Article Text
PDF
Abstract
Objectives In the early stage of sepsis, identifying high-risk paediatric patients with a poor prognosis and providing timely and adequate treatment are critical. This study aimed to evaluate the effect of average body temperature within 24 hours of admission on the short-term prognosis of paediatric patients with sepsis.
Design A retrospective cohort study.
Setting A single-centre, tertiary care hospital in China, containing patient data from 2010 to 2018.
Participants 1144 patients with sepsis were included.
Intervention None.
Primary and secondary outcome measures The main outcome measure was in-hospital mortality, which was defined as death from any cause during hospitalisation. The secondary outcome was the length of hospital stay.
Results The LOWESS method showed a roughly ‘U’-shaped relationship between body temperature on the first day and in-hospital mortality. Multivariate logistic regression showed that severe hypothermia (OR 14.72, 95% CI 4.84 to 44.75), mild hypothermia (OR 3.71, 95% CI 1.26 to 10.90), mild hyperthermia (OR 3.41, 95% CI 1.17 to 9.90) and severe hyperthermia (OR 5.15, 95% CI 1.84 to 14.43) were independent risk factors for in-hospital mortality. Compared with other variables, the Wald χ2 value of temperature on the first day minus the degree of freedom was the highest.
Conclusions Whether hypothermic or hyperthermic, the more abnormal the temperature on the first day is, the higher the risk of in-hospital death in children with sepsis.
- Paediatric intensive & critical care
- Paediatric infectious disease & immunisation
- Paediatric intensive & critical care
- PAEDIATRICS
Data availability statement
Data are available upon reasonable request. The full data set used in this study is available from the first author at wanghb53@mail2.sysu.edu.cn. However, reanalysis of the full data for other use requires approval by the PIC Institute.
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.
Statistics from Altmetric.com
- Paediatric intensive & critical care
- Paediatric infectious disease & immunisation
- Paediatric intensive & critical care
- PAEDIATRICS
Strengths and limitations of this study
We studied the average body temperature within 24 hours of paediatric intensive care unit admission rather than the initial value.
The sample size of this study was relatively large and adequate confounding factors were included that might interact with admission temperature.
It was a single-centre retrospective study.
Due to database limitations, we had no data on some relatively important variables.
Introduction
Sepsis is one of the most common critical illnesses in paediatric medicine. Sepsis is caused by dysregulation of the body’s response to infection and can lead to microcirculation disorders, immune dysfunction, and tissue and organ dysfunction in paediatric patients.1 The condition of paediatric patients with sepsis is complex, and the symptoms are often atypical and may change or progress at any time, thereby endangering life. Among hospitalised patients in the paediatric intensive care unit (PICU), the prevalence of sepsis can reach more than 8%,2 and approximately one-quarter of deaths are caused by sepsis.3 Therefore, identifying high-risk paediatric patients with a poor prognosis in the early stage of sepsis and providing timely and adequate interventions are critical.4
Although considerable effort has been invested in the study of sensitive prognostic biomarkers, identifying these high-risk patients is still challenging.5 Some cytokines, such as serum soluble triggering receptor expressed on myeloid cells-1, preprotease and interleukins, have received increasing attention from clinicians because of their ability to identify sepsis at an early stage and predict prognosis6–10; however, the detection of these cytokines is limited by high costs and high operational skill requirements, that is, they are not commonly detected in hospitals. The paediatric risk of mortality score III (PRISM III) has been widely used in PICUs and has been shown to be accurate in assessing disease severity and the prognosis of paediatric patients with sepsis.11 However, the PRISM III score involves many indicators, the application is cumbersome, and the prediction accuracy is variable, thus limiting its practical application.
Fever is caused by the release of pyrogens, such as acute phase proteins, and most often occurs in infected patients.12 Physiologically, a continuous thermoregulatory response, including hypothermia and fever, has been recognised in the process of sepsis.13 Animal experiments have shown that the purpose of hypothermia is to increase body tolerance and reduce energy consumption, and fever helps kill pathogenic micro-organisms.14 However, the impact of these body temperature-related manifestations on the prognosis of the disease is controversial, especially in paediatric patients. At present, few reports on the relationship between body temperature and the prognosis of PICU patients with sepsis are available, and simple clinical indicators for assessing the prognosis of paediatric patients with sepsis are limited. Therefore, the purpose of this study was to investigate the predictive value of different body temperatures on admission on the short-term prognosis of paediatric patients with sepsis and to provide reliable markers for the early identification of high-risk paediatric patients.
Materials and methods
Dataset and sample source
The Paediatric Intensive Care (PIC) database is a large-scale, paediatric-specific bilingual database available in both Chinese and English.15 The PIC database contains the clinical data (eg, diagnostic, testing and monitoring data) of 12 881 paediatric patients admitted to multiple intensive care wards in the Children’s Hospital, Zhejiang University School of Medicine, between 2010 and 2018. This Children’s Hospital has more than 1900 beds and is the National Clinical Research Centre for Child Health of China.
Inclusion and exclusion criteria
The inclusion criteria were as follows: (1) age between 1 month and 18 years; (2) only information from the first hospitalisation if a patient was hospitalised multiple times and (3) children with sepsis: suspected or confirmed infection complicated by systemic inflammatory response syndrome (SIRS) within 48 hours of admission to the PICU.16 The definition of SIRS is given in online supplemental table 1. The exclusion criteria were as follows: (1) patients admitted to the neonatal intensive care unit (this subset of patients are included in the PIC database, but they are outside the scope of this study); (2) patients without chart event data, which include routine vital signs, demographic data, input and output, and others; (3) patients without data on body temperature for the full 24 hours after PICU admission. The process of study subject screening is shown in figure 1.
Supplemental material
Flow diagram of patient recruitment. NICU, neonatal intensive care unit; ICU, intensive care unit.
Data extraction, missing value processing and grouping methods
The extracted demographic data included age and sex, and initial laboratory test data within 24 hours of PICU admission were collected. Comorbidities were diagnosed within 48 hours of PICU admission (see online supplemental table 1 for definitions of comorbidities). The main outcome measure was in-hospital mortality, which was defined as death from any cause during hospitalisation. The secondary outcome measure was the length of hospital stay.
For the variables extracted in this study, the proportion of missing values was less than 10%. For normally distributed variables, the mean was imputed for missing values; for variables with a skewed distribution, the median was imputed for missing values.
Measured ear temperature was taken as the body temperature. Since body temperature can fluctuate within a range under the influence of environment, medications, disease state, etc, the average value within 24 hours of admission to the PICU was used. A roughly ‘U’-shaped relationship between admission body temperature and in-hospital mortality was detected using the locally weighted regression scatterplot smoothing (LOWESS) method, with the lowest point of the curve at approx. 36.5°C–37.5°C (figure 2). Therefore, we divided all patients based on their admission body temperature level into the following groups: the hypothermia group (<36.5°C), the normothermia group (36.5°C–37.5°C) and the hyperthermia group (>37.5°C).
Relationship between temperature on the first day and inpatient mortality in critically ill children with sepsis using locally weighted scatterplot smoothing analysis.
Statistical analysis
Measurement data conforming to a normal distribution are expressed as mean±SD, and the differences between two groups were compared by Student’s t-test. Measurement data with a skewed distribution are expressed as median (IQR), and differences between two groups were compared by the Mann-Whitney U test. Count data are expressed as the number (percentage), and differences between two groups were compared by the χ2 test.
The LOWESS method was used to detect the approximate relationship between body temperature on the first day and in-hospital mortality. The variables affecting in-hospital mortality that were significant (p<0.20) in the univariate analyses were included in multivariate logistic regression analysis. Multicollinearity was tested using the variance inflation factor (VIF) method, and a VIF ≥5 was considered as severe multicollinearity. The goodness-of-fit test was used to test the model fitting. Covariate adjustment was performed using the extended model approach: model 1=body temperature on the first day; model 2=model 1+demographic data (age and sex); model 3=model 2+laboratory tests (peripheral white cell count, platelets, serum sodium, serum potassium, serum ionised calcium, serum glucose, arterial blood pH and lactate); model 4=model 3+complications (anaemia, hypertension, malignancy, diabetic ketoacidosis, liver dysfunction and pneumonia). According to the normal reference ranges, the laboratory results were converted into binary variables (i.e., normal laboratory results and abnormal laboratory results). The relative importance of each covariate in the multivariate logistic regression model was evaluated by subtracting the degree of freedom of the covariate from the Wald χ2 value. A larger difference corresponded to a more important covariate in the logistic regression model.
P<0.05 was considered statistically significant. Statistical analysis was performed using R V.4.2.1, STATA V.16 and SPSS V.24.
Patient and public involvement
There were no patients involved in the current study. Additionally, participants and members of the public were not involved in the design, conduct, reporting or dissemination plans of the current research.
Results
General information of the enrolled paediatric patients
Table 1 lists the baseline characteristics of paediatric patients with sepsis. A total of 1144 patients with sepsis were enrolled, including 601 men (52.5%), with a median age of 34 months. Among the comorbidities of the paediatric patients with sepsis, 796 cases of anaemia (69.6%), 350 cases of hypertension (30.6%), 85 cases of liver dysfunction (7.4%), 51 cases of malignancy (4.4%), 49 cases of diabetic ketoacidosis (4.3%) and 97 cases of pneumonia (8.5%) were noted. The median hospital stay among the paediatric patients with sepsis was 12 days, and 37 in-hospital deaths occurred.
Baseline characteristics
Comparisons of clinical data between paediatric patients with sepsis with different temperatures on admission
The clinical data of paediatric patients in the normothermia group were compared with those in the hypothermia group and the hyperthermia group. As shown in table 1, age (20 (6–55) vs 39 (17–85), p<0.001), systolic blood pressure (101 (84–113) vs 110 (100–122), p<0.001), diastolic blood pressure (60 (47–72) vs 66 (58–76), p<0.001), white cell count (10.0 (6.2–14.4) vs 11.6 (7.2–16.5), p=0.015) and platelet count (247 (171–320) vs 264 (193–354), p=0.008) were lower in the hypothermia group than in the normothermia group; the proportion of men (56.4% vs 52.1%, p=0.040), respiratory rate (28 (25–32) vs 26 (23–30), p=0.001) and proportion of diabetic ketoacidosis (5.1% vs 2.8%) were higher in the hypothermia group than in the normothermia group. Age (26 (15–65) vs 39 (17–85), p=0.003) and diastolic blood pressure (65 (55–74) vs 66 (58–76), p=0.030) were lower in the hyperthermia group than in the normothermia group; heart rate (134 (118–150) vs 122 (102–142), p<0.001), respiratory rate (30 (25–38) vs 26 (23–30), p<0.001), proportion of liver dysfunction (8.5% vs 6.0%, p=0.003) and proportion of pneumonia (16.2% vs 5.7%, p<0.001) were higher in the hyperthermia group than in the normothermia group.
In terms of clinical outcome indicators, no statistically significant differences in the length of hospital stay were identified between the normothermia group and the hypothermia group or between the normothermia group and the hyperthermia group. The in-hospital mortality rate in the hypothermia group was 7.5 times higher than that in the normothermia group (9.0% vs 1.2%, p<0.001), and the in-hospital mortality rate in the hyperthermia group was 5.0 times higher than that in the normothermia group (6.0% vs 1.2%, p<0.001).
Approximate relationship between body temperature on the first day and in-hospital mortality
The LOWESS method was used to detect the approximate relationship between average temperature on the first day and in-hospital mortality among paediatric patients with sepsis. As shown in figure 2, a roughly ‘U’-shaped relationship was observed between admission body temperature and in-hospital mortality. The in-hospital mortality rate was lowest when body temperature was in the range of 36.5°C–37.5°C. When body temperature on the first day was higher than 39.0°C, the in-hospital mortality rate was 23%. When body temperature on the first day was less than 34.0°C, the in-hospital mortality rate was 50%.
Multivariate logistic regression to further analyse the relationship between body temperature on the first day and in-hospital mortality
As shown in table 2, in the progressively expanded multivariate logistic regression model, with the normothermia group as the reference, hypothermia was significantly correlated with increased in-hospital mortality (model 1: OR 8.15, 95% CI 3.46 to 19.19; model 2: OR 6.95, 95% CI 2.92 to 16.55; model 3: OR 6.65, 95% CI 2.74 to 16.15; and model 4: OR 6.23, 95% CI 2.26 to 15.19). Moreover, hyperthermia was significantly correlated with increased in-hospital mortality (model 1: OR 5.24, 95% CI 2.24 to 12.26; model 2: OR 4.86, 95% CI 2.07 to 11.44; model 3: OR 4.48, 95% CI 1.88 to 10.65; and model 4: OR 4.15, 95% CI 1.73 to 9.95).
ORs for hospital mortality in septic children with three levels of average temperature on the first day
To further explore the relationship between abnormal body temperature and in-hospital mortality, we divided the hypothermia group into the mild hypothermia subgroup (36.0°C ≤body temperature <36.5°C) and severe hypothermia subgroup (<36.0°C) and the hyperthermia group into the mild hyperthermia subgroup (37.5°C <body temperature ≤38.0°C) and severe hyperthermia subgroup (>38.0°C). Mild hypothermia, severe hypothermia, mild hyperthermia and severe hyperthermia were all independent risk factors for in-hospital mortality in different models (table 3). The ORs of all the covariates in model 4 are shown in table 4. When the covariates were adjusted, age (OR 0.99, 95% CI 0.98 to 1.00), female sex (OR 0.45, 95% CI 0.21 to 0.96), hyperlactatemia (OR 2.89, 95% CI 1.35 to 6.20), severe hypothermia (OR 14.72, 95% CI 4.84 to 44.75), mild hypothermia (OR 3.71, 95% CI 1.26 to 10.90), mild hyperthermia (OR 3.41, 95% CI 1.17 to 9.90) and severe hyperthermia (OR 5.15, 95% CI 1.84 to 14.43) were independently associated with in-hospital mortality.
ORs for hospital mortality in septic children with five levels of average temperature on the first day
Univariate and multivariable logistic regression results of model 4
The statistical results of the Wald χ2 values minus df for each variable are shown in figure 3 and online supplemental table 2. The results showed that body temperature on the first day was the most important covariate for predicting in-hospital mortality (the value was 20.2), followed by hyperlactatemia, age and sex.
Supplemental material
The statistical results of the Wald χ2 values minus the df for each variable.
Discussion
Sepsis is associated with a high mortality rate in hospitalised paediatric patients. Risk stratification of sepsis patients is important for timely intervention for high-risk patients and is of great significance to improve prognosis. The main purpose of this study was to evaluate the effect of body temperature on the first day on the prognosis of PICU patients with sepsis. The LOWESS method showed a roughly ‘U’-shaped relationship between admission body temperature and in-hospital mortality. The multivariate logistic regression analysis showed that temperature on the first day was a significant influencing factor for in-hospital mortality. Whether hypothermic or hyperthermic, the more abnormal the temperature on the first day is, the higher the risk of in-hospital death in children with sepsis.
Paediatric patients have high incidence rates of infectious diseases. Clinically, changes in body temperature are often used to identify infection at an early stage to estimate the degree of infection and the effect of anti-infection treatment. However, studies on the correlation between body temperature and prognosis of paediatric patients with sepsis are relatively lacking. Whether fever or hypothermia is an adaptive response or a harmful response caused by maladaptation remains unresolved.17
This study found that hypothermia on admission was an independent risk factor for in-hospital death. Ahmad et al evaluated the relationship between temperature on the first day and mortality in 374 neonates with sepsis and found that the mortality rate in the hypothermia group was 5.4 times higher than that in the normothermia group (33.0% vs 6.1%) and 2.8 times higher than that in the fever group (33.0% vs 11.6%).18 Similarly, in two studies in adults with sepsis, compared with patients in the other groups, patients in the hypothermia group had lower levels of inflammatory cytokines19 and higher in-hospital mortality.20 The findings of this study are consistent with the results of the aforementioned studies. The possible reason is that under hypothermic conditions, the body remains in a low-motility state with low cytokine levels, and immunosuppression is closely related to prognosis.21 Yehya et al performed a secondary analysis on a prospective cohort of 191 paediatric patients with severe sepsis and classified the paediatric patients with sepsis based on the body temperature trajectory in the first 72 hours.22 In contrast with the above results, although the in-hospital mortality rate in paediatric patients in the hypothermia group (17%) was higher than that in patients in the other subtype groups (3%–11%), the difference was not statistically significant (p=0.260), which may be due to a lack of statistical power since the sample size was small.
Regarding the effect of fever on the prognosis of patients with sepsis, different studies have reported controversial results. In studies in adult patients, high body temperature may be unfavourable or beneficial, and some studies have found no correlation between fever and mortality.23–25 Unlike in adults, high fever is more likely to cause cardiopulmonary function or nervous system damage or failure in paediatric patients. As a normal immune response by the body after infection, fever helps inhibit bacterial growth, promote the synthesis of antibodies and cytokines, and activate T cells, neutrophils, and macrophages, thereby facilitating infection control.26 However, damage to cell components, local tissues and organs, and even the whole body, can also occur during fever. Children with sepsis are often immunocompromised and have underdeveloped thermoregulatory centres. Therefore, the adverse effects of fever on the body may outweigh the benefits that fever confers, thereby increasing the risk of mortality in patients.27
This study demonstrated the relationship between temperature on the first day and the short-term prognosis of critically ill children with sepsis. This study has certain strengths. First, since body temperature can fluctuate within a range affected by the environment, medications, disease state, etc, we included the average body temperature within 24 hours of PICU admission rather than the initial value. The average value is a more realistic reflection of patient’s initial temperature status. Second, we found a ‘U’-shaped relationship between body temperature on the first day and mortality in paediatric patients with sepsis rather than a simple linear relationship. Third, the sample size of this study was relatively large, giving it good statistical power.
This study also has certain limitations. First, it was a retrospective study limited to one centre and may be affected by selection bias. Second, limited by the database, we could not obtain data on some relatively important variables, such as drug dose and PRISM III score, which may affect the final results to some extent. Third, we could not know the cause of the abnormal body temperature or the measures taken to control it. It is necessary to verify the importance of these findings in future prospective studies.
Conclusion
Sepsis is a disease of general concern to PICU doctors. This study showed that body temperature on the first day was a promising independent prognostic marker for paediatric patients with sepsis and that hypothermia and hyperthermia were both significantly correlated with a poor prognosis. Although further studies are needed to determine the exact mechanism of the relationship between body temperature on the first day and mortality, this study provides a basis for considering risk stratification of paediatric patients with sepsis based on body temperature on the first day. Whether physical or pharmaceutical correction of body temperature can improve the prognosis of paediatric patients with sepsis warrants further study.
Data availability statement
Data are available upon reasonable request. The full data set used in this study is available from the first author at wanghb53@mail2.sysu.edu.cn. However, reanalysis of the full data for other use requires approval by the PIC Institute.
Ethics statements
Patient consent for publication
Ethics approval
This study involves human participants and this project has been approved by the Institutional Review Committee of the Children’s Hospital of Zhejiang University School of Medicine (ethics reference number: 2019-IRB-052). Because this study was a retrospective study and did not affect clinical decision-making, the need for informed consent from patients was waived.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Footnotes
HW, YC and MX contributed equally.
Contributors HW, YC, MX, ZL, BS and LG conceived and designed the study; HW extracted the data; HW, TH, LH and RZ analysed the data; HW, YC and MX drafted the manuscript; ZL, BS and LG critically reviewed and revised the manuscript; LG is responsible for the overall content as the guarantor; all authors approved for submission.
Funding This research was supported by Shandong Province Key Project of TCM Science and Technology (grant number Z-2022081), Key Research and Development Plan in Jining City (grant number 2020JKNS006), Post-doctoral Innovative Talent Support Program of Shandong Province (grant number SDBX2022020), Post-doctoral Programme of Affiliated Hospital of Jining Medical University (grant number JYFY321211) and Jining Medical University Research Fund for Academician Lin He New Medicine (grant number JYHL2018FZD08).
Competing interests None declared.
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.