Searching for Evidence in Healthcare

Yay! We get to finally talk about my favorite part of the evidence-based process, searching! We've already completed Step 1, which is developing a well-built clinical question by using PICO. Now it's time to search for what we need. There are all kinds of ways to search and so many places to find information. You can start off with a basic search in Google. We all do it, even librarians. You will find stuff, but you're not going to find everything. And there's the issue of reliability and the quality of what you find. So, the next place to try is a database. For nurses, CINAHL is one of the best. This is slightly more advanced than Google, but not too scary. Now, if you want to go really crazy, try searching something like this in PubMed. Don't let this freak you out. Most of us will never need to go into this level of detail. Okay, wait, wait, wait, wait, wait. We're not quite ready to jump into the searching yet. I know, I promise we'll get there soon. Now remember this little table from the first step of the evidence-based practice process? We want to figure out what our question type is so we know the best sort of study that will help us. These are the four biggies. Therapy or treatment, diagnosis, prognosis, and harm. It's all very good to list the types, but it doesn't make a lot of sense, at least to me, until we actually give concrete examples. of each question type. So the first one we're talking about what are the ways that smokers can quit smoking? That's pretty easy it actually has it right there in the title. That's a therapy question. Now the next one we're looking at what happens to premature babies once they are school-age and how does being born premature influence their learning? That is prognosis. And the third one we're looking to compare CT scans with other clinical findings in diagnosing intracranial hemorrhage. Again, that one's right there in the title. Diagnosis. And the final one. In adults, does cell phone use increase the risk of brain cancer? That is harm or etiology. Now, not all questions are going to fit really tidy in each question type, but these are at least the big areas. Once you know where your question best fits, that is what type of question it is, then you can figure out what studies can provide the best evidence. This kind of table is really helpful for that. You can see your question types on the left hand side here and your study types on the right side. The list starts with the strongest type of study and then moves down to research that's not as helpful for providing strong evidence. So for an example, the majority of health care questions tend to fall in the therapy category. These are best answered by randomized control trials. And if you can't find any of those, then you move down to cohort studies. If there are none of those either, then you move down to case control and so on. Now let's go into some detail about each of these study types. We're going to start with the weakest in terms of evidence-based practice and end with the strongest. So case reports and case series are articles written about one patient or a series of patients with the same issue. These are great for rare diseases, disorders, reactions, but not always the best evidence, mainly because it's not generalizable because we're only talking about a few people. But sometimes this is all you can find. When I was a hospital librarian, there were patients coming in to the hospital with pulmonary embolisms and a host of other really strange complications. The doctors finally discovered that these patients had been to what were called injection parties, where unauthorized people were injecting silicone into their bodies, a sort of cheap and illegal form of cosmetic surgery. There was absolutely no high-level evidence out there on this sort of topic, but there were some case reports. Case control studies look at a group of patients who have a disease, a condition, or exposure these are known as the cases, with those who don't have the same disease or exposure. They're known as the controls. This looks backwards in time, also known as retrospective. Sometimes this is the only ethical way to do a study. One of the most recognizable examples of case control research is the thalidomide babies. Doctors were seeing an alarming number of severe birth defects. So they compared those babies without the defects and found that those with defects had mothers who took thalidomide while pregnant. There was already a high degree of suspicion that thalidomide was causing these defects, but doctors could not randomly assign some expectant moms to take the drugs while others didn't, because totally uncool. So they had to look back in time. You will see these in a lot of the harm or etiology type of questions because of these ethical issues. A cohort study. follows groups, known as cohorts, that are linked in some way, and it follows them over time to see the effect of something. It's going forward in time, also known as prospective. Many of the well-known cohort studies go on for a very long time. In some cases, the original researchers and participants have already died and they're moving on to a second or third generation. An example of this is the Framingham Heart Study. It started in 1948, And it first followed about 5,000 residents of the town of Framingham, Massachusetts, looking for those who developed heart disease and those who didn't, and they wanted to see what was different between the two. This is the study where we discovered the ton of really important connections between lifestyle choices like smoking and diet and cardiovascular disease. This is still going on where the researchers are following the children and the grandchildren of the original participants. A prospective blind comparison to a gold standard is how many new diagnostic tools are tested. It compares the new tool to what is currently considered the gold standard. This is a controlled trial that looks at patients with varying degrees of an illness and administers both the new and the gold standard diagnostic tests. An example of this is finding easier, cheaper, and faster methods to diagnose strep throat in children. A randomized controlled trial is a study that assigns participants into a control group or into an experimental group. Obviously there is way more to it than that, but that's the two cent explanation. This type of study is truly the backbone of health sciences research. It's not an easy type of research to do, and it can be expensive in terms of both dollars and time. But a well done randomized control trial can control for differences in the participants and bias at many levels. If you throw a rock into PubMed, you're probably going to hit a randomized control trial. They are everywhere. When it comes to finding evidence-based information, this is one of the biggies, though a good randomized control trial can be a great source too, especially if there's no systematic reviews on your area of interest. A systematic review is usually written by a group of researchers who provide an unbiased, comprehensive review of all relevant studies on a particular clinical or health-related topic or question. They are supposed to review all the information from published studies, and unpublished studies, focusing on clinical trials or similar treatments. So that whole all the information should give you a pause. Seriously, who is going to review all the research on a topic? Well, these groups do. That's why it can take years to complete. It's important to distinguish a literature review from a systematic review. You'll generally see literature reviews in the intro section of articles. It lets you know that the authors reviewed some of the research related to their topic. You may even have to do a lit review yourself. Unless you are pulling all the research there is, you are not doing a systematic review and really that's okay. Now you'll generally see the word systematic review in the title of these articles just because it's so hard to do they want to make sure everyone knows that's what it is. In the bottom example you won't see the word systematic review in the title. but if you look at where it's coming from, it's coming from a database called the Cochrane Database of Systematic Reviews. Pretty much everything in there is a systematic review. So if you see Cochrane as the author or the source, it's a good bet that it's a systematic review. A meta-analysis is a subset of a systematic review. It's kind of like a super systematic review. It takes the systematic review process one step further by applying statistical analysis to evaluate all the different studies as if they were one. So with a systematic review, you could look at 10 separate studies on smoking cessation to draw a conclusion about effective treatment. With a meta-analysis, you treat the studies as one after a lot of really complicated statistics. So you could be looking at a total number, or N, in the tens or even hundreds of thousands. I've heard meta-analysis compared to creating a rope. Just one strand is not very strong, but when you combine many strands, you have something you can rely on. A meta-analysis is considered the best of the best when it comes to evidence-based information. So, how do you know what type of study a research article falls into? A lot of the times, it will tell you right there in the title. Very convenient. If it doesn't say so in the title, look in the abstract. If it doesn't say so in the abstract, pull the full article if you have access to it and look in the methods section. Now that we know a little more about what each of these studies actually entail, let's think about why the studies appear where they do on this chart. Randomized control trials are pretty much the gold standard, but you can't always find one. In some instances, it would be extremely unethical and probably illegal. So, you move on to a cohort. Now, if you can find a systematic review of randomized control trials on your topic, well, do a little dance because someone has done all the heavy lifting for you. Okay, summing up. You should have a pretty good feel for what the main question types are, which are therapy, prognosis, diagnosis, and harm. And you should also have an understanding of what each study type entails. Next, we'll talk about the varying levels of evidence and how we can use the library's resources to find great evidence-based information. To reward you for making it through a less than exciting lecture, here's an adorable baby pig.

Yay! We get to finally talk about my favorite part of the evidence-based process, searching! We've already completed Step 1, which is developing a well-built clinical question by using PICO.

Now it's time to search for what we need. There are all kinds of ways to search and so many places to find information. You can start off with a basic search in Google. We all do it, even librarians.

You will find stuff, but you're not going to find everything. And there's the issue of reliability and the quality of what you find. So, the next place to try is a database.

For nurses, CINAHL is one of the best. This is slightly more advanced than Google, but not too scary. Now, if you want to go really crazy, try searching something like this in PubMed.

Don't let this freak you out. Most of us will never need to go into this level of detail. Okay, wait, wait, wait, wait, wait.

We're not quite ready to jump into the searching yet. I know, I promise we'll get there soon. Now remember this little table from the first step of the evidence-based practice process?

We want to figure out what our question type is so we know the best sort of study that will help us. These are the four biggies. Therapy or treatment, diagnosis, prognosis, and harm.

It's all very good to list the types, but it doesn't make a lot of sense, at least to me, until we actually give concrete examples. of each question type. So the first one we're talking about what are the ways that smokers can quit smoking?

That's pretty easy it actually has it right there in the title. That's a therapy question. Now the next one we're looking at what happens to premature babies once they are school-age and how does being born premature influence their learning?

That is prognosis. And the third one we're looking to compare CT scans with other clinical findings in diagnosing intracranial hemorrhage. Again, that one's right there in the title. Diagnosis. And the final one.

In adults, does cell phone use increase the risk of brain cancer? That is harm or etiology. Now, not all questions are going to fit really tidy in each question type, but these are at least the big areas. Once you know where your question best fits, that is what type of question it is, then you can figure out what studies can provide the best evidence.

This kind of table is really helpful for that. You can see your question types on the left hand side here and your study types on the right side. The list starts with the strongest type of study and then moves down to research that's not as helpful for providing strong evidence. So for an example, the majority of health care questions tend to fall in the therapy category.

These are best answered by randomized control trials. And if you can't find any of those, then you move down to cohort studies. If there are none of those either, then you move down to case control and so on.

Now let's go into some detail about each of these study types. We're going to start with the weakest in terms of evidence-based practice and end with the strongest. So case reports and case series are articles written about one patient or a series of patients with the same issue. These are great for rare diseases, disorders, reactions, but not always the best evidence, mainly because it's not generalizable because we're only talking about a few people.

But sometimes this is all you can find. When I was a hospital librarian, there were patients coming in to the hospital with pulmonary embolisms and a host of other really strange complications. The doctors finally discovered that these patients had been to what were called injection parties, where unauthorized people were injecting silicone into their bodies, a sort of cheap and illegal form of cosmetic surgery.

There was absolutely no high-level evidence out there on this sort of topic, but there were some case reports. Case control studies look at a group of patients who have a disease, a condition, or exposure these are known as the cases, with those who don't have the same disease or exposure. They're known as the controls.

This looks backwards in time, also known as retrospective. Sometimes this is the only ethical way to do a study. One of the most recognizable examples of case control research is the thalidomide babies.

Doctors were seeing an alarming number of severe birth defects. So they compared those babies without the defects and found that those with defects had mothers who took thalidomide while pregnant. There was already a high degree of suspicion that thalidomide was causing these defects, but doctors could not randomly assign some expectant moms to take the drugs while others didn't, because totally uncool.

So they had to look back in time. You will see these in a lot of the harm or etiology type of questions because of these ethical issues. A cohort study. follows groups, known as cohorts, that are linked in some way, and it follows them over time to see the effect of something. It's going forward in time, also known as prospective.

Many of the well-known cohort studies go on for a very long time. In some cases, the original researchers and participants have already died and they're moving on to a second or third generation. An example of this is the Framingham Heart Study.

It started in 1948, And it first followed about 5,000 residents of the town of Framingham, Massachusetts, looking for those who developed heart disease and those who didn't, and they wanted to see what was different between the two. This is the study where we discovered the ton of really important connections between lifestyle choices like smoking and diet and cardiovascular disease. This is still going on where the researchers are following the children and the grandchildren of the original participants. A prospective blind comparison to a gold standard is how many new diagnostic tools are tested.

It compares the new tool to what is currently considered the gold standard. This is a controlled trial that looks at patients with varying degrees of an illness and administers both the new and the gold standard diagnostic tests. An example of this is finding easier, cheaper, and faster methods to diagnose strep throat in children.

A randomized controlled trial is a study that assigns participants into a control group or into an experimental group. Obviously there is way more to it than that, but that's the two cent explanation. This type of study is truly the backbone of health sciences research.

It's not an easy type of research to do, and it can be expensive in terms of both dollars and time. But a well done randomized control trial can control for differences in the participants and bias at many levels. If you throw a rock into PubMed, you're probably going to hit a randomized control trial. They are everywhere. When it comes to finding evidence-based information, this is one of the biggies, though a good randomized control trial can be a great source too, especially if there's no systematic reviews on your area of interest.

A systematic review is usually written by a group of researchers who provide an unbiased, comprehensive review of all relevant studies on a particular clinical or health-related topic or question. They are supposed to review all the information from published studies, and unpublished studies, focusing on clinical trials or similar treatments. So that whole all the information should give you a pause.

Seriously, who is going to review all the research on a topic? Well, these groups do. That's why it can take years to complete.

It's important to distinguish a literature review from a systematic review. You'll generally see literature reviews in the intro section of articles. It lets you know that the authors reviewed some of the research related to their topic. You may even have to do a lit review yourself.

Unless you are pulling all the research there is, you are not doing a systematic review and really that's okay. Now you'll generally see the word systematic review in the title of these articles just because it's so hard to do they want to make sure everyone knows that's what it is. In the bottom example you won't see the word systematic review in the title. but if you look at where it's coming from, it's coming from a database called the Cochrane Database of Systematic Reviews.

Pretty much everything in there is a systematic review. So if you see Cochrane as the author or the source, it's a good bet that it's a systematic review. A meta-analysis is a subset of a systematic review.

It's kind of like a super systematic review. It takes the systematic review process one step further by applying statistical analysis to evaluate all the different studies as if they were one. So with a systematic review, you could look at 10 separate studies on smoking cessation to draw a conclusion about effective treatment. With a meta-analysis, you treat the studies as one after a lot of really complicated statistics.

So you could be looking at a total number, or N, in the tens or even hundreds of thousands. I've heard meta-analysis compared to creating a rope. Just one strand is not very strong, but when you combine many strands, you have something you can rely on.

A meta-analysis is considered the best of the best when it comes to evidence-based information. So, how do you know what type of study a research article falls into? A lot of the times, it will tell you right there in the title.

Very convenient. If it doesn't say so in the title, look in the abstract. If it doesn't say so in the abstract, pull the full article if you have access to it and look in the methods section. Now that we know a little more about what each of these studies actually entail, let's think about why the studies appear where they do on this chart. Randomized control trials are pretty much the gold standard, but you can't always find one.

In some instances, it would be extremely unethical and probably illegal. So, you move on to a cohort. Now, if you can find a systematic review of randomized control trials on your topic, well, do a little dance because someone has done all the heavy lifting for you. Okay, summing up. You should have a pretty good feel for what the main question types are, which are therapy, prognosis, diagnosis, and harm.

And you should also have an understanding of what each study type entails. Next, we'll talk about the varying levels of evidence and how we can use the library's resources to find great evidence-based information. To reward you for making it through a less than exciting lecture, here's an adorable baby pig.

Transcript for:Searching for Evidence in Healthcare

Transcript for:
Searching for Evidence in Healthcare