Articles
Fracture risk in adult patients treated with growth hormone replacement therapy for growth hormone deficiency: a prospective observational cohort study Daojun Mo, Maria Fleseriu, Rong Qi, Nan Jia, Christopher Jeremy Child, Roger Bouillon*, Dana Sue Hardin*
Summary Background To our knowledge, no controlled studies of the effects of long-term growth hormone replacement on fracture risk in adult patients with growth hormone deficiency exist. We assessed the effect of growth hormone treatment on fracture risk in patients with growth hormone deficiency from the international Hypopituitary Control and Complications Study (HypoCCS) surveillance database.
Lancet Diabetes Endocrinol 2015
Methods In this prospective cohort study, patients with growth hormone deficiency were analysed from the HypoCCS database of adults with hypopituitarism from the USA, Canada, Japan, and 14 European countries. Patients were eligible if they were aged 18 years or older and had an established diagnosis of growth hormone deficiency, either alone or with multiple pituitary hormone deficiencies, as identified by clinical history and biochemical testing. Patients were assessed over a mean follow-up period of 4·6 years (SD 3·8). The effect of growth hormone treatment on fracture risk was assessed by Cox proportional hazard modelling with adjustment for several confounders.
See Online/Comment http://dx.doi.org/10.1016/ S2213-8587(15)00036-4
Findings Between Jan 3, 1996, and Dec 15, 2012, we enrolled 10 673 patients to this study. Of the enrolled patients, 1032 patients were excluded from assessment because of incomplete data, leaving 9641 in the analysis cohort. Of these patients, 8374 of received growth hormone and 1267 did not. Annual fracture incidence rate was lower in patients who received growth hormone than in those who did not (fracture incidence rate 1·19% vs 1·91%, hazard ratio [HR] 0·69, 95% CI 0·54–0·88). However, no difference in fracture risk was observed between patients who did and did not receive growth hormone treatment in the subgroup of patients with pre-existing osteoporosis (n=826; 0·97, 0·48–1·95). Interpretation Our results suggest that growth hormone replacement therapy could be protective against fracture for adult patients with growth hormone deficiency without previously reported osteoporosis. Starting growth hormone therapy before the onset of osteoporosis might be optimum for bone health of adult patients with growth hormone deficiency. Funding Eli Lilly and Co.
Introduction Low bone mass or density is a characteristic feature of adult patients with growth hormone deficiency. Findings from previous studies1–6 showed that adult patients with severe growth hormone deficiency had lower bone mineral density7 and were two to five times more likely to have a fracture than those without growth hormone deficiency.1–6 The effects of growth hormone replacement on biomarkers of bone turnover, bone mineral density, and bone mineral content have been assessed in clinical trials, including in patients with childhood-onset and adult-onset growth hormone deficiency. Results from these studies7 have shown that growth hormone has a biphasic effect on bone. In the first 12 months of growth hormone treatment, bone mineral density might not increase and might even decrease. After 24 months, bone mineral content and bone mineral density increased from baseline in the lumbar spine and proximal femur in adults with either childhood-onset or adult-onset growth hormone deficiency, as shown in a detailed meta-analysis8 of randomised controlled studies. In long-term
Published Online April 13, 2015 http://dx.doi.org/10.1016/ S2213-8587(15)00098-4
*Contributed equally Eli Lilly and Company, Lilly Research Laboratories, Indianapolis, IN, USA (D Mo MD, R Qi MS, N Jia PhD, D S Hardin MD); Northwest Pituitary Center, Departments of Medicine and Neurological Surgery, Oregon Health and Science University, Portland, OR, USA (M Fleseriu MD); Eli Lilly and Company, Lilly Research Laboratories, Windlesham, UK (C J Child PhD); and Clinic and Laboratory of Endocrinology, Gasthuisberg, KU Leuven, Leuven, Belgium (Prof R Bouillon MD) Correspondence to: Dr Daojun Mo, Lilly Corporate Center, Eli Lilly and Company, Indianapolis, IN 46285, USA [email protected]
prospective studies,8,9 growth hormone replacement therapy for up to 15 years induced a sustained increase in total body and lumbar spine (L2–L4) bone mineral content and bone mineral density. These findings suggest that long-term growth hormone treatment might help to prevent fractures in adult patients with growth hormone deficiencies. Moreover, growth hormone replacement therapy increases muscle mass and strength, which might have additional beneficial effects on fracture risk. One Swedish study3 reported a decreased relative fracture incidence in a subgroup of men with adult-onset growth hormone deficiency (n=389) compared with matched population controls during 6-year growth hormone and sex hormone replacement therapy, although this was not reported in women. Two other retrospective studies2,10 also suggested that growth hormone therapy in adults with growth hormone deficiencies might reduce vertebral fracture prevalence. However, no prospective controlled studies have reported effects of long-term growth hormone replacement on fracture occurrence in adult patients with growth hormone deficiencies.6,8
www.thelancet.com/diabetes-endocrinology Published online April 13, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00098-4
1
Articles
The Hypopituitary Control and Complications Study (HypoCCS),11 an international post-marketing surveillance study done by Eli Lilly and Co, prospectively obtained clinical outcome information in adult patients with growth hormone deficiency from the USA, Canada, Japan, and 14 European countries (Austria, Belgium, Iceland, Czech Republic, Denmark, France, Germany, Hungary, Italy, Norway, Spain, Sweden, the Netherlands, and UK). The study obtained information about the occurrence of cancer, mortality, cardiovascular diseases, diabetes, metabolic syndromes, and fractures and associated risk factors. The aim of this analysis was to assess growth hormone replacement effects on fracture incidence by comparing patients in the HypoCCS cohort who received growth hormone treatment with those that did not, while taking possible confounders into account.
Methods Study design and patients In this prospective cohort study, we enrolled patients who met the adult growth hormone deficiency indication for Humatrope (somatropin [rDNA origin] for injection; Eli Lilly and Co, Indianapolis, USA), according to the approved package insert for each participating country. Patients were included if they were aged 18 years or older and had an established diagnosis of growth hormone deficiency, either alone or with multiple pituitary hormone deficiencies, as identified by clinical history and biochemical testing.11 The investigating site physician decided what diagnostic approach was used and whether growth hormone therapy should be given to patients. According to the observational design of HypoCCS, the choice of whether to receive growth hormone replacement therapy was made by patients in consultation with their physicians. Patients were not eligible for enrolment if they had active or unresolved contraindication(s) to growth hormone therapy (including evidence or suspicion of active malignancy, evidence of sustained pituitary or other intracranial tumour activity, or pregnancy) or if they were breastfeeding. Patients did not receive growth hormone replacement therapy in HypoCCS for several reasons: they did not want injections, they felt well and did not feel the need for therapy, they were concerned about CNS or pituitary tumour recurrence, they had a history of malignancy or were concerned about malignancy, they had a history of diabetes mellitus, they had a history of adverse events related to growth hormone, reimbursement issues, other safety concerns, and other unknown reasons. HypoCCS was done in accordance with the Declaration of Helsinki and all applicable regulatory requirements in the participating countries. Ethics review board approval and written consent for data collection, electronic processing, and publication were obtained in accordance with national laws. Regular safety reviews were done by Eli Lilly and Co. 2
Procedures We assessed patients who had available information about age, sex, disease onset (adult onset or childhood onset), at least one follow-up visit after study entry, and whether they received growth hormone or not in HypoCCS. Fracture events were reported by the study sites during the study and were recorded in case report forms for pre-existing conditions and adverse events. Additionally, patients whose height had decreased by more than 5 cm at first appearance of height decrease since HypoCCS entry were regarded as having incident vertebral fractures.12,13 The identified fracture events were classified into vertebral fractures and non-vertebral fractures with the clinical information available in the event report. If several incident fractures per patient were reported, only the first event was included in this analysis. A patient was deemed to have pre-existing osteoporosis if their physician gave a diagnosis of osteoporosis or osteoporotic fracture at study entry. Some patients were receiving osteoporosis drugs without a classification of osteoporosis noted. No validation was done to support the reported diagnosis in the study (for example, using bone mineral density or radiography for validation). After case report forms were comprehensively reviewed, other confounders, in addition to pre-existing osteoporosis, were extracted for analysis: age; sex; weight; BMI (kg/m²); smoking; cause of growth hormone deficiency; duration of growth hormone deficiency; age at growth hormone deficiency diagnosis; history of fracture; family history of osteoporosis; growth hormone deficiency onset type (childhood onset or adult onset); impaired vision; history of growth hormone therapy before HypoCCS; average growth hormone dose (μg per day); concomitant use of corticosteroids, sex hormones, thyroid hormone, vasopressin analogs, lipid-altering medication, and osteoporosis medication; comorbid conditions (ie, diabetes mellitus and depression); postmenopausal status; and presence of multiple pituitary hormone deficiencies (ie, in addition to growth hormone deficiency, any of diabetes insipidus, secondary hypothyroidism, secondary hypoadrenalism, or secondary hypogonadism). Information about the concomitant medication of calcium, vitamin D, and bisphosphonates were not available.
Statistical analysis We calculated the crude incidence for fracture by dividing the number of first fractures reported during the study period by the total person-years at risk. Person-years were defined for patients with a new fracture in HypoCCS as the interval from HypoCCS entry to the first fracture event date and, for patients without a reported fracture, as the interval from HypoCCS entry to the last contact date. Initially, fracture incident rates were compared between growth hormone-treated and untreated groups by estimation of an unadjusted hazard ratio (HR) in a univariate Cox proportional hazard model in which growth hormone treatment was an independent variable. To assess possible confounding factors such as age, sex, smoking,
www.thelancet.com/diabetes-endocrinology Published online April 13, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00098-4
Articles
bodyweight, comorbid conditions (eg, depression), concomitant medications (eg, corticosteroids),3,14 and preexisting osteoporosis,15,16 the unadjusted HRs were estimated for all these possible risk factors. A full HR model was developed by entering the factors significant at α level 0·1 in the univariate models and those factors that are clinically relevant (growth hormone therapy, age, sex, cause of growth hormone deficiency, history of fracture, depression, corticosteroid use, bodyweight, family history of osteoporosis, smoking, multiple pituitary hormone deficiencies [hypoadrenalism, hypothyroidism, hypogonadism, or diabetes insipidus in addition to growth hormone deficiency], impaired vision, lipid medication, menopause status, growth hormone deficiency onset type [childhood onset or adult onset], age at growth hormone deficiency diagnosis, previous growth hormone treatment, duration of growth hormone deficiency, medication for osteoporosis treatment and prevention, oestrogen replacement, androgen replacement, thyroid hormone replacement, pre-existing osteoporosis, and diabetes mellitus). Backward model reduction strategy was used to derive a reduced model with factors significant at level 0·05. The results from reduced models are presented in this Article. Hazard proportionality assumption was checked and interaction of time-to-event (or censoring) was introduced into the model if the assumption did not hold.17 The predictive effect of pre-existing osteoporosis for fracture was estimated in a model for all patients. Growth hormone treatment might have different effects on fracture risk in those with and those without osteoporosis;15,16 therefore, separate models were also built for the populations of patients with and without preexisting osteoporosis. To test the robustness of growth hormone treatment effect, the same modelling process was repeated for prediction of vertebral and non-vertebral fractures separately. The magnitude of effect of growth hormone replacement therapy was speculated to vary by age. Therefore, the same modelling process was repeated in separate cohorts of patients older than 60 years and those aged 60 years or younger. The variables for insulinlike growth factor 1 standard deviation score, bone mineral density, lean body mass, and fat mass were not included in model because of a large number of patients with missing values for these measures. SAS version 9.2 was used for all statistical analyses.
Role of the funding source The funder of the study, in collaboration with the study authors, completed the tasks of study design, data collection, data analysis, data interpretation, and writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Results Between Jan 3, 1996, and Dec 15, 2012, we enrolled 10 673 patients to this study. 1032 patients were excluded
from assessment because of incomplete data (980 patients did not have a follow-up visits, whereas others had missing data for age, sex, type of onset, or growth hormone therapy). Thus, we analysed a total of 9641 patients, of whom 8374 received growth hormone and 1267 did not. Table 1 summarises factors possibly associated with fracture for both groups. This summary was used to identify possible baseline imbalances between treatment groups. The group that received growth hormone was younger than the group that did not and had a younger age at diagnosis of growth hormone deficiency; a longer mean follow-up; a higher percentage of women; and more patients with depression, oestrogen replacement, previous growth hormone treatment, diabetes insipidus, and idiopathic growth hormone deficiency. The group that received growth hormone also had lower percentages of pre-existing osteoporosis; adult-onset growth hormone deficiency; multiple pituitary hormone deficiencies; hypogonadism (5858 [70%] of 8374 vs 919 [73%] of 1267); diabetes mellitus; users of lipid-lowering medication, osteoporosis medication, and corticosteroids; and impaired vision. The group that received growth hormone had a higher percentage of patients with craniopharyngioma and a smaller percentage with pituitary adenoma. The mean dose for the group that received growth hormone was 447 μg per day. During a mean follow-up of 4·6 years (SD 3·8; median 3·4 years [IQR 1·6–6·4]) in HypoCCS, a total of 539 patients were reported to have at least one fracture event. Of the initial fracture events reported, 221 (41%) were vertebral fractures and 318 (59%) were non-vertebral fractures. Lower annual incidence rates of vertebral and non-vertebral fractures combined were reported in patients treated with growth hormone than in patients who did not receive growth hormone (1·19% vs 1·91%, p
Fracture risk in adult patients treated with growth hormone replacement therapy for growth hormone deficiency: a prospective observational cohort study Daojun Mo, Maria Fleseriu, Rong Qi, Nan Jia, Christopher Jeremy Child, Roger Bouillon*, Dana Sue Hardin*
Summary Background To our knowledge, no controlled studies of the effects of long-term growth hormone replacement on fracture risk in adult patients with growth hormone deficiency exist. We assessed the effect of growth hormone treatment on fracture risk in patients with growth hormone deficiency from the international Hypopituitary Control and Complications Study (HypoCCS) surveillance database.
Lancet Diabetes Endocrinol 2015
Methods In this prospective cohort study, patients with growth hormone deficiency were analysed from the HypoCCS database of adults with hypopituitarism from the USA, Canada, Japan, and 14 European countries. Patients were eligible if they were aged 18 years or older and had an established diagnosis of growth hormone deficiency, either alone or with multiple pituitary hormone deficiencies, as identified by clinical history and biochemical testing. Patients were assessed over a mean follow-up period of 4·6 years (SD 3·8). The effect of growth hormone treatment on fracture risk was assessed by Cox proportional hazard modelling with adjustment for several confounders.
See Online/Comment http://dx.doi.org/10.1016/ S2213-8587(15)00036-4
Findings Between Jan 3, 1996, and Dec 15, 2012, we enrolled 10 673 patients to this study. Of the enrolled patients, 1032 patients were excluded from assessment because of incomplete data, leaving 9641 in the analysis cohort. Of these patients, 8374 of received growth hormone and 1267 did not. Annual fracture incidence rate was lower in patients who received growth hormone than in those who did not (fracture incidence rate 1·19% vs 1·91%, hazard ratio [HR] 0·69, 95% CI 0·54–0·88). However, no difference in fracture risk was observed between patients who did and did not receive growth hormone treatment in the subgroup of patients with pre-existing osteoporosis (n=826; 0·97, 0·48–1·95). Interpretation Our results suggest that growth hormone replacement therapy could be protective against fracture for adult patients with growth hormone deficiency without previously reported osteoporosis. Starting growth hormone therapy before the onset of osteoporosis might be optimum for bone health of adult patients with growth hormone deficiency. Funding Eli Lilly and Co.
Introduction Low bone mass or density is a characteristic feature of adult patients with growth hormone deficiency. Findings from previous studies1–6 showed that adult patients with severe growth hormone deficiency had lower bone mineral density7 and were two to five times more likely to have a fracture than those without growth hormone deficiency.1–6 The effects of growth hormone replacement on biomarkers of bone turnover, bone mineral density, and bone mineral content have been assessed in clinical trials, including in patients with childhood-onset and adult-onset growth hormone deficiency. Results from these studies7 have shown that growth hormone has a biphasic effect on bone. In the first 12 months of growth hormone treatment, bone mineral density might not increase and might even decrease. After 24 months, bone mineral content and bone mineral density increased from baseline in the lumbar spine and proximal femur in adults with either childhood-onset or adult-onset growth hormone deficiency, as shown in a detailed meta-analysis8 of randomised controlled studies. In long-term
Published Online April 13, 2015 http://dx.doi.org/10.1016/ S2213-8587(15)00098-4
*Contributed equally Eli Lilly and Company, Lilly Research Laboratories, Indianapolis, IN, USA (D Mo MD, R Qi MS, N Jia PhD, D S Hardin MD); Northwest Pituitary Center, Departments of Medicine and Neurological Surgery, Oregon Health and Science University, Portland, OR, USA (M Fleseriu MD); Eli Lilly and Company, Lilly Research Laboratories, Windlesham, UK (C J Child PhD); and Clinic and Laboratory of Endocrinology, Gasthuisberg, KU Leuven, Leuven, Belgium (Prof R Bouillon MD) Correspondence to: Dr Daojun Mo, Lilly Corporate Center, Eli Lilly and Company, Indianapolis, IN 46285, USA [email protected]
prospective studies,8,9 growth hormone replacement therapy for up to 15 years induced a sustained increase in total body and lumbar spine (L2–L4) bone mineral content and bone mineral density. These findings suggest that long-term growth hormone treatment might help to prevent fractures in adult patients with growth hormone deficiencies. Moreover, growth hormone replacement therapy increases muscle mass and strength, which might have additional beneficial effects on fracture risk. One Swedish study3 reported a decreased relative fracture incidence in a subgroup of men with adult-onset growth hormone deficiency (n=389) compared with matched population controls during 6-year growth hormone and sex hormone replacement therapy, although this was not reported in women. Two other retrospective studies2,10 also suggested that growth hormone therapy in adults with growth hormone deficiencies might reduce vertebral fracture prevalence. However, no prospective controlled studies have reported effects of long-term growth hormone replacement on fracture occurrence in adult patients with growth hormone deficiencies.6,8
www.thelancet.com/diabetes-endocrinology Published online April 13, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00098-4
1
Articles
The Hypopituitary Control and Complications Study (HypoCCS),11 an international post-marketing surveillance study done by Eli Lilly and Co, prospectively obtained clinical outcome information in adult patients with growth hormone deficiency from the USA, Canada, Japan, and 14 European countries (Austria, Belgium, Iceland, Czech Republic, Denmark, France, Germany, Hungary, Italy, Norway, Spain, Sweden, the Netherlands, and UK). The study obtained information about the occurrence of cancer, mortality, cardiovascular diseases, diabetes, metabolic syndromes, and fractures and associated risk factors. The aim of this analysis was to assess growth hormone replacement effects on fracture incidence by comparing patients in the HypoCCS cohort who received growth hormone treatment with those that did not, while taking possible confounders into account.
Methods Study design and patients In this prospective cohort study, we enrolled patients who met the adult growth hormone deficiency indication for Humatrope (somatropin [rDNA origin] for injection; Eli Lilly and Co, Indianapolis, USA), according to the approved package insert for each participating country. Patients were included if they were aged 18 years or older and had an established diagnosis of growth hormone deficiency, either alone or with multiple pituitary hormone deficiencies, as identified by clinical history and biochemical testing.11 The investigating site physician decided what diagnostic approach was used and whether growth hormone therapy should be given to patients. According to the observational design of HypoCCS, the choice of whether to receive growth hormone replacement therapy was made by patients in consultation with their physicians. Patients were not eligible for enrolment if they had active or unresolved contraindication(s) to growth hormone therapy (including evidence or suspicion of active malignancy, evidence of sustained pituitary or other intracranial tumour activity, or pregnancy) or if they were breastfeeding. Patients did not receive growth hormone replacement therapy in HypoCCS for several reasons: they did not want injections, they felt well and did not feel the need for therapy, they were concerned about CNS or pituitary tumour recurrence, they had a history of malignancy or were concerned about malignancy, they had a history of diabetes mellitus, they had a history of adverse events related to growth hormone, reimbursement issues, other safety concerns, and other unknown reasons. HypoCCS was done in accordance with the Declaration of Helsinki and all applicable regulatory requirements in the participating countries. Ethics review board approval and written consent for data collection, electronic processing, and publication were obtained in accordance with national laws. Regular safety reviews were done by Eli Lilly and Co. 2
Procedures We assessed patients who had available information about age, sex, disease onset (adult onset or childhood onset), at least one follow-up visit after study entry, and whether they received growth hormone or not in HypoCCS. Fracture events were reported by the study sites during the study and were recorded in case report forms for pre-existing conditions and adverse events. Additionally, patients whose height had decreased by more than 5 cm at first appearance of height decrease since HypoCCS entry were regarded as having incident vertebral fractures.12,13 The identified fracture events were classified into vertebral fractures and non-vertebral fractures with the clinical information available in the event report. If several incident fractures per patient were reported, only the first event was included in this analysis. A patient was deemed to have pre-existing osteoporosis if their physician gave a diagnosis of osteoporosis or osteoporotic fracture at study entry. Some patients were receiving osteoporosis drugs without a classification of osteoporosis noted. No validation was done to support the reported diagnosis in the study (for example, using bone mineral density or radiography for validation). After case report forms were comprehensively reviewed, other confounders, in addition to pre-existing osteoporosis, were extracted for analysis: age; sex; weight; BMI (kg/m²); smoking; cause of growth hormone deficiency; duration of growth hormone deficiency; age at growth hormone deficiency diagnosis; history of fracture; family history of osteoporosis; growth hormone deficiency onset type (childhood onset or adult onset); impaired vision; history of growth hormone therapy before HypoCCS; average growth hormone dose (μg per day); concomitant use of corticosteroids, sex hormones, thyroid hormone, vasopressin analogs, lipid-altering medication, and osteoporosis medication; comorbid conditions (ie, diabetes mellitus and depression); postmenopausal status; and presence of multiple pituitary hormone deficiencies (ie, in addition to growth hormone deficiency, any of diabetes insipidus, secondary hypothyroidism, secondary hypoadrenalism, or secondary hypogonadism). Information about the concomitant medication of calcium, vitamin D, and bisphosphonates were not available.
Statistical analysis We calculated the crude incidence for fracture by dividing the number of first fractures reported during the study period by the total person-years at risk. Person-years were defined for patients with a new fracture in HypoCCS as the interval from HypoCCS entry to the first fracture event date and, for patients without a reported fracture, as the interval from HypoCCS entry to the last contact date. Initially, fracture incident rates were compared between growth hormone-treated and untreated groups by estimation of an unadjusted hazard ratio (HR) in a univariate Cox proportional hazard model in which growth hormone treatment was an independent variable. To assess possible confounding factors such as age, sex, smoking,
www.thelancet.com/diabetes-endocrinology Published online April 13, 2015 http://dx.doi.org/10.1016/S2213-8587(15)00098-4
Articles
bodyweight, comorbid conditions (eg, depression), concomitant medications (eg, corticosteroids),3,14 and preexisting osteoporosis,15,16 the unadjusted HRs were estimated for all these possible risk factors. A full HR model was developed by entering the factors significant at α level 0·1 in the univariate models and those factors that are clinically relevant (growth hormone therapy, age, sex, cause of growth hormone deficiency, history of fracture, depression, corticosteroid use, bodyweight, family history of osteoporosis, smoking, multiple pituitary hormone deficiencies [hypoadrenalism, hypothyroidism, hypogonadism, or diabetes insipidus in addition to growth hormone deficiency], impaired vision, lipid medication, menopause status, growth hormone deficiency onset type [childhood onset or adult onset], age at growth hormone deficiency diagnosis, previous growth hormone treatment, duration of growth hormone deficiency, medication for osteoporosis treatment and prevention, oestrogen replacement, androgen replacement, thyroid hormone replacement, pre-existing osteoporosis, and diabetes mellitus). Backward model reduction strategy was used to derive a reduced model with factors significant at level 0·05. The results from reduced models are presented in this Article. Hazard proportionality assumption was checked and interaction of time-to-event (or censoring) was introduced into the model if the assumption did not hold.17 The predictive effect of pre-existing osteoporosis for fracture was estimated in a model for all patients. Growth hormone treatment might have different effects on fracture risk in those with and those without osteoporosis;15,16 therefore, separate models were also built for the populations of patients with and without preexisting osteoporosis. To test the robustness of growth hormone treatment effect, the same modelling process was repeated for prediction of vertebral and non-vertebral fractures separately. The magnitude of effect of growth hormone replacement therapy was speculated to vary by age. Therefore, the same modelling process was repeated in separate cohorts of patients older than 60 years and those aged 60 years or younger. The variables for insulinlike growth factor 1 standard deviation score, bone mineral density, lean body mass, and fat mass were not included in model because of a large number of patients with missing values for these measures. SAS version 9.2 was used for all statistical analyses.
Role of the funding source The funder of the study, in collaboration with the study authors, completed the tasks of study design, data collection, data analysis, data interpretation, and writing of the report. The corresponding author had full access to all the data in the study and had final responsibility for the decision to submit for publication.
Results Between Jan 3, 1996, and Dec 15, 2012, we enrolled 10 673 patients to this study. 1032 patients were excluded
from assessment because of incomplete data (980 patients did not have a follow-up visits, whereas others had missing data for age, sex, type of onset, or growth hormone therapy). Thus, we analysed a total of 9641 patients, of whom 8374 received growth hormone and 1267 did not. Table 1 summarises factors possibly associated with fracture for both groups. This summary was used to identify possible baseline imbalances between treatment groups. The group that received growth hormone was younger than the group that did not and had a younger age at diagnosis of growth hormone deficiency; a longer mean follow-up; a higher percentage of women; and more patients with depression, oestrogen replacement, previous growth hormone treatment, diabetes insipidus, and idiopathic growth hormone deficiency. The group that received growth hormone also had lower percentages of pre-existing osteoporosis; adult-onset growth hormone deficiency; multiple pituitary hormone deficiencies; hypogonadism (5858 [70%] of 8374 vs 919 [73%] of 1267); diabetes mellitus; users of lipid-lowering medication, osteoporosis medication, and corticosteroids; and impaired vision. The group that received growth hormone had a higher percentage of patients with craniopharyngioma and a smaller percentage with pituitary adenoma. The mean dose for the group that received growth hormone was 447 μg per day. During a mean follow-up of 4·6 years (SD 3·8; median 3·4 years [IQR 1·6–6·4]) in HypoCCS, a total of 539 patients were reported to have at least one fracture event. Of the initial fracture events reported, 221 (41%) were vertebral fractures and 318 (59%) were non-vertebral fractures. Lower annual incidence rates of vertebral and non-vertebral fractures combined were reported in patients treated with growth hormone than in patients who did not receive growth hormone (1·19% vs 1·91%, p
