Validated Score Predicts Which Women Are at Risk of Midlife Declines in Physical Health and Function

Adverse health trajectories often begin during midlife. Interventions target exercise and other behaviors that might prevent future decline. However, it’s possible even more effective strategies could be developed that focus on high-risk groups.

Researchers at Brigham and Women’s Hospital used data from two large U.S. cohorts to derive and validate a risk score that predicts clinically meaningful midlife declines in physical health and function among thousands of women. Notably, the score is based on variables that are easy to obtain in clinical practice.

Daniel H. Solomon, MD, MPH, chief of the Section of Clinical Sciences in the Division of Rheumatology, Inflammation and Immunity at Brigham and Women’s Hospital, JoAnn E. Manson, MD, MPH, DrPH, chief of the Division of Preventive Medicine, and colleagues introduce the risk score in BMJ Open.


The primary outcome of the risk score was a clinically important decline in physical health and function. This was defined as an 8-point or greater decline in the Physical Component Summary (PCS) score of the 36-Item Short Form Survey (SF-36).

In a previous study, Dr. Solomon and colleagues used data from the Study of Women’s Health Across the Nation (SWAN) to identify seven significant predictors of PCS decline between ages 55 and 65:

  • Six risk factors—lower educational attainment, smoking, higher body mass index, history of cardiovascular disease, history of osteoarthritis, and depressive symptoms
  • One adjustment variable—baseline PCS level

SWAN is a racially diverse longitudinal cohort of 3,302 women enrolled at seven geographically diverse sites.

In this new study, the seven predictors were entered into a regression model that included 2,535 women from the Women’s Health Initiative (WHI), a longitudinal study investigating major determinants of chronic diseases among postmenopausal women enrolled at 40 centers.


When the two cohorts were compared, each predictor was associated with similar odds of 10-year decline in PCS. The area under the curves (ability to predict functional decline) was also similar: 0.71 in SWAN and 0.64 in WHI.

The regression coefficients from SWAN were then used to calculate a risk score for PCS declines in both SWAN and WHI. The median scores were very similar in the two cohorts:

  • SWAN—Median, 0.16; IQR, 0.10–0.25
  • WHI—Median, 0.13; IQR, 0.089–0.20

The team experimented with various risk categories to create a binary clinical test. Using a threshold of 0%–30% (very low to medium probability of ≥8-point decline in PCS over 10 years) and >30% (high to very high probability), the risk score was useful in both cohorts:

  • Specificity of 89%–93%
  • Negative predictive value of 76%–86%
  • Accuracy of 73%–79%
  • Sensitivity of 18%–36%
  • Positive predictive value of 43%–49%

Clinical Applications

The variables in the risk score may help women and their clinicians predict future health status and functional decline and therefore motivate behavior change.

In addition, the study results support system-wide strategies to improve middle-aged women’s health by addressing tobacco use, high body mass index and depressive symptoms.

Leave a Reply