Looking beyond hospital-based studies for postgraduate dissertation

 Invited guest lecture at the Research Methodology Workshop – Part I Protocol Writing for Postgraduates at the Aarupadai Veedu Medical College & Hospital, Vinayaka Missions Research Foundation, Puducherry on 07.02.2023. 

 

Vijayaprasad Gopichandran

Assistant Professor

Community Medicine

ESIC Medical College and Hospital,

KK Nagar, Chennai 600078

vijay.gopichandran@gmail.com

 

Acknowledgment: I would like to acknowledge the useful inputs of my colleague 
Dr. Vignesh Dwaraknath, Assistant Professor, Community Medicine at the ESIC Medical College & Hospital, KK Nagar, Chennai 600078 and Dr. Sudharshini Subramaniam, Associate Professor, Community Medicine at Madras Medical College, Chennai 600003 towards strengthening the arguments for conducting dissertation research studies outside of the hospital settings. 

 

 

Completing a research study and submitting a dissertation report is essential for obtaining a post graduate qualification in any specialty. The postgraduate resident is required to develop a research idea, design a study, conduct it, collect data, analyze it, and write up the findings in a comprehensive dissertation report. For most postgraduate residents working in clinical specialties, it becomes a norm that they carry out their research studies in the hospital setting. They develop research questions from their everyday work in the hospital, design a hospital-based study, collect date there and complete the study in the hospital. For those residents who work in the laboratory-based specialties like pathology or microbiology, their study setting is also often the hospital which feeds samples to the laboratories. Residents in the department of community medicine and rarely residents from some clinical departments like pediatrics and obstetrics and gynecology, conduct studies in communities. 

 

In this essay, I will try to outline the advantages of looking beyond the hospital settings for conducting postgraduate dissertation research studies. I will provide a brief description of where all we can look beyond the hospital for our studies and highlight some examples of research studies which can be conducted better outside the hospital settings compared to inside the hospitals. The main aim of this presentation is to showcase the strengths of moving beyond hospital settings for post graduate dissertations.

 

When we say, looking beyond hospital-based studies, we mean conducting studies in communities, schools, colleges, workplaces, virtual spaces (emails, telephone calls, video calls), public spaces like restaurants, bus stops, railway stations, shopping malls and coffee shops. Some of these places like school, colleges, workplaces offer us the advantage of having a database of potential participants from which we can sample randomly. 

 

Hospital based studies usually suffer from several limitations. For example, if we enroll patients into a clinical study from a hospital, it is almost impossible to do a random sampling. This is because usually there is no database of patients admitted to or attending the outpatient department of the hospital with the condition of interest. Therefore, usually the sampling is non-probabilistic, with consecutive patients coming with the condition of interest being enrolled. Such non-probabilistic sampling methods greatly limit the internal validity of the research. The non-probabilistic nature of the sample may introduce sample selection bias. Since the sample is non-probabilistic, it is also possible that their parameters are non-normally distributed, thus limiting the application of many statistical tests. A particular demography of patients usually frequent certain hospitals. For example, patients attending a government general hospital are usually from the poorer social-economic class. However, patients attending the corporate hospital settings usually belong to the upper class. Therefore, the sample is unlikely to be representative of the general population. This in turn precludes generalizability of the findings. When we do hospital-based studies there is, yet another important problem known as Berkeson’s bias. Let us take the example of a study to explain association between smoking and deep venous thrombosis. Cases are persons with DVT, and controls are persons without. If both cases and controls are taken from the hospital, then controls are also likely to have increased probability of smoking, because the smoking can lead to other conditions which drive patients to hospitals, e.g., lung diseases, coronary artery disease etc. However, when the study is done in the community, it gives accurate findings as this hospital-based sampling bias is removed. The Berkson’s bias is a major limitation of hospital-based studies which try to understand risk factors for an illness. 

 

These limitations are not the only reason for why we should look beyond hospitals for your dissertation. There are several inherent advantages of doing studies beyond the hospital. People, especially from the poor sections of the society and from rural areas, come to hospitals only in advanced stages of the disease. Therefore, hospital-based studies are biased by looking at only advanced end of the spectrum of any disease. If you are doing a test of diagnostic validity, this will seriously distort the results, as the sensitivity and specificity of the test will only be applicable to that end of the spectrum. Many times, patients have illnesses which are undetected. For example, a patient may have a cervical cancer, but it may be undetected. In a community-based study which actively screens for cervical cancer, even such undetected patients may have the chance of entering the study. Therefore, community-based studies ensure fairness and justice. When studies are conducted in communities, workplaces or schools and colleges, they are designed and developed in a manner that addresses the needs of the community and therefore are more relevant. Reaching out to people at their own homes, schools of workplaces empower the people and enables them to engage more actively in the research. In case of clinical trials, community-based trials are more pragmatic and realistic compared to conducting trials in the in-patient settings where patients are under close observation. The pragmatic context of the study gives us the true effect of the drug or device, which is more relevant for using it among people in the community after the drug is approved. These make community-based studies advantageous. 

 

Let us take some examples and discuss the differences of conducting those studies in hospitals versus communities. 

 

1.     What is the burden of lower limb chronic venous insufficiency in India? 

 

A postgraduate may design this study in the surgical OP. However, the prevalence of lower limb chronic venous insufficiency will be over-estimated because such patients would seek treatment in hospitals more often and therefore the probability of finding them is more in the hospital setting. If the postgraduate goes out to conduct this study in the workplace, there again, based on the nature of the work that is done there, the probability of finding persons with chronic venous insufficiency may be more in some workplaces involving prolonged standing. The most appropriate setting for doing this study would be the community setting. 

 

 

2.     What proportion of adults in India have vaccine hesitancy against COVID 19 vaccines?

 

A postgraduate may decide to conduct this study among those attending the OPD in the hospital. However, people attending hospital OPD may have comorbid conditions like diabetes and hypertension. People with these comorbidities may have a higher self-perception of risk and therefore have lesser vaccine hesitancy. Moreover, the hospital demographics usually are comprised of people in middle and older ages with less people from younger age groups. Usually, older individuals have a higher self-perception of risk for COVID 19 and therefore lesser vaccine hesitancy. Therefore, a hospital-based sample may grossly under-estimate the level of vaccine hesitancy. The postgraduate can sample from vaccine camps that are conducted in various places like malls, railway stations, bus stops etc. However, even here, only those people who are motivated to get vaccinated would reach these spots and therefore it may again under-estimate the level of vaccine hesitancy. Here again, the best setting to conduct this study would be workplace, or communities. 

 

3.     What is the incidence of device associated infections in Puducherry? 

 

A postgraduate cannot conduct this study beyond the hospital settings. People on urinary catheters, intravascular catheters, endotracheal tubes etc. are usually admitted in the hospitals. Very rarely patients may be on chronic in-dwelling urinary catheters at the home setting in the community. Moreover, device associated infections usually are due to hospital acquired microorganisms which thrive only in the hospital settings. Thus, this research can be conducted only in the hospital setting and cannot be taken beyond the hospitals. 

 

4.     Do children with bronchial asthma have a higher prevalence of stunting compared to children without bronchial asthma? 

 

 

A postgraduate may design this study from the pediatrics outpatient department and from children admitted in the pediatric ward. The postgraduate may pick up children with bronchial asthma from the ward and OP and children without asthma also from the ward and OP. However, this will underestimate the difference in nutritional status between the asthmatic and non-asthmatic children. This is because the children without asthma who are included in this study are not normal children. They are children attending the hospital for some other condition, which may itself lead to stunting. If the postgraduate designed this study to be conducted in schools, the postgraduate may have to cover many schools to get the required number of children with asthma. One alternative would be to get the children with asthma from the hospital and normal children as age and sex matched controls from the community or from schools. Here again, choosing controls from beyond the hospital setting will only give rise to meaningful results.

 

5.     Do working women have a higher incidence of low birth weight compared to home makers?

 

The postgraduate may take the list of women who had normal delivery and low birth weight delivery from the nominal register of the labor room of the medical college hospital and conduct the study by interviewing them for their job status. However, it is important to note that the Berkson’s bias will operate here. The other factors that influence the women to deliver in a medical college tertiary care hospital will also influence the birth weight. Therefore, it is difficult to prove that the low birth weight is indeed due to the work status of the woman. It would be better to sample from the nominal register of the ANM working under the PHCs attached to the hospital. Here the postgraduate will get normal women who deliver in PHCs as well as women who are referred and deliver in tertiary care centers. This will remove the Berkson’s bias. 

 

From these examples and the above discussion, I would like to posit that when it is feasible and meaningful to conduct a study outside the hospital setting, the postgraduate must attempt to conduct it there. There are very few conditions which require to sample from the hospitals. For example, if the study is on critical care patients, or events that occur typically in a hospital setting like surgeries and deliveries, there is no choice but to conduct those studies in the hospital. In all other situations, the postgraduate must consider looking beyond the hospital to conduct the study. While deciding on whether to conduct the study in the hospital settings or beyond, the postgraduate must consider the following series of questions:

 

 

1.     Who is the target population? Are they hospital patients? Are they the general population?

2.     Can a representative sample be obtained from the hospital? Do patients attending the hospital represent the target population?

3.     Will hospital-based sample bias the study? How will it bias the study?

4.     Are resources (time, money, human resource) available to conduct the study beyond the hospital settings?

 

Answers to these questions must guide the postgraduate. If these considerations allow the postgraduate to go beyond the hospital settings, the postgraduate must make those efforts.