Early Acid Suppression Therapy Exposure and Fracture in Young Children

A retrospective cohort of children born in the Military Healthcare System (MHS) who received continuous MHS care for at least the first 2 and up to 14 years of life was formed. The MHS provides health care to military members, retirees, and dependent family members. Care is provided around the world at military treatment facilities and by civilian providers. Records are maintained for all inpatient and outpatient medical care and outpatient prescription medications.

Children born in the MHS between October 1, 2001, and September 30, 2013, were included. Children with diagnoses that included osteogenesis imperfecta, cholestasis, or child maltreatment, identified via International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) codes, were excluded because these conditions independently predispose a child to fractures.²² Cholestasis is oftentimes associated with prolonged parenteral nutrition or chronic liver disease, both of which can increase the risk of metabolic bone disease.²⁴ Birth hospitalizations of ≥7 days were used as a surrogate for admission to the NICU, and children with these extended birth hospitalizations were excluded in an attempt to limit the study population to otherwise lower-risk infants.

Data extracted from civilian and military outpatient pharmacy records were used to identify the days supply and refills of all AST prescriptions in the first 5 years of life. Children were classified as having been prescribed either a PPI, an H₂RA, or both if they had ≥1 outpatient prescription for these medications during the first year of life. For children who started AST before 1 year of age, prescription data were followed for 5 years to track total length of treatment. These children were compared with children who were not prescribed any AST during the first 5 years of life. Those who initiated AST between the ages of 1 and 5 years were excluded from the primary analysis because later AST prescriptions are uncommon and are more likely related to chronic disease. Children who initiated AST at 12 to 24 months of age were included in a secondary analysis. Pharmacy records were also used to identify any children prescribed an anti-epileptic medication at any point in the study because anti-epileptics have been linked with increased fracture risk.^25,26 

ICD-9-CM codes were used to identify fractures in the outpatient record that occurred both before and after the age of 1 year by using the Agency for Healthcare Research and Quality Clinical Classification System category for fractures.²⁷ Any child with a fracture before 1 year of age was categorized as having a previous fracture, because previous fractures can be associated with increased fracture risk.²⁸ Visits for the same fracture type within 6 months were excluded because they were likely follow-up visits and not visits for new fractures. ICD-9-CM codes were used to identify preterm birth (<37 + 0/7 weeks’ gestation) and low birth weight (LBW) (weight ≤2500 g) in the inpatient record of any children who were not already excluded for a birth hospitalization of ≥7 days. ICD-9CM codes were also used to identify obesity or overweight status in the outpatient record, because increased BMI has been linked to increased fracture risk and decreased bone mineral density.^29,30 

Group differences were compared by using the Wilcoxon rank test and χ² analysis. A Cox proportional hazard model was used to assess the hazard ratio (HR) of fracture by AST use. An adjusted analysis was used to control for sex, preterm birth, LBW, obesity and/or overweight, anti-epileptics, and history of fracture before 12 months of age. When Schoenfeld residual testing revealed a violation of the proportionality assumption, an interaction with time was added to the model.

Two secondary analyses were used to explore the impact of duration of prescription AST use and age of AST medication initiation. Length of prescription use was divided into quartiles for both H₂RA and PPI groups and for the combination group. Each quartile was compared with controls who were not treated. A Cox proportional hazard model was used to assess the hazard of fracture by days on PPIs, H₂RAs, and both medications; adjusted analyses were used to control for sex, preterm birth, LBW, obesity and/or overweight, anti-epileptics, fracture before 1 year of age, and time. By using 2 models, a sensitivity analysis was performed to explore whether previous fracture affected the relationship between AST and fracture.

Prescription patterns were next grouped by age at initiation. To fully explore age at AST initiation, the final analysis population was expanded to include children who initiated an AST from age 0 to 24 months. Children were classified as being prescribed AST in the first 6 months of life, at 6 to 12 months, and at 12 to 24 months and were compared with controls not prescribed any AST in the first 5 years. An analysis of variance with Tukey comparison was used to examine the impact of AST age of initiation on length of use. A Cox proportional hazard model was used to assess the relative incidence of fracture by days on PPIs, H₂RAs, and both medications; adjusted analyses were used to control for male sex, preterm birth, LBW, obesity and/or overweight, anti-epileptics, fracture before 1 year of age, and time when indicated.

Stata Intercooled 13 (Stata Corp, College Station, TX) software was used for the statistical analysis. P <.05 was considered statistically significant. The study was reviewed and approved by the Uniformed Services University Institutional Review Board.

A total of 1 190 544 infants were born in the MHS between 2002 and 2015. Infants were excluded if they were classified as being at increased risk for facture because of an extended birth hospitalization of >7 days (40 099; 3.3%), maltreatment (9005; 0.74%), cholestasis (407; 0.03%), or osteogenesis imperfecta (174; 0.01%). Children also were excluded if they received MHS care for <2 years (266 309; 22%) or if they initiated AST after 1 year of age (22 919; 2%). Of the remaining 851 631 infants, 754 345 (89%) did not initiate AST and were study controls, and 97 286 (11%) initiated AST in the first year of life. Of those on AST, 7998 (9%) were prescribed a PPI, 71 578 (73%) were prescribed an H₂RA, and 17 710 (18%) were prescribed both (Supplemental Fig 2). Compared with those not prescribed an acid suppressant in the first 5 years, children who initiated AST by 1 year of age were enrolled and followed in the MHS for a shorter period of time and were less likely to have overweight or obesity. They were more likely to be male, born preterm, and born with LBW. AST was not associated with anti-epileptic prescription medication use (Table 1).

In the unadjusted analysis, the hazard of fracture after 1 year of age was associated with being prescribed a PPI (HR 1.23; 95% CI 1.15–1.32), being prescribed an H₂RA (HR 1.13; 95% CI 1.10–1.15), and being prescribed both an H₂RA and a PPI in the first year of life (HR 1.32; 95% CI 1.26–1.38; Table 2). An increased hazard of fracture was associated with male sex, overweight or obesity, and previous fracture; a decreased fracture hazard was associated with LBW (Table 2). Preterm birth and anti-epileptic medication were not associated with fracture (Table 2). A sensitivity analysis revealed that previous fracture did not affect the association between AST and fracture.

After adjustment for covariates, including time, the association between AST prescriptions and fracture remained. Fracture hazard increased 23% in children prescribed a PPI (HR 1.23; 95% CI 1.14–1.31) and 31% in children prescribed both an H₂RA and a PPI in the first year of life (HR 1.31; 95% CI 1.25–1.37; Table 2, Fig 1). The adjusted fracture hazard was increased with male sex and previous fracture, was decreased with LBW, and was not associated with overweight or obesity, with anti-epileptic use, or with preterm birth (Table 2). H₂RA initiation in the first year of life was not associated with increased fracture.

An adjusted analysis used to explore effects of prescription length in days suggests that fracture hazard increases with duration of AST exposure. Medication days were divided into quartiles for each medication class. The median prescription length was 60 days for monotherapy with PPIs or H₂RAs and 192 days for combination therapy (Table 3). Children prescribed a PPI for 0 to 30 days had a 19% increased fracture hazard, and those prescribed PPIs for >150 days had a 41% increased fracture hazard (Table 3). Children prescribed a combination of PPIs and H₂RAs for 0 to 120 days had a 17% increased fracture hazard compared with controls with no AST use in the first 5 years of life (Table 3). Use of PPIs and H₂RAs was associated with increased fracture hazard for medication use for 120 to 192 days (31%), 192 to 338 days (20%), and >338 days (50%; Table 3). Use of H₂RAs alone was associated with increased fracture hazard in those prescribed the medications for 0 to 30 days (14%), 60 to 120 days (16%), and >120 days (22%; Table 3).

Compared with controls with no AST prescriptions before age 5, PPI initiation at 0 to 6 months of life was associated with a 23% increased fracture hazard, PPI initiation at 6 to 12 months was associated with a 21% increase, and PPI initiation at 12 to 24 months was not associated with an increased fracture hazard in the adjusted analysis (Table 4). Children prescribed an H₂RA did not have an increased hazard of fracture regardless of when medication was initiated (Table 4). Fracture hazard was increased in children prescribed both a PPI and an H₂RA, compared with controls with no AST use; PPI and H₂RA initiation at 0 to 6 months was associated with a 32% increase, at 6 to 12 months a 23% increase, and at 12 to 24 months a 38% increase in adjusted analysis (Table 4). Duration of AST exposure decreased with age of initiation (P < .001), with the mean length of use differing for all 4 groups.

AST medication use during the first year of life was associated with increased fracture hazard in children. The greatest fracture hazard was associated with combination PPI and H₂RA use during the first year of life. PPI use alone was associated with an increased fracture hazard, whereas H₂RA use alone did not significantly impact fracture hazard. Fracture hazard increased with duration of AST use, which suggests a possible dose-dependent response, and with younger age of AST initiation with PPIs, alone or in combination with H₂RAs.

These findings are consistent with adult studies linking AST with increased risk for osteoporotic fracture.^31,32 Results are also consistent with research linking fracture with early PPI use in infants who were sick and healthy,^22, and with findings that PPIs did not increase fracture risk in older children,²¹ because AST initiation after age 1 was not associated with fracture.

Similar to adult studies,³³ fracture hazard increased with prescription duration. The median prescription length for combined therapy (192 days) was considerably longer than that for PPIs (60 days) or H₂RAs (60 days; Table 3), possibly suggesting that the increased fracture hazard is related to duration of medication use, or perhaps an unidentified confounding disease, rather than a synergistic effect of combined AST.

Fracture hazard increased with younger age of initiation, with children starting at <6 months of age having greater fracture hazard than those starting at 12 to 24 months of age. Results are consistent with the limited previous studies that revealed an increased fracture rate in infants but not in older children.²¹ This may be due to rapid bone turnover that occurs in the first year of life,³⁴ or this may be because infants who start earlier may continue AST treatment for longer periods of time. Although diagnostic codes indicating prescription rationale could not be linked, results may be related to different indications for medications in older versus younger children. Previous studies infer AST initiation at age ≤6 months is for treatment of symptoms of GER or GER disease,³⁵ which is difficult to diagnose in infants, and that the majority of children outgrow symptoms by age 1.

The exact mechanism linking AST with bone health is unclear; however, biologically plausible hypotheses exist. It was previously postulated that PPIs could impair calcium absorption because of their effect on gastric acid secretion,³⁶ leading to compensatory hyperparathyroidism and increased osteoclast activity.³⁷ Adult studies, however, revealed no significant difference in bone density or blood calcium levels in patients on long-term PPI therapy versus none.^38,39 Another theory is that PPIs change osteoclast activity and viability.⁴⁰ Osteoclasts have a proton pump on their surface, which, if inhibited, could lead to osteoclast dysfunction and unopposed osteoblast activity resulting in a disorganized and potentially more fragile bone matrix.² One study of pantoprazole on human osteoclasts in vitro revealed that exposed osteoclasts had significantly lower viability and activity, although no change was seen in osteoclast-specific gene expression.⁴¹ 

The effect of H₂RAs on bone development has been less studied compared with that of PPIs and is less strongly associated with fracture. Older adult users of H₂RAs showed decreased bone mineral density on dual-energy radiograph absorptiometry scans, but this was seen only in those not taking calcium supplements.³² One adult study revealed no association between H₂RAs and fracture,⁴² and another revealed reduced fractures in adults on combination H₂RA and PPI therapy, compared with PPI use alone.⁴³ The authors of these studies, however, explored the effect of AST on the adult bone, which may react differently to PPI and H₂RA exposure than the rapidly growing and remodeling bone of young children.

Although an association between AST use in infancy and increased fracture risk was found, findings do not explore fracture patterns or mechanism of injury. Children with ICD-9-CM codes for child maltreatment were excluded; however, the incidence of recorded maltreatment in this study population was low, likely indicating that all children who were maltreated were not excluded. Only more severe maltreatment is regularly coded in the electronic health record. In addition, infant victims of maltreatment often present with symptoms that include fussiness and GER or vomiting, which could result in an AST prescription.^44,45 

Study findings do not establish a causal relationship between PPI exposure and fracture. Practitioners should continue to take appropriate actions when they evaluate children with histories, injuries, or fracture patterns that cause concern for potential abuse, regardless of AST exposure.

This study is additionally limited by use of prescriptions as a proxy for taking medication; researchers were unable to assess if children consistently used the medication for the full prescription period and were unable to account for different dosing methods or prescribing patterns over time. In the study, we also excluded appointments for the same type of fracture within 6 months as follow-up appointments; this determination may have resulted in misclassification of some new cases of fracture as follow-up. Although factors most known to be associated with fracture in young children were accounted for, other potential confounders, including socioeconomic status, geographic region, breastfeeding status, use of other medications, or comorbid conditions besides those in this study’s exclusion criteria, were not captured or included. Finally, the use of such a large sample may have overpowered our analyses.

This study has multiple strengths, including a large population and the ablity to identify and exclude children with conditions that place them at highest risk for fractures. The MHS provides free health care and free or low-cost prescriptions, reducing access-to-care and ascertainment bias. Finally, in this study, we examined the impact of AST prescribed during infancy on fracture risk over a longer time period (compared with that in previously reported studies).

AST used in the first year of life, especially PPIs, are associated with increased fracture hazard in children. Results should not be interpreted to suggest that PPIs or H₂RAs alone explain fractures, which is important in suspected cases of nonaccidental trauma. Results indicate longer AST use and earlier initiation may increase fracture hazard. Practitioners should be aware of the potential for fracture when considering treatment with AST versus lifestyle changes and watchful waiting. If AST use is necessary, providers should limit prescriptions to a single drug and limit their duration when possible.

AST

acid suppression therapy

GER

gastroesophageal reflux

H₂RA

histamine H₂-receptor antagonist

HR

hazard ratio

ICD-9-CM

International Classification of Diseases, Ninth Revision, Clinical Modification

LBW

low birth weight

MHS

Military Healthcare System

PPI

proton pump inhibitor