Fewer Early Surgical Site Infections After Hip Fracture and Lessons Learned From the Safe Hands Project: A Five-Year Longitudinal Study of 3,553 Patients

Background: Surgical site infection after acute hip fracture surgery is a devastating complication associated with increased suffering and mortality. The aim of the study was to investigate early SSI, sepsis, pneumonia and urinary tract infections over ve years, before and after the implementation of the Safe Hands project. Methods: This was a single-centre observational study with a ve-year longitudinal design, investigating the effects of an infection-prevention intervention targeting the clinical care pathway of individuals with acute hip fracture. Statistical analyses were based on routinely collected patient outcome data comprising 3,553 patients. The study conforms to the criteria of the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE). Results: The incidence of early SSIs decreased from 2.5% in years 1-2 to 1.1% in years 4-5. Similar results were observed for sepsis (2.7% to 1.3%) and urinary tract infections (14.2% to 4.2%). The multivariable regression shows that, for every observed year, the odds of early SSIs decreased. Male gender, procedure time, sepsis and preoperative skin damage increased the odds signicantly. Conclusions Our preventive bundle, based on partnership between researchers, managers and clinicians and a strong commitment to change from the involved professions, appear to be effective in reducing the frequency of potentially devastating SSIs and other HAIs after hip fractures surgery. The use of external and internal facilitators was crucial to enable individual and organisational learning and overcoming barriers to improvements.

eligible for inclusion, (Year 1: baseline, year 2-3: intervention phase and year: 4-5 follow up). After Mars 2020, Covid-19 hit the hospital with full force, consequently not eligible for inclusion in the present study. The inclusion criteria were: i) age > 65 years, ii) surgically treated hip fracture and iii) perioperative care at surgery compared with primary hip or knee arthroplasty (12). A general lack of hand hygiene and aseptic techniques in the OR, especially during anaesthetic care, was identi ed (12)(13)(14), together with the fact that organisational structures, con icting goals and power issues often worked as barriers to change. In cases of success, supportive relationships between the managers from different professions and organisational levels were essential, along with a strong sense of ownership and control over the implementation process (15). Given the prevalence of hip fractures, the patients' vulnerability and their need for surgery to regain functional independence, there was a strong commitment among hospital management and researchers to improve the quality of care. To address the identi ed contextual problems, the Safe Hands project was initiated (ClinicalTrials.gov ID: NCT02983136) with the aim of testing and evaluating an implementation programme to increase awareness of the problem of hospitalacquired infections (HAI) and increase the use of preventive measures including hand hygiene, with the emphasis on the vulnerability of patients with hip fractures. The programme had an iterative, exible approach, allowing for the co-creation of solutions and adaptations to the speci c contextual factors.
The implementation process for components in the programme and the links to theory have previously been published (16,17). The initial project was expanded to involve the entire hip fracture care pathway, targeting catheter-related urinary tract infections, the risk of bladder distention and de cits in hand hygiene and aseptic techniques. The aim of the present study was to investigate early SSI, sepsis, pneumonia and urinary tract infections (UTI) over ve years, before and after the implementation of the Safe Hands project. an orthogeriatric ward in our hospital. Exclusion criteria: i) HLOS < 2 days, ii) pathological fracture, iii) excision arthroplasty and iv) re-fracture or a contralateral hip fracture., Variables and de nitions are presented in Box 1. The hospital's quality register contains prospectively collected in-hospital data related to orthogeriatric hip fractures.

The bundle intervention and changes in standard practices
Ongoing interventions in complex hospitals settings can yield both positive synergies and competing interests and con icting goals within the organisation. This paper considers this complexity by reporting all the relevant method and organisational changes occurring during the ve-year study period. The components of the Safe Hands project and other changes in hospital standard practices are presented in Table 1.

Data collection
Prospectively collected register data over ve years were used to analyse early SSIs and other infectious outcomes. In addition, other organisational changes unrelated to the Safe Hands project were recorded. The data les were scrutinised by a research nurse and missing data were corrected in the register whenever possible. Outliers were veri ed against source data (i.e. patient records). The data les were cleaned according to the inclusion and exclusion criteria, outlined in Table 2 for included and excluded patients.

Statistical methods
Primary outcome: early SSI, secondary outcomes: sepsis, pneumonia and UTI. For categorical variables, n (%) is reported and, for continuous variables, the mean (SD)/median (min;max). For comparisons between ordered groups, the Mantel-Haenszel chi square test was used for dichotomous and ordered categorical variables, while the Jonckheere-Terpstra test was used for continuous variables. Univariable and multivariable logistic regression were used for predictors of SSIs, unadjusted and adjusted for age and gender. A multivariable logistic regression analysis was performed. The variables were included together and selected, based on the results of the univariable analysis, and, for clinical relevance, years 1-5, gender, procedure time, sepsis and skin damage. "UTC more than once" was not included, because the association with SSI is multifactorial. It is probable that an IUC is the result of an SSI rather than a cause. P-values, odds ratios (OR) with a 95% con dence interval and area under the ROC curve are based on original values and not on strati ed groups. All signi cance tests were two-sided and conducted at the 5% signi cance level. Data were analysed with the SPSS statistical package version 25 (IBM Corp. Armonk, NY, USA) and SAS Version 9.4, SAS Institute, Cary, NC, USA.

Results
The results are based on an analysis of data comprising 3,553 patients. Patient and clinical characteristics are presented in Table 3. There were few changes in the case mix over the years, although there were more patients with cognitive impairments in years 4-5 compared with years 1-2. HLOS decreased by three days from years 1-2 to years 4-5.

Surgical site infections
The frequency of early SSIs decreased from 2.5% in years 1-2 to 1.1% in years 4-5 ( Table 4). The overall frequency SSI in hemiarthroplasty was 2.7%, while it was 1.5% for internal xation.
In the univariable analyses (adjusted for age and gender), an earlier year in the study period, gender, sepsis, severe pressure ulcer (Norton stage 4), skin damage, sepsis and IUC more than once signi cantly increased the odds of developing an SSI (
Of 77 patients with sepsis in this cohort, 20% were diagnosed on admission and 80% were hospital associated. Fifty-nine % were of unknown origin, 14% and 5% secondary to a urinary tract infection and pneumonia respectively. The mean HLOS for patients without sepsis was 12 days (95% CI 11.6-12.1), while it was 23 days for those with hospital-associated sepsis (95% CI 15.6-31.1).

Discussion
During the study period, we observed that in-hospital SSIs and other nosocomial infections following treatment and care for hip fractures can be signi cantly reduced by using the bundle approach based on the Safe Hands project. The initial early SSI rate in our cohort lies in the mid-range of previously reported rates (19) and the rate after the bundle interventions was in the lower range (20). Rates of pneumonia and unspeci ed infection increased slightly in the fth study year and this is probably attributable to Covid-19. For every year, the odds of an early SSI decreased, despite that there were signi cantly more patients with cognitive impairment and nursing home residents in the last two study years, indicating greater frailty in the cohort (21). In line with the literature, male gender, prolonged procedure time and more than one urinary catheterisation increased the odds of early SSIs (2,17,19). In contrast, age, diabetes and an ASA score did not predict SSI in our cohort. The identi cation of discrete modi able risk factors is clinically desirable to ensure optimal intervention. The strong association between sepsis and SSI was not surprising (22); it stresses the importance of handling all medical devices, such as venous and urinary catheters, with strict adherence to hand hygiene guidelines and aseptic techniques (23). S. aureus bacteremia, albeit to a lesser extent if hospital acquired, increases the risk of bacterial seeding to a previously inserted orthopaedic implant or another biomedical device, thereby compounding morbidity (24)(25)(26). There are studies from several centres including our hospital that report decreased mortality and hospital re-admission in key infections, including S. aureus bacteremia, following early Infectious disease consultations (27,28). To this end, S. aureus alarms and the increased availability of bedside ID consultations were introduced in the second quarter of the rst study year and in the rst quarter of the third study year, respectively. Skin lesions on admission and severe pressure ulcers constituted a very high risk of SSI. The latter is a modi able risk factor that shows the importance of a team effort preventing complications in surgery; by including RNs and emphasising the importance of optimal nursing care, pressure ulcers can be avoided. However there is a need for high-quality trials, establishing the optimal repositioning frequency in this patient group (29).

Strengths and limitations
As this is a single-centre observational study, there are caveats when it comes to the interpretation of these results. In addition, other changes in hospital standard practices, many of which were developed in relation to the growing organisational focus on this patient group, have probably impacted the results of the Safe Hands interventions. What we can see is a probable reciprocal effect where the different changes reinforce the results in terms of patient outcome. However, we have tried transparently to report all the changes that have occurred during the ve years included here to minimise the risk of overstating the in uence the Safe Hands project has had on clinical practice. The study also has its strengths, such as the large study cohort of 3,553 patients. To avoid imputation errors, the registered data have been validated against patient records. The local quality register started in 2015 and the number of patients included in it has uctuated over the years. Fewer patients were included in the rst years of the register. The estimated completeness in the rst year was approximately 60%, based on a median value of included patients in years 2 to 5. No systematic errors that can explain the lack of imputation in the rst year of the register have been found.
Using only routinely collected data to analyse outcome has its limitations. As a result, other important prognostic and confounding factors, such as blood transfusion, body weight and smoking, have not been controlled for in the statistical analysis.
Bundle approaches have inherent strengths and limitations. Previous bundle interventions have proven useful in improving the quality of care and reducing SSIs in HF patients (6, 20) and other serious HAIs, such as blood-stream infections (30) and ventilator-associated pneumonia (31). Others have criticised bundle approaches and challenged their usefulness, as it is di cult or even impossible to tease out the parts of the bundle that have contributed to the desired change and the extent. We argue that this criticism is less important than the potential bene ts of bundles. Moreover, it might be useful to move away from linear thinking where every single part can be measured and understood, to acknowledge the complexity of change and view the transformation process from a holistic perspective where the whole is greater than the simple sum of parts.

Lessons learned
At the start of the Safe Hands project, we aimed to create sustained improvements in the treatment and care of older individuals with hip fracture, with special emphasis on infection prevention. The results of the present study indicate sustained improvements and, moreover, the incidence of early SSIs, UTIs and bloodstream infections continued to decrease even after the interventions were implemented in year three and the research team left the site. It is common for most interventions to show an effect in the short run, but the challenge has been to create sustained improvements after the intervention(32), a challenge we were aware of when deciding on the implementation strategy. We see some explanations of our promising results and sustained effect. Implementation theories and frameworks has highlighted how contextual factors can both promote and hinder the uptake of evidence-based care (33)(34)(35)(36). For this reason, the results of our study cannot be understood without acknowledging contextual mechanisms such as leadership engagement, resources, an organisational safety culture and commitment to change.
To add another layer of complexity, the Safe Hands implementation programme was aimed at surgeons, RNs, specialised RNs and nurse assistants, leaders (formal and informal) and managers. To handle this complexity, the programme was based on facilitating mechanisms for contextual negotiation and collective action; 1) Building a strong partnership between researchers, management and clinicians based on mutual respect, 2) External and internal facilitation as a role and a process that focused on enabling and supporting individual and organisational learning (37,38). We found that the choice of facilitators was critical and needed to be adapted to match the context. To be perceived as trustworthy, these facilitators needed to have an in-depth understanding of the medical context and infection prevention. The internal facilitators were introduced step by step and represented all the professional categories. When the external facilitator left the site, the internal facilitators remained and were able to function as local championns (16). In this way, the improvements and learning in clinical practice could continue and may be one contributory factor in terms of the sustained and reduced infection rates. 3) Dialogue and cocreation, to facilitate organisational learning. Isaacs' (39) and Schein's (40-42) work has demonstrated the signi cance of creating space for dialogue. From their work, we used interprofessional dialogue to learn more about one's own and co-workers' ways of thinking about infection prevention and to inquire collectively about how available knowledge could best be transformed into co-creating and testing new ways of working together to reduce the risks of infection after surgery. As a result, the work aimed to create a cultural change instead of modifying behaviours. For this to occur, we found, in line with previous studies (41, 43) (p. 305), that the creation of psychological safety, mediated by respectful dialogue, was imperative to facilitate transformation.
Initially, very few people in the organisation appeared to acknowledge the magnitude of the problem with HAIs. Competing interests and other daily problems to resolve may have shadowed the infection issue. By using local quality data as a basis for dialogue with the management and clinicians lead to increasing awareness and a shared sense of urgency in relation to the problem. Most managers and clinicians developed the motivation to engage in the transformative work, even if not everyone was motivated to make changes. To sum up, the Safe Hands project changed the way risks, safety and infection prevention were perceived in relation to hip fracture patients (16) and signi cantly improved patient outcomes.

Conclusions
Our preventive bundle, based on partnership between researchers, managers and clinicians and a strong commitment to change from the involved professions, appear to be effective in reducing the frequency of potentially devastating SSIs and other HAIs after hip fractures. The use of external and internal facilitators was crucial to enable individual and organisational learning and overcoming barriers to improvements Declarations Ethical approval and consent to participate Ethical approval was obtained from the Regional Ethics Review Board in Gothenburg, Sweden, (reference number 166-15 and amendment 327-17). Patients received written information about the quality register, with information on who to contact about their register data if they did not want their data to be used for research purposes.

Consent to publication
Not applicable Availability of data and materials Data are available in response to reasonable requests.  Q3-4 -A new routine was implemented that formalised the practice that junior physicians in training should receive support from a senior surgeon to avoid prolonged surgical time for hip fracture surgery. The aim was to create a culture where it would be easy and appropriate to ask for help from a senior.

No competing interests declared
3 Intervention Q1-2 -Antibiotic rounds twice weekly led by a consulting infectious diseases specialist were introduced on the geriatric wards with the aim of promoting sound antibiotic use, e.g. reducing the number of prophylaxis-resistant bacterial strains on the wards.
-Accessibility to the consulting infectious diseases specialist was increased from two to four days a week for bedside assessments.
-The preoperative shower routine consisting of a double shower with 4% chlorhexidine gluconate (CHX) was changed from two showers before surgery to one shower before surgery.
If the patient had to wait for surgery for more than 48 hours after the rst shower, an additional CHX treatment was carried out.
-Expanding the Safe Hands project; a catheter-related urinary tract infection prevention strategy (Safe Bladder) was developed.
Q3-4 -Safe Bladder was implemented in the full care pathway ER, OR, PACU and the geriatric wards.