In May 2017, Natural Womanhood and FACTS put out a call for people to sign their petition asking CDC to “update its website to reflect the best data available and to cite the individual effectiveness rate of each unique type of evidence-based fertility awareness based method (FABM)”, asserting that the CDC website currently “misrepresents the actual effectiveness of individual modern FABMs” by reporting a typical failure rate for all FABMs of 24%. They state that data on effectiveness of FABMs on the CDC website is based on “one study with a low quality research design, which explains why its conclusions are so far from what quality studies show us”. They posit “bias” in this estimate as it is based on a retrospective survey, and as it lumps together variants of calendar rhythm plus other FABMs. The assertions in this petition raise multiple concerns (outlined below).
As background, information from a nationally-representative survey, the National Survey of Family Growth (NSFG), is regularly used to estimate contraceptive failure rates for the US. An analysis of 2002 NSFG data was published in 2008 by Kost et al. Information from that analysis is combined with an extensive review of other research on contraceptive effectiveness to produce a chapter on contraceptive efficacy in the regularly updated book, Contraceptive Technology. The table on CDC’s website is adapted from the contraceptive effectiveness table in Contraceptive Technology. (Note: an updated analysis using 2006-2010 NSFG data was recently published by Sundaram et al., and will be included in the next edition of Contraceptive Technology, currently being revised.)
Bullet points below provide clarification on issues that the petition does not adequately represent.
1. Typical use effectiveness estimates for several other kinds of contraceptive methods are derived from the same NSFG analysis. Typical use estimates for withdrawal, male condoms, oral contraceptive pills, injectable contraception, and FABMs are all drawn from analyses of NSFG data; FABMs are not treated differently or unfairly.
2. NSFG analyses are extremely useful for determining typical use contraceptive effectiveness rates in the general population of the United States. All study designs have strengths and limitations which must be considered in conjunction with the question of interest. For the purpose of deriving typical use estimates of contraceptive use applicable to the general population in the United States, the NSFG analyses referenced above (and cited in Contraceptive Technology and on CDC's website) are based on high quality data and are calculated using rigorous scientific methods specifically designed for population-based data.
Both typical and perfect use pregnancy rates can be measured in clinical trials, but experts generally rely on typical use rates calculated with population-based survey data, in order to monitor the actual experience across a range of contraceptive users with differing demographic characteristics. While one drawback of this approach is reliance upon women’s recall of key information, there are also important drawbacks to calculating typical use contraceptive failure rates in clinical trials. Populations in clinical trials are more highly selected (and thus less generalizable to the wider population), and their behaviors may be impacted by frequent contact with investigators and study staff (this may be particularly true for methods that are highly user-dependent, such as FABMs). Perfect use pregnancy rates are best measured in clinical trials (and it is worth noting that a large proportion of FABM effectiveness studies have incorrectly calculated perfect use rates, as pointed out by Trussell and Grummer-Strawn 1990).
3. FABMs are lumped together for calculation of typical use rates for statistical reasons. The petition is correct that NSFG estimates lump FABM methods together, and that >80% of use is composed of “calendar rhythm” (which likely includes a proportion of women who are not even formally using calendar rhythm, but are intending to time intercourse based on their [informed or uninformed] understanding of the fertile period). Thus, it is accurate that the 24% estimate may not be applicable to all individual FABMs. Rather, it is applicable to most women in the US who report that they are using either “rhythm or safe period by calendar” or a “safe period by temperature or cervical mucus test, natural family planning.” It is important to understand that this lumping is done for statistical reasons. Overall FABM use is low in the US (~1% of all women, ~2-3% of all contracepting women, according to NSFG data), meaning there are not enough episodes of use of each individual FABM to generate statistically stable estimates for each method separately. With low use, the choice is to either not calculate an effectiveness estimate for typical use of FABMs in NSFG data or to lump. In other words, data on use of FABMs in the NSFG have been combined in order to shed light on the typical experience of US women using these methods; not in order to mislead.
In the most recent NSFG analysis by Sundaram et al, even when lumping FABM methods, there was an insufficient number of use segments to derive stable estimates. Thus, the paper does not report an estimate for typical use effectiveness of FABMs because the estimate is highly unstable and sensitive to the characteristics of the small number of women who had data on use of these methods in the survey. Again, this is not a decision made with an intent to ignore FABMs, this is based on scientific rigor and integrity.
4. CDC does list FABM methods separately for perfect use rates. The petition asserts that CDC is misleading the public, but fails to mention that CDC’s table presents perfect use pregnancy rates separately for Standard Days Method (5%), Two Day Method (4%), Ovulation Method (3%), and Symptothermal Method (0.4%), derived from clinical trials. As perfect use estimates are derived from clinical trials, in most cases, they represent the “best case scenario” for contraceptive effectiveness if the method is used correctly and consistently, all of the time. This is useful, encouraging information for women who may want to try these methods, but perfect use rates are not expected to represent the experience many women will have. Typical use estimates, particularly those generated using population-based data, tell us what the experience of users has been among a range of women with differing demographic and socioeconomic characteristics and under varying conditions of use, and so, they are quite useful, as they represent the challenges women face in using methods perfectly.
5. Scientifically robust approaches to providing updated information about FABM effectiveness rates are underway. For example, Contraceptive Technology is revised every few years to reflect the most recent data. The upcoming edition of Contraceptive Technology (21st Edition) includes new clarifications to help readers understand why typical use effectiveness of every individual FABM method cannot be estimated currently in population survey data. In addition, a systematic review of the effectiveness of prospective, clinical studies of all FABMs is underway; more information on this effort is available at this link. (Disclosure: I am involved in both of these efforts).
In sum, the petition asserts bias and lack of scientific rigor in current research findings and implies that the CDC is misleading people, when in fact, there are sound scientific reasons for the CDC’s approach to communicating the range of contraceptive options available and the evidence to-date of their contraceptive effectiveness. It is certainly worthwhile to call for more research into the likely wide-ranging effectiveness of different FABMs, particularly among typical users, but it is not necessary to diminish the value of the research and current body of evidence to do so.
At the same time, it is certainly important to acknowledge that a great deal of misunderstanding exists around FABMs, and that clear communication about these methods is necessary. For example, helping people understand the distinctions between the rhythm method and other FABMs (such as the sympto-thermal method, the TwoDay method, Standard Days method, Billings Ovulation method, sympto-hormonal methods such as Marquette, and other methods) would be useful in dispelling the misunderstanding that all women using fertility awareness based methods (or "natural family planning methods") are using the rhythm method (though again, a very large proportion of women in the US who report using some kind of fertility awareness report using calendar rhythm, although those reports may reflect formal or informal use). Finding productive and accurate ways to communicate about these methods is important, and will take collaborative efforts to achieve.
Bullet points below provide clarification on issues that the petition does not adequately represent.
1. Typical use effectiveness estimates for several other kinds of contraceptive methods are derived from the same NSFG analysis. Typical use estimates for withdrawal, male condoms, oral contraceptive pills, injectable contraception, and FABMs are all drawn from analyses of NSFG data; FABMs are not treated differently or unfairly.
2. NSFG analyses are extremely useful for determining typical use contraceptive effectiveness rates in the general population of the United States. All study designs have strengths and limitations which must be considered in conjunction with the question of interest. For the purpose of deriving typical use estimates of contraceptive use applicable to the general population in the United States, the NSFG analyses referenced above (and cited in Contraceptive Technology and on CDC's website) are based on high quality data and are calculated using rigorous scientific methods specifically designed for population-based data.
Both typical and perfect use pregnancy rates can be measured in clinical trials, but experts generally rely on typical use rates calculated with population-based survey data, in order to monitor the actual experience across a range of contraceptive users with differing demographic characteristics. While one drawback of this approach is reliance upon women’s recall of key information, there are also important drawbacks to calculating typical use contraceptive failure rates in clinical trials. Populations in clinical trials are more highly selected (and thus less generalizable to the wider population), and their behaviors may be impacted by frequent contact with investigators and study staff (this may be particularly true for methods that are highly user-dependent, such as FABMs). Perfect use pregnancy rates are best measured in clinical trials (and it is worth noting that a large proportion of FABM effectiveness studies have incorrectly calculated perfect use rates, as pointed out by Trussell and Grummer-Strawn 1990).
3. FABMs are lumped together for calculation of typical use rates for statistical reasons. The petition is correct that NSFG estimates lump FABM methods together, and that >80% of use is composed of “calendar rhythm” (which likely includes a proportion of women who are not even formally using calendar rhythm, but are intending to time intercourse based on their [informed or uninformed] understanding of the fertile period). Thus, it is accurate that the 24% estimate may not be applicable to all individual FABMs. Rather, it is applicable to most women in the US who report that they are using either “rhythm or safe period by calendar” or a “safe period by temperature or cervical mucus test, natural family planning.” It is important to understand that this lumping is done for statistical reasons. Overall FABM use is low in the US (~1% of all women, ~2-3% of all contracepting women, according to NSFG data), meaning there are not enough episodes of use of each individual FABM to generate statistically stable estimates for each method separately. With low use, the choice is to either not calculate an effectiveness estimate for typical use of FABMs in NSFG data or to lump. In other words, data on use of FABMs in the NSFG have been combined in order to shed light on the typical experience of US women using these methods; not in order to mislead.
In the most recent NSFG analysis by Sundaram et al, even when lumping FABM methods, there was an insufficient number of use segments to derive stable estimates. Thus, the paper does not report an estimate for typical use effectiveness of FABMs because the estimate is highly unstable and sensitive to the characteristics of the small number of women who had data on use of these methods in the survey. Again, this is not a decision made with an intent to ignore FABMs, this is based on scientific rigor and integrity.
4. CDC does list FABM methods separately for perfect use rates. The petition asserts that CDC is misleading the public, but fails to mention that CDC’s table presents perfect use pregnancy rates separately for Standard Days Method (5%), Two Day Method (4%), Ovulation Method (3%), and Symptothermal Method (0.4%), derived from clinical trials. As perfect use estimates are derived from clinical trials, in most cases, they represent the “best case scenario” for contraceptive effectiveness if the method is used correctly and consistently, all of the time. This is useful, encouraging information for women who may want to try these methods, but perfect use rates are not expected to represent the experience many women will have. Typical use estimates, particularly those generated using population-based data, tell us what the experience of users has been among a range of women with differing demographic and socioeconomic characteristics and under varying conditions of use, and so, they are quite useful, as they represent the challenges women face in using methods perfectly.
5. Scientifically robust approaches to providing updated information about FABM effectiveness rates are underway. For example, Contraceptive Technology is revised every few years to reflect the most recent data. The upcoming edition of Contraceptive Technology (21st Edition) includes new clarifications to help readers understand why typical use effectiveness of every individual FABM method cannot be estimated currently in population survey data. In addition, a systematic review of the effectiveness of prospective, clinical studies of all FABMs is underway; more information on this effort is available at this link. (Disclosure: I am involved in both of these efforts).
In sum, the petition asserts bias and lack of scientific rigor in current research findings and implies that the CDC is misleading people, when in fact, there are sound scientific reasons for the CDC’s approach to communicating the range of contraceptive options available and the evidence to-date of their contraceptive effectiveness. It is certainly worthwhile to call for more research into the likely wide-ranging effectiveness of different FABMs, particularly among typical users, but it is not necessary to diminish the value of the research and current body of evidence to do so.
At the same time, it is certainly important to acknowledge that a great deal of misunderstanding exists around FABMs, and that clear communication about these methods is necessary. For example, helping people understand the distinctions between the rhythm method and other FABMs (such as the sympto-thermal method, the TwoDay method, Standard Days method, Billings Ovulation method, sympto-hormonal methods such as Marquette, and other methods) would be useful in dispelling the misunderstanding that all women using fertility awareness based methods (or "natural family planning methods") are using the rhythm method (though again, a very large proportion of women in the US who report using some kind of fertility awareness report using calendar rhythm, although those reports may reflect formal or informal use). Finding productive and accurate ways to communicate about these methods is important, and will take collaborative efforts to achieve.
RSS Feed