Quality assurance in anatomic pathology: automated SNOMED coding.

This case study details the set-up and implementation of the PathNet autocoder (Cerner Corporation) in a busy anatomic pathology laboratory. After initial start-up, procedures were developed to improve the system's performance. Four classes of software coding errors were identified, and an index was developed to measure the number of cases between errors (CBE). Through modifications in the program, the CBE increased sharply by the end of the six-month study period. During the last three months of the study, the efficiency of case retrieval was tested by comparing manual and electronic methods on the same reference cases. This demonstrated significant time saving and removed the variability of manual coding. The technique employed in this study may assist other institutions seeking to implement such a coding system within their respective environments.

Analysis of three decision-making methods: a breast cancer patient as a model.

PURPOSE: To compare three decision making techniques using a common clinical problem. METHODS: Two recently developed methods, the analytic hierarchy process (AHP) and the analytic network process (ANP), were compared with a Markov process in the evaluation of the optimal post-lumpectomy treatment strategy for an elderly woman with a mammographically detected, nonpalpable early-stage breast cancer. The following treatment alternatives were considered: observation, radiation, tamoxifen, combination radiation and tamoxifen, and simple mastectomy. All three decision methods incorporated patient preferences. RESULTS: The models agreed on the ranking of the preferred treatment, radiation and tamoxifen, but there were variations in the rankings of the other treatment choices. Individual differences between the three models were uncovered. The Markov process provided estimates of quality-adjusted life expectancy and distribution of health events. Both AHP and ANP required less development time than the Markov process. CONCLUSION: All three methods may be useful tools to the clinician in analyzing complex medical problems. The Markov is the most labor-intensive method but provides detailed results, whereas the AHP and the ANP give only rank orders of the alternatives. The most important considerations in choosing between these methods are time to project completion and the detail of information sought.

Optimal interval for triple-lumen catheter changes: a decision analysis.

A survey of 53 university and community hospitals revealed that 73% of the institutions had no standard policy for the replacement of triple-lumen catheters (TLCs). Since the maintenance of a TLC in place for a prolonged period may lead to infectious complications, it appeared warranted that standards of management be developed. A decision-tree model was constructed for evaluating the optimal time for changing a TLC that would minimize infection. Cost estimates and health effects at three-, five-, and ten-day change intervals were considered for catheter insertion and complications resulting from such insertion. The results suggested that prophylactic catheter changes should occur no later than every five days, provided that there are no signs of infection. However, sensitivity analysis of several variables suggested that individual institutions should establish policy timing changes based upon careful interpretation of their own data. A model was developed to assist in determining the optimal time to change a TLC based upon such data.

Treatment of early-stage breast cancer in the elderly: a health-outcome-based approach.

PURPOSE: To evaluate the post-lumpectomy treatment of a nonpalpable, stage I, T1b tumor, mammographically detected, in a 74-year-old woman without comorbidities. METHODS: A Markov process, through 120 monthly cycles, was used to model patient progression through a treatment program, employing literature data and a health-outcome utility. Treatments considered were: observation; radiation totaling 5,000 cGy over six weeks; tamoxifen, 20 mg/day, for five years; simple mastectomy; and radiation therapy plus tamoxifen. Health states included absence of disease (NED), loco-regional recurrence, distant metastasis, age-sex-race (ASR)-adjusted death, cancer mortality, treatment complications, and post-mastectomy death. Transition probabilities were established from the literature. Health-state utilities were determined from the responses of health care professionals to a basic reference gamble. RESULTS: Quality-adjusted life years (QALYs) were determined to be 8.19 for radiation plus tamoxifen, decreasing to 8.04 for mastectomy, a difference of only a 0.15 years (1.8 months). Sensitivity analysis, however, showed relative stability in the ranking among treatment options. CONCLUSION: Although the model showed little difference between QALYs with the treatments, the combination of radiation and tamoxifen provides the optimal therapy for this case.

Sequential test selection in the analysis of abdominal pain.

Numerous decision-making tools exist to assist physicians in diagnosis management. However, the accuracy of available clinical information is often ambiguous or unknown and current analytical models do not explicitly incorporate judgementally defined information. A model encompassing both physician judgment and probability analysis was developed to accommodate such data. A problem requiring sequential diagnostic testing was structured utilizing the analytic hierarchy process (AHP). The case presented involved a patient complaining of upper abdominal pain who, after initial evaluation, did not need immediate surgery. Physicians were faced with identifying the optimal sequence of diagnostic testing. The criteria used for test selection included minimizing risk, patient discomfort, and cost of testing and maximizing diagnostic capability. Although at the onset the "best" test choice was unknown, the clinical picture indicated four test alternatives: upper gastrointestinal series (GI), abdominal ultrasonography (US), abdominal computed tomography (CT), and upper gastrointestinal endoscopy (END). Based upon the relative preferences of the criteria utilized, the AHP analysis indicated that upper GI series was the optimal first test. Given a negative test, posterior probabilities were calculated using Bayes' theorem, resulting in a new estimate of diagnostic capability. The AHP analysis was reiterated, identifying abdominal ultrasonography as the optimal second test. This analysis may be repeated as many times as necessary. Sensitivity analysis demonstrated that changing criteria preferences may alter the choice of tests and/or their sequence.

Urea reduction ratio and urea kinetic modeling: a mathematical analysis of changing dialysis parameters.

Based mainly on the simplicity of its calculation, the urea reduction ratio (URR) has been suggested as an alternative to urea kinetic modeling (Kt/V) as a measure of hemodialysis adequacy. However, recent studies have raised questions concerning the accuracy of URR, particularly in the presence of residual kidney function (KrU). This study was initiated to evaluate the relationship between URR and Kt/V under a variety of dialysis conditions. Equations based on the variable-volume, single-pool model described by Gotch were used to construct a model incorporating the variables used in the estimation of URR and Kt/V. The model's prediction of URR correlated closely with measured URR in 30 patients (r = 0.9987, p < 0.000001). This analytic approach showed that changes in each of several dialysis parameters caused divergence in the values of URR and Kt/V. The model showed that URR could be less than 0.65, while total Kt/V was greater than 1.2, whether or not KrU was present. In fact, when KrU was greater than 1. 0, URR could be less than 0.65, while Kt/V might be 2.0 or higher. On the other hand, the model showed instances where URR could be greater than 0.65, when Kt/V was less than 1.2. This occurred only when KrU was less than 1.0. To determine the prevalence of these anomalies in clinical practice, 767 kinetic modeling determinations were evaluated in 207 patients. One of the above discrepancies was observed at least once in 30.9% of the patients, representing 12.1% of all determinations. In addition, it was found that omitting KrU from the calculation of Kt/V generally leads to a Kt/V <1.2. This, when associated with a URR <0.65, could erroneously imply inadequate dialysis. The patient data are consistent with the view that URR and Kt/V are the net result of several variables that may act together or even in opposing directions. Based on this mathematical model and the observed clinical data, the use of URR alone to assess dialysis adequacy, or neglecting the contribution of KrU to Kt/V, may lead to changes in the dialysis prescription that are neither correct nor necessary.