Interinstitutional database for comparison of performance in lung fine-needle aspiration cytology. A College of American Pathologists Q-Probe Study of 5264 cases with histologic correlation.

In 1990, the College of American Pathologists Q-Probes Quality Assurance Program studied performance in fine-needle aspiration (FNA) of pulmonary lesions derived by retrospective analyses of cases accessioned throughout 1989 by 436 institutions in North America. The aggregate database consisted of 13,094 lung FNA cases with 11,922 (91%) judged as satisfactory for cytologic evaluation. Of these satisfactory aspirates, 5264 (40%) had corresponding histologic tissue biopsy preparations and FNA diagnoses available for further evaluation and formed the basis for determining diagnostic accuracy. There was no significant difference in overall performance results derived from the data provided by all participants compared with the median of those reporting a greater number of correlated FNA cases. In the diagnosis of lung cancer by FNA, the following performance results were derived using the aggregate database: 89% sensitivity of FNA procedure, 99% sensitivity of FNA diagnosis, 96% specificity, 99% positive predictive value, 70% negative predictive value, 91% efficiency, 0.8% false-positive FNA interpretation, and 8% false-negative rate. The aggregate value and median performance values of sensitivity and specificity derived from this Q-Probe study, which reflects the general practices of mostly non-university hospitals in North America, compare very favorably with study results of similar design in the literature reflecting practices from academic centers. This appears to validate published rates from academic centers as reproducible in the general practice of pathology and validates the use of these values derived from an aggregate database as a benchmark to measure performance improvement in lung FNA.     

Interinstitutional comparison of frozen section consultation in small hospitals: a College of American Pathologists Q-Probes study of 18,532 frozen section consultation diagnoses in 233 small hospitals

OBJECTIVE: To study pathology intraoperative consultation practices and the accuracy of diagnoses made by frozen section. DESIGN: In 1994, participants in the College of American Pathologists Q-Probes laboratory quality improvement program each completed questionnaires and prospectively collected data on up to 20 frozen section procedures performed over a 5-month period. SETTING: Surgical pathology laboratories serving private and public hospitals with 300 or fewer occupied beds. PARTICIPANTS: Two hundred thirty-two North American institutions and one New Zealand institution. MAIN OUTCOME MEASURES: The discordance and deferral rates of frozen section diagnoses and the reasons for frozen section discordance relative to corresponding diagnoses made on permanent (paraffin) sections. Calculation of frozen section discordance rates excluded diagnoses of subtypes or grade of malignancy, biopsies on specimens in which there was no gross lesion (eg, mammographic specimens), thyroid follicular lesions, tissue taken only to determine adequacy for other studies (eg, estrogen-binding proteins), and frozen sections performed to evaluate margins of specimens oriented en face. RESULTS: Out of 18,532 frozen section diagnoses performed on 327,884 surgical cases, 859 (4.6%) diagnoses were deferred until permanent sections were available for review; 17,357 (98.2%) nondeferred diagnoses agreed with, and 316 (1.8%) disagreed with, those diagnoses rendered on permanent sections. The most common cause of discordance was underdiagnosis of neoplasia, usually due to block- or tissue-sampling errors. CONCLUSIONS: We recommend that laboratories routinely monitor frozen section discordance, cut additional sections deeper into the frozen block and/or sample additional tissue when the initial frozen section diagnosis is negative or nonproductive, reconcile all discordant frozen section diagnoses in the final report, and periodically assess the value of performing frozen section examinations.     

Laboratory phlebotomy. College of American Pathologists Q-Probe study of patient satisfaction and complications in 23,783 patients

Outpatients from 630 institutions participated in a phlebotomy module of Q-Probes, a quality assurance program of the College of American Pathologists, Northfield, Ill. This module assessed patient outcome measurements of complications, discomfort, and satisfaction with the phlebotomy procedure. Of the 29,700 ostensibly healthy individuals registered, 80.1% returned postcards containing measurements and assessments they made about the procedure and information recorded by the phlebotomist. The median time required for phlebotomy was 6 minutes, with 25% of patients requiring less than 5 minutes and 10% more than 21 minutes for completion of the procedure. The average number of phlebotomy attempts per patient was 1.03, with 95 patients (0.4%) experiencing three to 11 attempts. Ecchymoses occurred in 4048 (16.6%) attempts, with the median size of ecchymosis being 15.1 mm. On the average, an outstanding employee was identified by patients 46.6% of the time. The discomfort caused by the needle puncture was more than expected by 35.3% of patients. Although 98.6% of the patients were satisfied, 336 patients were dissatisfied with the procedure. We conclude that the technical skills of phlebotomists and patient satisfaction with phlebotomy are outstanding, but that patient discomfort from the procedure needs to be minimized.     

Operating room blood delivery turnaround time: a College of American Pathologists Q-Probe Study of 12647 units of blood components in 466 institutions

OBJECTIVES: To determine the normative distribution of time elapsed for blood bank personnel to fill nonscheduled operating room (OR) blood component orders in hospital communities throughout the United States, and to examine hospital blood bank practices associated with faster blood component delivery times. DESIGN: Participants in the College of American Pathologists Q-Probes laboratory quality improvement program collected data prospectively on the times elapsed for blood bank personnel to fill nonscheduled emergent orders from hospital ORs for red blood cell (RBC) products, fresh frozen plasma (FFP), and platelets (PLTs). Participants also completed questionnaires describing their hospitals' and blood banks' laboratory and transfusion practices. SETTING AND PARTICIPANTS: Four hundred sixty-six public and private institutions located in 48 states in the United States (n = 444), Canada (n = 9), Australia (n = 8), the United Kingdom (n = 4), and Spain (n = 1). MAIN OUTCOME MEASURES: The median time elapsed between requests for blood components by OR personnel and the retrieval of those components by blood component transport personnel, and the median time elapsed between requests for blood components by OR personnel and the arrival of those components in ORs. RESULTS: Participants submitted data on 12 647 units of RBCs, FFP, and PLTs. The median aggregate request-to-retrieval turnaround times (TATs) for RBCs, FFP, and PLTs ranged from 30 to 35 minutes, and the median aggregate request-to-arrival TATs for RBCs, FFP, and PLTs ranged from 33 to 39 minutes. Most of the TAT was consumed by events occurring prior to, rather than after release of components from blood banks. Shorter prerelease TATs were associated with having surgical schedules that listed patients' names and procedures available to blood bank personnel prior to surgeries, and having adequate clotted specimens in the blood bank and completed type-and-screen procedures performed before requests for blood components were submitted to blood banks. Among the fastest-performing 10% of participants (90th percentile and above), request-to-retrieval TATs ranged from 12 to 24 minutes for the 3 blood components, whereas among the slowest-performing 10% of participants (10th percentile and below), request-to-retrieval TATs ranged from 63 to 115 minutes for the 3 components. Median TATs ranged from 33 to 37 minutes for the 3 components. Institutions with TATs in the fastest-performing 25th percentile more frequently stored cross-matched RBCs in the OR daily, stocked PLTs for unexpected surgical use, stored PLTs in or near the OR, and had laboratory rather than nonlaboratory personnel deliver components to the OR than did those institutions with TATs in the slowest-performing 25th percentile. CONCLUSIONS: Hospital blood bank personnel can deliver blood components to the OR in slightly longer than 30 minutes, measured from the time that those units are requested by OR personnel. Practices aimed at saving time before components are released from blood banks will be more efficient in reducing overall TAT than those practices aimed at saving time after components are released from blood banks. Specific practices associated with shorter blood delivery TATs included providing blood bank personnel with access to the names of surgical patients potentially requiring blood components, having pretransfusion testing completed on those patients prior to surgery, having ample blood products on hand, and having laboratory personnel control blood product delivery.     

Outpatient order accuracy. A College of American Pathologists Q-Probes study of requisition order entry accuracy in 660 institutions

CONTEXT: Laboratory test order entry errors potentially delay diagnosis, consume resources, and cause patient inconvenience. OBJECTIVE: To evaluate the frequency and causes of computer order entry errors in outpatients. DESIGN: Cross-sectional survey and prospective sample of errors. Participants answered questions about their test order entry policies and practices. They then examined a sample of outpatient requisitions and compared information on the requisition with information entered into the laboratory computer system. Order entry errors were divided into 4 types: tests ordered on the requisition, but not in the computer; tests performed but not ordered on the requisition; physician name discrepancies; and test priority errors. PARTICIPANTS: Six hundred sixty laboratories enrolled in the College of American Pathologists Q-Probes program. MAIN OUTCOME MEASURE: Overall order entry error rate. RESULTS: A total of 5514 (4.8%) of 114 934 outpatient requisitions were associated with at least 1 order entry error. The median participant reported 1 or more order errors on 6.0% of requisitions; 10% of institutions reported errors with at least 18% of requisitions. Of the 4 specific error types, physician name discrepancies had the highest error rate, and test priority errors the lowest error rate. Four institutional factors were significantly associated with higher overall error rates: orders verbally communicated to the laboratory; no policy requiring laboratory staff to compare a printout or display of ordered tests with the laboratory requisitions to confirm that orders had been entered correctly; failure to monitor the accuracy of outpatient order entry on a regular basis; and a higher percentage of occupied beds (ie, a busier hospital). CONCLUSIONS: Computer order entry errors are common, involving 5% of outpatient requisitions. Laboratories may be able to decrease error rates by regularly monitoring the accuracy of order entry, substituting written and facsimile orders for verbal orders, and instituting a policy in which orders entered into computer systems are routinely rechecked against orders on requisitions.     

Duplicate laboratory orders: a College of American Pathologists Q-Probes study of thyrotropin requests in 502 institutions

OBJECTIVE: To examine the frequency and cause of duplicate thyrotropin (TSH) testing. METHODS: Five hundred two institutions, ranging in size from fewer than 100 to more than 600 beds, examined consecutively processed TSH assays to identify duplicate orders. Duplicates were defined as two or more TSH tests performed within 7 days. All together, participants submitted data on 221,476 TSH orders. RESULTS: The median institution reported that 1.5% of TSH tests duplicated a TSH order that had been received from the same patient within the previous 7 days. Ten percent of institutions reported that 4.5% or more of their TSH tests were duplicates. Institutions with higher duplicate rates tended to be larger (ie, they had a greater number of occupied beds) and to have duplicate tests that were more likely to be ordered by a physician other than the one who ordered the initial test. Participants reported that for 19% of duplicate orders, physicians were unaware that the first test had been ordered. Physicians also indicated that duplicate assays were ordered to see if a previous result had changed (15%) or to check on the accuracy of a previous result (13%). Participants reported that 11% of duplicate TSH assays that their laboratory performed had apparently never been ordered. CONCLUSIONS: A large number of institutions are performing duplicate TSH tests that, in most cases, appear to be medically unnecessary. Institutions aiming to reduce the frequency of duplicate testing should consider policies that decrease the opportunity for different physicians to order tests on a single patient and should increase the accuracy with which physician orders are transmitted to the laboratory.     

Postanalytical quality improvement: a College of American Pathologists Q-Probes study of elevated calcium results in 525 institutions

OBJECTIVE: To evaluate elevated patient calcium results as a postanalytic quality indicator of physician practices. DESIGN: Participants prospectively identified hypercalcemic patient results for 4 months or until they found 320 hypercalcemic results, and then, after at least 3 days, reviewed the medical records of these patients. Hypercalcemia was defined as a calcium value that exceeded the upper limit of each laboratory's reference range by 0.12 mmol/L or more. Participants, as well a subset of their physicians who did not acknowledge or respond to elevated results in the medical record, answered a questionnaire about their practices. PARTICIPANTS: Five hundred twenty-five laboratories enrolled in the College of American Pathologists Q-Probes program. MAIN OUTCOME MEASURES: The presence of hyercalcemic results in patients' medical records and physicians' acknowledgement and response to those elevated results. RESULTS: More than 5500 hypercalcemic results were identified, of which 53.2% represented a new finding. About 3.5% of results were not charted in the patients' records, and 23.1% of patient records did not contain clinician documentation of the abnormal result. Follow-up laboratory tests were not ordered for 13.8% of the elevated values. For 570 of the 808 results for which there was neither clinician documentation nor designated follow-up laboratory tests ordered, patients' physicians received written notification of the elevated calcium results along with a questionnaire. Responses were received from 386 physicians (68%). One hundred physicians indicated they did not order the specific calcium measurement, and of these 100, 85 responded it was part of a panel. The 286 physicians who ordered the test stated the results ultimately led to further testing (69%), a change of management (56%), or a new diagnosis (25%). CONCLUSIONS: We found that a high percentage of abnormal results (3.5%) were not documented in the patients' medical records, the diagnosis of hypercalcemia frequently was new (53.2%), and a high percentage of physicians did not respond to elevated calcium results by writing a note (23.1%) or ordering another test (13.8%). Opportunities for quality improvement at these postanalytical steps are far greater than at the analytical step. Laboratorians must help physicians identify and respond to clinically important laboratory results.     

Surgical pathology-based outcomes assessment of breast cancer early diagnosis: a College of American Pathologists Q-Probes study in 199 institutions

OBJECTIVE: To develop breast cancer outcomes data relating pathologic tumor variables at diagnosis with clinical method of detection. DESIGN: Anatomic pathologists assessed 30 consecutive breast cancers at each institution, resulting in an aggregate database of 4232 breast cancers. SETTING: Hospital-based laboratories from the United States (98%), Canada, Australia, and Belgium. PARTICIPANTS: One hundred ninety-nine laboratories in the 1999 College of American Pathologists Q-Probes voluntary quality improvement program. MAIN OUTCOME MEASURES: Pathologic variables indicative of favorable outcomes included percentage of carcinomas detected at the in situ stage, tumors < or = 1 cm in diameter, and invasive cancers with lymph nodes negative for metastases. RESULTS: All outcomes measures, including percent in situ carcinomas (26.9% vs 13.8%), tumor size < or = 1 cm (57.8% vs 36.5%), and lymph node-negative status (77.8% vs 64%), were more favorable when tumors were detected by screening mammography (P <.001) compared to all other detection methods. CONCLUSIONS: This study demonstrates an opportunity for pathologists to develop outcomes information of interest to health care organizations, providers, patients, and payers by integrating routine oncologic surgical pathology and clinical breast cancer detection data. Such readily obtained interim outcomes data trended and benchmarked over time can demonstrate the relative clinical efficacy of preventive breast care provided by health care systems long before mortality data are available.