Monitoring expert system performance using continuous user feedback

Medical Applications of Artificial Intelligence (Legal Aspects and Future Prospects)

Background: Cutting-edge digital technologies are being actively introduced into healthcare. The recent successful efforts of artificial intelligence in diagnosing, predicting and studying diseases, as well as in surgical assisting demonstrate its high efficiency. The AI’s ability to promptly take decisions and learn independently has motivated large corporations to focus on its development and gradual introduction into everyday life. Legal aspects of medical activities are of particular importance, yet the legal regulation of AI’s performance in healthcare is still in its infancy. The state is to a considerable extent responsible for the formation of a legal regime that would meet the needs of modern society (digital society). Objective: This study aims to determine the possible modes of AI’s functioning, to identify the participants in medical-legal relations, to define the legal personality of AI and circumscribe the scope of its competencies. Of importance is the issue of determining the grounds for imposing legal liability on persons responsible for the performance of an AI system. Results: The present study identifies the prospects for a legal assessment of AI applications in medicine. The article reviews the sources of legal regulation of AI, including the unique sources of law sanctioned by the state. Particular focus is placed on medical-legal customs and medical practices. Conclusions: The presented analysis has allowed formulating the approaches to the legal regulation of AI in healthcare.

No additional protection against ventriculitis with prolonged systemic antibiotic prophylaxis for patients treated with antibiotic-coated external ventricular drains

OBJECT External ventricular drains (EVDs) are commonly used for CSF diversion but pose a risk of ventriculitis, with rates varying in frequency from 2% to 45%. Results of studies examining the utility of prolonged systemic antibiotic therapy for the prevention of EVD-related infection have been contradictory, and no study to date has examined whether this approach confers additional benefit in preventing ventriculitis when used in conjunction with antibiotic-coated EVDs (ac-EVDs). METHODS A prospective performance analysis was conducted over 4 years to examine the impact of discontinuing systemic antibiotic prophylaxis after insertion of an ac-EVD on rates of catheter-related ventriculitis. Ventriculitis and other nosocomial infections were ascertained by a qualified infection disease nurse using definitions based on published standards from the Centers for Disease Control and Prevention, comparing the period when patients received systemic antibiotic therapy for the duration of EVD treatment (Period 1) compared with only for the peri-insertion period (Period 2). Costs were analyzed and compared across the 2 time periods. RESULTS Over the 4-year study period, 866 patients were treated with ac-EVDs for a total of 7016 catheter days. There were 8 cases of ventriculitis, for an overall incidence of 0.92%. Rates of ventriculitis did not differ significantly between Period 1 and Period 2 (1.1% vs 0.4%, p = 0.22). The rate of nosocomial infections, however, was significantly higher in Period 1 (2.0% vs 0.0% in Period 2, p = 0.026). Cost savings of $162,516 were realized in Period 2 due to decreased drug costs and savings associated with the reduction in nosocomial infections. CONCLUSIONS Prolonged systemic antibiotic therapy following placement of ac-EVDs does not seem to reduce the incidence of catheter-related ventriculitis and was associated with a higher rate of nosocomial infections and increased cost.

Peripherally Inserted Central Venous Catheter–Associated Bloodstream Infections in Hospitalized Adult Patients

Background.

Limited data on the risk of peripherally inserted central venous catheter-associated bloodstream infections (PICC BSIs) in hospitalized patients are available. In 2007, dedicated intravenous therapy nurses were no longer available to place difficult peripheral intravenous catheters or provide PICC care Barnes-Jewish Hospital.

Objectives.

To determine the hospital-wide incidence of PICC BSIs and to assess the effect of discontinuing intravenous therapy service on PICC use and PICC BSI rates.

Setting.

A 1,252-bed tertiary care teaching hospital.

Methods.

A 31-month retrospective cohort study was performed. PICC BSIs were defined using National Healthcare Safety Network criteria.

Results.

In total, 163 PICC BSIs were identified (3.13 BSIs per 1,000 catheter-days). PICC use was higher in intensive care units (ICUs) than non-ICU areas (PICC utilization ratio, 0.109 vs 0.059 catheter-days per patient-day for ICU vs non-ICU; rate ratio [RR], 1.84 [95% confidence interval {CI}, 1.78-1.91]). PICC BSI rates were higher in ICUs (4.79 vs 2.79 episodes per 1,000 catheter-days; RR, 1.7 [95% CI, 1.10-2.61]). PICC use increased hospital-wide after the intravenous therapy service was discontinued (0.049 vs 0.097 catheter-days per patient-day; P = .01), but PICC BSI rates did not change (2.68 vs 3.63 episodes per 1,000 catheter-days; P = .06). Of PICC BSIs, 73% occurred in non-ICU patients.

Conclusions.

PICC use and PICC BSI rates were higher in ICUs; however, most of the PICC BSIs occurred in non-ICU areas. Reduction in intravenous therapy services was associated with increased PICC use across the hospital, but PICC BSI rates did not increase.

Patient- and Device-Specific Risk Factors for Peripherally Inserted Central Venous Catheter—Related Bloodstream Infections

Objective.

To determine the patient- and device-specific risk factors for hospital-acquired peripherally inserted central venous catheter–related bloodstream infections (PICC BSIs) in adult patients.

Design.

Nested case-control study.

Setting.

Barnes-Jewish Hospital, a 1,252-bed tertiary care teaching hospital.

Patients.

Adult patients with PICCs placed from January 1, 2006, through July 31, 2008.

Methods.

PICC BSI cases were identified using the National Healthcare Safety Network definition. Uninfected control patients with PICCs in place were randomly selected at a 3: 1 ratio. Patient- and device-related variables were examined using multivariate analysis.

Results.

The overall PICC BSI rate was 3.13 per 1,000 catheter-days. Independent risk factors for PICC BSIs included congestive heart failure (odds ratio [OR], 2.0 [95% confidence interval (CI), 1.26–3.17];P= .003), intra-abdominal perforation (OR, 5.66 [95% CI, 1.76–18.19];P= .004),Clostidium difficileinfection (OR, 2.25 [95% CI, 1.17–4.33];P= .02), recent chemotherapy (OR, 3.36 [95% CI, 1.15–9.78];P= .03), presence of tracheostomy (OR, 5.88 [95% CI, 2.99–11.55];P< .001), and type of catheter (OR for double lumen, 1.89 [95% CI, 1.15–3.10];P= .01; OR for triple lumen, 2.87 [95% CI, 1.39–5.92];P= .004). Underlying chronic obstructive pulmonary disease (OR, 0.48 [95% CI, 0.29–0.78];P= .03) and admission to surgical (OR, 0.43 [95% CI, 0.24–0.79];P= .006) or oncology and orthopedic (OR, 0.35 [95% CI, 0.13–0.99];P= .05) services were less likely to be associated with having a PICC BSI.

Conclusions.

We identified several novel factors related to PICC BSIs. These factors may inform preventive measures.

Developing information technology for infection prevention surveillance

The potential to automate at least part of the surveillance process for health care-associated infections was seen as soon as hospitals began to implement computer systems. Progress toward automated surveillance has been ongoing for the last several decades. But as more information becomes available electronically in the healthcare setting, the promise of electronic surveillance for healthcare-associated infections has become closer to reality. Although true fully automated surveillance is not here yet, significant progress is being made at a number of centers for electronic surveillance of central catheter-associated bloodstream infections, ventilator-associated pneumonia, and other healthcare-associated infections. We review the progress that has been made in this area and issues that need to be addressed as surveillance systems are implemented, as well as promising areas for future development.

Surveillance for vancomycin-resistant enterococci: type, rates, costs, and implications

OBJECTIVE

To evaluate 2 active surveillance strategies for detection of enteric vancomycin-resistant enterococci (VRE) in an intensive care unit (ICU).

DESIGN

Thirty-month prospective observational study.

SETTING

ICU at a university-affiliated referral center.

PATIENTS

All patients with an ICU stay of 24 hours or more were eligible for the study.

INTERVENTION

Clinical active surveillance (CAS), involving culture of a rectal swab specimen for detection of VRE, was performed on admission, weekly while the patient was in the ICU, and at discharge. Laboratory-based active surveillance (LAS), involving culture of a stool specimen for detection of VRE, was performed on stool samples submitted for Clostridium difficile toxin detection.

RESULTS

Enteric colonization with VRE was detected in 309 (17%) of 1,872 patients. The CAS method initially detected 280 (91%) of the 309 patients colonized with VRE, compared with 25 patients (8%) detected by LAS; colonization in 4 patients (1%) was initially detected by analysis of other clinical specimens. Most patients with colonization (76%) would have gone undetected by LAS alone, whereas use of the CAS method exclusively would have missed only 3 patients (1%) who were colonized. CAS cost Dollars 1,913 per month, or Dollars 57,395 for the 30-month study period. Cost savings of CAS from preventing cases of VRE colonization and bacteremia were estimated to range from Dollars 56,258 to Dollars 303,334 per month.

CONCLUSIONS

A patient-based CAS strategy for detection of enteric colonization with VRE was superior to LAS. In this high-risk setting, CAS appeared to be the most efficient and cost-effective surveillance method. The modest costs of CAS were offset by the averted costs associated with the prevention of VRE colonization and bacteremia.

Prevention of hospital infection

The prevention of nosocomial infections is an important aspect of patient care, particularly in high-risk areas such as intensive care units (ICUs). Local hospital leadership needs to develop easily defined infection-control policies that are evidence-based. These infection-control policies also require the presence of a dedicated group of infection-control practitioners to provide education, collect surveillance data, and oversee the implementation of the local infection-control plan.

To gown or not to gown: the effect on acquisition of vancomycin-resistant enterococci

Infection-control recommendations include the use of gowns and gloves to prevent horizontal transmission of vancomycin-resistant enterococci (VRE). This study sought to determine whether the use of a gown and gloves gives greater protection than glove use alone against VRE transmission in a medical intensive care unit (MICU). From 1 July 1997 through 30 June 1998 and from 1 July 1999 through 31 December 1999, health care personnel and visitors were required to don gloves and gowns upon entry into rooms where there were patients infected with nosocomial pathogens. From 1 July 1998 through 30 June 1999, only gloves were required under these same circumstances. During the gown period, 59 patients acquired VRE (9.1 cases per 1000 MICU-days), and 73 patients acquired VRE during the no-gown period (19.6 cases per 1000 MICU-days; P<.01). The adjusted risk estimate indicated that gowns were protective in reducing VRE acquisition in an MICU with high VRE colonization pressure.

Clinical decision-making using computers: opportunities and limitations

This paper deals with the use of information technology in order to make clinical decisions in dentistry. Since there is a large number of available systems and ongoing projects using computers, we first identify and describe major streams, as well as singular specialties. We underline the need for standards for data format and exchange in addition to the ability of different systems to communicate with each other. We conclude with a five-step recommendation to transform a dental office into digital and how to implement a computer-aided clinical decision support system, which provides action-oriented support in order to reduce the likelihood of human error during dental treatment.

Acquisition of vancomycin-resistant enterococci during scheduled antimicrobial rotation in an intensive care unit

Scheduled rotation of treatment of gram-negative antimicrobial agents has been associated with reduction of serious gram-negative infections. The impact of this practice on other nosocomial infections has not been assessed. The purpose of this study was to determine if scheduled antimicrobial rotation reduced rates of acquisition of enteric vancomycin-resistant enterococci (VRE) among 740 patients admitted to an intensive care unit (ICU). The preferred gram-negative agent was ceftazidime during rotation 1 and ciprofloxacin during rotation 2. Unadjusted VRE acquisition rates were 8.5 cases per 1000 ICU days and 11.7 cases per 1000 ICU days during rotations 1 and 2, respectively (P<.01). However, scheduled antimicrobial rotation of ceftazidime with ciprofloxacin had no effect on the risk of acquiring VRE in the ICU after adjustment for known risk factors. Independent predictors of acquisition of VRE were enteral feedings, higher colonization pressure, and increased duration of anaerobic therapy. Our findings can confirm no additional beneficial or adverse effect on VRE acquisition among ICU patients as a result of this practice.

Environmental control to reduce transmission of Clostridium difficile

Restrictive antibiotic policies and infection control measures have been shown to reduce the incidence of Clostridium difficile-associated diarrhea (CDAD) among hospitalized patients. To date, the role of environmental disinfectants in reducing nosocomial CDAD rates has not been well studied. In a before-and-after intervention study, patients in 3 units were evaluated to determine if unbuffered 1:10 hypochlorite solution is effective as an environmental disinfectant in reducing the incidence of CDAD. Among 4252 patients, the incidence rate of CDAD for bone marrow transplant patients decreased significantly, from 8.6 to 3.3 cases per 1000 patient-days (hazard ratio, 0.37; 95% confidence interval, 0.19-0.74), after the environmental disinfectant was switched from quaternary ammonium to 1:10 hypochlorite solution in the rooms of patients with CDAD. Reverting later to quaternary ammonium solution increased the CDAD rate to 8.1 cases per 1000 patient-days. No reduction in CDAD rates was seen among neurosurgical intensive care unit and general medicine patients, for whom baseline rates were 3.0 and 1.3 cases per 1000 patient-days, respectively. Unbuffered 1:10 hypochlorite solution is effective in decreasing patients' risk of developing CDAD in areas where CDAD is highly endemic. Presumed mechanisms include reducing the environmental burden and the potential for C. difficile transmission among susceptible patients.

Enteric carriage of vancomycin-resistant Enterococcus faecium in patients tested for Clostridium difficile

OBJECTIVE

To identify independent risk factors for enteric carriage of vancomycin-resistant Enterococcus faecium (VREF) in hospitalized patients tested for Clostridium difficile toxin.

DESIGN

Retrospective case-cohort study.

SETTING

Tertiary-care teaching hospital.

PATIENTS

Convenience sample of 215 adult inpatients who had stool tested for C difficile between January 29 and February 25, 1996.

RESULTS

41 (19%) of 215 patients had enteric carriage of VREE Five independent risk factors for enteric VREF were identified: history of prior C difficile (odds ratio [OR], 15.21; 95% confidence interval [CI95], 3.30-70.10; P < .001), parenteral treatment with vancomycin for > or = 5 days (OR, 4.06; CI95, 1.54-10.73; P = .005), treatment with antimicrobials effective against gram-negative organisms (OR, 3.44; CI95, 1.20-9.87; P = .021), admission from another institution (OR, 2.95; CI95, 1.21-7.18; P =.017), and age > 60 years (OR 2.57; CI95, 1.13-5.82; P = .024). These risk factors for enteric VREF were independent of the patient's current C difficile status.

CONCLUSIONS

Antimicrobial exposures are the most important modifiable independent risk factors for enteric carriage of VREF in hospitalized patients tested for C difficile.

A randomized trial of povidone-iodine compared with iodine tincture for venipuncture site disinfection: effects on rates of blood culture contamination

BACKGROUND

Contamination of blood cultures creates problems in their interpretation and unneeded resource utilization. Because skin flora comprise the major group of contaminant species, more effective skin disinfection at the venipuncture site could reduce contamination.

SUBJECTS AND METHODS

We performed a randomized trial in adult inpatients at a tertiary care teaching hospital. Antecubital venipuncture sites were randomly disinfected with povidone-iodine or iodine tincture, and blood cultures (two bottles, 10 mL of blood) were drawn by professional phlebotomists. Scoring of contaminant species was restricted to skin flora. Hospital resource utilization was compared among patients with contaminated blood cultures and those with sterile blood cultures.

RESULTS

Of the 3,851 blood cultures collected during the study, 120 (3.1%) were contaminated with skin flora. The contamination rate for blood cultures collected after povidone-iodine was 3.8% (74 of 1,947), compared with a rate of 2.4% (46 of 1,904, P = 0.01) after iodine tincture. The difference in mean total hospital costs for patients with contaminated blood cultures and those with sterile blood cultures was $4,100 (95% confidence interval: $740 to $7,400, P = 0.02).

CONCLUSIONS

Iodine tincture is superior to povidone-iodine for venipuncture site antisepsis before blood culture sampling. Because of the high costs associated with contaminated blood cultures, hospitals should consider switching from povidone-iodine to iodine tincture. Reduction of the contamination rate may improve the quality of patient care and reduce hospital costs.

Statistical process control methods for expert system performance monitoring

The literature on the performance evaluation of medical expert system is extensive, yet most of the techniques used in the early stages of system development are inappropriate for deployed expert systems. Because extensive clinical and informatics expertise and resources are required to perform evaluations, efficient yet effective methods of monitoring performance during the long-term maintenance phase of the expert system life cycle must be devised. Statistical process control techniques provide a well-established methodology that can be used to define policies and procedures for continuous, concurrent performance evaluation. Although the field of statistical process control has been developed for monitoring industrial processes, its tools, techniques, and theory are easily transferred to the evaluation of expert systems. Statistical process tools provide convenient visual methods and heuristic guidelines for detecting meaningful changes in expert system performance. The underlying statistical theory provides estimates of the detection capabilities of alternative evaluation strategies. This paper describes a set of statistical process control tools that can be used to monitor the performance of a number of deployed medical expert systems. It describes how p-charts are used in practice to monitor the GermWatcher expert system. The case volume and error rate of GermWatcher are then used to demonstrate how different inspection strategies would perform.