Web application guides decision-making and strategies for mitigating presurgical morbidities affecting outcomes.

Multiple clinical variables, including individual comorbidities, must be contemplated when deciding whether to proceed with resection of a meningioma, yet there has been no means of rating these combined risk factors.

Peter J. Morone, M.D., an assistant professor of neurosurgery and director of neurosurgical oncology at Vanderbilt University Medical Center, was the principal investigator on a July 2023 study filling that gap and offering a web-based tool to help guide surgeons.

“Using this tool will allow physicians to comprehensively discuss surgical risks with their patients.”

By studying electronic health records of thousands of meningioma patients, he and his team developed and validated a model for predicting hospital length of stay (LOS), a viable surrogate for postoperative outcomes and cost, regardless of tumor stage or grade.

This work also adds new parameters to those traditionally considered.

“Using this tool will allow physicians to comprehensively discuss surgical risks with their patients,” Morone said. “It also lays out what are modifiable factors and what their significance is, so the option is there to address them prior to surgery.”

Full and Reduced Models

Meningiomas are common intracranial tumors that often warrant surgical intervention due to their potential to increase in size and exert a mass effect. New value-based health care models are necessitating quantification of the risks of these surgeries during the first patient encounter, Morone says.

To enable this, Morone’s team looked at National Inpatient Sample records from meningioma surgical patients from 2009-2013 and developed two prognostic models for time to favorable discharge after surgery. Of the 10,757 patients studied, 6,554 (60 percent) had a favorable discharge, defined as a discharge to home, and the median LOS was three days.

Their “full model” includes 37 (28 clinical and 9 socioeconomic) predictors associated with a higher likelihood of unfavorable discharge. Of the 28 clinical predictors, 13 were modifiable risk factors. A “reduced model” combines the 13 modifiable factors with age, sex and race.

“We looked at both patient-related and disease-related variables to most accurately risk-stratify,” Morone said.

The team found that all but two of the modifiable risk factors – drug abuse and obesity – negatively impacted the chances of a favorable discharge.

“These exceptions were somewhat surprising on the surface, but while obesity alone wasn’t a significant factor, its numerous comorbidities did account for clinical factors influencing LOS,” Morone said. 

The two models predicted time to favorable discharge with similar accuracy, but Morone says the simpler, reduced model offers the most practical utility. In fact, all the variables included in the reduced model can be incorporated into an EHR, allowing for the calculation of real-time discharge probabilities based upon preoperative patient characteristics, he says.

Prevailing Factors in Outcomes

The list of clinical and demographic factors most strongly associated with longer LOS in both models included older age and the presence of comorbidities, including paralysis, recent weight loss, electrolyte disorders, coagulopathy, diabetes, deficiency anemias, neurological and psychiatric disorders, and pulmonary circulation disorders.

Age turned out to be the largest single factor.

“For example, the probability of being discharged by hospital day three (median time to favorable discharge) is 48 percent for a 35-year-old male versus 27 percent for a 75-year-old male,” the authors wrote.

 Race was also a substantial factor, with Black patients averaging a longer length of stay than white patients, a gap that grew after about age 65.

Gender was an insignificant predictor of discharge in the full model, though in the reduced model, females had a 6 percent higher chance of favorable discharge than males.

As expected, hospitalization characteristics weighed significantly. Patients with private insurance and those admitted to private or non-profit hospitals or hospitals located outside the Northeastern United States had higher odds of favorable discharge outcomes. Interestingly, the chances of favorable discharge were similar in teaching versus non-teaching hospitals.

Modifiable Risks

Of greatest clinical importance were the findings about modifiable morbidities, which, if addressed, could turn a high-risk patient into a good surgical candidate.

 In order from greatest to least, top factors negatively impacting successful discharge were fluid and electrolyte disorders, coagulopathy, congestive heart failure, deficiency anemias, alcohol abuse and diabetes.

A history of substance abuse accounted for an approximately 15 percent lower probability of successful discharge, and the gap further tapered to under 5 percent for hypertension and obesity, Morone said.

“With this data, I am able to sit down with a patient and talk about improving their odds of a successful surgery if we work as a team over weeks to months to treat some of these co-morbidities,” Morone said.

The researchers deliberately limited the study to factors weighing in on the surgical decision, not those that attempt to predict overall survival.

“These models provide the information we need as a compass for the timing and, in many cases, for the most appropriate decision to do surgery or pursue other therapies,” Morone said. “This is what we were trying to accomplish.”

About the Expert

Peter Morone, M.D.

Peter J. Morone, M.D., M.S.C.I., is an assistant professor of neurological surgery and director of neurosurgical oncology at Vanderbilt University Medical Center. One aspect of his research focuses on meningioma risk-stratification and treatment optimization.