CancerSurvivalRates — It’s been quite a year

Just over a year ago, we launched CancerSurvivalRates.com. We were fortunate to hit the small window in between the holidays and when COVID-19 began to dominate news coverage. Despite the circumstances, the reaction to CSR was amazing.

After a few news outlets such as Medscape and Stat picked up the story, we started getting users from around the world. Oncologists and other physicians who grapple with providing guidance to patients thanked us for building something that was so much easier for physicians, nurses, and other healthcare providers to use. And in just one year, we’ve had almost 100,000 users, with 100 to 200 users on any given day. Usage is growing.

We were really pleased that users found the site valuable. Physicians on MedTwitter weighed in on many aspects of the site. Most reviews were positive and supportive. Physicians and others who focus on palliative care loved the site, seeing an opportunity to destigmatize conversations about life expectancy and get patients more familiar with their options.

Critical reaction was limited. A few physicians argued that it was the physician’s role to communicate prognosis and that our site shouldn’t be a substitute as a physician would have much more information about a specific patient’s medical profile to determine prognosis. We agreed at the time and still do.

We have always thought of patient usage of CSR as a “conversation starter.” The site is geared to help patients know what to ask their physician. If a patient ends up on CSR, it is likely because they found us through googling for information, indicating a gap in communication or understanding. As we said last year, we wish that there wasn’t a need for CSR.

What surprised us most was what we didn’t hear. Most users didn’t have any questions about how to use the site. No one asked how to manipulate the buttons or what they meant. This was puzzling at first — were people just coming to look and leaving? Over time, we began to get a few users emailing us to indicate they all understood how to use the site. We began to hear from doctors who used the site to explain prognosis to patients. It made us very happy to know we had created something valuable and part of the clinical tools used by physicians from around the world.

Making Connections

One of the best aspects of launching CSR were the connections we were able to make. A few of the leading voices in oncology, epidemiology, palliative care, and other areas of medicine were so kind in offering genuine feedback to improve the site. Dr. Vinay Prasad, an Oncologist, Professor of Epidemiology at UCSF, and host of Plenary Session, reviewed our site and suggested we include percentiles for survival times. Noted!

Dr. Belinda Kiely and Dr. Martin Stockler, both Oncologists at the NHMRC Clinical Trials Centre, University of Sydney, Australia who had spent many years researching and publishing articles about communicating cancer prognosis, reached out to express interest in talking about site enhancements. It was fantastic connecting, as we immediately began to discuss opportunities to make the site even easier for physicians and patients discussing what the numbers mean to someone in their situation.

Our collaboration with Drs. Kiely and Stockler as well as Dr. Andrew Martin, Senior Biostatistician at the NHMRC Clinical Trials Centre, University of Sydney, Australia led to the enhancements to CancerSurvivalRates.com that we’re pleased to announce today.

What’s New

The first enhancement to CSR was adding 10 year survival rate models. Dr. Emily Marlow, our epidemiologist, created new models using the U.S. NIH National Cancer Institute SEER data (cases from 2006–2017). Most other sites only have 5 year survival rates.

The biggest change was to add what we call “survival scenarios.” CancerSurvivalRates.com has completely changed the way in which survival “scenarios” are communicated.

For later stage cancers, the site offers “best case”, “worst case”, and “typical case” scenarios:

For patients and families, we hope survival scenarios offer a more clear sense of the outcomes for people who had a similar diagnosis.

For clinicians, we designed scenarios for use in communicating prognosis in a way that most people can relate to and understand.

For earlier stage cancers and cancers with better prognosis (as defined by having a 3 year median survival rate >50%), we show a slightly modified version of the scenarios:

We are so excited to release the enhanced version of CancerSurvivalRates.com to patients, family members, and healthcare professionals. We will continue to work hard keeping CSR as a useful, reliable, and trustworthy site.

We would love to know what you think. Feel free to contact us at csr@courage.health

Additional Thanks & Gratitude

In addition to our collaborators at the University of Sydney, we like to thank many who have helped and provided input, including:

Dr. Holly Prigerson, Professor of Geriatrics, Cornell Weill School of Medicine

Dr. Ramy Sedhom, Palliative Care Physician Fellow at Memorial Sloan Kettering Cancer Center

Dr. Jim Murphy, Radiation Oncologist, Associate Professor, Gastrointestinal (GI) Tumor Service Chief, UCSD Moores Cancer Center

Dr. Drew Bruggeman, Radiation Oncologist, Assistant Professor of Radiation Medicine and Applied Sciences, UCSD Moores Cancer Center

Dr. Aaron Simon, Resident Physician Radiation Medicine and Applied Science, UCSD Moores Cancer Center

Dr. Mark Lewis, Hematology/Oncology, Intermountain Medical Group

Dr. Vincent Rajkumar, Professor, Division of Hematology, Department of Internal Medicine, Mayo Clinic

Technical Details

  • The 1, 2, 3, and 5 year survival models used a contemporary subset of the SEER database of new first cancer diagnoses from 2004 up to 2016, consisting of 3 million patients that were followed until 2017
  • The 10 year survival models considered new first cancer diagnoses from 2000–2007, consisting of 1.7 million patients that were followed until 2017
  • All conditional survival rates are estimated using Cox Proportional Hazards Models
  • The survival scenarios are based on the survival time distribution from the Cox Proportional Hazards models, using the estimated survival rate at 3 years to determine whether to display survival rates at 1, 2, 5 and 10 years, or ranges for best, worst, and typical scenarios