Development and temporal evaluation of a multivariable prediction model to screen for late-diagnosed cystic fibrosis
This study developed and temporally evaluated a multivariable prediction model using seven clinical features that successfully identified all adult patients with likely or unresolved cystic fibrosis for diagnostic testing while minimizing false positives among unlikely cases, though further prospective multicenter validation is required.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer
For decades, the diagnosis of cystic fibrosis has been a story of early detection. Thanks to newborn screening programs, most children with the condition are identified shortly after birth, allowing families to begin life-saving treatments immediately. However, a significant number of people slip through the cracks. These are adults who carry the genetic changes for the disease but whose symptoms are subtle, unusual, or easily mistaken for other common respiratory or digestive problems. Because the signs are so varied and often appear late in life, doctors may not think to test for cystic fibrosis until years of illness have passed. This delay is critical; without a confirmed diagnosis, patients miss out on modern therapies that can dramatically improve their health and longevity. The challenge for clinicians is knowing which adult patient, presenting with vague symptoms like chronic cough or digestive trouble, actually needs a specific sweat test to confirm the disease.
To address this gap, researchers at Santa Barbara Cottage Hospital and Cedars-Sinai Medical Center developed a new tool designed to help doctors decide when to order these tests. They created a screening model that looks at a patient's medical history and current symptoms to calculate the likelihood of cystic fibrosis. The team gathered data from 256 adults who had been evaluated for the condition between 2005 and 2023. They divided these patients into two groups: a larger group used to build the model and a smaller, later group used to test if the model still worked over time. The researchers focused on seven specific clinical clues that often appear in adults with the disease: pancreatic insufficiency (where the body cannot digest food properly), acute or chronic pancreatitis, infections with specific types of bacteria, the presence of a family member with a cystic fibrosis gene variant, and specific patterns seen on lung scans like bronchiectasis or "tree-in-bud" opacities.
The researchers built a system that weighs these seven factors together. If a patient has a high enough score based on these factors, the model recommends a sweat chloride test. Additionally, the model includes a safety net: if a patient has the specific lung patterns mentioned above, even if they lack the other six factors, the model still recommends testing. When the researchers applied this two-part rule to their first group of patients, the model successfully identified every single person who was later confirmed to have cystic fibrosis. It also caught every patient who had an unresolved form of the disease, meaning it did not miss anyone who needed further investigation. In the second, later group of patients, the model performed just as well, identifying all confirmed cases and all unresolved cases.
The study also looked at how often the model might suggest a test for someone who does not have the disease. In the group of patients where doctors initially thought cystic fibrosis was unlikely, the model correctly decided not to test for the vast majority of them. In fact, among the patients the doctors had already dismissed, the model only flagged two people out of 126 as needing a test. This suggests the tool is precise enough to avoid unnecessary testing while remaining sensitive enough to catch every true case. The researchers noted that the model works best when used by clinicians who can accurately identify these specific medical conditions in their patients. While the study was conducted at a single center and involved a relatively small number of people, the results offer a promising, practical way to use everyday medical information to find adults who have been living with an undiagnosed condition for too long. The authors emphasize that this is a preliminary step and that the model needs to be tested in larger, more diverse populations before it can be widely adopted, but it represents a significant move toward solving the puzzle of late-diagnosed cystic fibrosis.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.