While literature on hormonal contraceptives and condoms is abundant, published work on withdrawal (also known as coitus interruptus) is scarce, even though withdrawal is one of the most widely used temporary methods of birth control in the world.1 The most recent available estimates suggest that withdrawal use is relatively common among U.S. teenagers and young adults. According to data from the 2002 National Survey of Family Growth (NSFG), more than half of sexually experienced women aged 15–24 have ever used withdrawal.2 Similarly, a Kaiser Family Foundation national survey of sexually active 18–24-year-old men and women showed that 43% have ever used withdrawal, 20% consider it a safer-sex behavior and 12% use it regularly.3 Data from the 2003 Youth Risk Behavior Surveillance System suggest that 11% of sexually active teenagers (grades 9–12) used withdrawal at last sex.4 Analyses of 2002 NSFG data indicate that 15% of sexually active males aged 15–19 used withdrawal at last sex, as did 10% of those aged 20–25.5

Studies in the United States6,7 and Australia8 have reported on attitudes toward withdrawal use among adolescents and young adults. In interviews about the benefits and drawbacks of condoms, conducted in two midsize U.S. cities, almost 20% of a sample of male and female black 14–19-year-olds spontaneously identified withdrawal as a popular alternative to condoms.6 Likewise, a majority of 13–25-year-old Latinas in a Chicago study said they would use withdrawal if their partner agreed or had a preference.7 Australian men and women aged 18–25 participating in a series of focus groups indicated that within “regular” relationships, withdrawal is sometimes preferable to condoms for birth control.8 The main reported benefits of withdrawal are that it is more convenient and allows greater intimacy than condoms;6 the main drawbacks are concerns about males’ lack of self-control6,8 and reduced sexual pleasure, primarily for males.8

Despite the widespread use and popularity of withdrawal, it has largely been ignored and devalued as a viable contraceptive option by reproductive health care providers in the United States, presumably because of negative assumptions regarding its effectiveness.1,9 Published estimates of failure rates for withdrawal have varied. For instance, a six-year prospective study involving married women aged 25–39 found the overall failure rate for withdrawal to be 7%.10 In contrast, analyses based on 1995 and 2002 NSFG data estimated failure rates in one year of typical use to be 27% and 18%, respectively (demographic differences between samples were implicated in the different findings).11–13 A well-regarded reproductive health reference, Contraceptive Technology, asserts that withdrawal’s failure rate with typical use is equivalent to those of barrier methods (e.g., 17% for male condoms).14 It is unclear how widely this comparability is appreciated by reproductive heath providers or educators.

In this article, we present withdrawal-related findings from a wide-ranging qualitative study of the contraceptive attitudes and experiences of young adults. The primary research aim was to explore the attitudes, norms and self-efficacy of young adults regarding the involvement of intimate partners in contraceptive decisions and the influence of these partners on contraceptive use. We focused on young adults because they exhibit high rates of STDs and unintended pregnancy3,4 but are relatively understudied. The qualitative approach permitted participants to share in their own words details of their evolving contraceptive experiences over a variety of relationships, and gave researchers gain an in-depth understanding of participants’ opinions, influences and experiences.15

METHODS

Sample

Recruitment for the study took place between March 2006 and June 2007. To be eligible, individuals had to be 18–25 years of age; self-identify as black, Hispanic or white; have ever engaged in heterosexual sex; have ever used a condom or another contraceptive; speak English; and live in Philadelphia or one of four surrounding suburban counties. Potential participants were recruited through street outreach, flyers posted in family planning clinics, advertisements on an online community forum and snowball sampling (i.e., participants were encouraged to refer their friends to the study). We employed a theoretical sampling approach calling for an ongoing process of concurrent data collection, coding and analysis; this approach facilitates the identification of evolving theories and explanations, and helps to prevent premature termination of data collection, data redundancies and insufficiencies, and unfocused conclusions.16

In all, 140 men and women were screened for the study, and 131 met eligibility requirements. Thirty-six of those eligible did not attend their interview appointment; screening variables did not significantly differ between these individuals and the 95 (41 males and 54 females) who enrolled in the study. Although the sample size was originally set at 120, recruitment was curtailed at 95 participants because the research team determined that saturation had been reached—i.e., that additional interviews would reveal little or no new thematic information regarding contraceptive use patterns.17

Procedure

Four female project staff members conducted life-history interviews with participants in private offices. Because adolescent males prefer female health care providers, including for genital exams,18 we anticipated that young men and women would be comfortable discussing their sexual and contraceptive experiences with a female interviewer. Interviewers received extensive training from an experienced qualitative interviewer, and the project manager conducted mock interviews with project staff and assessed their performance before data collection began. Quality control supervision continued through weekly meetings, in which staff discussed their experiences and provided each other with feedback and suggestions on interviewing strategies. Before each interview, participants gave written informed consent and permission to audio-record the interview, and completed a brief face-to-face survey covering demographic characteristics, sexual history and contraceptive use. Participation in study activities generally took 1.5–2 hours to complete, and participants received a $30 honorarium to compensate them for their time. Ethical approval for this study was obtained from the institutional review board of the Family Planning Council.

The theory of planned behavior19 helped to guide the study design and the development of the interview guide and codebook. According to this theory, a person’s intention is the primary determinant of a given volitional behavior (in the current context, contraceptive use). Three constructs represent the immediate determinants of intention: a person’s attitudes, subjective norms and perceived behavioral control. Attitudes refer to behavioral beliefs (e.g., advantages and disadvantages of contraceptive use). Subjective norms refer to beliefs about how specific individuals or groups (e.g., partners and family members) view the behavior. Perceived behavioral control refers to how easily and skillfully individuals think they can execute the behavior. In general, people intend to perform a behavior when they evaluate it positively, experience social pressure to perform it, and believe they have the means and opportunity to do so. The theory of planned behavior recognizes that background factors (e.g., age and gender) can indirectly influence intentions and behavior through their effects on more specific determinants.

The open-ended life-history interview guide covered contraceptive knowledge; communication and decision making regarding method selection, use and discontinuation; and reproductive values and goals. Participants were asked to describe their contraceptive history in their own words, starting with their consensual sexual debut and discussing up to five subsequent, self-selected intimate relationships. They were also asked to describe their encounters with providers regarding sexual health and family planning. Probes were used as necessary to focus the interview, transition to a new topic, elicit additional information and clarify information provided.

Pile-sort exercises augmented the qualitative interview by providing insight into participants’ knowledge of and attitudes toward contraceptive methods. Participants were given a set of 14 cards, each showing the name of a contraceptive method, and were asked to sort the cards into piles on the basis of specific criteria.20 They were first instructed to sort the cards into as many piles as they wished on the basis of perceived effectiveness in preventing pregnancy. Once this task was completed, they provided a brief description of each pile (e.g., very effective, somewhat effective, not sure how effective). The interviewer recorded which cards were in each pile and corresponding descriptions. A similar pile-sort exercise was then used to assess participants’ perception of methods’ popularity.

Data Management

•Quantitative. Screening and demographic data were analyzed using SPSS 16.0. Pile-sort data were organized using Microsoft Excel, and response categories were reviewed. For each pile-sort exercise, the number of times each method appeared in the “most” and “least” piles (i.e., most and least effective, most and least popular) was tabulated, and descriptions of the “most” and ”least” piles were reviewed.

•Qualitative. Interviews were transcribed by a professional transcription company. Transcripts were reviewed for accuracy and edited as required. To facilitate access to the data, a topical summary of each transcript was prepared by a member of the research team.15 Framework analysis21 was used to examine the qualitative data. Similar to grounded theory, framework analysis is a more structured approach that allows for the inclusion of a priori concepts in conjunction with themes that emerge from the reading of the transcripts. These concepts and themes, and the study’s primary aims, guided the research team as they developed a codebook.

Two members of the research team independently coded 18 of the 95 interview transcripts through seven coding rounds, using Atlas.ti software. Given the complexity of coding open-ended interviews, the research team leader divided the transcripts into segments (on the basis of the interviewer’s questions), and each segment became a unit of coding. Cohen’s kappa scores were calculated for each code in each round to determine the level of intercoder agreement. The research team met after each coding round to refine the codebook as necessary, given the coders’ experiences and the level of agreement. At completion of the final coding round, kappa scores for individual codes ranged from 0.63 to 0.87, indicating that intercoder agreement was reached for all codes.22 The remaining interviews were then divided between the two coders and coded independently.

The primary codes were communication; condoms; decision-making or influencing factors; pregnancy and children; contraceptive methods (not including condoms); and theory of planned behavior concepts pertaining to contraceptive use, decisions and communication. As part of a secondary analysis, codes were further subdivided (e.g., “withdrawal” was a subdivision of the broad code “contraceptive methods”). For this article, the research team reviewed the interview summaries and the complete transcript text to which the secondary code “withdrawal” had been applied. The team met regularly to discuss themes and patterns emerging from the data and, after all data had been reviewed, reached consensus regarding key themes and findings.23

Because the primary aim of the qualitative interviews was to explore participants’ experiences, the course of the discussion was determined by each participant’s contraceptive history. Thus, the use of withdrawal was not systematically explored in all interviews. However, as withdrawal emerged as a recurring topic in early interviews, withdrawal-specific probes were incorporated into the interview guide.

RESULTS

Quantitative Analyses

Forty-three percent of participants were black, 20% were Hispanic and 37% were white. Participants’ mean age was 21.7 years (standard deviation, 2.3). Eighty-four percent had at least a high school diploma or its equivalent, 57% had completed some college and 78% had worked in the three months prior to the interview. Only four participants had ever been married; three of these were separated or divorced. In the 30 days prior to interview, 91% had been sexually active, and 19% had had sex with two or more partners. While 56% of females had been pregnant, 38% of males had gotten a woman pregnant; 26% of females and 16% of males had at least one child. Forty percent of females and 12% of males had ever received an STD diagnosis.

In the first pile-sort exercise, 84% of participants placed withdrawal in their “least effective” pile (Table 1). Five percent of all participants cited user error as the reason for its ineffectiveness; 13% indicated that using withdrawal required luck. Seventeen percent of participants put withdrawal in a pile generally described as “not that effective, but worth a try.” While most participants placed withdrawal in their “least effective” pile, 73% placed it in their “most popular” pile. Only the male condom and the pill were placed in the “most popular” pile more often.

Table 1. Percentage of sexually experienced men and women aged 18–25 who had ever used a contraceptive, by perceptions of which methods are the most and least effective, and the most and least popular, Philadelphia, 2006–2007
Method Most (N=95) Least (N=95)
  1. Note: Percentages are based on results of a pile-sort exercise, in which participants sorted cards naming the various methods into as many piles as they wished to rank effectiveness and popularity.

Effective
Vasectomy 74  8
Tubal ligation 65  8
Injectable 56  7
Male condom 51  8
Pill 51  7
Emergency contraception 42 16
Female condom 32 14
Patch 31 11
IUD 23 13
Ring 17 12
Spermicide 13 20
Diaphragm 13 19
Withdrawal  8 84
Rhythm  2 47
Popular
Male condom 93  2
Pill 80  4
Withdrawal 73  8
Injectable 62  4
Patch 53 12
Emergency contraception 52 12
Female condom 19 45
Spermicide 12 41
Ring 12 40
Diaphragm  8 43
Vasectomy  6 72
Tubal ligation  6 70
IUD  4 48
Rhythm  3 48

Qualitative Analyses

•Reasons for use. Participants gave a variety of explanations for choosing withdrawal in various contexts. Dissatisfaction with commonly accepted contraceptive methods (e.g., hormonal birth control or condoms) could lead to inconsistent use or abandonment of these methods, and to reliance on withdrawal as the primary method of pregnancy prevention. Participants also described using withdrawal as a secondary method of pregnancy prevention in conjunction with hormonal contraceptives or condoms, rather than as a substitute for them. Withdrawal was occasionally used in lieu of a condom when one was not available or desired. Relationship development and establishment of trust also played a role in the decision to use withdrawal instead of condoms or hormonal methods.

Participants overwhelmingly acknowledged that they had at least basic knowledge of their contraceptive choices and understood the role of condoms as a means of preventing pregnancy and STDs, including HIV. However, both men and women mentioned that condoms cause physical discomfort and decrease pleasure and intimacy. Many participants saw withdrawal as a short- or long-term alternative to condom use. Some also expressed a negative opinion of hormonal contraception because of concern regarding adverse side effects (potential or experienced) or a belief that these methods are “unnatural.” For example:

“She was not on the pill. … It’s not something she wants to put into her body. We tried to use condoms. … That didn’t really work. So I switched to a combination of withdrawal and rhythm.”—White male, age 25

Even among participants who generally favored using commonly accepted methods, some reported using withdrawal in conjunction with a hormonal method or condoms to augment contraceptive efficacy. In a few cases, men did not fully trust either the effectiveness of the hormonal method their partner used or her ability to use it correctly. One man related:

“I really didn’t know nothing about the patch. … She said it’s, like, ninety-seven percent this and that, and I thought, whatever. … I pulled out anyway, just to be on the safe side. … I didn’t really have too much faith in the patch.”—Black male, age 22

Participants also described occasionally using withdrawal when they had intended to use condoms but condoms had been unavailable, they had gotten caught up in the heat of the moment or they had been intoxicated. A female participant recalled:

“We were out drinking all night or doing whatever, and just didn’t think about it. Then, midway through having sex … it would register that we are not using a condom, so he would pull out.”—White female, age 23

One of the main themes that emerged was the role of trust in decisions about preventive behavior within a relationship. Frequently, trust emerged in the context of participants’ perceptions of their partner’s level of STD risk. One participant (a black male, age 25) said that he would use withdrawal if he did not have a condom, but only “with someone I’ve been with for a long period of time, so I knew everything about her.”

Women often talked about trusting a partner to withdraw before ejaculation, as in the following remarks:

“I didn’t really care too much, because I trusted him to just pull out. He was very active in his religion, and he was a really sweet guy. He actually acted a lot older than his age. … He was knowledgeable and caring about other people, so I trusted him for that.”—White female, age 21

Although participants generally characterized withdrawal use as most common and acceptable in long-term or committed relationships, they sometimes reported using it in other contexts. One participant (a white female, age 21) recounted an occasion on which a partner had asked if she was on the pill, “I said no, I wasn’t on anything—so we just decided that [withdrawal] is what we would do.” When the interviewer asked whether this had been “just a date,” the participant replied affirmatively.

•Expectations and normative beliefs. Many participants viewed withdrawal as a popular and accepted part of sexual behavior. One participant explained:

“I guess once you reach a certain age—like, you get a certain sexual experience—you expect to do things. So she expected me to pull out—and I was going to pull out.”—Black male, age 18

For some, this expectation existed even when the woman was using a hormonal contraceptive. A hormonal method user commented:

“For a long time, I thought [using withdrawal] was something that everybody did. … It’s what I had always done—and especially when [my partner] just did it. I was like, oh, okay. Well, I guess everybody just does that.”—White female, age 21

Participants said that young people might not talk about using withdrawal because they assume that it will be used if they are not using a condom. For example, one participant offered the following recollection:

“I told her, ‘I don’t have a condom,’ and she didn’t say nothing. So that’s usually the okay [to use withdrawal].”—Black male, age 22

Some believed that a man should not ejaculate inside a woman unless she specifically gives him permission to do so, as the following quote illustrates:

“I was like, ‘Please tell me you just didn’t do that [ejaculate inside me]!’ He didn’t know that I didn’t want him to, so he was like, ‘What? What did I do?’ I was like, ‘You can’t do that! You’re not supposed to do that!’ I was angry at first, but he explained to me that he didn’t know that I didn’t want him to, and that he didn’t know that he couldn’t.”—White female, age 21

A few participants elaborated on this theme, indicating that a man who does not withdraw when he is expected to is committing a serious betrayal of trust. When asked what would have happened if his partner had asked him to pull out and he had not, one participant (a white male, age 22) replied, “[That] would have been rape.”

Few participants reported having received any education in school or from a health care provider about using withdrawal.

•Effectiveness and skill. While participants generally considered withdrawal use to be normative, some were concerned about its effectiveness. One (a white female, age 22) said, “It’s like Russian roulette … not at all enough to give you a safe feeling.” Another had this to say:

“I was always nervous about if withdrawal would actually work. … It was actually quite effective, since I didn’t ever get pregnant—but I would not recommend it. It was extremely nerve-wracking. I thought it was very risky, and every month, waiting for the period … it was scary.”—White female, age 21

Female participants with anxiety about withdrawal described themselves as being unsure or nervous about their partner’s ability to withdraw prior to ejaculation. One (a white woman, age 24) said that she was always “scared,”“paranoid” and “worried” about whether a partner withdrew in time. Another (a Hispanic woman, age 18) explained that it was hard to use withdrawal every time because “you don’t know if [your partner] can be quick enough.”

Similarly, a few men revealed instances when they were not able to withdraw in time or were unsure if they had done so. For example, one participant reported:

“We kicked it for a while, and then I did the same thing, withdrawal. I did that [withdrawal] and it was too late. I was like, ‘Oh, oh.’ I straight bust [ejaculated] up in her.”—Black male, age 25

Though some participants were unsure of withdrawal’s effectiveness or believed it to be ineffective, others expressed confidence in the method’s ability to prevent pregnancy. This belief was generally based on individuals’ perceptions of reproductive biology or their own (or peers’) experience of using it successfully. As one man explained it:

“I thought [withdrawal] was going to be very effective, just for the simple fact that I’m not ejaculating in her. … So I felt if that didn’t happen, then she wouldn’t get pregnant.”—Black male, age 22

Another man (black, age 23) recalled that he and his partner had repeatedly had sex without a condom and that “she never got pregnant, so we just decided not to use a condom.” The interviewer asked if they had taken any other measures to prevent pregnancy, “like pulling out,” and the participant replied, “Yes, that’s it.”

Participants who considered withdrawal effective reported confidence in a man’s ability to withdraw before ejaculation. In the words of one participant (a Hispanic female, age 22), “Oh, he knew his own body, just like I know mine.”

Additionally, some female participants felt they could ensure correct use by monitoring their partner’s level of arousal and anticipating when ejaculation would occur. One (a black woman, age 25) explained, “Like if you on top, you could feel when he about to … so you get off.” Another recalled that although she wanted to understand this, it was not easy to tell when her partner’s climax was occurring:

“My older friend, she used to always tell me, like …‘You can tell when a boy is coming.’… I’d be like, ‘How?’ She’d be like, ‘Because they jerk. …’ I’m trying to figure out, like, you jerking your body?”—Black female, age 19

•Barriers to motivation. Some participants who used withdrawal acknowledged that their use was inconsistent. One of the most common reasons expressed for inconsistent use was that sex was more pleasurable if the male did not pull out, as the following comment reflects:

“It’s just always, like, that moment of pleasure. … I’m like, ‘This is great, go. Just keep it there. Don’t move, just stay there.’”—Black female, age 19

Another commonly cited factor in inconsistent use was being intoxicated. One participant recalled:

“I didn’t pull out or anything. I was just high, and I was just relaxed. It was feeling so good … that I didn’t want to pull out. I just wanted to keep on going.”—Black male, age 23

•Education and knowledge. Few participants reported having received any education in school or from a health care provider about using withdrawal. One woman (black, age 18) recalled that although her doctor had told her about the IUD, condoms and hormonal methods, “He didn’t talk about withdrawal. I guess that wasn’t one of his options.” Another (a black woman, age 24) related that she and her doctor had never discussed withdrawal, but her doctor “just always said to use a condom so you don’t get a disease. The doctor said to use a condom and not just the birth control.”

Rather, participants had heard or learned about withdrawal primarily from friends, partners, older siblings, parents or entertainment media. As one recollected:

“No one ever, you know, talked to me about sex and everything like that, so I had to find out on my own—like, in the streets, from my friends. And all they talked about was condoms, and he pulled out.”—Black female, age 25

Some participants did not remember specifically learning about withdrawal and reported using it because it was instinctive—they had concluded that if ejaculation causes pregnancy, withdrawal prior to ejaculation would prevent pregnancy. In one man’s words:

“I just knew that it was possible to get a girl pregnant. I had no idea how or, you know … ovulation or anything like that. So I just knew not to come inside someone.”—White male, age 22

The need for education about withdrawal was evident. For example, some participants considered withdrawal to be ineffective because they believed that pre-ejaculate generally contains sperm:

“As I got older, I started to learn that pulling out don’t even work, because you got [pre-ejaculate]—and, yes, [that] can get somebody knocked up.”—Black male, age 24

Further, some women admitted feeling uncomfortable using withdrawal because they could not tell whether their partner ejaculated inside them, and this caused them embarrassment about their naiveté and concern about the effectiveness of withdrawal use. One admitted:

“I’m embarrassed to say this, but … he would try to say that he got soft in me. He was ejaculating in me the whole time, and I don’t know why I didn’t think about it. … After a while I asked him, and then he told me that he did every time. … I was upset, but I felt kind of stupid, just because I fell for it for so long.”—Hispanic female, age 18

Participants generally did not think of withdrawal as a legitimate contraceptive method. One participant (a white female, age 22) said that during the second year of one relationship, she and her partner had “completely unprotected sex … save the withdrawal method.” Another (a black male, age 23), explaining his use of withdrawal, said, “I don’t want to call it [a] contraceptive, but that’s the last option you have to using a condom to prevent a pregnancy.”

Specific probing was often necessary to encourage participants to share their experiences and attitudes about withdrawal. One male (white, age 25) told the interviewer that he and his partner were doing nothing to prevent pregnancy, but when the interviewer asked if he was “pulling out or anything,” he replied, “Yeah. … All the time.”

DISCUSSION

Our findings show that although the use of withdrawal as a contraceptive method is popular among teenagers and young adults, opinions about its effectiveness are mixed. Many of the themes that emerged in the interviews correspond with the main elements of the theory of planned behavior: attitudes (effectiveness, knowledge), norms (expectations), behavioral control (skill) and intentions (barriers to motivation). The reasons participants gave for using withdrawal combined aspects of all of these elements.

Participants detailed a variety of experiences about the use of withdrawal and expressed a range of attitudes toward it. Some described confidence and skill in using the method, and found the method effective in practice. Many of the positive reasons participants gave for using withdrawal-—its convenience, lack of side effects, connotations of trust and being more pleasurable than condoms—expand upon ones suggested in other studies.6,8,9 Our finding that withdrawal is an expected alternative to condoms in both casual and long-term relationships has also been noted by others.6,8 In addition, we have documented withdrawal’s use both as an alternative to and in conjunction with hormonal contraception.

De Visser’s study of Australian men suggested that anxiety about self-control when practicing withdrawal was a greater impediment to male sexual pleasure than the perceived discomfort associated with condom use.8 In contrast, few of our participants described withdrawal as reducing sexual pleasure, and the consensus was that it was more pleasurable than condoms. However, we have found evidence of ambivalent and negative attitudes toward withdrawal’s efficacy. Participants (especially women) expressed anxiety concerning potential lack of male self-control and the risk of pregnancy. This finding was also described by Horner at al.6 As we have noted, this relates to both the perceived ineffectiveness of the method itself and concerns about incorrect or incomplete use. Some participants believed that withdrawal is not reliable because pre-ejaculate fluid contains viable sperm—a belief that has not been supported by research.24 Participants did not think of withdrawal as legitimate contraception and generally did not talk about its use with their health care providers, even in discussions about contraception. As a result, participants generally acquired knowledge about withdrawal from less reliable sources.

Limitations

The participants in this study were recruited (using nonprobability sampling) from a major eastern U.S. city and its surrounding suburban counties; the generalizability of the results is therefore limited. Some literature suggests that self-reported data may be inaccurate, as participants may misremember or misrepresent information;25 other work, however, has found self-reported data to be reliable.26 Given the sensitive nature of our study, a variety of strategies were used to minimize bias due to recall and social desirability (e.g., a life-history framework guided discussion of events and decisions, and interviewers were nondirective and nonjudgmental). Given the retrospective nature of these interviews and the frequently long periods of recall (some participants’ contraceptive histories covered more than 10 years), some reporting of events likely was inaccurate. In addition, because withdrawal was not the primary focus of the contraceptive life-history method, some interviews explored withdrawal in more detail than others.

Because our interviews focused mainly on contraceptive experiences, participants were not always asked in detail about STD prevention or their perceptions of STD risk. In general, participants discussed withdrawal in the context of pregnancy prevention, but they often mentioned its use in relation to their trust of a partner, suggesting that they recognized that from an STD risk standpoint, withdrawal is tantamount to unprotected sex. Other studies have found that withdrawal is used for pregnancy prevention primarily when the perceived risk of STD is low.6,8

Implications

While we are not promoting the use of withdrawal over other contraceptive methods, our findings underscore the need for providers to assess and provide guidance about withdrawal use. Furthermore, providers may benefit from understanding the variety of reasons for, expectations regarding and patterns of withdrawal use. Addressing correct use of withdrawal is particularly important with family planning patients for whom it is the primary method of contraception (and for regular condom users, who may occasionally use withdrawal if a condom is unavailable or undesired9), because, for example, its effectiveness may diminish when a man ejaculates more than once during a sexual encounter.27 Patients whose health care needs suggest that they may have had unprotected sexual activity (e.g., those who request emergency contraception, pregnancy tests or STD tests) may, with suitable probing, reveal routine or occasional reliance on withdrawal. Furthermore, discussion of reasons for withdrawal use may uncover reasons for avoidance of more reliable methods that a health care provider could address. Hormonal contraceptive users could benefit from an accurate understanding of the correct use of and risks associated with withdrawal as a secondary, or backup, method, given high rates of inconsistent method use, method switching and gaps in use that leave many at risk for unintended pregnancy.13,28 In this context, it may be appropriate to promote withdrawal as a backup contraceptive method, as providers do with emergency contraception.9,29

As our findings suggest that providers may not be aware of their patients’ use of withdrawal, future research should examine provider knowledge, attitudes and practices regarding the provision of services related to withdrawal. Areas of potential inquiry include the extent to which the use of withdrawal is explored with patients, provider perceptions of its effectiveness, and the factors that influence whether providers introduce the subject of withdrawal in health care visits. Provider and patient perceptions about the relationship between withdrawal (when used as an alternative to condoms) and STD prevention also need to be investigated. Finally, better effectiveness