The PICO question framework is generally used when preparing a search strategy. PICO (Patient Population Comparator Outcome) was developed to assist clinicians in mapping out the question they wanted answered by the medical literature. When creating the search strategy, it is common practice to avoid searching for outcomes for a number of reasons: databases don’t search the full text (apart from the odd database like CINAHL FullText), so any mention of outcomes of interest in the full text would not be retrieved; outcomes are generally poorly described in abstracts and/or title; not all outcomes of interest are mentioned in abstracts; and specific outcomes of interest might not be covered at all (meaning that including that specific outcome in a search strategy might result in zero results, depending on how you structure the strategy).

Searching for outcomes can be relevant or even essential, when the question is something along the lines of causes of morbidity and mortality or deaths by (insert disease or condition here). In this case, you want to use text terms like: Mortality, Mortalities, Death, Deaths, Survival, Fatality, Fatalities, Outcome, Outcomes, Cause, Causes, PM, Post mortem, Post-mortem, Autopsy, Pathology, Pathologies and MeSH such as Mortality, Survival Analysis, and Survival Rate (thank you Douglas).

In systematic review search strategies, outcomes are not searched because it reduces the sensitivity of the retrieval. However, in some instances where the number to screen passes 10,000, adding outcomes to the strategy becomes tempting. When I was working on reviews and the lead researcher blanched when the number to screen passed 5000, I said to them that as databases are always growing, the number to screen will grow too. But! You can always add a publication date limit to address this problem, especially if it concerns a therapy or treatment that has only been around for x years or so.

Up next: P is for PICO

Nesting, otherwise known as searching with parentheses, is a technique used in searching as a sort of short cut. It can get complex fairly quickly though, when you have 3 or more inner nests of search terms with commands in the one string. In review search strategies, this technique is often used with adjacency commands.

Working out which nesting operation is actioned first is explained by mathematics – the inner equation is solved first, then the second from the inner etc. Databases work on mathematical principles, even bibliographic ones! (As an aside, I hated maths when I was at school, but when I started working with nesting searches, this type of equation became more understandable.)

Example 1: This line is from a review search strategy for cancer rehabilitation: ((speech or swallow* or dysphagi* or deglutit*) adj3 (rehab* or therap*)).tw,kf.

There are only 2 nests in this OVID Medline search strategy. The first line to be actioned is the rehab* or therap* search. Once this is done, the database will move onto the next operation to be actioned, which is the speech search. But the database will only retrieve items that have the words speech etc and rehab* etc occurring in the title, abstract or keywords only if there are 2 or less words between them.

Example 2: This line is from a search filter that aims to find reviews: ((quantitative adj3 (review* or overview* or synthes*)) or (research adj3 (integrati* or overview*))).ti,ab,kf.

In this OVID Medline search strategy, there are 5 nested searches. Finding out what the database actions first is the nest with the most parentheses at the end. (Correct me if I’m wrong.) In this search, it is the (integrati* or overview*) first, then the word research occurring 2 or words less between integrati* or overview*, then the review etc nest, then quantitative occurring 2 words or less between that and the review etc nest, and finally bringing results to the searcher with both these alternatives or one or the other in the title, abstract or keyword fields. This, almost 1 line search, is making the database do a lot of operations within seconds.

Nesting searches can get tricky when you have many searches going on – where do you put the closing brackets when you have more than 3?

How do you approach nested searching? What is the longest string you have used? Let me know in the comments!

Next up: O is for Outcomes

a] Four papers by Smith, Smart, Aleck and Right, were located by researcher B using keywords in Google Scholar. These papers were considered relevant for inclusion by others in the research team.

b] Four papers [Hip, Hop, Tic, and Toe] deemed relevant for inclusion by the research team, were located by researcher B by browsing in PubMed and Google Scholar using select keywords from the search strategy.

c] Four papers located by researcher B [Smith, Wesson, Pain, and Gore], were found by entering keywords from the search strategy in PubMed’s simple search box. The research team on consultation deemed these papers relevant for inclusion.

What librarians would really like to say: I wish you’d keep track and record how you came across these papers you are insisting on including. Why can’t you do this? Now I have to make up something that maybe could’ve happened. SIGH!!! How would you approach this issue? I’d love to know your thoughts.

PRISMA in search methodologies: doesn’t it annoy you to smithereens when you read in a systematic review that “… the search was conducted according to PRISMA…” or “… this systematic review has been conducted using PRISMA guidelines.” No no no and NO!!! Yes, PRISMA is a guideline, but it is a guideline for what should be reported. It is not a guideline for conducting a systematic review. There are a few handbooks and manuals out there that do that job.

And now for MeSH! MeSH (Medical Subject Headings) is the thesauri used to index biomedical literature in Medline. There was a period every year where MeSH was unavailable because the thesaurus was being updated. It now seems that MeSH is updated daily. Is this AI at work? The MeSH Browser is very useful for building searches without going into Medline. It is also fun to browse. Did you know there is a subject called Legendary Creatures? It’s under Anthropology and includes unicorns, werewolves and vampires. There was some concern on Bluesky #medlibs a few months ago that the terms Sexual and Gender Minorities and Transgender Persons would be altered or removed since the ultraconservative anti-vaxxer Kennedy was installed as head of Health & Human Services. It all looks OK so far. Has anyone noticed any changes?

A few years ago, not that long ago really, I participated in a voluntary effort to review and add to NLMs revised Population Groups. Before this project was undertaken, the NLM had reintroduced the concept ‘Blacks’, which resulted in a huge outcry. This project to look deeply at population groups was timely (and really, it was very very overdue). We finally have a MeSH term for Australian Aboriginal and Torres Strait Islander Peoples. This means all the work into updating the Aboriginal and Torres Strait Islanders filter has to be redone though. And my personal reading goal to read more African authors writing stories set in their own countries came in use – I knew what areas of the continent peoples were located.

Subheadings are great to use in MeSH, and when there isn’t a subheading you want to use with a subject, you can use a technique called floating subheadings. This is where you can search for a subject and then search for a subheading, then pair them together. Supplementary headings are like placeholder subject headings that are separate from MeSH. These headings are mostly used for substances like drugs, proteins, toxins, bacteria, viruses etc. The about page says this is updated daily in order to increase accessibility to new information. Does this mean that the annual MeSH refresh is still a thing?

Have you any favourite MeSH terms? Do tell!

Up next: N is for Nesting!

Many databases include non-English articles, articles written in English paired with another language, or index journals written in myrid languages. This is not a bad thing – it provides equitable academic research exposure to researchers in Europe, Africa and Asia. Some clinical and research areas have more foreign language articles than others. I seem to remember that allied health areas see this most often (I haven’t been able to find out where I got this from so if this rings a bell, please leave a comment). This might be no longer the case though. I was doodling around online recently and came across this article (Medical research production in native languages: A descriptive analysis of PubMed database) that details the numbers of article languages in PubMed. The top 5 languages are English, German, French, Russian, and Japanese. Chinese language articles are increasing while largely spoken languages such as Arabic, are poorly represented.

It is not unexpected that the majority of journals indexed are in English, but that should not be the reason for restricting to research in English. I asked a question on expertsearching about foreign language articles and got a reply from a Cochrane information specialist (not representing Cochrane policy). AI programs such as DeepL and Google’s own translate feature, have been used for abstracts, some full text, and trials records for the inclusion/exclusion screening stage. For included trials, native speakers have been asked to assist with data extraction and ROB analysis.

Farhad Shokraneh has written an excellent blog post about non-English language (NEL) studies that is well worth reading. The first thing that should be mentioned is that the processes for handling NEL trials should be fully documented in the protocol. Why document everything else and ignore this important aspect? You shouldn’t make up your methodology as you go along! AI translation tools are still in development, but it seems to be OK to use them for inclusion/exclusion purposes. If you use them to translate full text, get native speakers to assess accuracy. As Shokraneh found in a project he was working on, reviews of the translated texts found errors. Scientific translation services are expensive – they are costed per letter/character. If you have review funding, make sure that translation services, should they be needed, are costed.

And above all, treat translated texts the same way as you would English texts.

Up next: M is for MeSH (Medical Subject Headings)

Keywords are essential components of systematic review search strategies. It is good practice to mimic subject headings (as well as their ‘other terms’ or ‘entry terms’) as keywords in title and abstract fields. Some databases such as CINAHL, search full text. If you have this version of CINAHL, avoid searching full text and use the title and abstract fields (side note: you can now use XB to search for title AND abstract combined – what a time saver!). Consistency in searches across databases is important.

Keywords are very useful. When NLM suspends MeSH for the annual update, you can still search Medline using keywords. You can locate articles that haven’t been indexed yet (though how long this remains true with AI assisted automated indexing I don’t know). You can use keywords to access articles about countries that are not recognised by the United States and therefore not included in MeSH (eg Palestine (though interestingly, Palestinians is an Entry Term for Arabs)), and geographical areas that have changed names and borders numerous times (especially useful if you are looking for historical data). While we are discussing geography, you can use keywords to search for major cities in case the indexing/author keywords don’t mention the country. You can use keywords to search for institution names in addition to using the Affiliation field, use it to search for commonly used acronyms…. the list is endless!

Author supplied keywords (which Medline indexers never look at as far as I am aware) have been included in Medline records since 2013. They are located under the abstract and are also in the XML and MEDLINE record displays. In PubMed, you can search for them in an untagged search, or you can use the OT or TW tag. Author supplied keywords are also useful to look at when constructing search strategies.

Of course, sometimes problems with keyword searches arise. In a recent post to the expertsearching e-list, someone noticed that the copyright notices tagged at the end of abstracts in Web of Science contained words of interest (in this case, AI), which meant that irrelevant articles were being retrieved. PsycINFO has copyright notices in its abstracts and it probably will amend it due to rapidly changing technology soon. Other databases may follow. What will happen when systematic reviews and health technology assessments look at AI topics? Can you imagine the enormous retrieval sets? This is already happening with the term ‘open access’. Publishers who plaster OA and other terms all over their publications cause massive headaches for searchers looking at OA topics. Another reason not to search full text!

Up next: L is for Limits

JBI are running a free search methods webinar on the 2nd April – it’s not too late to sign up! Check it out at this link: https://events.humanitix.com/jbi-live-webinar It will be recorded so if you can’t make it on the day, you will be able to view it later.

This update is part of the overall review methodology handbook. It sits alongside the Cochrane Handbook, the CRD Manual of Reviews of Effectiveness (a bit dated now but still solid), and the EUnetHTA Review Manual.

This update is part of the overall review methodology handbook. It sits alongside the Cochrane Handbook, the CRD Manual of Reviews of Effectiveness (a bit dated now but still solid), and the EUnetHTA Review Manual.

JBI also have an extensive list of critical appraisal tools, ranging from CA Cohort Studies to CA Textual Expert Opinion/Policy/Narrative. Now that the NIH have removed their scientific integrity policy, critical appraisal is more important than ever. Check them out!

Up next: K is for Keywords (and KISS…)

I’ve been thinking about the contents of this blog post for a little while. Searching for papers written by employees at your institution is generally a headache (or more impolitely, a big pain in the ars*). Sometimes you get a situation where a researcher from your institution works on a project in partnership with another institution but neglects to add their own institution down in authorship details. This happened on a few occasions for a large group of final year medical students at my former workplace. And we were supposed to track all published research output authored by them. Ha – good luck with that! If your researchers are collaborators and not necessarily listed as authors, you can use the Medline IR (Investigator) field. From the OVID Medline database guide: “The Investigator (IR) field contains personal names of individuals (e.g., collaborators and investigators) who are not authors of a paper but rather are listed in the paper as members of a collective/corporate group that is an author of the paper. The format for authors is last name followed by one or more initials: Smith JC or Smith J.” Eg: FIR – Solomon, Benjamin IR – Solomon B

There are a number of things you can do with the Affiliation Data [AD] field (The affiliation of the authors, corporate authors and investigators appear in this repeating field; 1996- The primary author’s electronic mail (e-mail) address is included at the end of the Affiliation field, if present; December 2014- Multiple affiliations for each author or contributor are included. ) You can search for the institution and geographic locations in the AD field, and even professional qualifications. When searching for institutions, you have to have a think about all the ways it can be referred to. For example, Peter MacCallum Cancer Centre can be searched in full, but you could also search for PeterMac, PMCC, or Peter MacCallum. The following data is harvested for this field: division of the institution, institution name, city, state, postal/zip code, country (USA for the United States)[period space] then the e-mail address. Eg: Division of Vascular Surgery, University of Washington, Seattle, WA. Electronic address: szetterv@uw.edu., Peter MacCallum Cancer Centre, Melbourne, VIC, Australia. Electronic address: ben.solomon@petermac.org.

We once had a question that was a bit of a head scratcher: a bibliographic list of all papers within a certain timeframe authored by researchers from hospital radiology departments Australia-wide. You could use the AD field to search for some of this, but it would be a very involved search. Start with a state, such as Queensland[ad] AND radiology[ad] – 1561 results in PubMed. You’d have to check the full record to assess relevancy. You could go large by searching for Australia[ad] instead of Queensland and also include words like imaging[ad]. (Another option would be to find a directory or two).

What if you have a question like “I want to find some papers authored by nurses” but you don’t have CINAHL? You could use the AD field to search for credentials as well as department names, Eg: “RN”[ad] OR “CRNA”[ad] OR “MSN”[ad] OR “APRN”[ad] OR “CNS”[ad] OR “DNP”[ad] OR “NP”[ad] OR “CNE”[ad] OR “nursing”[ad] OR “nurse”[ad]. Some articles have: AD – Morgan Stein is a registered nurse currently pursuing an MSN-FNP degree at Ohio University, Athens | School of Nursing and Midwifery, Griffith University, Brisbane, Queensland, Australia.

Of course, you could also search Web of Science, Scopus or Google Scholar. However, searching for all alternative names and acronyms of your institution is still necessary, whatever database you use.

Up Next: J is for JBI

Horizon scanning! What is it? It is a proactive search for information on topics of interest. In health technology assessment (HTA), it is designed to alert payers and other invested organisations of new and emerging health technologies before they enter the market that could be potentially disruptive or have significant impact, prioritising which needs assessment. A new organisation called the International Horizon Scanning Initiative focuses on pharmaceuticals and plan to publish briefs on their website for public access (full reports for members only). You can read more about this new organisation here, and here.

The Australian and New Zealand Horizon Scanning Network which existed in the early 2000s, provided horizon scanning reports to the Medical Services Advisory Committee (MSAC), which then determined whether to commission a full HTA. At the moment, there isn’t an Australia-wide horizon scanning system, but the need has been identified.

So, where is the information gathered from to inform horizon scanning systems? The Patient Centered Outcomes Research Institute (PCORI) (operated by ECRI) has a manual; Appendix A lists the resources to check (a mix of journals, newsletters, and websites (eg:BioSpace, BusinessWire Health, Circulation, Alzheimer’s & Dementia, NEJM etc)) and also has a Embase search strategy for a focus area. This is key – horizon scanning focuses on particular topics of interest, according to the audience it is aimed at. The PCORI Horizon Scanning database is free for patients, carers and other interested persons to access.

If you do proactive searches for your clients in particular topics or for specific questions, you are doing a form of horizon scan. I did this when I was at RMH for some departments – I called these documents evidence updates.

Next post: I is for Institutional Names

But let’s get to Geography first. I’ve been thinking a little about this since working on the Australian Aboriginal and Torres Strait Islander search filter. When searching for literature that is site-specific, as well as searching for Australia exploded with all the Territories and States, you would search text words for each of those, as well as abbreviations of the States and Territories – with and without full stops (NSW / N.S.W.), as well as capital cities, major regional cities, and geographic areas (Central Tablelands, Arnhem Land, Darling Downs …). I no longer have access to OVID Medline/Embase, so I can’t do a lot of testing, but I do remember that OVID ignores full stops (but PubMed doesn’t – a search for N.S.W. comes up with 121 results).

If you are searching for literature from the United States, you can drill down to suburban areas in major cities. For example – a search in PubMed for Bronx OR Queens using text words, retrieves 6215 results. This is not really possible for cities in other countries. A search for Spruce Grove or Haddow (suburbs in Edmonton), doesn’t really result in anything useful. You’d have to search using major cities (such as Winnipeg or Ottawa).

Since NLM expanded the Geographic regions and Peoples thesauri trees, it is easier to look for new Information for specific places and people. However, as Medline doesn’t retrospectively index entries, this has to be kept in mind when searching. Also, geographic names change over time. Take the Czech Republic or the former Yugoslavia. Place names in Australia have also undergone change. Uluru, formerly known as Ayres Rock, is one example.

Grey literature! We hear a lot about it but what is it? I tend to think of it as literature that is outside commercial publishing. So, that would include clinical trial registrations and data, wouldn’t it? White papers from peak bodies, theses (you’d have to be careful here because some theses are published in full as books, or as ‘chapters’ in journals), protocols registered in PROSPERO, health technology assessments, government reports, policy briefings, legal rulings etc. Importantly, these documents are publicly accessible, so your health record, although outside commercial publishing, is not grey literature. Academic repositories are a source of grey literature (again, you will have to check if any literature has been commercially published) – OpenDOAR and OAIster are two sources you can check. There is also GreyNet and or course GreyMatters – that checklist (now a database) we all like from CADTH (Canada specific, but there are some good sources to check on it).

Next post: H is for Horizon Scanning

I’m currently doing a bit of work on the Aboriginal Australian Population filter. This is the first time I’ve been involved in a filter development project. I did a little presentation at the Information Retrieval Group (of HTAi) Advanced Searching Workshop last year in Adelaide about population filters and spoke about the original Lowitja filter. That, plus the international furore over the Medical Subject Heading population assignations, prompted the redevelopment of the filter.

But what are filters? Filters are highly developed search strategies that aim to find, or exclude, certain types of research literature. There are population filters, study type filters, geographical, age, … there are numerous filters available. A great resource for filters for medical and health literature is the ISSG Search Filter Resource. https://sites.google.com/a/york.ac.uk/issg-search-filters-resource/home I have used filters in systematic review search strategies, mostly to limit to certain types of study type. The one that most people are familiar with is the Cochrane RCT filter. Updated filters are in the Cochrane Handbook Technical Supplement to Chapter 4 Box 3a, 3b, 3c and 3d. There are also new filters for Emboss and CINAHL.

Can you use 2 or more filters in one search? You can, but you should test them together to make sure no unexpected results occur. I’ve put study filters and population filters together – including certain types of studies and excluding particular age groups (like children and teens). Common pairings is the Cochrane RCT filter with the humans not animals filter.

Filters can be of various lengths. The longest filter I’ve used is the SIGN Patient Issues filter https://www.sign.ac.uk/using-our-guidelines/methodology/search-filters/. It has since been refined and updated since I last used it. I would expect that if you are updating a systematic review, as well as checking for changes in thesauri terms, you should also check if the filter/s you used have been updated.

Next post: G is for Geography

I read something which made me think a lot recently. I’ve started reading John Berger’s ‘Ways of Seeing’, a text that followed his seminal 1972 television series of the same name. He states: …like all information, it is either put to use or ignored; Information has no special authority in itself (p24). That statement startled me for a moment but on reflection, it is true.

This was demonstrated to me when I thought about a project I was involved in at my previous place of work. We developed an evidence service where all new evidence was sent to people expressing interest in updates in specific clinical areas eg infectious diseases. They were ‘catalogued’ using ICD codes and an activity code that was being used at the time. However, for numerous reasons this service was not useful. On reflection, the main reason why it wasn’t useful was because the evidence provided did not fit an information need. The evidence provided had no special authority in itself. If it ended up being used in decision-making, it acquired authority; the authority is the context in which it can be used.

So – is an evidence information product (a systematic review, a health technology assessment or other similar synthesis) valueless? I am not saying evidence is without value. What I am clumsily trying to say is that if it isn’t answering a question or supporting a plan of action, it is not valueless, it is just information.

Next post: F is for Filters

Let’s go through this one by one. PubMed is a database platform (a service that hosts databases and provides a search interface). It has 3 components: Medline (a bibliographic database that provide links to full text if available); PubMed Central (a full text archive of journals curated by NLM), and Bookshelf (a full text books, book chapters, reports and other material archived by the NLM). NLM itself muddies the water by calling PubMed a database…

Google Scholar is a search engine for an ever expanding index of academic literature. It is not a database because it has no controlled vocabulary, the contents are not stable, and searches are not reproducible.

Embase IS a bibliographic database – correct! Why is it a bibliographic database? Citations are indexed with controlled vocabulary (Emtree), various platforms provide advanced search capabilities, there is stable access to academic content, and searches are reproducible.

EBSCO is a database platform. What database you search depends on what your institution subscribes to. As you can see from this list, there are a few versions of Medline available. Proquest is another database platform that provides access to Medline, as well as a few other databases.

Is ScienceDirect a database? No, it’s a full text index of journals and books published by Elsevier. It had an advanced search function, but that is limited. Access to full text depends on your institution’s subscriptions,

So – how should you refer to databases you searched to obtain data for your review? Make sure you state:

• what segment of the database you searched (Medline has many)
• when you searched the database
• what platform you used to search the database (this is essential for reproducibility)

I’ll close with this statement from a review that made #medlibs gasp in happiness on x.com: “This strategy was developed using Medline via the OVID interface (Ovid Medline(R) ALL) and then translated across five databases …”

Next post: Evidence

The oldest form of citation analysis is backwards/forwards (citing/cited) tracking. You would take your final included set of citations and either extract all the references used by your included articles manually or use databases like Scopus or Science Citation Index (available via Web of Science) to electronically locate them. You can now do this automatically through a service called citationchaser. Citationchaser uses the LENS.org database.

What about Google Scholar? Google Scholar used to be recommended for this, but it has fallen into disfavour after revelations that the system was being gamed by people selling citations. It’s not hard to find ‘citation boosting’ services.

Co-citation has been around since the 1970s. This type of citation analysis looks for individual papers that have citing papers in common. This is a forward looking process. Does your paper share citing papers with others? That other paper could be relevant! This sort of search relies on there being citations – if your paper hasn’t been cited, there will be no new papers to discover. Bibliographic coupling looks backwards, looking for cited papers in common; if paper A shares the same reference as paper B, they are coupled bibliographically. Check out the Wikipedia entries for co-citations and bibliographic coupling – it has a good explanation of the theory. What tools can you use to do this analysis? VOSviewer, Connected Papers and CiteSpace are still available from a mass of tools that appeared a few years ago (CoCites and Citation Gecko no longer function).

There are more citation-based mapping tools available – check out Aaron Tay’s blog post about them in all their variety: http://musingsaboutlibrarianship.blogspot.com/2022/08/citation-based-literature-mapping-tools.html Note the date – some of these tools will no longer be functioning.

Next post: Databases

There are 2 commonly used Boolean operators, and one that is less used. These are AND OR and NOT. AND is fairly straightforward – it is a joining operator between two different concepts. If you search for a AND b in the Title field, the database will quickly trawl through millions of records to find them. The words will appear in any order in any position. This can be useful when you want to quickly locate some maybe articles to establish a search strategy. When you include more fields, such as the Title Abstract (ti,ab), the search expands. What is happening here though, is that OVID, the platform I am using for Medline, ORs the fields in this AND search. So my search for (compassion fatigue).ti,ab is in fact looking for any items that have compassion as well as fatigue in the title OR abstract. The combinations searched for is all records with:

• compassion fatigue anywhere in the Title AND Abstract
• compassion fatigue anywhere in the Title OR
• compassion fatigue anywhere in the Abstract

If you wanted to search for compassion AND fatigue in the Title AND Abstract, you would have to force OVID to do this by searching compassion.ti AND fatigue.ti AND compassion.ab AND fatigue.ab. Why should you do this when you can just search as a phrase? I did a little experiment and there are records appearing in the ti,ab search that have compassion satisfaction and occupational fatigue that might be relevant in a potential search. (Anyway in Ovid, “compassion fatigue” and compassion adj fatigue” have the exact same results as compassion fatigue). Note: in OVID two words next to each other are searched as a phrase – there is no need to use “..” or adj.

OR has two meanings in the English language – there is the exclusive meaning and the other is inclusive. The exclusive meaning is often used when offering one of a limited set of alternatives: (you can buy apples or cheese, but not both; either you do it or no one will). Databases use the inclusive meaning, When you search the statement: a OR b OR c, the database will return any record containing:

• a by itself
• b by itself
• c by itself
• a and b
• a and c
• b and c
• a and b and c

In computer programming, Boolean uses true or false. All the items in the above list are true, and the non-returns are false. The above results would not return items with d in them, because d is false. The database rejects d. Another way to think of OR is to use it to combine synonyms in a set. This is very useful if something has many different ways of being referred to, for example generic drug names.

NOT can be a tricky operator. There is the potential to exclude records that could be of interest. NOT is an exclusion operator. When using NOT, the concepts you don’t want come after the NOT. Bathwater NOT Baby (a Boolean rephrasing of ‘don’t throw the baby out with the bathwater’ :D). A double NOT is used in reviews to exclude animal studies. If you see exp Animals/ NOT Humans.sh, then the following statement NOTing the results of the animal studies from the last preceding line, it looks like this:

• keep exp Animals/ set if they don’t have Humans in the Subject Heading field
• remove the records in the above set from the last line of the search strategy (the set with all the studies)

Note that this excludes animal-human studies. If you wanted to look at studies with humans and animals, you would have to use exp Animals NOT (Humans/ and exp Animals/), and NOT this set from your main search set.

So to close this post, you can see that Boolean searches are logical statements and can get very complex indeed! When you see something like (OR/3-5) AND (OR/6-8), it is worth thinking about what the searcher is asking the database to do and if it helps, write it out.

Next post: Citation Analysis

There are a number of As in searching and three I can think of right now are:

• Assumptions
• AND (the Boolean operator)
• Adjacency (or proximity)

So, let’s talk about adjacency! This search technique is almost like phrase searching, but is more flexible in that it allows for any number of additional words to occur between two ‘must have’ words. In OVID, adjacency is indicated by ADJ. Simple right? This can get a little confusing when you dig deeper. If you search word adj word, it will search for word 1 next to word 2 without any words in between in that order. But why do that if you can search word word and get the same result? Things start changing when you start adding numbers next to the ADJ. When you add 1, the search is not a strict order – any word can be first or last. There are no words in-between. So if you are reading a search strategy that uses (cancer* or tumo* or carcinoma* or sarcoma* or neoplasm*) ADJ5 (lung* or pulmonary or broncho*).ti, what does the ADJ5 do? It tells the database to return any entries that have lung etc and cancer* etc in the Title field IF there are 4 words or less occurring between them (if there are more words, the database will ignore them). In OVID, this manoeuvre includes what is known as STOP words (words like and, because, the, in with, kg etc), so keep this in mind when devising your search.

What if you are searching CINAHL, which is only available via the EBSCOhost platform, or some other database platform such as the Cochrane Library? In CINAHL, instead of ADJ, the command is N for NEAR. N5 means what it says – five words between word 1 and word 2 (it doesn’t matter what order you typed them in). If you are doing a phrase search where order is important, the command used is W (WITH). When W5 is used, the database looks for any result that has 5 words or less appearing between word 1 and word 2 in the order the words appear.

In the Cochrane Library, it is slightly different. NEAR spelt out in full without any numbers means words are searched in any order with 6 words or less between them. It is possible to search NEAR/5 (or whatever number of words you want to search for – keep in mind that the larger the number is, you might as well use AND instead). If you want the words to appear in a fixed order, type NEXT. No numbers mean words appear next to each other. Adding a /5 next to NEXT, will introduce more words between the first and second word. Note: NEXT/* does not appear in the Search Help, but it does work!

What about PubMed? PubMed is the NLM free interface to search Medline and other content. For a long time, it wasn’t possible to do an adjacency search. Then in November 2022, NLM introduced what it called adjacency searching. It can only be used in the Title or Title/Abstract fields. I haven’t experimented with it widely (I prefer constructing searches using databases via OVID), but have a go and let me know what you think of it.

Next post: Boolean!

1. Searching in just one database
2. Listing databases used
3. The search strategy
4. PRISMA
5. Not consulting a librarian

Searching in just one database. A few reviews stated that only one database was searched. One had the audacity to state that no duplicates were found. If only one database is used, this makes it a literature review, NOT a systematic review.

Listing databases used. A common issue is listing platforms and publishers as databases. Another is not specifying what database is being used when a collection of databases is being used. A review I looked at today listed Cochrane and Elsevier in the list of databases. Cochrane is an international organisation and Elsevier is the world’s largest STEM publisher. When you want to indicate that you used databases in the Cochrane Library, specify whether it was the Cochrane Database of Systematic Reviews (CDSR) or CENTRAL. It is important to list what platform was used to search. This is because databases are available on a variety of different platforms and informing you which platform was used increases reproducibility. A few others have listed Web of Science (WoS) as a database. WoS is a platform that contains many databases, including Medline. One review stated they searched PubMed, Medline and WoS (depending on their institution’s subscription, they probably searched Medline). They searched Medline 3 times!

The search strategy. Medline alone contains over 36 million citations. When your citations from all your database searching doesn’t even reach 500 (or is even less), something is seriously wrong with your search strategy. Other issues include very badly crafted search strategies (and I have seen some gawd awful ones, let me tell you). When someone is doing a critical appraisal of a systematic review, one of the first questions is about the search strategy. Is it robust enough to make it worthwhile continuing the appraisal? Many reviews have fallen at the first hurdle. Another strange issue is the number of people running the search strategies. One review stated all four authors ran the search strategy in all databases. Why? They would all get the same results (hopefully)!! Only one person has to run the searches and download all citations.

A mantra to keep in mind: the search strategy is the foundation of the systematic review. A bad foundation undermines the review.

PRISMA. PRISMA is a reporting guideline as per its full title: Preferred Reporting Items for Systematic Reviews and Meta-analysis. It is NOT a handbook or guideline for writing/conducting the review. When referring to PRISMA, please state that it was reported using/according to PRISMA.

Not consulting a librarian. All of these problems could have been avoided if the authors included a librarian on the team. If your institution or organisation has librarians on staff, please consult them! They can be on the sidelines reviewing your strategy and providing advice, right up to full authorship with screening and commenting on draft versions before journal submission. Do you want your review to pass the first test in critical appraisal? Get a librarian on board!!

After the initial clinic visit, I had to fill out an online form about vaccine side effects, which included taking my temperature every day for 7 days. Then nothing doing until the following month when I will go to one of the hospital pathology clinics to have a blood test. Then some more weeks will go by and I will have another blood test.

At the first appointment, I was asked if I would like a copy of the results when they become available. I demurred, saying that I will look it up and read the RCT. I asked which journal they were considering and they hadn’t decided on one yet. They have a wide choice due to the population they are studying. I also mentioned JANE (Journal Author Name Estimator) and that they could present at numerous conferences (AIDS, various cancer conferences, vaccinations/immunology and renal).

It will be years until I get to read it though!

If you are invited to join a trial or come across one that is recruiting that you are eligible for, I encourage you to join. It’s an interesting project to be involved in.

There are a number of steps to follow in order to assess value. I will outline them below with comments.

Is this article from a journal published by a reputable publisher? If you look at the article record on PubMed, on the right you will see full text links. There is a button for Elsevier. Elsevier is a giant in science publishing. Good so far!

Next up – let’s look at the paper itself. Yes that’s right, let’s look at the actual paper – NOT the abstract. You can’t make a judgement on abstract alone. Abstracts can give false impressions and leave out important information. Sometimes abstracts are incorrect. So, if you want to make a decision using this paper, get the full text and read it. I can’t emphasise this enough. If you want to look around for papers in order to make informed decisions, this is great. This is what healthcare staff want you to do! But you have to do the work. It is possible for you to get this paper for just a small fee. Go to your local public library and request it. This is called interlibrary loan.

Now you have the paper, what next? Well, we are going to do what is called a critical appraisal. It’s a process that anyone can do that helps unpick and understand the contents. It is basically a checklist for things to look out for and think about. This is an animal study (I know humans are animals, but still ..) so what we have to do is find a checklist for animal studies. These two webpages are from a paper[1] proposing a set of questions to answer [Table 1 Table 2]. I haven’t used this one yet as all the checklists I’ve used have been for human studies. So let’s begin!

1] “If the study was conducted in a manner that suggests little internal bias, will it be useful for the ‘next step’ because the population is relevant to ‘the next step?’” YES/N The paper authors state that there is a presumption that results of laboratory animal studies can be extrapolated to humans but this is not the case. The next step they say is to take it to other animal populations. So the question is – is it worthwhile going forward with this study and conduct another experiment in a different animal population?

This is a small early laboratory animal study in rats. Can I use it to make a health decision for myself or a family member? No. Let’s go back to the question we originally put: does ivermectin cause infertility in male humans? This paper will not answer that question. It is only asking if it causes infertility in male rats when used with verapamil. To answer this question, you need a paper that looks at side effects of ivermectin – a study about drug safety.

Safety of high-dose ivermectin: a systematic review and meta-analysis. [2] This title looks like a paper that could answer our question. There are lots of good starting points: the journal is indexed in Medline and it is a systematic review. Systematic reviews take studies found using a broad search strategy for a narrow question and analyse them together. Sometimes the studies the authors found are so similar, the individual results can be combined for statistical analysis – this is called meta-analysis. Systematic reviews still have to be critically appraised though and one tool I use is CASP. There are lots of hints as to where to look for the information in a paper and what to look for. If you have the full text, let’s do an appraisal.

1. Is there a clearly focussed question? Yes – look at the last sentence in the Introduction
2. Did the authors look for the right type of papers? – Ideally, RCTs should be used but in this case, it could be unethical. All types of studies were included that met the inclusion criteria.
3. Do you think that all important and relevant studies were included? Look at the Methods section. Relevant databases were searched, all languages included, reference papers sought and authors contacted. There is no ready access to the search strategy, which is a shame, but let’s say Yes and continue.
4. Did the authors do enough to assess the quality of the studies? There are two quality assessment sections – one for the meta-analysis which used only RCTs (a big plus) and the another for the other study types.
5. If the results have been combined, was it reasonable to do so? This means, was there a meta-analysis? Yes, there were two for different drug doses. Was it reasonable to do so? Yes. The authors state that there was low heterogeneity, which means the results were similar enough to be able to be combined.
6. What are the results? Adverse events were mild or moderate and not connected to the dosage but to the underlying condition the drug was being used to treat. There is a table of adverse effects and infertility is not there. The adverse effects listed are: eye, brain, skin and other (things like swelling and back pain),
7. How precise are the results? Looking at confidence intervals gives you an idea about how accurate the estimate is. The smaller the range, the better. The researchers are confident that if the analysis was repeated, similar results would appear. There are a range here – some are larger than others, depending on the condition.
8. Can the results be used in the local setting? This question is for health professionals. Are the populations in the study similar to mine?
9. Where all the important outcomes considered? For our purposes, yes! We want to know about infertility as a side effect.
10. Are the benefits worth the harm and costs? Ivermectin is a low cost drug and easily obtainable. But our question wasn’t about using it for treating parasitical infection.

Take-away: Ivermectin does NOT cause infertility in human males.

[1] Annette M. O’Connor, Jan M. Sargeant, Critical Appraisal of Studies Using Laboratory Animal Models, ILAR Journal, Volume 55, Issue 3, 2014, Pages 405–417, https://doi.org/10.1093/ilar/ilu038

[2] Navarro, Miriam, et al. “Safety of high-dose ivermectin: a systematic review and meta-analysis.” Journal of Antimicrobial Chemotherapy 75.4 (2020): 827-834. https://www.icpcovid.com/sites/default/files/2020-04/Safety%20of%20higher%20doses%20of%20Ivermectin%20JAC%202020.pdf

Epidemics and pandemics have been around since humans have, and even beyond that if we consider other animals and plants. In 2008, the Scout Report informed people about a new online exhibition at Harvard University Library about historical perspectives of contagious disease and epidemics. The exhibition is now expanded and the historical collection is available for view. All sorts of material is included from maps, photographs, letters, and others. A wide range of communicable diseases are included, including the bubonic plague. There is a browsable subject index which makes moving around this large collection easier.

One thing I noticed when looking at photographs is that mask wearing was de riguer from the early 20th century.

From the website:

Offering valuable insights to students of the history of medicine and to researchers seeking an historical context for current epidemiology, this collection contributes to our understanding of the global, social-history, and public-policy implications of disease.

The collection provides general background information on diseases and epidemics worldwide, and is organized around significant “episodes,” topics, and people concerned with contagious disease.

PRISMA-S was launched recently, outlining all the reporting requirements for literature searching in systematic reviews. Item 8 is: Include the search strategies for each database and information source, copied and pasted exactly as run. Note ‘exactly as run’. This is not possible in PubMed. Medline on the OVID platform (or via EBSCO or other aggregator) is preferred.

Searching Medline via a database aggregator platform has been the preferred practice for building and running search strategies for systematic reviews for over two decades now, mostly because of the ability to use proximity operators. Proximity operators are not available in PubMed and there are no plans to introduce them.

So, can you use PubMed at all? You can use it to search PMC articles, which can be useful for surgical (and other) questions. Care still needs to be taken though, and make sure you capture your search before logging off – PubMed no longer stores search histories.