DATA INTERPRETATION TIPS – DR.
RUPAK DESAI (470) 729 -2925
Data interpretation is a crucial step in any research study, and it involves analyzing and
drawing meaningful conclusions from the data collected. In this cohort study, you have
access to spreadsheets containing nationwide data in the PICO format. Here are some
steps and tips for interpreting the data in each sheet:
Sheet 1: Descriptive data on sample size or cohort divisions
Examine the sample size and cohort divisions: Begin by reviewing the total sample size
and how it is divided into different groups. Identify any patterns or trends that may be
relevant to your research question.
Analyze trends in the data: Look for trends in the data and identify any significant
differences between the groups. The data may be represented as numbers or
percentages, with or without p-values. If p-values are available, check for statistical
significance (p<0.05), which indicates that the observed differences are not due to
chance and are likely to be meaningful.
Check for pre and post propensity matching sample sizes: If there is information about
pre-propensity matching samples vs. post-propensity matched sample size, examine
the differences and identify any potential biases that may have been introduced by the
matching process.
Sheet 2: Cohort comparisons on mainly 3 variable categories
Sheet 2 in the shared spreadsheet contains mainly cohort comparisons based on three
variable categories: baseline patient and hospital level characteristics, pre-existing
Dr. Rupak Desai
(470) 729- 2925
relevant comorbidities, and in-hospital outcomes including healthcare resource
utilization.
Identify the groups being compared: Check which groups are being compared and how
they are defined. This will help you understand the context of the comparisons.
To interpret the data in Sheet 2, you should first familiarize yourself with the variables
being compared in each category. For example, in the baseline patient and hospital
level characteristics category, you may find variables such as age, sex, race, income,
insurance status, hospital type, and geographic location. In the pre-existing relevant
comorbidities category, you may find variables such as diabetes, hypertension, heart
disease, chronic lung disease, and cancer. In the in-hospital outcomes category, you
may find variables such as mortality rate, cardiovascular complications, and mechanical
ventilation use and resource utilization measures like patient discharge, hospital stay
and cost, etc.
Look for Chi-square or Student’s T test/Mann-Whitney U test p-values: These tests are
commonly used to compare categorical and continuous variables, respectively. Check
for any significant differences between the groups (p<0.05). Once you understand the
variables being compared, you can then look at the p-values from the chi-square test or
Mann-Whitney U test to determine if the differences observed between the groups are
statistically significant (p<0.05). If the p-value is less than 0.05, it suggests that the
difference between the groups is unlikely due to chance, and there may be a true
association between the variable and the outcome.
It's important to note that statistical significance doesn't necessarily mean clinical
significance. It's essential to interpret the findings in the context of the research question
and the study population. A statistically significant difference may not be clinically
Dr. Rupak Desai
(470) 729- 2925
meaningful, and a clinically meaningful difference may not always be statistically
significant.
Interpret the results: Once you have identified any significant differences, interpret the
results in light of your research question. What do these differences tell you about the
factors that may be associated with the outcome of interest?
Finally, when interpreting the results in Sheet 2, you should also consider the effect
sizes or magnitude of the observed differences. For example, if a variable has a
statistically significant difference between the two groups, but the difference is small
(e.g., a few percentage points), it may not be clinically significant. On the other hand, a
larger effect size may have a more significant impact on clinical decision-making.
In summary, to interpret Sheet 2 data, you should understand the variables being
compared, look at the p-values to determine statistical significance, consider clinical
significance, and examine the effect sizes of the observed differences.
Sheet 3: Logistic regression data with odds ratios and 95%
confidence interval
Understand logistic regression: Logistic regression is a statistical method used to model
the relationship between a binary outcome (e.g., presence/absence of a disease) and
one or more predictor variables.
Look for odds ratios and 95% CI: Odds ratios (ORs) and their 95% confidence intervals
(CIs) are commonly reported in logistic regression analysis. The OR tells you the
likelihood of the outcome occurring in one group compared to another, while the CI
gives you an estimate of the precision of the OR.
Dr. Rupak Desai
(470) 729- 2925
Identify significant associations: Check for any significant associations between the
predictor variables and the outcome (p<0.05). Interpret the results in light of your
research question and identify any potential risk factors or protective factors that may be
associated with the outcome.
Report positive or negative association: When reporting the association between
predictor variables and outcomes, make sure to specify whether the association is
positive (higher odds of the outcome in the predictor group) or negative (lower odds of
the outcome in the predictor group).
Consider whether to report only positive odds or positive and negative associations:
Depending on the research question and the novelty of the finding, it may be
appropriate to report only positive odds or both positive and negative associations.
In conclusion, interpreting data from spreadsheets in a cohort study requires careful
review, identification of trends and significant associations, and interpretation in light of
the research question. With these steps and tips, even beginners can accurately
interpret the data and write up high-quality abstracts. Close attention to the data will
result in better outcomes during abstract writing and ensure that all major highlights of
the analysis are included.
Dr. Rupak Desai
(470) 729- 2925
When the analysis of a cohort study finds odds ratios that are positive but with wide confidence
intervals and no significant p-values, there are different ways to report the results in clinical research
abstracts.
Here are some examples:
Emphasize the point estimate and uncertainty: "The odds ratio for the association between exposure
X and outcome Y was 1.3 (95% CI 0.8-2.4), indicating a possible positive association, but with
considerable uncertainty due to the wide confidence interval and lack of statistical significance (p=0.2)."
Highlight the limitations and potential implications: "Although we observed a positive association
between exposure X and outcome Y, with an odds ratio of 1.5 (95% CI 0.9-2.5, p=0.1), the wide
confidence interval suggests that the results are imprecise and the lack of statistical significance may be
due to chance or unmeasured confounding. These findings highlight the need for further research to
determine the true nature of this association and its potential clinical implications."
Discuss the clinical relevance and future directions: "Our study found a non-significant positive
association between exposure X and outcome Y, with an odds ratio of 1.2 (95% CI 0.7-2.0, p=0.3).
While these results are inconclusive, they suggest that exposure X may be a potential risk factor for Y.
Future studies with larger sample sizes and improved measurement of confounding factors are needed
to confirm these findings and establish the clinical relevance of this association."
Overall, it is important to present the results in a way that is clear and accurate, while
acknowledging the limitations and potential implications of the findings.
Using precise language to describe the confidence intervals and p-values can help readers better
understand the level of uncertainty in the results, while emphasizing the need for further research to
confirm or refute the observed associations.
Dr. Rupak Desai
(470) 729- 2925
