From Numbers to Narrative
A practical guide to bringing health data to life through storytelling
By Christian Otte and Michal Kapral
This resource is for health organizations that want to publish reports, data or insights to harness the power of storytelling and bring data to life. Drawing from journalism techniques and research on psychology, sociology and science communication, From Numbers to Narrative walks you step by step through the process of crafting high-impact, data-driven health reports and insights.
The Power of Storytelling
With a data storytelling approach, your readers will be able to easily understand, absorb, and act on the data you’re reporting.
Why are health reports often dense and boring? While they might be factually correct and contain nuggets of compelling data, they tend to have one glaring missing element: a storyline. Without a narrative, readers will quickly lose interest and won’t be able to gain insights from the data, nor will they be compelled to act on it.
4,100 years ago – The Epic of Gilgamesh, the first surviving written work of complete storytelling, is inscribed on clay tablets in Mesopotamia
22 years ago – John Mashey, at the time chief scientist at Silicone Graphics, begins to popularize the term “Big Data”
We no longer live in a world restricted to printed reports. Digital journalism has shown what’s possible with online storytelling. There’s an opportunity for health organizations to do the same.
A powerful example of data journalism and visual storytelling: Extensive Data Shows Punishing Reach of Racism for Black Boys, The New York Times’ The Upshot
What is this guide and who is it for?
For data to be meaningful, they need to tell a story.
The guide shows you how to find the sweet spot at the intersection of big data, visualization and storytelling, to bring health data to life and drive policy and behaviour change. It’s a practical resource for people who work in health care or health organizations to banish the boredom from their data-driven reports. With a data storytelling approach, readers will be able to easily understand, absorb, and act on the data you’re reporting.
It can be challenging to shift to a new style and format of health reports. But good news – you don’t need to change everything. This guide leaves plenty of leeway for your own decisions and creativity. It also covers a range of options. You can start by making a few valuable tweaks to your current reporting format, or push your reports forward with interactive data and advanced storytelling techniques. Where possible, we base our recommendations on the best evidence available, which will help you support the changes you make to your own products.
You can use this guide to:
- Improve existing reporting
- Start a new report
- Present data that was historically just a regular report in story format
- Tackle a complex health data topic you need to communicate
The 7 Steps to Storytelling
What is storytelling with data?
“Storytelling” has many meanings. In this guide, storytelling means telling a story through the way you select, organize, describe and visualize the health data in your overall report. At the most basic level, a story has a beginning, middle and end, in which “one thing happens in consequence of another.” You can use storytelling to tell stories within the broader report storyline, by including stories from patients, family caregivers, doctors, nurses and other health care providers, and even researchers.
Health data storytelling will only be effective if you can share your narrative through writing, data visualization, and design.
The 7-Step Process for Telling a Story with Health Data
1. Defining your report strategy and goals
From the outset, you and your team should articulate why you’re producing this report. A thoughtful strategy will guide your focus as you develop the report.
Why are you creating this report now? Consider these questions:
- Have you identified an important or trending topic in health?
- Are you shedding light on a new treatment or therapy?
- Is the report looking at a specific disease or condition?
- Will you be highlighting time trends, inequity issues for different populations, or geographic variations? How will this report help improve people’s lives?
- And importantly, do you have quality data available to support the right level of insight generation?
Once you’ve established the reasons for producing the report, you’re ready to set some specific goals and outcomes for the project. These should align with the strategy and mandate of your organization. Your goals should also consider the main audience for the report.
Think about who you are trying to influence or educate with this report:
- The public?
- Health care providers?
You can have multiple audiences, but narrowing it down to one or two main audience targets will help you achieve your goals and outcomes.
2. Selecting data and generating insights
Once you have a story strategy, the next step is to identify data sets that are needed for analysis. This means collecting, acquiring or accessing the data to begin analysis. Data can include health care indicators, or qualitative data such as patient interviews.
There are many ways to select data and generate insights for a report, but it tends to be an iterative process. First, make sure you have the right data at the start, otherwise you may look at the data that’s readily available and realize you can’t get the rich, meaningful insights that you need from it.
The question-based method is one of the best ways to organize and structure insights:
- Frame a series of specific, key questions in a hierarchy that you want to answer with your analysis. For example:
- What is the overall trend of this disease?
- Does the trend vary by geography, such as by region, city, or hospital?
- Do you see different growth rates by demographics or social determinants?
- What is driving the trend? What are the related or causal metrics or indicators that impact this trend?
- Identify which questions are answerable by the data (some data may not be available or reliable enough to answer the question).
- Consider titling your data visualization with the key question. For example: How do the annual rates of hospitalizations for chronic fatigue vary across regions?
- With the remaining questions you can answer, use a variety of data analysis techniques to generate the insights.
- Understand that most of the insights produced won’t make it into the final report.
The insights you generate can take on many forms, including simple statistics, percentages, regression analysis and even predictive models. In the “Designing Your Report” section of the guide we talk about how to visually present your insights. Regardless of the data’s visual form, we recommend refining the insight until it’s simple, specific and factual. For example:
“People in the north have a much shorter life expectancy than those in Ontario overall – 2.9 years lower than Ontario in the North West LHIN region and 2.5 years lower in the North East LHIN region.”
(Health Quality Ontario, Health in the North report, 2017)
With some insights, the reader (especially an analytical one) might wonder, “What’s the reason for that statistic or insight? Why was this included in the report? What’s the underlying driver?” It’s helpful to bring the answers to these questions forward as you continue to understand your audience and build your report.
Here are a few key considerations for generating insights:
- Refine: If you haven’t published a report before, you may need to refine your process and adjust how you do the analysis.
- Align: Begin the analysis with the goal of finding insights that line up based on the focus of your report.
- Ask questions: Generate a lot of insights. We’ve listed one method of structuring your analysis below. Finding the story in the data tends to be as much a science as it is an art.
- Focus: Distill down to the most impactful and meaningful insights based on the criteria you establish. What should you include in this report and what can be saved for later or used elsewhere?
- Fit the story: Evaluate the insights to decide whether they’ll help drive your narrative.
3. Conducting user research
While you’re getting the data you need and producing a good crop of insights, it’s useful to learn more about your audience. Since user research is a broad topic, we’re not going to delve into it in detail in this guide. A good understanding of your audience is one of the most important things you can do to produce a great story within your report, as it will help you frame the report to their interests. So if possible, get a user researcher to work on the project team. Here are a few key aspects of user research:
- Define your audience: Drawing from your report strategy and goals, clearly define your audience and prepare a user research plan to engage with and understand your audience, their information needs and interests. Identify what value your report will have for each audience.
- Define your objectives: Do the research with clearly defined objectives: who do you need to understand and what is most important to understand? Based on the objectives, select the appropriate research methods, such as interviews, surveys and focus groups.
- Gather user perspectives: Even if it’s not your current practice, take steps to incorporate user perspectives and feedback into the report creation process as it unfolds.
- Develop personas: Personas can be a good way to illustrate the types of people that make up your primary audiences and inform important decisions as you work to create your report.
- Scan the landscape: Consider doing a landscape scan up front to learn more about the types of reports already available, position your report and generate ideas.
- Perform user testing: As you progress through the project, we recommend coming back to your understanding of the user on an ongoing basis and actively testing what you create with users. For example, engaging clinicians and patients will help to ensure the insights are clear, understandable and impactful.
As you move to the next steps, continue to advocate for the needs of your audience, in particular patients, so that the report is tailored to speak to them and tell them a meaningful story that resonates.
4. Building a storyline
If you’ve done research into your audience, you’ll now have a good idea how to organize the data to build a storyline that meets their needs. There are a few things that would be good to know about your audience: their health literacy, their personal perspective on the topic, and their level of current knowledge and awareness of the topic.
Here are some tips to help craft your storyline as you assemble the report:
Make your headline a key finding
The simple technique of writing your key finding into a headline is commonplace in news and magazine features, because it works.
Journalists figured out ages ago that if you want people to read your content, you need to draw them in with an enticing headline. You can also add more context with a teaser – a bit of text under the headline that makes the reader want to read more. This will also force you to figure out exactly what you’re trying to say in each of your chapters or sections. Alternatively, you can write a longer headline that summarizes the entire key finding and complement it with a secondary finding as a teaser:
There were nearly 1.3 million new starts of opioid prescriptions in Ontario in 2016
New starts of opioids decreased slightly in 2016 from 2013, by about 25,000, or 2%.
(Health Quality Ontario, Starting on Opioids, 2018)
Think about how the text flows
Graceful transitions between paragraphs will help keep the reader’s attention. “Prose is architecture, not interior decoration,” Ernest Hemingway said. Build your narrative with a logical flow between sentences and paragraphs.
You can take the easy route as we’ve done here and stick a series of titles atop each paragraph. Other transition techniques include:
- The building block: Reinforce the idea set up in the previous paragraph, and then introduce a new concept.
- The comparison: Introduce a new paragraph by comparing the data points you’ve described in the previous paragraph.
- Cause and effect: Show how the data points in the previous paragraph influence the data described in the next paragraph.
- Ninja level: Paragraphs just flow naturally and logically, without transition sentences.
Whenever possible, make the data about people
People-izing your data will make it much more relatable than reporting abstract percentages. If you have a data point that 15.1% of Canadians are current smokers, you can say “more than 1 in 7 Canadians are current smokers.” Even better, include the actual total number of people, as in this report from the University of Waterloo’s Propel Centre for Population Health Impact:
“In 2017, the overall prevalence of smoking in Canada
was 15.1%, equivalent to approximately 4.6 million
Your health data will be much more impactful if you can show how many people are involved in a statistic. For example, a report that showed an increase in the percentage of family or friend caregivers who were experiencing additional distress included the number of additional distressed caregivers:
“More than 4 in 10 long-stay clients with caregivers (43.5%) had caregivers who experienced distress, anger or depression in relation to their caregiving role, or were unable to continue caregiving, in 2017/18. That was up from 33.8% in 2014/15, and represented more than 26,000 additional distressed caregivers.” (Ontario Health (Quality), Measuring Up 2019)
Introduce a plot twist to make your reader care
Your story outline should include an element of surprise. Data storytelling consultant Jay Golden says a story “is not the shortest distance between two points. A story is a journey with a twist, a problem approached in an interesting way, that makes us care.” Another way of thinking about the twist in your report’s narrative is to find the data points that are surprising. If you have subject matter experts reviewing the data, ask them what they find most surprising. You’d be surprised how often this simple technique works to identify twists in your narrative.
For a report on palliative care, the doctor reviewing the data said he was shocked to see that more than two-thirds of patients receiving palliative care had an emergency department visit in their last month of life. This became the plot twist. Most people said they wanted to die at home, but about two-thirds of those receiving palliative care ended up in the ER near the end of their life, and most died in hospital. The report’s narrative highlighted the tension between what people wanted, and the reality of what was happening in the health system.
Chunk your text
In 1956, cognitive psychologist George Miller discovered that people could only retain about seven chunks of information in their short-term memory. To make your narrative more effective, group your information into easily digestible chunks of related material.
Key resource: How Chunking Helps Content Processing, Nielsen Norman Group
Use pull quotes and callouts
Pull quotes and callouts are invaluable devices to captivate readers skimming your reports. Take a few of the most interesting nuggets of text or data points, blow them up into a large font, and intersperse them throughout the report. These callouts can even tell a mini story of their own for a reader who does a quick scan of the report.
“THE WAY WE RESPOND TO MISTAKES AS HOSPITALS IS AS TRAGIC AS THE MISTAKES THAT HAPPENED” (Vox.com article on hospital harm, by Sarah Kliff)
5. Writing a lived-experience story
Behind all health data, there are stories about people. In health care, those stories are often extremely powerful. So powerful that they can completely alter someone’s viewpoint on a topic.
You don’t need to have a professional writer on staff to include lived-experience stories in your reports. By following a few simple techniques, you can create great stories that capture the essence of what it was like for someone at the heart of your report topic.
“Patient stories offer valuable insights that go way beyond the statistics and the outcomes: they have the power to inspire, humanize, compel action, and challenge assumptions,” says Jennifer Schipper, Interim Communications Lead at Ontario Health. “Our colleagues tell us that hearing patients and their experiences, and being able to ask thoughtful questions, really gives them insights that they wouldn’t have by just looking at published literature or data.”
While generally referred to as patient stories, these lived-experience stories can include perspectives from family or friend caregivers, health care providers, and researchers.
Anatomy of a patient story
- Conflict: “There is no story without conflict,” says patient storytelling consultant Franklin Street. Even good-news patient stories have an element of conflict if you ask questions about fears, which can help with empathy and memorability.
- Empathy: “Find elements of a story that are just uncommon enough to be unique but no so far out of space that it is difficult to create empathy,” Street says.
- Setting: Describe the scene, and don’t overcomplicate the story with too many settings.
- Plot: Take a page from Shakespeare, says Street, and tell your story in three acts: establish the characters, explore a problem they face, and then solve that problem.
3 keys to story selection
- Diverse voices: “There’s no one universal patient voice, so you want to make sure you have a process wrapped around having ethnic, age, socioeconomic and gender diversity,” says medical communicator Dr. Seema Marwaha. “While difficult, that’s worth the effort.”
- Do a pre-interview by phone or email: A brief pre-interview will give you a sense of whether the patient’s story is a good fit for your report. It will also let you gauge the patient’s comfort level in re-living potentially traumatic experiences.
- Let the patient tell their own story: Give the patient leeway to tell the parts of their story that are important to them, rather than checking off boxes related to specific data points. The most impactful patient stories go beyond medical details.
7 Best Practices for Patient Stories
- Identify your goals – What’s your purpose in telling these patient stories?
- Write an invitation letter – Explain to the patient the context of the report, what kind of stories you’re looking for, and how the stories will be incorporated into the report.
- Consider emotions – Coach the patient through the possibility that recounting their story could make them angry, regret, grief or trauma.
- Give options for story-gathering methods – Let the patient choose how they would like to tell their story, such as by phone interview, in-person interview, or email.
- Ditch the jargon – Communicate with the patient in clear language, and avoid health care jargon and acronyms.
- Obtain written consent – Get a standard release form written up for your organization, with check boxes for different types of publication and consent, such as print, audio, video, communications documents.
- Assess risks – Ask the patient if publishing their story will create of any challenges or risks, and ask them to get publication approval from anyone else in the story who might be affected.
6. Designing your report
Design plays a critical role in creating a report that tells a compelling story. When it’s done really well, the narrative, insights, data visualization, and patient stories all intertwine seamlessly. High-quality design and visual appeal is not an easy feat. Getting design resources involved early on and throughout the process will have a substantial impact on the success of the project from concept to launch.
Focus the design process on the audience and user research.
Presenting insights and data visualization
So much of telling a story with data involves presenting the insights you have in a simple, accessible and intuitive way. By its nature, data is often complex to analyze and communicate. Data visualization is both an art and a science: it’s just as important to get the graph axes and units of measure right as it is to make the graph visually appealing.
Tip: to help your audience interpret the data, consider using a question as a chart title? For instance, “How are patient infection rates trending provincially?”
We suggest always using less complex visualization types, steering clear of trendy styles such as 3D graphs and bullet charts.
Infographic-style visualizations and report formats can be effective, but typically work best in narrower applications.
The visualizations you create should also match and be from the same family. Think of it as curating a number of art pieces as part of an art gallery exhibit.
Data visualization design is a topic in itself. Instead of covering it in this guide, here are some excellent resources:
- Google’s data visualization guidelines
- User research questions for dashboard design
- Design checklist for the perfect charts
- Data visualization best practices
For online reports, you can use static images of data visualizations, though they have constraints, including a loss of quality when zoomed, and potentially sluggish download speeds if large.
There are free and relatively simple embedded chart tools such as Google Charts that present data visualizations effectively.
For the report design itself, take an iterative approach by producing mock-ups, prototypes and then testing these with users before publication, and ideally at key points along the way as the report takes shape.
Depending on the report audience, you may need one or more formats: print, downloadable PDF (Portable Document Format), online digital or mobile format. Historically, reports have largely been designed to be printed. Though static formats such as PDF files have the benefit of being printable and available online, we recommend designing your report as digital-first rather than print and digital.
PDF-based reports don’t work well on mobile devices and are not as easy to navigate and read. When you design for digital, you’re creating a digital product that can be viewed in a web browser on a laptop, tablet or mobile phone. If you design your report for the small screen size of mobile devices (smart phones and tablets) it can act as a helpful constraint, keeping your content focused and succinct.
Across all formats think about the colour, theme and style that best suits the type of data and the audience. For instance, for a patient-oriented report on chronic disease, you may prefer to have a more informal theme versus a standard, corporate-looking layout.
Photography is a critical component of the report, especially if you’re including patient stories. Choosing a photo that fits with your story can be a challenging and time-consuming process. Whenever possible, use real photographs of the people in your stories, instead of stock photography. When in doubt, take it out: don’t include photos that could detract from the tone and message of your story.
Video, much like photography, can dramatically enhance the story you’re telling if done well. Planning, cost and time to produce a high-quality custom video should be clearly understood up front. One trend is to use informal, phone-camera style videos that are more real and much more inexpensive.
Animation for data visualizations and insights should be used in select instances where you really want to draw attention to data. Too much animation can draw attention away from the actual power and impact of the data and story itself.
All of the design choices you make should line up with the original vision and balance readability, content and length. If you plan to issue multiple reports over time, think about coming up with a design language. Design language is the overall style that guides the design of a set of associated products, such as a series of reports on different cancer therapies.
7. Publishing and socializing your report
Publishing and then socializing your report is the last step and should be part of your overall plan.
Earlier in your design process you’ll ideally have considered how your report audiences are going to receive and access the report. Here are three common ways that many organizations publish their reports, and their pros and cons:
- Printing: A printed report is great to pick up and read, but is often costly and reduces the potential reach of the report. Some types of reports will continue to benefit from being printed – it comes back to your overall strategy and audience.
- PDF: If you’re printing copies you’ll likely also have a PDF version available that can be posted on a website or distributed by email. Cost is much lower per reader than printing. That said, PDFs have a number of limitations that make your report less accessible and usable than designed-for-digital formats. For instance, viewing a PDF report on a mobile device can be cumbersome and frustrating, in particular having to zoom in to read text and charts. Also, PDFs are typically not created in a way that makes them accessible for people with disabilities.
- Digital: If you’re looking to maximize the reach and use of your report it’s worthwhile to invest in producing a digital version. Converting your report to a digital format such as web page or mobile app means that it’s designed to be viewed on a variety of devices, from laptops to mobile phones. The primary downside is that digital-first reports can require more resources and specialized expertise to produce than other formats. That said, for a large distribution report (e.g., over 100,000 readers) the cost per impression can be less than print, and you can ensure the report is easy to view across devices and accessible by all audiences.
Socializing and promoting your report to your audiences effectively will ensure that the right people read it and share it with others.
There are a variety of ways you can share and promote your report. For broad-scale reports and large audiences you’ll typically want to use a multi-channel approach, combining a mix of communications across social media, internet, email, and potentially events and conferences. For more niche reports, if you have the specific audience in mind, how can you best contact them and get the report in their hands?
From an audience perspective, think about how readers of your report will find it. People who haven’t received direct promotion via email will most likely discover it through an internet search.
Prepare to make your report more easily discoverable in search on Google, and optimize your report content and keywords to rank higher in search results.
One of the best ways to reach a broad audience is to pitch it to the news media. If your report tells a story, you’re much more likely to attract media coverage, and elevate your report findings, leading to widespread promotion and increased interest.
Lastly, we strongly recommend setting up some form of report and audience analytics. With minimal effort, you can at least approximate the reach and use of your report and estimate its value and impact. At a minimum it’s good to know how many views or downloads your report has amassed over time. You can also launch surveys at the time of the report launch to get direct feedback from readers on their experience to inform your work in future.
The story continues
In his book, Houston, We Have a Narrative: Why Science Needs Story, Randy Olson argues that “narrative deficiency is the biggest problem facing science,” and advocates for creating a narrative culture within organizations. This is a noble goal, but it doesn’t mean you need to follow every step of this guide. By the simple act of considering the story in the development of your health reports and insights, you’re already on your way toward building that narrative culture. One small step toward a narrative approach is a giant leap for health data impact.
The authors thank the team at Ontario Health (Quality) who helped inform the work on health data storytelling that inspired this guide, with special thanks to everyone in the Public Reports, Digital, and Patient Partnering branches. Thanks to Susan Brien, PhD, Interim VP of Health System Performance at Ontario Health, and Julia Zarb, PhD, Program Director of the Master of Health Informatics program at the University of Toronto’s Institute for Health Policy, Management and Administration, for their wisdom, guidance and support on this project.
- Lohr, S (2013). The Origins of “Big Data”: An Etymological Detective Story. Retrieved from The New York Times: https://bits.blogs.nytimes.com/2013/02/01/the-origins-of-big-data-an-etymological-detective-story/
- Frank, AW (2010). Letting Stories Breathe: A Socio-Narratology. University of Chicago Press: Chicago
- Mascarelli, A (2018). Good Transitions: A Guide to Cementing Stories Together. Retrieved from The Open Notebook: https://www.theopennotebook.com/2018/09/25/good-transitions-a-guide-to-cementing-stories-together/
- Moran, K (2016). How Chunking Helps Content Processing. Retrieved from Nielsen Norman Group: https://www.nngroup.com/articles/chunking/
- Street, F (2018, August 23). The Anatomy of a Patient Story. Retrieved from Franklin Street: https://www.franklinstreet.com/insights/2018/08/the-anatomy-of-a-patient-story/
- Milne V., Buchanan F, Tepper J, Petch J. (2017) How patient stories are re-shaping health care. Retrieved from Healthy Debate: https://healthydebate.ca/2017/09/topic/patient-centered-care-stories
- Olson, R. (2015). Houston, we have a narrative: Why science needs story. The University of Chicago Press.