Interpreting Research
Interpreting Research
If you have not been trained in how to read and interpret academic papers or research data, the terminology used and the ways in which data are presented can be confusing. In this section we attempt to de-mystify some of the concepts and phrases that can be difficult to understand (e.g. What is a ‘case-control study’? What are ‘odds ratios’? How do we interpret them?).
Case Control Studies
Case-control studies are a type of research used to explore what factors might be linked to a specific condition. They’re especially useful for studying rare outcomes, like SIDS (Sudden Infant Death Syndrome), where following thousands of babies in real time (as in cohort studies or clinical trials) would be impractical or too costly. Read more how SIDS and SUDI are categorised here.
Instead of tracking people over time, case-control studies look backward:
- They compare two groups:
- Cases – people who already have the condition
- Controls – similar people who do not
- Researchers then examine past exposures (e.g. sleep practices, smoking) to see what might differ between the groups.
A classic example: To explore the link between smoking and lung cancer, researchers compare people with lung cancer (cases) to those without (controls), and look at how many in each group smoked. If smoking is much more common among the cases, this suggests an association.
In SIDS research, this approach allows scientists to identify patterns by comparing SIDS cases versus cases where the baby didn’t die. By examining sleep position, feeding method, and household factors, case-control studies help reveal which circumstances are linked to higher or lower risk.
Next, we explore how researchers interpret these findings using statistics like odds ratios and confidence intervals.
Odds Ratios and Confidence Intervals
What Are Odds Ratios?
In case-control studies, odds ratios (ORs) are used to measure how strongly a certain factor is associated with an outcome. For example, they can show how sleep environment relates to the risk of SIDS (Sudden Infant Death Syndrome).
An odds ratio compares the odds of something happening in one group to the odds in another:
- OR = 1: No difference between groups.
- OR > 1: Higher risk in the first group.
- OR < 1: Lower risk in the first group (a protective effect).
In the table below, babies who shared a room with their parents (vs. sleeping alone) had a consistently lower risk of SIDS — shown by ORs less than 1. This protective effect remains even when adjusting for other factors like sleep position or smoking.
You may also see aOR (adjusted odds ratio) or mOR (multivariate odds ratio), which account for those other influences.
| Percent exposed | |||||
| Author | Country | Case | Control | Univariate OR | Multivariate OR |
| Scragg 1996 | New Zealand | 20.7 | 37.1 | 0.44 | 0.25 |
| Blair 1999 | England | 25.3 | 39.0 | 0.53 | 0.51 |
| Hauck 2003 | United States | 20.8 | 28.1 | 0.67 | Not reported |
| Carpenter 2004 | Europe | 28.0 | 44.5 | 0.49 | 0.32 |
| Tappin 2005 | Scotland | 35.8 | 63.5 | 0.32 | 0.31 |
Table after Mitchell 2009:1715.
What Are Confidence Intervals?
Every odds ratio comes with a confidence interval (CI) — a range showing where the true value likely lies. A 95% confidence interval means we’re 95% confident the real odds ratio is within that range.
Key things to know:
- If the confidence interval includes 1, the result is not statistically significant — we can’t be sure the difference isn’t due to chance.
- Narrower confidence intervals mean more precise, reliable results.
Example:
- OR = 0.32, 95% CI: 0.20–0.50 → statistically significant, protective effect
- OR = 0.90, 95% CI: 0.60–1.35 → not statistically significant
Interpreting Case Control Studies
…And the Rocky Road to Recommendations
Finding a statistical association doesn’t mean proving causation. For example, case-control studies show a strong link between smoking and lung cancer, but that doesn’t by itself prove smoking causes cancer. More evidence and different types of studies have been needed to confirm cause-and-effect.
Case-control studies on SIDS help identify risk factors — behaviours or conditions linked with higher or lower risk. When the same associations show up across many studies, in different populations and over time, researchers gain confidence in the findings. A good example is room-sharing: We don’t fully understand why it lowers SIDS risk, but the consistent evidence has led to public health guidance to keep babies in the same room as a parent or caregiver for sleep. Learn more about the research about safe sleep guidance here.
However, turning research into clear recommendations isn’t always straightforward. Interpreting odds ratios from single or small studies to say something like “this doubles the risk” or “this is three times more dangerous” can be misleading — especially when studies vary or don’t account for all relevant factors.
A key example is bed-sharing. Many studies suggest it carries more risk than room-sharing, but the story is complicated:
- Sofa-sharing (which is extremely risky) is often grouped in with bed-sharing.
- Early studies didn’t account for important factors like parental smoking, alcohol, or drug use.
- More recent research suggests that bed-sharing without those risks may be no more dangerous than room-sharing, but the small number of relevant cases makes it hard to confirm this statistically.
Because SIDS is now very rare, future studies may never have large enough samples to fully tease apart these factors — including the possible protective effect of breastfeeding while bed-sharing.
A meta-analysis in 2012 tried to combine data from many studies to assess the risk of bed-sharing, but because most of those studies didn’t include key variables, it couldn’t provide clearer answers (Vennemann et al., 2012).
Last Reviewed: August 2025