Pamidronate linked to fewer fractures in Romanian children with bone fragility

Children with osteogenesis imperfecta experienced substantially fewer fractures per year after intravenous pamidronate was introduced into their care, according to a 20-year retrospective study spanning three tertiary pediatric hospitals in Bucharest, Romania. The cohort included 49 children, and the adjusted analysis associated the post-treatment period with a 51% lower fracture incidence rate than the period before treatment began. The finding comes from a before-and-after observational design, so it cannot establish that pamidronate alone caused the reduction, but it provides unusually long-term real-world evidence from a region where pediatric data have been sparse.

Osteogenesis imperfecta, often called brittle bone disease, is an inherited disorder associated with fragile bones and recurrent fractures. Its severity varies widely, from mild forms with relatively few fractures to severe phenotypes involving marked skeletal deformity, growth impairment and impaired mobility. Children in the Romanian cohort were diverse: 28 were female and 21 male; the median age at diagnosis was 24.5 months, with an interquartile range of 0.25 to 81.25 months. Sillence type I was the most common subtype, affecting 21 children, followed by type III in 17, type IV in nine and type V in two.

Pamidronate, a bisphosphonate that inhibits osteoclast-mediated bone resorption, was administered intravenously in cyclic regimens of 0.5 to 1 mg/kg per day for three consecutive days, repeated every three to four months. Treatment began at a median age of 45 months, and its median duration was 72 months. Overall follow-up extended for a median of 144 months. The researchers analyzed each child’s experience both before and after treatment initiation, rather than comparing pamidronate-treated children with an untreated group. This approach reduced some between-patient variability because every participant contributed data from both periods.

Before pamidronate was started, 198 fractures were documented over 250.3 person-years, giving a crude fracture incidence rate of 0.79 per person-year. After initiation, 208 fractures occurred over 359.3 person-years, for a rate of 0.58 per person-year. The unadjusted post-treatment rate was therefore about 27% lower, with an incidence rate ratio of 0.73 and a 95% confidence interval of 0.60 to 0.89. Because the number of fractures rose while the rate fell, the finding reflects the longer observation period after treatment rather than a complete elimination of fractures.

The principal analysis used a negative binomial generalized estimating equations model with a log link. This approach accommodates overdispersed fracture counts, correlates the two observations contributed by each child and includes log person-years as an offset. After adjustment for Sillence type, sex and age at treatment initiation, the post-pamidronate period remained associated with lower fracture incidence, with an incidence rate ratio of 0.49, a 95% confidence interval of 0.35 to 0.68 and p below 0.001. In the authors’ interpretation, this represents an estimated 51% lower incidence rate during the post-initiation period, although the result remains an association rather than a causal treatment effect.

Disease subtype also tracked with fracture burden. Compared with type I osteogenesis imperfecta, type III was associated with more than twice the fracture incidence rate, with an incidence rate ratio of 2.09, a 95% confidence interval of 1.49 to 2.92 and p below 0.001. Type IV also had a higher adjusted rate, with a ratio of 1.72, a confidence interval of 1.01 to 2.94 and p of 0.048. The point estimate for type V was higher at 2.50, but its wide confidence interval of 0.82 to 7.59 and p of 0.106 did not meet the threshold for statistical significance; only two children had type V, sharply limiting precision.

Sex was not independently associated with fracture incidence. The estimated rate for females relative to males was 1.23, but its 95% confidence interval was 0.89 to 1.70 and p was 0.213. Older age at pamidronate initiation was associated with a lower overall fracture rate, at 0.992 per additional month, with a 95% confidence interval of 0.988 to 0.996 and p below 0.001. That result should not be read as evidence for delaying treatment. It may instead reflect differences in phenotype, diagnosis and referral patterns: children with milder disease may have entered treatment later, while age at initiation does not capture how fracture risk changes throughout follow-up.

Orthopedic care was part of the post-treatment landscape. Twenty-nine children, or 59.2%, underwent corrective osteotomies with telescopic intramedullary rodding, all after pamidronate had been initiated, and 36 received physiotherapy during observation. Adding the surgery variable to the model did not materially alter the treatment-period association, which remained at an incidence rate ratio of 0.49 with the same confidence interval. Surgery itself was not statistically associated with fracture incidence, at 1.02 with a 95% confidence interval of 0.67 to 1.54 and p of 0.936. That sensitivity analysis was exploratory because records did not allow the exact timing, number and anatomical sites of procedures to be modeled as time-varying exposures.

Several limitations constrain the interpretation of these results. The study had no concurrent untreated control group, and all children were older during the post-treatment observation than before it. Fracture risk in pediatric osteogenesis imperfecta can vary with growth, skeletal maturation, mobility and the natural course of the disorder, yet the model adjusted only for age at treatment initiation, not age throughout each period. Selection was also required for documented pre-treatment fractures, potentially creating regression toward the mean if children entered treatment during a period of unusually high fracture activity. Thirty-seven of 86 children identified in the hospital records were excluded because of missing data or other predefined reasons, introducing possible selection bias.

Retrospective ascertainment added further uncertainty. Some pre-treatment fractures were reconstructed from medical histories rather than continuously observed at the participating centers, meaning fractures treated elsewhere may have been missed. Bone mineral density measurements were unavailable for systematic analysis, so structural changes could not be connected directly to the fracture pattern. Genetic testing was not performed for 32 children, and two patients with negative targeted testing of COL1A1 and COL1A2 did not undergo broader sequencing. Despite these constraints, the use of person-time, repeated within-patient observations and a model suited to overdispersed counts gives the report more weight than a simple comparison of raw fracture totals. It offers regional evidence of a clinically meaningful association between pamidronate initiation and lower fracture incidence, while leaving the size of any causal drug effect unresolved.

Subject of Research: Long-term fracture outcomes after intravenous pamidronate initiation in 49 Romanian children with osteogenesis imperfecta

Article Title: Fracture outcomes following intravenous pamidronate initiation in children with osteogenesis imperfecta: a multicenter retrospective cohort study from Bucharest, Romania

Article References: Dinulescu, A., Stăiculescu, M.-E., Dijmărescu, I., Pavelescu, M.-L., Păcurar, D., & Ulici, A. (2026). Fracture outcomes following intravenous pamidronate initiation in children with osteogenesis imperfecta: a multicenter retrospective cohort study from Bucharest, Romania. Frontiers in Pediatrics, 14, Article 1973795. https://doi.org/10.3389/fped.2026.1973795

Image Credits: AI Generated

DOI: 10.3389/fped.2026.1973795

Keywords: osteogenesis imperfecta, pamidronate, pediatric fractures, bisphosphonates, retrospective cohort, Romania, bone fragility, fracture incidence, Sillence classification, pediatric orthopedics, Fracture, outcomes