Risk factors for intracranial infection after craniotomy: A case–control study

Abstract Background Intracranial infection, serving as a severe postoperative infection after craniotomy, poses significant problems for patients' outcomes. Objective To explore risk factors for intracranial infection after craniotomy. Methods A total of 2,174 patients who underwent craniotomy from 1 May 2018 to 30 June 2019 were retrospectively studied. Finally, 196 patients with intracranial infections were classified as case group, and 392 patients randomly selected from patients without intracranial infection were classified as control group. Demographic, clinical, laboratory, microbiological, and antimicrobial data were systemically recorded. The characteristics, pre‐ and postoperative variables, and other variables were evaluated as risk factors for intracranial infection by univariate analysis and binary logistic regression model. Results There was no significant difference in terms of demographics between two groups, except for gender, hypertension, length of stay (LOS), intraoperative blood loss, tumor, and trauma surgery. The independent risk factors were male, age ≤45, hypertension, tumor surgery, surgery in autumn (compared with spring), surgical duration ≥4 hr, intraoperative blood loss ≥400 ml, and postoperative oral infection, coma, and serum RBC > normal value. Trauma surgery (p < .001, OR = 0.05, 95% CI: 0.017–0.144) was an independent protective factor (p < .05, OR < 1) for intracranial infection. All 196 patients in the case group submitted specimens for cerebrospinal fluid (CSF) cultures, and 70 (35.71%) patients had positive results. Gram‐positive pathogens predominated (59 cases, 84.28%). Staphylococcus were the most common causative pathogens, and fully resistant to aztreonam, cefazolin, and benzylpenicillin, but not resistant to linezolid and minocycline. Conclusion Identifying the risk factors, pathogens, and pathogens' antibiotic resistance for intracranial infection after craniotomy plays an important role in the prognosis of patients.


| INTRODUC TI ON
Craniotomy, as a neurosurgical procedure, has been performed more than a century and is characterized by performing within the intracranial space (Adaaquah, Gates, & Van Gompel, 2018;Gonzalez-Darder, 2016).
White blood cell count or neutrophil count is still indicators of infection and clinical judgment; however, their specificity might not be high (Folyovich et al., 2014;Westendorp, Nederkoorn, Vermeij, Dijkgraaf, & van de Beek, 2011). Surgical drainage and antibiotics are also the effective therapy to treat intracranial infection (Dashti et al., 2008;Kural et al., 2019). However, many drugs are facing difficulty to penetrate into brain due to the blood-brain barrier (BBB) (Daneman & Prat, 2015). Meanwhile, cerebrospinal fluid (CSF) only can be drained a little through lumbar puncture. Therefore, it is of great significance to explore risk factors for intracranial infection after craniotomy in order to timely prevent it.
Currently, more than 10 high-risk factors have been reported, which were mainly for surgical site infection after craniotomy (Adaaquah et al., 2018;Fang, Zhu, Zhang, Xia, & Sun, 2017;Kourbeti et al., 2007). However, the risk factors of season and postoperative oral infection were not analyzed. In the 1982s, the seasonal variation in arterial blood pressure was reported (Brennan, Greenberg, Miall, & Thompson, 1982). In addition, Herweh et al have concluded that hypertensive intracerebral hemorrhage was associated with the increased air pressure by conducting a worldwide cohort (Herweh et al., 2017).
As regards the oral infection, one previous study has reported that brain abscesses might be the potential deadly complications of odontogenic infections through summarizing related cases (Moazzam, Rajagopal, Sedghizadeh, Zada, & Habibian, 2015). However, there was no case-control study performed to verify the relationship between oral infection and intracranial infection.
In our study, we calculated the prevalence of intracranial infection after craniotomy and further identified the risk factors for intracranial infection, especially some new risk factors such as surgical season and postoperative oral infection. Furthermore, microorganisms and its antibiotic resistance isolated from cerebrospinal fluid (CSF) were also determined.

| Participants
The enrolled patients were admitted from Department of (f) were treated by nonsurgical treatment, such as intravenous antibiotics; and (g) had adhesion of subarachnoid space to cerebrospinal fluid circulation disorder. tracranial infections were classified as case group. The data collection was difficult due to the poor condition of our hospital, and 392 cases in the control group, twice as many as in the case group, were randomly selected from the remaining 1,973 patients without intracranial infection. Case and control groups chose the same consecutive patients who underwent craniotomy. The flow diagram is shown in Figure 1. The prevalence of nosocomial intracranial infection was defined to be happened after 48 hr of admission to control the prevalence-incidence bias. In addition, the blind method was used to control the investigation bias, all patients or clinicians did not know the information to distinguish the case group from the control group.

| Definitions
Intracranial infection was diagnosed according to the definitions of Centers for Disease Control (CDC) (Horan et al., 2008). The prevalence of intracranial infection was defined to be happened: (a) after 48 hr of admission; (b) during operation or pathological examination; (c) from organisms cultured from brain tissue or dura; (d) considering at least the following two signs such as dizziness, fever (>38°C), headache, local neurosis, consciousness-changing, or confusion; (e) and from (a) microorganisms identified from brain tissue, abscess tissue, blood, or urine; (b) diagnostic single-antibody titer (IgM) or fourfold increase in paired sera (IgG) for pathogen; and (c) radio- Cerebrospinal fluid leak was defined as any leak of the fluid that surrounds the brain and spinal cord and escapes from the cavities within the brain or central canal in the spinal cord (Abuabara, 2007), and was reported to be associated with the development of meningitis (Jones & Becker, 2001).

| Data collection
The demographic, clinical, laboratory, microbiological, and antimicrobial data were systemically analyzed by same team, laboratory and healthcare department. The clinical data were further analyzed by reviewing electronic medical records and files. The other data were collected from three electronic surveillance systems: Nosocomial infection records were from the Ongoing Nosocomial Infection Surveillance of Xinglin; clinical data were from electronic medical records; and microbiological and antimicrobial data were from microbial systems. The normal values of postoperative indexes of ALB, hsCRP, RBC, and serum HGB were 40-55 g/L, 0-6 mg/L, 4.3-5.8 × 10 12 L (men) and 3.8-5.10 × 10 12 L (women), and 130-175 g/L (men) and 115-150 g/L (women), respectively. In addition, gender, age, hypertension, hyperlipidemia, diabetes mellitus, trauma, and tumor were also included.

| Determination of the cut-point of quantitative variables
As shown in Table 1, the average age of the targeted population was 48.87 ± 16.21 years. The age of 45 years was selected as cutpoint. The average surgical duration and preoperative LOS were 4.62 ± 2.03 hr and 7.44 ± 6.56 days, 4 hr and 7 days were selected as F I G U R E 1 Flow diagram of patient selection cut-points, respectively. The ASA score ranges from 1 to 5 (Saklad, 1941). The mean ASA score in the case and control group was 2.49 ± 0.89 and 2.51 ± 0.85, respectively, a score of 2 was selected as cut-points. Due to intraoperative blood loss, at least 400 ml was considered as a major bleeding in clinical diagnosis, 400 ml was selected as cut-points.

| The clinical diagnostic routine of microbiology methods
If patients in the case group were with the symptoms of headaches, fever, nausea, and vomiting, 3-5 ml CSF was collected from brain or abscess tissue by needle aspiration or biopsy during surgical operation or autopsy, and stored in sterile test tubes for detecting the infection within 1 hr. The microorganisms were cultured in the Vitek compact autokinetic microbe culture instrument (bioMerieux).
The quality control strains were Escherichia coli ATCC 25922,   with p < .05 were significant. EpiData 3.1 was used to input these data by two graduate students. We have another staff to verify these data.

| Univariate analysis of potential risk factors
The potential risk factors are shown in     Note: Normal value (postoperative RBC): the total number of normal men and women in the case group and control group, the normal value was 4.3-5.8 × 10 12 L (men) and 3.8-5.10 × 10 12 L (women); normal value (postoperative HGB): the total number of normal men and women in the case group and the control group, the normal value was 130-175 g/L (men) and 115-150 g/L (women).
TA B L E 2 (Continued) (Shi et al., 2017). In the present study, 2,174 patients who underwent craniotomy were enrolled; among them, 196 patients were infected with intracranial infection during hospitalization with a rate of 9.02%. The rate was similar to the previous study, which reported that the incidence of intracranial infection after craniotomy was from 1.4% to 9.5% (Shi et al., 2017). However, the infection rate was significantly higher than other reports, which might be due to the patient's severe condition with a variety of underlying diseases. Although antibiotics and surgical drainage can be used to treat intracranial infection, the BBB provides an obstacle for drug delivery to central nervous system, and CSF only can be drained a little through lumbar puncture. Therefore, it is important to determine the risk factors for intracranial infection after craniotomy. In this study, the results indicated gender, age, CSF leakage, ASA score, surgical duration, and others were the risk factors, which have been reported in previous studies (Fang et al., 2017;Kourbeti et al., 2007Kourbeti et al., , 2015Lin, Zhao, & Sun, 2015), while season and postoperative oral infection were firstly taken into account in this study.

Previously, Adaaquah et al have reported that season is related
to the variation of arterial blood pressure (Adaaquah et al., 2018).
In this study, Staphylococci isolated from infected CSF sample were identified as the major pathogen. Staphylococci, as the main infectious bacteria of intracranial infection, could widely distribute on human skin, especially in spring and summer, due to the moister and hotter weather (Kourbeti et al., 2015;Zhan et al., 2014). Furthermore, patients who underwent surgery in autumn were more likely to infect with intracranial infection than those who underwent surgery in spring (OR = 2.866, 95% CI: 1.592-5.159). Therefore, in this study, season was firstly explored, and it might be an independent risk factor for intracranial infections, which might recommend that patients after craniotomy should live in a constant temperature and humidity ward. The exact mechanism of seasonal factors for intracranial infection is still further research.
In this study, postoperative oral infection was also identified as an independent risk factor for intracranial infection after craniot-

omy. Moazzam et al reported bacteria could adversely disseminate
from oral infection to intracranial infection (Moazzam et al., 2015).
Previous studies have reported that 1,200 different types of microbes were isolated from the human mouth (Corson, Postlethwaite, & Seymour, 2001;Dewhirst et al., 2010). Microorganism in the mouth could enter into the cranial vault via blood, venous drainage, inoculation, or lymphatic drainage (Moazzam et al., 2015). Furthermore, Andersen and Carpenter et al showed the central nervous system (CNS) infection correlated with oral infection, as 32%-60% of brain abscesses have been shown to be polymicrobial (Andersen & Horton, 1990;Carpenter, Stapleton, & Holliman, 2007).
Furthermore, other risk factors for intracranial infections were also explored, such as gender, age, and surgical duration, but there was no consensus on why these factors were risk factors for intracranial infections (Kourbeti et al., 2015;Lin et al., 2015). This study showed men were 1.775 times likely to have intracranial infections than women after craniotomy, which might be because men were more likely to smoke and drink alcohol than women. Farrokhi et al have reported smoking could increase the risk of infection during deep brain stimulation surgery (Farrokhi et al., 2019). More young people choose neurosurgery than the elderly, due to the limitations of income, referral system, and religious beliefs for the elderly (Cassir et al., 2015;Inoue et al., 2015). In this study, the ultimately enrolled people were relatively young (48.87 ± 16.21 years), randomly selected to perform craniotomy, which will produce Berkson's bias. Thus, patients in multiple hospitals were needed to reduce Berkson's bias in the future study. Besides, the sample size of our study was still smaller than the multicenter ones and the confidence interval was narrow. In our future study, we will collect larger sample size to confirm the study.

TA B L E 3 Binary logistic regression analysis for intracranial infections after craniotomy
In conclusion, it was of great significance to explore risk factors for intracranial infection after craniotomy in order to timely prevent it. In this study, surgical season and postoperative oral infection were firstly determined as the new risk factors for intracranial infection. Meanwhile, microorganisms and its antibiotic resistance isolated from intracranial infections were determined. A better understanding of the targeted risk factor, microorganisms, and its antibiotic resistance of intracranial infection after craniotomy could encourage us to adopt more favorable prevention, treatment, and strategies, which contribute to the prognosis of patients.

ACK N OWLED G M ENT
None.

CO N FLI C T O F I NTE R E S T
All authors have no conflicts of interests to disclose.

E TH I C A L A PPROVA L
This study has been approved by Research Ethics Committee of the Second Hospital of Hebei Medical University (2018-R084).

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.