Flunisolide

Objective: To examine nonvertebral fracture risk in relation to inhaled corticosteroid (ICS) exposure among adults with respiratory disease. Design and patients: Nested case-control study within a cohort of 89,877 UnitedHealthcare members aged [greater than or equal to] 40 years with physician insurance claims for COPD or asthma, enrolled for [greater than or equal to] 1 year from January 1, 1997 to June 30, 2001.

Methods: Cases (n = 1,722) represented patients with a first treated nonvertebral fracture (the index date is the first fracture claim). Control subjects (n = 17,220) were randomly selected from the person-time and assigned a random index date. ICS exposure was ascertained I month, 3 months, 6 months, and 12 months before the index date, with estimated cumulative dose through 0 to 6 months, 7 to 12 months, and 0 to 12 months. Covariates included demographics, oral corticosteroid and other medication exposure, comorbidities, and indicators of respiratory disease severity. Odds ratios (ORs) adjusted for all covariates were estimated by logistic regression.

Results: No increased fracture risk with ICS exposure as a class or with fluticasone propionate alone was detected. ORs for exposure in the preceding 30 days were 1.05 (95% confidence interval [CI], 0.89 to 1.24), 1.13 (95% CI, 0.90 to 1.40), and 0.97 (95% CI, 0.78 to 1.21) for all ICS, fluticasone propionate, and other ICS, respectively. No dose-response effect was present. Among patients with COPD only (n = 6,932), no increased risk was found for recent ICS exposure (OR, 0.86; 95% CI, 0.59 to 1.25).

Conclusions: Concern about nonvertebral fracture risk should not strongly influence the decision to use recommended doses of ICS for adult patients with asthma or COPD in managed-care settings in the United States. This study could not evaluate very-high ICS dose, long-term ICS exposure, or vertebral fracture risk.

Key words: asthma; COPD; corticosteroids, inhaled; fractures; pulmonary diseases, chronic obstructive

Abbreviations: CI = confidence interval; ICD-9 = International Classification of Diseases, Ninth Revision; ICS = inhaled corticosteroid; OR = odds ratio

**********

Inhaled corticosteroid (ICS), recommended as first-line therapy for patients with persistent asthma, is increasingly being used for the management of COPD. (1,2) The benefit of ICS in the management of COPD is less clear than that for asthma, but several 3-year randomized trials (3-5) of ICSs in persons with COPD have demonstrated small improvements in forced expiratory volume, a reduced rate of exacerbations, and some improvement in respiratory symptoms and health status, despite no reduction in the rate of respiratory decline. The Global Initiative for Chronic Obstructive Lung Disease (6) in 2001 proposed ICSs for the treatment of moderate-to-severe COPD and recurrent exacerbations. The 2003 update (7) narrowed the recommendation to include only patients with severe or very severe COPD and frequent exacerbations. The 2004 American Thoracic Society/ European Respiratory Society COPD treatment guidelines (8) recommend the addition of ICSs to long-acting inhaled [beta]agonists for patients with more advanced disease (FE[V.sub.1] < 50% predicted) to reduce exacerbations and improve health status.

Extended use of systemic corticosteroids leads to osteoporosis and fracture, but evidence incriminating ICS use is weak. (9-14) Despite lower systemic absorption than oral preparations, ICSs have been associated with bruising, glaucoma, oral candidiasis, and cataracts. (2,15-19) Studies (2, 15,20) of the effects of ICS on bone show inconsistent results. There is a direct relation between higher doses of ICS and small reductions in bone mineral density over time in several studies, (21-23) but other researchers have found no bone loss after 1 year even with high-dose ICS therapy. (24,25) Of three long-term clinical trials of ICS among patients with COPD, one trial (4) showed increased bone demineralization among patients treated with triamcinolone, one trial (3) found no relationship between inhaled budesonide and change in bone mineral density or reported fractures, and the third trial reported no difference in fracture incidence between patients treated with fluticasone propionate and those receiving placebo. Cohort studies conducted in health-care administrative databases have failed to demonstrate increased fracture risk at recommended doses of ICS, (26-28) while a single case-control study (29) found a small positive association of ICS with hip fracture.

Chronic disease is itself a risk factor for fracture, (30) and obstructive lung disease is a risk factor for fracture irrespective of corticosteroid therapy. (31-33) Patients with COPD have an estimated 36 to 60% prevalence of osteoporosis. (34)

We sought to investigate the relation of ICS use to risk of fracture by conducting a nested case-control study of nonvertebral fracture within a large cohort of adult patients with insurance claims for asthma and/or COPD identified from an automated database. To address the problem that COPD severity might be independently linked to both risk of osteoporotic fracture and to the use of ICS, the study was conducted within a population of patients with chronic respiratory disease. We identified many indicators of underlying respiratory disease severity, and corrected for their effects on the analysis. Nonvertebral fracture, with its clear indication for medical services and relatively uncomplicated anatomy, is a suitable condition for study within an automated database, (35) unlike vertebral fracture, which has a gradual and often painless onset that not may lead to claims for medical care utilization. (36)

MATERIALS AND METHODS

Source Population

Ingenix Epidemiology (Auburndale, MA) maintains a research database of UnitedHealthcare members who have both medical and prescription benefit coverage. For the time of this study, the research database included complete health services utilization information (eg, claims for all dispensations, inpatient and outpatient services, and procedures including the associated diagnoses and costs) on approximately 5 million persons in 17 states in the United States. Health insurance claims data are an accurate record of the clinical events described in medical and hospital records. (37,38) We assembled a cohort of adults aged [greater than or equal to] 40 years and enrolled in a health plan for at least 12 continuous months from January 1, 1997, through June 30, 2001, and with at least two claims for a physician visit in an outpatient setting or one claim in an inpatient setting with International Classification of Diseases, Ninth Revision (ICD-9) codes for asthma (ICD-9 code 493), or COPD (chronic bronchitis [ICD-9 code 491], emphysema [ICD-9 code 492], or chronic airway obstruction, not elsewhere classified [ICD-9 code 496]). The cohort eligibility date was the latest of the following dates: January 1, 1997, 12 months since joining UnitedHealthcare, 40th birthday, or first fulfillment of asthma or COPD definition.

Case Ascertainment

We evaluated treated nonvertebral fractures for all subjects from the date of cohort enrollment through June 2001. We identified nonvertebral fracture by a diagnosis code (ICD-9 codes 807, 808, 810-829, and 733.1, excluding 733.13) in association with a claim for physician outpatient or emergency department care or an inpatient hospitalization, and required that the patient receive fracture treatment within 2 weeks of the diagnosis. Hip fractures were counted only if there was inpatient care. The fracture date was the service date of the first claim in the fracture sequence. We excluded anyone with a fracture in the year before the cohort eligibility date, and those with malignancy other than nonmelanoma skin cancer. We also excluded those with insurance claims indicating malunion or nonunion, or late effects of fracture, as these most likely did not represent incident fractures.

Control Selection

The control group was sampled from the person-time of the respiratory cohort by two-tiered random sampling with replacement. Members of the respiratory cohort were selected at random, and for each a random day was chosen within the study interval. The sampling continued until the number of control subjects was 10 times the number of cases. The randomly assigned day served as the index date for control subjects. Future cases were eligible as control subjects.

Corticosteroid Exposure

ICSs were identified from pharmacy claims occurring in the 365 days before the fracture date in the cases or the randomly assigned date in the control subjects. All ICSs were grouped together as a class, and separated into those containing fluticasone propionate, alone or in combination with salmeterol, and all other types (beclomethasone dipropionate, budesonide, flunisolide, and triamcinolone). We identified dispensing in time windows before the index date: 0 to 30 days, 0 to 90 days, 0 to 180 days, and 0 to 365 days.

To estimate the cumulative dose of ICS, we multiplied the number of containers in each dispensing (canisters or disk packs) by the number of doses per container, and then multiplied by the drug strength to obtain the total dose in micrograms. Because pharmacy data do not directly reflect actual use, we classified the cumulative dose only in broad intervals, in the periods of 0 to 180 days, 181 to 365 days, and 0 to 365 days before the index date.

To standardize ICS dose, we applied a beclomethasone equivalent multiplier to the cumulative dose of each ICS in the three broad periods described above. The multiplier was calculated by dividing the relative topical potency of ICS, as derived from the MacKenzie skin-blanching test, (1) by the relative topical potency of beclomethasone dipropionate. We estimated the average daily dose for each period by dividing the standardized cumulative dose by the number of days in the time interval, and created categories representing very low, low, medium, and high average doses of beclomethasone dipropionate (1 to 167 [micro]g, 168 to 504 [micro]g, 505 to 840 [micro]g, and > 840 [micro]g, respectively). (1)

Oral corticosteroid exposure was estimated in a similar manner with cumulative dose in milligrams expressed in prednisone equivalents. Nasal corticosteroids of all types were ascertained separately from ICS, and grouped into one variable to indicate at least one dispensing in the 30 days before the index date.

Covariate Ascertainment

Covariates included demographic variables (age at index date, sex, geographic region of health plan, calendar year of cohort eligibility, season at index date, and year of joining health plan) and type of underlying respiratory disease (asthma, COPD, or both) at the cohort eligibility date. Potential confounding variables were ascertained in the year before the index date and expressed as dichotomous indicator variables. These included medical conditions (hypertension, hyperlipidemia, diabetes mellitus, stroke, obesity, rheumatoid/psoriatic arthritis, Cushing syndrome, hyperparathyroidism, inflammatory bowel disease, dementia, systemic lupus erythematosus, osteoporosis, osteopenia, osteomalacia, vitamin D deficiency, anemia, depression, back pain, and hormonal deficiency); medications (bisphosphonates, statins, anticonvulsants, estrogen, raloxifene, and calcitonin); indicators of intensity of medical care utilization for underlying respiratory disease (physician office visits, emergency department visits, and hospitalizations for asthma and/or COPD, outpatient and inpatient encounters for acute bronchitis, acute exacerbation of COPD, and pneumonia, respiratory therapy and procedures, antibiotics prescribed for asthma, COPD, pneumonia, or bronchitis, and noncorticosteroid respiratory medications). We also assessed total medical care costs excluding prescription costs in the year before the index date, divided into respiratory and nonrespiratory care costs, and determined the total costs of noncorticosteroid prescription medications. We divided all cost variables into quintiles for analysis.

Statistical Analysis

Relation of Individual Covariates to Fracture: Analyses were conducted using a statistical software package (Statistical Analysis Software, version 8.01; SAS Institute; Cary, NC). We first cross-tabulated individual variables by fracture case and control status, then estimated odds ratios (ORs) for each individual variable by regressing fracture on each covariate, adjusting for demographic variables.

Relation of Corticosteroid Exposure to Fracture: Corticosteroid use within each time window before the index date was represented by indicator variables, separately for all ICSs as a class, fluticasone propionate alone, ICSs other than fluticasone propionate, and oral corticosteroids. We used logistic regression to estimate the OR for fracture for each corticosteroid group separately for each time window before the index date, adjusting for demographic variables. Similar analyses were performed for cumulative dose. To determine whether oral corticosteroid use modified the effect of ICS on fracture risk, the analysis of cumulative dose of ICS in the year before the index date was repeated separately for those with no use and any use of oral corticosteroids in the year before the index date.

Multivariate Analysis: We used multiple logistic regression to model the association of ICS with fracture risk including oral corticosteroid exposure and all other covariates in the model, thus adjusting for underlying respiratory disease status and other fracture risk factors. Because control subjects were sampled at random from the person-time of the source population, the ORs from the logistic models provide valid estimates of the corresponding incidence rate ratios. (39) We constructed separate regression models for each of the four time windows of exposure for all ICSs as a class and for fluticasone propionate-containing preparations separated from all others. Separate models were constructed for cumulative dose of ICS related to fracture risk. To investigate whether the relation of ICS exposure to fracture risk differed by underlying respiratory disease, we repeated the analyses stratified by those with COPD and those with asthma as the underlying disease at cohort entry.

RESULTS

The respiratory cohort totaled of 89,877 people. Approximately 40% of the cohort had insurance claims evidence of COPD, while 56% had asthma, and approximately 4% had both at cohort entry. Just over three fourths of the cohort members were 40 to 59 years old, while 14% were 60 to 64 years old, and 10% were [greater than or equal to] 65 years old. We identified 1,722 persons with a treated nonvertebral fracture: 609 among 36,190 persons with COPD only, 1,033 among 50,313 persons with asthma only, and 80 among 3,374 persons with both conditions. For single fractures, the most common fracture sites were the lower (38.5%) and upper (25.2%) limbs. The least common sites were the pelvis (n = 10), hip (n = 26), and rib/sternum (n = 30). Multiple fractures were ascertained in 258 individuals (n = 533 fractures), with most combinations representing an upper limb and wrist fracture in the same claims sequence.

Underlying respiratory disease was distributed similarly between cases and the 17,220 control subjects, but a higher proportion of cases than control subjects were female (70.6% vs 58.9%). Although mean age was similar in cases and control subjects (52.9 years vs 52.2 years), respectively, a slightly higher percentage of cases (10.5%) than control subjects (7.7%) were [greater than or equal to] 65 years old. Cases and control subjects were distributed fairly evenly with respect to the other demographic variables (data not shown). As shown in Table 1, medical conditions individually related to increased fracture risk were bone disorders, vertebral fracture, conditions requiring corticosteroid therapy, depression, back pain, obesity, and diabetes mellitus. Increased medical utilization, including that related to respiratory disease, was also associated with increased fracture risk, as were exposure to anticonvulsants, bisphosphonates, hormones, and calcitonin. An inverse association was noted with statin use. Exposure to oral corticosteroids was not strongly related to fracture risk, nor was a dose effect seen for cumulative exposure over 6 months or 12 months. The risk associated with oral corticosteroids in the 6 months before the index date was somewhat elevated for the lowest dose (1 to 30 mg) category only. Nasal steroid use in the preceding 30 days was not associated with fracture. Fracture risk increased with quintile of cost for all three medical care cost indicators, with persons in the highest quintile of nonrespiratory care costs twice as likely as those in the lowest quintile to experience a fracture (data not shown).

Approximately 35% of the study population was exposed to ICS (15% fluticasone propionate and 22% other ICSs) and 27% to oral cortieosteroids in the year before the index date. Exposure to ICS was highest for those with claims for both asthma and COPD (46%) than asthma (29%) or COPD (21%) alone. Multivariate ORs for any exposure to ICS in time windows before the index date are shown in Table 2 for the full sample (n = 18,942) and those with COPD only (n = 6,932) and asthma only (n = 11,277) at cohort entry. For the full sample, ORs for recent exposure (0 to 30 days before the index date) were 1.05, 1.13, and 0.97 for all inhaled, fluticasone propionate alone, and other ICS, respectively, with the 95% CIs including the null value of one. There was no evidence of increased fracture risk for exposure to ICS as a class, for fluticasone propionate alone, or for all other ICSs in any of the time windows for the full sample, or either of the separate respiratory disease categories (Table 2).

There was no dose-response trend for fracture risk according to ICS dose (Table 3). Adjusted OR estimates were close to the null value for all dose categories in the year and in the 6 months before the index date. Only a small number of patients were exposed to the highest doses; thus; estimates for these categories show a high degree of variability, and these results should be interpreted with caution.

ORs of nonvertebral fracture were similar across ICS dose among users and nonusers of oral corticosteroids, except for an elevated risk among oral corticosteroid users in the next-to-highest (505 to 840 [micro]g) dose category of ICS (Table 4). Among the strata of oral corticosteroid users, the ORs associated with the medium ICS dose were 1.48, 1.45, and 1.75 for all ICSs, fluticasone propionate, and other ICSs, respectively.

DISCUSSION

This study showed elevations in the incidence of nonvertebral fractures in association with the previously reported risk factors of older age, female sex, chronic medical conditions associated with osteoporosis, depression, and anticonvulsant use, (40-42) but did not find a detrimental effect of ICS. A number of indicators of medical care utilization for respiratory disease were related to increased fracture risk, consistent with prior research showing that chronic respiratory disease is an independent predictor of fracture, (32,33) Our study differed from several prior studies (26,27,29) of fracture and ICS in that it was conducted completely within a cohort of patients with physician-diagnosed asthma and COPD. Comparisons were made between fracture cases and control subjects with the same underlying respiratory disease, and presumably a similar magnitude of fracture risk elevation associated with the underlying disease.

Studies have found little or no risk for nonvertebral fracture in association with ICS use in the doses that we were able to observe in a US managed-care population. Suissa and colleagues (28) studied Canadian patients with an average age of 81 years and accounted for the underlying respiratory disease; they reported similar findings of no increased fracture risk at recommended dose levels of ICS. Increased fracture risk was only detected at daily doses > 1,000 [micro]g; for those followed up for > 8 years, the relative risk of hip fracture after exposure to [greater than or equal to] 2,000 [micro]g of ICS was 1.61 (95% CI, 1.04 to 2.50). (28) Van Staa and colleagues, (26) using General Practice Research Database patients, failed to find an increased fracture risk (OR, 1.0; 95% CI, 0.94 to 1.06) when comparing ICS users to respiratory disease control subjects; only the comparison with nonrespiratory disease control subjects revealed a small elevation of risk (OR, 1.15; 95% CI, 1.10 to 1.20).

Lau and colleagues (27) studied hip fracture in elderly Canadian women not limited to those with respiratory disease, and did not find evidence of increased fracture risk with ICS (hazard ratio, 0.92; 95% CI, 0.75 to 1.12). Hubbard and colleagues, (29) using a case-control design in the General Practice Research Database (mean age, 79 years; 79% women) found an elevated OR of hip fracture related to ICS of 1.19 (95% CI, 1.10 to 1.28) after adjusting for oral corticosteroid use. However, a positive and statistically significant trend was reported for increasing dose of ICS and fracture risk driven by a very small group exposed to the highest dose, and this trend was not modified by oral corticosteroid use. (29) As only 3% of cases and 2% of control subjects had COPD, the underlying respiratory disease probably did not strongly influence the results. Our study, conducted in a younger population of managed-care enrollees, only identified 26 hip fractures, and did not have the power to evaluate this outcome separately from other types of fracture.

Several study limitations inherent to the data source should be noted. Classification of ICS exposure required estimation of the average daily dose from container size and number of containers dispensed. In calculating cumulative doses, we proceeded as if the entire dose occurred in the time window of dispensing. Accordingly, we used relatively large windows of time (6 months or 1 year) for estimating dose; almost certainly, the windows included periods without exposure. Because of the relatively high turnover in health plan membership, it was not practical to extend the baseline window to capture > 1 year of historical information, limiting our ability to evaluate the effects of recent ICS exposure adjusted for the long-term effects of oral corticosteroid or ICS use. However, an analysis restricted to health plan members with [greater than or equal to] 3 continuous years of enrollment (n = 7,340), yielded results similar to the entire sample (data not shown). Our ability to detect increased risk among individuals exposed to high doses of ICS was limited because so few individuals in our study population fell into this category.

The claims database does not include direct measures of lifestyle factors that affect bone mass, such as body weight adjusted for height, exercise, diet, or smoking, but we adjusted for physician-diagnosed medical conditions that predispose to inactivity and could be correlated with bone mass. Because the study population was derived from managed-care enrollees, the number of elderly persons (aged [greater than or equal to] 65 years) is underrepresented. Accordingly, our results may not be able to describe associations that would hold in a more elderly population or one without workplace benefits.

The strengths of the present study include the large number of COPD and asthma patients and the ability to ascertain the exposure status and other potential fracture risk factors from the computerized medical claims data prior to fracture occurrence without regard to case or control status, thus eliminating the chance of bias due to differential ascertainment. By using disease-matched control subjects (patients with COPD or asthma), and adjusting for many indicators of underlying disease severity, we could assess the effects of ICS use while controlling for confounding by indication and factors related to the underlying respiratory disease, such as smoking and inactivity, that are themselves risk factors for fracture.

After adjustment for severity of the underlying respiratory disease, neither patients with COPD nor patients with asthma using ICSs show a substantially increased risk of nonvertebral fracture compared to corresponding patients without ICS use. The results suggest that for contemporary managed-care settings in the United States, the concern about nonvertebral fracture should not be a factor that strongly influences the decision to use ICS at recommended doses for the management of patients with COPD. This study, however, could not evaluate the increased risk associated with high doses or very long-term ICS exposure, and did not evaluate the risk of vertebral fractures.

REFERENCES

(1) National Asthma Education and Prevention Program. Guidelines for the diagnosis and management of asthma: expert panel report II. Bethesda, MD; National Institutes of Health, 1997; NIH publication No. 97-4051

(2) Bonay M, Bancal C, Crestani B. Benefits and risks of inhaled corticosteroids in chronic obstructive pulmonary disease. Drug Safety 2002; 25:57-71

(3) Pauwels RA, Lofdahl C-G, Laitinen LA, et al. Long-term treatment with inhaled budesonide in persons with mild chronic obstructive pulmonary disease who continue smoking. N Engl J Med 1999; 340:1948-1953

(4) The Lung Health Study Research Group. Effect of triamcinolone on the decline in pulmonary function in chronic obstructive pulmonary disease. N Engl J Med 2000; 343: 1902-1909

(5) Burge PS, Calverley PMA, Jones PW, et al, on behalf of the ISOLDE study investigators. Randomised, double blind placebo controlled study of fluticasone propionate in patients with moderate to severe chronic obstructive pulmonary disease: the ISOLDE trial. BMJ 2000; 320:1297-1303

(6) Hurd S, Pauwels R. Global initiative for chronic obstructive pulmonary disease. Pulm Pharmacol Ther 2002; 15:353-355

(7) National Institutes of Health, National Heart, Lung, and Blood Institute. Global initiative for chronic obstructive pulmonary disease: global strategy for the diagnosis, management, and prevention of chronic obstructive pulmonary disease, updated 2003. Available at: http://www.goldcopd.com. Accessed September 16, 2004

(8) American Thoracic Society. Standards for the diagnosis and management of patients with COPD. Available at: http:// www.thoracic.org/COPD. Accessed September 16, 2004

(9) van Staa TP, Leufkens HG, Abenhaim L, et al. Use of oral corticosteroids and risk of fractures. J Bone Miner Res 2000; 15:1001-1005

(10) Adachi JD. Corticosteroid-induced osteoporosis. Int J Fertil Womens Med 2001; 46:190-205

(11) Naganthan V, Jones G, Nash P, et al. Vertebral fracture risk with long-term corticosteroid therapy. Arch Intern Med 2000; 160:2917-2922

(12) Baltzan MA, Suissa S, Bauer DC, et al. Hip fracture attributable to corticosteroid use [letter]. Lancet 1999; 353:1327

(13) van Staa TP, Leufkens HG, Abenhaim L, et al. Oral corticosteroids and fracture risk: relationship to daily and cumulative doses. Rheumatology 2000; 39:1383-1389

(14) Reid R. Glucocorticoid-induced osteoporosis. Baillieres Clin Endocrinol Metab 2000; 14:279-298

(15) Barnes PJ, Pedersen S, Busse WW. Efficacy and safety of inhaled corticosteroids: new developments. Am J Respir Crit Care Med 1998; 157:S1-S53

(16) Garbe E, Suissa S, LeLorier J. Association of inhaled corticosteroid use with cataract extraction in elderly patients. JAMA 1998; 280:539-543

(17) Walsh LJ, Wong CA, Oborne J, et al. Adverse effects of oral corticosteroids in relation to dose in patients with lung disease. Thorax 2001; 56:279-284

(18) Kelly HW, Nelson HS. Potential adverse effects of the inhaled corticosteroids. J Allergy Clin Immunol 2003; 112: 469-478

(19) Lipworth BJ. Systemic adverse effects of inhaled corticosteroid therapy: a systematic review and meta-analysis. Arch Intern Med 1999; 159:941-955

(20) Wong CA, Subakumar G, Casey PM. Effects of asthma and asthma therapies on bone mineral density. Curr Opin Pulm Med 2002; 8:39-44

(21) Israel E, Banerjee TR, Fitzmaurice GM, et al. Effects of inhaled glucocorticoids on bone density in premenopausal women. N Engl J Med 2001; 345:941-947

(22) Tattersfield AE, Town GI, Johnell O, et al. Bone mineral density in subjects with mild asthma randomized to treatment with inhaled corticosteroids or non-corticosteroid treatment for two years. Thorax 2001; 272-278

(23) Boulet L-P, Milot J, Gagnon L, et al. Long-term influence of inhaled corticosteroids on bone metabolism and density. Am J Respir Crit Care Med 1999; 159:838-844

(24) Medici TC, Grebski E, Hacki M, et al. Effect of one year treatment with inhaled fluticasone propionate or beclomethasone dipropionate on bone density and bone metabolism: a randomized parallel group study in adult asthmatic patients. Thorax 2000; 55:375-382

(25) Hughes JA, Conry BY, Male SM, et al. One year prospective open study of the effect of high dose inhaled steroids, fluticasone propionate, and budesonide on bone markers and bone mineral density. Thorax 1999; 54:223-229

(26) van Staa TP, Leufkens HGM, Cooper C. Use of inhaled corticosteroids and risk of fractures. J Bone Miner Res 2001; 16:581-588

(27) Lan E, Mamdani M, Tu K. Inhaled or systemic corticosteroids and the risk of hospitalization for hip fracture among elderly women. Am J Med 2003; 114:142-145

(28) Suissa S, Baltzan M, Kremer R, et al. Inhaled and nasal corticosteroid use and the risk of fracture. Am J Respir Crit Care Med 2004; 169:83-88

(29) Hubbard RB, Smith CJP, Smeeth L, et al. Inhaled corticosteroids and hip fracture: a population-based case-control study. Am J Respir Crit Care Med 2002; 166:1563-1566

(30) Mussolino ME, Looker AC, Madans JH, et al. Risk factors for hip fracture in white men: the NHANES I Epidemiologic Follow-up Study. J Bone Miner Res 1998; 13:918-924

(31) van Staa TP, Abenhaim L, Cooper C, et al. Public health impact of adverse bone effects of oral corticosteroids. Br J Clin Pharmacol 2001; 51:601-607

(32) Iqbal F, Michaelson J, Thaler L, et al. Declining bone mass in men with chronic pulmonary disease. Chest 1999; 116:16161624

(33) Sin DD, Man JP, Man SFP. The risk of osteoporosis in Caucasian men and women with obstructive airways disease. Am J Med 2003; 114:10-14

(34) Biskobing DM. COPD and osteoporosis. Chest 2002; 121: 609-620

(35) Ray WA, Griffin MR, Fought RL, et al. Identification of fractures from computerized Medicare files. J Clin Epidemiol 1992; 45:703-714

(36) Centers for Disease Control and Prevention. Incidence and costs to Medicare of fractures among Medicare beneficiaries aged [greater than on equal to] 65 years-United States, July 1991-June 1992. MMWR Morb Mortal Wkly Rep 1996; 45:877-883

(37) Quam L, Ellis LB, Venus P, et al. Using claims data for epidemiologic research: the concordance of claims-based criteria with the medical record and patient survey for identifying a hypertensive population. Med Care 1993; 31:498-507

(38) Lanza LL, Dreyer NA, Schulz NJ, et al. Use of insurance claims in epidemiologic research: identification of peptic ulcers, GI bleeding, pancreatitis, hepatitis and renal disease. Pharmacoepidemiol Drug Saf 1995; 239-248

(39) Rothman KJ, Greenland S. Modern epidemiology. New York, NY: Lippincott Williams & Wilkins, 1998; 94-96

(40) Espallargues M, Sampietro-Colom L, Estrada MD, et al. Identifying bone-mass-related risk factors for fracture to guide bone densitometry measurements: a systematic review of the literature. Osteoporos Int 2001; 12:811-822

(41) Feldkamp J, Becker A, Witte OW, et al. Long-term anticonvulsant therapy leads to low bone mineral density: evidence of direct drug effects of phenytoin and carbamazepine on human osteoblast-like cells. Exp Clin Endocrinol Diabetes 2000; 108:37-43

(42) Ray JG, Papaioannou A, Ioannidis G, et al. Anticonvulsant drug use and low bone mass in adults with neurodevelopmental disorders. QJM 2002; 95:219-223

Catherine B. Johannes, PhD; Gary A. Schneider, MSPH; Timothy J. Dube, BA; Tanya D. Alfredson; Kourtney J. Davis, PhD; and Alexander M. Walker, MD, DrPH

* From Ingenix Epidemiology (Drs. Johannes and Walker Mr. Schneider, Mr. Dube, and Ms. Alfredson), Auburndale, MA; and GlaxoSmithKline (Dr. Davis), Research Triangle Park, NC. Dr. Davis is an employee of GlaxoSmithKline. The work was performed at Ingenix Epidemiology, supported by a research contract with GlaxoSmithKline. Manuscript received February 23, 2004; revision accepted July 21, 2004.

Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (e-mail: permissions@chestnet.org).

Correspondence to: Catherine B. Johannes, PhD, Ingenix Epidemiology, One Riverside Center, Suite 3-120, 275 Grove St, Auburndale, MA 02466; e-mail: kjohannes@epidemiology.com

COPYRIGHT 2005 American College of Chest Physicians
COPYRIGHT 2005 Gale Group