Sepsis Awareness at the University Hospital Level: A Survey-Based Cross-Sectional Study

Background: Sepsis is a leading cause of morbidity and mortality. Prompt recognition and management are critical to improved outcomes. In 2019, the Lausanne University Hospital (LUH) launched a quality of care project aiming to improve sepsis management. As part of this effort, we aimed to assess sepsis awareness among nurses and physicians of the LUH and among the local paramedics. Methods: We conducted a survey on nurses and physicians at our institution and local paramedics between January and October 2020 representing over 1,000 professionals distributed over all hospital departments. The survey assessed professionals’ knowledge of sepsis epidemiology, denition, recognition and initial evaluation (nurses and paramedics) or sepsis epidemiology, diagnosis, and management (physicians). Pediatrics and the neonatal unit were excluded. Results: A total of 1,116 of 1,216 contacted persons among the 4417 targeted population participated and completed the survey (participation rate 91.8%). This included 619 of 2,463 (25.1%) of hospital nurses, 348 of 1,664 (20.9 %) of physicians and 149 of 290 (51.4%) of canton paramedics. Our nurse and physician sample was slightly imbalanced for sex and age. Thirteen percent of participants (28.4% of physicians, 5.9% of nurses, 6.8% of paramedics) correctly identied the Sepsis-3 consensus denition. Similarly, 48.6% of physicians and 10.0% of nurses identied the SOFA (sequential organ failure assessment) score as a sepsis dening score for infected patients. Furthermore, 24% of participants identied the Quick Sepsis-related Organ Failure Assessment (qSOFA) score as a predictor of increased mortality; 6% identied correctly the components of the score. For a patient with suspected sepsis, 96.1%, 91.6% and 75.8% of physicians respectively identied blood cultures, broad-spectrum antibiotics and uid resuscitation as required interventions; 76.4% and 18.2% of physicians requested these initial measures within 1 and 3 hours, respectively. For physicians, recent training correlated with awareness regarding denitions, SOFA score and qSOFA score use and components:


Background
Sepsis is a syndrome de ned as a dysregulation of the host response to an infection. 1 Its incidence has increased over the past decades and accounted in 2017 for an estimated 48.9 million cases and 11 million deaths globally, more deadly than stroke and myocardial infarction combined. 2 Sepsis is also associated with signi cant long-term morbidity, including cognitive impairment, recurrent septic episodes and increased mortality amongst survivors. 3,4 In the absence of speci c targeted therapy blunting the dysregulated host response to infection, optimal sepsis management relies on rapid recognition, initiation of antimicrobial therapy and intensive supportive care. Since 2002, the Surviving Sepsis Campaign (SSC) has aimed to reduce sepsis-related mortality and morbidity by increasing sepsis awareness among professionals and providing consensus management guidelines structured into bundles. [5][6][7] Sepsis awareness and prompt recognition by healthcare professionals (HCPs) are critical components of the management of septic patients. Sepsis awareness includes basic notion of epidemiology, de nition of sepsis, and the familiarity with the implementation of bedside scoring tools. 8 In the last three decades, sepsis de nitions have been reviewed twice since the initial round of 1991 with the last iteration being the 2016 Sepsis-3 consensus de nitions. 1 These changes in de nitions have been accompanied by changes in the clinical score and diagnostic criteria. As an example, Systemic In ammatory Response Syndrome (SIRS) is now replaced by the sequential [Sepsis-related] organ failure assessment (SOFA) score. Despite being introduced more than four years ago, there is a dearth of article on the degree of actual knowledge about the actual content of the de nition among various HCPs. We identi ed only three studies on sepsis awareness amongst HCPs, limited in size and scope of HCPs which tested Sepsis-3 consensus de nitions knowledge. [9][10][11] Studies of previous sepsis de nitions have revealed gaps in sepsis recognition and management amongst medical and paramedical staff. 8,10,12−18 Most studies, however, focus on a single HCP subset, have limited participation (50-200 participants) and are restricted to a single department. Further, few studies have been conducted in wards despite nosocomial sepsis representing 20-30% of all cases. [19][20][21] Methods Study aim, design and setting In 2019, the Lausanne University Hospital (LUH) launched a quality of care program to improve sepsis management. As a part of the effort, this study aims to quantify Sepsis-3 consensus awareness amongst nurses and physicians of various clinical units at LUH and local paramedics and identify potential de cits that should be addressed in continuing education.
This cross-sectional study was conducted through an anonymous, on-line survey measuring the awareness, knowledge, and attitudes about sepsis among nurses and physicians of the LUH and local paramedics. The LUH is a 1568-bed tertiary care university hospital, serving the city of Lausanne (population circa 300,000 inhabitants) and the tertiary care reference medical centre for the Canton de Vaud (799,145 inhabitants) in Switzerland. Participants were identi ed through random visits on the work place or at continuing education seminars.

Measures
The research team designed a survey inspired from previously published surveys assessing knowledge and awareness of sepsis. 22,23 The questions were tailored to the profession (clinical scenario adapted activity sector -medicine, surgery, emergency department or gynecology). The survey was written and completed in French. Each section of the survey (paramedics', nurses' and physicians' section) was submitted to three focus groups consisting of 3 to 6 participants of each profession, commonly involved in care of patients with sepsis. These focus groups assessed the applicability and appropriateness (validity) of the survey. The questionnaire was revised using feedback from the groups. Surveys of nursing staff and paramedics were more focused on screening and initial evaluation whereas physicians were also tested on diagnosis and management. Responses options included Likert-type scales, binary (e.g. "yes/no") or multiple choices. Each question was locked upon answering, which prevented post hoc changes that could be in uenced by information provided at a later stage of the survey. The nal survey contained questions on participants' demographic characteristics (5/7/6 questions for nurses/paramedics/physicians), sepsis continuous education (3/3/3 questions), self-evaluation of sepsis knowledge and clinical management (2/2/2 questions), de nitions, scores and epidemiology (11/12/14 questions), and sepsis management (4/4/5 questions). The survey was developed in REDCap (Research Electronic Data Capture) software so as to automatically export participants' responses to a database. 24,25 Surveys are provided as supplementary material (supp. meth. survey).

Data collection and recruitement
Participants were recruited between January 20 and October 10, 2020. We aimed for a large and representative sample size of 1,000 persons distributed over all departments (Emergency department (ED), intensive care unit (ICU), Medicine, Paramedic, Psychiatry, or Surgery) and professions (paramedics, nurses and physicians) to reach 25% of LUH staff considered HCPs. Pediatrics and neonatology staff (not covered by Sepsis-3 consensus de nitions) as well as nurses and physicians not in daily contact with patients (i.e., who were working in research team or in administration) were excluded. Thus, participants were screened amongst the medical (n=1664) and nursing staff (n=2463) in daily contact with patients of LUH and amongst paramedics of the Canton of Vaud (n=290) during the screening period. Participation was voluntary and anonymous. In order to maximize data reliability, participants answered the online survey under investigator supervision so as to avoid biased responses (internet queries, discussions between colleagues). Participants were screened during scheduled patient hand-offs, seminars or group meetings, as permitted by heads of units. Participants completed the online survey using tablets or smartphones (participants' or provided by the investigators).

Statistical analysis
We described participants characteristics and survey responses across professions: 1) paramedic, 2) nurses, and 3) physicians. Continuous variables were summarised as means and standard deviations [SD] and categorical variables as frequencies and proportions. We also evalutated study participants representativeness of the LUH population of nurses and physicians using Student t-test and Pearson chi2 test for comparing mean ages and proportions of female professionals. In order to assess associations between sepsis awareness and proxies of prior medical and sepsis training, we used univariate logistic regression models with age, continuing education (yes vs. no or last training < 3 years), professional experience ( > 5 years vs. ≤ 5 years), knowledge self-evatuation of sepsis knowledge (good-very or good vs. others) and eld of practice (ED, ICU, Medicine, Paramedic, Psychiatry, or Surgery) as explanatory variables. For each model, we estimated the odds ratio (OR) of correct vs. incorrect answer as well as is 95% con dence interval (95%CI). All tests for statistical signi cance were two tailed (p<0.05). We performed statistical analyses using the computing environment R version 4.0.2 (R Development Core Team, 2005) and Prism version 9.0.0 (Graphpad Software).

Participants
Of the 4,417 eligible health-care professionals (HCPs) comprising 290 paramedics, 2,463 nurses and 1,664 physicians, 1,216 were screened (see methods) for participation. Among the 1,216 selected HCPs, 1,116 completed the survey while 46 refused to participate and 54 were excluded because of incomplete answer (91.7% of screened HCPs and 25.2% of total HCPs) ( Figure 1A). All clinical areas were represented, though representation of profession and specialty varied ( Figure 1B). Table 1 shows the characteristics of the participants. Participating nurses's mean age was not different from the institutional nurses mean age (p = 0.1), while participating physicians' mean age was lower than the institutional mean (p = 0.001). Gender distribution revealed an overrepresentation of male participants for nurses (p = 0.03) but was balanced for physicians (p = 0.1). T-test comparison for sample age distribution and (c 2 ) for sex difference analysis in order to assess representativity. Data from paramedics companies not available for representativity analyse N/A s.

Participant's training and perceptions
We next assessed sepsis training. In general 69.4% of HCPs reported a prior training on sepsis. Because our study launched in January 2020, we looked at the 2017-2019 as the period for training including the 3rd draft of consensus de nitions. The majority of participants (73.7%) reported no sepsis-speci c training in the last 3 years and 31.6% reported never having attended a sepsis-speci c ( Table 2). Conversely, 26.3% of participants reported a training within the last three years respectively. Nurses (82.9%) and paramedics (75.8%) reported more often no training or a training more than 3 years prior compared to physicians (56.6%). Next, participants were asked to evaluate their knowledge and management skills on sepsis using a 5-category (very good/good/average/fair/poor) Likert scale ( Figure  2). Overall, 26.3 % of participants graded their knowledge as very good and good (Figure 2A). Similarly, 35.8% graded their management skills as very good and good. An analysis by category of health care professionals revealed similar trends ( Figure 2B) although statistically signi cant differences between professions were noted with physicians selfevaluating best and paramedics selfevaluating worst, whether regarding knowledge or management. We then asked participants to provide answers regarding their perception of sepsis (medical emergency, morbidity/mortality, evaluation, its link to organ dysfunction/propensity to develop under antimicrobial thearpy). Participants were cognizant of the severity and the necessity for emergent management of sepsis (87.4 and 95.6%, respectively, strongly agree or agree) (Figure 3). They estimated sepsis and septic shock mortality to be 40% and 50%, respectively. They recognized the association between organ dysfunction and infection for sepsis can arise under antimicrobial therapy (Figure 3). A majority of participants (74.9%, 67.7% and 96.1% respectively) identi ed age, active cancer and immunosuppression as risk factors but only half (52.8%) recognized a prior septic event as such. volume resuscitation/SIRS score > 2 points/ bacteremia / blood lactate > 2 mmol/l / SOFA score > 10 points), 17.0% of physicians de ned septic shock according to Sepsis-3 (hemodynamic instability requiring vasopressors despite adequate volume resuscitation and serum lactate of more than 2 mmol/l). Finally, nearly 50% of the physicians associated the qSOFA ( Figure 4B) and SOFA ( Figure 4C) scores with sepsis. Yet, only 42.1% of physicians reported having computed the SOFA score previously and 17.0% correctly identi ed the components of the qSOFA score.

Management
Participants were next asked what recommended timing for intervention was (choice: within 1h/3h/6h/12/24h), the vast majority of participants (88.5%) chose interventions within one to three hours of sepsis recognition. Figure 5 is a clinical vignette of a patient with suspected sepsis and a qSOFA score of 2 assessing the use of diagnostic tools and management skills show by the participants according to profession. Nearly all paramedics (90.6%) recognized the need for a rapid transfer to ED ( Figure 5A). However, 42.3% considered vital signs monitoring as warranted. The vast majority of nurses recognized the need for immediate medical assessment (93.1%), monitoring of vital signs (82.3%), most requested blood cultures (70.1%) and half requested drawing blood for laboratory analysis (51.2%) ( Figure 5B). Physicians identi ed vital signs monitoring (92.0%), blood culture draw (96.0%), lactate measurement (89.1%) and imaging (77.9%) as critical diagnostic steps ( Figure 5C, diagnostic tests). Once sepsis was con rmed (presence of infection plus a SOFA score of 3), the majority of physicians chose the assessment for intravenous access (87.1%), administration of broad-spectrum antibiotics (91.7%) and uid resuscitation (76.1%) as immediate therapeutic interventions ( Figure 5C therapeutic interventions).

Discussion
Our study is a foundational analysis of the sepsis quality of care improvement project at LUH for the strategic development plan of the 2019-2023 period. We identi ed signi cant de ciencies in sepsis awareness amongst nurses and physicians of our university tertiary care center and local paramedics. A minority of healthcare professionals in our institution are aware Sepsis-3 consensus de nitions for sepsis. Similarly, a minority of staff are aware SOFA and qSOFA scores. Correspondingly, a minority of paramedics, nurses, and physicians self-evaluated as good or very good for sepsis knowledge and management. Importantly, these ndings are associated with a lack of continuing education.
Despite the fact that Sepsis-3 consensus was released more than four years ago, 1 despite its incorporation in the core of the Lausanne medical school curriculum or in institutional tools such as the LUH's guide for empirical antimicrobial therapy, our results show a lack of penetrance of the latest sepsis de nition. 26, 27 The lack of speci c continuing education accounts primarily for this. Only 18.5% of participants reported having attended sepsis-speci c training in the previous three years. Thus, the vast majority of participants have not been exposed to training on the new sepsis de nitions and are not familiar with the qSOFA score. This was striking for both paramedics and nurses that are at the front line of sepsis recognition. Nurses spend comparatively more time than physicians at the patient bedside 28 and early recognition of nosocomial sepsis by nurses increases 30-days survival. 29 Similarly, only one-third of physicians are aware of the current sepsis de nition. One-fth of physicians using the de nition of hemodynamic instability in addition to infection, may lead to delays in the recognition of septic patients. Furthermore, the low rate of calculation of a SOFA score by physicians implies that documentation of sepsis in discharge summaries and electronic medical records is also compromised. We are accumulating evidence in the context of our quality of care project (data not shown). As a consequence, sepsis epidemiology at the institutional level may be severely affected.
These observations support further -and regular -training incorporating Sepsis-3 consensus de nitions in our institution as studies support continuous training to improve sepsis awareness amongst participants. 30,31 Because a minority of participants, whether nurses, paramedics or physicians rated their knowledge and management skills as good or very good, there is a major opportunity for continuing education.
To the best of our knowledge, the present study is the rst assessing sepsis-3 knowledge surveying large sample size, multiple professions across all adult departments of a tertiary care center, thus representing all persons implicated in adult sepsis care. Multiple studies have assessed sepsis awareness, 8,10,12−18 however, only three probed Sepsis-3, all of which were limited in scope: Nucera and coworkers assessed Sepsis-3 awareness among nurses and physicians and found similar de ciencies, however the study was limited to 181 persons and excluded oncology wards. Consistent with our study, they identi ed major de ciencies in awareness particularly pertaining to scores and de nitions. However, the large sampling in our study enables a better resolution of de ciencies. As an example, the capacity to de ne sepsis according to sepsis signi cantly better in ICU, ED and internal medicine compared to surgery and psychiatry. Mulders and co-workers assessed a very different setting, interviewing general practitioners, but found similar observations with very low penetrance of SOFA score-based sepsis de nitions and qSOFA score-based assessment. Finally, a survey limited to ICU physicians in China revealed a limited familiarity with only 16% of 366 physicians using Sepsis-3 consensus de nitions. 11 Studies relating to Sepsis-2 de nitions had already identi ed signi cant de ciencies: Seymour and co-workers found paramedical staff struggling to de ne sepsis. 8 Abdul Rahman and colleagues identi ed de ciencies among nurses and physicians in the ED. 12 However, sepsis-speci c training is associated with signi cant improvement in such de ciencies. 18 This study's strengths include the number of participants, the participation rate, the combined assessment of nurses, physicians and paramedics and the breadth departments of adult medicine assessed. Furthermore, methodology with direct supervision of participants taking the survey ensures high-quality data collection. It also has limitations: The survey was built on knowledge, attitude and practice of health care professionals towards sepsis based on literature review and focus groups of experts clinicians. 32 It was tested in iterative pilots and revisions among intended respondents. However, we did not perform subsequent reliability (internal consistency, test-retest reliability, or inter-rater reliability) or construct validity assessment through a Crohnbach's alpha test due to the various formats of questions. Second, it is limited to a single center and results may not be generalizable, although they are consistent with previous studies. Third, we have a slight imbalance towards younger age for participants and male sex for nurses. The exclusion of staff not having daily contact with patients likely accounts in part for the age bias. The propensity of male nurses to take the test is more di cult to explain; it might re ect a more prevalent part-time activity amongst females compared to males (average full time equivalent 0.73 vs. 0.82). Fourth, we had signi cant discrepancies in the various hospital departments. This was strongly in uenced by differences in availability (seminars, availability on the ward).

Conclusion
Our study reveals signi cant de ciencies in sepsis awareness at an institutional level, in all professions and departments four years after the introduction of Sepsis-3 consensus de nitions. Their penetrance is limited and bedside tools are not mastered. It is associated with a lack of speci c training, setting the roadmap for sepsis-education, targeting all professions tailored to their activity. The improved recognition and monitoring among nurses and paramedics and de nition implementation among physicians with sustained continuing education is a critical step to our quality of sepsis care improvement program. The local institutional review board (CER-VD, Lausanne, Switzerland) waived written consent for this research project.

Consent for publication
All participants, upon taking the survey, were asked to consent for publication (e-consent within the survey).

Availability of data and materials
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request. Sepsis awareness. Assessment of sepsis characteristics or features (i.e. urgency of care, severity, need for prompt evaluation and context of appearance) according to a 5-category Likert scale by study participants.