Epidemiology, management and costs of sepsis in Spain (2008–2017): a retrospective multicentre study

Abstract Objective: To update the profile of patients attended with sepsis in specialised care centres in Spain, to analyse in-hospital mortality, disease management and costs between 2008 and 2017. Methods: Admission records registered between 1 January 2008 and 31 December 2017 obtained from a Spanish National hospital discharge database for public and private hospitals. Centres are responsible for data codification, evaluation and confidentiality. The database is validated internally and is subjected to periodic audits. Files corresponding to patients with sepsis and septic shock were selected by means of the International Statistical Classification of Diseases and Related Health Problems, 9th version and 10th version codes. These criteria claimed 311,674 records of 288,211 patients. Direct medical costs of secondary healthcare include expenses derived from the admission: examination, medication, treatment and costs of nutrition, personnel, medical equipment and resources. Results: More than 53% of all patients were males, with a mean age of 73.0 years. Fifty-one percent of the identified admissions were due to a sepsis without organ dysfunction, 21.5% to sepsis with organ dysfunction, and 27.3% registered a septic shock. The incidence of sepsis increased 2.7 times between 2008 and 2017, reaching a hospital incidence of 5.7 per 10,000 inhabitants in 2017. Case fatality rate (CFR) was 23.2% and 35.0% in patients without and with organ dysfunction in 2017, respectively, and 42.9% in patient with septic shock, decreasing over time. Mean annual direct medical costs of specialised care over the study period were €6664 and €8084 per patient in patients with sepsis without and with organ dysfunction, respectively, and €11,359 per patient in those with septic shock. Conclusions: The social and economic burden of sepsis in Spain continues to grow (incidence, total costs). Despite its general decreasing trend, CFR remains elevated, thus, patients could benefit from further research and protocol revision.


Introduction
Sepsis is defined as a life-threatening organ dysfunction caused by a dysregulated host response to infection in the Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3) 1 . Sepsis can be caused by a wide range of organisms, and have varied origins, principally respiratory, abdominal or urinary 2 . It is a major cause of morbidity and mortality, remaining a challenge for clinicians and healthcare systems. Sepsis is assumed to represent a significant burden worldwide, with a global incidence of 43.7 per 10,000 persons-year, with several studies indicating an increasing trend over the past decades 3,4 .
In Spain, the incidence of sepsis has been evaluated several times over the past decades, revealing a similar trend and an incidence of 8.7 per 10,000 persons between 2006 and 2011, as evaluated in the Spanish national hospital discharge database including sepsis diagnosed in patients hospitalised for other reasons 5,6 . On the contrary, mortality appears to be decreasing, which serves as an indicator of the success of recent awareness campaigns and updated treatment guidelines 6 . Globally, the implementation of specific sepsis treatment guidelines and protocols has been critical for the decrease in sepsis-related mortality. The guidelines developed by the Surviving Sepsis Campaign (SSC) published in 2004 and updated in 2016 have been associated with improved outcomes worldwide 7 ; these focused on a series of bundles to be completed within 3 or 6 h after the admission including blood culture, the administration of broad spectrum antibiotics and treatment to control hypotension and the repeated examination of vital signs and cardiovascular function [7][8][9] . Similarly, Sepsis-3 recommends the use of the Sequential Organ Failure Assessment (SOFA), which correlates with an increased probability of mortality, to clinically characterise patients 1,10 . These criteria have provided a tool for early recognition that needs to be continuously updated 11,12 .
The revision and improvement of protocols is one of the applications of epidemiology revisions, hence the growing interest on obtaining detailed and updated epidemiological data including sepsis and septic shock via this study, we aim to provide updated data regarding the characteristics of patients attended with sepsis in hospitals and specialised care centres in Spain, incidence, in-hospital mortality, disease management and costs, and to evaluate any temporal trends between 2008 and 2017.

Study design and setting
Admission records of patients admitted in specialised care centres (hospitals and ambulatory) were analysed in a retrospective multicentre study, with a study period set from 1 January 2008 to 31 December 2017. Records were obtained from a Spanish National discharge database including public and private hospitals, which covers 90% of hospitals in Spain and is representative of all Spanish regions. The database includes anonymised admission data with diagnoses codified by means of the International Statistical Classification of Diseases and Related Health Problems, 9th version (ICD9) prior to 2016 and 10th version (ICD10) the years 2016 and 2017. Data codification is achieved at the hospital level by specialised doctors by using the Spanish ICD codification guides made available to health professionals. Hospitals are responsible for data codification, evaluation and confidentiality. The database is validated internally, and is subjected to periodic audits. In this process, errors and unreliable data are eliminated. Data inclusion was limited to admissions explicitly registered as sepsis or septic shock cases.

Data extraction
Files corresponding to admissions with sepsis or septic shock as primary diagnosis (cause of admission) were selected via the ICD9 codes: 038.0, 038.1x, 038.2, 038.3, 038.4x, 038.8, 038.9, 995.91, 995.92, 785.52, and ICD10 codes: A40.x, A41.x, R65.2x. A total of 311,674 records complied with these criteria, corresponding to 288,211 patients that were classified according to sepsis severity: with or without organ dysfunction (without septic shock), and those that registered septic shock during the admission. Subsequently cases were classified according to the pathogen into streptococcal sepsis, staphylococcal sepsis, sepsis due to anaerobes, sepsis due to other Gram-negative organisms, other specified sepsis (i.e. sepsis due to Enterococcus) and sepsis due to unspecified organisms.

Study variables
The database codifies hospital discharge data on the patient profile and admission details: patients' sex and age, Spanish region, date of admission, type of admission, date of discharge, type of discharge (including death), service that discharged the patient, length of stay, readmission rate, admission motive, secondary diagnoses registered during the admission, medical procedures performed and cost of the admission.

Data analysis
Single-patient data, obtained grouping recurrent admissions per each single patient, was used to characterise the population and calculate case fatality rate (CFR) defined as the proportion of in-hospital deaths registered in the study population. Sepsis age-adjusted incidence was extracted from the ministry of health database and was measured as the proportion of sepsis cases registered in specialised care centres within the population adjusted per age. All admission files were used to analyse the nature of admissions, length of stay, readmission (understood as a subsequent admission for the same cause within 30-days after discharge), medical procedures and costs. All the registered secondary diagnoses were identified by using ICD9 and ICD10 codes.
Descriptive values are presented in mean (standard deviation [SD]), length of stay is presented in mean (standard error [SE]). Odds ratio (OR) with a 95% confidence interval (CI) were calculated for deceased patients, using non-deceased patients as the reference group. Two-tailed T-student or one-way analysis of variance were used as appropriate and two-sample Z tests were used to test for differences in sample proportions, with a p < .05 considered statistically significant. Statistical analyses were performed using IBM SPSS Statistics for Windows, Version 20. Direct medical costs of specialised healthcare were calculated based on admission costs indicated in the database; these are calculated according to the standardised average expenses of admissions and medical procedures determined by the Spanish Ministry of Health, which includes all expenses derived from the admission: examination, medication, treatment and costs of nutrition, personnel, medical equipment and resources.
This study did not involve human participants and the database contains anonymised data with no identifying parameters, complying with the principles of Good Clinical Practice and the Declaration of Helsinki. In this context the Spanish legislation does not require patient consent and ethics committee approval 13 .

Results
The 288,211 patients admitted with sepsis and septic shock over the ten year period were included in the analysis of patient characteristics (Table 1). Across all groups, more than 50% of the patients were males, with a mean age of 73.0 years. Over 51% of all registered admissions were due to a sepsis without organ dysfunction, 21.5% to sepsis with organ dysfunction without septic shock and 27.3% registered a septic shock. In most of the cases the pathogen was not specified; when specified, gram-negative organisms were predominant. The patients with a streptococcal sepsis (principally E. coli and Streptococcus pneumoniae) were the youngest. Sepsis incidence measured in specialised care facilities was 5.7 per 10,000 persons in 2017, increasing 2.7 times over the study period ( Figure 1(A)). Most of the cases were registered among patients aged 70-90 years (Figure 1(B)).
CFR was 23.2% and 35.0% in patients without and with organ dysfunction in 2017, respectively, and 42.9% in patient with septic shock. CFR decreased significantly over the study period in patients with organ dysfunction and septic shock (2008 vs. 2017, p < .001) (Figure 2(A)). CFR was higher in older patients throughout the study period (Figure 2(B)).
Secondary conditions diagnosed upon admission are listed in Table 2. Acute conditions associated with organic dysfunction were predominant. Chronic conditions as essential hypertension, diabetes mellitus and malignant neoplasms, were found in between 20% and 30% of admissions, with slight differences between patient groups. In addition, potential associations appeared in the group of deceased patients versus the remaining population. Acute respiratory failure was more common in deceased patients ( In terms of healthcare use, over 96% of admissions in all groups were due to emergencies and attended by internal medicine departments in 63.5%, 67.1% and 47.1% of the cases in patients admitted due to sepsis without and with organ dysfunction and septic shock, respectively (Table 3). Mean length of stay was of 10.7 (SE ¼ 0.03) in patients without organ dysfunction and increased to 11.8 (SE ¼ 0.05) days in patients with organ dysfunction (p < .001). In admissions that registered septic shock this was of 14.3 (SE ¼ 0.06) days (septic shock vs. no septic shock, p < .001). Deceased patients registered a mean length of stay of 8.6 days (SE ¼ 0.04) (deceased patients vs. non deceased, p < .001). The 30-day readmission rates oscillated between the 10 and 25%.
The estimated mean annual direct costs of specialised care of sepsis were e6664 and e8084 per patient in patients with sepsis without and with organ dysfunction, respectively, and e11,359 per patient in those with septic shock (between groups, p < .001) (Table 4), which, considering all patients (N ¼ 288,211), equalled a total of e247,412,912 per year.   Overall, costs per patient decreased between 2011 and 2014 and increased again between 2014 and 2017 ( Figure 3). Mean admission cost was e7940 over the study period and varied with the length of hospital stay. Mean admission cost per stays up to Q1 and longer than Q3 were e5825-7779; e6703-10,132; and e8035-17,735 in admissions due to sepsis without or with organ failure and septic shock, respectively (within groups, p < .001).

Discussion
Several studies worldwide have described an increasing trend in the number of sepsis cases, which has correlated with an increasing interest in analysing its incidence, nature, management and costs. The incidence of sepsis has been previously evaluated in Spain, covering a period of 14 years (2000-2013) 5,6 . Data showed an increase that went from 3.3 per 10,000 personsyear (2000-2004) to 4.5 (2010-2013), similarly to estimations in a global scale 2,6,14 . The same trend has been observed in smaller studies focused on patient subpopulations as those diagnosed with type 2 diabetes mellitus 15 . Herein the incidence per 10,000 was 2.1 in 2008, increasing to 5.7 in 2015. A small decrease was observed the year 2016 after the introduction of ICD10, which limits comparability; yet, between 2016 and 2017 the increasing trend continued. Overall, this increase in incidence could be explained by the improved coding of sepsis that has been demonstrated in certain European countries, as Germany, as well as the increased sepsis awareness 16 .
On the contrary, data confirms a decrease in mortality, measured as the reduction of CFR. This parameter was 45.4% in Spain in 2006, and decreased to 40.2% in 2011 5 ; our data estimated a CFR of 23.2% and 35.0% in patients without and with organ dysfunction in 2017, respectively, and 42.9% in patients with septic shock, overall, decreasing over the study period presumably due to the improved treatment algorithms and guidelines promoted at the European and global level 1,7,8 . In Spain, European protocols are used, with compliance rates that have increased considerably since 2008 directly affecting health outcomes, reducing mortality rates in patients with severe sepsis and septic shock [17][18][19][20] . In addition, an effect of the increased register of sepsis cannot be discarded.
Regarding patients' profile, mean age registered upon admission was 73.0 (18.6) years. Data from the period 2000 to 2013 showed a significant temporal increase of patients' age, that in that case averaged 69.7 (20.0) years 6 .
A predominance of cases caused by Gram-negative bacteria, principally E. coli, would coincide with previous estimations, while the origin of the infection could not be directly evaluated 5,21 . Most admissions registered symptoms of organ dysfunction, although chronic conditions including hypertension and diabetes were also present. Renal, cardiac and respiratory symptoms were associated with in-hospital mortality, as well as the disorders of fluid electrolyte and acidbase balance and malignant neoplasms. On the other hand, even though liver dysfunction is considered a common  symptom of sepsis, liver affectation was not depicted clearly in this population 22 ; the lack of laboratory test results impeded a formal analysis of laboratory abnormalities. Most of the admissions were not scheduled, with hospital stays over the 10 days. Relatively high 30-day readmission rates have been previously described in patients with sepsis, attributed primarily to infections and pulmonary complications 23 .
These factors had an influence on the medical costs of the disease in specialised care, known to correlate with age, severity, admission and attention characteristics 24,25 . Overall, the mean annual cost per patient in the study period was e6664 and e8084 in patients with sepsis without and with organ dysfunction, respectively, and e11,359 per patient in those with septic shock; whereas in the period 2000-2013 average costs reached the e9090 following an increasing trend 6 . Herein, cost per patient appeared to decrease between the years 2011 and 2014, likely to be an effect of the economic crisis that had great effects on the Spanish medical and pharmaceutical expenditure in that period 26,27 . Data suggest a recovery of the increasing trend after 2014. As expected, admission costs increased with longer hospital stays.
A series of factors limited the results of this study. The update of the database codification to ICD10 in 2016 generated inconsistencies in the data, likely to derive from centres not adhering to the new system. Incidence was calculated using only patients admitted with sepsis in specialised care centres; this could origin an underestimation of this rate. Data selection method should be taken into account for data interpretation. Finally, the lack of laboratory test results in admission files hampered the evaluation of organ dysfunction and overall severity of the symptoms.

Conclusions
Sepsis is a major cause of mortality in Spain and represents a significant economic burden that continues to grow. Patients with sepsis with organ dysfunction and septic shock present elevated case fatality rates (35.0% and 42.9% in 2017), despite its globally decreasing trend presumably due to the successful application and revision of clinical protocols. Research to corroborate and update these results will be necessary in the upcoming years.

Declaration of funding
This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.