New calculator tool estimates fracture risk on dialysis

VANCOUVER – A new calculator that predicts short-term fracture risk at both 1 year and 3 years in patients on dialysis performed well in a study presented at the annual meeting of the American Society for Bone and Mineral Research.

The tool will soon be available on QxMD Calculate, which provides free decision-support tools for physicians, said presenter Andrea Cowan, MD, an assistant professor of medicine at the University of Western Ontario, London.

Dialysis patients have an approximately fivefold increased risk for fracture, Dr. Cowan noted, compared with the general population. However, treatments to prevent fracture risk are relatively limited and can have significant side effects. Therefore, “you really want to make sure that the person you’re targeting for treatment is actually going to be at a reasonable risk of fracture,” she said.

The Fracture Risk Assessment Tool (FRAX) is useful, but it estimates 10-year fracture risk, which is too long of a time frame to be useful for dialysis patients who experience a 50% 5-year mortality, according to Dr. Cowan. It does not take kidney failure or severe hyperparathyroidism into account, and it also requires information like bone mineral density, which poses an additional burden for a dialysis patient already undergoing multiple tests.

The new calculator could also be useful for research because it doesn’t rely on clinical data that might not be generally available, such as parental fracture, smoking status, or body mass index. “There’s a move towards things like pragmatic trials, which use more routinely collected data, have broader inclusion criteria, and are often more cost efficient to run. This calculator should be relatively easy to implement in trials using routinely collected data to perhaps define a subgroup of patients who may be at high risk of fracture without having to apply really cumbersome tools,” Dr. Cowan said.

The researchers included 11,599 patients between ages 40 and 89 years who were treated at a single center in Ontario between 2010 and 2017. The mean age was 66.18 years, 38.6% were women, 64.1% had diabetes, 11.9% had liver disease, and median time on dialysis was 0.81 years. The patients’ median parathyroid hormone level was 30 pmol/L.

At 3 years, the cumulative incidence of any fracture was 7.36% (95% confidence interval, 6.89-7.85), including 2.62% for hip fracture (95% CI, 2.34-2.93), 1.36% for spine fracture (95% CI, 1.16-1.59), 1.93% for wrist or forearm (95% CI, 1.69-2.20), and 2.15% for the pelvis (95% CI, 1.89-2.43). The incidence for all fractures at 1 year was 2.93 (95% CI, 2.62-3.26).

Variables associated with fracture risk included female sex (hazard ratio, 1.46; 95% CI, 1.27-1.67), a previous fracture more than 1 year in the past (HR, 1.65; 95% CI, 1.37-2.00), a fracture in the past year (HR, 3.63; 95% CI, 2.86-4.60), and proton pump inhibitor use (HR, 1.23; 95% CI, 1.04-1.45). After inclusion of vitamin D use, steroid use, time on dialysis, calcium levels, phosphate levels, presence of diabetes, rheumatoid arthritis, and chronic liver disease, the full model had an area under the curve of 77.7 at 1 year (95% CI, 73.3-84.4) and 69.9 at 3 years (95% CI, 68.0-72.2). For hip fracture, the model had an AUC of 80.1 at 1 year (95% CI, 77.0-83.5) and 71.9 at 3 years (95% CI, 70.1-74.2).

During the Q&A session, Dr. Cowan was asked how the tool could be implemented clinically. She said that it could have value in discussing fracture prediction and prevention with patients, but it could also increase fracture risk awareness among nephrologists. “I need to convince a lot of my colleagues because they’re focused on other things, so having this [calculator] I think is both good from a patient as well as a practitioner perspective. And the treatments that we have in people with end-stage renal disease are limited, so you want to know that you’re really targeting the high-risk person before you potentially put them on denosumab and increase the risk of severe hypercalcemia and things like that,” Dr. Cowan said.

The study points out the challenges of predicting fracture risk for specific populations, according to session comoderator Evelyn Hsieh, MD. She noted that the study needs follow-up. “I don’t think they had gotten to a validation [in a separate cohort] yet,” said Dr. Hsieh, an associate professor of medicine (rheumatology) and epidemiology (chronic diseases) at Yale University, New Haven, Conn.

Dr. Cowan and Dr. Hsieh have no relevant financial disclosures.

A version of this article first appeared on Medscape.com.

VANCOUVER – A new calculator that predicts short-term fracture risk at both 1 year and 3 years in patients on dialysis performed well in a study presented at the annual meeting of the American Society for Bone and Mineral Research.

The tool will soon be available on QxMD Calculate, which provides free decision-support tools for physicians, said presenter Andrea Cowan, MD, an assistant professor of medicine at the University of Western Ontario, London.

Dialysis patients have an approximately fivefold increased risk for fracture, Dr. Cowan noted, compared with the general population. However, treatments to prevent fracture risk are relatively limited and can have significant side effects. Therefore, “you really want to make sure that the person you’re targeting for treatment is actually going to be at a reasonable risk of fracture,” she said.

The Fracture Risk Assessment Tool (FRAX) is useful, but it estimates 10-year fracture risk, which is too long of a time frame to be useful for dialysis patients who experience a 50% 5-year mortality, according to Dr. Cowan. It does not take kidney failure or severe hyperparathyroidism into account, and it also requires information like bone mineral density, which poses an additional burden for a dialysis patient already undergoing multiple tests.

The new calculator could also be useful for research because it doesn’t rely on clinical data that might not be generally available, such as parental fracture, smoking status, or body mass index. “There’s a move towards things like pragmatic trials, which use more routinely collected data, have broader inclusion criteria, and are often more cost efficient to run. This calculator should be relatively easy to implement in trials using routinely collected data to perhaps define a subgroup of patients who may be at high risk of fracture without having to apply really cumbersome tools,” Dr. Cowan said.

The researchers included 11,599 patients between ages 40 and 89 years who were treated at a single center in Ontario between 2010 and 2017. The mean age was 66.18 years, 38.6% were women, 64.1% had diabetes, 11.9% had liver disease, and median time on dialysis was 0.81 years. The patients’ median parathyroid hormone level was 30 pmol/L.

At 3 years, the cumulative incidence of any fracture was 7.36% (95% confidence interval, 6.89-7.85), including 2.62% for hip fracture (95% CI, 2.34-2.93), 1.36% for spine fracture (95% CI, 1.16-1.59), 1.93% for wrist or forearm (95% CI, 1.69-2.20), and 2.15% for the pelvis (95% CI, 1.89-2.43). The incidence for all fractures at 1 year was 2.93 (95% CI, 2.62-3.26).

Variables associated with fracture risk included female sex (hazard ratio, 1.46; 95% CI, 1.27-1.67), a previous fracture more than 1 year in the past (HR, 1.65; 95% CI, 1.37-2.00), a fracture in the past year (HR, 3.63; 95% CI, 2.86-4.60), and proton pump inhibitor use (HR, 1.23; 95% CI, 1.04-1.45). After inclusion of vitamin D use, steroid use, time on dialysis, calcium levels, phosphate levels, presence of diabetes, rheumatoid arthritis, and chronic liver disease, the full model had an area under the curve of 77.7 at 1 year (95% CI, 73.3-84.4) and 69.9 at 3 years (95% CI, 68.0-72.2). For hip fracture, the model had an AUC of 80.1 at 1 year (95% CI, 77.0-83.5) and 71.9 at 3 years (95% CI, 70.1-74.2).

During the Q&A session, Dr. Cowan was asked how the tool could be implemented clinically. She said that it could have value in discussing fracture prediction and prevention with patients, but it could also increase fracture risk awareness among nephrologists. “I need to convince a lot of my colleagues because they’re focused on other things, so having this [calculator] I think is both good from a patient as well as a practitioner perspective. And the treatments that we have in people with end-stage renal disease are limited, so you want to know that you’re really targeting the high-risk person before you potentially put them on denosumab and increase the risk of severe hypercalcemia and things like that,” Dr. Cowan said.

The study points out the challenges of predicting fracture risk for specific populations, according to session comoderator Evelyn Hsieh, MD. She noted that the study needs follow-up. “I don’t think they had gotten to a validation [in a separate cohort] yet,” said Dr. Hsieh, an associate professor of medicine (rheumatology) and epidemiology (chronic diseases) at Yale University, New Haven, Conn.

Dr. Cowan and Dr. Hsieh have no relevant financial disclosures.

A version of this article first appeared on Medscape.com.

VANCOUVER – A new calculator that predicts short-term fracture risk at both 1 year and 3 years in patients on dialysis performed well in a study presented at the annual meeting of the American Society for Bone and Mineral Research.

The tool will soon be available on QxMD Calculate, which provides free decision-support tools for physicians, said presenter Andrea Cowan, MD, an assistant professor of medicine at the University of Western Ontario, London.

Dialysis patients have an approximately fivefold increased risk for fracture, Dr. Cowan noted, compared with the general population. However, treatments to prevent fracture risk are relatively limited and can have significant side effects. Therefore, “you really want to make sure that the person you’re targeting for treatment is actually going to be at a reasonable risk of fracture,” she said.

The Fracture Risk Assessment Tool (FRAX) is useful, but it estimates 10-year fracture risk, which is too long of a time frame to be useful for dialysis patients who experience a 50% 5-year mortality, according to Dr. Cowan. It does not take kidney failure or severe hyperparathyroidism into account, and it also requires information like bone mineral density, which poses an additional burden for a dialysis patient already undergoing multiple tests.

The new calculator could also be useful for research because it doesn’t rely on clinical data that might not be generally available, such as parental fracture, smoking status, or body mass index. “There’s a move towards things like pragmatic trials, which use more routinely collected data, have broader inclusion criteria, and are often more cost efficient to run. This calculator should be relatively easy to implement in trials using routinely collected data to perhaps define a subgroup of patients who may be at high risk of fracture without having to apply really cumbersome tools,” Dr. Cowan said.

The researchers included 11,599 patients between ages 40 and 89 years who were treated at a single center in Ontario between 2010 and 2017. The mean age was 66.18 years, 38.6% were women, 64.1% had diabetes, 11.9% had liver disease, and median time on dialysis was 0.81 years. The patients’ median parathyroid hormone level was 30 pmol/L.

At 3 years, the cumulative incidence of any fracture was 7.36% (95% confidence interval, 6.89-7.85), including 2.62% for hip fracture (95% CI, 2.34-2.93), 1.36% for spine fracture (95% CI, 1.16-1.59), 1.93% for wrist or forearm (95% CI, 1.69-2.20), and 2.15% for the pelvis (95% CI, 1.89-2.43). The incidence for all fractures at 1 year was 2.93 (95% CI, 2.62-3.26).

Variables associated with fracture risk included female sex (hazard ratio, 1.46; 95% CI, 1.27-1.67), a previous fracture more than 1 year in the past (HR, 1.65; 95% CI, 1.37-2.00), a fracture in the past year (HR, 3.63; 95% CI, 2.86-4.60), and proton pump inhibitor use (HR, 1.23; 95% CI, 1.04-1.45). After inclusion of vitamin D use, steroid use, time on dialysis, calcium levels, phosphate levels, presence of diabetes, rheumatoid arthritis, and chronic liver disease, the full model had an area under the curve of 77.7 at 1 year (95% CI, 73.3-84.4) and 69.9 at 3 years (95% CI, 68.0-72.2). For hip fracture, the model had an AUC of 80.1 at 1 year (95% CI, 77.0-83.5) and 71.9 at 3 years (95% CI, 70.1-74.2).

During the Q&A session, Dr. Cowan was asked how the tool could be implemented clinically. She said that it could have value in discussing fracture prediction and prevention with patients, but it could also increase fracture risk awareness among nephrologists. “I need to convince a lot of my colleagues because they’re focused on other things, so having this [calculator] I think is both good from a patient as well as a practitioner perspective. And the treatments that we have in people with end-stage renal disease are limited, so you want to know that you’re really targeting the high-risk person before you potentially put them on denosumab and increase the risk of severe hypercalcemia and things like that,” Dr. Cowan said.

The study points out the challenges of predicting fracture risk for specific populations, according to session comoderator Evelyn Hsieh, MD. She noted that the study needs follow-up. “I don’t think they had gotten to a validation [in a separate cohort] yet,” said Dr. Hsieh, an associate professor of medicine (rheumatology) and epidemiology (chronic diseases) at Yale University, New Haven, Conn.

Dr. Cowan and Dr. Hsieh have no relevant financial disclosures.

A version of this article first appeared on Medscape.com.