Introduction

A number of cohort studies have detailed the distribution of osteoporosis-related fractures and their accompanying risk factors in different regions of the world [1–18]. As these studies varied in their objectives and methods of data collection, the comparability of the information obtained may be limited. Furthermore, existing reports do not compare regional differences in patterns of patient management and outcomes of fracture.

The Global Longitudinal study of Osteoporosis in Women (GLOW) is an observational longitudinal study designed to improve understanding of international patterns of susceptibility, recognition, management, and outcomes of care in women aged 55 years and older at risk for fragility fractures. The aim of the GLOW study is to collect uniform data to: (1) describe the distribution of risk factors for osteoporosis-related fracture; (2) apply published fracture risk assessment tools in a population of older women; (3) identify differences in physician patterns of diagnosis and management of osteoporosis (e.g., how health care providers are identifying individuals for treatment; characteristics of women being treated); (4) characterize factors that influence patient persistence with treatment, including patient characteristics, awareness of fracture risk and comorbid conditions; (5) assess the real-world effectiveness of care on the incidence of fracture; and (6) evaluate the cost effectiveness of interventions for the prevention and management of osteoporosis from the perspective of the health care provider.

Study design

Study site selection

GLOW is being conducted in physician practices in 17 study sites in ten countries (Australia, Belgium, Canada, France, Germany, Italy, Netherlands, Spain, UK, and USA) in Australia, Europe, and North America. These sites are located in major population centers (Table 1). A Scientific Advisory Board comprising investigators at each of the 17 sites was constituted to provide scientific oversight and study management. These individuals are independent university-based investigators with content expertise in osteoporosis, who represent the disciplines of endocrinology, rheumatology, geriatric medicine, and epidemiology. These sites were selected based on the ability of the local investigators to consistently administer the survey methodology, on the availability of a wide spectrum of osteoporosis treatment options and bone densitometry, and the existence of prior studies in those regions, which would provide data for comparison with the GLOW sample. Practical considerations concerning the number of survey translations and number of countries in which the survey process could be supervised restricted the number of sites to those chosen for this study.

Table 1 Study sites and number of patients enrolled per site
Full size table

Physician sample selection

Practices typical of each region were recruited through primary care networks organized for administrative, research, or educational purposes or by identifying all physicians in a geographic area. Physician networks included regional health-system-owned or managed practices, health maintenance organizations, independent practice associations, and other primary care practice networks. Networks established for the purpose of general medical research were used only if they were not established exclusively for osteoporosis research and did not consist of physicians whose primary focus was academic.

Primary care physicians were defined as physicians who spent the majority of their time providing primary health care to patients. Depending on the country in which the study site was located, this included internists, family practitioners, and general practitioners who provide primary care. If the physician network or study area included more eligible physicians than were required to recruit a sufficient number of patients, a random sample of those physicians within the network or study was invited.

Each physician completed a standardized form that collected data on their demographics and practice characteristics (Table 2).

Table 2 Physician data
Full size table

Patient selection

Each physician practice provided a list of the names and addresses of women aged 55 years and older who had consulted their physician in the past 24 months. These lists comprised the sampling frame. Sampling was stratified by age to ensure that two thirds of the women surveyed were 65 years of age and older. In each practice, we recruited from all eligible women 65 and over and a random sample of half that number under age 65 years. Sample size estimates were generated to detect a 30% difference in 5-year fracture incidence between treated and untreated patients with a power of 80%. On this basis, a sample of approximately 3,000 patients was sought at each site. Patients were excluded if they were unable to complete the study survey due to cognitive impairment, language barriers, or institutionalization or were too ill.

Instrument development

Questionnaires were designed to be self-administered and cover the domains outlined in Table 3. Where possible, items from published validated instruments were used, including the National Health and Nutrition Examination Survey (NHANES) [19], EuroQol (EQ-5D) [20], and SF-36 [21] (physical function component). Questions that had not been used previously were tested cognitively in the context of the complete questionnaire in a sample of women in the study age group. The complete baseline questionnaire was also pilot-tested before being finalized to gauge subject comprehension and completion time. Questionnaires were translated into five languages (French, Spanish, German, Italian, and Dutch) in addition to English by the University of Massachusetts-Amherst Translation Center. Where items from existing questionnaires had been translated previously, these items were incorporated directly. Translations were reviewed by study coordinators at each site for accuracy and consistency with local idiom. Because NHANES is administered to a representative sample of US residents, it was possible to compare responses to items that were similar in the GLOW survey to assess the similarity of the populations. Data from NHANES conducted in 2005 and 2006 were used for this purpose.

Table 3 Baseline questionnaire items
Full size table

Survey administration

Each study site obtained ethics committee approval to conduct the study in the specific location. Baseline questionnaires along with invitations to participate in the study signed by the local principal investigator were mailed to all potential patients. Nonrespondents were followed up with a series of postcard reminders, second questionnaires, and telephone interviews, as outlined in Fig. 1. Women who responded will be resurveyed annually for the next 4 years. In addition to repeating questions about medications, quality of life, and functional status, the follow-up surveys will include questions about persistence with medication, reasons for nonadherence, and detail about fracture-associated treatment.

Fig. 1
figure 1

Recruitment/enrollment flow chart. Asterisk, age-stratified sampling not feasible in Sydney, Paris, or Lyon

Full size image

Patient identity is safeguarded by the local study coordinator, who assigns an ID number to each participant at enrollment and maintains the site’s participant list locally. The names of patients are stored separately from study data transmitted to the central coordinating center (Center for Outcomes Research at the University of Massachusetts Medical School). Thus, unique patient identifiers are confidential to the investigators at each study site.

The process for entering, verifying, and managing survey data is uniform across all study sites. Completed questionnaires are sent to the central coordinating center, where they are scanned electronically, and data fields are audited visually by a person trained to process the forms. The data entry software is designed to detect out-of-range values, inconsistencies, and omissions and to document any resolutions. Scanned data are entered into a database stored on a secured password-protected computer. As a quality control measure, each study site maintains an administrative database that tracks surveys mailed and received, and scanned surveys are checked against these databases. Twice yearly meetings are held with study coordinators from each of the study sites to review survey administration and ensure uniformity of the process. For study sites using telephone follow-up in addition to mail, a standard telephone script is used and reviewed with each site to ensure consistency of telephone survey administration.

Results

A total of 723 physicians agreed to participate in the GLOW study and supplied practice lists. The number of physicians ranged from 14 to 72 per site (median 40). In the US, 298 participating physicians comprised 103 family physicians and 195 internal medicine physicians. All Canadian, Australian, and European participants were general practitioners.

Baseline surveys were mailed between October 2006 and February 2008 to 140,416 potential subjects (Fig. 1). After the exclusion of 3,265 patients who were either ineligible or had died, 60,393 women agreed to participate. The median response rate among the 17 study sites was 62% (range 15–75); 76% of the study sites had a response rate of 50% or greater. Two sites experienced notably lower response rates than were typical. In both Italy and Spain, the lower response to mailed surveys had been anticipated based on prior experience with mail surveys in those regions; accordingly, higher numbers of surveys were sent to potential participants in order to collect the targeted 3,000 responses.

The characteristics of the 60,393 women who participated in GLOW are displayed in Table 4. The mean age was 69 years and mean weight 148 lb (67.2 kg). Among characteristics known to place women at increased risk of fragility fracture, weight <125 lb (57 kg) was present in 16%, history of maternal hip fracture in 13%, and personal history of a fracture of the wrist, spine, or hip in 12%. Twenty-two percent had been told by a doctor or health professional that they had osteoporosis; 11% reported asthma, and 11% rheumatoid arthritis; 23% of women said their health status was “fair” or “poor.”

Table 4 Characteristics of women participating in GLOW, US women participating in GLOW, and NHANES women aged 55 years and older for 2005 to 2006
Full size table

Comparisons of demographic characteristics and risk factors for the US GLOW subjects and for women aged 55 and older sampled in the NHANES study (2005 to 2006) are also displayed in Table 4. Although the mean ages for the two groups were similar, women in the GLOW sample had received a higher level of education, were more often white, and had better self-reported health than women in the NHANES study. History of wrist fracture was also somewhat lower in the GLOW population than in the NHANES population. However, many of the risk factors were similar among the two samples, for example low weight, osteoporosis diagnosis, fracture of the spine or hip, and maternal fracture. The prevalence of common comorbid conditions, such as hypertension, high cholesterol, and asthma, was also similar.

When women were asked how concerned they were about osteoporosis, 54% expressed “some” concern and 25% said they were “very concerned” about the condition (Table 5). Overall, 43% said they had spoken with their doctor about osteoporosis testing, treatment, or prevention in the past 12 months. Twenty-one percent of women reported having been told by their doctor or health provider that they had osteoporosis; 19% said they were told they had osteopenia. When asked to rate their own risk of getting osteoporosis compared with women their own age, 33% rated their risk as lower and 19% as higher.

Table 5 Subjects’ awareness of osteoporosis
Full size table

Discussion

GLOW is designed to provide an international perspective on fracture risk in women, patient management practices, patient awareness, physical and emotional function following fracture, application of risk assessment models, and functional outcomes following fracture. Previous cohort studies of osteoporosis were designed primarily to identify factors associated with fracture incidence and document the distribution of low bone mineral density and its association with fracture risk. These efforts have been limited to specific regions or areas. GLOW will provide the first description of patterns of risk from an international perspective. Further, the data from GLOW will be used to assess not only fracture risk and incidence, but will identify patient concern and awareness and clinical management at a time when significant efforts have been made to implement treatment guidelines and educate patients about osteoporosis and fracture risk. In these baseline results, a minority of GLOW subjects (43%), among women 55 years and older, indicated having discussed osteoporosis with their physician in the past year, yet 79% of women in the study were somewhat or very concerned about osteoporosis. Future analyses of GLOW data will examine the link between perceived risk, concern, and physician encounters on treatment risk of fracture and quality of life.

Prior studies have reported undertreatment and underdiagnosis of osteoporosis [22]. However, since these studies were conducted, many new therapies have become more widely used than in the past. GLOW will report on contemporary treatment prevalence according to fracture risk and self-reported diagnosis of osteoporosis at a time when a wider range of patient management options have been generally accepted and are available

Previously collected risk factor data form the basis for risk-scoring algorithms designed to predict fracture risk and aid physicians in targeting treatment to those most in need [23–26]. GLOW will update data on these factors and allow the calculation of patterns of international fracture risk. Because the data will be gathered in a consistent manner, confidence will be increased that the variations seen in the distribution of risk factors between regions is due to differences in the populations rather than in the methods of data collection.

The study will also assess a spectrum of physician–patient interactions including discussion of osteoporosis, advice concerning falls, bone mineral density screening, diagnosis of osteoporosis, and pharmacological treatments. Regional and international comparisons of diagnosis and treatment patterns will be possible, with adjustment for region- and country-specific characteristics, such as the availability of health insurance, reimbursement for prescriptions, and treatment protocols. A number of items that assess subjects’ physical and emotional status have also been incorporated in the questionnaire. These include the mobility and vitality scales from SF-36 and the five subscales of the EQ-5D. Such measures will enable comparisons of functional outcomes for women who suffer various types of incident fractures in differing geographic regions over time.

Whereas most studies of patient persistence focus on a single drug, GLOW will include the full range of currently available pharmacological treatments for osteoporosis (alendronate, calcitonin, estrogen, etidronate, ibandronate, pamidronate, parathyroid hormone [1–84], raloxifene, risedronate, strontium ranelate, teriparatide, tibolone, and zoledronate). We will also be able to include any newly available osteoporosis medications in the questionnaire. The study will also examine the reasons why patients stop and switch medications.

GLOW data will allow assessment of the effectiveness of treatment on the incidence of fracture in a “real-world” setting. In contrast to randomized clinical trials, GLOW did not exclude women who had previously been diagnosed with osteoporosis or treated with bone drugs. Consequently, analysis of the treated population will include those women who stop or switch medications, as well as those who have a high degree of persistence. Adjustment will be possible for potential confounding of the relationship between treatment and fracture using fracture risk factors and risk scores. While the study is not designed to evaluate the effectiveness of any single bone drug, it will allow comparison of fracture rates among treated and untreated patients across all classes of interventions. Such head-to-head comparisons have not been evaluated in randomized controlled trials.

Analysis will also be carried out to estimate the relative cost effectiveness of various classes of interventions used in the management of fractures, using the usual principles set out for cost-effectiveness analysis [27–29]. An economic model based on the epidemiological evidence of treatment outcomes recorded in GLOW will be constructed [30].

GLOW is a practice-based rather than a population-based study and is subject therefore to biases in both the selection of physicians and the sampling and recruitment of patients. Practical considerations limited our sample selection to women from 17 study locations in ten countries. An expanded sample that included a broader representation of racial and ethnic groups was not pursued because of the complexity of administering a recruitment methodology involving oversampling of particular racial and ethnic groups in ten different countries, where the definitions of these groups were likely to vary considerably. Physicians at each site who agreed to participate may not be representative of all physicians in an area with respect to osteoporosis recognition and management. We attempted to avoid altering physician practice by minimizing doctors’ awareness of the study. There were no clinical interventions and physicians had no involvement in patient recruitment other than supplying practice lists.

Unlike studies that excluded women because of prior fracture, diagnosis of osteoporosis, or current treatment for osteoporosis, GLOW attempted to enlist all women 55 years and older who were active patients in each physician’s practice. By doing so, the study will provide a more complete picture of care received by women in this age group. Nonetheless, some participation biases are likely. It is possible that participants will have greater interest in bone health issues and seek information, screening, and treatment more actively. We attempted to reduce selection bias by creating a survey process that imposed low respondent burden. Participation required no clinic visits (by not requiring patients to schedule a clinic visit or face-to-face interview, we avoid requirements that might make participation difficult for women who are in poor health or have no or limited access to transportation) and questionnaires were mailed directly to the subject’s home and typically required only 15–20 min to complete. High response rates at most sites (median 62%) suggest that this strategy was successful. Comparison of characteristics for the sample of US women with those of the nationally representative sample of comparably aged NHANES women demonstrated that although GLOW women were better educated, more likely to be white, and reported better health, the prevalence of risk factors for fracture was similar.

All data are collected by patient self-report. While this approach is subject to limitations of recall and recall bias, it has the advantages of efficiency and methodological consistency. The combination of mail and telephone surveys is amenable to collection of data on quality of life, health status, and fracture risk factors of interest. The efficiency of the mail and phone survey approach also makes it feasible to obtain a substantial sample size and to provide adequate statistical power for the analysis of fracture outcomes, which are relatively rare events. The survey format also allows standardized administration that reduces the issues of noncomparability and variation in data quality that would arise if medical records and public health care databases from several different countries were used. Reports that have examined the validity of self-report of prescription medication use and fractures have shown reasonable accuracy [31–36]. Self-report may be preferable to the abstraction from medical records of data on diagnosis and treatment, given inconsistencies in record keeping between physicians and between study regions and countries. Additionally, records from primary care physicians may not include evidence of treatment initiated by a specialist physician. Validation of self-reports of variables such as fractures and bone mineral density examinations may be possible for subsets of subjects in sites where electronic medical records are available.

Conclusions

GLOW will provide important information on the patterns of management of fracture risk in older women over a 5-year period. The collection of data in a similar fashion in ten countries will allow comparisons of patient experience with prevention and treatment, and an understanding of differences in the distribution of risk among older women on an international basis.