INTRODUCTION

Sepsis consists of potentially fatal organic dysfunction triggered by the dysregulation of the host's immune response to infections. Sepsis is diagnosed from the association between multiple organ failure and suspected or confirmed infectious processes.^1–3 Classically, the characteristic organic dysfunction of sepsis is diagnosed by scores of two or more points on the Sequential Organ Failure Assessment (SOFA) scale, which assesses the functioning of the central, cardiovascular, respiratory and renal nervous systems, in addition to liver and platelet function¹. Sepsis and septic shock are associated with significant morbidity and mortality in individuals admitted to the intensive care unit (ICU) because they acquire hospital infections more frequently.^1,4

Risk factors for sepsis are extremes of age, a sedentary lifestyle, immunosuppression, alcohol consumption, cancer, diabetes, and prolonged use of catheters or other conditions that compromise skin integrity. However, among patients with established infections, other risk factors for sepsis have been reported, such as comorbidities and genetic factors of the host and infectious agent¹. ICU admission, previous hospitalizations, bacteremia, obesity and community-acquired pneumonia have also been described as potential risk factors for sepsis⁴.

Today, sepsis represents a growing challenge for health systems in countries around the world. With the advancement of health technologies, the longevity of the populations most vulnerable to sepsis, such as elderly and immunocompromised patients have increased, thus leading to a subsequent increased incidence of sepsis. It is estimated that 17 million cases of sepsis occur worldwide each year, but estimates are likely to underestimate the actual number of cases⁵.

It is believed that 600,000 of the annual cases of sepsis occur in Brazil, where the lethality rates of sepsis and septic shock exceeded 30% and 60%, respectively, in 2003⁵. In 2015 alone, 110,049 hospitalizations for sepsis were recorded in the country, of which only 925 occurred in the Federal District. However, the federative unit had the highest average length of hospital stay, at 20.11 days, and ranked eighth in the sepsis mortality rate, which corresponded to 50.38%⁶.

In addition to triggering significant morbidity and mortality, sepsis negatively impacts health systems in relation to bed occupancy and hospital costs. Estimates indicate that approximately 30% of Brazilian ICU vacancies are occupied by individuals with sepsis or septic shock. In addition, for each hospitalized patient, the average hospitalization cost is $10,595.00⁵.

Although sepsis has higher lethality rates in underdeveloped or developing countries5, the vast majority of scientific publications on the subject come from developed countries. Perceived disproportionality skews the available evidence regarding sepsis, which may compromise understanding of the subject in contexts of countries with limited resources¹. Therefore, it is likely that the underestimation of the incidence of sepsis in underdeveloped countries compromises the direction of resources for the study and treatment of sepsis regionally.

Therefore, especially in countries with limited resources, such as Brazil, the expansion of epidemiological data on sepsis, along with the implementation of public policies aimed at the prevention and proper management of sepsis nationally, and determining the underlying cause of death would all help to improve healthcare for septic patients. Thus, this study, conducted in the Federal District, aimed to evaluate the accuracy of mortality indicators due to sepsis in 2018.

METHODS

Study design and context

A validation study was conducted in the Federal District, which has the smallest territory and one of the smallest populations among the federative units in Brazil. The estimated population of the Federal District in 2021 corresponded to 3,094,325 individuals⁷. At the same time, the federative unit has the highest population density and the highest income in the country., in addition to the second highest rate of life expectancy at birth, which was estimated to be 77.6 years in 2014⁸.

Data source, investigation steps and data collection procedures

Data collection for this research was carried out at the State Department of Health of the Federal District (SES-DF). This department manages mortality information systems, as well as diseases, injuries and compulsory notification events in the Federal District⁹. In Brazil, mortality data are regulated through the Mortality Information System (SIM), whose standard document corresponds to death certificates.¹⁰

The information was collected through a standardized form (elaborated by the authors) based on the screening and investigation of potentially eligible death certificates. During screening, the garbage codes used in field 40 of the document, referring to the causes of death related to sepsis, were identified. Medical records, examinations and reports issued by health institutions linked to death certificates were accessed by the exclusive SES-DF system and analyzed. The information collected is presented and described in the variables section.

Population and eligibility criteria

In this study, the population was composed of people who died in the Federal District in 2018, whose underlying causes of death indicated sepsis, represented in the International Classification of Diseases version 10 (ICD 10) by codes A40, A40.0, A40.1, A40.2, A40.3, A40.9, A41.0, A41.1, A41.2, A41.3, A41.4, A41.5 and A41.9; all death certificates of children under one year of age were excluded. Sepsis is considered a garbage code¹¹. Garbage codes are basic causes of death that do not provide fundamental information about the triggering circumstance of death, so a valid underlying cause of death is not understood.

Variables

The main variable of this study is sepsis, as the underlying cause of death, which is recorded on the death certificate. In addition, other variables were used to characterize the population, which included age ranges (1 to 11 years, 12 to 18 years, 19 to 39 years, 40 to 59 years, and 60 years or more), sex (male and female), race or skin color (Asian, white, indigenous, African American and black), years of study (none, 1 to 3 years, 4 to 7 years, 8 to 12 years, 12 years or more), marital status (divorced/single/widowed and married/stable union), and place of death (hospital, other health facilities and other places).

Data analysis

After data collection, the underlying causes of death from sepsis were compared, recorded in the original statements and the causes defined after the investigation by the multidisciplinary team, which was considered the gold standard. Subsequently, the death certificates were divided into four groups: true-positive, false-positive, true-negative and false-negative.

In the true-positive group, the death certificates that remained with the same underlying cause after the investigation, were grouped. The false-positive group included death certificates that initially had the underlying cause of death from sepsis but were reclassified to others after being investigated. In the true-negative group, all declarations that did not present the cause of sepsis and remained with the original underlying cause of death after the investigation were grouped. In the false-negative group, all statements that had another underlying cause and were reclassified for sepsis after investigation were included.

Finally, sensitivity, specificity, positive predictive value, negative predictive value, positive likelihood ratio and negative likelihood ratio were estimated, with 95% confidence intervals. Data analysis was performed using the data analysis andstatistical software (STATA) statistical package, version 16.

The research was conducted respecting the ethical standards required in Resolutions 466/2012, 510/2016 and 580/2018 of the Ministry of Health and was approved by the Research Ethics Committee (CEP) of the School of Health Sciences (ESCS), CAAE:15457719.0.0000.5553, under opinion number 4,482,501. The researchers involved claim no conflicts of interest related to the development of the study.

RESULTS

At the end of the data analysis, 6,244 death certificates were evaluated. The population analyzed had a mean age of 62.78 years (standard deviation ± 22.35) and a median of 67 years, whose amplitude was 1 to 112 years. Most deaths occurred among patients aged 60 years or older, male, white skin color, divorced, single or widowed, and with 1 to 3 years of study. In most deaths, the place of occurrence was the hospital. Before being investigated, 233 declarations presented sepsis as the underlying cause, which corresponds to 3.74% of the population. After the investigation, only 35 statements had sepsis maintained as the underlying cause, which represents 0.56% of the results, as shown in Table 1. Therefore, the investigation of death certificates rectifies the underlying cause of 84.97% of the declarations.

Table 2 shows that the sensitivity of the evaluated statements about sepsis was approximately 0.93% (_95% CI: 0.64 to 1.29). On the other hand, there was a high probability of an individual who did not develop sepsis obtaining a statement that did not present sepsis as the underlying cause, since the specificity found was 92.00% (_{95% CI}: 90.90 to 93.10). The chance of a death certificate whose underlying cause was sepsis belonging to a septicemic individual was 15.00% (_{95% CI}: 10.70 to 20.30), which consists of the positive predictive value of the evaluated statements. The probability of a declaration that the individual who did not develop sepsis as the underlying cause was 38.00% (_95% CI: 36.80 to 39.30), which represents the negative predictive value.

Additionally, according to Table 2, it was found that the chance of a death certificate indicating sepsis as the underlying cause is 0.12 times higher (_{95% CI}: 0.08 to 0.16) for a septicemic person than for an individual with another underlying cause of death, which consists of the positive likelihood ratio of the evaluated statements. At the same time, the probability of a declaration not indicating sepsis as the underlying cause is 1.08 times higher for septicemic people than for individuals who have not developed sepsis, which corresponds to the negative likelihood ratio found.

.Confidence interval at 95%; N: sample; CI: confidence interval.

DISCUSSION

A total of 233 deaths that occurred in the Federal District in 2018 presented sepsis as the underlying cause in the death certificate before being investigated. Generally, the profile of the population studied consisted of males; White; divorced, single or widowed; 60 years or older; and low educational level. After the investigation, there were 35 statements left in which sepsis was the underlying cause of death. Thus, the investigation rectifies the underlying cause of 198 statements, which corresponded to 84.97% of the original 233. The death certificates evaluated showed low sensitivity and high specificity.

The profile of the participants in this study is similar to that described in the literature. In 2015, 53 patients were admitted to the ICU due to sepsis or septic shock at the University Hospital of Brasília (HUB). Most of the hospitalized patients were also male and older than 60 years, in addition to living in the Federal District or Goiás. Additionally, 90.60% of them had pathological antecedents, including systemic arterial hypertension (SAH), diabetes mellitus (DM), maternal diseases and previous episodes of sepsis.¹²

Another investigation corroborated the findings of the survey, conducted in 2017, where 37,082 death certificates were investigated, which occurred in 17 municipalities in northern Brazil. During the screening of the documents, some of the underlying causes of death considered garbage codes were classified as priority, which included sepsis. After the investigation, 79.1% of the statements coded with garbage codes had the underlying cause rectified, and the number of deaths whose underlying cause was sepsis decreased by 24.9%.¹³

However, inadequate completion of the underlying cause in death certificates is not restricted to the aforementioned sites. Between 2007 and 2016, more than 4 million of the deaths in Brazil were reported with garbage codes as the underlying cause, which represented 34.0% of the total number of deaths. Thus, the investigation of these statements is essential for the improvement of health data.¹³ According to DATASUS, in 2015 only, sepsis was responsible for 110,049 hospitalizations in the country, which incarcerated costs of R$400,387,078.76.6 In the event that health actions and public policies promote the effective improvement of the population's health conditions, in the tangent of sepsis to other diseases and diseases with high prevalence in Brazil, the quality of available health information is crucial.^13-15

Moreover, this study presented limitations. Among them, the main one consists of inadequate completion of the death certificates analyzed, which was verified both in the face of unfilled fields and incorrect information. Of the 6,244 statements, 694 did not present the field referring to the years of study of the deceased completed, corresponding to 11.11% of the sample. It is likely that the misconceptions observed in the statements are due to medical ignorance about the importance of proper filling of all fields of the document, as well as the malpractice of the medical class in informing the chain of pathological events questioned in the field of cause of death.^13,16

In addition, the second limitation of this study is the existence of outdated, incomplete or even blank medical records, referring to individuals who died, which was also reported in the study conducted at the hub in 2015⁶. Inadequate medical records may be related to the negligence of the responsible health professionals and the existence of multiple disintegrated electronic medical records systems, since care provided in certain health institutions may not be recorded in the medical records to which the researchers had access.

However, all death certificates whose underlying cause was a garbage code were investigated individually by a multidisciplinary and previously qualified team. Thus, in addition to rectifying the underlying epidemiological cause in the Mortality Information System by investigating the medical records and medical examination reports of individuals who died, the other fields filled out incorrectly or neglected were also corrected during the typing of the document whenever possible. The death review committees (CROs) of health institutions where the deaths occurred were also triggered in the face of outdated, incomplete or blank medical records to obtain the information necessary for the correct investigation of deaths. Moreover, this is the first study in the Federal District to evaluate the accuracy of death certificates that indicate sepsis as the underlying cause.

Death certificates that occurred in the Federal District in 2018 that indicated sepsis as the underlying cause of death, present low sensitivity and positive predictive value, which indicate the low reliability and low usefulness of mortality indicators due to sepsis for planning health actions and public policies. Nevertheless, the low accuracy of the statements demonstrates the lack of reliability of the diagnostic investigation of sepsis and, above all, of the completion of the declarations. The need for health institutions to implement programs aimed at improving the proper completion of death certificates and care for septicemic individuals is evident. Moreover, it is crucial that undergraduate medical courses and medical residency programs include the task of correctly completing death certificates as an educational objective.

References

1. Cecconi M, Evans L, Levy M, et al. Sepsis and septic shock. The Lancet. 2018;392(10141):75–87. https://doi.org/10.1016/S0140-6736(18)30696-2

2. Napolitano LM. Sepsis 2018: Definitions and Guideline Changes. Surgical Infections. 2018;19(2):117–25. https://doi.org/10.1089/sur.2017.278

3. Taeb AM, Hooper MH, Marik PE. Sepsis: Current Definition, Pathophysiology, Diagnosis, and Management. Nutr Clin Pract. 2017; 32(3): 296–308. https://doi.org/10.1177/0884533617695243

4. Neviere R. Sepsis syndromes in adults: Epidemiology, definitions, clinical presentation, diagnosis, and prognosis. UpToDate. 2020. https://www.uptodate.com/contents/sepsis-syndromes-in-adults-epidemiology-definitions-clinical-presentation-diagnosis-andprognosis?search=sepse%20fatores%20de%20risco&source=search_result&selectedTitle=1~150&usage_type=default&display_rank=1

5. Instituto Latino-Americano para Estudos da Sepse. Sepsis: a public health problem. Brasília: Conselho Federal de Medicina; 2015. 89 p.

6. Miquelin PRS, Reis GR. Comparison between the morbidity and mortality rates of patients with septicemia in all federation states and the Federal Districtl. Amazônia: Science & Health. 2016;4(4):20–4. http://ojs.unirg.edu.br/index.php/2/article/view/1374

7. Instituto Brasileiro de Geografia e Estatística. Cidades. https://cidades.ibge.gov.br/brasil/df/panorama

8. Instituto Brasileiro de Geografia e Estatística. SIDRA. Sustainable Development Indicators. https://sidra.ibge.gov.br/tabela/1174.

9. Secretaria de Saúde do Distrito Federal. Vigilância à Saúde. Brasília: Secretária de Saúde do Distrito Federal 2022. .https://www.saude.df.gov.br/vigilancia-a-saude

10. Ministério da Saúde. (BR), Fundação Nacional de Saúde. Instructions Manual for Filling the Death Certificate. 3ª edição – Brasília: Assessoria de Comunicação e Educação em Saúde (ASCOM); agosto de 2001.

11. Santos MR, Cunha CC, Ishitani LH, et al. Deaths from sepsis: underlying causes of death after investigation in 60 Brazilian municipalities in 2017. Revista Brasileira de Epidemiologia. 22(supl. 3), 2019. https://doi.org/10.1590/1980-549720190012.supl.3

12. Pereira JM. Profile of patients diagnosed with sepsis in Intensive Care Unit in a university hospital in Distrito Federal. [Brasília]: Universidade de Brasília (UnB); 2018.

13. Benedetti MSG, Saraty SB, Martins AG, et al. Evaluation study of the garbage codes research project in the Northern region of Brazil. Rev Bras Epidemiol. 2019;22(supl. 3):e19006. https://doi.org/10.1590/1980-549720190006.supl.3

14. Akhade SP, Dash SK, Akhade KS. The knowledge assessment and reducing the errors of medical certificate of cause of death with sensitization training of physicians: A quality improvement intervention study. Journal of education and health promotion. 2022; 11, 1-7. https://doi.org/10.4103/jehp.jehp_502_21

15. Makinde OA, Odimegwu CO, Udoh MO, et al. Death registration in Nigeria: a systematic literature review of its performance and challenges. Global health action, 2020; 13(1). https://doi.org/10.1080/16549716.2020.1811476

16. Nyondo T, Msigwa G, Cobos D, et al. Improving quality of medical certification of causes of death in health facilities in Tanzania 2014-2019. BMC health services research, 2021; 21(Suppl 1). https://doi.org/10.1186/s12913-021-06189-7