Presentation on theme: "Pearls and Pitfalls of Medical Publishing Stephen L. Corson, M.D. Editor-in-Chief The Journal of Minimally Invasive Gynecology The Official Journal of."— Presentation transcript:
Pearls and Pitfalls of Medical Publishing Stephen L. Corson, M.D. Editor-in-Chief The Journal of Minimally Invasive Gynecology The Official Journal of the AAGL Advancing Minimally Invasive Gynecology Worldwide Published in partnership with Elsevier Inc.
Considerations Have something meaningful to say IRB approval if applicable Consult a statistician first!!!!! If RCT, power analysis based on assumptions, necessary numbers in each arm, is protocol valid? Adequate literature search
Manuscript Preparation Follow instructions on website Senior author must sign attestation form All authors need to complete disclosure Where appropriate supply abstract, precis and key words Separate legend sheet for figures; titles for tables Consider a copy editor with English skills
Case Reports Must be truly unique—pathology, diagnostic methods, management, complications. Is the information clinically useful? Limit on number of authors and references
Instruments and Techniques Really new or a minor modification? Are claims of increased efficacy/safety supported by data (statistically, if compared)? Photo documentation and/or video is preferred; limit of 4.5 minutes for video. Industry Support? Full disclosure
Pilot Study Deals with new instruments, diagnostic methods, new procedures with small patient numbers. Statistics not an issue No claims of safety or efficacy can be made.
Review Article Data bases used; Medline, Cochrane Review Balanced approach to literature review Unbiased presentation as much as possible Does it fully encompass the current and older literature?
Type of Study Design Case control: Usually retrospective, matched. Cohort: prospective, not randomized, parallel Crossover: two groups or one group with placebo Randomized Control Trial (RCT) is the gold standard of investigations
Canadian Task Force Classification of Level of Evidence I Properly designed randomized control trial II-1 Properly designed controlled, not randomized trial II-2 Well designed cohort case controlled study, usually multicentered, prospective or retrospective II-3 Comparisons between times/places; uncontrolled Opinions can be published in “Clinical Opinions” Section
Original Research Papers Title must reflect an accurate description of the content. Keywords are important for citations in literature Abstract must contain study objective, design classification, setting, patients, interventions, measurements, main results and conclusions. The abstract must be clear, concise and accurate without discrepancy with the text. It is the “teaser.”
Introduction Rationale for the Study Relevant references State the question or the issue to be studied Is there a hypothesis to be tested?
Materials and Methods Study design Ethical Review Board approval if applicable Selection process; how randomized? Inclusions and specific exclusions Single or double blind? (pain/pathology/adhesion study data) Sample size: power analysis (>80%) Method of data collection Definitions of measurement norms and methods such as for EBL and histologic data
Materials and Methods (cont.) Statistical Methods If data are normally distributed, use mean, standard deviation, 95% confidence intervals. If data are skewed, use median, range, and if desired quartiles. Student’s “t” test may be used to compare means. Wilcoxon, or Mann-Whitney analysis for ordered categories such as continence/pain studies Contingency Tables (chi-square) for numbered data
Results Population demographics and experimental data Relationship between independent and dependent variables Multivariable analysis where appropriate Tables should be used to help the reader digest data, especially comparisons between groups or literature results. Do not repeat all the information given in the text. Tables need titles. Figures should elucidate. Furnish legends on a separate legend/title page. Photos must be of good quality; arrows to depict points of interest or structures are helpful. Videos demonstrating anatomy/techniques are desired (4.5 minutes maximum) Statistical significance of the results. Try to avoid use of “trends”. Adequate follow-up interval, especially for onco and urogyn
Discussion No new data from your study The first sentence summarizes the results Concise statement of findings Was the question answered or the hypothesis proved? Explain any unexpected results as best possible Compare results with the literature giving both pro and con Present limitations of the study Comment: are statistical differences relevant clinically?
Evolution of Data Reporting Case Report Pilot Study Consecutive Series (prospective) Matched case control (retrospective) Randomized Controlled Trial
Consecutive Series Useful in the early phases of study such as evaluating docking time, EBL and total operating time and hospital bed days for early experience with robotic surgery comparing first x cases with last x cases. Problems include “Intent to treat” statistics and lost to follow-up representing a skewed population (of not satisfied patients).
Crossover Study Limited usefulness because of carryover effect; mostly for pharmacologicals One group with A then B and one group with B then A (randomized). Could also be one group with different treatments of bilateral structures (tubes) Problems include patient assignment and sequencing, timing of measurements and dealing with dropouts.
Randomization Not every other patient or every other day Should be made from a computer generated list of random numbers. Important to follow “block randomization” in multicenter studies
Statistical Tidbits: Power Analysis Ideally this should precede the study The power analysis determines the sample size necessary to show either a difference or similarity between groups. It depends on the assumption of the magnitude of the difference Usually a value of >80% is needed for validity A power analysis has become almost mandatory for all RCT’s
Statistical Tidbits: “p” values Used to express the degree of dissimilarity Values show probability of results occurring by chance Type 1 error: falsely rejecting the null hypothesis Type 2 error: failure to accept the hypothesis The “p” value is not a substitute for power One sided-two sided analysis Report data as decimal without zero (.02) Usual criterion of <.05
Statistical Tidbits: Confidence Interval Shows the extremes at which the “true value” lies 95% of the time Gives information not available from “p” values The values are based on actual experimentally derived data
Evidence Based Screening Screening test 95% accurate False positive 1% Prevalence 0.5% 100 000 screened, 500 with disease and 475 (.95X500) positive tests Of 99,500 without disease, 1% positive (.01 X 99,500) = 995 false positives Therefore with 1470 positive tests only 32% have the disease.
Advanced Statistics Linear Regression Multivariable Analysis Life Table Analysis—Survival Curves Analysis of Variance (1 and 2 way) Fisher’s Exact Test, Bonferroni Correction Receiver Operating Curves Markov Tree Correlation Coefficient
Trends in Medical Publishing Reduced hard copy; increased electronic Speed and economics More video content International multicenter studies with focus on gender and ethnic differences Large central sources of journals such as medical school libraries Readers such as Kindle and IPad
Conclusions I am your advocate. My job is to help you present your data in the best possible format The editor also has a responsibility to the readers Language problems and poor writing usually can be fixed Statistical issues and fatal protocols are often unfixable
The Journal of Minimally Invasive Gynecology welcomes your submissions Submit your manuscript at www.ees.elsevier.com\jmig www.ees.elsevier.com\jmig For English-language editing services, please consult http://webshop.elsevier.com/languageediting/ Thank you for your attention.