Biomedical subjects
Publications and source records attributed to G H Guyatt.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Explore the source record for details and available documents.
Clinicians can often find treatment recommendations in traditional narrative reviews and the discussion sections of original articles and meta-analyses. Making a treatment recommendation involves framing a question, identifying management options and outcomes, collecting and summarizing evidence, and applying value judgments or preferences to arrive at an optimal course of action. Each step in this process can be conducted systematically (thus protecting against bias) or unsystematically (leaving the process open to bias). Clinicians faced with a plethora of recommendations may wish to attend to those that are less likely to be biased. Therefore, we propose a hierarchy of rigor of recommendations to guide clinicians when judging the usefulness of particular recommendations. Recommendations with the highest rigor consider all relevant options and outcomes, include a comprehensive collection of the methodologically highest quality data with an explicit strategy for summarizing the data (that is, a systematic review), and make an explicit statement of the values or preferences involved in moving from evidence to action. High rigor recommendations come from systematically developed, evidence-based practice guidelines or rigorously conducted decision analyses. Systematic reviews, which typically do not consider all relevant options and outcomes or make the preferences underlying recommendations explicit, offer intermediate rigor recommendations. Traditional approaches in which the collection and assessment of evidence remains unsystematic, all relevant options and outcomes may not be considered, and values remain implicit, provide recommendations of weak rigor. In an era in which clinicians are barraged by recommendations as to how to manage their patients, this hierarchy provides a potentially useful set of guides.
BACKGROUND: We know little about how physicians assess perioperative cardiac risk in patients undergoing noncardiac surgery. OBJECTIVES: To evaluate preoperative medical consultations and determine the extent to which consultants used validated cardiac risk indices and specialized noninvasive cardiac tests, and to assess agreement between physician ratings of cardiac risk (low, moderate, or high) and risk estimates derived using validated cardiac risk indices or, in the case of vascular surgery, a risk index. METHODS: This observational study was conducted at 5 Canadian teaching hospitals affiliated with 2 universities. We retrospectively evaluated 308 preoperative consultations performed in 297 patients and examined the frequency with which consultants recorded the use of validated cardiac risk indices. We used K statistics to quantify the extent to which physician ratings of cardiac risk agreed with risk estimates derived using validated cardiac risk indices. RESULTS: Physicians recorded use of a risk index in 31% of the consultations, but the index used was almost always the suboptimal classification of the American Society of Anesthesiologists. The agreement between physician estimates of cardiac risk and the validated cardiac risk indices was only fair, with a weighted K of 0.38 (95% confidence interval, 0.28-0.49). Overestimation and underestimation of cardiac risk occurred in 16% and 13% of the consultations, respectively. Consultants did not order dipyridamole thallium imaging or dobutamine stress echocardiography for any moderate-risk patients undergoing vascular surgery. CONCLUSIONS: Physicians underuse validated cardiac risk indices, and the agreement between the cardiac risk estimates and risk as determined by validated cardiac indices is suboptimal. Physicians are also underusing dipyridamole thallium imaging and dobutamine stress echocardiography for moderate-risk patients undergoing vascular surgery.
Explore the source record for details and available documents.
OBJECTIVES: To evaluate the efficacy of isoniazid for the prevention of tuberculosis in tuberculin skin test-positive and negative individuals with HIV infection. DESIGN: Meta-analysis of randomized controlled trials. SETTING: Seven trials from Mexico, Haiti, the United States, Zambia, Uganda and Kenya. PATIENTS: Individuals free from tuberculosis, 2367 persons in the intervention and 2162 in the control groups. INTERVENTION: Comparison of isoniazid with placebo or no prophylaxis. METHODS: A systematic search of the literature was carried out from 1985 to October 1997 for randomized controlled trials of isoniazid prophylaxis in HIV-infected persons. Two reviewers evaluated the relevance of each candidate study and the validity of eligible trials. Studies were pooled using a random effect model, conducting secondary analyses for tuberculin skin test-positive and negative persons. RESULTS: Mean follow-up in trials varied between 0.4 and 3.2 years. Pooling all seven trials, a risk ratio was found for persons treated with isoniazid for developing tuberculosis of 0.58 [95% confidence interval (CI), 0.43-0.80] and 0.94 (95% CI, 0.83-1.07) for death. In groups of tuberculin skin test-positive and negative persons, the risk ratio of tuberculosis was 0.40 (95% CI, 0.24-0.65) and 0.84 (95% CI, 0.54-1.30), respectively, and the difference in the effectiveness of isoniazid versus placebo between these groups was statistically significant (P = 0.03, for the difference of summary estimates). Consistency of results was found across trials (P > 0.10, heterogeneity value) for all comparisons. CONCLUSIONS: Prophylaxis with isoniazid reduces the risk of tuberculosis in persons with HIV infection. The effect is restricted to tuberculin skin test-positive persons.
The objective of the study was to evaluate a shortened osteoporosis quality of life questionnaire (OQLQ) in osteoporotic women with back pain due to vertebral fractures. From the longer 30-item OQLQ (four to nine items per domain) we created the mini-OQLQ by choosing the two items with the highest impact in each of five domains (symptoms, physical function, activities of daily living, emotional function, leisure). We administered the OQLQ, the Sickness Impact Profile, the SF-36 and the Brief Pain Index to patients at baseline, after 2 weeks and after 6 months. The intraclass correlations between baseline and the 2-week follow-up for the five mini-OQLQ domains ranged from 0.72 to 0.86. Cross-sectional correlations between the domains of the mini-OQLQ and other health instruments were moderate to large (0.35-0.80) and greater than predicted. The mini-OQLQ items showed moderate to large correlations with items omitted from the shortened questionnaire (0. 44-0.88). Correlations between the OQLQ domains and the other three instruments were greater than those of the mini-OQLQ, and partial correlations between OQLQ items omitted from the mini-OQLQ and the other three instruments after considering mini-OQLQ items were substantial (0.19-0.71) and statistically significant. Sample sizes of less than 200 per group should be required to detect minimally important differences in parallel-group clinical trials. Longitudinal correlations between the mini-OQLQ and the other measures were often significant but generally lower than predicted (0.10-0.49). The partial correlations revealed that the omitted items explained a significant portion of the longitudinal variance in each domain. We conclude that in a selected group of patients with back pain caused by vertebral fractures, the mini-OQLQ demonstrated good discriminative and adequate evaluative properties. The mini-questionnaire should be useful in clinical settings.
BACKGROUND: In patients with apparently operable non-small cell lung cancer (NSCLC), clinicians often omit investigation for M disease in asymptomatic patients. Previous investigations have not specified in detail what is meant by "symptomatic," and this could differ between surgeons. We have investigated the extent to which surgeons' criteria differ for presence of symptoms. METHODS: Participating surgeons from seven centers, enrolled patients they judged "asymptomatic" in a randomized trial of investigational strategies for NSCLC. Patients completed a structured questionnaire describing symptoms of the central nervous system (CNS). In 685 patients, we documented CNS symptom recurrence after resectional surgery over 1 year of follow-up. RESULTS: Two centers enrolled only patients without even the mildest symptoms. Three centers took an intermediate approach, occasionally classifying patients with mild symptoms as "asymptomatic" and thus enrolling them in the trial. Two centers classified an appreciable number of patients with minimal symptoms, and occasionally with more than minimal symptoms, as "asymptomatic." Patients with even mild CNS symptoms were more likely to subsequently present with CNS metastases. CONCLUSIONS: Thoracic surgeons differ in their ideas of what may constitute the symptoms of M disease. Patients with structured questionnaire results that suggest symptoms of CNS disease are more likely to have CNS symptom recurrence after resectional surgery.
The purpose of this study was to compare the performance of measures of health-related quality of life in a randomized controlled trial of respiratory rehabilitation versus conventional community care for patients with chronic airflow limitation. The study included 89 stable patients with moderate to severe chronic airflow limitation with measurement of health status at 12, 18, and 24 weeks. Outcomes included two disease-specific (the Oxygen Cost Diagram and the Chronic Respiratory Questionnaire [CRQ]) measures, a generic health profile (the Sickness Impact Profile [SIP]), and two utility measures (the Standard Gamble and the Quality of Well-Being index [QWB]). Of the measures, only the four domains of the CRQ (dyspnea, fatigue, mastery, and emotional function) showed statistically significant differences (P < or = 0.05) between treatment and control groups. Correlation between change in the CRQ and change in other relevant measures, including the 6-minute walk test and global ratings of change in dyspnea, fatigue, and emotional function were generally weak to moderate (from 0.19 to 0.51). All correlations between change in the QWB, SIP, and Standard Gamble and other measures were very weak or weak (up to 0.30). Correlation between change in the three generic measures were all very weak (<0.15). The results suggest that unless investigators include responsive and valid disease-specific measures of health-related quality of life in controlled trials in chronic diseases, they risk misleading conclusions about the effect of treatments on health status.
The objective of this study was to compare the impact of closed- versus open-ended question formats on the completeness and accuracy of demographic data collected in a mailed survey questionnaire. We surveyed general internists in five Canadian provinces to determine their career satisfaction. We randomized respondents to receive versions of the questionnaire in which 16 demographic questions were presented in a closed-ended or open-ended format. Two questions required respondents to make a relatively simple computation (ensuring that three or four categories of response added to 100%). The response rate was 1007/1192 physicians (80.0%). The proportion of respondents with no missing data for all 16 questions was 44.7% for open-ended and 67.0% for closed-ended formats (P < 0.001). The odds of having missing items remained higher for open-ended response options after adjusting for a number of respondent characteristics (2.67, 95% confidence interval 2.01 to 3.55). For the two questions requiring computations focused on professional activity and income, there were more missing data (P = 0.02, 0.02, respectively) but fewer inaccurate responses (P = 0.009, 0.20, respectively) for the open-ended compared to the closed-ended format. Investigators can achieve higher response rates for demographic items using closed format response options, but at the risk of increasing inaccuracy in response to questions requiring computation.
We updated our previous systematic review of the effect of supplemental calcium on blood pressure. We extended our previous searches on MEDLINE and EMBASE to May 1997 and examined citations from relevant articles. We contacted the authors of eligible trials to ensure the accuracy and completeness of data, and to identify unpublished trials. We included any study in which investigators randomized hypertensive or normotensive people to calcium supplementation or alternative therapy and measured blood pressure for at least 2 weeks. In addition to 32 trials included in the prior metaanalysis, 10 new trials contributed to this metaanalysis. Two pairs of independent reviewers abstracted data and assessed the validity of the study data according to six quality criteria. We calculated the differences in blood pressure change between the calcium supplementation and control groups and pooled the estimates with each trial weighted with the inverse of the variance using a random effects model. The predictors of blood pressure reduction that we examined included method of supplementation, baseline blood pressure, and the methodologic quality of the studies. The pooled analysis shows a reduction in systolic blood pressure of -1.44 mm Hg (95% confidence interval -2.20 to -0.68; P < .001) and in diastolic blood pressure of -0.84 mm Hg (95% confidence interval -1.44 to -0.24; P < .001). We found statistically significant heterogeneity of results across trials (P < or = .02), which persisted when we looked at the nondietary trials alone, but not when we restricted our analysis to dietary trials. Although there was a trend toward larger effects with dietary interventions, none of the possible mediators of blood pressure reduction explained differences in treatment effect. We conclude that calcium supplementation leads to a small reduction in systolic and diastolic blood pressure. The effect of supplemental calcium in the diet is at least as great as nondietary supplementation.
The 32-item Asthma Quality of Life Questionnaire (AQLQ) has shown good responsiveness, reliability and construct validity; properties that are essential for use in clinical trials, clinical practice and surveys. However, to meet the needs of large clinical trials and long-term monitoring, where efficiency may take precedent over precision of measurement, the 15-item self-administered MiniAQLQ has been developed. The MiniAQLQ was tested in a 9-week observational study of 40 adults with symptomatic asthma. Patients completed the MiniAQLQ, the AQLQ, the Short Form (SF)-36, the Asthma Control Questionnaire and spirometry at baseline, 1, 5 and 9 weeks. In patients whose asthma was stable between clinic visits, reliability was very acceptable for the MiniAQLQ (intraclass correlation coefficient (ICC)=0.83), but not quite as good as for the AQLQ (ICC=0.95). Similarly, responsiveness in the MiniAQLQ (p=0.0007) was good but not quite so good as for the AQLQ (p<0.0001). Construct validity (correlation with other indices of health status) was strong for both the MiniAQLQ and the AQLQ. Criterion validity showed that there was no bias between the instruments (p=0.61) and the correlation between them was high (r=0.90). The Mini Asthma Quality of Life Questionnaire has good measurement properties but they are not quite as strong as those of the original Asthma Quality of Life Questionnaire. The choice of questionnaire should depend on the task at hand.
OBJECTIVE: To compare results of a specific capacity assessment administered by the treating clinician, and a Standardized Mini-Mental Status Examination (SMMSE), with the results of expert assessments of patient capacity to consent to treatment. DESIGN: Cross-sectional study with independent comparison to expert capacity assessments. SETTING: Inpatient medical wards at an academic secondary and tertiary referral hospital. PARTICIPANTS: One hundred consecutive inpatients facing a decision about a major medical treatment or an invasive medical procedure. Participants either were refusing treatment, or were accepting treatment but were not clearly capable according to the treating clinician. MEASUREMENTS AND MAIN RESULTS: The treating clinician (medical resident or student) conducted a specific capacity assessment on each participant, using a decisional aid called the Aid to Capacity Evaluation. A specific capacity assessment is a semistructured evaluation of the participant's ability to understand relevant information and appreciate reasonably foreseeable consequences with regard to the specific treatment decision. Participants also received a SMMSE administered by a research nurse. Participants then had two independent expert assessments of capacity. If the two expert assessments disagreed, then an independent adjudication panel resolved the disagreement after reviewing videotapes of both expert assessments. Using the two expert assessments and the adjudication panel as the reference standard, we calculated areas under the receiver-operating characteristic curves and likelihood ratios. The areas under the receiver-operating characteristic curves were 0.90 for specific capacity assessment by treating clinician and 0.93 for SMMSE score (2p =.48). For the treating clinician's specific capacity assessment, likelihood ratios for detecting incapacity were as follows: definitely incapable, 20 (95% confidence interval [CI] 3. 6, 120); probably incapable, 6.1 (95% CI 2.6, 15); probably capable, 0.39 (95% CI 0.18, 0.81); and definitely capable, 0.05 (95% CI 0.01, 0.29). For the SMMSE, a score of 0 to 16 had a likelihood ratio of 15 (95% CI 5.3, 44), a score of 17 to 23 had a likelihood ratio of 0. 68 (95% CI 0.35, 1.2), and a score of 24 to 30 had a likelihood ratio of 0.05 (95% CI 0.01, 0.26). CONCLUSIONS: Specific capacity assessments by the treating clinician and SMMSE scores agree closely with results of expert assessments of capacity. Clinicians can use these practical, flexible, and evaluated measures as the initial step in the assessment of patient capacity to consent to treatment.
PURPOSE: To determine whether framing questions positively or negatively influences residents' apparent satisfaction with their training. METHOD: In 1993-94, 276 residents at five Canadian internal medicine residency programs responded to 53 Likert-scale items designed to determine sources of the residents' satisfaction and stress. Two versions of the questionnaire were randomly distributed: one in which half the items were stated positively and the other half negatively, the other version in which the items were stated in the opposite way. RESULTS: The residents scored 43 of the 53 items higher when stated positively and scored ten higher when stated negatively (p < .0001). When analyzed using an analysis-of-variance model, the effect of positive versus negative framing was highly significant (F = 129.81, p < .0001). While the interaction between item and framing was also significant, the effect was much less strong (F = 5.56, p < .0001). On a scale where 1 represented the lowest possible level of satisfaction and 7 the highest, the mean score of the positively stated items was 4.1 and that of the negatively stated items, 3.8, an effect of 0.3. CONCLUSIONS: These results suggest a significant "response acquiescence bias." To minimize this bias, questionnaires assessing attitudes toward educational programs should include a mix of positively and negatively stated items.
PURPOSE: Periodic diagnostic tests and continuous and intermittent monitoring are integral to critical care medicine. The focus of this article is understanding the impact of existing diagnostic technology, as well as that of new diagnostic technology. DATA SYNTHESIS: We use literature about gastric tonometry to illustrate eight steps for assessing the value of diagnostic technology. METHODS: These steps focus on how the technology works in the laboratory, its range of uses and diagnostic accuracy, its impact on healthcare workers, the decision making process, and patient outcomes, as well as issues of access, cost, and application in your own setting. CONCLUSIONS: Awareness of the scope and quality of research evaluating new and existing diagnostic technology is central to modern critical care practice.
BACKGROUND: Persisting controversy surrounds the use of transthoracic needle aspiration biopsy (TNAB) stemming from its uncertain diagnostic accuracy. A systematic review and meta-analysis was therefore conducted to evaluate the accuracy of TNAB for the diagnosis of solitary or multiple localised pulmonary lesions. METHODS: Searches for English literature papers in Index Medicus (1963-1965) and Medline (1966-1996) were performed and the bibliographies of the retrieved articles were systematically reviewed. Articles evaluating the accuracy of TNAB in series of consecutive patients presenting with solitary or multiple pulmonary lesions were considered. Only papers in which >/=90% of patients were given a final diagnosis according to an appropriate reference standard were included in the meta-analysis. RESULTS: A total of 48 studies were included and five meta-analyses were conducted according to four diagnostic thresholds. From the pooled sensitivity and specificity corresponding to each diagnostic threshold, associated likelihood ratios (LRs) were derived for malignant disease as follows: (1) malignant versus all other categories, LR = 72; (2) malignant or suspicious versus all others, LR = 49; (3) suspicious versus all categories but malignant, LR = 15; (4) benign versus all others, LR = 0.07; and (5) specific benign diagnosis versus all others, LR = 0.005. Differences in methodological quality of the studies, needle types, or whether a cytopathologist participated in the procedure failed to explain the heterogeneity of the results found in almost every meta-analysis. Given a 50% probability of malignancy prior to the TNAB, post-test probabilities of malignancy upon receiving the results would be malignant, 99%; suspicious, 94%; non-specific benign, 7%; and benign with a specific diagnosis, 0.6%. CONCLUSIONS: Given the intermediate pre-test probabilities that would probably lead to performing TNAB, findings of "malignant" or of a specific diagnosis of a benign condition provide definitive results. Findings of "suspicious" markedly increase the probability of malignancy, and "benign" markedly decreases it but may not be considered definitive.
Meta-analyses have investigated the efficacy of cholesterol-lowering interventions in relation to the underlying risk of coronary heart disease and the extent and duration of cholesterol reduction. We systematically reviewed the efficacy of antilipidemic interventions on major mortality outcomes in relation to drug classes. We searched MEDLINE and EMBASE from 1966 through October 1996 for randomized, controlled trials of any cholesterol-lowering interventions reporting mortality data. We included 59 trials involving 85 431 participants in the intervention and 87 729 participants in the control groups. We pooled these trials into 7 pharmacological categories of cholesterol-lowering interventions: statins (13 trials), fibrates (12 trials), resins (8 trials), hormones (8 trials), niacin acid (2 trials), n-3 fatty acids (3 trials), and dietary interventions (16 trials). Of the cholesterol-lowering interventions, only statins showed a large and statistically significant reduction in mortality from coronary heart disease (risk ratio, 0.66; 95% confidence interval [CI], 0.54 to 0. 79) and from all causes (risk ratio, 0.75; 95% CI, 0.65 to 0.86). For both all-cause and cardiovascular mortality, the difference between statins and the combined estimate of the other classes of agents was unlikely to be due to chance (P<0.02 for both comparisons). Meta-regression analysis demonstrated that variability in results across trials could be largely explained on the basis of differences in the magnitude of cholesterol reduction. Statins have the largest effect on the reduction of cardiovascular and all-cause mortality, and this result recommends their use in preference to other antilipidemic agents. The greater effect of statins is likely due to the larger reduction in cholesterol.