Worldwide, it is estimated that for every 16 surgical patients one will experience a surgical site infection (SSI), a healthcare-associated infection that carries an increased risk of morbidity, mortality and contributes to antimicrobial resistance (World Health Organization, 2016). In the UK, the likely cost of managing SSI exceeds £1 billion per annum (updated to 2020 costs) (Frampton, 2010) and institutions are not paid for readmissions of SSI above an agreed threshold of 30 days (NHS Improvement, 2019). Patients with SSI are six times more likely to be readmitted to hospital impacting on all aspects of safety, quality, cost and productivity in health care (Sanger et al, 2014; Shah et al, 2017; Getting It Right First Time, 2019). That the majority of SSI present after discharge (Woelber et al, 2016a; Public Health England, 2019), but patients and carers perceive information for surgical wound care and SSI as of low value (Sanger et al, 2014; Zellmer et al, 2015; Tartari et al, 2017) warrants particular attention. Indeed, across surgical specialities including cardiothoracic surgery there is a clear programme directed by the National Wound Care Strategy Programme's surgical stream to improve the information provided at discharge to patients and carers, as well as for referral pathways if wound concerns arise.
To address these challenges, the Photo at Discharge (PaD) strategy was developed within the healthcare setting (Rochon et al, 2016). On the day the patient leaves hospital, the nurse takes a colour picture of the patient's surgical wound, completes an electronic assessment, and provides tailored advice on SSI prevention and wound protection. The patient takes away the PaD form with all this information as a bespoke discharge resource. As a patient-centred strategy, PaD appears to meet patient needs and preferences for information regarding surgical wound care (Rochon and Morais, 2019) and may reduce the risk of readmission for SSI (Rochon et al, 2018). Increasingly, cardiac centres are adopting PaD as part of their standard discharge process (Cardiothoracic Interdisciplinary Research Network (CIRN) et al, in press).
PaD is not a simple intervention and similar to any intervention aiming to reduce 30-day readmissions, there are many challenges to successful implementation (Hansen et al. 2011). PaD involves changes to the processes and practices within the healthcare system (Thompson and Dobbs, 2018), as well as patient behaviour because of its explicit purpose of engaging patient responsibility in the transition of care (Coleman et al, 2004). A similar quality improvement (QI) approach of using nursing staff to take a baseline image was used in the USA to assist with SSI surveillance (Cannon et al, 2016). However, the remainder of the interventions seem to focus on the use of (primarily patient-generated) photos after the patient leaves hospital (Evans et al, 2019).
PaD has been identified as a positive area of practice during the Royal College of Nursing (RCN) consultation on the digital future of nursing (RCN, 2018). However, without a top-down systematic planned approach, the spread of PaD has been an unpredictable process, with the innovation spreading in a non-centralised and informal framework (Greenhalgh et al, 2004; Greszczuk et al, 2018). Thus little is understood about the implementation at different sites. Here we refer to fidelity as how closely the programme is implemented as intended by the developer, based on Carroll et al's (2007) conceptual framework. Systematic evaluation of fidelity is the best way to approximate implementation quality (Breitenstein et al, 2010). Conversely, low fidelity during the spread of interventions can reduce the intended benefits to care quality (Breitenstein et al, 2010). Fidelity may be determined by adherence (ie whether the protocol has been fully implemented) and can be evaluated by measuring content, coverage, frequency and duration (Carroll et al, 2007; Breitenstein et al, 2010). This work will use these elements, but departs from Carroll et al's proposed moderators of fidelity and instead refers to domains of sustainability within the Long-Term Success Tool (LTST). LTST was developed by Lennox et al (2017) to examine 13 factors or domains in order to identify potential barriers or opportunities to sustain improvement. The inclusion of LTST was deemed to be important as the majority of quality improvement projects fail to be sustained (Szabla, 2007). Thus, in addition to fidelity, we sought to study the methods that centres used to increase uptake into routine practice (Curran et al, 2012).
In this study, the researchers sought to consult with other centres using PaD in order to conduct an evaluation of outcomes (fidelity and sustainability) to determine implementation of PaD within different settings. This service evaluation was registered (CIRIS ID0033795).
Methods
Study setting and sample
Four cardiothoracic centres in the UK that provide PaD were contacted via email. Three centres (one teaching hospital and two specialist tertiary referral hospitals) agreed to take part. The fourth centre declined on the basis of workload/pressures.
Survey design
The service evaluation comprised two structured surveys for self-completion designed by the team who created PaD (ie ‘the developer’). Survey 1 was an Excel-based survey on site-specific information, based on a logic model (Table 1) and evaluation plan (Table 2). Survey 2 consisted of the LTST. The LTST was added to a dedicated secure online survey tool to collect anonymous stakeholder feedback. Both surveys were reviewed by team members before the surveys were piloted on the first site. Following minor grammatical changes to the pilot study, the surveys were finalised.
| Core inputs | Immediate impacts | Short-term impacts | Impacts | Health outcomes |
|---|---|---|---|---|
| Nurse takes picture and completes e-assessment of surgical wound | Enhanced discharge information | Healthcare providers have increased information regarding surgical wound and care | Improved discharge possibilities for better wound monitoring, earlier detection of wound concerns | Increased self-care of surgical wounds and monitoring, increased satisfaction |
| Patient/carer receives PaD | Increased confidence for self-care | Surgical patients will have more knowledge of who to contact with concerns | ||
| PaD accessible by multidisciplinary team | Improved continuity of care | Increased engagement in SSI prevention | Reduced number of readmissions for SSI, better communication between health settings, reduced risk of overtreatment |
| Area to measure | Standard to measure against | |
|---|---|---|
| Content | Was each of the PaD components implemented as planned? | Colour image + e-assessment, provided to patient and available to multidisciplinary team |
| Duration/dose | Was each of PaD components implemented as often and as long as planned? | Day of discharge |
| Coverage | What % of eligible patients received PaD? | Discharged or transferred to other care facility after cardiac surgery |
Survey distribution and data collection
Data were collected in the period 1-29 February 2020. Surveys were distributed electronically (via email) to named lead recipients at the hospitals identified through collaboration with Cardiac SSI Network members. Survey 1 was completed once for each centre. Leads were asked to share Survey 2 (link to electronic online survey: https://123formbuilder.com/form-5289305/PaD-2020) with key stakeholders for anonymous completion.
Data storage and governance
Data were collected and stored on a secure cloud-based server during the survey period. No patient identifiable information was collected. This study is not considered research as defined by the UK Policy Framework for Health and Social Care Research as outlined by the NHS Research Authority and so ethical approval was not required.
Analysis of data
Findings from the surveys are displayed using descriptive analysis. Nonparametric data are presented with median and range. Frequency distribution of aggregate scores are provided for ordinal data subgroups (Likert scale) for Survey 2 in graphical form (Sullivan and Artino, 2013). A gradient six-point scale was used to reflect the ordered opinions (very good, good, fair, poor, very poor) and middle values (‘no opinion’ and ‘don't know’) were included (Bowling, 2014). Cronbach's alpha was used to test the intercorrelation of grouped items and measure the underlying variable, sustainability (Sullivan and Artino, 2013). Intraclass Correlation Coefficient (ICC) was used to measure an average reliability, modelled with the same accessors (respondents) rating the same domains, wherein the systematic differences between the raters were not relevant (ie validity rather than absolute agreement model) and no opinion/don't know were excluded from this agreement/responsiveness testing. For consistency and reliability testing, a value of 0.7 or above was set for significance. Data were analysed using MedCalc Statistical Software version 19.2 (MedCalc Software Ltd, Ostend, Belgium).
Outcomes
Three centres agreed to take part (75% response rate) on this study. Participants completed Survey 1 in full, with no missing data (Table 3). Overall, the centres reported a combined total of 9007 patients received PaD since implementation began. The number of wards involved with PaD varied across the sites (ranging from one to six wards), with an overall bed-to-staff ratio of 100:237. Centres reported high concordance with fidelity measures. This included content (100% reported combining the colour image with bespoke wound assessment and advice), coverage (all centres reported high compliance rates, ranging between 92% and 100%), duration and dose (once, ie baseline picture, not a series of pictures, provided at discharge/transfer by the discharging healthcare provider). The centres were all currently undertaking PaD (none had discontinued the scheme) and sustained over the long term (>4 years), mid term (>2 years) and shorter term (<6 months). From time of proposal, the IT system took an average of 16 months to implement (range 11-21 months) with centres modifying existing in-house databases and colour printers. All three indicated a camera or tablet was purchased to assist with image capture.
| Centre 1 | Centre 2 | Centre 3 | |
|---|---|---|---|
| IT system | |||
| IT system used | PATS/Dendrite | Lorenzo | PATS/Dendrite (Iweb) |
| Date proposal/request for PaD went to IT | August 2014 | February 2018 | August 2016 |
| Date database available | July 2015 | November 2019 | January 2018 |
| Approximate time from IT proposal to implementation (months) | 11 | 21 | 17 |
| Implementation | Totals (n) | ||
| Number of ward(s) providing PaD | 4 | 1 | 6 |
| Number of beds | 76 | 28 | 96 |
| Number of nursing colleagues who can deliver PaD (system users) | 265 | 15 | 194 |
| Since implementing. how many PaD records in total (n) are there? | 5121 | 98 | 3788 |
| In January 2020, how many patients were eligible for PaD (ie discharged in January alive)? | 77 | 65 | 144 |
| In January 2020, how many patients received PaD? | 77 | 65 | 133 |
| Monthly compliance (%) snapshot, January 2020 | 100% | 100% | 92.4% |
| Resources | Totals (n) | ||
| Please complete this data for ONE ward only for equipment used for PaD | |||
| Camera | 1x Sony 20.1 megapixels (new) | 1x Sony Cybershot (new) | 0 |
| Tablet | 1 | 1 | 1 |
| Other (eg dedicated ward mobile phone) | 0 | 1 | 0 |
| Colour printer* | 1 | 1 | 1 |
| Practice | |||
| Is PaD taken routinely on the day of discharge (not other day eg D5)? | Y | Y | Y |
| Do you provide a colour copy to the patient/carer? | Y | Y | Y |
| Are you able to email the copy to the patient? | planned | possible but not routine | N |
| Is PaD on a patient portal to allow the patient to view the form remotely? | planned | N | N |
| Do you routinely provide a copy of PaD to the GP/other healthcare provider? | Y | GP has access to system | N |
| Is the PaD form saved on EPR (electronic patient record, so that all staff can view)? | Y | Y | Y |
| Does PaD include assessment/advice? | Y | Y | Y |
| Is the PaD just a standalone image (no printed assessment or advice on the same form as image)? | N | N | N |
Survey 2 was completed by 19 anonymous respondents from the three centres (ten from Centre 1, four from Centre 2 and five from Centre 3). Overall, responses were consistent in LTST (Cronbach's alpha with standardised variables was 0.92 (95% lower confidence limit 0.8511)); although reliability as determined by average measures in ICC only just reached significance (0.6971; 95% confidence interval 0.3926 to 0.8902). Favourable responses in the domains, such as ‘very good’ or ‘good’, were observed most frequently (n=114 (46.0%) and n=88 (35.5%), respectively). ‘Fair’ (n=4 (11.7%)), ‘poor’ (n=2 (2.0%)) or ‘very poor’ (n=1 (0.4%)) less commonly, with 4.4% of responses as ‘no opinion’/’don't know’. Figure 1 provides an overview of the 13 domains believed to influence the sustainability of quality improvement projects. The more outward the solid colour line, the higher the proportion (%) of respondents. Conversely, lines closer to the centre of the radar graph received a lower proportion (%) of responses. Using this model, respondents selected very good most frequently for ‘commitment to the improvement’ and ‘evidence of benefits’ (63% and 63%, respectively). Only one respondent selected very poor for any of the domains, in this case ‘alignment with external political and financial environment’.
Discussion
Approximately 1 in 75 coronary artery bypass graft (CABG) patients are readmitted with SSI (Lamagni et al, 2020). In many cases, early detection and sufficient therapy will reduce the burden of direct and indirect costs associated with SSI (World Health Organization, 2016) and it is notable that acute care and GPs find PaD helpful in wound review and management (Rochon et al, 2018). Furthermore, as patients are becoming more involved and better informed in their health and care (Tartari et al, 2017), the value of interventions to reduce readmissions within 30 days and patients' capacity for self-care in their transition from hospital to home should not be underestimated (Leppin et al, 2014). However, as Balasubramanian et al (2015) suggested, ‘demonstration’ projects such as PaD may have their impact evaluated (listed in Table 1 under health outcomes), but the iterative process of system change, including an evaluation of core outputs, is often not measured. As a result, when quality improvements fail to achieve their projected outcomes, the mechanism(s) for the failure may not be well understood (Balasubramanian et al, 2015).
In this study, the authors consulted with three centres in different organisations to evaluate implementation fidelity and possible factors influencing sustainability for PaD. This has been guided by the work of Hasson (2010) and the fidelity adherence theoretical framework proposed by Carroll et al (2007) and included factors for sustainability proposed by Lennox et al (2017). The purpose is to contribute to a greater understanding of fidelity and its processes, which may inform a broader knowledge base to create sustainable interventions (Breitenstein et al, 2010). The results of this study may be of particular interest to centres planning to implement PaD (McCabe et al, 2018; Tyrer, 2019).
Fidelity is understood to moderate the relationship between the intervention and its outcome (Carroll et al, 2007). For PaD, the ‘essential elements’ or content (Carroll et al, 2007) refer to enhanced surgical wound advice; frequency, duration and dose provided objective measures for effectiveness, with high levels of concordance (integrity) reported with the intended intervention. The majority of surgical wounds heal well, and the intention is to balance the practical application of the digital image without increasing patient workload unnecessarily (Zellmer et al, 2015) or that of the healthcare service (Sanger et al, 2016). PaD does not prevent infection (nor does any digital image), but can improve patient confidence in caring for the wound, provides important reassurance and acts as a tool to monitor the wound for changes (Rochon and Morais 2019). Poor or incomplete implementation can impact on an intervention's efficacy (‘type III error’ where a potentially useful intervention appears ineffective due to failed or diminished fidelity) (Dobson and Cook, 1980) thus findings from these surveys are reassuring in terms of successful spread of PaD.
Interestingly, there is a very high compliance within the centres involved despite slow progress of technological advancement within the NHS (Royal College of Nursing, 2018). Information technology in healthcare may be limited by internal barriers that may prevent uptake, including unsupportive organisation structure, irreconcilable differences of stakeholders and poor understanding of the process (Walley and Davies, 2002). In contrast, the experience of centres implementing PaD demonstrated strong nurse leadership, commitment to the improvement, involvement of stakeholders and evidence of benefits. Further, common adherence barriers to guidelines implementation such as education and training (Fischer et al, 2016) have been addressed by the majority of stakeholders (84%) felt skills and capabilities were ‘very good’ or ‘good’ for the use of images and digital assessment in wound care.
To the authors' knowledge, this is the largest collaborative implementation evaluation of PaD, and undertaken in three separate, ‘real world’ settings in order to better understand the care and quality benefits of the scheme. A strength of this study is that in conjunction with implementation fidelity evaluation, the authors have opted to use the LTST to examine 13 factors. The LTST, which can be used in combination with other tools and theories (Lennox et al, 2017), was developed for ‘real world settings’. The findings of LTST may prompt local action planning and may be combined with advice for developing policies and standard operating procedures for the use of digital images in wound care (Rochon et al, 2017). From a practical perspective, centres looking to implement PaD may wish to review findings particularly in relation to the resources in place (involving IT early in the implementation process and increasing cameras or consoles to more than one in each ward), skills and capabilities of those involved (new online learning course, ‘The use of digital images in wound care’ endorsed by the RCN is now available), and alignment with external and financial environment (Rochon and Morais, 2019).
This study has a number of limitations. First, the cross-sectional design means that results are reflective of a certain point in time. The authors attempted to control for this by using tools within the surveys that could be repeated/comparable (Adamson, 2005). Drawbacks of using a Likert-type scales include its ordinal nature, and that different combinations of the multiple items can result in the same aggregated scale or score (Adamson, 2005). Therefore, although factor analysis was to check internal consistency, it remains the case that some information may have been lost from components of the score (Bowling, 2014). Another important limitation of this work is the small sample size and incomplete coverage of centres performing PaD, and this potential source of bias may threaten the generalisability of the data (Bowling, 2014; Moore et al, 2015). A UK 2019 variation in practice audit conducted in cardiothoracic centres found approximately one-third (n=7) of the 19 anonymised centres that participated in the survey indicated they were undertaking PaD (CIRN et al, 2020). In this study, the authors have attempted to use validated implementation tools to improve the validity and reliability of findings; however, the tools have not collected qualitative data, which could provide deeper insight into the variables of interest, nor did the study use direct observation of those delivering the intervention to evaluate fidelity to improve the validity of self-reported fidelity (Carroll et al, 2007; Bowling 2014). Finally, this study examines the experience of PaD in cardiothoracic centres, which may not be readily transferrable to other specialist or acute services.
Conclusion
Traditional models of surgical wound care and assessment are not always aligned with the needs of patients or onward care providers. Within the cardiac specialism, PaD is an emerging strategy to share a baseline visual record and enhance discharge information. In the absence of external support for change on a large scale, PaD implementation has been nurse-led and accompanied by ‘on-the-ground’ learning. For this reason, the work here has focused on implementation, using a systematic approach to determine whether there are differing levels of fidelity, compliance and/or sustainability factors between three hospital sites. Although arising from a relatively small number of centres, this design enables comparison of core inputs (identified within the PaD logic model) reliably and consistently to produce actionable information, which in turn can be used to enhance, scale and accelerate PaD in other contexts.