathways for people with SWHSIs, how do they vary nationally and can they be optimised to improve efficiencies and outcomes?
- Is the development of evidence-based practice guidelines for the management of SWHSI patients possible, in order to benefit and inform patients, clinicians and NHS commissioning?