Data were obtained from the Canadian Database of Osteoporosis and Osteopenia (CANDOO) patients. CANDOO consists of 8065 patients and involves 8 sites across Canada (Calgary, n = 20; Saskatoon, n = 357; Winnipeg; n = 4; Hamilton (2 sites), n = 6033; Montreal (2 sites), n = 1527; Quebec City, n = 124). This database is a prospective registry of patients designed to capture a comprehensive set of osteoporosis-related clinical data during the course of routine specialist care . In particular, the CANDOO contains electronically stored data regarding basic patient demographics, fracture history, gynecological history, past use of osteoporosis-related drug treatment, drug side effects, past use of corticosteroids and other medications, dietary calcium intake, smoking habits, type and quantity of physical activities, fall history, past medical history and family history including fractures, a self administered osteoporosis HRQL instrument, basic laboratory results, and bone density measurements. One database record, with over 400 data fields per patient, is generated for each patient at each clinical visit.
For the current analysis, the database was searched for postmenopausal women 50 years of age and older who were seen at one Hamilton site and who were registered in CANDOO from 1994 to 1999. During this time period, this location was the only CANDOO site that regularly administered the osteoporosis HRQL instrument. At the Hamilton site, all patients seen in the course of routine specialist care were entered in CANDOO. On average patients were seen annually and the HRQL instrument was administered at each clinical visit. Participants were grouped according to incident fracture status. Group I: patients who experienced clinically recognized incident vertebral fractures. Group II: patients who sustained incident non-vertebral fractures. Incident fractures were registered based on the patient's response to an item ("How many new fractures have you had since your last visit?") from the CANDOO questionnaire. Incident vertebral fractures may or may not have been confirmed by x-ray. Incident non-vertebral fractures included the ankle, arm, clavicle, elbow, heel, hip, knee, leg, nose, pelvis, rib, shoulder, sacrum, and wrist. Group III: patients who did not report an incident fracture.
Patients with incident fractures were eligible for study entry if they completed at least a portion of the HRQL questionnaire prior to (pre assessment) and following their fracture (post assessment). Patients who did not report incident fractures also had to complete the HRQL questionnaire twice. The duration between pre and post HRQL measurements was matched between the three groups.
The mini-Osteoporosis Quality of Life Questionnaire (mini-OQLQ) was used to measure HRQL. The instrument is self-administered but is reviewed by the specialist or a nurse clinician during each visit. The mini-OQLQ was derived from the original 30 item Osteoporosis Quality of Life Questionnaire (OQLQ)  to enhance the instrument's usefulness in clinical practice. The mini-OQLQ includes the two items from the OQLQ with the highest impact in each of the five domains (symptoms, physical functioning, emotional functioning, activities of daily living, and leisure), for a total of 10 items (Additional File). Each item is associated with a seven-point scale. The total score for the instrument can vary from 10 to 70, while the domain scores can vary from 2 to 14. To standardized all scores, the total and the domain scores were divided by the number of items that were used to generate the values. A standardized rating of 1 represents the worst possible function (extreme difficulties, constant fear, extreme distress) and a rating of 7 represents the best possible function (no difficulties, no fear, no distress). It has been suggested, from previous work, that changes within each domain score of approximately 0.5 represents a clinically important difference in quality of life [19, 20]. The mini-OQLQ instrument has been evaluated in osteoporotic women with back pain due to vertebral fractures and has been found to be reliable and valid both for measuring differences in HRQL between patients with osteoporosis and changes within these patients over time .
Our primary objective was to determine the effect of vertebral and non-vertebral fractures on HRQL. Multiple regression analyses were performed to explain the relationship between incident fractures and HRQL. From these analyses, we determined regression coefficient estimates (which represent differences between pre and post HRQL measurements for both fractured and non-fractured patients), as well as 95% confidence intervals (CI) of the difference scores for the mini-OQLQ. The independent variables entered in the models included group; baseline mini-OQLQ score; group by baseline mini-OQLQ score interaction; number of years between pre and post administration of the mini-OQLQ; age; baseline height; baseline weight; prevalent vertebral and non-vertebral fracture status; number of prevalent vertebral and non-vertebral fractures; time spent exercising; dietary calcium intake; baseline bone mineral density of the lumbar spine and femoral neck; medications such as bisphosphonates (etidronate or alendronate), estrogen, or others (calcitonin, fluoride or raloxifene); and co-morbidities (including heart attack; stroke; breast, ovarian, cervical, uterine and colon cancer, and multiple myeloma; diabetes, epilepsy, arthritis, and thyroid disease). Regression models were selected based on Mallows C(P) statistic. Statistical significance was defined as p<0.05. All analyses were preformed on a Dell personal computer using a SAS/STAT (version 7.0; SAS Institute Inc., Cary, NC, USA) software package.