Critically Appraised Topic
From MindLinc Wiki
Contents |
Definition
Critically Appraised Topics are brief critiques that use evidence based medicine principles to evaluate journal articles. They are also called Quests, for QUestion, Evaluate, Synthesize.
To review Duke Residency CATs, click here.
Duke Residency Format for Writing a Critically Appraised Topic
Body
Citation:
Clinical Question:
Clinical Case or Background Info: BRIEF- no more than 5 lines
Question Type: pick one: Therapy, Diagnosis, Harm, Prognosis, Meta-Analysis/Systematic Review, Qualitative, Other
Validity Criteria for Appropriate Question Type: (see A below)
Study Design Type: pick one: Randomized Controlled Trial, Case Control, Cohort, Meta-Analysis/Systematic Review, Cross-Sectional, Qualitative, Case Report, Other
Relevant Criteria for Appropriate Design: (see B below)
Main Results: Try to limit this to no more than 10 lines. To do so, select the 1 or 2 main results relevant to your clinical question and report or calculate (if possible) the best summary statistic for these main results. If available, number-needed-to-treat or effect size is preferred to absolute risk reduction, which is preferred to an odds ratio, which is preferred to a relative risk reduction, which is preferred to other descriptive statistics or raw data. You do not need to report all results; instead, try to answer the question: What would you want to tell a colleague if they asked you, “So what did this study show, anyway?” You can include no more than 2 tables or figures IF you think they are very important. Do NOT copy all of the figures and tables in the paper. Strongly consider making your own summary table if the article’s tables are confusing. Finally, consider stating the meaning of the results in a plain English sentence which patients and families could understand.
To help with some of the harder study designs…if the study is about:
- Diagnostic tests: you might include the sensitivity or specificity reported, positive and negative likelihood ratios (if these are not presented, try to calculate them), positive predictive value/negative predictive value, adverse outcomes
- Harm: you might include the strength of association between exposure and outcome, risk estimates for the exposure and the precision of these estimates, the absolute level of risk from the exposure, RRR, OR
- Prognosis: you might include RRR, OR, Kaplan Meier Curves, likelihood of outcomes in specified time period and precision of these estimates.
Conclusions: try to limit this to no more than 10 lines. Focus your comments on the clinical application or impact of your appraisal. For example, given the strengths and weaknesses of the results, what inferences are fair to make, how confident are you that the data are conclusive, what action(s) should be taken, and how do you balance risks/benefits. Most importantly, how will this study change your practice of medicine?
Synopsis: one or two sentences summarizing your CAT, including the type of trial, duration of trial, number of patients, intervention, results, and threats to validity. If you think you can’t put all that in 1-2 sentences, see C for examples. This should NOT be your conclusions section.
A. Validity Criteria for Appropriate Study Type
(also available in worksheet form on Duke EBM webpage or in Users Guide to the Medical Literature). Thinking about the study overall, report how valid you think it is based on the answers to these questions. Give your reasoning, not yes/no answers to each question. Summarize your critique of the relevant validity criteria AND your other observations (i.e. comments about adequate sample size, other sources of bias such as industry sponsorship, other study limitations or weaknesses, etc…) to describe the overall validity of the article.
Therapy: Follow-up: Was follow-up complete? Were all patients who entered the trial properly accounted for and attributed at its conclusion? Randomization: Were pts randomized? Was randomization concealed? Intention to treat: Were pts analyzed in the groups to which they were randomized? Were all randomized patient data analyzed? Similar groups: Were pts in treatment and control groups similar with respect to known prognostic factors? Blinding: Were pts aware of group allocation? Were clinicians aware of group allocation? Were assessors aware of group allocation? Equal treatment: Aside from the intervention, were groups treated equally?
Harm: Did investigators demonstrate similarity in possible determinants of outcome; did they adjust for differences in the analysis? Were groups treated similarly outside of the putative exposure? Were the outcomes measured in the same way in the groups being compared? Was follow-up sufficiently complete? Is there a reasonable temporal relationship between exposure and outcome? Is there a dose-response gradient?
Differential Diagnosis: Did the investigators enroll the right patients? Was the patient sample representative of those with the clinical problem? Was the definitive diagnostic standard appropriate? Was the diagnostic process credible? For initially undiagnosed patients, was follow-up sufficiently long and complete?
Diagnostic Tests: Did the clinicians face diagnostic uncertainty? Was the tested population representative of that for which the diagnostic test will be used in practice? Was there a blind comparison between the test and an independent gold standard (briefly describe test and gold standard if needed)? Was there blinding of those performing and those undergoing the tests? Did all the pts receive both the tests (test being evaluated and gold standard) or was the decision to perform one test influenced by the results of the other? Was follow-up complete?
Prognosis: Was the sample of patients representative of clinical practice? Were the patients sufficiently homogenous with respect to prognostic risk? Were patients at a similar point in disease progression or severity? Was follow-up sufficiently complete? Were all patients treated similarly? Were all patients accounted for? Were objective and unbiased outcome criteria used?
Reviews and Meta-Analysis: Did the review explicitly address a sensible clinical question? Was the search for relevant studies detailed and exhaustive? Were relevant studies likely to be omitted? Were the primary studies of high methodological quality? Were the author’s assessments of the primary studies reproducible? How homogenous were the various study results?
Qualitative: Was the choice of patients explicit and comprehensive? Was data collection sufficiently comprehensive and detailed? Were the data analyzed appropriately and the findings corroborated adequately? Was follow-up complete?
B. Relevant criteria for Appropriate Design
Randomized Controlled Trial
- Blinding: You can omit this if you already addressed it in the Validity Criteria
- Duration: i.e. 8 week study, one year study. This is NOT the same as Follow-up in Validity Criteria above, which addresses how complete follow up was (essentially, were all patients accounted for at the end of the trial). This is trial length, including length of follow-up.
- Setting: Location and type of setting (academic, primary care, community hospital, etc)
- Patients/Population: Report total n, very brief description of the population (i.e., outpatient clinic pop in New Hampshire, mostly white, 60% men). You may also want to consider: were inclusion and exclusion criteria appropriate (you DO NOT need to list all inclusion and exclusion criteria unless you think this is particularly relevant), what were the screening or enrollment methods if relevant, what was the number screened vs. number enrolled if relevant.
- Intervention/exposure:
- Outcomes: Do not describe actual results, just which outcomes were evaluated! State whether the outcomes were determined a priori.
Systematic Review/Meta-analysis
- Data Sources: Peer reviewed journal articles, books, non-peer reviewed articles, government documents, pharmaceutical company studies, etc.
- Study Selection Criteria: Describe criteria used to decide which studies to include. You may want to describe the search strategy used.
- Outcomes: Do not describe actual results, just which outcomes were evaluated! State whether the outcomes were determined a priori.
Cohort or Case Control
- Setting: Location and type of setting (academic, primary care, community hospital, etc)
- Population/Patients: Report total n, very brief description of the population (i.e., outpatient clinic pop in New Hampshire, mostly white, 60% men). You may also want to consider: were inclusion and exclusion criteria appropriate (you DO NOT need to list all inclusion and exclusion criteria unless you think this is particularly relevant), what were the screening or enrollment methods if relevant, what was the number screened vs. number enrolled if relevant.
- Prognostic factors, Risk factors, and/or Exposures:
- Outcomes: Do not describe actual results, just which outcomes were evaluated! State whether the outcomes were determined a priori.
- Follow-up Period: This is NOT the same as Follow-up in Validity Criteria above, which addresses how complete follow up was (essentially, were all patients accounted for at the end of the trial). This is the period of time over which cases were followed. In cohort studies, this may be years. These studies can also follow patients backward in time, and that time still counts. Be careful not to confuse this with how long it took to assemble data for the study. For example, it might take 2 years for researchers to recruit the total number of participants or to review all the case files. Then they might follow the cohort for 12 years, or they might review cases back for 12 years. The follow-up would be 12 years in both situations.
Cross-sectional or Qualitative
- Setting: Location and type of setting (academic, primary care, community hospital, etc)
- Population: Report total n, very brief description of the population (i.e., outpatient clinic pop in New Hampshire, mostly white, 60% men). You may also want to consider: were inclusion and exclusion criteria appropriate (you DO NOT need to list all inclusion and exclusion criteria unless you think this is particularly relevant), what were the screening or enrollment methods if relevant, what was the number screened vs. number enrolled if relevant.
- Assessment: Describe how participants were evaluated and what tools were used to conduct evaluations.
- Outcomes: Do not describe actual results, just which outcomes were evaluated! State whether the outcomes were determined a priori.
C. Examples of Article synopses
Randomized Controlled Trials
In this 24 week RCT of 166 pts receiving buprenorphine-naloxone for opiate dependence, the efficacy of brief weekly counseling and once-weekly medication dispensing did not differ significantly from extended weekly counseling and thrice-weekly medication dispensing. However, because of the small sample size and the number of drop outs the study was underpowered to detect small differences.
In this randomized controlled clinical trial lasting 1 year, relapse rates for 93 women who met DSM-IV criteria for anorexia nervosa and who had completed a weight restoration program were not significantly different between those receiving fluoxetine (57% rate of relapse) and those receiving placebo (55%). This study met all appropriate validity criteria.
In this double-blind randomized controlled trial of 295 adult out-patients with generalized social phobia, all active treatments (fluoxetine alone, CCBT alone, fluoxetine plus CCBT, and placebo plus CCBT) were found to be superior to placebo after fourteen weeks, and all had similar efficacies.
In this multi-center RCT evaluating divalproex for treatment of aggression in 246 outpatients with either Cluster B personality disorders, intermittent explosive disorder, or PTSD, divalproex improved scores on aggression scales for those with Cluster B personality disorders but no treatment effect was observed for intermittent explosive disorder, PTSD, or across the entire combined population. High drop-out rates impaired study validity.
In this 24 week RCT of elderly patients (70yo or greater), paroxetine was effective in the treatment of Major Depression as well as prevention of recurrence. However, adding interpersonal therapy to paroxetine therapy did not appear to confer added treatment efficacy. In this placebo-controlled pilot study of 15 patients with atypical depression, pts treated with chromium picolinate 600 mcg had a significantly higher response rate (7/10) as compared to placebo (0/5), with minimal side-effects. This study is limited by a small sample size, which may have not been sufficient to balance baseline characteristics; nonetheless, this study is promising, and chromium picolinate may be an affordable, well-tolerated alternative in treating atypical depression. In this RCT evaluating the effects of problem-solving skills training (PSST) and parent management training (PMT) in children (age 7-13) with antisocial behavior, the combined treatment lead to a greater proportion with range of "nonclinical" functioning than either alone. NNT was 3, which appears clinically significant; however major limitations include failure to include drop-outs in analysis, as well as other typical limitations in behavior studies: self-report phenomenon, selection bias, regression to mean, lack of placebo, lack of long-term outcome
Cohort/Case Control Studies
In this retrospective cohort study of 147 patients with a discharge diagnosis of alcohol dependence, a history of DTs and tachycardia at admission were the two variables significantly associated with development of DTs during hospitalization (positive likelihood ratio for the presence of both predictors = 2; negative likelihood ratio = 0.5). Validity concerns include the non-representative nature of study participants compared to Duke Hospital patients, the lack of information on whether patients were at a similar point in disease progression and severity, and the lack of randomization present in any cohort study.
In this prospective, 3 year follow-up study of 54 patients with co-occurring bipolar and substance use disorders in the New Hampshire public mental health system, patients showed significant improvements in remission from substance abuse, independent living, employment, and subjective quality of life after receiving either of two forms of integrated dual disorders treatment. “Recovery” from substance abuse was noted to be a multi-dimensional concept and recovery in one domain did not necessarily transfer to other domains.
This community-based, multi wave, prospective cohort study of a randomly sampled rural elderly population shows that depressive symptoms associated with Alzheimer's disease and overall dementia are prodromal or early manifestations of the dementia, rather than predictors of the subsequent development of dementia.
Reviews
Despite the potential sources of bias implicit in the types of study included, this systematic literature review of 19 studies reporting in utero exposure to SSRIs found that signs across the majority of studies were similar and included tremors, jitteriness, shivering, increased muscle tone, feeding or digestive disturbances, irritability or agitation, and respiratory distress. Symptoms were generally mild and resolved with supportive care.
In this unsystematic review article, the authors compare the diagnostic advantages and disadvantages of the Wender Utah Criteria, the DSM-IV criteria, and various laboratory assessment strategies in the diagnosis of adult ADHD and conclude that adult ADHD should remain a clinical diagnosis and that much broader areas of functioning need to be assessed when evaluating adults for this disorder.
This review of randomized, double-blind, placebo-controlled trials finds evidence that treatment with donepezil in people with Alzheimer’s disease, for 12, 24, or 52 weeks, is associated with improvements in cognitive function and clinician-rated global clinical state.
Cross-sectional Studies
This period prevalence study looks at the 1-year prevalence of atypical antipsychotic use by age and sex among 6 million individuals in the US with private insurance. Twenty percent of all prescriptions for atypicals were for pts 19 y/o or younger, prescription claims for atypicals in children were more than twice as high for males, and over 25% of male pts with at least one claim for atypicals were <9y/o.
