Predicting active pulmonary tuberculosis using an artificial neural network.

BACKGROUND: Nosocomial outbreaks of tuberculosis (TB) have been attributed to unrecognized pulmonary TB. Accurate assessment in identifying index cases of active TB is essential in preventing transmission of the disease. OBJECTIVES: To develop an artificial neural network using clinical and radiographic information to predict active pulmonary TB at the time of presentation at a health-care facility that is superior to physicians' opinion. DESIGN: Nonconcurrent prospective study. SETTING: University-affiliated hospital. PARTICIPANTS: A derivation group of 563 isolation episodes and a validation group of 119 isolation episodes. INTERVENTIONS: A general regression neural network (GRNN) was used to develop the predictive model. MEASUREMENTS: Predictive accuracy of the neural network compared with clinicians' assessment. RESULTS: Predictive accuracy was assessed by the c-index, which is equivalent to the area under the receiver operating characteristic curve. The GRNN significantly outperformed the physicians' prediction, with calculated c-indices (+/- SEM) of 0.947 +/- 0.028 and 0.61 +/- 0.045, respectively (p < 0.001). When the GRNN was applied to the validation group, the corresponding c-indices were 0. 923 +/- 0.056 and 0.716 +/- 0.095, respectively. CONCLUSION: An artificial neural network can identify patients with active pulmonary TB more accurately than physicians' clinical assessment.

Impact of nosocomial infection on length of stay and functional improvement among patients admitted to an acute rehabilitation unit.

OBJECTIVE: To identify factors predictive of length of stay (LOS) and the level of functional improvement achieved among patients admitted to an acute rehabilitation unit for the first time, with special reference to the role of nosocomial infection. SETTING: A 40-bed acute rehabilitation unit within a 300-bed, tertiary-care, public, university-affiliated hospital. STUDY POPULATION: All patients admitted to the unit between January 1997 and July 1998. DESIGN: Prospective cohort study in which demographic and clinical data, including occurrence of nosocomial infection, were collected during the entire unit admission of each patient. Multivariate linear regression analysis was used to identify factors predictive of unit LOS or improvement in functional status as measured by the change in the Functional Independence Measure (FIM) score between admission and discharge (deltaFIM). RESULTS: There were 423 admissions to the rehabilitation unit during the study period, of which 91 (21.5%) had spinal cord injury (SCI) as a principal diagnosis. One hundred seven nosocomial infections occurred during 84 (19.9%) of the 423 admissions. The most common infections were urinary tract (31.8% of all infections), surgical-site (18.5%), and Clostridium difficile diarrhea (15%). Only one patient died of infection. After controlling for severity of illness on admission, functional status on admission, age, and other clinical factors, the significant positive predictors of unit LOS were as follows: SCI (P<.001), pressure ulcer (.002), and nosocomial infection (<.001). Significant negative predictors of deltaFIM were age (P<.001), FIM score on admission (<.001), prior hospital LOS (.002), and nosocomial infection (.007). CONCLUSIONS: Several variables were identified as contributing to a longer LOS or to a smaller improvement in functional status among patients admitted for the first time to an acute rehabilitation unit. Of these variables, only nosocomial infection has the potential for modification. Studies of new approaches to prevent infections among patients undergoing acute rehabilitation should be pursued.

Antibiotic-resistant organisms among long-term care facility residents on admission to an inpatient geriatrics unit: Retrospective and prospective surveillance.

BACKGROUND: There is limited information about antibiotic-resistant organisms in community long-term care facilities (LTCFs). The objective of this study was to obtain data on resistant organisms in residents from community LTCFs admitted to an inpatient acute geriatrics service (AGS). METHODS: Two studies were performed. In the first study, bacteriology records of all admissions to the AGS for the period from November 1, 1998, through June 30, 2000, were reviewed for resistant organisms (methicillin-resistant Staphylococcus aureus [MRSA], vancomycin-resistant enterococci [VRE], and resistant gram-negative bacilli). In the second study, residents admitted to the AGS during a 2-month period (N = 92 admissions) had surveillance cultures (nares, gastrostomy site, wounds, and urine) for resistant organisms done within 72 hours of admission. RESULTS: In the retrospective study, there were 727 admissions, of which 437 (60%) had 928 cultures within 72 hours of admission; 590 (64%) cultures grew 1 or more pathogens. Urine (65%) and blood (26%) cultures accounted for 91% of all cultures done. Rates of resistance by culture site were as follows: urine (resistant organism in 16.6% of 373 cultures), blood (6.7% of 60 cultures), wound (52% of 23 cultures), and sputum (40% of 20 cultures). MRSA and enterococci with high-level gentamicin resistance were the most common resistant organisms identified. No VRE were isolated; only 3% of 421 gram-negative isolates were considered resistant strains compared with 19% (P <.001) of gram-positive isolates. In the prospective study, 17% of 92 residents were found to have a resistant organism in 1 or more surveillance cultures; the most common resistant organisms were MRSA and high-level gentamicin-resistant enterococci. Only 1 resident was found to have VRE in a rectal swab culture; resistant gram-negative bacilli also were uncommon. CONCLUSIONS: Among residents of community LTCFs admitted to an AGS, resistant organisms were identified infrequently (<20% of admissions). MRSA was the most common resistant organism; VRE and resistant gram-negative bacilli were rare. These findings vary from other studies suggesting that there may be geographic variation in the epidemiology of resistant organisms among residents of community LTCFs.

Prospective surveillance for antibiotic-resistant organisms in patients with spinal cord injury admitted to an acute rehabilitation unit.

BACKGROUND: The objective of this study was to evaluate the epidemiology of antibiotic-resistant organisms among patients with spinal cord injury admitted to an acute rehabilitation unit for the first time. METHODS: After informed consent, 63 patients with spinal cord injury admitted to an acute rehabilitation unit between January 1997 and July 1998 had surveillance cultures (nares, urine, wounds or ulcers, and perineum) done on admission and every 2 weeks thereafter until discharge or as long as 6 weeks after admission. RESULTS: Of the 4 surveillance sites, perineal cultures most commonly grew one or more potential pathogens; however, antibiotic-resistant organisms were most often isolated from wounds or ulcers and least often in urine cultures. Staphylococcus aureus (methicillin-sensitive plus methicillin-resistant) and enterococci represented 44% of all organisms isolated in surveillance cultures. Methicillin-resistant S aureus was the most common resistant organism isolated. Less than 30% of the gram-negative bacilli isolated were considered antibiotic-resistant strains. Nosocomial infection as a result of any resistant organism was infrequent. After adjusting for various confounding factors in a logistic regression model, only the presence of a pressure ulcer predicted carriage of any resistant organism on admission to the rehabilitation unit. Acquisition of a resistant organism after admission to the unit at one or more surveillance sites occurred in 8 (22%) of 36 patients not colonized on admission. CONCLUSIONS: Methicillin-resistant S aureus was the most common resistant organism colonizing patients with spinal cord injury at the time of admission to an acute rehabilitation unit and throughout their stay. However, acquisition of any resistant organism after admission was uncommon on this unit, which used Standard Precautions in the routine care of patients. These findings have implications for the type of isolation procedures on acute rehabilitation units. The low rate of acquisition and infection with MRSA after admission argues against attempts at eradication as a method of control.

The effects of oxygen and hyperinflation on arterial oxygen tension after endotracheal suctioning.

The effects of hyperinflation and increasing the FIO2 to 1.0 on PaO2 after endotracheal suctioning were examined in 28 patients after cardiac surgery. The changes in PaO2 immediately after and at 5- and 10-minutes after suctioning were compared with control values. Hyperinflation at the patient's baseline FIO2 produced a mean decrease in PaO2 of 18 mm Hg immediately after suctioning in over 75% of the sample. Preoxygenation with an increase in the ventilator FIO2 to 1.0 produced a mean increase in PaO2 of 48 mm Hg in 75% of the sample. Delivering preoxygenation by hyperinflation provided greater protection for all patients with a mean increase in PaO2 of 100 mm Hg. No statistically significant differences existed between the 5- and 10-minute PaO2 value and the control PaO2 values for any of the procedures. Experimental hyperinflation was aborted in two patients who experienced significant hypotension in association with the hyperinflation. The hypotension was transient, with arterial blood pressure immediately returning to baseline values on termination of the experimental hyperinflation. In conclusion, 100% oxygen administered by hyperinflation for 1 minute before and after endotracheal suctioning protects patients after cardiac surgery from a fall in PaO2. Because hyperinflation can cause alterations in arterial blood pressure and heart rate, these vital signs should be visually monitored during the endotracheal suctioning procedure.

Minimum discard volume from arterial catheters to obtain coagulation studies free of heparin effect.

With the use of the canine experimental model, the accuracy and reliability of coagulation test results were compared between arterial samples and samples obtained by venipuncture. The age of the catheter, the concentration of the heparinized flush solution, and the minimum discard volume were evaluated to obtain coagulation study results free of heparin effect. Thirty-seven paired samples of PT, aPTT, and TT were obtained. Analyses revealed no significant difference in test results between new lines and 7-day-old lines, nor between heparin concentrations of 1 unit/ml, 2 units/ml, or 4 units/ml. A minimum discard volume of five times the dead space (measured from catheter tip to sampling proximal stopcock) resulted in accurate and reliable PT, aPTT, and TT test results from the arterial catheters.

Epidemiology of nosocomial infection and resistant organisms in patients admitted for the first time to an acute rehabilitation unit.

The objectives of this study were to define the epidemiology of nosocomial bacterial colonization and infection and to define predictors of nosocomial infection among a cohort (n=423) of admissions to an acute rehabilitation unit. Overall, methicillin-resistant Staphylococcus aureus (MRSA) and enterococci were the most commonly identified colonizing organisms. Escherichia coli and Pseudomonas aeruginosa were the most commonly identified colonizing gram-negative bacilli. During 70 (16.5%) of the 423 hospitalizations in the unit, 94 nosocomial infections occurred. The most common infections were those of the urinary tract (30% of 94 infections) or a surgical site (17%), Clostridium difficile diarrhea (15%), and bloodstream infection (12.8%). Antibiotic-resistant bacteria most commonly caused bloodstream infection (41.7%) and surgical site infection (56.3%). Independent predictors of nosocomial infection at the time of admission were functional status (measured with the functional independence measure), APACHE III score, and spinal cord injury. In conclusion, gram-positive organisms were the predominant strains causing nosocomial colonization and infection. The logistic model, if verified, may be useful in defining patients who should be targeted for measures to prevent nosocomial infection.