Financing U.S. Graduate Medical Education: A Policy Position Paper of the Alliance for Academic Internal Medicine and the American College of Physicians.

In this position paper, the Alliance for Academic Internal Medicine and the American College of Physicians examine the state of graduate medical education (GME) financing in the United States and recent proposals to reform GME funding. They make a series of recommendations to reform the current funding system to better align GME with the needs of the nation's health care workforce. These recommendations include using Medicare GME funds to meet policy goals and to ensure an adequate supply of physicians, a proper specialty mix, and appropriate training sites; spreading the costs of financing GME across the health care system; evaluating the true cost of training a resident and establishing a single per-resident amount; increasing transparency and innovation; and ensuring that primary care residents receive training in well-functioning ambulatory settings that are financially supported for their training roles.

Health policy basics: health insurance marketplaces.

Starting on 1 October 2013, most individuals and small businesses will be able to shop for and enroll in health insurance coverage through their state's health insurance marketplace, also known as an exchange. The health insurance marketplaces will serve as a one-stop resource to help the uninsured and the underinsured find comprehensive health coverage that fits their needs and budget and determine whether they qualify for health insurance tax credits provided by the Patient Protection and Affordable Care Act. Physicians may benefit because insured patients are more likely to have a regular source of care, adhere to medical regimens, and access preventive care. However, implementation of the marketplaces may prove challenging if enrollment numbers are insufficient, technical problems arise, and patients are unable to access providers. Despite these potential issues, physicians are encouraged to educate themselves about how the marketplaces work so they can direct their patients to find the coverage that best meets their medical needs.

Thyroid disease awareness is associated with high rates of identifying subjects with previously undiagnosed thyroid dysfunction.

BACKGROUND: Conventional screening for hypothyroidism is controversial. Although hypothyroidism is underdiagnosed, many organizations do not recommend screening, citing low disease prevalence in unselected populations. We studied attendees at a thyroid health fair, hypothesizing that certain patient characteristics would enhance the yield of testing. METHODS: We carried out an observational study of participants at a Michigan health fair that focused on thyroid disease. We collected patient-reported symptoms and demographics by questionnaire, and correlated these with the TSH values obtained through the health fair. RESULTS: 794 of 858 health fair attendees participated. Most were women, and over 40% reported a family history of thyroid disease. We identified 97 (12.2%) participants with previously unknown thyroid dysfunction. No symptom or combination of symptoms discriminated between hypothyroid and euthyroid individuals. Hypothyroid and euthyroid participants in the health fair reported each symptom with a similar prevalence (p > 0.01), a prevalence which was very high. In fact, when compared with a previously published case-control study that reported symptoms, the euthyroid health fair participants reported a higher symptom prevalence (range 3.9% to 66.3%, mean 31.5%), than the euthyroid individuals from the case-control study (range 2% to 54%, mean 17.4%). CONCLUSIONS: A high proportion of previously undiagnosed thyroid disease was identified at this health fair. We initially hypothesized symptoms would distinguish between thyroid function states. However, this was not the case in this health fair screening population. The prevalence of reported symptoms was similar and high in both euthyroid and hypothyroid participants. Because attendees were self-selected, it is possible that this health fair that focused on thyroid disease attracted participants specifically concerned about thyroid health. Despite the lack of symptom discrimination, the much higher prevalence of hypothyroidism in this study (12%) compared with the general population (<2%) suggests that screening may be appropriate and effective in certain circumstances such as thyroid health fairs.

Gender differences in patient-provider symptom agreement in reporting respiratory complaints on a questionnaire.

BACKGROUND: Men and women communicate differently, but it is unclear whether this influences health care outcomes. OBJECTIVE: Because women patients frequently choose women health care providers, we examined whether this preference was affected by communication styles. We focused on communication of disease-specific symptoms, hypothesizing that symptom agreement between women patients and women health care providers would be greater than between other patient-provider gender combinations. METHODS: Patients attending outpatient clinics were recruited as part of a study of respiratory illness at 7 university-affiliated sites during 3 consecutive influenza seasons (2000-2003). Individuals aged > or = 19 years were offered enrollment if they sought care for cold or flu symptoms at a participating study site. Patients were eligible to participate in the study if they reported any 1 of 6 symptoms: cough, runny nose, fever (subjective), muscle aches, sore throat, and/or exhaustion. Using separate questionnaires, patients and their health care providers recorded the patients' respiratory symptoms (as present or absent). Patients recorded their symptoms before visiting their health care provider, and providers recorded patient symptoms after the visit. Symptom agreement was compared using general estimating equations across all gender combinations. RESULTS: A total of 327 patients (220 women, 107 men) and 84 health care providers (37 women, 47 men) participated in the study. Overall symptom agreement for all patient-provider gender combinations was 81.9% (95% CI, 79.6%-84.2%). For each symptom, the observed agreement significantly exceeded the agreement expected by chance alone (P < 0.001 for all symptoms except "no energy," which was P = 0.023). The male-male pairing of patient and provider was more likely to agree on a symptom than were the other gender combinations, although not statistically significantly more so than the female-female pairing. CONCLUSIONS: In this survey of patients with respiratory illness, there was no significant difference in symptom agreement for most symptoms between the male-male and female-female patient-provider combinations. Based on these findings, symptom agreement alone does not explain why women patients select women health care providers.

Predictors of urinary tract infection based on artificial neural networks and genetic algorithms.

BACKGROUND: Among women who present with urinary complaints, only 50% are found to have urinary tract infection. Individual urinary symptoms and urinalysis are not sufficiently accurate to discriminate those with and without the diagnosis. METHODS: We used artificial neural networks (ANN) coupled with genetic algorithms to evolve combinations of clinical variables optimized for predicting urinary tract infection. The ANN were applied to 212 women ages 19-84 who presented to an ambulatory clinic with urinary complaints. Urinary tract infection was defined in separate models as uropathogen counts of > or =10(5) colony-forming units (CFU) per milliliter, and counts of > or =10(2) CFU per milliliter. RESULTS: Five-variable sets were evolved that classified cases of urinary tract infection and non-infection with receiver-operating characteristic (ROC) curve areas that ranged from 0.853 (for uropathogen counts of > or =10(5) CFU per milliliter) to 0.792 (for uropathogen counts of > or =10(2) CFU per milliliter). Predictor variables (which included urinary frequency, dysuria, foul urine odor, symptom duration, history of diabetes, leukocyte esterase on urine dipstick, and red blood cells, epithelial cells, and bacteria on urinalysis) differed depending on the pathogen count that defined urinary tract infection. Network influence analyses showed that some variables predicted urine infection in unexpected ways, and interacted with other variables in making predictions. CONCLUSIONS: ANN and genetic algorithms can reveal parsimonious variable sets accurate for predicting urinary tract infection, and novel relationships between symptoms, urinalysis findings, and infection.

Can internal medicine residents master microscopic urinalysis? Results of an evaluation and teaching intervention.

PURPOSE: Although microscopic urinalysis (micro UA) is commonly used in clinical practice, and residents are trained in micro UA, proficiency in this procedure has not been studied. METHOD: In 1996-97, 38 residents in the University of Nebraska Medical Center's internal medicine (IM) residency program were evaluated on their technical ability to perform micro UA, and on their cognitive skills in recognizing common micro UA findings. After identifying deficits in the residents' cognitive competency, two educational interventions were applied and residents were tested after each intervention. RESULTS: A total of 24 residents (63%) correctly prepared the specimen for analysis (the technical portion). On the cognitive portion, only one of the 38 residents correctly identified 80% of all micro UA findings in the urinary sediment, although 11 (29%) residents identified UA findings specific to urinary tract infection (UTI). The first educational intervention did little to improve residents' performance. A second more intensive intervention resulted in 10 (45%) residents identifying 80% of all micro UA findings, and 19 (86%) residents correctly identifying UTI findings. CONCLUSIONS: Many residents were not proficient in performing micro UA, even after intensive educational interventions. Although micro UA is a simple procedure, residents' mastery cannot be assumed. Residency programs should assess competency in this procedure.

The effects of proficiency and bias on residents' interpretation of the microscopic urinalysis.

OBJECTIVE: This study aims to determine whether residents are influenced by clinical information when interpreting microscopic urinalysis (UA) and estimating the probability of a urinary tract infection (UTI), and to determine the accuracy and reliability of UA readings. DESIGN: Residents estimated the UA white blood cell count and the probability of a UTI in vignettes using a fractional factorial design, varying symptoms, gender, and the white blood cell count on preprepared urine slides. RESULTS: Individual-level results indicated a clinical information bias and poor accuracy. Seventeen of 38 residents increased the white blood cell count in response to female gender; 14 increased the white blood cell count in response to UTI symptoms. Forty-nine percent of the readings were inaccurate; agreement ranged from 50% to 67% for white and red blood cells and bacteria. CONCLUSION: Many residents gave inaccurate UA readings, and many readings varied with clinical information. A significant portion of residents needs assistance in objectively and accurately interpreting the UA.

Prediction of community-acquired pneumonia using artificial neural networks.

BACKGROUND: Artificial neural networks (ANN) have been used in the prediction of several medical conditions but have not been previously used to predict pneumonia. The authors used ANN to predict the presence or absence of pneumonia among patients presenting to the emergency department with acute respiratory complaints and compared the results with those obtained using logistic regression modeling. METHODS: Feed-forward back-propagation ANN were trained on sociodemographic, symptom, sign, comorbidity, and radiographic outcome data among 1,044 patients from the University of Illinois (the training cohort) and were applied to 116 patients from the University of Nebraska (the testing cohort). ANN trained using different strategies were compared to each other and to main-effects logistic regression. Calibration accuracy was measured as mean square error and discrimination accuracy as the area under a receiver operating characteristic (ROC) curve. RESULTS: A 1 hidden-layer ANN trained using oversampling of pneumonia cases had an ROC area in the training cohort of 0.895, which was greater than the area of 0.840 for logistic regression (P = 0.026). This ANN had an ROC area in the testing cohort of 0.872, not significantly different from its area in the training cohort (P = 0.597). Operating at a threshold of 0.25, the ANN would have detected 94% to 95% of patients with pneumonia in the 2 cohorts while correctly excluding 39% to 50% of patients with other conditions. ANN trained using other strategies discriminated equally in the 2 cohorts but no better than did logistic regression. CONCLUSIONS: Among adults presenting with acute respiratory illness, ANN accurately discriminated patients with and without pneumonia and, under some circumstances, improved on the accuracy of logistic regression.

The relationship between treatment objectives and practice patterns in the management of urinary tract infections.

OBJECTIVE: To describe physicians' goals when treating uncomplicated urinary tract infections (UTIs) and the relationship between goals and practice patterns. STUDY DESIGN: Analysis of survey results. POPULATION: Primary care physicians. OUTCOMES MEASURED: Self-reported treatment objectives and practice patterns. RESULTS: Most physicians reported their UTI management was convenient for the patient (81.3%). Fewer stated they minimized patients' costs (53.4%), made an accurate diagnosis (56.7%), or avoided unnecessary antibiotics (40.9%). Physicians who stressed convenience or minimizing patient expenses were less likely to use many resources (urine culture, microscopic urinalysis, followup visits and tests, and prolonged antibiotic treatment) and more likely to use telephone treatment. Physicians who stressed accurate diagnoses or avoiding unnecessary antibiotics were more likely to use the same resources and less likely to use telephone treatment. CONCLUSION: UTI management goals vary across physicians and are associated with different clinical approaches. Differences in treatment objectives may help explain variations in practice patterns.

Use of genetic algorithms for neural networks to predict community-acquired pneumonia.

BACKGROUND: Genetic algorithms have been used to solve optimization problems for artificial neural networks (ANN) in several domains. We used genetic algorithms to search for optimal hidden-layer architectures, connectivity, and training parameters for ANN for predicting community-acquired pneumonia among patients with respiratory complaints. METHODS: Feed-forward back-propagation ANN were trained on sociodemographic, symptom, sign, comorbidity, and radiographic outcome data among 1044 patients from the University of Illinois (the training cohort), and were applied to 116 patients from the University of Nebraska (the testing cohort). Binary chromosomes with genes representing network attributes, including the number of nodes in the hidden layers, learning rate and momentum parameters, and the presence or absence of implicit within-layer connectivity using a competition algorithm, were operated on by various combinations of crossover, mutation, and probabilistic selection based on network mean-square error (MSE), and separately on average cross entropy (ENT). Predictive accuracy was measured as the area under a receiver-operating characteristic (ROC) curve. RESULTS: Over 50 generations, the baseline genetic algorithm evolved an optimized ANN with nine nodes in the first hidden layer, zero nodes in the second hidden layer, learning rate and momentum parameters of 0.5, and no within-layer competition connectivity. This ANN had an ROC area in the training cohort of 0.872 and in the testing cohort of 0.934 (P-value for difference, 0.181). Algorithms based on cross-generational selection, Gray coding of genes prior to mutation, and crossover recombination at different genetic levels, evolved optimized ANN identical to the baseline genetic strategy. Algorithms based on other strategies, including elite selection within generations (training ROC area 0.819), and inversions of genetic material during recombination (training ROC area 0.812), evolved less accurate ANN. CONCLUSION: ANN optimized by genetic algorithms accurately discriminated pneumonia within a training cohort, and within a testing cohort consisting of cases on which the networks had not been trained. Genetic algorithms can be used to implement efficient search strategies for optimal ANN to predict pneumonia.