- Email: [email protected]
Modeled estimates of myocardial infarction and venous thromboembolic disease in users of second and third generation oral contraceptives
Modeled Estimates of Myocardial Infarction and Venous Thromboembolic Disease in Users of Second and Third Generation Oral Contraceptives Pamela J. Schwingl*
and James Shelton+
Consistent reports from several recent studies suggestthat users of third generation oral contraceptives (OCs) containing gestodene and desogestrelmay be at increased risk of venous thromboembolic disease (VTE). Paradoxically, other reports indicate that these usersmay be at decreased risk of acute myocardial infarction (MI) compared with usersof secondgeneration OCs. To determine whether the potentially increased risk of VTE would outweigh the potentially reduced risk of MI in users of third generation OCs, we conducted an analysis to quantify the trade-offs providers and users may be faced to make between these formulations. The baseline rates of VTE and MI among non-users were calculated using US data on incidence and mortality of these conditions and estimates of the proportion of women exposed to these formulations in the US. These were multiplied by relative risks published in recent studies on third generation progestins to produce age- and formulationspecific risks. Results indicate that there would be small differences in diseaseburden between users of second and third generation OCs under the model assumptions at younger ages.However, among women 35-44 years of age, modeling results indicate that the potentially decreased incidence of MI among usersof third generation OCs more than offsets the potentially increased risk of VTE at this age. 0 1997Elsevier ScienceInc. All rights reserved. CQNTRACEPTION
ELSEVIER
1997;55:125-129
KEY WORDS: venous thromboembolism,
tion, attributed risk, third generation benefit analysis
myocardial infarcprogestins, risk/
*Family Health International. Division of Contraceptive Use and Epidemiology, Benefits Risk Unit, P.O. Box 13950, Research Triangle Park, NC 27709, and TOffice of Population, US Agency for International Development, 1601 N. Kent Street, #820m, Rosslyn, VA 22209 Name and address for correspondence: Pamela J. Schwingl, Ph.D., Family Health International, P.O. Box 13950, Research Triangle Park, NC 27709. Tel: (919) 544-7040, x403; Fax: (919) 544-7261; Internet: [email protected] Submitted for publication July 29, 1996 Revised December 19, 1996 Accepted for publication December 30, 1996
0 1997 Eisevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
Introduction everal recently published articles have consistently reported that users of third generation oral contraceptives (OCs) containing gestodene and desogestrel may be at increased risk of venous thromboembolic disease (VTE).le5 On the other hand, data from other sources indicate that these users may be at decreased risk of acute myocardial infarction (MI),“* compared with users of second generation OCs. The data on these relationships are relatively new and the publication of additional analyses from one of the key studies,5l9 has called into question the observed increased risks.” Lewis et a1.9 suggest that the findings of an increased risk of VTE in third generation users may be due, at least in part, to when the pills were introduced, rather than to the particular progestin contained in the pills. They reported a trend of increasing risk for the most recently introduced pills compared with the older levonorgestrel formulations. In addition, the authors found that OCs containing norgestimate (a progestin which is partially metabolized to levonorgestrel), but which were introduced some years after levonorgestrel, were associated with a higher relative risk for VTE compared with levonorgestrel-containing pills. Furthermore, an increased risk of VTE was noted among users of third generation OCs containing desogestrel combined with 20 ug of estradiol (chronologically newer], compared with pills containing the same amount of desogestrel combined with 30 ug of estradiol. To explain these observations, the authors argue that several types of bias may have been operating to produce these data. They argue that users of third generation products may constitute a set of individuals who are either less healthy than others, have experienced problems with OCs in the past, or else constitute a group of young first users, some of whom will be at genetically higher risk for VTE than the cohort of long-term users whose high risk members have been removed.
S
ISSN OOIO-7824/97/$17.00 PII SOOlO-7824(97)00026-7
126
Schwingl
and Shelton
Notwithstanding these concerns of potential bias, US practitioners are faced with the decisions about whether to prescribe third generation OCs. We believe that the potential trade-off between the risks and benefits of third generation OCs with respect to thromboembolic disease and MI should be evaluated now, at least to provide estimates of risk using the newly published estimates-’ If third generation OCs confer a lower risk of arterial thrombosis than other OCs currently in use, there may be a potential benefit to third generation users in terms of a lower risk of arterial thrombosis (including MI and potentially stroke). Because acute MI and stroke are both more common as women age, and also because the casefatality rate for MI is considerably higher than that of venous thromboembolism, we should quantify the trade-offs providers and users may now be faced to make between using these formulations for women in different age groups.
Materials
and Methods
We conducted a risk/benefit modeling analysis of the annual incidence of and mortality due to MI and VTE in the US according to age and generation of OC use. The purpose of this analysis was to determine whether the excess risk of VTE would outweigh the suggested reduced risk of MI in users of third generation OCs compared with users of second generation OCs, and whether this effect would vary by age. In our model, we calculated age-specific rates of VTE and MI among users of second and third generation OCs. The baseline risk for women using no OCs was calculated from the prevailing disease-specific incidence and mortality data, adjusting for the estimated proportion of women exposed to various OC formulations. We computed the generation-specific risks by multiplying the baseline risk by the estimated relative risk for incident or fatal disease for each generation of OCs. Estimates of the relative risk used in our model for incident and fatal MI and VTE (expressed as odds ratios in all cited case-control studies) among users of second and third generation OCs compared with nonusers are shown in Table 1. For users of second and third generation OCs compared with non-users, respectively, we used the estimates of relative risk of incident MI reported in Lewis et al. (RR = 3.1jl.l), or RR = 0.36 for users of third generation OCs compared with users of second generation OCS.~ We used estimates of the relative risk of fatal MI (RR = 1.9) from Thorogood’s study of fatal cardiovascular disease (CVD), which primarily included users of second generation OCs. l1 We assumed that the relative risk of fatal MI among third generation users compared with non-users was of the same magnitude as the inci-
Contraception 1997;55: 125-I 29
1. Estimates of relative risk of fatal and incident MI and VTE used in model
Table
MI OC Type
Third generation vs non-user Second generation vs non-user Third generation vs second generation ratio Non-user
Fatal
VTE
Incident
Fatal
Incident
1.16
1.16
4.2l”
7.213
1.91’
3.16
2.1L2
3,6’-2,“5
0.6
0.36 1.0
2.0 1.0
2.0 1.0
1.0
dence of MI among users of third generation OCs compared with non-users, since no reliable data are available to suggest otherwise. The relative risk of incident VTE among users of second generation OCs compared with non-users was computed as the average estimated relative risk reported in four recent studies (RR = 3.6).1-2,“5 For fatal VTE, the relative risk among users of second generation OCs was taken from Thorogood’s study of OCs and fatal VTE.i2 We assumed that the magnitude of the risk of VTE among users of third generation OCs compared to non-users would be twice as large as the risk in users of second generation OCs compared to non-users based on a review of data from these cited studies (incident VTE RR = 7.2; fatal VTE RR = 4.2). 2,4~5~9,13However, to ascertain the sensitivity of these analyses, we also ran a model under the assumption that users of third generation OCs were only at 1.5 times the risk compared with users of second generation OCs, based on data from the only study that was designed specifically to assess this association.5 Data on age- and formulation-specific prevalence of contraceptive use in the US come from the 1995 Ortho Birth Control Survey (personal communication, Ray Suehnholz, Ortho Pharmaceutical, New Jersey). The prevalence of second generation OCs is derived from the combined prevalence of formulations containing levonorgestrel, norethindrone, and norgestimate. Approximately half of all women using second generation pills take norethindrone formulations in the US, a much larger proportion than in the UK. While norgestimate is sometimes considered a third generation progestin, primarily based on its chronological introduction into the OC market, we included norgestimate as a second generation progestin in this analysis since norgestimate is largely metabolized to Desogestrel is the only third genlevonorgestrel.i4!15 eration progestin available in the US. To correctly apportion risk to users of second and third generation pills, the distribution of all pill users was incorpo-
Contraception 1997;55:125-129
Mortality
rated into the calculations. Only a small proportion of women take first generation formulations (those pills with 50 ug or more estrogen plus a progestagen), and these calculations are not presented. The average annual US mortality rates for the years 1991-1992 for MI (ICD-9 410) and VTE, including deep venous thromboembolism (ICD-9 451-453) and pulmonary embolism (415.1), come from National Center on Health Statistics; data on the average annual US incidence of these conditions are derived from the US National Hospital Discharge Database for 1988-1990. Because women who use OCs may have fewer risk factors due to medical screening, and because as women age, an increasing proportion of women with cardiovascular disease (CVD) have other known risk factors for MI and VTE, it is necessary to make some adjustments to the mortality and incidence data for these conditions. In particular, most cases of VTE are attributable to other risk factors that tend to increase according to age, leaving the rate of idiopathic disease fairly constant across age.16 We adjusted downward the number of reported deaths and cases according to the proportion of women with no other risk factors at each age reported in Stolley et al.” For MI, we adjusted downward the number of reported deaths and cases by the proportion with no other risk factors reported in Thorogood et al.‘l
Results Under the assumptions given and benefits of the different dependent. As shown in Table that total mortality from VTE lar in users of second and third Table women farction second
in the model, the risks OC regimens are age2, our results indicate and MI would be simigeneration OCs 1534
2. Average annual mortality per 100,000 healthy 15-44 years of age in the US from myocardial in(MI) and thromboembolic disease (VTE) in users of and third generation OCs
and Morbidity
in Users of Third Generation
Total mortality* Non-users Users of second generation OCs Users of third generation OCs Non-users MI VTE Users of second generation OCs MI VTE Users of third generation OCs MI
Table 3. Average annual incidence per 100,000 healthy
women 15-44 years of age in the US from myocardial infarction [MI) and thromboembolic disease (VTE) in users of second and third generation OCs
VTE ‘Differences
due
to rounding.
As
15-24 25-34 35-44 0.1 0.3 0.4
0.5 1.0 0.9
2.6 4.9 3.4
0.1
K
24 0:2
0.1 0.2
0.8 0.2
4.5 0.4
0.1 0.3
0.4 0.5
2.6 0.7
0.1
127
years of age, but that women 35-44 who use third generation OCs would have a lower rate of death (3.4 per 100,000) than users of second generation OCs (4.9/100,000). Older users would have a lower risk of death because the estimated increase in the mortality rate of VTE in users of third generation OCs (0.7/ 100,000) compared with users of second generation OCs (0.4/100,000) would be offset by a larger decrease in the mortality rate of MI in users of third generation OCs (2.6/100,000) compared with users of second generation OCs (4.5/100,000). The total incidence rate from VTE and MI among the women 15-24 and 25-34 is estimated to be somewhat higher among women 15-24 years of age who use third generation OCs (17.6/100,000) compared with those who use second generation OCs (12.1,’ 100,000) (Table 3). Similarly, among women 25-34 years of age, users of third generation OCs would have a higher total incidence rate (37.3/100,000) compared with users of second generation OCs (24.7/100,000). However, among the oldest women (35-44), the total incidence rate is estimated to be lower in users of third generation OCs (50.3/100,000) compared with users of secondgeneration OCs (62.6/100,000). This is explained by the lower incidence of MI among the oldest users of third generation OCs, which offsets the increased risk of VTE in this group. When we made the more conservative assumption that users of third generation OCs were only at 1.5 times the risk of VTE compared with users of second generation OCs, the relationships between the user groups and the direction of the results remained the same. However, under this assumption, while the total incidence of disease decreases among both users of
Age Diseaseand OC formulation
OCs
Diseaseand OC formulation
Total incidence* Non-users Users of second generation OCs Users of third generation OCs Non-users MI VTE
Users of second generation OCs MI VTE
Users of third generation OCs MI VTE ‘Differences
due
to rounding
15-24 25-34 3.5 12.1
35-44
17.6
7.2 24.7 37.3
19.4 62.6 50.3
1.3 2.2
2.4 4.8
14.7 4.7
4.0 8.1
7.4 17.4
45.5 17.1
1.4 16.2
2.6 34.7
16.1 34.2
128
Schwingl
and Shelton
second and third generation OCs 15-24 and 25-34 years of age, the comparative disadvantage to users of third generation OCs becomes smaller, and the comparative advantage among users 3544 years of age becomes larger. The comparative advantage in total mortality to users of third generation OCs slightly increases for both women 25-34 and 35-44 years of age, while for women 15-24 there are no material differences in mortality under the more modest assumption (data not shown).
Discussion The validity of our analysis depends on the validity of estimates of risk in the new studies, and in particular, on Lewis et al.‘@ preliminary result that third generation OCs reduce the risk of nonfatal MI compared with second generation OCs (RR = 0.36). Jick et al.’ reported a lower risk of MI, which was not statistically significant, among users of gestodene and desogestrel compared with users of second generation OCs (RR = 0.6-0.7, respectively). These estimates are based on only two exposed cases, and the confidence limits are, as a consequence, very wide. Lidegaard8 reported in a study of 91 cases of acute myocardial infarction that the relative risk ranged from l- to 4fold increase for users of second compared to users of third generation OCs, depending on the duration of use, but the differences were not statistically significant. While there are several sources of data which suggest that the incidence of VTE may be twice as high in users of third generation OCs as compared with users of second generation OCs, no such data exist concerning the risk of VTE mortality in third generation users, and it may be that mortality from VTE is not 50-100% higher, as assumed in our analysis. A small study of cardiovascular mortality reported by Jick et a1.,3 reports that among users of desogestrel, the only third generation progestin available in the US, the estimate of the relative risk of total cardiovascular mortality compared with users of levonorgestrel was 0.4 (0.1-2.1). In this study, one death among desogestrel users was due to pulmonary embolism, the second death was due to subarachnoid hemorrhage; while half of the eight levonorgestrel deaths were due to these conditions, the other half were due to ischemit and coronary conditions. The doubling of mortality risk from VTE in users of third generation OCs that we assumed, is compatible with the upper boundary of this confidence limit, and is based upon a worst case scenario. Our estimates of incidence and mortality of idiopathic VTE are consistent with published studies. Overall, the rate of VTE among non-users 1544 years
Contraception 1997;55:125-129
of age has been estimated to be 4/100,000, among second generation users 1O-l 5/l 00,000, and among third generation users 20-30/100,000 (based on unpublished data from Food and Drug Administration, 1995). We believe that these results, while based on data from the US where third generation OCs are less prevalent, are likely to be generalizable to the UK, since results from additional analyses (not shown) using prevalence data from the British studies, where prevalence of third generation progestins is much higher, yielded similar results. A caveat concerning the use of data from the European studies to infer incidence and mortality in the US is related to the fact that at least half of the second generation pills in the US contain norethindrone, whereas a very large proportion of second generation pills in the UK contain levonorgestrel. Based on previous studies suggesting that norethindrone and levonorgestrel have essentially similar cardiovascular risks at low doses, we grouped together users of these two second generation formulations.18,19 The classification of norgestimate as a second generation formulation in our analyses does not yield materially different results than if we had classified norgestimate as a third generation progestin, primarily because the proportion of women using norgestimate in the US is relatively small. Furthermore, the relative risk estimate of VTE published by Lewis et a1.,9 classifying norgestimate as a third generation progestin (RR = 1.6) as compared to classifying it as a second generation progestin (RR = 1.5), is subsumed in our sensitivity analysis of 2.0. When we incorporated this higher relative risk estimate into our analysis, our conclusions remained the same. Data are not available to conduct a formulationand smoking-specific analysis for the US, and it is not clear whether and to what degree cigarette smoking interacts with third generation OCs based on existing data. However, since the risk of VTE is constant across smokers and non-smokers,20,21 non-smoking users would be at the same increased risk of VTE as smokers, but would not be at substantially decreased risk of MI, given their already low risk of MI. On the other hand, OC users who smoke and are at greatly increased risk of MI, may stand to benefit more from decreased rates of MI, if third generation OCs are shown to decrease the risk in these users. Thus, third generation OCs might be most beneficial among users who continue to smoke, or possibly, among users who have other risk factors for MI. The impact of third generation OCs on the risk of stroke, particularly thrombotic stroke, which appears to be similar to MI epidemiologically and pathophysiologically, remains to be determined. While this condition is less common than MI among reproductive
Contraception 1997:55:125-l 29
age women, if, in future studies, we find that third generation OCs reduce the risk of all arterial thrombosis, then the balance of risks may be more definitively tipped toward third generation OCs, particularly in women over the age of 35. The risk/benefit assessment of oral contraceptives is complex, and our model addresses only some of the multiple factors that need to be taken into consideration by clinicians and decision makers. However, our model suggests that, based on the limited available data, and based on assumptions that third generation OCs may increase the risk of VTE by 50100%, there are very small differences in mortality and incidence of VTE and MI between users of second and third generation oral contraceptives. Women over 35 years of age and women with other MI risk factors, based on current data, may benefit from third generation pills. Given the significantly higher case fatality associated with MI as compared with VTE, a reduction in the incidence of these events could have a very important public health impact.
Mortality and Morbidity in Users of Third Generation OCs
6.
7. 8. 9.
10. 1I.
Acknowledgments Financial support for this study was provided by Family Health International (FHI) with funds from the United States Agency for International Development (USAID cooperative agreement #CCP-3079-A-005022-00), although the views expressed in this article do not necessarily reflect those of FHI or USAID.
14.
References
15.
1. World Health Organization Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Venous thromboembolic disease and combined oral contraceptives: results of international multicentre case-control study. Lancet 1995;346:1575-82. 2. World Health Organization Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Effect of different progestagens in low oestrogen oral contraceptives on venous thromboembolic disease. Lancet 1995;346:1582-8. 3. Jick H, Jick SS, Gurewich V, Myers MW, Vasilakis C. Risk of idiopathic cardiovascular death and non-fatal venous thromboembolism in women using oral contraceptives with differing progestagen components. Lancet 1995;346:1589-3. 4. Bloemenkamp KWM, Rosendaal FR, Helmerhorst FM, Buller HR, Vandenbroucke JP. Enhancement by factor V Leiden mutation of risk of deep-vein thrombosis associated with oral contraceptives containing thirdgeneration progestagen. Lancet 1995;346:1593-6. 5. Spitzer WO, Lewis AL, Heinemann LAJ, Thorogood M,
12. 13.
16. 17. 18. 19.
20. 21.
129
MacRae KD on behalf of Transnational Research Group on Oral Contraceptives and the Health of Young Women. Third generation oral contraceptives and risk of venous thromboembolic disorders: an international case-control study. BMJ 1996;312:83-8. Lewis AL, Spitzer WO, Heinemann LAJ, MacRae KD, Bruppacher R, Thorogood M on behalf of Transnational Research Group on Oral Contraceptives and the Health of Young Women. Third generation oral contraceptives and risk of myocardial infarction: an international casecontrol study. BMJ 1996;312:88-90. Jick H, Jick S, Myers MW, Vasilakis C. Risk of acute myocardial infarction and low-dose combined oral contraceptives. [Letter]. Lancet 1996;347:627-8. Lidegaard 0, Edstrom B. Oral contraceptives and myocardial infarction: a case-control study. [Abstract]. Eur J Contraception Reprod Health Care 1996;1:74. Lewis MA, Heinemann LAJ, MacRae KD, Bruppacher R, Spitzer WO. The increased risk of venous thromboembolism and the use of third generation progestagens: role of bias in observational research. Contraception 1996;54:5-13. Lidegaard 0, Milsom I. Oral contraceptives and thrombotic diseases: impact of new epidemiological studies. [Letter]. Contraception 1996;53:135-139. Thorogood M, Mann J, Murphy M, Vesey M. Is oral contraceptive use still associated with an increased risk of fatal myocardial infarction? Report of a case-control study. Br J Obstet Gynaec 1991;98:1245-53. Thorogood M, Mann J, Murphy M, Vessey M. Risk factors for fatal venous thromboembolism in young women: a case-control study. Int J Epid 1992;21:48-52. McPherson K. Third generation oral contraception and venous thromboembolism. BMJ 1996j312:68-9. Stanczyk FZ, Roy S. Metabolism of levonorgestrel, norethindrone and structurally related contraceptive steroids. Contraception 1990;42:67--96. Shenfield GM, Griffin JM. Clinical pharmacokinetics of contraceptive steroids: an update. Clin Pharmacokinet 1991j20:1537. Stadel B. Oral contraceptives and cardiovascular disease, Part I. N Engl J Med 1981j305:612-8. Stolley P, Tonascia J, Tockman M, Sartwell P. Thrombosis with low-estrogen oral contraceptives. Am J Epid 1975j102:197-208. Kay CR. Progestogens and arterial disease-Evidence from the Royal College of General Practitioners’ Study. Am J Obstet Gynecol 1982j142:762-5. Wahl P, Walden C, Knopp R, Hoover J, Wallace R, Heiss G, and Rifkind B. Effect of estrogen/progestin potency on lipid/lipoprotein cholesterol. N Engl J Med 19833 308:862-7. Lawson DH, Davidson JF, Jick H. Oral contraceptive use and venous thromboembolism: absence of an effect of smoking. BMJ 1977j2:729-30. Petitti DB, Wingerd J, Pellegrin F, Ramcharan S. Oral contraceptives, smoking, and other factors in relation to risk of venous thromboembolic disease. Am J Epidemiol 1978;108(6):480-5.
and James Shelton+
Consistent reports from several recent studies suggestthat users of third generation oral contraceptives (OCs) containing gestodene and desogestrelmay be at increased risk of venous thromboembolic disease (VTE). Paradoxically, other reports indicate that these usersmay be at decreased risk of acute myocardial infarction (MI) compared with usersof secondgeneration OCs. To determine whether the potentially increased risk of VTE would outweigh the potentially reduced risk of MI in users of third generation OCs, we conducted an analysis to quantify the trade-offs providers and users may be faced to make between these formulations. The baseline rates of VTE and MI among non-users were calculated using US data on incidence and mortality of these conditions and estimates of the proportion of women exposed to these formulations in the US. These were multiplied by relative risks published in recent studies on third generation progestins to produce age- and formulationspecific risks. Results indicate that there would be small differences in diseaseburden between users of second and third generation OCs under the model assumptions at younger ages.However, among women 35-44 years of age, modeling results indicate that the potentially decreased incidence of MI among usersof third generation OCs more than offsets the potentially increased risk of VTE at this age. 0 1997Elsevier ScienceInc. All rights reserved. CQNTRACEPTION
ELSEVIER
1997;55:125-129
KEY WORDS: venous thromboembolism,
tion, attributed risk, third generation benefit analysis
myocardial infarcprogestins, risk/
*Family Health International. Division of Contraceptive Use and Epidemiology, Benefits Risk Unit, P.O. Box 13950, Research Triangle Park, NC 27709, and TOffice of Population, US Agency for International Development, 1601 N. Kent Street, #820m, Rosslyn, VA 22209 Name and address for correspondence: Pamela J. Schwingl, Ph.D., Family Health International, P.O. Box 13950, Research Triangle Park, NC 27709. Tel: (919) 544-7040, x403; Fax: (919) 544-7261; Internet: [email protected] Submitted for publication July 29, 1996 Revised December 19, 1996 Accepted for publication December 30, 1996
0 1997 Eisevier Science Inc. All rights reserved. 655 Avenue of the Americas, New York, NY 10010
Introduction everal recently published articles have consistently reported that users of third generation oral contraceptives (OCs) containing gestodene and desogestrel may be at increased risk of venous thromboembolic disease (VTE).le5 On the other hand, data from other sources indicate that these users may be at decreased risk of acute myocardial infarction (MI),“* compared with users of second generation OCs. The data on these relationships are relatively new and the publication of additional analyses from one of the key studies,5l9 has called into question the observed increased risks.” Lewis et a1.9 suggest that the findings of an increased risk of VTE in third generation users may be due, at least in part, to when the pills were introduced, rather than to the particular progestin contained in the pills. They reported a trend of increasing risk for the most recently introduced pills compared with the older levonorgestrel formulations. In addition, the authors found that OCs containing norgestimate (a progestin which is partially metabolized to levonorgestrel), but which were introduced some years after levonorgestrel, were associated with a higher relative risk for VTE compared with levonorgestrel-containing pills. Furthermore, an increased risk of VTE was noted among users of third generation OCs containing desogestrel combined with 20 ug of estradiol (chronologically newer], compared with pills containing the same amount of desogestrel combined with 30 ug of estradiol. To explain these observations, the authors argue that several types of bias may have been operating to produce these data. They argue that users of third generation products may constitute a set of individuals who are either less healthy than others, have experienced problems with OCs in the past, or else constitute a group of young first users, some of whom will be at genetically higher risk for VTE than the cohort of long-term users whose high risk members have been removed.
S
ISSN OOIO-7824/97/$17.00 PII SOOlO-7824(97)00026-7
126
Schwingl
and Shelton
Notwithstanding these concerns of potential bias, US practitioners are faced with the decisions about whether to prescribe third generation OCs. We believe that the potential trade-off between the risks and benefits of third generation OCs with respect to thromboembolic disease and MI should be evaluated now, at least to provide estimates of risk using the newly published estimates-’ If third generation OCs confer a lower risk of arterial thrombosis than other OCs currently in use, there may be a potential benefit to third generation users in terms of a lower risk of arterial thrombosis (including MI and potentially stroke). Because acute MI and stroke are both more common as women age, and also because the casefatality rate for MI is considerably higher than that of venous thromboembolism, we should quantify the trade-offs providers and users may now be faced to make between using these formulations for women in different age groups.
Materials
and Methods
We conducted a risk/benefit modeling analysis of the annual incidence of and mortality due to MI and VTE in the US according to age and generation of OC use. The purpose of this analysis was to determine whether the excess risk of VTE would outweigh the suggested reduced risk of MI in users of third generation OCs compared with users of second generation OCs, and whether this effect would vary by age. In our model, we calculated age-specific rates of VTE and MI among users of second and third generation OCs. The baseline risk for women using no OCs was calculated from the prevailing disease-specific incidence and mortality data, adjusting for the estimated proportion of women exposed to various OC formulations. We computed the generation-specific risks by multiplying the baseline risk by the estimated relative risk for incident or fatal disease for each generation of OCs. Estimates of the relative risk used in our model for incident and fatal MI and VTE (expressed as odds ratios in all cited case-control studies) among users of second and third generation OCs compared with nonusers are shown in Table 1. For users of second and third generation OCs compared with non-users, respectively, we used the estimates of relative risk of incident MI reported in Lewis et al. (RR = 3.1jl.l), or RR = 0.36 for users of third generation OCs compared with users of second generation OCS.~ We used estimates of the relative risk of fatal MI (RR = 1.9) from Thorogood’s study of fatal cardiovascular disease (CVD), which primarily included users of second generation OCs. l1 We assumed that the relative risk of fatal MI among third generation users compared with non-users was of the same magnitude as the inci-
Contraception 1997;55: 125-I 29
1. Estimates of relative risk of fatal and incident MI and VTE used in model
Table
MI OC Type
Third generation vs non-user Second generation vs non-user Third generation vs second generation ratio Non-user
Fatal
VTE
Incident
Fatal
Incident
1.16
1.16
4.2l”
7.213
1.91’
3.16
2.1L2
3,6’-2,“5
0.6
0.36 1.0
2.0 1.0
2.0 1.0
1.0
dence of MI among users of third generation OCs compared with non-users, since no reliable data are available to suggest otherwise. The relative risk of incident VTE among users of second generation OCs compared with non-users was computed as the average estimated relative risk reported in four recent studies (RR = 3.6).1-2,“5 For fatal VTE, the relative risk among users of second generation OCs was taken from Thorogood’s study of OCs and fatal VTE.i2 We assumed that the magnitude of the risk of VTE among users of third generation OCs compared to non-users would be twice as large as the risk in users of second generation OCs compared to non-users based on a review of data from these cited studies (incident VTE RR = 7.2; fatal VTE RR = 4.2). 2,4~5~9,13However, to ascertain the sensitivity of these analyses, we also ran a model under the assumption that users of third generation OCs were only at 1.5 times the risk compared with users of second generation OCs, based on data from the only study that was designed specifically to assess this association.5 Data on age- and formulation-specific prevalence of contraceptive use in the US come from the 1995 Ortho Birth Control Survey (personal communication, Ray Suehnholz, Ortho Pharmaceutical, New Jersey). The prevalence of second generation OCs is derived from the combined prevalence of formulations containing levonorgestrel, norethindrone, and norgestimate. Approximately half of all women using second generation pills take norethindrone formulations in the US, a much larger proportion than in the UK. While norgestimate is sometimes considered a third generation progestin, primarily based on its chronological introduction into the OC market, we included norgestimate as a second generation progestin in this analysis since norgestimate is largely metabolized to Desogestrel is the only third genlevonorgestrel.i4!15 eration progestin available in the US. To correctly apportion risk to users of second and third generation pills, the distribution of all pill users was incorpo-
Contraception 1997;55:125-129
Mortality
rated into the calculations. Only a small proportion of women take first generation formulations (those pills with 50 ug or more estrogen plus a progestagen), and these calculations are not presented. The average annual US mortality rates for the years 1991-1992 for MI (ICD-9 410) and VTE, including deep venous thromboembolism (ICD-9 451-453) and pulmonary embolism (415.1), come from National Center on Health Statistics; data on the average annual US incidence of these conditions are derived from the US National Hospital Discharge Database for 1988-1990. Because women who use OCs may have fewer risk factors due to medical screening, and because as women age, an increasing proportion of women with cardiovascular disease (CVD) have other known risk factors for MI and VTE, it is necessary to make some adjustments to the mortality and incidence data for these conditions. In particular, most cases of VTE are attributable to other risk factors that tend to increase according to age, leaving the rate of idiopathic disease fairly constant across age.16 We adjusted downward the number of reported deaths and cases according to the proportion of women with no other risk factors at each age reported in Stolley et al.” For MI, we adjusted downward the number of reported deaths and cases by the proportion with no other risk factors reported in Thorogood et al.‘l
Results Under the assumptions given and benefits of the different dependent. As shown in Table that total mortality from VTE lar in users of second and third Table women farction second
in the model, the risks OC regimens are age2, our results indicate and MI would be simigeneration OCs 1534
2. Average annual mortality per 100,000 healthy 15-44 years of age in the US from myocardial in(MI) and thromboembolic disease (VTE) in users of and third generation OCs
and Morbidity
in Users of Third Generation
Total mortality* Non-users Users of second generation OCs Users of third generation OCs Non-users MI VTE Users of second generation OCs MI VTE Users of third generation OCs MI
Table 3. Average annual incidence per 100,000 healthy
women 15-44 years of age in the US from myocardial infarction [MI) and thromboembolic disease (VTE) in users of second and third generation OCs
VTE ‘Differences
due
to rounding.
As
15-24 25-34 35-44 0.1 0.3 0.4
0.5 1.0 0.9
2.6 4.9 3.4
0.1
K
24 0:2
0.1 0.2
0.8 0.2
4.5 0.4
0.1 0.3
0.4 0.5
2.6 0.7
0.1
127
years of age, but that women 35-44 who use third generation OCs would have a lower rate of death (3.4 per 100,000) than users of second generation OCs (4.9/100,000). Older users would have a lower risk of death because the estimated increase in the mortality rate of VTE in users of third generation OCs (0.7/ 100,000) compared with users of second generation OCs (0.4/100,000) would be offset by a larger decrease in the mortality rate of MI in users of third generation OCs (2.6/100,000) compared with users of second generation OCs (4.5/100,000). The total incidence rate from VTE and MI among the women 15-24 and 25-34 is estimated to be somewhat higher among women 15-24 years of age who use third generation OCs (17.6/100,000) compared with those who use second generation OCs (12.1,’ 100,000) (Table 3). Similarly, among women 25-34 years of age, users of third generation OCs would have a higher total incidence rate (37.3/100,000) compared with users of second generation OCs (24.7/100,000). However, among the oldest women (35-44), the total incidence rate is estimated to be lower in users of third generation OCs (50.3/100,000) compared with users of secondgeneration OCs (62.6/100,000). This is explained by the lower incidence of MI among the oldest users of third generation OCs, which offsets the increased risk of VTE in this group. When we made the more conservative assumption that users of third generation OCs were only at 1.5 times the risk of VTE compared with users of second generation OCs, the relationships between the user groups and the direction of the results remained the same. However, under this assumption, while the total incidence of disease decreases among both users of
Age Diseaseand OC formulation
OCs
Diseaseand OC formulation
Total incidence* Non-users Users of second generation OCs Users of third generation OCs Non-users MI VTE
Users of second generation OCs MI VTE
Users of third generation OCs MI VTE ‘Differences
due
to rounding
15-24 25-34 3.5 12.1
35-44
17.6
7.2 24.7 37.3
19.4 62.6 50.3
1.3 2.2
2.4 4.8
14.7 4.7
4.0 8.1
7.4 17.4
45.5 17.1
1.4 16.2
2.6 34.7
16.1 34.2
128
Schwingl
and Shelton
second and third generation OCs 15-24 and 25-34 years of age, the comparative disadvantage to users of third generation OCs becomes smaller, and the comparative advantage among users 3544 years of age becomes larger. The comparative advantage in total mortality to users of third generation OCs slightly increases for both women 25-34 and 35-44 years of age, while for women 15-24 there are no material differences in mortality under the more modest assumption (data not shown).
Discussion The validity of our analysis depends on the validity of estimates of risk in the new studies, and in particular, on Lewis et al.‘@ preliminary result that third generation OCs reduce the risk of nonfatal MI compared with second generation OCs (RR = 0.36). Jick et al.’ reported a lower risk of MI, which was not statistically significant, among users of gestodene and desogestrel compared with users of second generation OCs (RR = 0.6-0.7, respectively). These estimates are based on only two exposed cases, and the confidence limits are, as a consequence, very wide. Lidegaard8 reported in a study of 91 cases of acute myocardial infarction that the relative risk ranged from l- to 4fold increase for users of second compared to users of third generation OCs, depending on the duration of use, but the differences were not statistically significant. While there are several sources of data which suggest that the incidence of VTE may be twice as high in users of third generation OCs as compared with users of second generation OCs, no such data exist concerning the risk of VTE mortality in third generation users, and it may be that mortality from VTE is not 50-100% higher, as assumed in our analysis. A small study of cardiovascular mortality reported by Jick et a1.,3 reports that among users of desogestrel, the only third generation progestin available in the US, the estimate of the relative risk of total cardiovascular mortality compared with users of levonorgestrel was 0.4 (0.1-2.1). In this study, one death among desogestrel users was due to pulmonary embolism, the second death was due to subarachnoid hemorrhage; while half of the eight levonorgestrel deaths were due to these conditions, the other half were due to ischemit and coronary conditions. The doubling of mortality risk from VTE in users of third generation OCs that we assumed, is compatible with the upper boundary of this confidence limit, and is based upon a worst case scenario. Our estimates of incidence and mortality of idiopathic VTE are consistent with published studies. Overall, the rate of VTE among non-users 1544 years
Contraception 1997;55:125-129
of age has been estimated to be 4/100,000, among second generation users 1O-l 5/l 00,000, and among third generation users 20-30/100,000 (based on unpublished data from Food and Drug Administration, 1995). We believe that these results, while based on data from the US where third generation OCs are less prevalent, are likely to be generalizable to the UK, since results from additional analyses (not shown) using prevalence data from the British studies, where prevalence of third generation progestins is much higher, yielded similar results. A caveat concerning the use of data from the European studies to infer incidence and mortality in the US is related to the fact that at least half of the second generation pills in the US contain norethindrone, whereas a very large proportion of second generation pills in the UK contain levonorgestrel. Based on previous studies suggesting that norethindrone and levonorgestrel have essentially similar cardiovascular risks at low doses, we grouped together users of these two second generation formulations.18,19 The classification of norgestimate as a second generation formulation in our analyses does not yield materially different results than if we had classified norgestimate as a third generation progestin, primarily because the proportion of women using norgestimate in the US is relatively small. Furthermore, the relative risk estimate of VTE published by Lewis et a1.,9 classifying norgestimate as a third generation progestin (RR = 1.6) as compared to classifying it as a second generation progestin (RR = 1.5), is subsumed in our sensitivity analysis of 2.0. When we incorporated this higher relative risk estimate into our analysis, our conclusions remained the same. Data are not available to conduct a formulationand smoking-specific analysis for the US, and it is not clear whether and to what degree cigarette smoking interacts with third generation OCs based on existing data. However, since the risk of VTE is constant across smokers and non-smokers,20,21 non-smoking users would be at the same increased risk of VTE as smokers, but would not be at substantially decreased risk of MI, given their already low risk of MI. On the other hand, OC users who smoke and are at greatly increased risk of MI, may stand to benefit more from decreased rates of MI, if third generation OCs are shown to decrease the risk in these users. Thus, third generation OCs might be most beneficial among users who continue to smoke, or possibly, among users who have other risk factors for MI. The impact of third generation OCs on the risk of stroke, particularly thrombotic stroke, which appears to be similar to MI epidemiologically and pathophysiologically, remains to be determined. While this condition is less common than MI among reproductive
Contraception 1997:55:125-l 29
age women, if, in future studies, we find that third generation OCs reduce the risk of all arterial thrombosis, then the balance of risks may be more definitively tipped toward third generation OCs, particularly in women over the age of 35. The risk/benefit assessment of oral contraceptives is complex, and our model addresses only some of the multiple factors that need to be taken into consideration by clinicians and decision makers. However, our model suggests that, based on the limited available data, and based on assumptions that third generation OCs may increase the risk of VTE by 50100%, there are very small differences in mortality and incidence of VTE and MI between users of second and third generation oral contraceptives. Women over 35 years of age and women with other MI risk factors, based on current data, may benefit from third generation pills. Given the significantly higher case fatality associated with MI as compared with VTE, a reduction in the incidence of these events could have a very important public health impact.
Mortality and Morbidity in Users of Third Generation OCs
6.
7. 8. 9.
10. 1I.
Acknowledgments Financial support for this study was provided by Family Health International (FHI) with funds from the United States Agency for International Development (USAID cooperative agreement #CCP-3079-A-005022-00), although the views expressed in this article do not necessarily reflect those of FHI or USAID.
14.
References
15.
1. World Health Organization Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Venous thromboembolic disease and combined oral contraceptives: results of international multicentre case-control study. Lancet 1995;346:1575-82. 2. World Health Organization Collaborative Study of Cardiovascular Disease and Steroid Hormone Contraception. Effect of different progestagens in low oestrogen oral contraceptives on venous thromboembolic disease. Lancet 1995;346:1582-8. 3. Jick H, Jick SS, Gurewich V, Myers MW, Vasilakis C. Risk of idiopathic cardiovascular death and non-fatal venous thromboembolism in women using oral contraceptives with differing progestagen components. Lancet 1995;346:1589-3. 4. Bloemenkamp KWM, Rosendaal FR, Helmerhorst FM, Buller HR, Vandenbroucke JP. Enhancement by factor V Leiden mutation of risk of deep-vein thrombosis associated with oral contraceptives containing thirdgeneration progestagen. Lancet 1995;346:1593-6. 5. Spitzer WO, Lewis AL, Heinemann LAJ, Thorogood M,
12. 13.
16. 17. 18. 19.
20. 21.
129
MacRae KD on behalf of Transnational Research Group on Oral Contraceptives and the Health of Young Women. Third generation oral contraceptives and risk of venous thromboembolic disorders: an international case-control study. BMJ 1996;312:83-8. Lewis AL, Spitzer WO, Heinemann LAJ, MacRae KD, Bruppacher R, Thorogood M on behalf of Transnational Research Group on Oral Contraceptives and the Health of Young Women. Third generation oral contraceptives and risk of myocardial infarction: an international casecontrol study. BMJ 1996;312:88-90. Jick H, Jick S, Myers MW, Vasilakis C. Risk of acute myocardial infarction and low-dose combined oral contraceptives. [Letter]. Lancet 1996;347:627-8. Lidegaard 0, Edstrom B. Oral contraceptives and myocardial infarction: a case-control study. [Abstract]. Eur J Contraception Reprod Health Care 1996;1:74. Lewis MA, Heinemann LAJ, MacRae KD, Bruppacher R, Spitzer WO. The increased risk of venous thromboembolism and the use of third generation progestagens: role of bias in observational research. Contraception 1996;54:5-13. Lidegaard 0, Milsom I. Oral contraceptives and thrombotic diseases: impact of new epidemiological studies. [Letter]. Contraception 1996;53:135-139. Thorogood M, Mann J, Murphy M, Vesey M. Is oral contraceptive use still associated with an increased risk of fatal myocardial infarction? Report of a case-control study. Br J Obstet Gynaec 1991;98:1245-53. Thorogood M, Mann J, Murphy M, Vessey M. Risk factors for fatal venous thromboembolism in young women: a case-control study. Int J Epid 1992;21:48-52. McPherson K. Third generation oral contraception and venous thromboembolism. BMJ 1996j312:68-9. Stanczyk FZ, Roy S. Metabolism of levonorgestrel, norethindrone and structurally related contraceptive steroids. Contraception 1990;42:67--96. Shenfield GM, Griffin JM. Clinical pharmacokinetics of contraceptive steroids: an update. Clin Pharmacokinet 1991j20:1537. Stadel B. Oral contraceptives and cardiovascular disease, Part I. N Engl J Med 1981j305:612-8. Stolley P, Tonascia J, Tockman M, Sartwell P. Thrombosis with low-estrogen oral contraceptives. Am J Epid 1975j102:197-208. Kay CR. Progestogens and arterial disease-Evidence from the Royal College of General Practitioners’ Study. Am J Obstet Gynecol 1982j142:762-5. Wahl P, Walden C, Knopp R, Hoover J, Wallace R, Heiss G, and Rifkind B. Effect of estrogen/progestin potency on lipid/lipoprotein cholesterol. N Engl J Med 19833 308:862-7. Lawson DH, Davidson JF, Jick H. Oral contraceptive use and venous thromboembolism: absence of an effect of smoking. BMJ 1977j2:729-30. Petitti DB, Wingerd J, Pellegrin F, Ramcharan S. Oral contraceptives, smoking, and other factors in relation to risk of venous thromboembolic disease. Am J Epidemiol 1978;108(6):480-5.