Writing the Abstract

From Addiction's revised checklist for writing up research reports

Use the correct structured abstract format, which is normally Aims (or Background and Aims, if appropriate), Design, Setting, Participants/Cases, Intervention and Comparator (if appropriate), Measurements, Findings, Conclusions.

Aims (or Background and Aims, if appropriate): State the aims as a series of estimation or comparison problems clearly, rather than using general terms such as assess, investigate or explore.
Design: Describe the design, including the sampling or data acquisition design and control groups or conditions.
Setting: Give the geographic location (including country) and the study setting (if appropriate).
Participants/Cases: Give the number of participants involved in the analyses with dates, gender balance and some brief individual characteristics to position the study.
Intervention and Comparator: For trials, give a summary description of the interventions and control conditions.
Measurements: Describe the measurement of the main (primary) outcome and summarize any the predictors/covariables included in the analysis. Do not include a description of the statistical methods, unless these are novel and of methodological interest.
Findings: Summarize the main findings as they relate to the aims, supported with appropriate effect sizes/P-values/confidence intervals (usually 95%) when giving estimates or inference statements. These supporting statistics should appear in results tables.
Conclusions: State the conclusions clearly as a stand-alone statement of the key results as described under Findings, avoiding abbreviations, conjecture, recommendations or musings on what future studies might try to determine.

More specific advice for writing abstract conclusions

Addiction publishes abstracts that are clear, accurate and succinct. Each abstract conclusion must provide the main generalisable statement resulting from the study. In other words, abstract conclusions should function as stand-alone statements that report the study’s main findings in terms that are meaningful.

Abstract conclusions should consist of one or two sentences that contain no abbreviations, no items that belong in an earlier section of the abstract and no musings on what future studies might try to determine (though statements that explain how the current research will affect future research are fine). None of the following or similar empty phrases should be used in the abstract conclusions (or in the findings section):

• Our findings support/indicate/suggest/show…
• The results support/indicate/suggest/show…
• This research supports/indicates/suggests/shows…
• The results from this study support/indicate/suggest/show
• We conclude…
• It is argued that…
• More research is needed to…
• Future research/studies should…

To illustrate these guidelines, below are eight abstract conclusions taken from papers accepted for publication. In each case, we edited the original conclusion to provide a better citable statement.

1. ORIGINAL ABSTRACT CONCLUSION

At present, it is not possible to interpret the evidence with any degree of certainty. Future research should consider that the distribution of alcohol consumption data is likely to be skewed and that appropriate measures of central tendency are reported.

PROBLEMS: The conclusion does not explain what the current study found and talks unnecessarily about what future studies should accomplish. The main findings are probably buried in the results section of the abstract.

REVISED CONCLUSION

Computer-based interventions may reduce alcohol consumption compared with assessment only, but the conclusion remains tentative because of methodological weaknesses in the studies.

2. ORIGINAL ABSTRACT CONCLUSION

We conclude that the public should be informed that the addition of caffeine to alcohol does not appear to enhance driving or sustained attention/reaction time performance relative to alcohol alone.

PROBLEMS: The conclusion begins with the unnecessary phrase ‘we conclude that’. Removing this phrase provides a shorter, clearer, more easily citable statement.

REVISED CONCLUSION

The addition of caffeine to alcohol does not appear to enhance driving or sustained attention/ reaction time performance relative to alcohol alone.

3. ORIGINAL ABSTRACT CONCLUSION

The rapidly rising densities of private liquor stores in BC during the study period appears to have had a significant local-area effect on rates of alcohol-related death.

PROBLEMS: The phrase ‘in BC during the study period’ is vague and potentially confusing. Spelling out ‘British Columbia’ and providing the period of study creates a clearer, more useful abstract.

REVISED CONCLUSION

The rapidly rising densities of private liquor stores in British Columbia from 2003 to 2008 appears to have had a significant local-area effect on rates of alcohol-related death.

4. ORIGINAL ABSTRACT CONCLUSION

The major finding in this descriptive study of American Indian smokers is that traditional use of tobacco is not a detriment to quitting, and may in fact be correlated with greater cessation. However, this protective effect appears to diminish considerably if the person smokes traditional tobacco. Significantly more research is needed, both to verify these findings related to the influence of traditional tobacco use and to create more effective, culturally-tailored smoking cessation programs for American Indian smokers.

PROBLEMS: The conclusion is wordy: the phrase ‘The major finding in this descriptive study’ is unnecessary and the final sentence reflects upon what future studies might achieve. Also, one of the major findings of this study was buried in the results section.

REVISED CONCLUSION

American Indians appear to show low levels of awareness of effective pharmacotherapies to aid smoking cessation but those who use 'traditional tobacco' report somewhat longer periods of abstinence from past quit attempts.

5. ORIGINAL ABSTRACT CONCLUSION

Patterns of alcohol and drug use vary across industry and occupational groups indicating the need for tailored and targeted interventions. Alcohol use at work is widespread, in contrast to the relatively low prevalence of drug use. A substantial proportion of workers who use these substances may under estimate the extent to which their drug and alcohol use may negatively affect workplace safety.

PROBLEMS: The conclusion does not provide detailed findings. A better conclusion was constructed using data from the findings section.

REVISED CONCLUSION

More than 1 in 20 Australian workers admit to having been intoxicated at work by alcohol and almost 1 in 30 report working while intoxicated by psychoactive drugs. The rates are higher for some industries, such as the hospitality industry, than others.

6. ORIGINAL ABSTRACT CONCLUSION

Pregnant HIV-infected IDUs in Ukraine had worse clinical status, poorer access to PMTCT prophylaxis and HAART, more adverse pregnancy outcomes and higher risk of MTCT than non-IDU women.

PROBLEMS: The conclusion contains several undefined abbreviations.

REVISED CONCLUSION

Pregnant HIV-infected injecting drug users (IDUs) in Ukraine have worse clinical status, poorer access to prevention of mother-to-child transmission prophylaxis and highly active antiretroviral therapy, more adverse pregnancy outcomes and higher risk of mother-to-child transmission than non-IDU women.

7. ORIGINAL ABSTRACT CONCLUSION

While most menthol smokers said they would not be happy with a ban on menthol cigarettes, more said that they would respond by quitting smoking altogether than seeking out a black market for menthol cigarettes.

PROBLEMS: The conclusion does not provide a generalizable statement but is instead phrased to reflect the study’s specific findings.

REVISED CONCLUSION

Preliminary evidence suggests that a significant minority of adolescent smokers of menthol cigarettes in the US would try to stop smoking altogether if such cigarettes were banned.

8. ORIGINAL ABSTRACT CONCLUSION

The 1976-1985 birth cohorts for men and the 1981-85 birth cohort for women appear to drink more heavily than earlier or later birth cohorts indicating the need for prevention efforts targeting these groups and the expectation of increased future alcohol-related problems and harms in these cohorts.

PROBLEMS: The conclusion does not explain that the study was conducted in the United States and ends with general observations better expressed in the body of the paper. Phrasing is also awkward.

REVISED CONCLUSION

In the United States, men born between 1976 and 1985, and women born between 1981 and 1985 appear to have higher alcohol consumption than in earlier or later years.