Why most research manuscripts get rejected by leading medical journals?- Part 2: Objectives and study design

In the first part of this series here, we saw a general overview of the reasons why most articles on medical research submitted to indexed journals with a high impact factor get rejected. We will now get into the details of each with some examples.

Perhaps, the commonest yet most overlooked causes for rejection are the study objectives and design. Several factors need to be considered before the objective(s) and design/type of the study are decided. Objectives and study type and design go hand-in-hand and cannot be discussed exclusively from each other. What often happens is that researchers either have data that they wish to get published as a retrospective study (because I have good and large amount of data, why not publish it!), or they observe some trend in their clinical practice and think this can be conducted as a prospective study and published. However, in reality, these are endpoints and not starting points for a study. I will start with a simple example that all of us can relate to, and follow the same example to see how the objectives and study type/design need to be planned. Please note here that this is a hypothetical example and the data mentioned regarding the outcomes is not necessarily true. Further, in this part, we will focus on a retrospective study and will take up the prospective studies in part 3.

Let us say you wish to publish a study that shows metformin still plays an important role in diabetes management in the era of DPP4s. You might already have the data of several patients on metformin or a DPP4 alone or on a combination of both, or were switched from metformin to DPP4, or were on one of them and the other was added. Let us assume the data shows better glycemic control with a combination of DPP4 with Metformin rather than DPP4 alone. Since the data already exists, it is handed over to a statistician and after the analysis is done, the manuscript writing starts with an objective to submit it to the best journals with the highest impact factor. This is exactly where the plot goes wrong. A simple search might show that there are already numerous studies that have reached the same conclusion. Why would the journal want to publish your data? At the same time, since there are many studies with similar outcomes, it is obvious that despite similar outcomes numerous such studies were published. This is where the study design becomes important.

A good way to start is to take a look at the studies with identical outcomes (better glycemic control with a combination of DPP4 and Metformin rather than DPP4 alone) published in the leading journals with high impact factors say in the last 10 years, especially those that have been cited frequently, and to go through the methodologies and limitations of these studies. This not to replicate the study designs that have been followed, but to know what parameters make each of these studies different from each other, although their conclusions appear similar. The outcomes might have been better/poor in some patient profiles than in others or there might be a correlation of the outcomes with some factors. Some examples could be:

1. Age group
2. Male/female sex
3. A particular DPP4 rather than DPP4 as an entire class
4. Region/race/ethnicity of the subjects
5. Number of years since the diagnosis of diabetes
6. Duration of followup
7. Associated secondary outcomes like renal function, cardiovascular events, weight loss/gain, lipid profiles, etc.
8. Association with coexisting factors like weight, physical activity, etc at baseline.
9. Pre-existing factors that could have affected the outcomes e.g. hypertension, history of myocardial infarction, etc.

Now with a fair idea of what data is already available and where the gaps are, it might be good to go back to your data and think about what can be the highlight of your study based on the data you have and what is already published. As an example, maybe most studies have included all the DPP4s as an entire class of drugs. Whereas in your data maybe 50-60% of the cohort was on a particular DPP4. That can be the strength rather than the weakness of the study. It might make sense to include only this cohort and leave out the rest so that the strength of the data is robust. Another example - maybe it has been already published that the combination of metformin with DPP4 shows better outcomes in patients who were obese at baseline. Can you make it more specific by analyzing the outcomes by different BMI levels rather than obesity in general? Again, it would depend on your data- what parameters have been monitored and documented that can be used to create a study design that makes a meaningful contribution to research, which the journal editors think their readers will be interested in. The flipside here could be that if you filter down the data to such levels, your sample size might become very small, which will discuss in the next paragraph. However, the positive is that you now have a clear focus for your study, which means a higher chance for publication, and if the sample size is small, it is better to wait until you have data from more patients before starting work on the analysis. Once there is clarity about the focus area, it is important to have a research question (study hypothesis) in mind, as that will decide the design that would be best. Some examples could be:

1. Metformin + DPP4 has better glycemic control in those with BMI 25-30 kg/m2 than in those with BMI > 30 kg/m2
2. Metformin + DPP4 has better glycemic control when started at the time of diagnosis than when the 2nd drug is added later.
3. Metformin + DPP4 has better glycemic control in those with diastolic blood pressure <80 mm Hg than in those with diastolic BP> 80 mm Hg or Metformin + DPP4 abcgliptin has better glycemic control in those with diastolic blood pressure <80 mm Hg while Metformin + DPP4 xyzgliptin has better glycemic control in those with diastolic BP> 80 mm Hg
4. Among patients who did not show response to Metformin or DPP4 montherpay, more patients below 40 years of age were prescribed Metformin + DPP4 abc gliptin while those above 40 years were prescribed Metformin + DPP xyz gliptin.
5. Metformin + DPP4 showed improved glycemic control within 3 months in those with normal serum creatinine while the same results were seen in 6 months in those with elevated serum creatinine.
6. Metformin + DPP4 showed improved glycemic control in patients with no hyperlipidemia but not in those with high lipid levels.

Among the examples of study hypotheses above, you can see that for the qts 5, at least a 6 months' data is required (longitudinal study). In example 4, data a single timepoint would suffice (cross-sectional study). For examples 1-3, the researcher would need to decide the cutoff time period that will be considered for the study (cohort studies). Example 6 compares the outcomes in those with the presence of a particular risk factor vs those who do not have that risk factor. (case controlled study).

Next, the sample size and statistician comes in. You need to provide him/her details of the focus and objective of your study as also the study hypothesis, and share the available literature so that the statistician can suggest an appropriate sample size with a rationale for the same. Not all journals ask for the rationale for sample calculation for a retrospective study, but many do. If you wish to target specific journals for your manuscript, it is good to go through the journal guidelines at this stage to know if sample size calculation is required.

At the end of all the above work, you now clearly know what is the data and information required if you wish to submit the planned study for publication in an indexed journal. Thus, there is clarity on what is already available with you, what is useful,what should be discarded, what is still missing.

We will look at prospective studies in the next part of this series.

Link to part 1: https://www.dhirubhai.net/pulse/why-most-research-manuscripts-get-rejected-leading-medical-dhanuka

Link to part 3: https://www.dhirubhai.net/pulse/why-most-research-manuscripts-get-rejected-leading-medical-dhanuka-2e/