PFIZER INC. These results are supplied for informational purposes only. Prescribing decisions should be made based on the approved package insert. For publications based on this study, see associated bibliography. PROPRIETARY DRUG NAME /GENERIC DRUG NAME: Depo- Medroxyprogesterone Acetate (DMPA) PROTOCOL NO.: Not applicable. TITLE: The Incidence of Bone Fractures in Depo-Medroxyprogesterone Acetate (DMPA) Users and Non-users in Women in the General Practice Research Database (GPRD) Study Center(s): Not applicable. This web synopsis is based on the report The Incidence of Bone Fractures in Depo-Medroxyprogesterone Acetate (DMPA) Users and Non-users in Women in the General Practice Research Database (GPRD) dated 14 Dec 2009 Study Initiation Date and Primary Completion or Completion Dates: Study start, 01 January 1987. End of the study period, 31 December 2005. Phase of Development: Not applicable Study Objective(s): The objective of this study was to increase understanding of the possible effects of depo-medroxyprogesterone acetate (DMPA) on bone health in women. Of particular interest were the following questions: METHODS What is the difference in incidence of fracture in women who ever or never used DMPA for contraception? How does incidence of fracture vary by cumulative DMPA exposure? Is there a delayed effect of DMPA on fracture rates at two or more years after the end of exposure? Study Design: The study followed a cohort of women who were in the General Practice Research Database (GPRD). Their fracture experience was assessed until the earliest of 31 December 2005, the date of fracture, death, or departure from the general practice. Using Page 1
prescription records, current, recent, and past periods of exposure following an injection of DMPA were defined for each woman. Users of DMPA were also classified by cumulative exposure as low (1 to 7 injections) or high ( 8 injections). Number of Subjects (Planned and Analyzed): The analyzed study cohort comprised 312,395 women and 1,722,355 person-years of follow-up. The analyzed sub-cohort (women with fracture data available for at least 6 months prior to the Index date) comprised 166,367 women. This subcohort was used for comparisons of fracture rates before and after DMPA treatment was started and for assessment of potential confounders for which information was available in the GPRD. Diagnosis and Main Criteria for Inclusion: Female subjects who met the following criteria were extracted from the GPRD: Known year of birth; Woman classified as an acceptable patient by GPRD standard criteria relating to practice registration details and other measures of data quality; and Having at least one contraceptive prescription record dated between 1 January 1987 and 31 December 2005 and occurring before she reached 50 years of age. The contraceptive record also must have been dated after the practice up-to-standard date, when the data recorded by the practice met standard GPRD criteria for quality and completeness. The contraceptive prescription also was required to be after the patient s registration in the practice but before the practice s last data collection date and the patient s transferred-out or death date (if applicable). The patient registration date is the date when the woman was first recorded by the National Health Service (NHS) as a patient of a particular general practitioner, GP. The following contraceptives were acceptable: Depo-Provera (injection 150 mg/ml) or for medroxyprogesterone acetate contraceptive injection (150 mg/ml), Nonhormonal prescription contraceptives [e.g., intrauterine device (IUD), cervical cap, or diaphragm], or Hormonal prescription contraceptives that are not DMPA (e.g., other estrogen or progesterone contraceptives or combination estrogen and progesterone contraceptives). This group includes IUDs that release estrogen or progesterone. The study cohort was divided into women who did or did not have a minimum of 6 months of history before the index contraceptive date in the same GPRD practice in which the index contraceptive had been prescribed. Study Treatment: Not applicable. Efficacy Evaluations: No efficacy evaluations were performed for this study. Page 2
Safety Evaluations: Rules were created to define the extent of each woman s longitudinal GPRD experience that would be used in the study. These rules applied to all the women in the large cohort and subcohort, regardless of the type of contraceptives used or amount of baseline history. The index date for observation was the date of a woman s first prescription record with a contraceptive code that occurred on or after the latest of the following events: The study start date (01 January 1987), Her registration date, The up-to-standard date for the practice in which she was registered. The termination date was the first of the following dates: The practice s last date of contributing data to GPRD; The end of the study period (31 December 2005); The woman s first fracture date after entering the study; or The date when the woman terminated from the practice due to relocation, transfer to another practice, or death. Statistical Methods: Incidence of fracture was reported by 3 types of exposure: Ever use of DMPA (any exposure, yes or no): Cumulative DMPA exposure (low vs high); and Time since exposure (current, recent, or past). For cumulative DMPA exposure, low was defined as the equivalent of 1 to 7 injections and high as the equivalent of 8 injections. For time since exposure, the definitions were: Current: Current time was defined as equivalent to active DMPA exposure time, which occurred during the DMPA episodes described above (ie, the active contraceptive time). Current exposure continued as long as DMPA prescriptions continued to occur in the data (subject to the rules for adjusting for overlaps and duplicates). Recent: This was time up to 640 days after each active DMPA episode, and before a subsequent DMPA episode or the end of follow-up. Recent days were defined as ending at 730 days after the date of the last injection (inclusive). Recent time could have been interrupted if a woman received a subsequent DMPA injection or reached the end of follow-up. Past: Past time was defined as occurring after periods of recent exposure (ie, more than 730 days after the date of the last injection) and before any subsequent DMPA episode or the end of follow-up. Presumably, the effect of DMPA on bone mineral density (BMD) Page 3
was virtually absent during past exposure time. Time in the past category accumulated until the start of a subsequent DMPA episode or the end of follow-up. For incidence rates, observation was terminated at the first bone fracture that occurred after a woman entered the study cohort. Fractures were classified according to the DMPA exposure of the woman on the date of the fracture. If fractures at more than one bone site occurred on the same day, only one fracture event was counted for the incidence rates. Fracture events were identified using the Read and Oxford Medical Information System (OXMIS) dictionary terms used by GPs to record fracture occurrence. Definitions of fractures were the same as those developed in the earlier feasibility study. The following individual types of fractures and categories of fractures were analyzed: Ankle Arm Clavicle Finger/toe Foot Hand Hip Leg Multiple/extensive Pelvis Ribs/sternum Shoulder Skull/face Vertebra Wrist Unspecified (no fracture site was provided in the GPRD) Unspecified osteoporotic (no fracture site; coded osteoporotic ) Long bones of the arm; Page 4
All fractures (any reported fracture) All fractures except fractures of the hip, fractures of the long bones of the arm and fractures of the vertebrae. All fractures except finger/toe, skull/face and unspecified Axial skeleton fractures (vertebra + pelvis) Appendicular skeleton fractures (arm + leg + wrist + ankle + hand + foot + rib/sternum + clavicle + shoulder + hip) The primary outcome was fracture at any bone site. For all analyses of exposure, the reference category was nonexposure to DMPA (non-users). RESULTS Subject Disposition and Demography: Subject disposition is not applicable since this was a retrospective database study. Subject ages at study index date are summarized for DMPA users and non-users in Figure 1. Figure 1. Age at Study Index Date in Users and Nonusers of DMPA Abbreviation: DMPA = Depo-Medroxyprogesterone Acetate Note: Users had one or more DMPA injections at any time during the study period (with or without, before or after any other contraceptives). Nonusers did not have any DMPA prescriptions during the study period. Page 5
Subject contraceptive use is summarized in Table 1. Table 1. Number of Women by Contraceptive Combination Used Contraception During Study Period Number of Women Percent of Cohort Nonhormonal only 15,092 4.8 Other hormonal only 196,770 63.0 Nonhormonal and other hormonal 21,468 6.9 Total with no DMPA exposure 233,330 74.7 DMPA only 23,084 7.4 DMPA and other hormonal 51,479 16.5 DMPA and nonhormonal 663 0.2 DMPA, nonhormonal and other hormonal 3,839 1.2 Total with any DMPA exposure 79,065 25.3 Total women in cohort 312,395 100.0 Abbreviation: DMPA = Depo-Medroxyprogesterone Acetate Efficacy Results: No efficacy evaluations (assessment of contraceptive efficacy) were performed in this study. The primary objective of the study was patient safety as reflected by bone fracture risk. Safety Results: Table 2 summarizes the duration of study observation for the cohort. The average follow-up from index date was 5.9 years for users and 5.4 years for nonusers of DMPA. Table 2. Duration of Observation in Years Number and Percentage of Person-Years by Duration of Follow-Up Number of Women % Total Person- Years <1 45,002 14.4 23,733 0.5 1 to <2 41,250 13.2 61,239 1.5 2 to<3 33,149 10.6 82,534 2.5 3 to <5 54,329 17.4 216,187 4.0 5 to <10 82,208 26.3 586,936 7.1 10 to <15 42,204 13.5 527,075 12.5 15 14,253 4.6 224,652 15.8 Total 312,395 100.0 1,722,356 5.5 Average Person- Years per Woman Table 3 shows the potential confounders which were assessed using data available in GPRD. Each of these confounders was associated with a higher risk for fracture in this population, and was more prevalent in the DMPA user cohort; estrogen HRT has been shown to decrease fracture risk in postmenopausal women and was more prevalent in the non-user cohort. Page 6
Table 3. Odds Ratio for High Risk Patients to be Present in the DMPA User Cohort vs. the Non-User Cohort Risk Factor DMPA (N=41,876) Control (N=124,491) With W/O With W/O Odds Ratio 95% CI Alcohol abuse/dependence 55 41821 120 124371 1.36 0.99-1.87 Drug abuse 71 41805 74 124417 2.85 2.06-3.95 Inflammatory bowel disease 51 41825 75 124416 2.02 1.42-2.88 Epilepsy 101 41775 113 124378 2.66 2.03-3.47 Asthma 870 41006 1744 122747 1.48 1.37-1.61 Oral corticosteroids 432 41444 955 123536 1.34 1.20-1.51 Fall 179 41697 320 124171 1.66 1.39-2.00 Estrogen HRT 75 41801 689 123802 0.32 0.26-0.41 Smoking (current) 7977 33899 14526 109965 1.63 1.59-1.67 Pregnancy < age 20 844 41032 662 123829 3.79 3.43-4.19 Abbreviations: DMPA = Depo-Medroxyprogesterone Acetate; N = number; CI = confidence interval; HRT = hormone replacement therapy; W/O = without. Table 4 shows Incidence Rate Ratios (IIR) for DMPA use in the subcohort with available pre-treatment data that would allow assessment of confounders, standardized for age and other selected factors. These data show that after standardizing for age and each potential confounder, the IRRs for any DMPA were close to the value of 1.40 that resulted from standardizing by age alone. While these risk factors were associated with increased fracture rate and were more prevalent in the DMPA users cohort, the absolute numbers of women with each risk factor (or any risk factor) were relatively small compared to the overall number of women in the study and, as shown in Table 4, did not materially affect the study results. It was therefore concluded that there was little evidence of confounding from the baseline variables for which data were available in GPRD. Unassessed confounders, for which information was not available in GPRD, may also have been present and could have affected the study results. Page 7
Table 4. Incidence Rate Ratios for DMPA Use, Standardized for Age and Other Selected Factors (Subcohort) Any DMPA a Potential Confounding Factors IRR 95% CI Crude IRR 1.37 1.29-1.45 Standardized for age only Age (5-year groups) 1.40 1.32-1.49 Standardized for age and each baseline factor b Oral corticosteroids 1.40 1.32-1.49 Pregnancy < age 20 1.40 1.32-1.49 Smoking, current 1.39 1.31-1.47 Alcohol abuse 1.41 1.32-1.49 Asthma 1.39 1.31-1.48 Drug abuse 1.40 1.32-1.48 Epilepsy 1.40 1.32-1.48 Estrogen HRT 1.40 1.32-1.49 Fall 1.40 1.32-1.49 Past fracture, any site 1.41 1.32-1.50 Inflammatory bowel disease 1.40 1.32-1.48 Abbreviations: DMPA = depo-medroxyprogesterone acetate; HRT = hormone replacement therapy; IRR = incidence rate ratio; CI = Confidence Interval. a. Ever used DMPA in the study period, compared with never used DMPA. b. IRRs are standardized for age in 5-year groups and for each factor (one at a time). There was a total of 11,822 incident fractures in the study population, for an overall incidence rate of 6.9 per 1,000 person-years. After the start of treatment, the crude incidence of any fracture with any use of DMPA was 9.0 per 1,000 person-years, which was higher than that in nonusers (6.4 per 1,000 person-years). The crude incident rate ratio (IRR) for any fracture with any use of DMPA was 1.41 (95% Confidence Interval or CI: 1.35-1.47). Table 5 presents the relative risks for fracture that were determined for the subcohort by comparing DMPA users and non-users, based on these subjects having data available before and after DMPA use was started. The results show that DMPA users had a higher fracture risk than non-users before DMPA treatment was started. The higher risk for fracture in the DMPA cohort was similarly present after DMPA use started. Within the DMPA cohort, comparing the period after DMPA treatment started to the period before DMPA treatment started, the risk ratio was 1.08 (95% CI: 0.92, 1.26) indicating that there was no significant increase in fracture risk associated with starting DMPA treatment. The non-user cohort also showed no change in fracture rate associated with starting contraception (other than DMPA). Page 8
Table 5. Incidence of Fractures Before and After DMPA Treatment (Pre- and Post- Index Date) in the Full Study Cohort and/or in the Sub-Cohort of Patients Having 6 Months of Pre-Index Date Data) DMPA vs Non- User: PRIOR to Initiation of DMPA or Other Index Contraceptive (subcohort)* DMPA vs Non- User: AFTER INITIATION of DMPA or Other Index Contraceptive (subcohort)* DMPA vs Non- User: AFTER INITIATION of DMPA or Other Index Contraceptive (full cohort) DMPA Users: PRIOR to vs AFTER INITIATION of DMPA Treatment (subcohort)* Non-Users: PRIOR to vs AFTER INITIATION of Index Contraceptive Method (not DMPA) N DMPA User Number With Any Fx/ Fracture 1000PY Never User (Control) Number With Any Fx/ N Fracture 1000PY 41,876 176 8.41 124,491 409 6.57 41,876 1574 9.06 124,491 4939 6.64 79,065 2935 8.97 233,330 8887 6.37 Prior to Initiation After Initiation 41,876 176 8.41 41,876 1574 9.06 124,49 1 409 6.57 124,491 4939 6.64 IRR (95% CI) p-value 1.28 (1.07-1.52) 0.007 1.36 (1.29-1.44) <0.001 1.41 (1.35-1.47) <0.001 RR (95% CI) p-value 1.08 (0.92-1.26) 0.349 1.01 (0.91-1.12) 0.849 (subcohort)* Abbreviations: DMPA = depo-medroxyprogesterone acetate; Fx = fracture; PY = person years; HRT = hormone replacement therapy; IRR = incidence rate ratio; N = number; RR = relative risk; CI = Confidence Interval. * The 'subcohort' is the 53% of the study population who had fracture data available for the 6-month period immediately prior to the Index Date (baseline) and, therefore, could be included in comparisons of fracture r 'before' vs 'after' the first recorded contraceptive prescription (Index Date). All patients (full cohort) had dat available after the Index Date and could be included in the post-index Date comparison of DMPA Users vs N Users. Page 9
Figure 2 shows the fracture rate in the DMPA cohort as a function of time after starting DMPA. The fracture rate for the non-user group is represented as a flat line since these subjects did not start DMPA. The fracture rate in the DMPA cohort was elevated from time=0, with no induction period evident. If the greater fracture rate in the DMPA cohort was due to the BMD decline that occurs with DMPA use, and recognizing that the BMD decline requires at least 2 years to become fully present, then an induction period would be expected before the fracture rates in the two cohorts (DMPA users and non-users) diverged. Figure 2. Crude incidence of fracture during current-recent exposure to DMPA from time of first DMPA injection (all fractures) Abbreviations: DMPA = depo-medroxyprogesterone acetate; pyrs = person years Table 6 shows the fracture risk ratios for various types of fractures. If the greater rate of fracture in the DMPA cohort was due to low BMD, then those fracture sites that are known Page 10
to be most sensitive to low BMD and not the result of trauma (e.g., axial skeletal fractures) would be expected to show the greatest difference between the cohorts in terms of fracture risk. In contrast, fractures of the appendicular skeleton are more likely to involve trauma, to variable degrees. Fractures of the face, skull, fingers and toes are generally recognized as not being osteoporotic fractures and are usually due to trauma. Table 6. Incident Rate Ratios (DMPA Users vs. Non-Users) in the Full Study Population During Follow-Up by Fracture Type DMPA No DMPA Fx/ Fx/ Fracture Groups N N 1000PY 1000PY IRR 95% CI All fractures 2935 8.97 8887 6.37 1.41 1.35-1.47 All fractures (except finger/toe, skull/face, unspecified) 1700 5.19 5345 3.83 1.36 1.28-1.43 Axial skeleton 65 0.20 289 0.21 0.96 0.73-1.25 Appendicular skeleton 1632 4.99 5050 3.62 1.38 1.30-1.46 Osteoporosis-associated fractures Vertebra 35 0.11 141 0.10 1.06 0.73-1.53 Hip 8 0.02 38 0.03 0.90 0.42-1.92 Wrist 331 1.01 899 0.64 1.57 1.38-1.78 Pelvis 30 0.09 148 0.11 0.86 0.58-1.28 Unspecified osteoporotic 1 0.003 1 0.001 4.28 ND Other fractures by site Ankle 248 0.76 729 0.52 1.45 1.26-1.67 Arm 374 1.14 1205 0.86 1.32 1.18-1.49 Clavicle 54 0.16 178 0.13 1.29 0.95-1.75 Finger/toe 528 1.61 1676 1.20 1.34 1.22-1.48 Foot 193 0.59 653 0.47 1.26 1.07-1.48 Hand 152 0.46 432 0.31 1.50 1.25-1.80 Leg 140 0.43 516 0.37 1.16 0.96-1.39 Multiple/extensive 2 0.01 5 0.004 1.70 0.33-8.79 Rib/sternum 124 0.38 355 0.25 1.49 1.21-1.83 Shoulder 8 0.02 45 0.03 0.76 0.36-1.61 Skull/face 167 0.51 502 0.36 1.42 1.19-1.69 Unspecified site 540 1.65 1364 0.98 1.69 1.53-1.86 Abbreviations: DMPA = depo-medroxyprogesterone acetate; Fx = fracture; PY = person years; HRT = hormone replacement therapy; IRR = incidence rate ratio; CI = Confidence Interval; N = number; ND = not determined. Table 7 shows the effect of duration of DMPA exposure on fracture risk, where high exposure was defined as 8 or more DMPA injections and low exposure was defined as fewer than 8 cumulative DMPA injections. Overall, low exposure to DMPA was associated with a higher risk for fracture, compared to non-users, than high exposure. The Page 11
study results do not provide a reason why an inverse dose response relationship was observed. Table 7. Effect of DMPA Exposure on Risk Rate Ratios After Initiation of Treatment Cohort (DMPA exposure level) DMPA (HIGH) DMPA (LOW) Non-Users Observation (PY x 1000) 83.04 244.29 1395.04 Fracture Groups IRR 95% CI IRR 95% CI Fx/1000PY All Fractures 1.22 1.13-1.30 1.47 1.40-1.54 6.37 All Fractures (except finger/toe, skull/face, unspecified) 1.21 1.09-1.34 1.41 1.32-1.49 3.83 Axial Skeleton 0.87 0.52-1.46 0.99 0.73-1.33 0.21 Appendicular Skeleton 1.23 1.11-1.37 1.43 1.34-1.52 3.62 Abbreviations: DMPA = depo-medroxyprogesterone acetate; Fx = fracture; PY = person years; IRR = incidence rate ratio; CI = Confidence Interval. Table 8 shows crude fracture incidence as a function of both DMPA exposure (low, high) and time since last DMPA injection ( current use of DMPA at the time of fracture; recent use, fracture occurred within the 2 years following the last DMPA injection; and past use, where the fracture occurred more than 2 years after the last DMPA injection). Compared with the incidence of 6.4 per 1,000 person-years in the unexposed women, crude fracture rates were higher in all 3 categories of exposure time since last injection, with an overall trend toward higher rates in the current usage category compared to the past usage category, but the confidence intervals for current, recent and past usage categories all overlapped and were not different. Rates were generally similar for low (1-7 injections) and high (8 or more injections) cumulative exposure in all 3 exposure-time categories, except for current time, where the fracture rate in the low cumulative exposure category (12.1) was greater than the rate in the high cumulative exposure category (6.9). Page 12
Table 8. Crude Incidence of Any Fracture per 1,000 Person-Years by Time from Last DMPA Injection and Cumulative Number of Injections Exposure Fractures P-Y Rate per 1,000 Crude IRR 95% CI Total 11,822 1,722,356 6.9 None a 8,887 1,395,040 6.4 1.00 - Total past a 1,011 119,928 8.4 1.32 1.24-1.41 Low b 892 106,400 8.4 1.32 1.23-1.41 High b 119 13,528 8.8 1.38 1.15-1.65 Total recent a 931 103,984 9.0 1.41 1.31-1.50 Low b 751 84,755 8.9 1.39 1.29-1.50 High b 180 19,228 9.4 1.47 1.27-1.70 Total current a 993 103,404 9.6 1.51 1.41-1.61 Low b 645 53,123 12.1 1.91 1.76-2.06 High b 348 50,281 6.9 1.09 0.98-1.21 Abbreviations: CI = confidence interval; DMPA = depo-medroxyprogesterone acetate; IRR = incidence rate ratio; P-Y = person-years. a. The reference level of nonexposure is composed of person-years of women who did not use DMPA and of person-years of DMPA users before their first DMPA prescription. Current exposure time is during episodes of active DMPA contraception, based on the interleaving of active 90-day exposures generated by each injection. Recent exposure time is up to 640 days after the last actively exposed day, and represents time after current exposure but less than 2 years since the last injection (not including the 90 days of current exposure time associated with the last injection). Past exposure begins after the last recently exposed day, and represents time 2 years or more after the last injection. b. Cumulative exposure was low if the woman had accumulated a history of 1 to 7 DMPA injections, and high if she had received 8 or more. Table 9 shows the rates standardized by age, using weights from the age distribution of the entire cohort person-time. The age-standardized rates, like the crude rates, were higher than the unexposed rate, and suggested a trend toward increased fracture rate from past to recent to current exposure time. With age standardization, fracture rate was markedly higher in the low exposure category among current users, compared to current users with high DMPA exposure, but this was not seen for either recent or past users. Page 13
Table 9. Age-Standardized Incidence of Fracture per 1,000 Person-Years by Time from Last DMPA Injection and Cumulative Number of Injections Exposure Standardized Rate per 1,000 P-Y Standardized IRR 95% CI None a 6.3 1.00 - Total past a 8.5 1.34 1.25-1.44 Low b 8.5 1.34 1.25-1.45 High b 7.4 1.16 0.96-1.41 Total recent a 9.2 1.45 1.34-1.56 Low b 8.8 1.39 1.27-1.52 High b 9.4 1.49 1.27-1.74 Total current a 9.6 1.51 1.41-1.62 Low b 13.7 2.16 1.75-2.67 High b 6.7 1.06 0.95-1.18 Abbreviations: CI = confidence interval; DMPA = depo-medroxyprogesterone acetate; IRR = incidence rate ratio; P-Y = person years. a. The reference level of nonexposure is composed of person-years of women who did not use DMPA and of person-years of DMPA users before their first DMPA prescription. Current exposure time is during episodes of active DMPA contraception, based on the interleaving of active 90-day exposures generated by each injection. Recent exposure time is up to 640 days after the last actively exposed day, and represents time after current exposure but less than 2 years since the last injection (not including the 90 days of current exposure time associated with the last injection). Past exposure begins after the last recently exposed day, and represents time 2 years or more after the last injection. b. Cumulative exposure was low if the woman had accumulated a history of 1 to 7 DMPA injections, and high if she had received 8 or more. CONCLUSION(S): This study demonstrated that the DMPA users did not experience an increase in fracture risk after starting DMPA treatment, compared to their pre-treatment level of risk. This finding is relevant to the 5 to 6 year period after first DMPA use, which was the average observation time in this study. The study also confirmed that DMPA users differ significantly from nonusers at baseline with respect to the rate at which fractures occur, which was the primary endpoint in the study. Overall, the DMPA cohort s excess of appendicular fractures, typically related to trauma, in the face of no excess of axial fractures more likely to be related to low BMD in the absence of trauma, suggests that the mechanism for the observed difference in fracture rates at baseline could possibly be related to exposure to trauma, but information on exposure to trauma was not available in the study database; the observed differences in fracture rates do not appear to be related to use of DMPA, and hence to the BMD decline that is know to occur with DMPA use. Page 14