Association between rating of respiratory distress and vital signs, severity of illness, intubation, and mortality in acutely ill subjects.

BACKGROUND: When deciding whether mechanical ventilation is indicated, physicians integrate their findings on physical examination in a gestalt known as respiratory distress. Despite its importance, this gestalt is poorly understood. This study aims to describe the association between the rating of the severity of respiratory distress and vital signs, severity of illness, use of mechanical ventilation, and death. A prospective observational study with 1,134 consecutive subjects with uncertain triage evaluated by a critical care consult team was carried out in a public inner city teaching hospital. METHODS: After the initial evaluation of each patient, a critical care physician rated the level of respiratory distress. We recorded vital signs, diagnosis, and laboratory results and calculated the Acute Physiology and Chronic Health Evaluation (APACHE) II score. We recorded if mechanical ventilation was initiated by 72 h and if the subject died during the hospitalization. RESULTS: The most common diagnoses were respiratory illnesses. Higher distress levels were associated with higher breathing frequency (20, 22, 27, and 30 breaths/min, P < .001) and heart rate (96, 101, 109, and 116 beats/min, P < .001) and lower S(pO2) (97, 95, 93, and 92%, P < .001). These variables explain only a small portion of the variance of distress. Distress correlated weakly with the APACHE II score (r = 0.22, P = .001). Blood pressure, temperature, Glasgow coma scale score, and laboratory data were unrelated to the levels of distress. However, higher levels of distress correlated with intubation rates (5, 13, 27, and 41%, P < .001). The area under the receiver operating characteristic curve for respiratory distress predicting intubation (0.72) was larger than that for breathing frequency (0.65). Distress was an independent predictor of intubation but not of death. CONCLUSIONS: A physician's rating of respiratory distress is independently predictive of intubation in 72 h. Vital signs explain only a small proportion of variance in distress; the other observations contributing to a physician's rating of distress must be determined.

Diagnostic error in medicine: analysis of 583 physician-reported errors.

BACKGROUND: Missed or delayed diagnoses are a common but understudied area in patient safety research. To better understand the types, causes, and prevention of such errors, we surveyed clinicians to solicit perceived cases of missed and delayed diagnoses. METHODS: A 6-item written survey was administered at 20 grand rounds presentations across the United States and by mail at 2 collaborating institutions. Respondents were asked to report 3 cases of diagnostic errors and to describe their perceived causes, seriousness, and frequency. RESULTS: A total of 669 cases were reported by 310 clinicians from 22 institutions. After cases without diagnostic errors or lacking sufficient details were excluded, 583 remained. Of these, 162 errors (28%) were rated as major, 241 (41%) as moderate, and 180 (31%) as minor or insignificant. The most common missed or delayed diagnoses were pulmonary embolism (26 cases [4.5% of total]), drug reactions or overdose (26 cases [4.5%]), lung cancer (23 cases [3.9%]), colorectal cancer (19 cases [3.3%]), acute coronary syndrome (18 cases [3.1%]), breast cancer (18 cases [3.1%]), and stroke (15 cases [2.6%]). Errors occurred most frequently in the testing phase (failure to order, report, and follow-up laboratory results) (44%), followed by clinician assessment errors (failure to consider and overweighing competing diagnosis) (32%), history taking (10%), physical examination (10%), and referral or consultation errors and delays (3%). CONCLUSIONS: Physicians readily recalled multiple cases of diagnostic errors and were willing to share their experiences. Using a new taxonomy tool and aggregating cases by diagnosis and error type revealed patterns of diagnostic failures that suggested areas for improvement. Systematic solicitation and analysis of such errors can identify potential preventive strategies.

Effect of education on hand hygiene beliefs and practices: a 5-year program.

To evaluate infection control and hand hygiene understanding at 3 public hospitals, we surveyed 4,345 healthcare workers (HCWs) 3 times during a 5-year infection control intervention. The preference for the use of alcohol hand rub for hand hygiene increased dramatically; in nurses, it increased from 14% to 34%; in physicians, 4.3% to 51%; and in allied HCWs, 12% to 44%. Study year, infection control interactive education-session attendance, infection control knowledge, and being a physician or allied HCW independently predicted a preference for alcohol hand rub.

Antimicrobial consumption data from pharmacy and nursing records: how good are they?

OBJECTIVE: To determine whether randomly selected intravenous (IV) antimicrobial doses dispensed from an inpatient pharmacy were administered. DESIGN: This was a prospective, cross-sectional study in which dose administration was confirmed by direct observation and by assessment of the medication administration record (MAR). A retrospective analysis of the return rate of unused IV antimicrobial doses was performed subsequently. SETTING: Medical and surgical intensive care units (ICUs) and non-ICUs of a 550-bed urban public teaching hospital. PARTICIPANTS: Hospitalized patients with an order in the pharmacy database for an IV antimicrobial during 9 non-consecutive weekdays in June 1999. RESULTS: Of 397 doses, 221 (55.7%) assessed by bedside observation and 238 (59.9%) assessed by MAR review were classified as administered; 139 doses (35.0%) were dispensed but changes in the drug order or the patient's status prevented their administration. In the subsequent assessment, of 745 IV antimicrobial doses dispensed during 24 hours, 322 (43.2%) were returned to the pharmacy unused; 423 (56.8%) of the doses-consistent with our prior observations-were presumably administered. CONCLUSIONS: Because computerized pharmacy data may overestimate actual antimicrobial consumption, such data should be validated when used in studies of hospital antimicrobial use. Dispense-return analysis offers a simple validation method.

Missed hypothyroidism diagnosis uncovered by linking laboratory and pharmacy data.

BACKGROUND: Although diagnostic errors are important, they have received less attention than medication errors. Timely follow-up of abnormal laboratory test results represents a critical aspect of the diagnostic process, and failures at this step are a cause of delayed or missed diagnosis, resulting in suboptimal clinical outcomes and malpractice litigation. We linked laboratory and pharmacy databases to (1) explore the potential for linking laboratory and pharmacy databases to uncover diagnostic errors, and (2) determine the frequency of failed follow-up of elevated levels of thyroid-stimulating hormone (TSH). METHODS: We downloaded TSH test results for 2 consecutive years from a laboratory database and linked this database with a pharmacy database to screen for patients with TSH levels of 20 mU/mL or higher who were not receiving levothyroxine. Patients with elevated TSH levels lacking prescriptions were followed up by telephone and record review. RESULTS: During the 2-year period, 982 (2.7%) of 36 760 unique patients tested for TSH level had elevated TSH levels. Of these patients, 177 (18.0%) had no recorded levothyroxine prescriptions. We attempted to contact 177 patients with high TSH levels who were not taking thyroid medications and reached 123 (69.5%). Of the 123 patients we were able to reach, 12 in 2000 and 11 in 2001 were unaware of their abnormal test results or a diagnosis of hypothyroidism, representing 2.3% of 982 patients with elevated TSH levels. We were unable to reach another 54 patients (5.5% of the total number of patients with elevated TSH levels) by either telephone or mail. CONCLUSIONS: By linking laboratory and pharmacy databases, we uncovered patients who did not undergo follow-up for abnormal TSH results. Conservatively, there was no follow-up for abnormal TSH results in more than 2% of patients, and another 5% of patients were lost to follow-up and possibly unaware of their results. Uncovering patients with missed diagnosis illustrates a potential use of linking laboratory and pharmacy databases to identify vulnerabilities in the care system and improve patient safety.

Computer algorithms to detect bloodstream infections.

We compared manual and computer-assisted bloodstream infection surveillance for adult inpatients at two hospitals. We identified hospital-acquired, primary, central-venous catheter (CVC)-associated bloodstream infections by using five methods: retrospective, manual record review by investigators; prospective, manual review by infection control professionals; positive blood culture plus manual CVC determination; computer algorithms; and computer algorithms and manual CVC determination. We calculated sensitivity, specificity, predictive values, plus the kappa statistic (kappa) between investigator review and other methods, and we correlated infection rates for seven units. The kappa value was 0.37 for infection control review, 0.48 for positive blood culture plus manual CVC determination, 0.49 for computer algorithm, and 0.73 for computer algorithm plus manual CVC determination. Unit-specific infection rates, per 1,000 patient days, were 1.0-12.5 by investigator review and 1.4-10.2 by computer algorithm (correlation r = 0.91, p = 0.004). Automated bloodstream infection surveillance with electronic data is an accurate alternative to surveillance with manually collected data.

Unnecessary use of central venous catheters: the need to look outside the intensive care unit.

We developed criteria for justifiable CVC use and evaluated CVC use in a public hospital. Unjustified CVC-days were more common for non-ICU patients compared with ICU patients. Also, insertion-site dressings were less likely to be intact on non-ICU patients. Interventions to reduce CVC-associated bloodstream infections should include non-ICU patients.

Antibiotic combinations with redundant antimicrobial spectra: clinical epidemiology and pilot intervention of computer-assisted surveillance.

Redundant antibiotic combinations are a potentially remediable source of antibiotic overuse. At a public teaching hospital, we determined the incidence, cost, and indications for such combinations and measured the effects of a pharmacist-based intervention. Of 1189 inpatients receiving >or=2 antibiotics, computer-assisted screening identified 192 patients (16.1%) receiving potentially redundant combinations. Chart reviews showed that 137 episodes (71%) were inappropriate. Physician overprescribing errors were found in 77 episodes (56%); most involved redundant coverage for gram-positive or anaerobic organisms. In 76 episodes (55%), lapses in the medication ordering and distribution system led to the persistence in the pharmacy records of regimens no longer active according to the patient charts. The incidence of redundant antibiotic combinations was significantly higher in the intensive care unit and surgery services, compared with medical services. Interventions to discontinue redundant agents were successful in 134 (98%) of the 137 episodes. Potential drug cost savings and reduction in redundant antibiotic combination days were 10,800 dollars and 584 days, respectively; pharmacist time for patient review and intervention cost 2880 dollars. Use of redundant antibiotic combinations was common, and a pharmacist-based intervention was feasible, with a potential annualized cost savings of 48,000 dollars.

Development of a clinical data warehouse for hospital infection control.

Existing data stored in a hospital's transactional servers have enormous potential to improve performance measurement and health care quality. Accessing, organizing, and using these data to support research and quality improvement projects are evolving challenges for hospital systems. The authors report development of a clinical data warehouse that they created by importing data from the information systems of three affiliated public hospitals. They describe their methodology; difficulties encountered; responses from administrators, computer specialists, and clinicians; and the steps taken to capture and store patient-level data. The authors provide examples of their use of the clinical data warehouse to monitor antimicrobial resistance, to measure antimicrobial use, to detect hospital-acquired bloodstream infections, to measure the cost of infections, and to detect antimicrobial prescribing errors. In addition, they estimate the amount of time and money saved and the increased precision achieved through the practical application of the data warehouse.

Electronic interpretation of chest radiograph reports to detect central venous catheters.

OBJECTIVE: To evaluate whether a natural language processing system, SymText, was comparable to human interpretation of chest radiograph reports for identifying the mention of a central venous catheter (CVC), and whether use of SymText could detect patients who had a CVC. DESIGN: To identify patients who had a CVC, we performed two surveys of hospitalized patients. Then, we obtained available reports from 104 patients who had a CVC during one of two cross-sectional surveys (ie, case-patients) and 104 randomly selected patients who did not have a CVC (ie, control-patients). SETTING: A 600-bed public teaching hospital. RESULTS: Chest radiograph reports were available from 124 of the 208 participants. Compared with human interpretation, SymText had a sensitivity of 95.8% and a specificity of 98.7%. The use of SymText to identify case- and control-patients resulted in a sensitivity of 43% and a specificity of 98%. Successful application of SymText varied significantly by venous insertion site (eg, a sensitivity of 78% for subclavian and a sensitivity of 3.7% for femoral). Twenty-six percent of the case-patients had a femoral CVC. CONCLUSIONS: Compared with human interpretation, SymText performed well in interpreting whether a report mentioned a CVC. In patient populations with less frequent CVC placement in femoral veins, the sensitivity for CVC detection likely would be higher. Applying a natural language processing system to chest radiograph reports may be a useful adjunct to other data sources to automate detection of patients who had a CVC.