Impact of complications on outcomes following aortic and mitral valve replacements in the United States.

AIM: Heart valve replacement surgeries account for 20% of all cardiac procedures. In-hospital mortality rates are approximately 6% for aortic valve replacements and 10% for mitral valve replacements. The objectives of the study are to provide nationally representative estimates of complications following aortic and mitral valve replacements and to quantify the impact of different types of complications on in-hospital outcomes. METHODS: The Nationwide Inpatient Sample was analyzed for years 2000-2003. The effect of complications on in-hospital mortality, length of stay (LOS), and hospital charges were examined using bivariate and multivariable logistic and linear regression analyses. The confounding effects of age, sex, primary diagnosis, type of valve replacement, type of admission, comorbid conditions, and hospital characteristics were adjusted. RESULTS: A total of 43,909 patients underwent aortic valve replacement as the primary procedure during the study period and 16,516 patients underwent mitral valve replacement. Complications occurred in 35.2% of those undergoing aortic valve replacements and in 36.4% of those undergoing mitral valve replacements. Almost half of these are cardiac complications and a quarter involve hemorrhage/hematoma/seroma. Complications were significantly associated with in-hospital mortality, LOS, and hospital charges even after adjusting for patient and hospital characteristics. CONCLUSION: Complications are prevalent and exert a considerable influence on outcomes following aortic and mitral valve replacements. Quality initiatives should focus on minimizing complications and improving processes of care that would enable complications to be better resolved if they occur.

An introduction to time-trend analysis.

Healthcare professionals often are presented with data that appear to indicate an upward or downward trend over time. For example, admissions of acquired immunodeficiency syndrome (AIDS) patients appear to be increasing, cesarean section rates appear to be decreasing, or nosocomial pneumonia rates appear to be increasing. Critical decisions sometimes are based on such trends, which often are presented without a statistical analysis. Those responsible for decision making may be left wondering whether these apparent trends represent only chance variation. Graphs showing trends over time generally present one of three kinds of outcome data: counts (eg, three AIDS admissions), proportions (eg, 10 cesarean sections per 100 total deliveries), or person-time data (eg, 13 cases of nosocomial pneumonia per 10,000 patient days). Using familiar examples and a minimum of technical language, we illustrate the analysis of time trends.

Community characteristics associated with elder abuse.

OBJECTIVES: To help define the relationship between elder abuse rates and counties' demographics, healthcare resources, and social service characteristics. DESIGN: County-level data from Iowa were analyzed to test the association between county characteristics and rates of elder abuse between 1984 and 1993 using univariate correlation analysis and stagewise linear regression. SETTING: Ninety-nine counties in Iowa. PARTICIPANTS: Iowa residents aged 65 years and older. MEASUREMENTS: County-level population-adjusted numbers of abused elderly, abused children, children in poverty, high school dropouts, physicians and other healthcare providers, hospital beds, social workers and caseworkers in the Department of Human Services (DHS). RESULTS: Community characteristics that had a positive association with rates of reported or substantiated elder abuse at the P < .001 level were population density, children in poverty, and reported child abuse. Lower substantiated elder abuse rates were associated at P < .05 with higher community rates of high school dropouts, number of chiropractors, and number of nurse practitioners. After adjusting for number of DHS caseworkers and reported child abuse rates (a surrogate for workload) a district effect persists for substantiated elder abuse cases (P = .002). CONCLUSION: County demographics are risk factors for reported and substantiated elder abuse. The strongest risk factor for reported elder abuse was reported child abuse. The difference in districts may reflect differences in resources and/or differing characteristics of caseworkers who substantiate elder abuse. The risk factors may reflect conditions that influence the amount of elder abuse or the detection of existing elder abuse.

The physician's decision to use tube feedings: the role of the family, the living will, and the Cruzan decision.

OBJECTIVE: To determine the relative importance of factors influencing physicians to use tube feedings in patients lacking decision-making capacity. DESIGN: Survey. SETTING AND PARTICIPANTS: Four hundred thirty-nine members of the Missouri Academy of Family Physicians. MEASUREMENTS: Using a mailed questionnaire, physicians were asked for a decision about feeding tube placement in an 89-year-old man who was unable to swallow or communicate after a stroke. Changing the conditions of the scenario, we then evaluated the influence of patient age, duration of disability, a living will, the Cruzan decision, and family preferences on the physician's decision. RESULTS: After reading the initial case history, 47% of physicians opposed tube feedings. Physicians who were told that the patient signed a living will specifically excluding tube feedings were more likely to oppose tube feedings than those who were told that he signed only a standard living will (53% vs 42%; P = 0.02). Forty-two percent of physicians who initially suggested a feeding tube changed their recommendation if the family opposed it. Sixty-six percent of physicians who initially opposed a feeding tube changed their recommendation if the family "pushed" for it. Thirty-three percent of physicians who initially opposed tube feedings under the living will scenario would favor tube feedings if the patient had not signed a living will. Twenty-two percent of physicians who initially opposed tube feedings would change to favor them if the issue had arisen before the Cruzan decision. CONCLUSIONS: Among this group of physicians, there was no consensus on whether tube feedings should be initiated. However, it was found that the family's opinion was the most influential factor affecting the physician's recommendation about tube feedings. Most physicians endorsed family preferences for tube feedings even when this intervention was specifically opposed in the patient's living will.

The association between manual removal of the placenta and postpartum endometritis following vaginal delivery.

OBJECTIVE: To determine whether manual removal of the placenta after vaginal delivery is a risk factor for postpartum endometritis. METHODS: A retrospective cohort study of vaginal deliveries compared 1052 patients who had manual removal of the placenta with 1085 patients whose placentas delivered spontaneously. Subjects were selected randomly from the 25,687 vaginal deliveries at the University of Iowa Hospitals during 1979-1992. The presence of endometritis was determined using information in medical records. The data were analyzed using odds ratios (OR) and multiple logistic regression. RESULTS: After controlling for confounding variables, manual removal of the placenta was associated with postpartum endometritis (adjusted OR 2.9, 95% confidence interval [CI] 1.7-4.9). Other risk factors for endometritis included maternal age less than 17 years (OR 3.3, 95% CI 1.5-7.2), postpartum anemia (OR 2.9, 95% CI 1.9-4.5), and membranes ruptured longer than 24 hours (OR 2.5, 95% CI 1.4-4.3). CONCLUSION: Manual removal of the placenta is a risk factor for postpartum endometritis after vaginal delivery.

Clinical diagnosis and the order of information.

BACKGROUND: Information order can influence judgment. However, it remains unclear whether the order of clinical data affects physicians' interpretations of these data when they are engaged in familiar diagnostic tasks. METHODS: Of 400 randomly selected family physicians who were given a questionnaire involving a brief written scenario about a young woman with acute dysuria, 315 (79%) returned usable responses. The physicians had been randomized into two groups, and both groups had received the same clinical information but in different orders. After learning the patient's chief complaint, physicians received either the patient's history and physical examination results followed by the laboratory data (the H&P-first group) or the laboratory data followed by the history and physical examination results (the H&P-last group). The results of the history and physical examination were supportive of the diagnosis of UTI, while the laboratory data were not. All physicians judged the probability of a urinary tract infection (UTI) after each piece of information. RESULTS: The two groups had similar mean estimates of the probability of a UTI after learning the chief complaint (67.4% vs 67.8%, p = 0.85). At the end of the scenario, the H&P-first group judged UTI to be less likely than did the H&P-last group (50.9% vs 59.1%, p = 0.03) despite having identical information. Comparison of the mean likelihood ratios attributed to the clinical information showed that the H&P-first group gave less weight to the history and physical than did the H&P-last group (p = 0.04). CONCLUSIONS: The order in which clinical information was presented influenced physicians' estimates of the probability of disease. The clinical history and physical examination were given more weight by physicians who received this information last.

Identifying left lower lobe pneumonia at chest radiography: performance of family practice residents before and after a didactic session.

RATIONALE AND OBJECTIVES: The authors sought to determine whether the lateral chest radiograph is helpful in identifying left lower lobe pneumonia among inexperienced readers. MATERIALS AND METHODS: The authors selected all patients who presented to a family practice training program with radiologic and clinical evidence of left lower lobe pneumonia (n = 65). They then selected an equal number of patients in whom chest radiographs were taken to "rule out pneumonia" and were found to be normal. Eight 1st-year family practice residents were asked to read the radiographs before and after a didactic session that emphasized lateral chest radiograph interpretation. The radiographs were presented under two viewing conditions: posteroanterior (PA) only versus PA and lateral. Receiver operating characteristic (ROC) curve methods were used to compare the effect of both the didactic session and the viewing condition on diagnostic accuracy. RESULTS: There were no significant differences in performance before and after the didactic session and no differences between the two viewing conditions. After including only abnormal radiographs that demonstrated the "spine sign" (an apparent increased opacification of the lower vertebral bodies on the lateral view), the residents performed better when presented with both PA and lateral radiographs than when presented with the PA radiograph only (area under ROC curve, .8158 vs .7418, respectively; P = 0.24). CONCLUSION: In patients with left lower lobe pneumonia whose radiographs demonstrated the spine sign, diagnostic accuracy improved when the lateral chest radiograph was viewed.

Confounding bias and effect modification in epidemiologic research.

The medical literature contains many studies of the clinical effectiveness of diagnostic tests and therapeutic interventions. Common problems in experimental design occur that influence the usefulness of original research. Confounding bias and effect modification are two important factors that affect whether clinicians ought to apply the findings of clinical research to the care of their patients. Investigators should minimize confounding biases in their work. Effect modification should be described so that readers can decide which of their patients will benefit from a particular study. This article uses a number of clinical examples to help the clinician and investigator understand the influences of bias and effect modification on clinical research.

Home visit program for teaching elder abuse evaluations.

BACKGROUND AND OBJECTIVES: A home visit program was designed to teach family practice residents how to evaluate patients for elder abuse and capacity (the ability to make one's own decisions). METHODS: Residents assessed potential abuse victims reported to Arizona's Adult Protective Service (APS) in their homes. Written evaluations prepared immediately following each home visit were abstracted for diagnoses (including abuse), recommendations, and patient demographics. Follow-up surveys by APS case workers determined whether the home visit recommendations were accomplished. Graduates of the residency were surveyed about their perceptions of the educational value of the program and their practice characteristics. RESULTS: The residents evaluated 201 patients. The mean age was 77, and 73% of patients were female. Seventy-five percent were incapacitated, 65% of these because of dementia. Ninety-one percent were abused, and the types of abuse included neglect (69%), exploitation (20%), physical abuse (8%), and unknown (3%). Recommendations were accomplished in the majority of cases: medical advice (68%), services (65%), medical evaluations (58%), guardian (53%), and conservator (52%). Graduates who participated in this program (1985-1992) rated their ability to diagnose elder abuse and to assess the patient's home environment significantly higher than earlier graduates who did not participate in the program (1977-1984). Earlier graduates made more home visits and provided more statements for guardianship than later graduates. CONCLUSIONS: The home visit program gave residents exposure to a population of elderly who were abused, demented, and living at home. This program provided clinical substance to build an effective teaching experience and furnished APS with a needed service.

Russian family practice training program: a single step on a long journey.

BACKGROUND AND OBJECTIVES: Since 1992, when the Russian government recognized family practice as a medical specialty, efforts have begun to progress from the idea stage to the delivery of patient care via family practice methods. We describe an educational effort to help teach Russian physicians family practice skills. METHODS: Five young Russian physicians were selected from an initial pool of 15 candidates on the basis of standardized testing, English language skills, and their potential to teach future Russian family physicians. Clinical, teaching, and business curricula were developed and used during the 6-month training period for the five selected physicians. Trainees were evaluated by mentors' and preceptors' written evaluations and by the American Board of Family Practice In-training Examination before, during, and at completion of the training. Subsequently, a fully equipped family practice office was opened in St Petersburg to serve as an on-site training facility. RESULTS: The trainees' self-perceived knowledge in community medicine, geriatrics, medical decision making, patient education, behavioral science, preventive medicine, and general family practice topics improved over the course of training. The composite scores on the in-training examinations improved from baseline (30 versus 308). Preceptors noted the greatest improvements in the use of clinical instruments, proficiency in physical exams, accessing medical information, and formulating differential diagnoses. The St Petersburg family practice office opened on October 1, 1996. The trainees now participate in the care of patients in this office and teach a new class of family medicine interns. CONCLUSION: The training program we describe has allowed Russian physicians to acquire new skills and knowledge that they can use and adapt to training future Russian family physicians.

Understanding logistic regression analysis through example.

Logistic regression is a valuable statistical tool that is often used in primary care research. When researchers explore the association between a possible risk factor and a disease, they attempt to control the effects of extraneous factors (confounders) that can obscure the true association. Using logistic regression, researchers can simultaneously control for the effects of multiple confounders. When investigators use logistic regression, they make subjective decisions about which factors to include in the analysis and in the final predictive model. Critical readers must understand basic concepts of logistic regression and potential problems with its use before they can accurately interpret study results. This article uses a familiar example to explain the principles of logistic regression to make it understandable to nonstatisticians.

The effect of physician characteristics on compliance with adult preventive care guidelines.

BACKGROUND AND OBJECTIVES: This study identified physician characteristics and attitudes related to self-reported compliance with adult prevention guidelines. METHODS: A questionnaire was mailed to family practice and internal medicine residents and faculty at the University of Iowa (n = 209). The questionnaire's 78 items fell into seven categories, including physician demographics, history-taking practices, counseling practices, self-perceived effectiveness in changing patient behavior, beliefs about preventive care, knowledge about preventive care, and perceived barriers to the delivery of preventive care. RESULTS: Compliance with history-taking recommendations was independently associated with high knowledge scores, female physician gender, and high self-perceived effectiveness in changing patient behavior. The only factor that was independently associated with counseling efforts was self-perceived effectiveness in changing patient behavior. CONCLUSIONS: Factors that were independently associated with self-reported preventive care efforts include female physician gender, knowledge about preventive care guidelines, and perceived effectiveness in changing patient behavior. After controlling for these factors, other variables such as lack of time, lack of reminder systems, attitudes about preventive care, and amount of formal preventive care education were not related to self-reported compliance with counseling and history-taking recommendations.

A taxonomy of generic clinical questions: classification study.

OBJECTIVE: To develop a taxonomy of doctors' questions about patient care that could be used to help answer such questions. DESIGN: Use of 295 questions asked by Oregon primary care doctors to modify previously developed taxonomy of 1101 clinical questions asked by Iowa family doctors. SETTING: Primary care practices in Iowa and Oregon. PARTICIPANTS: Random samples of 103 Iowa family doctors and 49 Oregon primary care doctors. MAIN OUTCOME MEASURES: Consensus among seven investigators on a meaningful taxonomy of generic questions; interrater reliability among 11 individuals who used the taxonomy to classify a random sample of 100 questions: 50 from Iowa and 50 from Oregon. RESULTS: The revised taxonomy, which comprised 64 generic question types, was used to classify 1396 clinical questions. The three commonest generic types were "What is the drug of choice for condition x?" (150 questions, 11%); "What is the cause of symptom x?" (115 questions, 8%); and "What test is indicated in situation x?" (112 questions, 8%). The mean interrater reliability among 11 coders was moderate (kappa=0.53, agreement 55%). CONCLUSIONS: Clinical questions in primary care can be categorised into a limited number of generic types. A moderate degree of interrater reliability was achieved with the taxonomy developed in this study. The taxonomy may enhance our understanding of doctors' information needs and improve our ability to meet those needs.

Analysis of questions asked by family doctors regarding patient care.

OBJECTIVES: To characterise the information needs of family doctors by collecting the questions they asked about patient care during consultations and to classify these in ways that would be useful to developers of knowledge bases. DESIGN: Observational study in which investigators visited doctors for two half days and collected their questions. Taxonomies were developed to characterise the clinical topic and generic type of information sought for each question. SETTING: Eastern Iowa. PARTICIPANTS: Random sample of 103 family doctors. MAIN OUTCOME MEASURES: Number of questions posed, pursued, and answered; topic and generic type of information sought for each question; time spent pursuing answers; information resources used. RESULTS: Participants asked a total of 1101 questions. Questions about drug prescribing, obstetrics and gynaecology, and adult infectious disease were most common and comprised 36% of all questions. The taxonomy of generic questions included 69 categories; the three most common types, comprising 24% of all questions, were "What is the cause of symptom X?" "What is the dose of drug X?" and "How should I manage disease or finding X?" Answers to most questions (702, 64%) were not immediately pursued, but, of those pursued, most (318, 80%) were answered. Doctors spent an average of less than 2 minutes pursuing an answer, and they used readily available print and human resources. Only two questions led to a formal literature search. CONCLUSIONS: Family doctors in this study did not pursue answers to most of their questions. Questions about patient care can be organised into a limited number of generic types, which could help guide the efforts of knowledge base developers.

An audit of obstetric care in a university family medicine department and an obstetrics-gynecology department.

The care of obstetric patients in a university family medicine department was compared with that in the obstetrics-gynecology department of the same university. The obstetric service patients tended to be at higher risk due to a higher black population (24.2 percent vs 6.3 percent), greater prepregnancy weight (mean 154.0 lbs vs 113.9 lbs), and a greater number of patients referred from the community because of prenatal complications. However, the family medicine patients had a higher incidence of premature rupture of membranes (26 percent vs 11 percent), and were therefore at risk for several complications. Family medicine nulliparas had first stages of labor which lasted an average of 12.2 hours as opposed to obstetric service nulliparas whose first stages averaged only 9.2 hours. There were more family medicine than obstetric service patients who received no anesthesia (18.0 percent vs 10.2 percent). Elective low forceps were used more often by obstetric service physicians than by family physicians (28.2 percent vs 15.3 percent). Mothers on the family medicine service had more puerperal complications than those on the obstetric service (16.0 percent vs 5.6 percent). No serious discrepancies in quality of care could be found between the two services.

Legal implications of birth videos.

There is little information available in the peer-reviewed literature on the medical and legal aspects of videotaping obstetric procedures. To manage legal risks, some large medical centers do not allow families to videotape the birth. One liability insurer is now attempting to limit video cameras in labor and delivery suites throughout its state. These policies can have significant implications for physicians and their patients. In an effort to examine approaches to the problem, we gathered the experiences of physician and attorney members of the American College of Legal Medicine through letters and telephone conversations, and we performed a review of the available medical and legal literature. Based on this research and review, we present the benefits and risks of permitting families to videotape the birth process, and we make recommendations for reducing potential liability.

What clinical information resources are available in family physicians' offices?

BACKGROUND: When faced with questions about patient care, family physicians usually turn to books in their personal libraries for the answers. The resources in these libraries have not been adequately characterized. METHODS: We recorded the titles of all medical books in the personal libraries of 103 randomly selected family physicians in eastern Iowa. We also noted all clinical information that was posted on walls, bulletin boards, refrigerators, and so forth. Participants were asked to describe their use of other resources such as computers, MEDLINE, reprint files, and "peripheral brains" (personal notebooks of clinical information). For each physician, we recorded how often the resources were used to answer clinical questions during 2 half-day observation periods. RESULTS: The 103 participants owned a total of 5794 medical books, with 2836 different titles. Each physician kept an average of 56 books in the office. Prescribing references (especially the Physicians' Desk Reference) were most common (owned by 100% of the participants), followed by books on general internal medicine (99%), adult infectious disease (89%), and general pediatrics (83%). Books used to answer clinical questions were more likely to be up to date (copyright date within 5 years) than unused books (74% vs 27%, P <.001). Items posted on walls included drug dosage charts and pediatric immunization schedules. Only 26% of the physicians had computers in their offices. CONCLUSIONS: Drug-prescribing textbooks were the most common type of book in family physicians' offices, followed by books on general internal medicine and adult infectious diseases. Although many books were relatively old, those used to answer clinical questions were generally current.

The information needs of family physicians: case-specific clinical questions.

BACKGROUND: Physicians must be able to rapidly obtain information that answers specific patient-related clinical questions. This study describes the information-seeking process in the office practices of family physicians. METHODS: We observed and recorded the information-seeking and information-obtaining behavior of 30 family physicians in their offices. RESULTS: Based on 172 hours of observation and 602 patient visits, family physicians sought answers to an average of only one clinical question for every 15 patients seen. Urban physicians sought answers to more questions than rural physicians (one question for every 9 patients, as compared with one question for every 24 patients; P less than .05). The frequency of seeking information was not related to the physician's age. Busier physicians (those seeing more patients per hour) tended to ask fewer questions (correlation coefficient (r) = -.34, P = .06). Drug-prescribing questions were the most common type; second most common were orthopedic questions. Colleagues and the Physicians' Desk Reference were the most often used resources. Eight percent of the questions were not answered. CONCLUSIONS: Among family physicians, patient-related questions are infrequently asked and highly specific. Most questions are rapidly answered using colleagues and books, not journals or computers.

Neonatal respiratory distress in the community hospital: when to transport, when to keep.

Neonatal respiratory distress is a common problem confronting family physicians. Although respiratory distress may represent a benign, self-limited process, it may also be the first sign of sepsis or serious cardiopulmonary disease. Because it is crucial to differentiate the two, we offer a practical approach to the treatment of neonatal respiratory distress at community hospitals. Our method, the Rule of 2 Hours, is based on readily accessible clinical findings. We believe it will help physicians detect babies at risk for serious problems, but will not result in unnecessary referral of neonates that are simply adapting to extrauterine life.

Benign fever following vaginal delivery.

BACKGROUND: In patients with fever following vaginal delivery, physicians must differentiate benign self-resolving fevers from fevers with more serious causes, especially endometritis. To help differentiate these clinical entities, we explored the characteristics and risk factors for benign "single-day" postpartum fever. METHODS: We conducted a retrospective cohort study of 2137 vaginal deliveries. Patients were randomly selected from the 25,687 vaginal deliveries that took place between 1979 and 1992 at The University of Iowa Hospitals and Clinics. The data were analyzed using odds ratios and multiple logistic regression. RESULTS: Benign fevers occurred in 3.3% of patients, while endometritis was diagnosed in 1.6%. After controlling for confounding variables, two clinical factors were independently associated with single-day fever: primiparity (odds ratio [OR], 3.4; 95% confidence interval [CI], 2.0 to 5.7) and use of a uterine pressure catheter (OR, 2.4; 95% CI, 1.5 to 3.7). These factors were not associated with endometritis. The first postpartum temperature elevation ( > or = 38.0 degrees C) occurred earlier in patients with single-day fever than in patients with endometritis (4.0 +/- 4.6 hours postpartum vs 30.2 +/- 27.0 hours postpartum, P < .001). The maximum temperature elevation was lower, on average, in patients with single-day fever than in patients with endometritis (38.2 degrees +/- 0.2 degrees C vs 38.9 degrees +/- 0.6 degrees C, P < .001). CONCLUSIONS: Single-day fever was more likely to occur in primiparous women and in women who were monitored with a uterine pressure catheter. Most women with benign single-day fevers had early low-grade fevers, whereas women with endometritis had higher fevers that occurred later in the postpartum period.