Systematic Review and Meta Analysis Session Manuscript Writing | SRMA Session 6 — Transcript
Full transcript
- 0:00get done.
- 0:02AI to write the discussion. We can take
- 0:03help from AI to write the intro. And
- 0:05even a person who has no experience of
- 0:08research, he will just message that I
- 0:10can write intro, I can write discussion,
- 0:11I can write this.
- 0:13And of course, analysis everyone thinks
- 0:15it's hard and methodology everyone
- 0:17thinks it's it's hard of course to write
- 0:19the whole thing.
- 0:21But intro, discussion everyone thinks
- 0:23it's very very easy thing.
- 0:25But the main point is that this is the
- 0:27most important thing for a meta.
- 0:29What they want from you is to make good
- 0:33conclusions from what you have analyzed.
- 0:36And a good meta analysis is one that has
- 0:38a good discussion.
- 0:40So after the whole checking, after the
- 0:43editors are done with checking that the
- 0:45all the articles are there included in
- 0:47the screening,
- 0:49analysis has been done properly or not
- 0:51and anything like that, their main focus
- 0:53is on the discussion. How have you
- 0:55justified that your meta is important?
- 0:59Okay? So we don't ignore the step, but
- 1:01as you guys will see, many students they
- 1:04use a lot of AI and that will cause that
- 1:07causes high AI rate
- 1:10and also high plagiarism. And these two
- 1:12things again are the things that we have
- 1:14to avoid. If
- 1:16if an article is submitted to a journal
- 1:18and it is it has an AI or plagiarism
- 1:20above 20% then it is rejected on the
- 1:23spot.
- 1:24Okay? They don't even that does not even
- 1:26go to the editor. It is rejected at the
- 1:29technical stage check.
- 1:31That is done by the simple process term
- 1:34processing. Okay? Whenever an article is
- 1:36published, they first just quickly check
- 1:38AI plagiarism and if it is more than 20%
- 1:41it is rejected. Okay? And the more the
- 1:44rejection we face, the more of course we
- 1:46are losing chances of getting published
- 1:48in a
- 1:49in the journals that can easily accept
- 1:51our article. Okay? So from the start we
- 1:53have try to make our article in a way
- 1:56that it will get published. Okay? So
- 1:58let's start with the manuscript I have
- 2:00shared with you guys some points also in
- 2:03the main group
- 2:05that you guys have to
- 2:07always take in your
- 2:09take in consideration when you are
- 2:10writing the manuscript.
- 2:12Let me see, let me go through them.
- 2:14[clears throat]
- 2:15They were suggested by the our
- 2:16corresponding author and he said that
- 2:18the article that are submitted to me,
- 2:20they lack all this stuff.
- 2:22First thing is you never add any
- 2:24reference in the abstract. References
- 2:27are there only in the intro and the
- 2:29discussion and sometimes in the
- 2:31methodology section. Okay, and
- 2:33references should be added in a proper
- 2:36format.
- 2:38It should be proper Vancouver style.
- 2:40And to do that, I have showed you guys
- 2:42multiple times how to go about it, how
- 2:44to add a reference.
- 2:47Sorry.
- 2:50Okay, how to add reference in
- 2:53Vancouver style?
- 2:55You will do like this. Let me show you.
- 3:01Either you copy the article and just go
- 3:03on to Vancouver generator and you can go
- 3:06for generating the link from here. Okay?
- 3:09And then you can by the end you will
- 3:11have all the articles written in the
- 3:14Vancouver form. Then you can get a
- 3:16little bit formatting done from the
- 3:18cloud, from cloud or any other AI
- 3:20software. But if you add all the
- 3:22references from AI softwares,
- 3:24it will have problems. Okay? So mostly
- 3:26try that to use only perplexity and and
- 3:30Claude for the for generating the
- 3:32references.
- 3:34Gemini, Google Scholar, Gemini and even
- 3:37ChatGPT, they are all they all generate
- 3:39fake responses. They all generate fake
- 3:41references.
- 3:42So if you are adding a fake reference,
- 3:43it will always get picked up. And even
- 3:46if you are using
- 3:47Claude or perplexity, you have to
- 3:49counter check it in a way.
- 3:51Sometimes [snorts]
- 3:52the reference that Claude has suggested,
- 3:55it is not what it it is stating. Okay?
- 3:58So you your task is made easy by AI, but
- 4:01what you have to do is you have to
- 4:02counter check this step, okay?
- 4:04Otherwise, these are the things that are
- 4:06costing
- 4:07a lot of articles, many problems.
- 4:10Okay, let me go through it. The next
- 4:12step
- 4:14You will not add any references in the
- 4:16abstract. Proper formatting of the
- 4:18references should be there. Prospero
- 4:20registration we will do at this stage
- 4:22now, okay? Abstract should be in 250 to
- 4:24300 words. That's the limit mostly
- 4:27journals they require it. AI plagiarism
- 4:29should be reduced below 20%. Discussion
- 4:32is the most critical part and you have
- 4:34to address it. I will share with you how
- 4:37to go about writing the discussion
- 4:39writing. Prisma flowchart you have to
- 4:41make properly. Okay, and I have also
- 4:44shared with you guys the template for
- 4:46the Prisma flowchart. Prisma flowchart
- 4:48the main problem that the students make
- 4:50is that that the numbers don't add up.
- 4:53For example, if you are saying the 200
- 4:55articles were screened in primary
- 4:57screening and 300 articles were screened
- 4:59in
- 5:01were articles screened in secondary
- 5:02screening, you should mention that 300
- 5:05articles were excluded in primary
- 5:07screening. The numbers should be adding
- 5:08up. Then if 300 articles were included
- 5:11in secondary screening, 100 were
- 5:13excluded because of wrong population, 50
- 5:16were excluded because of wrong
- 5:17intervention, 20 were excluded because
- 5:19of wrong outcome.
- 5:21Just 14 were included. But if it is not
- 5:24adding up, all these numbers are not
- 5:26adding up to 300, then that means you
- 5:28are making a problem.
- 5:30So the editor will not go and check
- 5:33whether 100 were wrong population or 100
- 5:35were wrong intervention or anything like
- 5:37that. But your numbers should add up.
- 5:40Okay? So even if there is like an up and
- 5:42down, of course you're not keeping that
- 5:45much track of the stuff. If there's an
- 5:47up and down, you don't know how many
- 5:48articles were excluded because of loss
- 5:50of intervention or wrong intervention,
- 5:52wrong population.
- 5:53You can just a little bit move it up and
- 5:56down. Like you can change it by 50 to 55
- 5:58or anything like that. But the numbers
- 6:00should add up. Okay? That's the main
- 6:03concern.
- 6:05Proper labeling and referring to the
- 6:06supplementary file and I have shared
- 6:08with you guys also
- 6:10a manuscript that you can use as a
- 6:12template.
- 6:14What is important is let me
- 6:16let me open up one manuscript.
- 6:19I guess I have it.
- 6:23Yes, this is the manuscript.
- 6:27Here what is the supplementary file when
- 6:29you go in the
- 6:31when you go in the result section
- 6:34here
- 6:35when you have mentioned specifically
- 6:37that figure one it should actually be
- 6:39referring to the figure one in the
- 6:40supplementary file and all whenever you
- 6:43are
- 6:43wherever you have mentioned the
- 6:45supplementary file you have to mention
- 6:46for example of course you are not adding
- 6:48the stuff here in the in the main
- 6:50manuscript you are adding the stuff in
- 6:52the supplementary file. So whenever
- 6:54you're talking about quality assessment
- 6:55you have to mention that it is mentioned
- 6:58in it is shown in the figure this one
- 7:00or figure seven
- 7:02figure S1 of the supplementary file then
- 7:05forward
- 7:06let me show you here they would also
- 7:08have mentioned it.
- 7:15Let me show you where else
- 7:19supplementary figures S10 to 11 okay so
- 7:22all the time you are continuously
- 7:24referring the supplementary files and
- 7:25adding the citations here. So that will
- 7:28make it easier for the editor to
- 7:31correctly check side by side from the
- 7:33manuscript and the supplementary file
- 7:35side by side okay.
- 7:37So proper labeling and referring to the
- 7:39supplementary file needs to be done. All
- 7:41the supplementary tables should be cited
- 7:44in the main manuscript okay I will tell
- 7:45you guys what stuff to add in the
- 7:47supplementary file and what to add in
- 7:49the main manuscript okay.
- 7:51Keep one parent article and try to
- 7:53follow it for that you can even keep
- 7:56for example if you are trying for mostly
- 7:58we target Springer journals
- 8:00or even you can go for any journal like
- 8:03related to whatever you want and you can
- 8:05open up any meta-analysis that they
- 8:07published and see how they have written
- 8:09it.
- 8:10And try to copy their pattern.
- 8:12Okay or at least the manuscript that I
- 8:14have shared with you guys in the group
- 8:16you can go about it that you copy the
- 8:18pattern of that manuscript okay.
- 8:21So this is the stuff that you have to
- 8:23keep in mind when you are writing the
- 8:24manuscript proper formatting should be
- 8:26there.
- 8:27So let's start. What is a good
- 8:29manuscript? It has a clear and exciting
- 8:31scientific message.
- 8:33It should have a logical flow.
- 8:36It is written in a clear and
- 8:37understandable language and is relatable
- 8:39to the reader.
- 8:40Once again if you are getting the things
- 8:41done from AI
- 8:43you can but just giving it a prompt help
- 8:45me write an intro with these many
- 8:47references of this topic and just copy
- 8:49pasting it that's not the right stuff to
- 8:51do.
- 8:52Okay you have to counter check the
- 8:54references you have to counter check
- 8:55what the content is there and what it is
- 8:57referring to actually.
- 8:59And is is it having a depth or not or is
- 9:02it making a good case or not?
- 9:05When you are writing a manuscript you
- 9:07write in this order. First is the title
- 9:10then we have the abstract
- 9:12then we have the intro then we have the
- 9:13methodology and results, discussion and
- 9:16conclusion.
- 9:17And how they are written
- 9:20how how you do you do a meta analysis?
- 9:23First of course you do methodology when
- 9:25you have done the screening and the risk
- 9:26of assessment and all that stuff you can
- 9:28write the methodology then you can go
- 9:31about results after you are done with
- 9:33the result writing or analysis on revman
- 9:35you can go about writing results. After
- 9:37results you will write the discussion
- 9:39that will be based on results.
- 9:41The conclusion will be based on all
- 9:43these three components together and
- 9:45especially on the discussion.
- 9:47The introduction is a separate part. You
- 9:49can start it while you are writing the
- 9:51intro that's totally fine and an
- 9:53abstract is a total summary of all these
- 9:55components.
- 9:57And then finally you decide the topic
- 9:59after you are done with the whole
- 10:01article okay but this is the format
- 10:03where we write the manuscript but
- 10:05actually it's written in this way. You
- 10:06cannot write the abstract
- 10:08before the whole article.
- 10:11You cannot finalize the title before the
- 10:13whole article is done. You cannot write
- 10:15discussion before the results. You
- 10:17cannot write the conclusions before the
- 10:19discussion.
- 10:20If you are doing that, that means you
- 10:22are just writing a general discussion.
- 10:24You are not discussing the results.
- 10:27You're just generally writing about the
- 10:28intervention and the control and the
- 10:30population and the outcomes.
- 10:33In the methodology section, it's an easy
- 10:35part. Okay? I will share with you guys
- 10:37now we'll go about it how to write the
- 10:39methodology. You just have to change the
- 10:41details according to what you have done.
- 10:44If you have performed any specific
- 10:46subgroup analysis, sensitivity analysis,
- 10:47you have to explain that. Okay? So
- 10:50starting with first thing is your search
- 10:52strategy.
- 10:54In your main manuscript, you will just
- 10:55write the main keywords that you have
- 10:58used,
- 10:59the main mesh words.
- 11:02In the full search strategy and the PICO
- 11:04table that we made, that will go in the
- 11:06supplementary file.
- 11:07Then how did you decided the studies?
- 11:10How was data extraction done? How was
- 11:12risk of bias done? And how was data
- 11:14analysis done? You have to explain all
- 11:16these things.
- 11:17Okay? It's a recipe for your research.
- 11:19You tell the reader exactly what you did
- 11:21in your systematic review
- 11:23with enough detail that anyone can
- 11:25replicate it without room for
- 11:28misinterpretation. Okay? Your search
- 11:31search strategy should be that much
- 11:33transparent that if somebody just copies
- 11:35it and pastes it on PubMed, it should
- 11:37generate the same articles.
- 11:39Okay? Even the method that you have done
- 11:41for data extraction, risk assessment,
- 11:43data analysis, study selection, it
- 11:45should all be as detailed as it can be.
- 11:48You will talk about which databases you
- 11:50used, any additional details, which mesh
- 11:52term you used, the full search string
- 11:55that will go in the supplementary file.
- 11:57I have also shared with you guys the
- 11:58template of the supplementary file. You
- 12:00can follow it along. Okay? Also, you
- 12:02have to mention the end date of your
- 12:05research. For example, which month was
- 12:07your research
- 12:09was your screening done? Okay? You don't
- 12:12have to go about 12th October 2026. You
- 12:15can just mention October 2026 or
- 12:18November 2026. Okay? So just that that a
- 12:22person if he is updating the meta
- 12:24analysis, he should know that you have
- 12:25done screening till November 2026.
- 12:28Okay?
- 12:29If you have hand searched any articles,
- 12:31you have to mention them. Forward
- 12:33citation means checking the references
- 12:35of the articles. You have to mention
- 12:37them. How many articles you found by
- 12:38hand searching, by forward citation,
- 12:41backward citation we did not.
- 12:44Backward citation we did, forward
- 12:45citation we did not do. Okay? Backward
- 12:47citation means checking the references.
- 12:49Forward citation is very hard.
- 12:52Forward citation actually means that you
- 12:54go on research gate, you open up that
- 12:56article and see which articles have
- 12:59cited that article. So it's a very hard
- 13:02process. Okay? We don't go about it
- 13:03mostly. We just go with these two steps
- 13:06hand searching and
- 13:09backward citation. Okay. Even if you
- 13:11don't remember all of the stuff of
- 13:13methodology we have discussed so far,
- 13:15you just have to go about copying these
- 13:19these paragraphs.
- 13:21You just have to change according to
- 13:23what you have done. Okay? For example,
- 13:25an electronic search was performed on
- 13:27PubMed and Base Cochrane clinical trials
- 13:29.gov from inception up to your time.
- 13:33We all used PubMed and Base Cochrane
- 13:36clinical trials .gov. We use these four
- 13:38databases.
- 13:39From inception means the time the
- 13:42articles start getting published on
- 13:44these databases up to which month
- 13:47you have done the screening.
- 13:49So you'll just change this time.
- 13:51Which mesh term you used, you will
- 13:53mention that.
- 13:54The detailed search strategy is
- 13:55available in the supplementary table.
- 13:57Okay? So the whole search strategy
- 13:59search strategy you will add in the
- 14:01supplementary file.
- 14:03The conference proceedings they have
- 14:04mentioned we checked them, then we ref
- 14:06checked the references of the articles.
- 14:08We also did forward citation. You will
- 14:11just mention that we did manual
- 14:12searching
- 14:14and bibliographies were searched.
- 14:19Now this part, references were imported
- 14:21to EndNote, duplicates were removed,
- 14:23what was your inclusion exclusion
- 14:24criteria, which authors did the
- 14:26screening?
- 14:27That is all summed up in this para. So
- 14:29this is the first para.
- 14:31This would be the second para of your
- 14:32manuscript.
- 14:34All of the articles were imported into
- 14:36EndNote. You can mention it here Ryan
- 14:38and duplicates were removed.
- 14:40Two authors just mentioned their
- 14:41initials, screened the titles and the
- 14:44abstracts.
- 14:45Full text of the remaining articles were
- 14:46reviewed against the eligibility
- 14:48criteria. Conflicts were discussed and
- 14:50resolved with the third author. You will
- 14:52mention the third author name or the
- 14:54supervisor or the mentor.
- 14:56Then the inclusion exclusion criteria
- 14:57they have mentioned. Okay? Study designs
- 15:00other than RCTs, animal studies and
- 15:03studies not reporting survival were
- 15:04excluded.
- 15:06Okay? So that means only RCTs were
- 15:08included.
- 15:09We did not apply any language
- 15:10restriction. You can mention that you
- 15:12applied the language restriction of
- 15:15English. Okay? But you can also mention
- 15:18this thing that those articles
- 15:20that were searched through manual
- 15:21searching and that needed translation we
- 15:23used a Google translator to get the
- 15:26result because sometimes when you
- 15:27mention that we applied a filter of
- 15:28English language,
- 15:31mostly the journals say that why that
- 15:33means that you missed out on many
- 15:35articles that were in other language.
- 15:37Okay? So you have to mention that those
- 15:40trials that were really important but
- 15:42they were in different languages, we
- 15:44translated them.
- 15:45Okay?
- 15:46So what variables were did you
- 15:48extracted? What were the outcomes?
- 15:51Events and total, mean standard
- 15:53deviation, how was risk assessment
- 15:55performed? Okay? This is the third para
- 15:56now.
- 15:58This was the second para, this was the
- 16:00first para.
- 16:01Now this is the third paragraph. If you
- 16:04have any more information other than
- 16:06this paragraph, of course you can
- 16:07mention it. Okay?
- 16:09Data extraction was done by two
- 16:10researchers. They have mentioned their
- 16:11name. In case of disagreements it was
- 16:13resolved by the third researcher and
- 16:16what data you extracted you have to
- 16:18mention it, what variables and what
- 16:20outcomes.
- 16:22And if those outcomes have a proper some
- 16:23definition or anything, you can mention
- 16:25that. Okay?
- 16:27For categorical variables,
- 16:30number of events and totals were
- 16:32extracted. For continuous outcomes,
- 16:34means and a deviation sample were size
- 16:36were extracted. You also guys did the
- 16:38same thing.
- 16:39Then this is the fourth paragraph. Two
- 16:41authors assess the risk of bias, that is
- 16:44the quality assessment using the risk of
- 16:45bias ROB 2 assessment tool. ROB 2
- 16:48assessment tool address five domains.
- 16:50This is all same stuff.
- 16:53You can just copy paste it, but of
- 16:54course, if you will copy paste it, it
- 16:56will cause plagiarism. So just change
- 16:57the
- 16:58change the words a little bit, okay?
- 17:01Disagreements were resolved by the third
- 17:02author.
- 17:04Then which uh model you use for
- 17:06analysis, if you did any subgroup
- 17:08analysis, you all have to mention that.
- 17:09Then this is the fifth paragraph about
- 17:12analysis.
- 17:13Risk ratio, mean difference were
- 17:14calculated. You also calculated this.
- 17:18[clears throat] Random effect model was
- 17:19used, chi-square test and heterogeneity
- 17:22statistics were used. This is almost the
- 17:25same stuff
- 17:26till here. Then if you have done any
- 17:29subgroup analysis, you will mention it
- 17:30here.
- 17:31Otherwise, it's all same.
- 17:34If there is any mismatch between the
- 17:36number of studies in the result section
- 17:38and the prisma flow diagram, that is
- 17:40very very important. So this is
- 17:42important in maintaining the integrity
- 17:44in reporting the study selection
- 17:45process, okay?
- 17:47So that is the most important thing.
- 17:48Then there should be no mismatch between
- 17:50the numbers. The numbers should be
- 17:52adding up.
- 17:53If you have mentioned we included 13
- 17:55articles uh
- 17:56in our meta analysis, then 13 articles
- 17:59should be there in the baseline sheet.
- 18:01If there is any mismatch, then it will
- 18:03be considered a wrong thing.
- 18:07Okay, [clears throat] let me open up
- 18:08this article.
- 18:09Wait a minute.
- 18:14Let me then show you guys uh
- 18:17[clears throat] result
- 18:23Okay. So, methodology we have done so
- 18:26far. Now we are on result writing. In
- 18:28result writing, you don't need to report
- 18:31everything about
- 18:33this was the result of this one study,
- 18:34the second study and all. You just have
- 18:37to report the major result of the whole
- 18:39analysis done. Okay? So how you will
- 18:41arrange your forest plots? You will
- 18:43arrange them in order of importance.
- 18:45Primary outcomes first and then
- 18:47secondary outcomes.
- 18:49So what are primary outcomes? Primary
- 18:51outcomes are those outcomes that are
- 18:53reported by majority of the articles.
- 18:56Okay? So if an article,
- 18:58if an outcome is reported by 10 of the
- 19:00studies, then that means it's a primary
- 19:02outcome. According to your articles, you
- 19:05decide your primary outcome. If all the
- 19:07studies are reporting one main outcome
- 19:09and they all are saying this is our
- 19:11primary outcome, then that is your
- 19:13primary outcome of your meta analysis.
- 19:15Another criteria that we keep is primary
- 19:18outcome is the one that is having
- 19:20significance and less heterogeneity.
- 19:22So you have to arrange in this order
- 19:24first significant results and less
- 19:27heterogeneity and then the insignificant
- 19:29results.
- 19:31Then how will you write it? Let me show
- 19:33you. First, how many studies are
- 19:35reporting it?
- 19:36What was the final result and any
- 19:38subgroup analysis if you have done?
- 19:40Okay.
- 19:42Let me show you guys here.
- 19:45This is the same methodology that we
- 19:46have
- 19:47gone through right now. The meta was
- 19:49conducted according to the PRISMA
- 19:51guidelines and even the PRISMA
- 19:53guidelines they have attached it in the
- 19:55supplementary file. You all have to do
- 19:56this too.
- 19:58And the Prospero number they have
- 19:59mentioned it here. Okay? This is the
- 20:01Prospero ID. When you will register on
- 20:03Prospero, you will be given this ID.
- 20:07How to register on Prospero? Let me just
- 20:09show you guys that too.
- 20:11You will open up, you will go to
- 20:13Prospero website.
- 20:16Wait a minute.
- 20:21You You log in.
- 20:26[clears throat]
- 20:27You will register the review.
- 20:30I accept the terms.
- 20:35It's of people and humans.
- 20:37It will be published on the side.
- 20:41What type of review are you planning?
- 20:43Intervention review.
- 20:47And then
- 20:50I opened it up.
- 20:53Okay. Now here in Prospero, they ask you
- 20:56all the information that you have about
- 20:58your manuscript, okay? What is the
- 21:00title? What are the details? What are
- 21:02the eligibility criteria? What is the
- 21:04inclusion criteria? Here in this case,
- 21:06you can take help from AI totally, okay?
- 21:08And even if you are taking help from AI,
- 21:11it is considered a better thing, okay?
- 21:13Because there are less chances of error.
- 21:15For example, if I'm doing a registration
- 21:17of Prospero, what I would do is I would
- 21:19just simply open up any AI software. I
- 21:22will open up Plot and I will just
- 21:24mention I am
- 21:26registering my
- 21:28protocol of Sarma on
- 21:34GABA pentane
- 21:39for Parkinson disease.
- 21:43Help me fill these domains.
- 21:45And it will give you almost the right
- 21:48stuff. You will just copy the title. You
- 21:49will you will write your title review
- 21:51title here, whatever it is.
- 21:53Then you will go back. You will go for
- 21:55review type. It's a systematic review.
- 21:58What what is your population? What's
- 21:59your rationale? What's your objectives?
- 22:01You have to answer all these questions,
- 22:03okay? This is very easy. If anyone of
- 22:05you when doing this Prospero
- 22:07registration, you face any problem, you
- 22:09can always ask me, okay? But this is a
- 22:11very easy process. You just have to go
- 22:13about it. It takes time. It takes 30
- 22:15minutes because you have to add all of
- 22:17this stuff, but still
- 22:19it's not hard, okay? Then later on,
- 22:24they ask you the question
- 22:26eligibility criteria let me see
- 22:30they will ask you about your population
- 22:31intervention control what's your
- 22:33timeline
- 22:35what's your availability you have to you
- 22:36can mention we don't have it okay it's
- 22:38totally fine
- 22:40how will you do the screening which
- 22:41outcome will you analyze which method
- 22:43will you use for analysis
- 22:46then you have to add
- 22:49your members these team members you
- 22:51don't need to add the whole team all 15
- 22:53members
- 22:54here you will just add the person who is
- 22:56registering the meta and the person who
- 22:59is a mentor okay that's it or you can
- 23:01add yourself and only me okay because
- 23:04you just want to register it we don't
- 23:06need to mention the names of all the
- 23:08authors because it becomes really hard
- 23:10and if even one person does not approve
- 23:12it it will not get published okay so we
- 23:15want just those people who are active to
- 23:17be
- 23:18part of these this prospero registration
- 23:21because this does not decide the
- 23:23authorship authorship is decided on the
- 23:26basis of when you will submit the
- 23:27article into the journal okay so this is
- 23:30again the same stuff that we just saw on
- 23:33the slides
- 23:36outcomes data extraction risk of bias
- 23:39data synthesis
- 23:42and the now we are moving off to the
- 23:43results writing you can guys can just
- 23:46copy this also you can copy this
- 23:48template of this
- 23:50meta-analysis okay you can see how they
- 23:53have written it and you can go about in
- 23:55the same way writing your whole meta
- 23:57the literature search retrieve how many
- 23:59results first they have mentioned this
- 24:01thing in the results after the removal
- 24:03of duplicates reviews
- 24:05we had a total of just 11 rcts okay
- 24:09and they have made this prisma diagram
- 24:11you can see how neatly they have made it
- 24:13records identified records removed
- 24:16duplicate records then primary screening
- 24:19done how many were removed
- 24:21how many were sought
- 24:23those that were not available as full
- 24:25articles and then secondary screening
- 24:29what's the reason for exclusion and then
- 24:31final thing.
- 24:33Okay, so in this way, your whole prisma
- 24:35checklist should be made and this should
- 24:37be made by the person who was very
- 24:38actively involved in the screening
- 24:40process.
- 24:42Then then the next paragraph on the
- 24:43results would be about the baseline
- 24:45characteristics.
- 24:47All of the studies were published
- 24:48between this to this timeline. What was
- 24:51the location? What was the site?
- 24:54What was the intervention?
- 24:56And what was the control? What else
- 25:00we have here
- 25:02what was the setting
- 25:04and any different thing about the study
- 25:06like a monitored acute care unit was
- 25:08used and anything like that. The first
- 25:10paragraph, the baseline sheet about the
- 25:11baseline sheet and then they have
- 25:13mentioned the detailed baseline sheet is
- 25:16shown in table one and they have added
- 25:18the table baseline sheet here.
- 25:20Okay. Then there the paragraph is on
- 25:23quality assessment. You can see here
- 25:25that they are not going into too much
- 25:27detail. They have just mentioned risk of
- 25:29bias was conducted using ROB two
- 25:32and we found high risk of bias in three
- 25:34studies and some concern in six studies.
- 25:37The most common issue was this
- 25:40and the remaining two RCTs were low
- 25:41risk.
- 25:42That's it.
- 25:44And then they have mentioned the
- 25:45supplementary file that you if you want
- 25:47further info, you can go go and search
- 25:49that. Okay, you can look at that the
- 25:52whole ROB assessment table.
- 25:55From here, the results are actually
- 25:56starting results and the primary
- 25:58outcomes. So first thing, how many
- 26:01studies are reporting it? Okay.
- 26:03That was the first thing. Wait a minute.
- 26:05Where was it?
- 26:10Number of studies reporting that
- 26:11outcome.
- 26:13All 11 studies reported the need of
- 26:14intubation as an outcome. So that's why
- 26:16this is a primary outcome. Okay. Then
- 26:19what was the result? What was the risk
- 26:21ratio?
- 26:22They have mentioned the patients in the
- 26:24APB group were at a lower risk of
- 26:27needing intubation as compared to the
- 26:29control group.
- 26:30Why? Because the risk ratio is less than
- 26:32one.
- 26:33I told you guys even in the last lecture
- 26:35that if the risk ratio is equal to one,
- 26:38that means there is no difference
- 26:39between the intervention and the control
- 26:41group.
- 26:42the risk ratio is less than one, that
- 26:44means the intervention group has a lower
- 26:47risk
- 26:48of that outcome as compared to the
- 26:49control group.
- 26:51Okay? And if it is more than one, that
- 26:53means that the intervention has a higher
- 26:55risk.
- 26:57Then they have mentioned the confidence
- 26:58interval, the p value and heterogeneity.
- 27:01That's it.
- 27:02Okay? You are not applying the eggers
- 27:04test that is a test that is applied on
- 27:07our software. So we have not applied it.
- 27:09So we will just exclude it for now this
- 27:11line.
- 27:12You will move on to this line.
- 27:14Grade assessment was done and this
- 27:16outcome was found judged to be moderate
- 27:18with concerns about this. Okay? So
- 27:21alongside the outcome they have
- 27:22mentioned the grade assessment.
- 27:24Okay? The last lecture was on quality
- 27:26assessment and grade assessment. If you
- 27:28guys have watched it, you will be
- 27:29knowing this thing that quality
- 27:30assessment we do of each study
- 27:33and grade assessment we do of each
- 27:34outcome.
- 27:35So along the outcome they have mentioned
- 27:37the grade. Okay? And then they have
- 27:39added the forest plot.
- 27:44Wait a minute. Sorry.
- 27:47[laughter]
- 27:52Wait a minute. Let me just message.
- 27:55Okay. Then if if you have performed any
- 27:58subgroup analysis, you will mention
- 27:59here. Otherwise move on to the next
- 28:02outcome, mortality.
- 28:0311 studies reported this mortality. What
- 28:06was the mortality? It was less than zero
- 28:09it was 0.93. So it was almost equal to
- 28:12one. So it was you can say it is
- 28:14balanced between the intervention and
- 28:15the control group. Okay? It's not like
- 28:17that the intervention has a lower risk.
- 28:20Okay? And then what was the grade
- 28:21assessment?
- 28:23You can see how they have very nicely
- 28:25referred to the supplementary file
- 28:28at each step. Okay? Then the forest plot
- 28:30they have added it. Then if you have
- 28:32performed any subgroup analysis, you
- 28:34will mention it here.
- 28:35Here, you can see in the secondary
- 28:37outcomes, they are not going into too
- 28:40much detail.
- 28:41They are just they don't even add the
- 28:44forest plots. Only the forest plots of
- 28:46the primary outcomes are added in in the
- 28:48main manuscript. The forest plots of the
- 28:50secondary outcomes, they all go in the
- 28:53supplementary file.
- 28:55Okay, for the secondary outcome, you
- 28:57don't go for doing
- 28:59writing the
- 29:00adding the forest plot because that will
- 29:02make your article very long.
- 29:06Okay, let's go back.
- 29:08[clears throat]
- 29:09So just the number of studies reporting
- 29:10that outcome, what was the result, and
- 29:13grade assessment, and if you have
- 29:15performed any subgroup analysis, okay?
- 29:20Also four point
- 29:23grade assessment of that outcome.
- 29:35Okay, now let's move on.
- 29:39So now this is about the discussion.
- 29:41Okay, so so far for the methodology, you
- 29:43just have to change the words and to
- 29:45write replace it with what you have
- 29:47done.
- 29:48Okay, analysis is very hard is very
- 29:50easy. Doing analysis is a hard work, but
- 29:53writing it is very easy. You just have
- 29:55to write how many studies are reporting
- 29:56it, what is the final result, and what's
- 29:59the grade assessment of it. That's it.
- 30:00Okay. So now the most important thing is
- 30:03the discussion writing.
- 30:04In discussion writing, you discuss each
- 30:07and every outcome in detail.
- 30:09And you have to repeat this whole thing
- 30:12for each outcome.
- 30:14Okay.
- 30:15First, if there is any previous Sharma
- 30:17or you are doing an updated Sharma, you
- 30:19will have to compare with that.
- 30:20Okay. Then compare with the main
- 30:22clinical trial. For example, mortality.
- 30:25I am discussing first outcome,
- 30:27mortality.
- 30:28I will see if how will I compare it with
- 30:30the main clinical trials. The same
- 30:32trials that you have included in your
- 30:34Sharma, you will mention that this trial
- 30:37done by this person they mentioned that
- 30:39mortality is better in the intervention
- 30:42group as compared to the control group.
- 30:44But this another trial that was done by
- 30:46the other person he mentioned something
- 30:49else okay so whatever they the trials
- 30:52talked about that outcome you have to
- 30:54mention it here.
- 30:55Okay and if the previous sarma is there
- 30:57you will have to talk what they talked
- 30:58about this outcome.
- 31:00So first in a way you are just stating
- 31:02that this was our result
- 31:04but this was the different opinions of
- 31:06the different trials on this outcome.
- 31:09If you can find any reason of why is
- 31:12there a difference in the findings of
- 31:13these different trials that have been
- 31:15conducted you will mention it.
- 31:17If you will comment if you can find any
- 31:19mechanism you can mention it.
- 31:22How will this help in policy and
- 31:24practice okay pick the most important
- 31:27finding and then repeat this whole thing
- 31:29for example the main thing you have to
- 31:31do is
- 31:33bring the data from the main trials
- 31:36what they are talking about it okay and
- 31:37you can do this from your extraction
- 31:40sheet let us go back and see one
- 31:42outcome.
- 31:45For example in the first paragraph they
- 31:47have just summarized the whole analysis.
- 31:50The results show a reduction in the risk
- 31:52of intubation with the intervention
- 31:54group as compared to the control group.
- 31:56For other outcome this was comparable
- 31:59the setting was this and that so in in a
- 32:01way
- 32:02they're just summarizing all the
- 32:03outcomes in the first paragraph but now
- 32:05here they are starting the first
- 32:07outcome.
- 32:09Prone positioning improves oxygen
- 32:11delivery and reduce mortality
- 32:14in patients on mechanical ventilation.
- 32:17Then they have explained the mechanism
- 32:19why does this happen okay and then
- 32:22firstly they have just explained the
- 32:24whole mechanism why this happens okay
- 32:26now they are bringing up each trial
- 32:29separately our results which are
- 32:31influenced by the meta by this person
- 32:34were consistent with the reduction in
- 32:36the risk of intubation.
- 32:38A similar previous meta-analysis was
- 32:40done and they mentioned this thing.
- 32:42Okay? And it was another meta-analysis
- 32:45and they mentioned this thing.
- 32:46We only included RCTs to remove
- 32:49confounding. Then another trial, again
- 32:52they are mentioning another trial.
- 32:54So in this way you have to do about it.
- 32:57And you can see that this is not
- 32:59something that AI can write. Yes, you
- 33:01can take help from AI to write the
- 33:03stuff, but at the end of the day you
- 33:05have to go back and check the articles
- 33:07yourself. Okay? Otherwise
- 33:09AI can just make the things very wrong.
- 33:12So this is how you will repeat it for
- 33:15each outcome.
- 33:16You will first quickly
- 33:18summarize the whole analysis
- 33:21in in one paragraph, not mentioning the
- 33:23numbers, just writing
- 33:25intervention was better if with regards
- 33:27to these outcomes and not with regards
- 33:29to these outcomes. Okay? And then you
- 33:32will compare with the previous
- 33:33systematic reviews and you will compare
- 33:35with the previous clinical trials. And
- 33:37if you can bring out any reason for
- 33:39this, sorry I
- 33:41I have joined again. Okay.
- 33:43Then after writing the methodology,
- 33:46results, discussion,
- 33:48discussion writing should be taken up by
- 33:51three to four persons. Result writing
- 33:53can be done by two persons. Even
- 33:55methodology can be written by two
- 33:56persons. Okay? But discussion three to
- 33:58four people should take it.
- 34:00Then final limitations, conclusions,
- 34:02even just a single person can do it.
- 34:04Be honest in mentioning the limitations
- 34:06which could affect your results. Okay?
- 34:09But you don't have to point out the
- 34:11limitations of yourself.
- 34:13For example, if you say that we did not
- 34:15do screening properly,
- 34:17of course you are giving a point to the
- 34:18editor.
- 34:19They will say you should have done it.
- 34:21We included only English studies.
- 34:23That is a limitation that you could have
- 34:26controlled. Okay? So you don't have to
- 34:27mention your limitation. You have to
- 34:30mention the limitation of the studies.
- 34:32You can mention if they had done a
- 34:34longer follow-up time.
- 34:36If the patients were compared with a
- 34:39control group or placebo,
- 34:41if the population was more
- 34:42heterogeneous, if we had a
- 34:46trial with more sample size, that would
- 34:48be a good limitation. Okay, this is
- 34:50giving a way forward
- 34:52to the article, to the meta analysis,
- 34:54limitations in the review process
- 34:55itself.
- 34:56And also try if you are mentioning any
- 34:59mistake that you have done, then try to
- 35:01mention the steps that you have taken to
- 35:03overcome them.
- 35:06In the conclusion, final conclusion,
- 35:08just write a quick para
- 35:10finalizing the whole analysis and the
- 35:12discussion. Okay, it will just mention
- 35:15that the intervention was better as
- 35:17compared to the control in these these
- 35:19outcomes and not good in these outcomes.
- 35:21Okay.
- 35:23Then the introduction writing.
- 35:24Introduction writing is an easy part,
- 35:26but once again, this is a part that is
- 35:28like that comes
- 35:30at the top of the whole manuscript.
- 35:32Should be written in a very proper way
- 35:34that it it is able to grab the attention
- 35:36of the
- 35:38editor.
- 35:39Your most important thing of your
- 35:41introduction is to convince them that
- 35:43your meta analysis is important.
- 35:46Okay, even if you are doing an updated
- 35:48meta analysis in your introduction, you
- 35:50should be able to convince the other
- 35:52person that your meta analysis is
- 35:54important for the research world. Okay.
- 35:57So first you will describe your
- 35:58population.
- 36:00Then you will describe what is already
- 36:02known, your control group,
- 36:04what is unknown, what is the
- 36:06intervention you are bringing, and then
- 36:08we have the rationale. Okay. So what is
- 36:11known? That is the procedure that is
- 36:13already being used.
- 36:14For example, any drug that's already
- 36:16being used, for example, your control
- 36:17group
- 36:18that you are comparing against.
- 36:20Of course, that control group, that drug
- 36:23has been used for a long period of time.
- 36:25So there should be something good in it.
- 36:27So you will first cover it that it it
- 36:28was a good drug and it it helped in this
- 36:31this this regard, but it was not good
- 36:33for treating AKI or anything like that.
- 36:36Okay. And then you will move on to your
- 36:38intervention. But this drug that is a
- 36:40new drug,
- 36:41it it has less side effects and it's
- 36:43better or anything like that. You have
- 36:45to cover it like that. Focus on your
- 36:47treatment of choice, any efficacy and
- 36:49safety data and highlight the knowledge
- 36:51gap.
- 36:53In a way again introduction five
- 36:55paragraphs
- 36:56population, intervention, control,
- 36:59rationale.
- 37:01And if you are updating your meta then
- 37:03you in your rationale you have to
- 37:05explain why is your meta important if
- 37:08you are updating it.
- 37:10What is different in this meta from the
- 37:12previous meta? Okay.
- 37:15In the supplementary file all the thing
- 37:17all this stuff will go.
- 37:19It will contain all the extra tables and
- 37:21figures. Your prisma checklist will go
- 37:23in the supplementary file. Your complete
- 37:25search string will go in the
- 37:26supplementary file. Any subgroup
- 37:29analysis, sensitivity analysis you have
- 37:30performed that will go in the
- 37:32supplementary file. Only the main forest
- 37:34plot will be in the main manuscript and
- 37:35only of the primary outcome.
- 37:38The forest plot of the secondary
- 37:39outcomes will go in the supplementary
- 37:41file.
- 37:42Quality assessment, grade assessment it
- 37:44will all go in the supplementary file.
- 37:46Okay.
- 37:48Then we have the abstract. It provides a
- 37:50brief overview of the studies
- 37:52objectives, methods, results and
- 37:54conclusion allowing the readers to
- 37:56quickly grasp the key concept. It can be
- 37:58structured and unstructured.
- 38:00For example, structured means you are
- 38:02giving headings background, methodology,
- 38:05results, discussion, conclusion.
- 38:06Unstructured means you are not giving
- 38:09any headings.
- 38:11Then your title it will follow the full
- 38:13pico format. It should not be like a
- 38:15very catchy thing. It should be just
- 38:17explaining what is your population, what
- 38:19is your intervention, what is your
- 38:21control and what are the outcomes.
- 38:23So it should be a concise, clear and
- 38:25informative. We will just write
- 38:27the efficacy or the role of this drug in
- 38:33AKI patients
- 38:34in improving patient survival.
- 38:37A meta-analysis of randomized controlled
- 38:40trials or cohort studies whatsoever you
- 38:42have done.
- 38:45Don't exaggerate the novelty of your
- 38:46results. Do not use words like
- 38:48groundbreaking, first ever, and all that
- 38:51stuff. Just prior to our knowledge, it
- 38:53there has been no previous
- 38:55meta-analysis on this topic.
- 38:58Don't use informal language. Don't don't
- 39:01use any difficult language. For example,
- 39:02mostly when you try to get it done from
- 39:04AI, it's a very hard language. Don't do
- 39:06that.
- 39:07Spelling and grammatical mistakes are
- 39:09totally non not tolerable.
- 39:11So even in that case, you can use AI to
- 39:14correct the spelling and the grammatical
- 39:15mistake, but you have to get it done.
- 39:18Make an outline first, explain each word
- 39:21into three or four points, and then
- 39:23write away. Okay? Just let's go for a
- 39:25quick
- 39:27recap of the stuff
- 39:28and then you guys will
- 39:30we will go for ending the session.
- 39:33First thing, the title. Okay, you can
- 39:35see here this was their intervention.
- 39:37Awake prone positioning for
- 39:39non-intubated COVID-19 patient with
- 39:41respiratory failure,
- 39:43a meta-analysis of randomized controlled
- 39:45trials. Okay? Then introduction
- 39:48abstract. This is a structured abstract
- 39:51because they are giving headings.
- 39:55And if you are not giving if you are not
- 39:57having if you are if the general
- 39:59requires writing a non-structured
- 40:01abstract or unstructured abstract, in
- 40:03that case you will not give headings.
- 40:04Okay? So firstly, they have explained
- 40:07the intro.
- 40:08You can see how they have done it. Then
- 40:10you will write the methodology.
- 40:13Then you will write the results.
- 40:16And then we will have conclusions and
- 40:19that's it.
- 40:20Then we have come onto the introduction
- 40:22section. First you see they have
- 40:24explained COVID-19, the population.
- 40:27Then they have explained
- 40:29what is their intervention that they are
- 40:31bringing in.
- 40:32Then the studies have been conducted to
- 40:34found and it was found that it was
- 40:36feasible.
- 40:37Because they are updating it, so they
- 40:40have mentioned the previous systematic
- 40:42reviews.
- 40:43They were not having that much good
- 40:44trials. There has been a new trial and
- 40:47it has not been covered. So that's why
- 40:49we are doing an updated meta-analysis.
- 40:52Methodology, results, and discussion we
- 40:54have seen. Then we have at the end
- 40:57conclusion and meta-analysis showed that
- 40:59this intervention lowers the risk of
- 41:01intubation just quickly summarizing the
- 41:03whole
- 41:05analysis and the discussion part.
- 41:07Then they have attached a supplementary
- 41:09file and you can even take help from
- 41:12this supplementary file to see the
- 41:13template.
- 41:15Okay, and that's it
- 41:18for the whole thing. So manuscript
- 41:20writing will take some time, but if you
- 41:22guys will have a good
- 41:24group effort, then it won't take that
- 41:26much time. Okay? So but try to do, try
- 41:29to have a good quality manuscript. Okay?
- 41:32Because this is the thing that will go
- 41:33forward. Nobody can see how you did the
- 41:35screening. Nobody can see how you did
- 41:37the analysis. But if you are able to
- 41:39explain in your methodology, in your
- 41:40results, in your discussion, how you
- 41:42have done, how you have put effort into
- 41:44this topic, then that will show. Okay?
- 41:47Let me know if you have any question.
- 41:52[clears throat]
- 42:05Wait a minute.
- 42:20No question, then you guys can start
- 42:22writing it. Your coordinators will
- 42:24create a poll
- 42:25in the group and you guys have to choose
- 42:27which whichever you are taking. Of
- 42:29course, those people who are taking the
- 42:30discussion, their contribution is
- 42:32higher.
- 42:33If you are taking methodology, your your
- 42:35task is a little bit easier. Result
- 42:37writing is a hard thing. Introduction
- 42:39writing is not that hard. Okay? So in a
- 42:42way, I guess if they would divide it
- 42:43like this thing, two for intro, two for
- 42:45methodology, four for discussion, and I
- 42:47guess two for result writing.
- 42:48Supplementary file making is a very hard
- 42:51task. okay? So those tasks that require
- 42:53more effort, of course, they would be
- 42:55having a more
- 42:57credit to the contributions.
- 43:00So you all have to keep the points that
- 43:03I have mentioned in the group in your
- 43:05mind and that was it for today's
- 43:07session.
- 43:09I guess let's go for ending it. Thank
- 43:11you so much for attending. Allah Hafiz.
About this transcript
This page contains the full transcript of Systematic Review and Meta Analysis Session Manuscript Writing | SRMA Session 6 by Dr Sumia - Learn Research With Us, generated from the public captions YouTube serves with the video. The transcript has 7,038 words across 1,201 segments, with the original timestamps preserved so you can click any line to jump to that moment in the embedded player.
What you can do with it
Use the transcript to take notes, quote the speaker, build a study guide, generate a summary with ChatGPT or Claude via the YouTube Summary tool, or export it as a timed subtitle file with YouTube to SRT. You can also re-open it in the transcriber to choose a caption track available on the video.
Free YouTube transcript tool
YouTube2Text is a free YouTube transcript generator — no signup, no daily limit. Paste any YouTube link and get the full transcript instantly, with timestamps, click-to-jump, caption language selection, AI prompts for ChatGPT, Claude, and Gemini, and exports to TXT, SRT, VTT, or Markdown.