Interventional versus observational study – moral contention or mere semantics?

Vijayaprasad Gopichandran, 15.06.2023

 

I am part of the scientific review committee of our institution and today we had a meeting to discuss research proposals and review them. The meeting ended on an academically surcharged note. I will present the research study which was the cause for this interesting debate and then lay out the premise of the debate. 

 

One researcher from the department of gynecology proposed a study in which she wants to explore the association between obesity and puberal menorrhagia, excessive bleeding during menstruation in adolescent girls. She said that she would enroll adolescent girls who have pubertal menorrhagia in the study and assess whether they are obese or not. The hypothesis was that pubertal menorrhagia is more common among obese girls. In addition to this as a second part of this study, she said she would give these adolescent girls standard treatment for the excessive bleeding and then see whether the obesity influences response to the treatment. The researcher said that this is an observational study. 

 

The debate started when I intervened and said that this is not an observational study, but it is an interventional study. I said it is an interventional study because:

1.     The study looks at the response to treatment (the treatment being the intervention) and how the response to treatment is influenced by obesity.

2.     Without the treatment (intervention) there is no study

3.     While the first part which looks only at association between obesity and occurrence of excessive bleeding during puberty is an observational study, the second part which looks at the response to treatment is an interventional study. 

 

The head of ophthalmology, the head of emergency medicine and faculty of community medicine, all of whom are members of the scientific review committee argued against this notion. Their arguments were:

1.     These adolescent girls with pubertal menorrhagia would be treated anyways, irrespective of the study. The study only observes the outcomes. 

2.     The girls are not recruited to the study for the intervention. The intervention would be given anyways. They are recruited only for observing the outcomes. 

3.     As the treatment would be given anyways, they do not incur any additional risks because of participating in the study. 

 

Subsequently another faculty from department of ophthalmology and the biostatistician member of the committee all joined in the debate and the debate became more exciting. While there are valid points on both sides of the argument, I would like to express my views on this in greater detail starting from the basics of what is an interventional study. 

 

What is an interventional study?

 

In epidemiology, we broadly have two types of studies. One is observational study and the other interventional study. The main difference between these two types of studies is that while in observational studies the researcher does not manipulate or control the environment in which the medical phenomenon is unraveling, in interventional studies the researcher manipulates the research environment by producing or removing the exposure and controlling some or many conditions in the research environment. 

 

A typical example of an observational study is the first part of the study described above. The study of association between obesity and pubertal menorrhagia. The researcher in this case only watches the phenomena of obesity and pubertal menorrhagia. She observes whether these two are associated with one another. She does not manipulate the research environment in any way, she does not control the research environment. She just watches as things happen on their own. In the second part of the study, this is not the case. She enters the research environment and implements a change. She creates a change in the research environment by providing a treatment to the girls to stop or reduce their bleeding. Then she watches them as they respond to the treatment and studies whether the response to the treatment is influenced by obesity. While in the former study she just watches, in the latter, she enters the research environment, does a manipulation and observes the outcome. Observation of outcomes is common in both the settings, but the difference between phase 1 of the study and phase 2 of the study is that in phase 2 she observes after introducing an intervention. She herself introduces the intervention. The intervention is done with an intention to observe whether obesity influences the outcome of the intervention. 

 

What is a therapeutic intervention? 

 

An intervention is either a positive or a negative action in the treatment of a patient. In case I provide a treatment for a disease, it is a positive intervention. Alternatively, I could be removing a risk factor, or reducing the exposure to a factor, these are negative actions. A patient may have diabetic retinopathy and I may give an injection into the eye to treat this condition. This is a therapeutic intervention. A person may be a chronic smoker, and as a result of smoking might have developed chronic lung disease. If I help the patient stop smoking, it is a therapeutic intervention. However, these interventions are treatment and prevention interventions that are done routinely in practice. Are they also research interventions? To answer this question, we must know what is a research intervention? 

 

What is a research intervention?

 

A research intervention is manipulation of the research environment as described earlier. All research interventions are either positive or negative therapeutic or preventive interventions, but all treatment or preventive interventions are not necessarily research interventions. The difference between a treatment and research interventions is that, treatment interventions / prevention interventions are part of everyday clinical or public health practice. The participant in a treatment or preventive intervention comes with the expectation of cure or prevention. They submit themselves to the intervention in a belief that it will help them and they will become better. The main beneficiary of a therapeutic or preventive interventions is the patient. The outcomes of the interventions benefit the patient and they don’t lead to new knowledge. They don’t provide generalizable results that can benefit the society at larger. On the other hand, research interventions are done with a dual aim, namely benefit to the patient and then larger benefit to the society through the knowledge gained from the participating patient. Many times, research interventions are unproven and uncertain and only after the research is completed one can conclude whether the intervention works or not or on whom the intervention works better. In fact, if an intervention is well proven and its parameters well known, then it is not permitted to do research on that intervention, rather it must be offered as a definitive treatment. 

 

Given this major difference between a therapeutic intervention and a research intervention, let us explore the second part of the proposed study on obesity and pubertal menorrhagia. The researcher wants to give a therapeutic intervention which is routinely practiced. But she wants to do a study on who responds better, obese girls or non-obese girls. This implies that there is a certain level of uncertainty on who would respond better to this therapeutic intervention. Because of this uncertainty, and because she wants to resolve this uncertainty, this therapeutic intervention becomes a research intervention in this context. If the effectiveness of the treatment was established beyond doubt, there would be no scope for research there. The very fact that she wants to do research to study the effect of obesity on treatment response, implies that there is a certain level of uncertainty, and she wants to resolve this uncertainty by conducting this research and generating new knowledge. Therefore, it is a research intervention. Any research which comprises of a research intervention is interventional research. 

 

Another study which had a similar controversial debate was one in which an ophthalmologist wanted to give injection of a certain drug in the eye and study the various factors which influence the response to this injection. The same argument as above can be applied to this study also. The injection into the eye is established treatment. Therefore, the ophthalmologist argued that it is only an observational study, as the injection would be given routinely anyways as treatment. They argued that the intervention is not for the sake of the study. However, I would like to pose the same argument as above. The therapeutic intervention of injection into the eye, becomes a research intervention the moment that they have an uncertainty of which factors would favorably influence the response and which factors would adversely influence the response. The reason the researcher is conducting the study is to resolve this uncertainty. Therefore, the therapeutic intervention is no longer a simple treatment procedure, but a research intervention with a certain level of uncertainty that needs to be resolved and new knowledge generated. 

 

The community medicine faculty raised another interesting question during this debate. If a researcher does an audit of admissions to the emergency department of a hospital to understand the mortality rates and agreement between predicted and observed mortality rates, can it be called an interventional study? He argued that anyways these patients admitted to the emergency department would undergo some form of treatment. So, is that an interventional study too? My response to this would be to again go back to the nature of the intervention. Is the intervention therapeutic, or is it done for research? When I am studying an audit of mortality rates in an emergency department the focus of my enquiry is overall performance of the emergency department and not any treatment intervention. People with different emergency conditions may attend the emergency department. Some may be there for accidents and trauma, some for acute coronary events, some for snake bites and poisoning etc. The treatments are varied. The effect of treatment, or the factors influencing treatment are not the focus of enquiry, but the overall performance of the emergency department is the focus. Therefore, the treatments provided in the emergency department are therapeutic interventions and not research interventions. The interventions are not provided to generate knowledge about the intervention per se, but to help patients. The research question is broader and looks at whether the emergency room performs well. Therefore, it is not an interventional study. 

 

The biostatistician argued that sometimes students perform observations on outcomes of treatments provided by some other experts in their respective departments. For example, a student may want to study the outcomes of laparoscopic cholecystectomies done in her department. She is not performing the laparoscopic cholecystectomies. But she is only making observations of the outcomes such as duration of hospital stay, fever post operatively, port site infections, etc. Is this an observational or an interventional study? I would go back to the question of whether the laparoscopic cholecystectomy is a therapeutic intervention alone or is it also a research intervention. The research is being done because there is a genuine uncertainty about what would be the outcomes of the laparoscopic cholecystectomies. Therefore, the surgeries are being performed to see the response or success rates or occurrence of complications. Therefore, they are no longer just therapeutic interventions, but they are research interventions with genuine uncertainty about how many of them succeed, how many fail, how many have complications, which the student is studying. Therefore, it is an interventional study. 

 

Therefore, when there is a doubt whether a study is an interventional or an observational study, I propose the following framework for assessment. Ask the following questions serially:

1.     Is there an intervention in the study? If no – then it is purely an observational study. If yes, then ask next question. 

2.     Is the intervention an established treatment? If no – then it is purely interventional study. If yes, then ask next question. 

3.     Is there genuine doubt about any aspect of this treatment such as what factors influence response to it, on what type of patients it will work, whether it will work in the current study context, what are the side effects or complications and is that the current research question? If yes – it is interventional. 

 

The debate closed with a lot of uncertainties, and we all agreed to disagree on several aspects of the debate. My last point is, in the research world, it really doesn’t matter what we call it. Whether we call it an interventional study or an observational study, the rigors required are the same. The ethical protections are the same. So, this is largely an argument over semantics. But the debate helped me think this through clearly and articulate them as this essay. I am certain that the debate has equally strong points on the other side too. What matters ultimately is that the research is done rigorously and ethically.