The Basics of Research Articles
By Live Well Teens Team
Research articles are known for confusing language, complex scientific terms and explanations, and contradictory results. However, we make many household decisions based on these confusing, complex, and contradictory primary resources supposedly for scholars. In this blog post, we will help you to interpret research articles (pertaining to medicine) like a pro.
Why Reading Research Articles is Important
Research articles, although intended for scholars, have information that we can use to improve our daily lives. Although information might be hidden with complex scientific terms, being able to read these articles can allow you to better fact-check media claims and to understand the truth about a topic rather than blindly following headlines. Learning to read research is a lifelong skill, and knowing some of these basic terms can be beneficial.
The Different Types of Studies (Medical)
These types of studies examine a population and attempt to draw conclusions from this examination.
Case-control studies retrospectively compare patients with a disease (cases) and patients without a disease (controls) to determine if a risk factor has an effect on disease. This type of study is useful when investigating disease outbreaks or cases of rare diseases.
Cohort studies are part of a wider category called longitudinal studies, in which researchers follow participants for many years. Cohort studies typically observe large groups of individuals, recording their exposure to certain risk factors to find clues as to the possible causes of disease. After this is completed, researchers analyze their data and draw conclusions based on it. The main advantage with cohort studies is that they are extremely large (generally), so researchers can be confident in their conclusions being true across a wide range of people. However, several participants often drop out of these studies, which can increase the risk of bias. Finally (and notably), these studies can be expensive and time-consuming.
Cross-sectional studies are a type of observational study. In this study, participants are recruited and either included or excluded based on whether or not one participant meets certain criteria. Then, a researcher looks at both exposure (for example, recreational drug use) and the outcome of that exposure (Were they affected or unaffected?). For example, in a study on the effect of fruits on blood sugar levels, an investigator might look at who ate fruits and their subsequent blood sugar levels. These studies are inexpensive, can be conducted quickly, and are useful in public health. However, these studies can only establish associations and not causal relationships between exposure and outcome. In addition, these studies can be prone to biases due to variables that are unaccounted for. This does not necessarily make the conclusions of cross-sectional studies right or wrong. Instead, it just means that we should keep in mind that these studies only prove correlations and not causes.
A double-blinded study is one in which neither the participants nor the researchers know who receives a treatment and who receives a placebo or the experiment (for research purposes). In this case, let’s say that a double-blinded study has 10,000 participants. 5,000 receive the treatment (unknowingly) and 5,000 receive the fake substance (placebo). In this case, everyone takes the pill they were given and were examined for any symptoms of disease. Afterwards, based on their data, researchers come to a conclusion on how one variable may have impacted another.
Longitudinal studies follow subjects for years or even decades. These studies are often observational studies, with data collected on any number of exposures and outcomes. Longitudinal studies are often useful for looking at the associations between risk factors and development of disease and outcomes after treatments over time. In general, longitudinal studies are either cross-sectional, prospective, or retrospective. Longitudinal studies are extremely comprehensive and provide new insights.
In a placebo-controlled study, participants are randomly divided into different groups. Participants of one group receive a placebo (a substance that has no effect) and the participants in the other group are the experimental group which is examined in the study.
Prospective studies analyze groups of people who receive a substance (expected to have medical benefit) over a long period of time before they develop the disease or outcome being studied. Then, the outcomes of each group are compared. In general, prospective studies analyze a cohort of people. These studies are advantageous when studying multiple diseases at once. In addition, researchers don’t have to be concerned about ethical issues. However, because they are cohort studies, prospective cohort studies are subject to the limitations of cohort studies. In addition, selection bias is often an issue with these studies and confounding variables can be unaccounted for.
Randomized Controlled Trials
In a randomized controlled trial, participants are randomly placed into two or more groups. One group is the control group, while the other is the experimental group that is given a new treatment or exposed to something. The advantage with these trials is that the study population in each group is diverse and clustering of people with a common characteristic is less likely due to randomization. The main disadvantage with these trials, however, is cost – conducting these trials is extremely expensive. However, despite this, randomized controlled trials are considered to be the gold standard in medicine.
Retrospective studies find people who already have a disease or condition and analyze past events to form a conclusion about the disease or condition. For example, let’s say there is a study on MMR vaccines and autism. Researchers would look back in time to determine if any participants had any symptoms of autism after receiving the MMR vaccine. Case-control studies (see above), case reports (see below), and retrospective cohort studies are the main types. However, these studies are often placed below prospective studies when considering quality of evidence.
In a single-blinded study, researchers know who receives a treatment but the participants won’t. Take, for example, a study of an antidepressant and its effects on people with depression. Only the researchers would know who is taking the antidepressant and who isn’t, but the participants wouldn’t.
Observational studies analyze a group of people and attempt to make conclusions based on this. These types of studies are conducted when it is impossible or unethical to conduct a randomized controlled trial – for example, an RCT on smoking and lung cancer. Cohort, case-control, and cross-sectional studies are all types of observational studies (see the sections on these types of studies).
Studies of Studies
These types of studies examine all available studies on a topic. Next, based on certain criteria, studies are systematically screened and either included or excluded. The remaining studies are then analyzed and compared in order to draw conclusions.
A meta-analysis is a way to examine the results of several individual studies. The word meta-analysis refers to the analysis of the methods, results, and conclusions of different studies. A meta-analysis is conducted after a systematic review because this review is necessary to find the studies that are then analyzed.
A systematic review occurs prior to a meta-analysis. In this stage, researchers look at all papers on a topic – published or unpublished (to avoid publication bias) – and screen them. Then, based on what criteria the researchers have for study selection, many papers will be excluded in initial screening. Afterwards, all remaining studies may be subject to a second round of screening in which more studies may be excluded. After the second round of screening is over (depending on the study), the researchers read the papers left and meta-analyze them (see meta-analysis).
These articles can be shorter or longer than reviews or original articles but are generally included in many research journals. Some of these articles could be compared to comments on a social media post. In most cases, these articles are peer-reviewed like review articles and original articles are but are considered low-quality evidence.
A case report is an article in which a researcher reports unusual incidents of a medical problem. For example, if someone had a rare disease which affects only 2,000 people worldwide, a case report might provide insight into some of their symptoms and problems that a larger trial could not show existed.
These articles often include a researcher or researchers’ perspective on an article, generally in the same issue of a journal. While this does not always happen, the journal’s editor or article authors often respond(s) to the researcher’s comment. These articles may be useful if you want to understand the potential flaws or highlights of a study, as these editorials are not written by the article authors. Many times, these comments and/or editorials are written by peers, superiors to the authors of the article mentioned, or experts or leaders in the field of discussion.
The following terms are important to understand if you wish to understand papers in more detail.
A confidence interval (CI) is a probability that a study’s results will fall between two values in a set time. For example, if a study examines the effect of exercise on risk of depression in 100 people, the risk might fall between, for example, 72 and 80 percent. (This is not the real amount, just an example.) With a confidence interval of 95% (the most commonly used), this means that 95 of the 100 people examined should have the risk of depression increase by between 72 and 80 percent by exercising (which would be a vast misrepresentation of the published scientific papers on the topic, to put it lightly).
A P-value, put simply, is a way to measure whether or not the results of a trial were actually based on data and not on chance. P-values are generally expressed in decimals.
Relative risk and actual risk percentages are different. To calculate relative risk, use the Mayo Clinic’s scale at: https://www.mayoclinic.org/diseases-conditions/cancer/multimedia/relative-risk/img-20006446
The Different Types of Biases in Research
Understanding biases in research can help you to draw more accurate conclusions from the research. While there are many potential biases in research, here are a few:
When a journal is prone to publication bias, it is less likely to publish research with negative findings, thus skewing the conclusions of research overall. For example, an esteemed journal paid by an antidepressant company may be less likely to accept papers critical of antidepressants, thus potentially skewing the opinion of research in general. However, there are many reasons that publication bias can occur beyond the scope of this blog post.
Recall bias is when a participant does not give information on something related to the topic at hand because they do not recall it. For example, let’s examine a fake survey of someone about their cholesterol levels. When a researcher asks this person about their eating habits 5 years ago, the person does not remember. Unhealthy behaviors are most likely to be forgotten and underreported, which is a problem in studies based on self-reporting.
Some Tips from the Live Well Teens Team
Here are some tips that we have on reading research articles.
Don’t Read Just the Abstract (Except in Some Cases)
Many people read only the abstract of a research article and forget to read the main body of an article. While the abstract is a great summary of the article, it cannot provide you with in-depth information about the study in the main body of the article (e.g., the type of study, the data, or an in-depth discussion about other experiments). If you need to summarize findings, then reading the abstract only is fine. However, for in-depth review and analysis, you should read the whole paper to find out the true conclusion.
Examine Selection Criteria
For Original Studies:
- Population examined
- Gender or ethnicity
- People with or without a condition
- People within a specific age group
- Type of study
For Systematic Reviews and Meta-Analyses:
- Types of studies examined (e.g. randomized controlled trials, cohort studies, or even other systematic reviews)
- Language of studies
- Databases searched
- Search keywords
Read the Limitations Section of Any Paper
Many people make the mistake of jumping to conclusions without reading the limitations section of a paper. The limitations section contains the author(s)’ opinion of the gaps in the article. Remember that no study is perfectly representative of everyone and that limitations must be accounted for when evaluating a study.
Read as Many Papers as You Can on a Topic
The reason that you find media website articles with contradictory headlines is that people forget to look at a study in context. One study might be an outlier, but we might not know because we base our life decisions on that one study. If you read many papers, you will not only better understand the topic, but also examine what conclusions those papers have in comparison to the individual study. If you find that one study contradicts a wide body of literature, then you should read the methods section of the paper to find out why the study may have come to its conclusion.
Always Ask These Questions
For Original Studies:
- What are the authors looking at?
- What conclusions does this study come to?
- Why does this study come to this conclusion?
For Systematic Reviews and Meta-Analyses:
- What studies did the reviewers examine?
- What studies were included/excluded
- How many studies did the researchers base their conclusions on?
- What conclusion did the meta-analysis come to in the end?
For Other Articles:
- Does the author point out legitimate strengths and weaknesses of a study (editorials, commentary, scholarly comments)?
- Is the case in a case report actually suffering from these symptoms? (CLUE: Read other case reports on the topic.)
Don’t Be Intimidated by Scientific Terms
If you don’t know a scientific term, you should go to your search engine and look it up. If you know someone who can tell you what a term means, ask them to help you. You might find that a “useless” article has some insight you never knew about.
Understand the Scientific Method
Understanding the scientific method is, unequivocally, a surefire way to learn from research articles better. When you look at a paper through the lens of the scientific method, you can walk, in the words of one common saying, “in someone else’s shoes.” You can understand what prediction the authors made, the experiment they conducted, the results, and whether or not their hypothesis was supported based on the results.
Reading Papers Quickly
What if you don’t have time to read massive numbers of papers? One way to systematically look through papers is to read the abstract, conclusions and limitations sections only. In general, limitations will be in the section titled “Discussion”. Use Ctrl or Command + F and search for words such as “limitations”, “drawbacks”, or “disadvantages” and you should be able to find them. While this is a great method when in a hurry, we still recommend that you read through papers as thoroughly as possible to understand a topic best and make the most informed decision.
While research articles might turn some away, they are actually a great source of constantly evolving knowledge. While it might take time for you to enjoy reading these articles, you will eventually find great fun in finding out the nuances of a study or finding out that you are confused when the paper with high-quality methods and results which you just read is contradicted by ten others which have low-quality evidence for their conclusions. This week, we challenge you to read one research paper on any topic and be able to summarize its strengths, limitations, conclusions, and accuracy. Enjoy reading the scientific literature!
Thank you for reading this post, and we will see you next time on the Live Well Teens blog!