The intrauterine device (IUD) provides safe1 and effective2 contraception, and 12% of married women of reproductive age worldwide use it.1 In the United States, however, only 0.8% of women using contraception use the IUD.3 Before the debacle involving the Dalkon shield, 9.5% of married, white US women using contraception used the IUD.4 After the Dalkon Shield, manufacturers withdrew most devices from the US market, and IUD use decreased.5 Physicians and the public developed a persistent fear that all IUDs cause pelvic inflammatory disease (PID) despite evidence that PID after modern copper IUD insertion occurs rarely, at a rate of 1.6 per 1000 woman-years of use.6 Reanalysis of the early studies linking IUDs and PID question their methodology and generalizability.7,8 The relative underuse of the IUD in the United States reflects these public and professional concerns. The reluctance of US physicians to recommend IUDs except in narrowly selected patients9 contributes to this underuse. Many factors contribute to this reluctance, including a lack of training in use of IUDs during residency, a fear of litigation, and a belief that the IUD creates a high risk for PID.10–12
Little is known about how obstetrician-gynecologists use the IUD in clinical practice, what their attitudes toward the IUD are, or how they select IUD candidates in their practices. In 1989, Kooiker and Scutchfield11 surveyed obstetrician-gynecologists and family and general internal medicine physicians in San Diego County, CA, shortly after the 1988 release of the Paragard T380A Intrauterine Copper Contraceptive device (Ortho Pharmaceutical Corporation, Raritan, NJ) in the United States. They asked the physicians whether they recommended this new IUD to patients and whether they planned to use it in their practices. In their sample, 40% were not recommending this IUD to any patients. Respondents with a low knowledge score about this IUD, limited experience with IUD insertion, and non–obstetrician-gynecologist specialty had a more negative attitude toward this IUD and a lower willingness to recommend it. The top two reasons given for not recommending this IUD were fear of legal liability and a belief that the IUD was not medically safe.
No surveys of IUD use by US physicians have been published in the peer-reviewed literature since the study by Kooiker and Scutchfield. Recent advances in IUD technology, exemplified by the entry of a new type of IUD into the US market—the Mirena levonorgestrel-releasing intrauterine system (Berlex Laboratories, Montville, NJ)—make our study particularly timely. Understanding how physicians feel about and use the IUD in practice is important in designing targeted educational programs for the current generation of physicians. We aimed to assess obstetrician-gynecologists' current use of the IUD in practice, their attitudes toward the IUD, and the factors they consider when selecting patients as IUD candidates. Further, we aimed to test the hypotheses that limited IUD training in residency, fear of litigation, and a belief in a strong causal link between IUDs and PID are associated with lower IUD use in practice.
MATERIALS AND METHODS
We designed a self-administered written survey to ascertain knowledge, attitudes, and practices among obstetrician-gynecologists with regard to the IUD, as well as demographic, training, and practice information. We constructed the sampling frame from the geographic listing of the membership directory of the ACOG, including all members in active practice (Fellow or Junior Fellow in Practice) who had a mailing address in the 50 United States or the District of Columbia and were not listed in the military. Starting at a random number, we selected a systematic sample of every 30th name, yielding a list of 811 names. We mailed the questionnaire to each of these obstetrician-gynecologists with a cover letter signed by the primary investigator and a postpaid return envelope. We conducted two mailings between July and September of 2000.
We tested for nonresponse bias by comparing respondents and nonrespondents on county-level demographic information linked by ZIP code to data in the Area Resource File (http://www.arfsys.com). Specifically, we looked at urban or rural practice location, county per capita income, several physician workforce variables, and variables reflecting the demographic structure of the communities served by the physicians (eg, total births and the percentage of women in the population). We had no demographic information on nonresponders from our original sampling frame, the ACOG Membership Directory. To address this issue, we compared a random sample of 100 responders and 100 nonresponders. We obtained information on sex and year of graduation from medical school in nonresponders by using the American Medical Association's Web-based physician directory (http://www.ama-assn.org). We added four to the year of medical school graduation to estimate the year of residency graduation. In analyses comparing responders with nonresponders, we used a two-sample Student t test for normally distributed continuous variables, the Wilcoxon rank-sum test for non-normal variables, and the Pearson χ2 test for categorical variables.
We performed descriptive analyses of the main outcome of interest—the reported number of IUDs inserted in the past year—and demographic and attitudinal variables. We tested for associations between the reported number of IUDs inserted in the past year and demographic and attitudinal variables. Because the main outcome had a non-normal distribution, we performed both parametric and nonparametric tests; data are reported as means when results were similar. We used a two-sample Student t test and Spearman correlation for continuous independent variables and one-way analysis of variance and the Kruskal–Wallis test for nominal independent variables. We also created a scale of conservative-to-liberal IUD candidate selection by summing responses to five variables, measured on a Likert scale, that queried respondents' attitude toward parity, history of sexually transmitted disease (STD) and PID, marital status, and monogamy. We analyzed this scale in three categories: most conservative, moderate, and most liberal. This scale had an acceptable Cronbach α value of .70. We performed all analyses using Stata 6.0 (Stata Corporation, College Station, TX).
Of the 811 ACOG fellows to whom we mailed our survey, 400 returned complete responses, 17 had their mailing returned as undeliverable, and 26 returned invalid or incomplete responses, for a final response rate of 50%. This response rate is only slightly lower than the 52% response typical of large national physician surveys13 and is higher than that reported in previous random surveys of general ACOG members, which averaged 40%.14–22 Respondents and nonrespondents did not differ in county-level demographic information on income; rural or urban mix; or physician workforce, including obstetrician-gynecologist workforce. Respondents completed residency training more recently than nonrespondents did (mean year of graduation 1984 versus 1981, P = .04) and were more likely to be female (37% versus 26%, P = .07). The difference in sex distribution was not significant after adjustment for year of graduation, and the 3-year difference in the mean year of residency graduation was small.
Among respondents, 37% were female, the mean age was 47 years, and the mean year of residency completion was 1984 (Table 1). Seventy percent were in private practice and 14% were in academic departments, and 83% were generalists. They reported seeing an average of 88 patients in the office during a normal week. These characteristics are similar to those observed in recent surveys of ACOG members.14–16,18,19,23
Most respondents (80%) reported inserting IUDs in the last year, although 79% reported inserting 10 or fewer. The mean number of IUDs inserted in the past year was seven. Almost one third (32%) of respondents inserted 1 to 4 IUDs, 31% inserted 5 to 10 IUDs, and 17% inserted more than 10 IUDs. Respondents reported that they discussed a new or different contraceptive method with an average of 15% of their patients seen in the office. When asked to estimate the proportion of their patients using contraception whom they consider to be IUD candidates, half of respondents said 15% or less. During residency, most respondents reported inserting some IUDs; 13% inserted none, 38% inserted 1 to 20, 22% inserted 21 to 50, and 27% inserted more than 50.
Respondents had a remarkably positive attitude toward the IUD in general as a contraceptive method. Ninety-eight percent agreed that the Copper T380A IUD is effective, and 95% agreed that it is safe. Most (79%) agreed that they had enough time to counsel their patients on contraceptive options, and 64% believed that their patients were receptive to learning about the IUD. However, 20% percent agreed that the IUD was an abortifacient, and 16% agreed that it would lead to lawsuits against them.
Respondents restricted IUD candidates most tightly on the basis of monogamy and PID history. When presented with hypothetical characteristics of patients, 84% agreed that a woman in a nonmonogamous relationship should not have an IUD, and 81% agreed that a woman with a history of PID should not have an IUD. About two thirds of respondents did not recommend IUDs to nulliparous women or those with a history of STDs. Respondents were least restrictive about marital status; 31% agreed that a woman who is not married should not have an IUD. We then asked them how strongly different characteristics affect their selection of IUD candidates. Most (81%) were strongly affected by monogamy status. About two thirds (68%) responded that parity status had a strong effect on their decision, and a minority felt that marital status and the patient's education level affected their decision (40% and 30%, respectively).
We asked, “By what percent do you believe that a Copper T380A IUD increases the risk of PID over 10 years?” Twenty percent responded “zero,” indicating that this minority does not believe in a causal relationship between IUDs and PID. Nearly 40% responded that the increased risk is 1% to 9%, 17% that the risk is increased 10%, and 12% that the increase in risk is greater than 10%. Thus, nearly one third of physicians responded that IUDs increase the risk of PID by 10% or more, indicating a strong belief in a long-term causal relationship.
We tested for associations between the reported number of IUDs inserted in the last year and demographic characteristics of respondents. Sex and geographic region were not associated with the reported number of IUDs inserted. Age and year of graduation of residency were highly correlated, and both were associated with IUD insertion. Younger and more recent graduates inserted more IUDs. Respondents aged 31 to 45 inserted a median of 5 IUDs, those aged 46 to 55 inserted a median of 4, and those aged 56 to 73 inserted a median of 3 (P = .04). Each practice type had similar reported median numbers of IUDs inserted, ranging from 2 to 5, except for one: Physicians in health maintenance organizations reported inserting a median of 10 IUDs, but only 8 respondents were in this group.
To test our main hypotheses about underuse of the IUD in the United States, we tested for associations between the reported number of IUDs inserted in the past year and residency training in IUDs, fear of litigation, and a belief in a causal link between IUDs and PID (Table 2). We found no association between the number of IUDs inserted during residency and the reported number inserted in the last year. We found a significant association between fear of litigation and reported number of IUDs inserted in the last year. Sixteen percent of respondents agreed that using the IUD in practice puts them at risk for litigation. These respondents reported inserting a mean of 4 IUDs, whereas those who disagreed inserted 10 (P < .001).
We found a significant overall association (P = .004) between a belief in a causal link between IUDs and PID and reported IUD use in practice when we asked the question, “By what percent do you believe that a Copper T380A IUD increases the risk of PID over 10 years of use?” Physicians who responded “zero” had a higher mean number of IUDs inserted in the past year than did those who responded “>1%” (nine versus seven, two-sample Student t test P = .008).
We developed a score of conservative-to-liberal IUD candidate selection behavior using five questions on patient characteristics (nulliparity, STD and PID history, marital status, and monogamy). On the basis of score, we categorized respondents as conservative, moderate, or liberal. Reported use of the IUD in practice was associated with this score. The conservative group reported inserting fewer IUDs in the past year than the liberal group (four versus nine, P < .001).
Most obstetrician-gynecologists insert IUDs in clinical practice, but they report inserting few per year. Although attitudes toward the safety and effectiveness of the IUD are very positive, most respondents believe that a long-term causal relationship exists between the modern copper IUD and PID. Twenty-nine percent of respondents considered this increased risk to be 10% or greater.
The evidence argues against such a conclusion. A transient sixfold increased risk exists for 21 days after insertion, after which the rate of PID decreases to 0.059 per 100 woman-years of use with the Copper T380A device.6 The evidence does not demonstrate a long-term risk of PID definitively attributable to the IUD.6 Instead, sexual behavior and resultant exposure to chlamydia and gonorrhea produce the largest attributable risk for PID24 (Shelton JD. Risk of clinical pelvic inflammatory disease attributable to an intrauterine device [Research Letter]. Lancet 2001;357:443). No study has answered the question of whether the risk of progression to PID is greater or lesser in a woman with cervicitis and a modern copper IUD than in a woman with cervicitis but no IUD.8 Nor does evidence indicate that having an IUD in place will worsen a case of PID. In contrast to current evidence that the IUD does not cause PID in the long term, manufacturer recommendations (Paragard® T380A Copper Contraceptive. Prescribing information. Raritan, NJ: Ortho Pharmaceutical Corporation, 1997) list any lifetime history of PID as a contraindication to use of a copper IUD. Similarly, the ACOG Technical Bulletin on IUDs9 describes the ideal IUD candidate as having no history of PID.
Our survey supports the hypothesis that a fear of litigation limits use of the IUD in practice among obstetrician-gynecologists. Fortunately, a majority of respondents believed that IUDs do not lead to lawsuits, an opinion supported by reviews of physician litigation experience.12 Our results do not support the hypothesis that the number of IUDs inserted in residency is associated with current reported use of the IUD in clinical practice. Rather, the association of a respondent's conservative-to-liberal score in patient selection with IUD use may indicate that the teaching of candidate selection drives physicians' future use of the IUD.
Our survey has several limitations. Because it is cross-sectional, we can find associations but not cause-and-effect relationships. Thus, respondents may have developed more liberal criteria for IUD candidate selection after inserting many IUDs and becoming comfortable with the low rate of complications rather than by developing more liberal criteria that then cause them to insert higher numbers. We also relied on respondents' recall of the number of IUDs inserted in the past year and during residency. Our results may not be generalizable to all obstetrician-gynecologists in the United States. Because respondents graduated slightly more recently from residency than did nonrespondents, our survey may overestimate the use of IUDs by practicing obstetrician-gynecologists. Finally, our survey addresses neither the decision to use an IUD from the woman's perspective nor issues of insurance coverage for contraceptives.
Our findings nevertheless have important implications for physician training and education. Because most obstetrician-gynecologists are inserting few IUDs, educational programs should target these physicians to expand their IUD use. Such programs should highlight the evidence for IUD safety and the rarity of litigation. The number of IUDs inserted in residency may be less important than the development of less conservative, more evidence-based criteria for selecting IUD candidates.
The IUD is a safe and effective method of contraception. Its broader use in the United States is limited, not because obstetrician-gynecologists believe that the IUD is ineffective, or that they lack skill or knowledge in its use, but because of a persistent belief that IUDs cause PID, which results in lawsuits. The evidence does not support these fears about modern devices, and such tight restrictions fail to weigh the competing risk of unintended pregnancy from use of less effective methods. Failure to weigh this risk has profound public health implications in a country in which 49% of pregnancies are unintended and 53% of unintended pregnancies happen after contraceptive failure or misuse.25
Many women could safely use the IUD but are not offered this contraceptive method because physicians' selection of IUD candidates is unduly restrictive. Educating physicians about the safety of IUDs may expand their use of IUDs. More women would then be offered a method of convenient, safe, and highly effective long-term contraception.
1. Rinehart W. IUDs—An update. Popul Rep B 1995:1–35.
2. Trussell J, Vaughan B. Contraceptive failure, method-related discontinuation and resumption of use: Results from the 1995 National Survey of Family Growth. Fam Plann Perspect 1999;31:64–72, 93.
3. Piccinino LJ, Mosher WD. Trends in contraceptive use in the United States: 1982–1995. Fam Plann Perspect 1998; 30:4–10, 46.
4. Westoff CF, Jones EF. Contraception and sterilization in the United States, 1965–1975. Fam Plann Perspect 1977;9:153–7.
5. Forrest JD. The end of IUD marketing in the United States: What does it mean for American women? Fam Plann Perspect 1986;18:52–7.
6. Farley TM, Rosenberg MJ, Rowe PJ, Chen JH, Meirik O. Intrauterine devices and pelvic inflammatory disease: An international perspective. Lancet 1992;339:785–8.
7. Kessel E. Pelvic inflammatory disease with intrauterine device use: A reassessment. Fertil Steril 1989;51:1–11.
8. Grimes DA. Intrauterine device and upper-genital-tract infection. Lancet 2000;356:1013–9.
9. American College of Obstetricians and Gynecologists. The intrauterine device. ACOG technical bulletin no. 164. Washington, DC: American College of Obstetricians and Gynecologists, 1992.
10. Westhoff C. The IUD in evolution. Obstet Gynecol Surv 1996;51:S20–4.
11. Kooiker CH, Scutchfield FD. Barriers to prescribing the Copper T 380A intrauterine device by physicians. West J Med 1990;153:279–82.
12. Westhoff C, Marks F, Rosenfield A, Roig A, Nicholas A. Clinical and legal factors influencing IUD use in the United States. In: Bardin C, Mishell D, eds. Proceedings from the Fourth International Conference on IUDs. Boston: Butterworth-Heinemann, 1994:100–8.
13. Cummings SM, Savitz LA, Konrad T. Reported response rates to mailed physician questionnaires. Health Serv Res 2001;36:1347–55.
14. Erickson K, Schmidt L, Santesso DL, Schulkin J, Gregory K, Hobel C. Obstetrician-gynecologists' knowledge and training about antenatal corticosteroids. Obstet Gynecol 2001;97:140–6.
15. Power ML, Holzman GB, Schulkin J. Knowledge and clinical practice regarding folic acid among obstetrician-gynecologists. Obstet Gynecol 2000;95:895–8.
16. Diekman ST, Floyd RL, Decoufle P, Schulkin J, Ebrahim SH, Sokol RJ. A survey of obstetrician-gynecologists on their patients' alcohol use during pregnancy. Obstet Gynecol 2000;95:756–63.
17. Wilkins-Haug L, Hill LD, Power ML, Holzman GB, Schulkin J. Gynecologists' training, knowledge, and experiences in genetics: A survey. Obstet Gynecol 2000;95:421–4.
18. Power ML, Holzman GB, Schulkin J. Knowledge and clinical practice regarding calcium nutrition among obstetrician-gynecologists. Obstet Gynecol 1999;94:421–6.
19. Wilkins-Haug L, Hill L, Schmidt L, Holzman GB, Schulkin J. Genetics in obstetricians' offices: A survey study. Obstet Gynecol 1999;93:642–7.
20. Horan DL, Chapin J, Klein L, Schmidt LA, Schulkin J. Domestic violence screening practices of obstetrician-gynecologists. Obstet Gynecol 1998;92:785–9.
21. Scroggs JA, Griffin LP, Bayerl M, Schulkin J. Obstetrician-gynecologists as primary care physicians: The perspectives of health maintenance organization medical directors and obstetrician-gynecologists. Obstet Gynecol 1997;90:291–5.
22. Schmidt LA, Greenberg BD, Holzman GB, Schulkin J. Treatment of depression by obstetrician-gynecologists: A survey study. Obstet Gynecol 1997;90:296–300.
23. American College of Obstetricians and Gynecologists. Profile of Ob-gyn Practice, 1991–1998. Washington, DC: American College of Obstetricians and Gynecologists, 2000.
24. Lee NC, Rubin GL, Borucki R. The intrauterine device and pelvic inflammatory disease revisited: New results from the Women's Health Study. Obstet Gynecol 1988; 72:1–6.
25. Henshaw SK. Unintended pregnancy in the United States. Fam Plann Perspect 1998;30:24–9.