How to write the results and discussion

Michael P. Dosch CRNA PhD
May 2022

Results

Be happy! You're getting there. Just a small amount of writing to go from this point. The results and discussion are (relatively) cut and dried. But be sure to run them by all committee members and your chair before publishing or creating the poster, to make sure you haven't overlooked anything. And make sure they are congruent with your research purpose, objectives, hypothesis, and methods.

"Who's in, who's out"

Shows informed consent, and that exclusions were not arbitrary
After institutional review board approval and written informed consent were obtained, 100 subjects were recruited for the study. These were randomly assigned into two groups with the aid of a computer-generated table of random numbers. Of this group, six were excluded. Two were lost to follow up (one died on the second postoperative day from causes unrelated to the protocol, and one could not be reached by phone), and four had their anesthetic plan changed so that the protocol could not be performed (one had their surgery cancelled, three had spinal anesthesia). The characteristics of the remaining 94 are shown in Table 1. Table 2 shows... Table 3 shows...

Here's a sample "Table 1":

Table 1 Characteristics of the sample

Variable	Control n = 45	Heat & moisture exchanger n = 49	Probability
Age years¹	32.7 + 3.5	36.3 + 2.7	.08
Height m¹	1.72 + 0.6	1.67 + 0.8	NS
Weight kg¹	76.6 + 12.8	72.3 + 16.2	NS
Gender (number of males)²	21	26	NS
ASA Physical Status³	2 + 1	2 + 1	NS
OR room temperature (C)¹	21.1 + 3.6	20.6 + 2.9	NS

Notes:

Data is expressed as mean + one standard deviation. Probability determined using a two-tailed, unpaired Student's t test.
Data is expressed as number within the sample who possess the characteristic. Probability determined using Chi square (or Fisher's Exact test for 2 x 2 tables).
Data is expressed as median + one interquartile range. Probability determined using a Mann- Whitney U test.

Why is Table 1 in most studies?

Shows that demographic variables were evenly balanced in the process of random allocation of subjects to experimental and control groups.

Components of Results section

Results should answer main hypothesis or research question(s)

Order of presenting results is arbitrary
May be done in Table 1 in less-complicated studies; or be set apart to emphasize its importance.
Results that are "sidelights" should not receive equal weight
When presenting the results for the main hypotheses, consider:
- Clear, concise, simple
- Is enough detail presented to allow the reader to determine whether the effect of the experimental treatment (vs. chance alone, not bias or sloppy technique) produced the significant statistical value?
- Were adverse effects reported?
Do not state any differences were present between groups unless a significant P value is attached.
State "Cardiac output was less in the beans-and-franks group (P = .03). See Table 2." NOT "There was a significance between the beans-and-franks (B&F) group and the corn dog group. See Table 2."
You may note trends if you like (.05 < P <.10).
Don't comment on results.
Don't attach equal importance (or even bother to include) the entire statistical output. You select those descriptive and inferential statistics you wish to use, and place them in the order that seems reasonable to you.

Tables and Graphs

Presentation

Tables and graphs must stand alone (Can a member of your department unfamiliar with the study pick up your graph and explain its meaning to you?)
Text should highlight the importance or meaning of the figures and tables, not repeat the data contained within them.
Tables and figures both carry a necessary part of the message- use both
Do the numbers add up?
Are baseline values for the groups similar?
Is the degree of variability reported (and whether it is an SD or SEM identified)?
Are tables and graphs clearly labeled and appropriately scaled?
Are the results of statistical analysis presented?
Can one determine what statistical test produced the result?

Choosing figure types

To compare proportions and relative amounts (How big?), use a pie chart, a horizontal bar chart, or a table
To show trends (How do things change over time?), use a column chart or line graph
To show what's typical vs. exceptional (particularly how two groups compare in some dimension or variable), use a histogram, a cumulative percentage chart, or a box plot.
To show correlations (how well does one thing predict another?), use a scatterplot or multiplot chart.

Discussion

Presentation

Don't repeat results
Order simple to complex (building to conclusion); or may state conclusion first
Conclusion should be consistent with study objectives/research question. Explain how the results answer the question under study
Emphasize what is new, different, or important about your results
Consider alternative explanations for the results
Limit speculation
Avoid biased language or biased citation of previous work
Don't confuse non-significance (large P) with "no difference" especially with small sample sizes
Don't confuse statistical significance with clinical importance
Never give incidental observations the weight you attach to conclusions based on hypotheses generated before the study began

Components of the Discussion section

Look back

Answer whether the results make sense in terms of
- your expectation as expressed in the hypothesis?
- what you read before beginning (texts & research articles)?
- clinical practice?
- theoretical considerations?
If your results agree with previous work, fine. If they do not, explain why not, or you may leave it unresolved "We cannot account for the difference seen in..."
Were there limitations (sample size of course but what else)? Were there any problems with carrying out the method as originally planned? Not enough men in the study? Unanticipated amounts of side effects or pain? Low response rate? Failure to look at a crucial time interval?
Any unsettled points in results?

Look forward

Implications for patient care, or for theory
Suggestions for future research ("If I had to do it over I would..."). Be specific.

Conclusion

Beware inappropriate conclusions (beyond the range of the data, beyond the design of the study)

Abstract

Length 250 words
Contains all sections of paper
- Introduction with clinical importance and a key reference or two
- Methods in pertinent detail
- Results of testing the main hypothesis and most significant other results only
- Discussion a sentence or two on main implications or conclusion
Structured abstracts (See current abstracts in Anesth Analg or Anesthesiology)?
Include numbers for the main hypothesis (with descriptive statistics for central tendency & variability) so that readers will have a sense of the size of the treatment effect, and later researchers will have a basis for power and sample size calculations

Here's a sample Abstract.

Is ondansetron as effective as droperidol in prevention of postoperative nausea and vomiting?

Pamela J. Mencken RN BSN, Debra J. Blalock RN BSN, Wayne R. Miller PharmD, Michael P. Davis CRNA MS, Peter D. Hamm CRNA MS

The incidence of postoperative nausea and vomiting (PONV) remains 20 to 30% despite the availability of newer antiemetics such as ondansetron and other 5-HT3 antagonists. The cost of these drugs often results in the use of less expensive antiemetics such as droperidol. Common practice is to treat nausea and vomiting only after it has occurred. The few studies which have examined prophylaxis of PONV have had small sample sizes (Grond et al. Anesth Analg 1995; 81:603-7). The purpose of this study was to determine if there was a difference between ondansetron and droperidol in preventing PONV.

After institutional review board approval and with written informed consent, a controlled, double-blinded study was conducted with 105 male and female patients, ASA status I to III, randomly assigned into 2 groups with the aid of a computer-generated table of random numbers. All patients underwent elective intra abdominal procedures. Exclusion criteria included weight exceeding body mass index of 30 kg/m², nasogastric tube prior to induction, history of motion sickness or postoperative nausea and vomiting, antiemetic use within 24 hours of surgery, pregnancy, and subjects with contraindications to either study drug. All patients received a standardized induction with d-tubocurarine, succinylcholine, thiopental sodium, and fentanyl (2 to 20 mcg/kg). Anesthesia was maintained with isoflurane or desflurane in oxygen. Five minutes prior to induction of general anesthesia, patients received either ondansetron 4 mg intravenously (IV), or droperidol 1.25 mg IV. Syringes of identical appearance containing either agent were prepared by the satellite pharmacist, who alone was aware of group assignment. All data was collected by the principal investigators in a blinded fashion, rating PONV using a visual analogue scale of 0 to 10.

Five patients were eliminated from the study; 1 was lost to follow up, 2 patients exceeded the surgical time limit of 4 hours, 1 patient did not receive general anesthesia, and 1 patient did not receive the general anesthesia protocol as described. The groups did not differ significantly in age, weight, height, ASA status, or doses of intraoperative drugs. Patients in the droperidol group showed a trend (P=.078) toward less PONV (0.37 ± 0.038; mean ± one standard deviation) than the ondansetron group (1.0 ± 2.362). The patients who received droperidol had a trend towards a higher incidence of post discharge antiemetic use than the patients in the ondansetron group (P=0.091). Patients in the droperidol group did not spend a longer time in PACU (87 ± 62 min) as compared to the ondansetron group (102 ± 62 min; P=.443). Pretreatment with droperidol resulted in an overall 11.8% incidence of PONV, compared to 26.5% incidence in the ondansetron group (P=.07).

In conclusion, pretreatment with droperidol reduced the incidence of PONV in this sample, and patients did not stay longer in the PACU with the droperidol treatment. Further study is needed to determine if a combination of droperidol and ondansetron would decrease PONV more effectively than either agent used alone.

Miscellaneous

Report off closure to IRBs
Format is important- follow AMA!
Posters

Reading list

Cuddy PG, Elenbaas RM, Elenbaas JK. Evaluating the medical literature Part I: Abstract, Introduction, Methods. Ann Emerg Med 1983;12:549-55. https://doi.org/10.1016/s0196-0644(83)80296-0
Elenbaas JK, Cuddy PG, Elenbaas RM. Evaluating the medical literature Part III: Results and discussion. Ann Emerg Med 1983;12:679-86. https://doi.org/10.1016/S0196-0644(83)80416-8
Eger EI. A template for writing a scientific paper. Anesth Analg 1990;70:91-6. https://journals.lww.com/anesthesia-analgesia/Citation/1990/01000/A_Template_for_Writing_a_Scientific_Paper.16.aspx
Sessler DI, Shafer S, Writing Research Reports. Anesth Analg 2018;126:330-7. https://www.doi.org/10.1213/ANE.0000000000002597
Freeman J. How to Choose the Right Chart - A Complete Chart Comparison (rev. 12/15/2021). Accessed 5/17/2022 from https://www.edrawsoft.com/chart/choose-right-chart.html
Johnson SH. Avoiding the "school paper style" rejection. Nurse Anesthesia 1993;4(3):130-5. https://onlinelibrary.wiley.com/doi/10.1111/j.1750-4910.1991.tb00266.x
Elenbaas RM, Elenbaas JK, Cuddy PG. Evaluating the medical literature Part II: Statistical analysis. Ann Emerg Med 1983;12:610-20. https://doi.org/10.1016/S0196-0644(83)80205-4

Questions?
Return to Anesthesia Course Notes Site at UDM.