We used hospital data to demonstrate RT-NISS functions. On 17 June 2019 the ICD identified suspicious NI outbreaks using the RT-NISS in a tertiary general teaching hospital with 1,900 beds and initiated an epidemiologic investigation. The significance of RTNISS in preventing nosocomial outbreaks was assessed in the current study.
Bacterial colonization refers to the isolation of bacteria from a patient without symptoms of infection and is a prerequisite for active infection. An infection is defined as the presence of bacteria that cause pathologic changes, albeit to different degrees (Correa and Fortaleza 2019; Arzilli et al. 2022). As defined by the Chinese Ministry of Public Health in 2009, hospital outbreaks require at least three infectious cases with the same pathogenic agent in the same hospital department within seven days (The Ministry of Public Health 2009). The RT-NISS (the Xinglin Hospital Infection Real-time Monitoring System, V12.0,
In addition, real-time surveillance data were collected to guide clinical practice, and an interactive platform was developed to establish communication between the clinical staff and the ICD to improve case reporting, patient isolation, and intervention measures. Therefore, with basic and epidemiologic information, the ICD staff and clinicians work together to determine the source of NIs. The RT-NISS was used to extract clinical data from the HIS and LIS, including central venous catheters and antibacterial agents, pathogenic bacteria, and underreporting rates, which were analyzed by the ICD staff to confirm suspiciousness cases and guide clinical practice. A previous study showed that the specificity and sensitivity for diagnosing NIs in patients by the RT-NISS were 93.0% and 98.8%, respectively (Du et al. 2014).
The surveillance path of the RT-NISS was as follows: NI-related information was achieved from the HIS, hospital LIS, and radiology information system by data access middleware, including ventilator use, fevers, central venous catheterization, urinary tract catheterization, routine blood test results, routine urine test results, other laboratory test results (e.g., procalcitonin), bacteria-positive cultures, multidrug-resistant (MDR) bacteria, surgery, antibacterial agents, and medical advice regarding isolation. In this way, the prognosis for patients with infections was generated to provide warning for the clinical medical staff and the hospital infection management staff, and detect occult hospital infection outbreaks in a timely fashion. The NI algorithm screening proceeded to send out outbreak and NI case alerts. The Chinese Ministry of Public Health defined hospital outbreaks in 2009 as at least three infection cases with the same pathogenic agent in the same hospital department within seven days (The Ministry of Public Health 2009). Algorithms were based on statistical process control (SPC). The alert threshold not only defined elementary threshold, which was according to the management standards promulgated by the Ministry of Public Health, but the alert was triggered when cases exceeded a threshold of ≥ 3 infections in two weeks on the same ward. The alert threshold was also based on past statistical variations of case frequency.
The SPC offers the possibility to monitor different types of statistical parameters, such as the incidence count or rate, cumulative sums, or moving average. An outbreak database was set up in response to an outbreak alert, and clinical intervention was implemented to control the outbreak. In the absence of outbreak alerts, the surveillance ended. Assessments by infection control personnel (ICP) proceeded following NI case alerts. The surveillance ended if the ICP judges deemed no issues with the alerts. An NI database was set up if the ICP judges considered the alerts abnormal. Another path for NI case alerts or complicated NIs was discussing with physicians. The physician’s platform differentiated the NI cases and entered the NI cases into a NI database. If the physician confirmed that the case was not an NI, the surveillance was ended. After the database was established, a hospital-wide NI analysis was performed, and NI-targeted surveillance data were collected. The details can be reviewed on the flowchart of NI cases and outbreak pre-warning and confirmation in the Du et al. (2014) study.
When it was shown that the same type of bacteria appeared in a department for a short period and > 3 nosocomial infection cases were confirmed, it was suspected that NI outbreaks might occur, and clinical intervention should be arranged. The interaction platform constructed through the hospital infection real-time monitoring system was applied to realize the real-time report of infection cases, precise diagnosis, intervention, and feedback. In so doing, the special staff and clinicians involved in the diagnosis and control of infections work together and are aware of an occult hospital infection outbreak in a timely fashion.
This study was conducted in our medical center, which is in the middle of China. Our hospital has 1,900 beds and five ICUs. The neurosurgical ICU has 17 beds with 450 patients per year. Large general hospitals must prevent infectious clusters. The current study results were intended to serve as a reference for other general hospitals. From 11–17 June 2019, a total of 42 patients were admitted to the neurosurgery ICU and included in the study. The infection rate was counted. When the outbreak alerts were sent out, 32 of the 42 patients had been discharged. The ten hospitalized patients were alerted of a suspicious outbreak. The clinical intervention was initiated to prevent the outbreak. The use of antibiotics was reviewed and adjusted. The distribution of the bacteria in patients was analyzed.
The Chinese Ministry of Public Health defined hospital outbreaks in 2009 as at least three infection cases with the same pathogenic agent in the same hospital department within seven days. After the clinical intervention, five of the ten patients with suspicious NI outbreak alerts were clinically diagnosed with a suspicious NI according to the management standards promulgated by the Ministry of Public Health. AB was detected in the sputum samples collected from these patients. According to the diagnostic criteria of nosocomial infection, the other five patients were excluded in the absence of an outbreak. These five patients did not meet the diagnostic criteria, such as admission within 48 h, and no bacterial growth occurred in the cultures (The Ministry of Public Health 2009).
The Institutional Ethical Committee approved the project of our hospital (2020-Ethics Approval-No. 12). The study was proceeded following the ethical standards of the Helsinki Declaration, as revised in 2013. The patients were informed about the study’s purpose and procedures and provided written consent forms.
Bacterial antibiotic susceptibility analysis was performed using the minimum inhibitory concentration (MIC) method (Andrews 2001), and the results were judged according to the CLSI (2010) standard of the American Committee for Clinical Laboratory Standardization. Strains were tested for susceptibility to ampicillin/sulbactam, ceftazidime, cefepime, imipenem, gentamicin, tobramycin, ciprofloxacin, and sulfamethoxazole.
Thirty-six environmental samples were collected from staff hands, humidified liquid, curtains, stethoscope, water bottle stopper, and water bottle shell, injection pump, faucet, bed cup and water, water feeding syringe, the railing of bed, lift the patient board. Cotton swabs were densely coated on blood plates, and the blood plates were placed in an air environment at 37°C and cultivated for 48 h, after which the strains were identified by a mass spectrometer (Repizo et al. 2017). Environmental samples were kept, collected, and tested every day within one week after the hospital infection management workers intervened in the investigation.
Statistical analyses were performed using SPSS software (version 20.0) on real-time surveillance data.
Based on RT-NISS warnings, the five patients diagnosed with suspected AB infections included four males and one female, with a median age of 62 years (46–84 years). Most patients were older and had received a combination of invasive surgery and antibiotics. The baseline characteristics of the five patients are shown in Table I. A monitoring procedure (Fig. 1) was initiated to identify the source, and path of the NIs and the RT-NISS and clinical intervention were used to confirm suspected cases.
Baseline characteristics of the study population.
Case 1 | Case 2 | Case 3 | Case 4 | Case 5 | |
---|---|---|---|---|---|
Age (y) | 84 | 46 | 62 | 65 | 57 |
Gender | male | male | male | female | male |
Bed code | J01 | 23 | 18 | 22 | J07 |
Diagnosis | traumatic brain injury | spontaneous intracerebral hemorrhage | spontaneous intracerebral hemorrhage | traumatic brain injury | thalamic hemorrhage |
Date of ICU admission | 9 June | 3 June | 1 June | 31 May | 14 May |
Analyzed sample | sputum | sputum | sputum | sputum | sputum |
Date of sample analysis | June 13 | June 17 | June 13 | June 11 | June 17 |
Pathogen | AB (+++) | AB (+++) | AB (+++) | AB (++) | AB (++) KP (+++) |
Date of tracheostomy | June 18 | June 6 | no | May 31 | May 17 |
Duration of invasive mechanical ventilation | 3 | 30 | N/A | 8 | 23 |
Antibiotic combination therapy | yes | yes | yes | yes | yes |
NI or colonization | NI | colonization | colonization | NI | colonization |
AB –
This figure is a part of the visual time-series chart of inpatients shown by the software in use. Temporary clinical data shown were collected using a real-time nosocomial infection monitoring system. The data are displayed in different colors to facilitate comparison between patients. The symptoms represented in the system are displayed in Chinese and each symptom was translated into English, which was connected to the corresponding box with arrows to facilitate understanding. “More details” mean the detailed clinical data of the patients. The content of this interface only included the operation of patients with a suspected infection during hospitalization. The numbers 1–42 represent the duration of the hospitalization in days. A patient with an early warning appears as a red mark of multi-drug resistance bacteria on the warning interface. The possibility of a nosocomial infection outbreak was ruled out for multiple (≥ 3) patients, which was marked in red for multidrug resistance bacteria at any point in time. Because the warning interface contains a lot of patient information, the warning interface was not displayed completely.
Based on the HIS data, the number of NIs showed that the alert rate for suspicious infections or outbreaks from 11–17 June 2019 was 11.9% (5/42), which remained constant with the rates in 2018 and 2019 (Table II). The outbreak alert rate was similar to other times throughout the year; thus, the RT-NISS could access the data quickly. Most patients with infections had severe injuries, including craniocerebral injuries and lung contusions, required intensive care, and increased the risk of cross-infections or secondary infections. Therefore, clinical examination confirmed suspicious cases to avoid underreporting and erroneous reports.
Suspected alert rates for infections or outbreaks in different periods.
Periods | Number of suspected infections | Number of ICU patients | Rate of suspected infections | χ2 | |
---|---|---|---|---|---|
11–17 June 2019 | 5 | 42 | 11.9% | 0.267 | 0.875 |
20 May – 20 June 2018 | 6 | 38 | 15.8% | ||
20 April – 20 May 2019 | 6 | 41 | 14.6% |
The five patients were clinically diagnosed and confirmed as below: one patient was confirmed to have a NI by the RT-NISS; one patient was suspected of having a NI by the RT-NISS and confirmed to be a NI by clinical interventions, and there were three patients with bacterial colonization originating in the hospital. The data excluded the possibility of an outbreak. The results of drug susceptibility testing are shown in Table III. The drug susceptibility test results suggested that the same pathogenic strain might be involved in the infections, potentially increasing the risk of an outbreak (Table III).
Drug susceptibility of multidrug-resistant bacteria isolated from the patients.
Antibiotics | Case 1 | Case 2 | Case 3 | Case 4 | Case 5 |
---|---|---|---|---|---|
Ampicillin/sulbactam | R | R | R | R | R |
Ceftazidime | R | R | R | R | R |
Ceftriaxone | R | R | R | R | R |
Cefepime | R | R | R | R | R |
Imipenem | R | R | R | R | R |
Gentamicin | R | R | R | R | R |
Tobramycin | S | R | R | S | S |
Ciprofloxacin | R | R | R | R | R |
Levofloxacin | R | R | R | R | R |
Sulfamethoxazole | S | R | R | S | R |
R – resistant, S – susceptible
An 84-year-old man was admitted on June 9, 2019 with a craniocerebral injury and lung contusion. The patient was treated with amoxicillin and levofloxacin for four days, then cefoperazone-sulbactam from June 13 until discharge from the hospital. He had an elevated white blood cell (WBC) count on June 13, and increased interleukin-6 and procalcitonin (PCT) levels on June 14. On June 13 XDR-AB was detected in a sputum sample, and rales were present bilaterally. Invasive surgery was performed, including chest drainage, a tracheotomy, and ventilator support. On June 13 he was reported to have a hospital infection in the lower respiratory tract.
A 46-year-old man was admitted on June 3, 2019 with a spontaneous cerebral hemorrhage. He was treated with amoxicillin from 3–15 June, piperacillin-sulbactam from 15–20 June, and ceftazidime from 20–30 June. On June 8 the patient was reported to have a lower respiratory tract infection associated with
A 62-year-old man was admitted on June 1, 2019 with a spontaneous cerebral hemorrhage. He was diagnosed with a lung infection upon admission and was given piperacillin-sulbactam and combination therapy (piperacillin-sulbactam and levofloxacin) on June 6. The testing of bacterial culture upon admission was
A 65-year-old woman was admitted on May 31, 2019 with a severe craniocerebral injury. Ventilator support was provided upon admission, followed by combination antibiotic treatment with amoxicillin, gentamicin, and levofloxacin.
A 57-year-old man was admitted on May 14, 2019 with a thalamic hemorrhage. On 17 May the patient was reported to have a lower respiratory tract infection with
Case analysis showed a large number of MDR-AB strains in the ICU environment. To determine the source and path of infection, the ICD screened 36 environmental samples, including surfaces, hospital staff, and ventilators, of which 20 samples (55.5%) were contaminated. MDR-AB was detected in six samples, including a stethoscope, curtains, injection pump, a water faucet, and hands from two nurse workers. No new infections were detected during the one-week environmental samples tests in the following seven days.
The early warning data on the five cases were acquired through the RT-NISS. RT-NISS monitored the infections in real-time. Outputting data on bacterial infections, suspected cross-infections, and the prevalence of outbreaks are exported from RT-NISS. The RT-NISS captured the basic demographics about the patient and the corresponding epidemiologic information, which can lead to preliminary conclusions and save the investigation time. XDR-AB pneumonia remains a significant challenge in ICUs, and few drug treatment options are available for XDR-AB pneumonia treatment (Li et al. 2017). Ma et al. (2013) reported that mortality due to MDR and XDR-AB infections in China ICUs was 29% from 2011 to 2013. Indeed, the RT-NISS has an active role in providing accurate data to ICDs, diagnosing infections, and preventing outbreaks (Ma et al. 2013).
MDR-AB is rapidly becoming a global threat due to resistance to major classes of antibiotics (Nasr 2020). MDR-AB infections often occur in healthcare settings, especially in intensive care settings.
According to the Hospital Infection Outbreak Control guideline (WS/T524-2016; The Ministry of Public Health 2009), there were no outbreaks in our hospital-based on this criterion. After intervention by the clinicians, only two cases were diagnosed with a NI, and it was not an outbreak within the hospital. The other three cases were regarded as colonization. It demonstrated that the system was used effectively to collect and analyze the clinical information of infected patients, help ICD confirm the diagnosis, actively treat the patients, and implement control measures (Du et al.2014), such as antibiotic use instructions, analyzing the distribution of the bacteria within the patient, and contemplating bed adjustment. As demonstrated, data on the nosocomial transmission of drug-resistant bacteria can be retrieved accurately and rapidly with a real-time system, and variations in bacterial susceptibility to antibiotics can be determined (Chen et al. 2018). Compared with the original manual report, which generally requires confirmation of the infection and sending a report form to the Hospital Management Department, the system report can issue an early warning in real-time within 24 h and confirmation in the system (Huang et al. 2010). These data are compared with NI control targets established by health authorities to strengthen the actions of the ICD for preventing and controlling MDR bacterial infections. The system automatically extracts data and calculates the detection and infection rates and antimicrobial drug use in each hospital department every day (Leclère et al. 2017).
The RT-NISS was used to monitor infections in real-time. At the same time, the ICD promptly extracted clinical and epidemiologic information using the RT-NISS and instituted targeted measures to prevent infections and outbreaks (Zingg et al. 2015). The ICD systematically monitors the on-site status of clinical departments to avoid physician misconduct and initiates infection prevention and control measures, including epidemiologic investigations, patient isolation, environmental sanitation sampling, training, and measures to prevent and control multi-drug resistant bacteria. Moreover, the ICD should also cooperate with other departments to identify risk factors for spreading infections, control NIs, and promote the rational use of antibiotics (Zingg et al. 2015). The drug susceptibility data of pathogenic bacteria in the hospital, the bedside monitoring results of patients infected with XDR bacteria, and the prevention and control of drug-resistant bacteria spread can be obtained from the ICD monthly (Wieland et al. 2007). Each hospital department’s pathogenic bacteria susceptibility data can be released every six months. Data collected daily and monthly by the RT-NISS on the number of NIs, use of antibacterial drugs, number of patients infected with XDR bacteria, drug susceptibility of pathogenic bacteria, and the prevalence and control of NI and NI outbreaks of drug-resistant bacteria can be shared between hospitals.
Furthermore, epidemiologic surveys and environmental sanitation sampling and identification should be carried out (Zingg et al. 2015). In our study, 55.5% of the environmental samples were contaminated, and MDR-AB was detected in five samples; however, no new infections were detected in the following seven days after the hospital infection management workers began to intervene in the investigation. This finding confirms that RT-NISS and the ICD can cooperatively monitor and prevent NIs in the hospital to some degree.
The research limitations of this study were as follows. First, the system generated false alarms; thus, an effective and reproducible framework is needed to evaluate and compare these algorithms. The computer screening algorithm did not include medical records; thus, the terminology application may not be uniform. Second, polymerase chain reaction tests of the samples were not completed due to the lack of hardware equipment in the hospital, and the classification of bacteria was achieved from the microbiology laboratory. Bacterial resistance is generally only identified using the MIC method in routine clinical work due to the costs involved. The antibiotic susceptibility test method based on the MIC values can guide the treatment and control of infections macroscopically. However, without genotyping or phenotyping tests, this is not an acceptable approach for accurately judging whether there is an infection outbreak cluster. Also, PCR detection from environmental swabs was not performed due to the cost, which might affect the apparent correlation between AB strains isolated from patients and the environment. Finally, no automatic alarm appeared when the specimens were limited, and the early warning standard was not reached.
In summary, the RT-NISS integrated with the clinical diagnoses made by physicians enhances the timeliness and effectiveness of NI control and surveillance. The universal significance should be researched and explored by multiple hospitals.