skip navigation nih record
Vol. LXVII, No. 6
March 13, 2015
cover

previous story

next story


Health Care Decisions
Risk Calculators Help Doctors, Patients Choose Best Treatment

In 1989, young Michael Kattan was in business school working on a Ph.D. when he was diagnosed with Hodgkin’s disease. Unhappy with the chart his doctor used to predict his 5-year survival rate, Kattan decided to turn his quantitative skills to developing better prediction tools.

“I was told I was stage IV but the ‘little old lady’ with the walker and oxygen tank in the waiting room was also stage IV. The prediction model put us on the same survival curve,” said Kattan at a recent lecture at NIMH. “I knew I wanted a more tailored model that predicted likely outcomes based on information about me.” His goal was to produce patient-specific risk scores that would help doctors and patients choose treatments suited to each patient’s needs.

Dr. Tyrone Cannon Dr. Michael Kattan

Dr. Tyrone Cannon (l) and Dr. Michael Kattan have pioneered risk calculation science.

Photos: Jules Asher

His first tool, called a “prognostic nomogram,” was based on data from 983 prostate cancer patients. The nomogram outperformed clinicians in predicting 5-year survival rates and its success inspired him to continue his work. Today, Dr. Michael Kattan is chairman of the department of quantitative health sciences at the Cleveland Clinic. He has developed a number of statistics-based risk calculators to predict health risks and outcomes that help doctors and patients weigh health care options. These nomograms, available for free online (http://rcalc.ccf.org), give doctors and patients more information on which to make health care decisions. NIH funding also helped Kattan offer an online tool to support scientists who want to develop their own models.

Most recently, Kattan teamed up with Dr. Tyrone Cannon, professor of psychology and psychiatry at Yale University, and other investigators involved in the North American Prodrome Longitudinal Study (NAPLS 2). Their joint goal was to use NAPLS data to create a practical tool for predicting first-episode psychosis among youth at heightened risk for psychotic disorders. “Our initial goal was to develop a calculator based on risk factors that are easily assessed in standard clinical settings and then to refine it by incorporating information on biological risk factors as those come to light.”

The Kattan-Cannon collaboration resulted in a new “2-Year Probability of Conversion to Psychosis ” tool. This nomogram can help clinicians who have been trained on the interview used to diagnose prodromal risk syndromes to estimate an individual patient’s risk of moving from the prodromal phase of psychosis (showing early symptoms) to actual psychosis in the coming 24 months. Based on the patient’s risk score, the clinician and patient can decide on the best care strategy.

To develop the nomogram, Cannon and colleagues chose 8 variables that previous research has associated with psychosis risk, including the existence of a first-degree relative with psychosis, the number and types of traumas, difficulties with memory and verbal learning and problems with unusual thoughts and suspiciousness. Data from all eight variables are run through Kattan’s nomogram to create the individual’s risk score.

“Tools like the psychosis risk calculator can help patients and doctors discuss options and develop personal treatment plans,” said Dr. Robert Heinssen, director of the Division of Services and Intervention Research at NIMH. “There are several proven interventions available for people at risk for psychosis, such as cognitive behavior therapy, counseling and support for family members and outreach to schools or work settings. Using the nomogram, doctors and patients can discuss the patient’s level of risk and distress and decide on the best plan of action.”

In the future, Kattan and Cannon say they would like to see duration of prodromal symptoms included in the 2-year probability model and perhaps biological test results. They are also considering the possibility of developing a nomogram that can estimate the likelihood of remission and encourage use of the risk calculator in treatment studies. They hope that risk estimates can be strengthened by refining known risk factors and also by accounting for a particular intervention program.


back to top of page