Discussion - Welcome to Research Informatics & Data Management

Discussion - Welcome to Research Informatics & Data Management

Number of replies: 12

Welcome to Research Informatics & Data Management, a one-week intensive course built around an important question:

How do we know that the data we use to generate evidence can actually be trusted?

This is not a traditional research-methods or statistics course. Instead, we will focus on what happens before the analysis—how a research question becomes a set of data requirements, where those data come from, what happens to them along the way, and how data quality, meaning, governance, and bias can ultimately affect our findings.

One idea will guide us throughout the course:

A research finding is only as trustworthy as the information pathway that produced it.

We will work through five connected modules. We start by translating research questions into clearly defined, computable data requirements. From there, we explore healthcare data sources, provenance and lineage, data models and dictionaries, metadata, data quality, missingness, and fitness for purpose. We will also look at governance, privacy, ethical use of data, FAIR principles, and an issue that is sometimes overlooked: how bias can enter our data long before statistical analysis begins.

We will then bring these concepts into LMIC and resource-constrained settings, where research may depend on a mix of paper and digital records, fragmented systems, inconsistent patient identifiers, and limited interoperability. The goal is not to lower our standards, but to think practically about how we can create reliable and sustainable research-data processes within these realities.

You will put these ideas into practice through the “Can We Trust the Dataset?” exercise and conclude the course by developing an applied Research Informatics Plan.

As you work through the course, I encourage you to move beyond simply asking, “Do we have the data?”

Instead, ask:

Where did the data come from? What do they really mean? What happened to them along the way? Who might be missing? And are they actually fit for the question we are trying to answer?

That way of thinking is at the heart of Research Informatics.

In reply to First post

Re: Discussion - Welcome to Research Informatics & Data Management

by Aromolaran Precious Adebisola -
This course description hits on a critical reality that is too often overlooked in health research: data quality and reliability are forged long before we open a statistical software package.

In resource-constrained and low- to middle-income country (LMIC) settings, this focus on the "information pathway" is particularly vital. When you are navigating hybrid environments—where paper charts, fragmented digital tools, parallel reporting registries, and inconsistent patient identification systems overlap—the gap between what a variable is intended to measure and what is actually recorded can widen rapidly.

Tracing data lineage, evaluating fitness for purpose, and uncovering structural bias early on isn't just about protecting internal validity; it's an issue of health equity. If missingness disproportionately affects specific patient populations or if local documentation workflows distort clinical meaning, our downstream evidence will be compromised no matter how advanced the analytical modeling is.

I’m looking forward to diving into the "Can We Trust the Dataset?" exercise and building out the Research Informatics Plan. Thinking through practical governance and data-provenance strategies that maintain rigorous standards while remaining feasible within real-world clinical realities is exactly where the field needs to focus.
In reply to Aromolaran Precious Adebisola

Re: Discussion - Welcome to Research Informatics & Data Management

by Purshottam Hoovayya -
The place where I work as a hospital administrator, I have observed that the quality of hospital data depends heavily on the quality of clinical documentation. For example, incomplete patient histories, missing diagnosis details, or poorly documented treatment and discharge summaries can affect patient care as well as the reliability of research.

Even when data are available in the Hospital Information System, they may not accurately reflect what happened during patient care if the original documentation is incomplete or inconsistent. This can lead to misleading conclusions when analysing clinical outcomes, hospital performance, or disease trends.

This is why research informatics is important. We need to improve documentation practices, standardise data entry, and understand how information moves from clinical records to research datasets. Reliable analysis begins with reliable documentation, not with statistical methods alone.

This is why I find research informatics important. Before analysing hospital data, we need to understand where the data came from, how they were recorded, and whether they are complete and consistent. I look forward to learning practical ways to improve data quality so that the evidence generated at our hospital can support reliable research and better patient care.
In reply to First post

Discussion - Welcome to Research Informatics & Data Management

by Ishola Ridwan Femi -

I find the idea that “a research finding is only as trustworthy as the information pathway that produced it” very relevant to Health Information Management. In malaria surveillance, having data is not enough; we need to know where the data came from, how it was recorded, validated, and reported. This is important at OAUTHC, where paper and electronic systems may coexist and issues such as missing information, duplicate records, inconsistent identifiers, and delayed reporting can affect data quality.

This course has helped me understand that reliable research depends on reliable data. These concepts will guide my applied informatics project on improving the quality of malaria surveillance data at OAUTHC.

In reply to First post

Discussion - Welcome to Research Informatics & Data Management

by Salako Omolabake Fathiat -

Hi everyone!

I’m Omolabake, and I’m so happy to be here with you all.

That line about a finding only being as trustworthy as the path that produced it really touched my heart. Working as a nurse and researcher, I see that behind every value or diagnosis in a chart, there’s a real person and a story. 

Honestly, it’s easy to focus on getting the answers, providing care objectively and only doing what make everyday better. But, this course is a gentle reminder to slow down, care for the data, and make sure we’re truly honoring our patients' journeys. And more importantly, to never lose the humanity in our work.

I’m really looking forward to learning about research informatics because research is a huge part of what I do!

In reply to First post

Discussion - Welcome to Research Informatics & Data Management

by Khadijah Adefunke Jimoh -

“Research findings are only as trustworthy as the information pathway that produced them.”

As someone who has collected and managed data for randomized controlled trials and other research studies, this statement deeply resonates with me. Research findings can influence clinical decisions, health policies, and ultimately the lives of patients, yet the reliability of those findings depends on the quality and integrity of the data collected along the way. In my experience, even seemingly small errors in data collection, documentation, entry, or management can affect the accuracy of a study's findings and potentially have far-reaching consequences.

This has made me appreciate that data integrity is not simply a technical responsibility for a data manager. It is an essential part of research ethics. Every piece of information collected represents a participant's experience, and ensuring that it is accurately captured, appropriately managed, and protected is part of our responsibility to participants and to the wider scientific community. The more I have worked with research data, the more I have understood that trustworthy evidence begins long before statistical analysis—it begins with a trustworthy information pathway.

In reply to First post

Discussion - Welcome to Research Informatics & Data Management

by Ishola Ridwan Femi -

This introduction highlights an important point: having data does not automatically mean that the data can be trusted. In healthcare, information passes through several stages, from the patient encounter to documentation, data entry, validation, and reporting. Errors or missing information at any stage can affect the reliability of the final dataset and the findings produced from it. In my experience, issues such as incomplete documentation, duplicate patient records, inconsistent identifiers, and delayed reporting can affect data quality. This course will help me better understand how data provenance, metadata, data quality, governance, and fitness for purpose can be applied to identify and reduce these problems.

I particularly agree with the idea that a research finding is only as trustworthy as the information pathway that produced it. Understanding that pathway is essential for producing reliable evidence and making better healthcare decisions.

In reply to First post

Re: Discussion - Welcome to Research Informatics & Data Management

by Victorine Maikem -
Hello Everyone, I think one thing that really stands out for me is the questions that we ask, to capture the nuances in handling and managing patient data.
I find the fact that we ask beyond if we have the data, to understanding the process of obtaining that data and what it actually means for that community brings a more personalized and specific perspective to the analysis.
I look forward to the learnings in the session.
In reply to First post

Discussion - Welcome to Research Informatics & Data Management

by Emelike Prince Samson -

During my MPH thesis, I worked with data from 229 frontline PHC workers across 30 facilities. It made me realize that trustworthy findings depend on decisions made long before statistical analysis from defining variables and collecting data to checking completeness and consistency.

My experience with colorectal cancer symposium data collected for the objective of assessing attitudes towards colorectal cancer screening also reinforced the importance of understanding where health data come from and what they actually represent.

I’m particularly interested in learning how these principles can be applied practically in LMIC settings where paper records, spreadsheets, fragmented systems and limited interoperability are common.

In reply to First post

Discussion - Welcome to Research Informatics & Data Management

by Janiel Johnson -
I agree that a dataset is only as trustworthy as the process behind it, and the breast cancer treatment example makes that clear. The data came from four unlinked sources within a hospital without a unified EHR, so each variable's origin varied by record and was not consistently tracked. Some fields, most importantly "Diagnosis Date," carry inconsistent meaning because they represent different clinical events depending on the patient, and "Stage" mixes sources without a retained staging system. The research assistant standardized some values during extraction but never documented which transformations were applied where, leaving no way to verify or reproduce the dataset. Because it includes only patients captured in the hospital's own registry, patients who received care elsewhere or were lost to follow-up- likely the same patients most at risk of delayed treatment- are undercounted or missing. These gaps mean the dataset is not yet fit to answer the research question reliably; it can support exploratory review, but not a trustworthy finding, until the diagnosis-date definition is standardized, duplicate and implausible dates are verified, and missing-data handling is made explicit.
In reply to First post

Re: Discussion - Welcome to Research Informatics & Data Management

by Olatunde Olaniyi Abiodun Oluwafemi -
I cannot agree more with the idea that a research finding is only as trustworthy as the information pathway that produced it. This principle is closely related to the common saying in pathology: “Garbage in, garbage out.” If the information entering a system is incomplete, inaccurate, or poorly managed, the final result may also be unreliable, regardless of how sophisticated the subsequent analysis is.
As a pathologist, I recognize that no single professional has access to all the information required to reach a definitive diagnosis. Therefore, when a patient’s specimen is submitted for histopathological examination, the request form should contain accurate and comprehensive patient biodata, the details of the requesting physician, and relevant clinical information. Where available, previous radiological and pathological reports should also be provided, as they will assist the pathologist in interpreting the specimen and arriving at an accurate diagnosis.
The specimen must then be submitted for gross examination, with attention to whether it was placed in the appropriate container and adequately fixed. It subsequently passes through several stages, including tissue processing (manual or automated), followed by embedding, sectioning, flotation, drying, staining, and mounting with a coverslip. These processes should be performed by trained and competent laboratory personnel, including medical laboratory scientists and technicians, under appropriate supervision.
Each stage of this workflow can affect the quality and reliability of the final pathology report. Errors in patient identification, incomplete clinical information, inadequate fixation, poor tissue processing, or improper staining may compromise the interpretation of the specimen. Similarly, in research informatics and data management, problems with data collection, documentation, storage, transformation, or interpretation can influence the validity of the final research findings.
The histopathology workflow is therefore analogous to the five connected modules discussed in Research Informatics and Data Management. Both emphasize the importance of clearly defined requirements, reliable sources, accurate documentation, appropriate processing, quality assurance, governance, and fitness for purpose. This reinforces the need to look beyond the question of whether data are available and to ask where they came from, what they mean, how they were handled, and whether they are sufficiently reliable to answer the research question.
Overall, the histopathology workflow demonstrates that the reliability of a final diagnosis or research finding depends on the quality, accuracy, completeness, and proper management of information throughout the entire pathway
In reply to First post

Discussion - Welcome to Research Informatics & Data Management

by Iwuanyanwu Anselm Chizurum -

Hi. Thank you for the warm welcome 

The framing that "a research finding is only as trustworthy as the information pathway that produced it" cuts directly to the core of what often gets overlooked in traditional research design. In practice, so much emphasis is placed on downstream statistical modeling, while the upstream realities—system fragmentation, inconsistent identifiers, and undocumented transformations—are treated as mere administrative details rather than critical determinants of evidence validity.  

I am particularly looking forward to Module 5 and the focus on resource-constrained environments. Designing robust, reproducible data architectures amidst hybrid paper-digital workflows and interoperability gaps is where informatics theory meets real-world necessity. Asking not just "do we have the data?" but critically evaluating provenance, semantic preservation, and who might be systematically excluded is a discipline I am excited to sharpen through the applied exercises.  

In reply to Iwuanyanwu Anselm Chizurum

Re: Discussion - Welcome to Research Informatics & Data Management

by Chamberlain Nwanne -
Hi Anselm - Please excuse my delayed response.
You raise an important point about how much attention we give to statistical analysis while overlooking the information processes that produce the data. In resource-constrained settings such as ours in Nigeria, these challenges become even more significant, particularly when paper and digital systems coexist within the same system.
One question worth considering is: How do we design a trustworthy research dataset when the underlying information systems are fragmented and some patient encounters may never be captured? That is where research informatics thinking becomes especially valuable.