Effect of a real-time automatic nosocomial infection surveillance system on hospital-acquired infection prevention and control - BMC Infectious Diseases - BMC Infectious Diseases

Surveillance of HAIs is a fundamental and essential aspect of infection prevention and control programs [6,7,8,9,10]. Traditional surveillance is a manual time-consuming infection case-finding and reporting program that involves infection control physicians looking over lists of positive infection information of pathogen culture and patient histories daily and HAI case reports by clinicians. In this study, the results showed low infection incidences of HAIs (1.48%) and certain infection sites, including respiratory infection (0.93%), UTI (0.25%), SSI (0.09%), BSI (0.14%) and skin and soft tissue infection (0.12%), reported by traditional manual surveillance methods in period 1. These low reported infection incidences echo the inaccurate report demonstrated by other studies [9, 12]. Seifi's study presented the results of under- and overreporting of HAIs by traditional surveillance [9]. Many hospitals have a large number of beds but insufficient numbers of infection control physicians, and infection control physicians need to collect large amounts of data daily, which will result in false negative infection reports. Many cases are excessive, or unnecessary antibiotic usage will lead to false-positive infection findings. In addition, many cases are not ordered for pathogen culture to cause false negative infection reports. Moreover, the lack of knowledge regarding complicated infection cases of clinicians will lead to underreporting and mistaken reporting of infection events. Given their high workloads and need for multitasking, clinicians can make mistakes and lack sufficient time to report infection cases. All these factors make this type of manual surveillance unreliable and lack accuracy. There is no doubt that this frustrating situation would have serious consequences, including absent treatment for underreporting HAI patients, inappropriate treatment of false-positive infection patients, transmission of pathogens from HAI patients to other uninfected patients, and many HAIs and outbreaks that are not monitored and uncontrolled. Therefore, this predicament has catalysed the development and implementation of RT-NISSs.

Many studies in the past have demonstrated accurate and efficient electronic automatic surveillance systems [11, 12, 14,15,16]. Our analysis showed higher infection incidences of HAIs and certain infection sites, including respiratory infection, UTI, SSI, BSI and skin and soft tissue infection, reported by RT-NISS (both in period 2 and period 3) compared with those reported by traditional surveillance in period 1 (both p < 0.05), which is consistent with the accurate surveillance results based on RT-NISSs reported in other studies. Mingmei's study presented the sensitivity and specificity of a RT-NISS with an approximately 200-fold time-saving capacity compared with manual surveys [12]. Studies in this field are more concerned with defects by traditional manual surveillance and the virtues by RT-NISSs on the surveillance of all HAIs. However, few analyses have evaluated the use of RT-NISSs for surveillance of different types of MDRO or have only focused on one type of MDRO at a time [17, 18]. In the present study, we demonstrated a more accurate surveillance result of MDRO reports by the RT-NISS in both periods 2 and 3 compared with traditional surveillance in period 1 (Tables 2, 3). In period 1 of the present study, clinicians and infection control physicians obtained MDRO case information based on microbiology laboratory physicians' daily reporting. However, microbiology laboratory physicians underreported MDRO cases or reported cases in an untimely manner due to their high workload, that led to the high FN numbers of both kinds of MDROs. Furthermore, there weren't sufficient trainings for microbiology laboratory physicians led to erroneous judgement or mis-classification of MDROs that made the FP and FN numbers high in the period 1. We present the results of a lower number of FPs and FNs and higher Se, Sp, PPV and NPV of 5 different varieties of MDROs, including ESBL, MRSA, CRAB, CRPA and CRE, since RT-NISS implementation. Compared with period 1, the MDRO reporting process and surveillance changed, and the MDRO information and data became reliable and accurate. MDROs are the main pathogenic bacteria of SSI, UTI, BSI and pneumonia, and these types of nosocomial infections are intractable and complex and the factor of high mortality for patients. Moreover, MDROs easily lead to nosocomial transmission and nosocomial cluster infection events [19,20,21]. Thus, timely and accurate monitoring of MDROs is an important targeted monitoring project for infection prevention and control programs [12]. The results of the present study about MDROs surveilled by RT-NISS could benefit and inform targeted prevention and control programs.

Another strength of our study was the investigation of the effect of the daily implementation of RT-NISS on HAI surveillance. After RT-NISS implementation in period 2, we analysed the long-term data collected by RT-NISS and acquired comprehensive knowledge about HAI risk factors. Then, we implemented some interventions (Table 4). And for those wards didn't implement prevention and control measures for MDROs and/or inappropriately administered antibiotics and/or without standard pathogen culture, we reported monthly utilization rate of antibiotics and rate of pathogen culture in each ward to remain them, provided special education for the healthcare workers and harsh economic measures were implemented for the serious situation. The incidences of HAIs and hospital-acquired MDROs in period 3 were significantly lower than those in period 2 (0.41% vs. 0.52%, p = 0.021). These results demonstrated that the adoption of RT-NISS coupled with the implemented interventions could reduce the rate of HAIs. In addition, regarding the results of the MDROs reported by RT-NISS in period 2, we further found 3 mistakes associated with the use of the RT-NISS. 1. There were 24 other kinds of muti-drug resistance Pseudomonas species misclassified as CRPA resulted in 24 FP of CRPA. 2. 14 CRE misclassified as ESBL resulted in 14 FP of ESBL and 14 FN of CRE. 3. 13 ESBL, 19 MRSA, 7 CRAB and 5 CPPA were under-reported because of the LIS connection fails by RT-NISS. Based on those above mistakes, we checked and corrected different varieties of MDROs' rule definition of RT-NISS, we also required RT-NISS engineers regularly check the connection with other hospital information systems and timely fixed the connection fails in period 3. After modifying the rule definition and timely correcting the data input, the number of FPs and FNs of 5 different varieties of MDROs, including ESBL, MRSA, CRAB, CRPA and CRE, in period 3 was reduced to 0, and the Se, Sp, PPV and NPV of 5 different varieties of MDROs in period 3 were both improved to 100%.

Our study still has a limitation. Due to all patients' information in the period 1 (the manual surveillance phase) were collected only by manual surveillance, it is difficult to obtain all of the sufficient information of large number of patients. Thus, would the composition of the patient population or the specialty of the patients affect the individual infection types in 3 periods of the present study is not sure.

Effect of a real-time automatic nosocomial infection surveillance system on hospital-acquired infection prevention and control - BMC Infectious Diseases - BMC Infectious Diseases