Writing is thinking on paper
Over the past 15 years or so, I have had the pleasure of conducting many short courses (usually 1 week with 10 lectures) in scientific writing in Australia, Vietnam and Thailand. These courses were designd to help my colleagues, young and old, to get their papers published in peer-reviewed journals. Over the years, I have come up with a number of principles of scientific writing that I would now like to share with you all. In this note, I am going to focus on the writing of title and abstract of an impact paper.
To me, an impact paper is a piece of scientific communication that reports Patient-Oriented Evidence that Matters (POEM). I work in clinical research, and my examples are therefore mainly taken from clinical studies. I got the idea from BMJ editor-in-chief Richard Smith (BMJ 2002). A large proportion (~70%) of papers are rejected because they don’t have POEM. If you listen to editors of high impact journals such as New England Journal of Medicine, you will find that they are constantly searching for the best work that is reported dispassionately. So, the first thing to keep in mind is that we are talking about writing a POEM paper in this series; we are not talking about how to write a boring paper. My formula for an ‘impact paper’ is very simple:
Impact Paper = Good Science + Good Writing
Now, assuming that your science is great, the next and really important question is ‘is your writing good?’ Perhaps, you cannot answer that question. Judging from my personal experience as an author and journal editor / reviewer, I can say that most papers are poorly crafted. In this note, I would like to share with you some tips on how to write a good and impact scientific paper.
First, just 3 side notes (warnings):
Writing is difficult, but scientific writing is even more difficult. I would say that scientific writing is more difficult than doing an experiment, because the former requires intensive mental work but the latter is largely manual work. Scientific writing takes a lot of time, and your manuscript is likely to go through multiple iterations before it is suitable for submitting to a journal.
Writing a boring paper is pretty easy, but writing an impact paper is very hard. A boring paper follows the standard IMRaD format (Introduction + Methods + Results and Discussion) filled with common words and sentences describing the content that is incremental at best. That is why most scientific papers are not read, and even if they are read, few are cited. Impact writing requires strategic thinking. Thus, by impact writing I mean the manuscript must be structured to concisely convey the essence of your data/message to the world. Impact writing is aimed at demonstrating an advance in knowledge that qualifies your manuscript as a research paper.
Many scientists are not trained in scientific writing. When I started my doctoral study, I did not have a clue about writing, let alone scientific writing, but I was lucky to have two wonderful mentors whoare excellent writers (one of them later became an editor-in-chief of JBMR). I have learned (and I am still learning) a lot from my mentors about the choice of words and impact writing. So, as you can guess, I have learned scientific writing on the job. I have also read a lot of books and journal articles on scientific writing and scientific communication to further garner my writing skills. And, in the process of learning, I realize that most scientists — even professor level scientists — are just like me, that they too have no training in scientific writing, and that they usually make a lot of mistakes that they are not aware of. Some professors have the tendency of writing long and rambling sentences with colorful words. This tendency is perhaps rooted from their cultural background, but it is not appropriate for scientific papers.
How to write an impact title?
Having published many papers over the years, I have come to the conclusion that the two important components of a paper are the title and the abstract. For every person who reads the whole content of a paper, about 500 read only the title and may be the abstract. Unfortunately, most authors do not pay adequate attention to the two components, and that is a disadvantage to them. Here, I would like to offer some general tips and principles of writing an impact title.
A good paper should have a strong title. There are 3 types of title: declarative, descriptive, and interrogative. Declarative title is typically a summary statement of your study’s key finding and/or message (eg “Activated macrophages are essential in a murine model for T cell–mediated chronic psoriasis“, JCI 2006). Descriptive title is a neutral statement with no clear message (eg “Prospects for using risk scores in polygenic medicine”, Genome Med 2017). Interrogative title, as the name implies, is actually a question (eg “Are there re-arrangement of hotspots in the human genome?” PLoS Comp Biol 2007). Papers with a question-style titles have more downloads that those with declarative or descriptive titles; however, when it comes to citation, papers with a declarative or descriptive titles seem to attract more cites than those with a question-style titles (Jamali et al Scientometric 2011; Fig 1). Journal editors and readers seem to prefer papers with a strong declarative title.
Fig 1: Association between types of title and citations and downloads
Fig 2: Correlation between title length and citations
What is the optimal number of words in a title? There is really no golden rule here. However, papers with short titles seem to have better citations than those with long titles (Letchford et al. J Royal Soc 2015; Fig 2). Personally, I prefer titles with 20 words or less.
Now, the next question is: how to write an impact title? I have come up with 5 ‘principles of titling’ as follows:
Principle 1: an impact title should convey a new information or a key and impact message (eg Essential hypertension: effect of calcium antagonist felodipine);
Principle 2: an impact title should start with an important word or concept. Consider “Smoking is associated with post-fracture mortality” and “Genetics and the individualized assessment of fracture”, which one you want to emphasize?
Principle 3: emphasis of method or methodology. Consider the following 3 titles, “Zinc supplementation for growth”, “Zinc supplementation for growth in preterm infants”, and “Zinc supplementation for growth in preterm infants: a randomized controlled trial“, which one you prefer?
Principle 4: an impact title should have some keywords. Your paper will be indexed in Pubmed and other bibliometric databases, and those databases use keywords for classification purpose. Therefore, having at least one keyword in the title will help readers search for relevant papers. Consider the following title, “Fracture risk assessment: the role of ultrasound“, it will attract attention of those who are interested in ‘fracture‘ and ‘ultrasound.’
Principle 5: an impact title should be informative or interesting. Consider the following titles, “An Investigation of Hormone Secretion and Weight in Rats”, “Fat Rats: Are Their Hormones Different?”, and “The Relationship of Luteinizing Hormone to Obesity in the Zucker Rat”, which one is more informative?
You can view the above 5 principles as 5 DOs. Now, there are 4 DON’Ts in the title:
- Don’t use jargons in the title;
- Don’t use abbreviations and acronyms;
- Don’t use non-informative words (eg “Calcium supplementation for growth“)
- Don’t include too many details or distracted information (eg “A study of association between statin and bone loss in women aged 60-90 years in District 1, Ho Chi Minh City, Vietnam.”)
Checklist: Ask yourself the following questions when you are writing a title. You may, if you like, see this as a check list:
- Is it a declarative?
- Is it less than 20 words?
- Does it convey a message?
- Is it started with an important word?
- Does it have a keyword?
- Is it informative or interesting?
If you tick “yes” to those questions, then your title qualifies as an impact title. If not, go back and re-title your paper until you can tick yes to all 6 questions!
How to write an impact abstract?
Abstract, as the name implies, is a summary of contents of a paper. The word abstract is actually from Latin, Ab means “out”, and trahere means “to pull”. So, abstract literally means “to pull out”. A good abstract should extract all key information of a paper in typically less than 250 words. The key information here includes the background, aim, methods, results, and conclusion of a paper.
It should be emphasized that the abstract is an independent part of a scientific paper. Let me say that again: the abstract is a stand alone entity of a paper. You have to write the abstract so that readers do not need to read the full paper, but they can still capture the “story” of the paper by reading the abstract. Writing a good abstract can be a challenge, because you have too much information to cover, but you have to limit the whole story in <250 words!
Thus, after the title, the abstract is also an important element of a paper. Readers are more likely (10-500 times) to read the abstract than the full length paper. Therefore, you shoud (actually, you must) invest enough time and intellectual power to write a good abstract. An editor of JAMA used to remark that “The abstract is the single most important part of a manuscript, yet the most often poorly written“. Here, I can offer some tips for abstract writing.
There are two types of abstract: structured and unstructured. A structured abstract has subtitles such as background, aim, methods, results, and conclusion. A unstructured abstract is a paragraph with typically 10-15 sentences. Some journals prefer structured abstract, while others ask for unstructured abstract. However, whether it is structured or unstructured, the abstract should provide an overview of the paper and answer the following 4 key questions:
- Why did you do the study?
- How did you do it?
- What did you find?
- What does your finding mean and why should readers care?
Here is an example of a structured abstract (Ho-Pham et al 2015):
“Background. The burden of obesity in Vietnam has not been well defined because there is a lack of reference data for percent body fat (PBF) in Asians. This study sought to define the relationship between PBF and body mass index (BMI) in the Vietnamese population.
Methods. The study was designed as a comparative cross-sectional investigation that involved 1217 individuals of Vietnamese background (862 women) aged 20 years and older (average age 47 yr) who were randomly selected from the general population in Ho Chi Minh City. Lean mass (LM) and fat mass (FM) were measured by DXA (Hologic QDR 4500). PBF was derived as FM over body weight.
Results. Based on BMI ≥30, the prevalence of obesity was 1.1% and 1.3% for men and women, respectively. The prevalence of overweight and obesity combined (BMI ≥25) was ~24% and ~19% in men and women, respectively. Based on the quadratic relationship between BMI and PBF, the approximate PBF corresponding to the BMI threshold of 30 (obese) was 30.5 in men and 41 in women. Using the criteria of PBF >30 in men and PBF >40 in women, approximately 15% of men and women were considered obese.”
Conclusion. These data suggest that body mass index underestimates the prevalence of obesity. We suggest that a PBF >30 in men or PBF >40 in women is used as criteria for the diagnosis of obesity in Vietnamese adults. Using these criteria, 15% of Vietnamese adults in Ho Chi Minh City was considered obese.”
and a unstructured abstract (Styrkarsdottir et al. Nature 2013):
“Low bone mineral density (BMD) is used as a parameter of osteoporosis. Genome-wide association studies of BMD have hitherto focused on BMD as a quantitative trait, yielding common variants of small effects that contribute to the population diversity in BMD. Here we use BMD as a dichotomous trait, searching for variants that may have a direct effect on the risk of pathologically low BMD rather than on the regulation of BMD in the healthy population. Through whole-genome sequencing of Icelandic individuals, we found a rare nonsense mutation within the leucine-rich-repeat-containing G-protein-coupled receptor 4 (LGR4) gene (c.376C>T) that is strongly associated with low BMD, and with osteoporotic fractures. This mutation leads to termination of LGR4 at position 126 and fully disrupts its function. The c.376C>T mutation is also associated with electrolyte imbalance, late onset of menarche and reduced testosterone levels, as well as an increased risk of squamous cell carcinoma of the skin and biliary tract cancer. Interestingly, the phenotype of carriers of the c.376C>T mutation overlaps that of Lgr4 mutant mice.”
Now, I propose 5 principles of abstract writing as follows:
Principle 1: novelty and challenge. A good abstract should provide a brief background of the problem, a gap in knowledge, and a specific aim to fill that gap. Still, a better abstract should have something that challenges the current dogma or status quo. Example: “Many medical societies propose that the BMI threshold for defining obesity in Asians should be lower than the threshold for whites. This study sought to challenge that proposal by …”
Principle 2: quantitative. To me, a good and informative abstract should provide some quantitative data and/or evidence. Consider the following statement, “The study involved two cohorts of patients from two independent hospitals”, and “The study involved 1765 patients in the development cohort and 1728 in the validation cohort”. Which one is more informative?
Principle 3: the conclusion must be consistent with the data. This is extremely important, because if the conclusion is deviated from the reported data, your paper may be viewed as a coercision. Consider a recent paper (Inouye, et al JACC 2019), the authors report a very modest result, “The metaGRS had a higher C-index (C = 0.623; 95% CI: 0.615 to 0.631) for incident CAD …“, but when it comes to the conclusion they write a very strong statement: “The genomic score developed and evaluated here substantially advances the concept of using genomic information to stratify individuals with different trajectories of CAD risk and highlights the potential for genomic screening in early life to complement conventional risk prediction”!
Principle 4: a good abstract should have a take-home message. A take-home message here is the meaning of your finding, not a summary of the study. It can also be an advice. Imagine that you look straight in the eye of readers and make a unambiguous statement about your study. For instance, “patients and providers should consider the potential for serious adverse cardiovascular effects of treatment with rosiglitazone for type 2 diabetes“.
Don’t write “require further studies” because it is both unnecessary and meaningless. No single study can reach the truth.
Principle 5: concise and clarity. This is a default principle because you are allowed to write your “story” in 200 to 300 words. You have to be very strategic choosing words and sentence construction. Consider the following two statements, “Several recent studies have suggested that type II diabetic patients on rosiglitazone treatment have a greater risk of cardiovascular morbidity and mortality“, and “In type II diabetes, rosiglitazone treatment is associated with a greater risk of cardiovascular morbidity and mortality“.
Checklist: Before moving to another section, ask yourself the following questions in relation to your abstract:
- Have I stated the background and the research question/aim clearly?
- Have I concluded with an answer to the research question?
- Have I stated the design of the study?
- Have I reported the number of participants/animals/subjects and how they were ascertained in the study?
- Have I reported the method of measurement?
- Have I defined the primary outcome and how was it analyzed?
- Have I included key numerical results relating to the outcome?
If you can confidently tick “yes” to all of the above questions, then you can start writing the next section.
Another important but probably less attentive entity is the cover letter. In the cover letter, authors need to provide a brief background of the study, a key finding, and explain why the finding is important. Also, you need to explain why you choose the journal for your paper. It is useful that you tell editors the problem/disease you study is clearly the most important, but you should write it in a way does not require fancy statistics or superlatives.
Consider this, “Osteoporosis is a degenerative bone disease marked by overresorption of bone by osteoclasts”, and this, “Osteoporosis is the 54th leading killer of Americans and is a major public health threat for an estimated 44 million Americans, or 55 percent of the people 50 years of age and older. In the U.S., 10 million individuals are estimated to already have the disease and almost 34 million more are estimated to have low bone mass, placing them at increased risk for osteoporosis.” (Ulhma Neill, JCI). The first statement is much more concise than the second statement which is filled with so many statistics.
The cover letter should have no more than 5 paragraphs and it should get straight to the point. I have seen cover letters with phrases like “As you know“, “As you are aware“, “the topic is hot“, “the finding is sexy” (yes, sexy!) I don’t think you need to use those phrases in the cover letter, because it does not mean much. As you know? No, I don’t know, you tell me. You may think your paper is sexy, others may think differently.
Here is a typical example of a cover letter:
In summary, title, abstract, and cover letter are 3 important entities of a scientific paper. I have provided a series of principles that you can use to come up with an impact title and a good abstract. I hope that these principles (and tips) are helpful to you. I will come back with a series of notes on how to compose a scientific paper section by section within the IMRaD framework.