- Email: [email protected]
Determinants of intrauterine contraception provision among US family physicians: a national survey of knowledge, attitudes and practice
Contraception 83 (2011) 472 – 478
Original research article
Determinants of intrauterine contraception provision among US family physicians: a national survey of knowledge, attitudes and practice☆,☆☆ Susan E. Rubina,⁎, Jason Fletchera , Tara Steina , Penina Segall-Gutierrezb , Marji Golda b
a Department of Family and Social Medicine, Albert Einstein College of Medicine, Bronx, NY 10461, USA Department of Obstetrics and Gynecology, Keck School of Medicine, University of Southern California, Los Angeles, CA 90033, USA Received 23 March 2010; revised 12 October 2010; accepted 13 October 2010
Abstract Background: Poor contraception adherence contributes to unintended pregnancy. Intrauterine contraception (IUC) is user-independent thus adherence is not an issue, yet few US women use IUC. We compared family physicians (FPs) who do and do not insert IUC in order to ascertain determinants of inserting IUC. Study design: We surveyed 3500 US FPs. The primary outcome variable was whether a physician inserts IUC in their current clinical practice. We also sought to describe their clinical practice with IUC insertions. Results: FPs who insert IUC had better knowledge about IUC (adjusted OR 1.85, 95% CI 1.32–2.60), more comfort discussing IUC (adjusted OR 2.35, 95% CI 1.30–4.27), and were more likely to believe their patients are receptive to discussing IUC (adjusted OR 2.96, 95% CI 2.03–4.32). The more IUC inserted during residency, the more likely to insert currently (adjusted OR 1.44, 95% CI 1.12–1.84). Only 24% of respondents inserted IUC in the prior 12 months. Conclusions: US FPs have training and knowledge gaps, as well as attitudes, that result in missed opportunities to discuss and provide IUC for all eligible patients. © 2011 Elsevier Inc. All rights reserved. Keywords: Intrauterine contraception; Intrauterine device; Clinician knowledge; Survey; Insertion; Family physician
1. Introduction Although contraceptive prevalence in the US is high [1,2] and most methods are nearly 100% effective when used perfectly [3,4], unintended pregnancy remains a significant public health issue. Healthy People 2010 includes the objectives that 70% of pregnancies are intended and that only 7% of women will experience pregnancy despite using a reversible contraceptive method [5]. Yet, at least half of US women will experience an unintended pregnancy by age 45 ☆
This project was support by an anonymous foundation that wishes to remain anonymous. Dr. Rubin's salary was supported by a fellowship grant from the New York State Empire Clinical Research Investigator Program. ☆☆ A poster presentation of portions of this data was presented at the Association of Reproductive Health Professionals National Conference in Los Angeles, CA, USA, Sept 2009, and the Society of Teachers of Family Medicine North East Regional Meeting in Rye Brook, NY, Oct 2009. ⁎ Corresponding author. Tel.: +1 718 430 2752; fax: +1 718 430 8645. E-mail address: [email protected] (S.E. Rubin). 0010-7824/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.contraception.2010.10.003
[6] and over 50% of women receiving abortion services in the US were using contraception at the time of their unintended pregnancy [7]. Improper and inconsistent use of contraception contributes to the rate of unintended pregnancy [8,9]. Since nearly one quarter of US women seeking private family planning care see a family physician (FP) [8] and the average US woman spends 30 years trying to avoid pregnancy [10], it is critical that FPs counsel and provide all appropriate contraceptive methods. Many of the factors associated with poor contraceptive adherence (forgetting, method unavailability, misunderstanding of correct use) [11–13] are obviated by userindependent methods. Intrauterine contraception (IUC) is user independent, highly reliable [4], safe [14–16], and costeffective [17,18]. There are few contraindications to IUC use [19], yet IUC use in contracepting US women is only 5% [1] as compared to 15% worldwide [20]. Increasing IUC use has the potential to reduce unintended pregnancy rates [21]. Two types of IUC are currently available in the US, the Copper T
S.E. Rubin et al. / Contraception 83 (2011) 472–478
380A (Teva Pharmaceuticals Inc, marketed as ParaGard®), and the levonorgestrel-releasing intrauterine system (Bayer HealthCare Pharmaceuticals, marketed as Mirena®). The issues behind low IUC utilization are manifold and include patient, clinician and health systems factors. Since clinician recommendation and insertion are necessary for IUC provision, elucidating clinician factors is vital. We know little, however, about FPs practices with IUC. One national survey reports that 99% of FP respondents “dispense, prescribe or recommend” oral contraceptives, but only 39% do so with IUC [8]. A survey of Indian Health Service providers found that all providers perform contraception counseling and have a generally favorable attitude towards IUC, but FPs have less IUC knowledge and experience than obstetriciangynecologists [22]. We do not know of any prior published study identifying possible determinants of whether or not FPs insert IUC. In order to understand potential facilitators and barriers to FPs inserting IUC, we sought (1) to compare the knowledge, attitude and practice of FPs who insert IUC to those who do not, and (2) to describe FPs practice with IUC insertion. We hypothesized that FPs who insert IUC had more insertion training in residency [22,23], graduated from residency more recently, are more likely to be female, perform cervical cancer screening (Pap smears) in their current clinical practice, do not believe it takes more time to counsel about IUC and have increased knowledge of, comfort counseling about, and positive beliefs towards IUC.
473
medium effect size (d=.3, one third of a standard deviation difference between groups) in knowledge, attitude and practice behaviors between inserters and non-inserters. A post hoc multivariate logistic regression was conducted with members of the sample with data on all predictors (completecase analysis). The retained sample size and distribution of variables provided sufficient statistical power for reliable parameter estimates and significance tests. 2.3. Survey instrument
We surveyed a random national sample of FPs identified from the American Medical Association's Physician Master File. This database is intended to include all US physicians [24,25]. Our sample of 3500 FPs was selected by Medical Marketing Services (a database licensee of the American Medical Association) computer-generated algorithm from a total population of approximately 95,000 FPs. Exclusion criteria included Puerto Rican practice sites and classification as a resident, administrator or retiree.
The majority of our questions were adapted from those used in prior survey studies exploring clinicians' knowledge, attitude and clinical practice with IUC [22,23]. We developed additional questions related to FPs and to the levonorgestrel-releasing intrauterine system. All questions were piloted on a sample of FPs at our home institutions, and the survey was modified appropriately. The final 45-item self-administered survey contained demographic questions including training in and current provision of reproductive health services. We measured training in IUC insertion by asking how many IUCs respondents inserted during residency (none, 1–9, 10–19, 20–49 or more than 49). Likelihood to recommend IUC was measured by asking “How likely are you to recommend IUC to a woman with each of the following characteristics?” followed by eight unique patient scenarios, none of which precludes IUC use. These scenarios included nulliparity, distant history of a sexually transmitted infection (STI) or pelvic inflammatory disease (PID), unmarried status, nonmonogamous sexual relationship, abnormal Pap with colposcopy pending, age younger than 20 years and history of ectopic pregnancy. Responses were recorded on a five-point Likert scale. The four knowledge questions were derived from evidence-based literature and package inserts [4,26]. Questions assessing attitudes and beliefs covered: comfort in discussing IUC with patients (very comfortable to very uncomfortable); perception of patients' receptivity to learning about IUC (very receptive to not at all receptive); IUC safety (true, false, unsure); efficacy (very effective to very ineffective) and, as compared to other contraceptive methods, time needed to discuss IUC with patients (more, same or less). For those respondents who indicated they had inserted IUC in the prior 12 months, we asked about the frequency and type(s) of IUC(s) they inserted and their practice with STI testing, cervical cancer screening and pregnancy evaluation prior to insertion.
2.2. Sample size determination
2.4. Data collection
We planned to compare those physicians who had inserted at least one IUC in the past 12 months called “inserters” and those who had not inserted IUC in the past 12 months called “non-inserters.” We speculated that the inserters and non-inserters would be represented in a 1:4 ratio. A priori power analyses indicated that 600 responders would provide adequate power to detect differences of small/
In May 2008, we mailed 3500 FPs a cover letter, survey, $1 incentive and business reply envelope. FPs could respond either via mail or secure Internet survey site. Three weeks later, we sent a follow-up reminder mailing with a replacement survey and another $1 incentive [27,28]. FPs who did not respond 6 weeks after the second mailing were considered nonresponders.
2. Materials and methods The Institutional Review Board of Montefiore Medical Center in the Bronx, NY, and the University of Southern California in Los Angeles approved this project. 2.1. Study population and eligibility
474
S.E. Rubin et al. / Contraception 83 (2011) 472–478
Using information from the Master File list, we compared responders versus nonresponders on variables including age, years since residency training, board certification status, urban vs. nonurban clinical practice site (based on metropolitan statistical area [MSA] codes) and gender.
2.5. Data analysis Primary analyses compared responses between inserters and non-inserters using t-tests for continuous data and chisquare tests for categorical variables. In order to calculate a “likelihood of recommending IUC” variable, we averaged responses to the eight individual questions into a single summed score. Inserter and non-inserter group means were compared. Responses to the four knowledge items were summed into a single knowledge score; group means were calculated and compared. Attitude and belief questions were each analyzed separately. We generated descriptive statistics for inserter practices. Variables that were significantly related to inserter group in bivariate analyses were included as predictors for a post hoc logistic regression model predicting inserter status. The logistic regression was conducted as a complete-case analysis using list-wise exclusion for missing data since missing data did not meet the requirements of missing at random required for imputation. The selected predictors were organized into four blocks: nonmodifiable provider characteristics, IUC beliefs, knowledge and clinical experience. We sequentially entered blocks in the regression model to assess additive effects of each predictor block. The block with nonmodifiable provider characteristics variables was the initial block entered. Entry order for the remaining blocks was determined empirically using model fit indices. At each stage in the model, the block that yielded the greatest contribution to the model was entered. The results reported are for the full model including all predictors. Surveys were included in our bivariate analysis if the respondent completed at least 12 of the 45 items. All data analyses were performed using SPSS version 15.0 (SPSS, Chicago, IL, USA).
3. Results Of the 3500 FPs invited to participate, 973 surveys were returned. We received 869 valid completed surveys, 53 incomplete surveys, 51 returned incentives and/or blank surveys and 45 invalid addresses, and two FPs were deceased. This yields a final response rate of 25%. Only 42 FPs responded via the Internet. Twenty-four percent (n=199) of respondents inserted at least one IUC in the prior 12 months. Responders and non-responders did not differ in terms of age, years since residency training, board certification status or population size of their practice setting. A larger proportion of responders than nonresponders were female (38.7% vs. 31.2%, pb.001). Bivariate analyses comparing inserters and non-inserters were conducted on data from all responders (n=869). Logistic regression analyses was conducted on data from responders with complete data on the selected predictors (n=460). The sample retained for regression analysis had a significantly greater proportion of inserters (29.8%) than the full sample (23.9%). 3.1. Inserters versus non-inserters As Tables 1–3 demonstrate, in unadjusted bivariate analyses, the groups differed significantly on every hypothesized factor except perceived time needed to counsel
Table 1 Demographics, professional background and practice variablesa Total
Female Number of IUDs inserted during residency
None 1–9 10 or more
Years since completed residency training In an average week, % patients female between 15–44 years old In those women, % on whom perform Pap smears In those women, % with whom discuss contraception
Inserters
Non-inserters
p
Adjusted OR (CI)b
35.4 44.3 35.1 20.6
.002 b.001
1.22 (.69–2.15) 1.44 (1.12–1.84)
Mean
(S.D.)
p
Adjusted OR (CI)
9.4 21.9
15.8 31.4
9.7 19.0
b.001 b.001
.997 (.97–1.03) 1.01 (1.00–1.02)
67.5
28.5
45.5
34.6
b.001
1.01 (1.00–1.02)
68.2
30.6
47.6
36.0
b.001
1.00 (.99–1.01)
n
(%)
n
(%)
n
(%)
315 313 307 191
38.6 38.6 37.9 23.6
91 32 89 68
48.1 16.9 47.1 36.0
214 262 208 122
Mean
(S.D.)
Mean
(S.D.)
14.8 33.7
9.1 20.2
13.1 41.1
50.8
34.5
52.7
35.9
Values in bold italics indicate those results that are statistically significant. IUD, intrauterine device. a Percent calculated for number of respondents to each individual question. b Adjusted odds ratios reported are from regression model containing all predictors — gender, years since completing residency, % of practice that includes women of reproductive age, % of patients on whom perform Pap smears, % of patients with whom discussed contraception, number of IUDs inserted during residency, comfort discussing IUC, belief about patients' receptivity to learning about IUC, belief that IUC is effective, time required to discuss IUC, likelihood to recommend IUC and knowledge score.
S.E. Rubin et al. / Contraception 83 (2011) 472–478
475
Table 2 Attitudes, beliefs and likelihood to recommend IUCa Inserter
Very/somewhat comfortable discussing IUDs Believe patients are very or somewhat receptive to discussing IUDs Believe IUDs are safe
Believe IUDs are effective (four-point Likert scale; 1=Very Effective to 4=Very Ineffective) Likelihood to recommend IUC (eight items on a five-point Likert scale; 1=Very Unlikely to 5=Very Likely) As compared to other contraception methods, time needed to discuss IUC with patients (three-point Likert; 1=More Time to 3=Less Time)
Noninserter
p
Adjusted OR (CI)c
n
(%)
n
(%)
181 167 189
95.8 89.3 96.9
476 323 530
79.2 54.8 86.0
b.001 b.001 b.001b
2.35 (1.30–4.27) 2.96 (2.03–4.32)
Mean
S.D.
Mean
S.D.
p
Adjusted OR (CI)
1.07
0.26
1.27
0.57
b.001
1.48 (.58–3.80)
2.72
0.86
2.27
0.79
b.001
.87 (.63–1.20)
1.55
0.49
1.52
0.54
NS
b
a
Percent calculated for number of respondents to each individual question. Variables not included in the logistic model. c Adjusted odds ratios reported are from regression model containing all predictors — gender, years since completing residency, % of practice that includes women of reproductive age, % of patients on whom perform Pap smears, % of patients with whom discussed contraception, number of IUDs inserted during residency, comfort discussing IUC, belief about patients' receptivity to learning about IUC, belief that IUC is effective, time required to discuss IUC, likelihood to recommend IUC and knowledge score. b
patients about IUC. Reported adjusted odds ratios are from the logistic model containing all predictors.
that inserters were statistically more likely than non-inserters to recommend an IUC (2.72 vs. 2.27) (Table 2).
3.1.1. Demographics and practice factors In bivariate analyses, many variables were correlated with inserting IUC, including female gender, inserting IUC during residency, completing residency more recently, caring for more women of reproductive age, performing Pap smears and discussing contraception more frequently. However, in multivariate analysis, only inserting IUC during residency remained a significant predictor: an increase of one category in number of IUCs inserted during residency (0, 1–9, 10+) corresponded to a 1.44 increase in the likelihood of being an inserter (Table 1).
3.1.3. Knowledge On the four-item summed knowledge score, inserters scored .5 point higher than noninserters (3.31 vs. 2.82); this difference is significant in both bivariate and multivariate analyses. Table 3 illustrates responses to individual questions. 3.1.4. Attitudes and beliefs As Table 2 illustrates, while both inserters and noninserters had a positive attitude toward IUC safety and efficacy, inserters rated them as safer and more effective. In our unadjusted model, the absolute difference between inserters and noninserters' belief that currently available IUCs are safe is 10.9%; the difference between groups in feeling very/somewhat comfortable discussing IUC with patients was 16.6%. Eighty-nine point three percent (89.3%) of inserters as compared to 54.8% of noninserters believe
3.1.2. Likelihood of recommending IUC Comparison of bivariate mean composite scores (measured on a five-point Likert scale) from the eight separate patient scenarios, none of which precluded IUC use, showed Table 3 Knowledge about IUCa
Inserter
Knowledge sum scale (all four items)
Non-inserter
Mean
S.D.
Mean
S.D.
3.31
0.74
2.82
0.88
Correct answer:
n
(%)
n
(%)
IUDs do not increase the long-term risk of cervical dysplasia IUDs do not have to be inserted when a woman is menstruating IUDs do not increase the long-term risk of PID IUDs can be used in women with no history of previous pregnancy
172 170 113 162
88.7 86.7 58.2 84.8
430 377 172 441
69.7 60.5 27.6 71.6
p
Adjusted OR (CI)b
b.001
1.85 (1.32–2.60)
PID, pelvic inflammatory disease. a Percent calculated for number of respondents to each individual question. b Adjusted odds ratios reported are from regression model containing all predictors - gender, years since completing residency, % of practice that includes women of reproductive age, % of patients on whom perform Pap smears, % of patients with whom discussed contraception, number of IUDs inserted during residency, comfort discussing IUC, belief about patients' receptivity to learning about IUC, belief that IUC is effective, time required to discuss IUC, likelihood to recommend IUC and knowledge score.
476
S.E. Rubin et al. / Contraception 83 (2011) 472–478
Table 4 Practice patterns of clinicians who insert intrauterine contraception (n=199)a
# IUDs inserted in past 12 months (range 1–50)
Type(s) of IUC inserted
Frequency of antibiotic prescription at time of IUC insertion Maximum amount time from last STI screen
If no prior STI screen, would you… If last normal Pap smear were 2 years ago, would you… To rule out pregnancy prior to inserting, do you check…
CuT 380A only Levonorgestrel IUS only Both Always Sometimes Rarely Never 1 month 3 months 6 months 9 months N9 months Same day screen and insert Screen, follow-up for insertion Other Same day screen and insert Screen, follow-up for insertion Other Urine hCG only Serum hCG only Both urine and serum hCG Other
Mean
S.D.
9.36
9.7
n 41 59 85 6 4 31 146 30 46 28 6 71 56 104 20 81 90 16 147 20 9 9
(%) 22.2 31.9 45.9 3.2 2.1 16.6 78.1 16.6 25.4 15.5 3.3 39.2 31.1 57.8 11.1 43.3 48.1 8.5 79.5 10.8 4.9 4.9
STI, sexually transmitted infection; hCG, human chorionic gonadotropin. a Percent calculated for number of respondents to each individual question.
their patients are receptive to discussing IUC. Significant predictors in the multivariate model included greater odds of feeling very/somewhat comfortable discussing IUC (OR 2.35, 1.30–4.27) and an increasing belief that their patients are receptive to IUC (OR 2.96, CI 2.03–4.32). Belief that IUC is safe was not included in the logistic model since it was not a significant predictor of inserter status in the reduced sample. 3.2. Inserters practice patterns Inserters reported that 41.1% of their patients were women of reproductive age; they performed Pap smears in 67.5% and discussed contraception with 68.2% of these patients. Despite discussing contraception with a median of 1140 female patients annually, they inserted a mean of just nine IUC in the prior 12 months. Only 22.6% (n=45) placed more than 10 devices in the previous 12 months. The inserters were generally up-to-date with their IUC insertion practices. For example, 31.1% would screen for STIs and insert IUC at the same visit (Table 4). 4. Discussion US FPs have training and knowledge gaps, as well as attitudes, that result in missed opportunities to discuss and provide IUC to all eligible patients. FPs who insert IUC are using up-to-date guidelines, yet they insert few devices per
year. Further work needs to be done to understand why FPs do not insert more IUC and to establish effective programs to increase insertion rates. It is encouraging that the factors we found differentiating inserters from noninserters are potentially modifiable. Our results support previous findings that an important predictor of being an inserter is IUC insertion experience during residency [29]. However, FPs lack adequate training in IUC insertion [30,31]. In 1995, 66% of family medicine chief residents had no clinical experience with IUC insertion and only 6% had managed more than six cases during residency [32]. In order to increase IUC use in the US, professional organizations have called for an approach based on the European model where IUC insertion training and education is routinely offered as part of the medical curriculum and for established practitioners [33]. In our sample, the more recent graduates are more likely to insert, indicating that FP residencies may be improving IUC training. It would be informative to have updated data from family medicine residency programs. The other factors associated with inserting in our multivariate model — knowledge, comfort discussing and believing patients are receptive to IUC — are influenced throughout a FPs career. Therefore, these factors are more complex to assess and modify than gaps in residency training. Since we collected data at one point in time, it is uncertain whether, for example, increased knowledge increases the likelihood that a FP inserts IUC, or if inserting IUC leads to better knowledge. Yet, this association between inserting IUC and contraception knowledge is consistent with prior studies [34]. It is also unclear how different clinical practice settings affect these factors. While we found that FPs with certain personal practice characteristics were more likely to insert, many practice setting and systems factors we did not measure including access to other providers for insertion, cost of the device, reimbursement, insurance coverage and limited scope of practice may also affect the ability of a FP to discuss and insert IUC. The most effective methods to change practicing physicians' knowledge, attitudes and practice have not yet been determined and are influenced by a myriad of practice and personal factors [35]. Increasing the proportion of FPs who insert IUC is particularly challenging because in addition to changing knowledge and attitudes, it also involves integrating a new procedure into practice [36]. It may be prudent to focus future efforts on FPs who already provide reproductive health care, including Pap smears and endometrial biopsies. Over half (52.5%) of FPs perform endometrial biopsy [37], the diagnostic procedure most akin to IUC insertion. Further studies of FPs who already provide reproductive health care could clarify details of why some are doing IUC insertions and others are not. Training initiatives could establish how easily FPs with endometrial biopsy skills can learn IUC insertions. Finally, competency in insertion and believing that IUC is safe and effective does not automatically equate with
S.E. Rubin et al. / Contraception 83 (2011) 472–478
inserting IUC. The vast majority of respondents from a national survey of obstetrician-gynecologists agree IUC is safe and effective, yet respondents are unduly restrictive in their eligibility criteria and insert a mean of seven devices annually [23]. A 2007 study in California found that obstetrician-gynecologists are familiar with, have a positive attitude towards and have good knowledge of IUC, but insertion rates are relatively low [38]. Physicians trained to insert IUC may provide it infrequently because they ascribe to out-of-date eligibility criteria [21,39,40]. For example, despite the fact that IUC is an option for women with a history of ectopic pregnancy [19,41], for nulligravid women and adolescents [19,42,43] and for women with a history pelvic inflammatory disease (PID) or STI more than 3 months prior to insertion [14,15,44], many of our respondents were neutral or unlikely to recommend IUC for a women with any one of these issues in their medical history. Our survey has a number of limitations. First, our response rate is low, though it does fall within the range of published physician survey studies [45–47]. Since a topic's perceived relevance affects response rate [48], our response rate may reflect nonresponders' perception that IUC does not pertain to their practice. Thus, our result showing that 24% of FPs insert IUC is likely higher than the actual percentage of FPs who insert. Second, by conducting the logistic regression as a complete-case analysis, there was a substantial drop in the sample size. This reduced the statistical power of the model, and yielded a potentially biased sample. Support for the complete-case approach can be found in the consistency of bivariate associations in the full and subgroup sample; all but one of the bivariate associations found to be significant in the full sample were also significant in the reduced sample. Therefore, despite potential bias in the sample, results from the multivariate analysis provide important information about identifying differences between inserters and non-inserters when controlling for other variables in the model. Third, as with all survey studies, a response bias is of concern. However, due to the relative homogeneity of physicians, it has been suggested that response bias is less concerning as compared to layperson surveys [28]. Finally, we analyzed our data comparing inserters and non-inserters. The non-inserters are likely heterogeneous in the frequency with which they counsel about IUC. We did not measure knowledge, attitudes and beliefs that affect whether a non-inserter is more or less likely to discuss and recommend IUC. Studies examining barriers and facilitators of IUC counseling, not simply insertion, at both the individual provider and larger systems level would be informative. Despite these limitations, our study demonstrates that significant knowledge and training gaps, as well as attitudes, prevent FPs from offering IUC to all eligible patients. More IUC education and insertion training in residency, as well as directing interventions at FPs who already provide reproductive health services, could increase the proportion of women who have access to IUC. Increasing use of this
477
effective, user-independent contraceptive method could ultimately reduce the rate of unintended pregnancy, bringing us closer to the objectives of Healthy People 2010.
References [1] Mosher W, Jones J. Use of contraception in the United States: 19822008. Natl Cent Health Stat Vital Health Stat 2010;23:84. [2] Bensyl DM, Iuliano DA, Carter M, et al. Contraceptive use—United States and territories, Behavioral Risk Factor Surveillance System, 2002. [erratum appears in MMWR Morb Mortal Wkly Rep. 2005;54:1214] Morb Mortal Wkly Rep Surveill Summ 2005;54:1–72. [3] Trussell J. Contraceptive failure in the United States. Contraception 2004;70:89–96. [4] Hatcher R, Trussell J, Nelson A, et al. Contraceptive Technology. 19th edition. New York: Ardent Media; 2008. [5] Healthy People 2010. http://www.healthypeople.gov/Document/html/ volume1/09Familyhtm [cited 2010 March 1]; Available from: http:// www.healthypeople.gov/Document/html/volume1/09Family.htm. [6] Jones RK, Singh S, Finer LB. Repeat abortion in the United States. Occasional Report. New York: Guttmacher Institute; 2006. [7] Jones RK, Darroch JE, Henshaw SK, et al. Contraceptive use among U.S. women having abortions in 2000-2001. Perspect Sex Reprod Health 2002;34:294–303. [8] Landry DJ, Wei J, Frost JJ. Public and private providers' involvement in improving their patients' contraceptive use. Contraception 2008;78:42–51. [9] Moreau C, Cleland K, Trussell J. Contraceptive discontinuation attributed to method dissatisfaction in the United States. Contraception 2007;76:267–72. [10] Frost JJ, Darroch JE, Remez L. Improving contraceptive use in the United States. Issues Brief (Alan Guttmacher Inst) 2008:1–8. [11] Nelson AL, Westhoff C, Schnare SM. Real-world patterns of prescription refills for branded hormonal contraceptives: a reflection of contraceptive discontinuation. Obstet Gynecol 2008;112:782–7. [12] Smith JD, Oakley D. Why do women miss oral contraceptive pills? An analysis of women's self-described reasons for missed pills. J Midwifery Women's Health 2005;50:380–3. [13] Rosenberg MJ, Waugh MS, Meehan TE. Use and misuse of oral contraceptives: risk indicators for poor pill taking and discontinuation. Contraception 1995;51:283–8. [14] Shelton JD. Risk of clinical pelvic inflammatory disease attributable to an intrauterine device. Lancet 2001;357:443. [15] Grimes DA. Intrauterine device and upper-genital-tract infection. Lancet 2000;356:1013–9. [16] Grimes DA, Lopez LM, Manion C, et al. Cochrane systematic reviews of IUD trials: lessons learned. Contraception 2007;75(6 Suppl):S55–9. [17] Trussell J, Lalla AM, Doan QV, et al. Cost effectiveness of contraceptives in the United States. Contraception 2009;79:5–14. [18] Chiou CF, Trussell J, Reyes E, et al. Economic analysis of contraceptives for women. Contraception 2003;68:3–10. [19] World Health Organization. Medical eligibility criteria for contraceptive use - 3rd edition. Intrauterine devices (IUDs); 2004. with 2008 update [cited 2010 June 22]; Available from: http://www.who.int/ reproductivehealth/publications/family_planning/9241562668index/ en/index.html. [20] Population Reference Bureau. Family Planning Worldwide, 2008 Data Sheet; 2008 [cited 2010 February 10]; Available from: http://www.prb. org/pdf08/fpds08.pdf. [21] Speidel JJ, Cynthia CH, Wayne CS. The potential of long-acting reversible contraception to decrease unintended pregnancy. Contraception 2008;78:197–200. [22] Espey E, Ogburn T, Espey D, et al. IUD-related knowledge, attitudes and practices among Navajo Area Indian Health Service providers. Perspect Sex Reprod Health 2003;35:169–73.
478
S.E. Rubin et al. / Contraception 83 (2011) 472–478
[23] Stanwood NL, Garrett JM, Konrad TR. Obstetrician-gynecologists and the intrauterine device: a survey of attitudes and practice. Obstet Gynecol 2002;99:275–80. [24] Konrad TR, Slifkin RT, Stevens C, et al. Using the American Medical Association physician masterfile to measure physician supply in small towns. J Rural Health 2000;16:162–7. [25] Williams PT, Whitcomb M, Kessler J. Quality of the family physician component of AMA Masterfile. J Am Board Fam Pract 1996;9:94–9. [26] Paragard FDA package insert. 2009 [cited 2009 June 13]; Available from: http://www.paragard.com/health_care_professional/global/pdf/ Prescribing-Info.pdf. [27] Edwards P, Roberts I, Clarke M, et al. Increasing response rates to postal questionnaires: systematic review. Bmj 2002;324(7347):1183. [28] Kellerman SE, Herold J. Physician response to surveys. A review of the literature. Am J Prev Med 2001;20:61–7. [29] Harper CC, Blum M, de Bocanegra HT, et al. Challenges in translating evidence to practice: the provision of intrauterine contraception. Obstet Gynecol 2008;111:1359–69. [30] Cheng D. Family planning training in Maryland Family Practice and Obstetrics/Gynecology residency programs. J Am Med Womens Assoc 1999;54(4):208–10. [31] Schreiber CA, Harwood BJ, Switzer GE, et al. Training and attitudes about contraceptive management across primary care specialties: a survey of graduating residents. Contraception 2006;73:618–22. [32] Steinauer JE, DePineres T, Robert AM, et al. Training family practice residents in abortion and other reproductive health care: a nationwide survey. [see comment]Fam Plann Perspect 1997;29:222–7. [33] Sonfield A. Popularity disparity: attitudes about the IUD in Europe and the United States. Guttmacher Policy Rev 2007;10:19–24. [34] Dehlendorf C, Levy K, Ruskin R, et al. Health care providers' knowledge about contraceptive evidence: a barrier to quality family planning care? Contraception 2010;81:292–8. [35] Cabana MD, Rand CS, Powe NR, et al. Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999;282:1458–65. [36] Litaker D, Ruhe M, Flocke S. Making sense of primary care practices' capacity for change. Transl Res 2008;152:245–53.
[37] American Academy of Family Physicians, Practice Profile II Survey. Performance of diagnostic procedures in family physicians. Table #64. American Academy of Family Physicians, Practice Profile II Survey. July 2008 [cited 2010 January 9]; Available from www.aafp.org/ online/en/home/aboutus/specialty/facts/64.html. [38] Postlethwaite D, Shaber R, Mancuso V, et al. Intrauterine contraception: evaluation of clinician practice patterns in Kaiser Permanente Northern California. Contraception 2007;75:177–84. [39] Madden T, Allsworth JE, Hladky KJ, et al. Intrauterine contraception in Saint Louis: a survey of obstetrician and gynecologists' knowledge and attitudes. Contraception 2010;81:112–6. [40] Rubin SE, Fletcher J, Stein T, et al. Underuse of the IUD in contraceptive care and training. Fam Med 2010;42:387–8. [41] Skjeldestad FE. How effectively do copper intrauterine devices prevent ectopic pregnancy? [Article] Acta Obstet Gynecol Scand 1997;76: 684–90. [42] Hubacher D, Lara-Ricalde R, Taylor DJ, et al. Use of copper intrauterine devices and the risk of tubal infertility among nulligravid women. N Engl J Med 2001;345:7. [43] National Institute for Health and Clinical Excellence. Long-acting reversible contraception. 2005 [cited 2010 March 5]; Available from http://www.nice.org.uk/Guidance/CG30/NiceGuidance/pdf/English. [44] Steen R, Shapiro K, Steen R, et al. Intrauterine contraceptive devices and risk of pelvic inflammatory disease: standard of care in high STI prevalence settings. Reprod Health Matters 2004;12:136–43. [45] Cummings SM, Savitz LA, Konrad TR. Reported response rates to mailed physician questionnaires. Health Serv Res 2001;35: 1347–55. [46] Moeller MP, White KR, Shisler L. Primary care physicians' knowledge, attitudes, and practices related to newborn hearing screening. Pediatrics 2006;118:1357–70. [47] Astin JA, Soeken K, Sierpina VS, et al. Barriers to the integration of psychosocial factors in medicine: results of a national survey of physicians. J Am Board Fam Med 2006;19:557–65. [48] Dillman D. Constructing the questionnaire. Mail and Internet Surveys: The Tailored Design Method, 2nd ed. 2000. p. 87.
Original research article
Determinants of intrauterine contraception provision among US family physicians: a national survey of knowledge, attitudes and practice☆,☆☆ Susan E. Rubina,⁎, Jason Fletchera , Tara Steina , Penina Segall-Gutierrezb , Marji Golda b
a Department of Family and Social Medicine, Albert Einstein College of Medicine, Bronx, NY 10461, USA Department of Obstetrics and Gynecology, Keck School of Medicine, University of Southern California, Los Angeles, CA 90033, USA Received 23 March 2010; revised 12 October 2010; accepted 13 October 2010
Abstract Background: Poor contraception adherence contributes to unintended pregnancy. Intrauterine contraception (IUC) is user-independent thus adherence is not an issue, yet few US women use IUC. We compared family physicians (FPs) who do and do not insert IUC in order to ascertain determinants of inserting IUC. Study design: We surveyed 3500 US FPs. The primary outcome variable was whether a physician inserts IUC in their current clinical practice. We also sought to describe their clinical practice with IUC insertions. Results: FPs who insert IUC had better knowledge about IUC (adjusted OR 1.85, 95% CI 1.32–2.60), more comfort discussing IUC (adjusted OR 2.35, 95% CI 1.30–4.27), and were more likely to believe their patients are receptive to discussing IUC (adjusted OR 2.96, 95% CI 2.03–4.32). The more IUC inserted during residency, the more likely to insert currently (adjusted OR 1.44, 95% CI 1.12–1.84). Only 24% of respondents inserted IUC in the prior 12 months. Conclusions: US FPs have training and knowledge gaps, as well as attitudes, that result in missed opportunities to discuss and provide IUC for all eligible patients. © 2011 Elsevier Inc. All rights reserved. Keywords: Intrauterine contraception; Intrauterine device; Clinician knowledge; Survey; Insertion; Family physician
1. Introduction Although contraceptive prevalence in the US is high [1,2] and most methods are nearly 100% effective when used perfectly [3,4], unintended pregnancy remains a significant public health issue. Healthy People 2010 includes the objectives that 70% of pregnancies are intended and that only 7% of women will experience pregnancy despite using a reversible contraceptive method [5]. Yet, at least half of US women will experience an unintended pregnancy by age 45 ☆
This project was support by an anonymous foundation that wishes to remain anonymous. Dr. Rubin's salary was supported by a fellowship grant from the New York State Empire Clinical Research Investigator Program. ☆☆ A poster presentation of portions of this data was presented at the Association of Reproductive Health Professionals National Conference in Los Angeles, CA, USA, Sept 2009, and the Society of Teachers of Family Medicine North East Regional Meeting in Rye Brook, NY, Oct 2009. ⁎ Corresponding author. Tel.: +1 718 430 2752; fax: +1 718 430 8645. E-mail address: [email protected] (S.E. Rubin). 0010-7824/$ – see front matter © 2011 Elsevier Inc. All rights reserved. doi:10.1016/j.contraception.2010.10.003
[6] and over 50% of women receiving abortion services in the US were using contraception at the time of their unintended pregnancy [7]. Improper and inconsistent use of contraception contributes to the rate of unintended pregnancy [8,9]. Since nearly one quarter of US women seeking private family planning care see a family physician (FP) [8] and the average US woman spends 30 years trying to avoid pregnancy [10], it is critical that FPs counsel and provide all appropriate contraceptive methods. Many of the factors associated with poor contraceptive adherence (forgetting, method unavailability, misunderstanding of correct use) [11–13] are obviated by userindependent methods. Intrauterine contraception (IUC) is user independent, highly reliable [4], safe [14–16], and costeffective [17,18]. There are few contraindications to IUC use [19], yet IUC use in contracepting US women is only 5% [1] as compared to 15% worldwide [20]. Increasing IUC use has the potential to reduce unintended pregnancy rates [21]. Two types of IUC are currently available in the US, the Copper T
S.E. Rubin et al. / Contraception 83 (2011) 472–478
380A (Teva Pharmaceuticals Inc, marketed as ParaGard®), and the levonorgestrel-releasing intrauterine system (Bayer HealthCare Pharmaceuticals, marketed as Mirena®). The issues behind low IUC utilization are manifold and include patient, clinician and health systems factors. Since clinician recommendation and insertion are necessary for IUC provision, elucidating clinician factors is vital. We know little, however, about FPs practices with IUC. One national survey reports that 99% of FP respondents “dispense, prescribe or recommend” oral contraceptives, but only 39% do so with IUC [8]. A survey of Indian Health Service providers found that all providers perform contraception counseling and have a generally favorable attitude towards IUC, but FPs have less IUC knowledge and experience than obstetriciangynecologists [22]. We do not know of any prior published study identifying possible determinants of whether or not FPs insert IUC. In order to understand potential facilitators and barriers to FPs inserting IUC, we sought (1) to compare the knowledge, attitude and practice of FPs who insert IUC to those who do not, and (2) to describe FPs practice with IUC insertion. We hypothesized that FPs who insert IUC had more insertion training in residency [22,23], graduated from residency more recently, are more likely to be female, perform cervical cancer screening (Pap smears) in their current clinical practice, do not believe it takes more time to counsel about IUC and have increased knowledge of, comfort counseling about, and positive beliefs towards IUC.
473
medium effect size (d=.3, one third of a standard deviation difference between groups) in knowledge, attitude and practice behaviors between inserters and non-inserters. A post hoc multivariate logistic regression was conducted with members of the sample with data on all predictors (completecase analysis). The retained sample size and distribution of variables provided sufficient statistical power for reliable parameter estimates and significance tests. 2.3. Survey instrument
We surveyed a random national sample of FPs identified from the American Medical Association's Physician Master File. This database is intended to include all US physicians [24,25]. Our sample of 3500 FPs was selected by Medical Marketing Services (a database licensee of the American Medical Association) computer-generated algorithm from a total population of approximately 95,000 FPs. Exclusion criteria included Puerto Rican practice sites and classification as a resident, administrator or retiree.
The majority of our questions were adapted from those used in prior survey studies exploring clinicians' knowledge, attitude and clinical practice with IUC [22,23]. We developed additional questions related to FPs and to the levonorgestrel-releasing intrauterine system. All questions were piloted on a sample of FPs at our home institutions, and the survey was modified appropriately. The final 45-item self-administered survey contained demographic questions including training in and current provision of reproductive health services. We measured training in IUC insertion by asking how many IUCs respondents inserted during residency (none, 1–9, 10–19, 20–49 or more than 49). Likelihood to recommend IUC was measured by asking “How likely are you to recommend IUC to a woman with each of the following characteristics?” followed by eight unique patient scenarios, none of which precludes IUC use. These scenarios included nulliparity, distant history of a sexually transmitted infection (STI) or pelvic inflammatory disease (PID), unmarried status, nonmonogamous sexual relationship, abnormal Pap with colposcopy pending, age younger than 20 years and history of ectopic pregnancy. Responses were recorded on a five-point Likert scale. The four knowledge questions were derived from evidence-based literature and package inserts [4,26]. Questions assessing attitudes and beliefs covered: comfort in discussing IUC with patients (very comfortable to very uncomfortable); perception of patients' receptivity to learning about IUC (very receptive to not at all receptive); IUC safety (true, false, unsure); efficacy (very effective to very ineffective) and, as compared to other contraceptive methods, time needed to discuss IUC with patients (more, same or less). For those respondents who indicated they had inserted IUC in the prior 12 months, we asked about the frequency and type(s) of IUC(s) they inserted and their practice with STI testing, cervical cancer screening and pregnancy evaluation prior to insertion.
2.2. Sample size determination
2.4. Data collection
We planned to compare those physicians who had inserted at least one IUC in the past 12 months called “inserters” and those who had not inserted IUC in the past 12 months called “non-inserters.” We speculated that the inserters and non-inserters would be represented in a 1:4 ratio. A priori power analyses indicated that 600 responders would provide adequate power to detect differences of small/
In May 2008, we mailed 3500 FPs a cover letter, survey, $1 incentive and business reply envelope. FPs could respond either via mail or secure Internet survey site. Three weeks later, we sent a follow-up reminder mailing with a replacement survey and another $1 incentive [27,28]. FPs who did not respond 6 weeks after the second mailing were considered nonresponders.
2. Materials and methods The Institutional Review Board of Montefiore Medical Center in the Bronx, NY, and the University of Southern California in Los Angeles approved this project. 2.1. Study population and eligibility
474
S.E. Rubin et al. / Contraception 83 (2011) 472–478
Using information from the Master File list, we compared responders versus nonresponders on variables including age, years since residency training, board certification status, urban vs. nonurban clinical practice site (based on metropolitan statistical area [MSA] codes) and gender.
2.5. Data analysis Primary analyses compared responses between inserters and non-inserters using t-tests for continuous data and chisquare tests for categorical variables. In order to calculate a “likelihood of recommending IUC” variable, we averaged responses to the eight individual questions into a single summed score. Inserter and non-inserter group means were compared. Responses to the four knowledge items were summed into a single knowledge score; group means were calculated and compared. Attitude and belief questions were each analyzed separately. We generated descriptive statistics for inserter practices. Variables that were significantly related to inserter group in bivariate analyses were included as predictors for a post hoc logistic regression model predicting inserter status. The logistic regression was conducted as a complete-case analysis using list-wise exclusion for missing data since missing data did not meet the requirements of missing at random required for imputation. The selected predictors were organized into four blocks: nonmodifiable provider characteristics, IUC beliefs, knowledge and clinical experience. We sequentially entered blocks in the regression model to assess additive effects of each predictor block. The block with nonmodifiable provider characteristics variables was the initial block entered. Entry order for the remaining blocks was determined empirically using model fit indices. At each stage in the model, the block that yielded the greatest contribution to the model was entered. The results reported are for the full model including all predictors. Surveys were included in our bivariate analysis if the respondent completed at least 12 of the 45 items. All data analyses were performed using SPSS version 15.0 (SPSS, Chicago, IL, USA).
3. Results Of the 3500 FPs invited to participate, 973 surveys were returned. We received 869 valid completed surveys, 53 incomplete surveys, 51 returned incentives and/or blank surveys and 45 invalid addresses, and two FPs were deceased. This yields a final response rate of 25%. Only 42 FPs responded via the Internet. Twenty-four percent (n=199) of respondents inserted at least one IUC in the prior 12 months. Responders and non-responders did not differ in terms of age, years since residency training, board certification status or population size of their practice setting. A larger proportion of responders than nonresponders were female (38.7% vs. 31.2%, pb.001). Bivariate analyses comparing inserters and non-inserters were conducted on data from all responders (n=869). Logistic regression analyses was conducted on data from responders with complete data on the selected predictors (n=460). The sample retained for regression analysis had a significantly greater proportion of inserters (29.8%) than the full sample (23.9%). 3.1. Inserters versus non-inserters As Tables 1–3 demonstrate, in unadjusted bivariate analyses, the groups differed significantly on every hypothesized factor except perceived time needed to counsel
Table 1 Demographics, professional background and practice variablesa Total
Female Number of IUDs inserted during residency
None 1–9 10 or more
Years since completed residency training In an average week, % patients female between 15–44 years old In those women, % on whom perform Pap smears In those women, % with whom discuss contraception
Inserters
Non-inserters
p
Adjusted OR (CI)b
35.4 44.3 35.1 20.6
.002 b.001
1.22 (.69–2.15) 1.44 (1.12–1.84)
Mean
(S.D.)
p
Adjusted OR (CI)
9.4 21.9
15.8 31.4
9.7 19.0
b.001 b.001
.997 (.97–1.03) 1.01 (1.00–1.02)
67.5
28.5
45.5
34.6
b.001
1.01 (1.00–1.02)
68.2
30.6
47.6
36.0
b.001
1.00 (.99–1.01)
n
(%)
n
(%)
n
(%)
315 313 307 191
38.6 38.6 37.9 23.6
91 32 89 68
48.1 16.9 47.1 36.0
214 262 208 122
Mean
(S.D.)
Mean
(S.D.)
14.8 33.7
9.1 20.2
13.1 41.1
50.8
34.5
52.7
35.9
Values in bold italics indicate those results that are statistically significant. IUD, intrauterine device. a Percent calculated for number of respondents to each individual question. b Adjusted odds ratios reported are from regression model containing all predictors — gender, years since completing residency, % of practice that includes women of reproductive age, % of patients on whom perform Pap smears, % of patients with whom discussed contraception, number of IUDs inserted during residency, comfort discussing IUC, belief about patients' receptivity to learning about IUC, belief that IUC is effective, time required to discuss IUC, likelihood to recommend IUC and knowledge score.
S.E. Rubin et al. / Contraception 83 (2011) 472–478
475
Table 2 Attitudes, beliefs and likelihood to recommend IUCa Inserter
Very/somewhat comfortable discussing IUDs Believe patients are very or somewhat receptive to discussing IUDs Believe IUDs are safe
Believe IUDs are effective (four-point Likert scale; 1=Very Effective to 4=Very Ineffective) Likelihood to recommend IUC (eight items on a five-point Likert scale; 1=Very Unlikely to 5=Very Likely) As compared to other contraception methods, time needed to discuss IUC with patients (three-point Likert; 1=More Time to 3=Less Time)
Noninserter
p
Adjusted OR (CI)c
n
(%)
n
(%)
181 167 189
95.8 89.3 96.9
476 323 530
79.2 54.8 86.0
b.001 b.001 b.001b
2.35 (1.30–4.27) 2.96 (2.03–4.32)
Mean
S.D.
Mean
S.D.
p
Adjusted OR (CI)
1.07
0.26
1.27
0.57
b.001
1.48 (.58–3.80)
2.72
0.86
2.27
0.79
b.001
.87 (.63–1.20)
1.55
0.49
1.52
0.54
NS
b
a
Percent calculated for number of respondents to each individual question. Variables not included in the logistic model. c Adjusted odds ratios reported are from regression model containing all predictors — gender, years since completing residency, % of practice that includes women of reproductive age, % of patients on whom perform Pap smears, % of patients with whom discussed contraception, number of IUDs inserted during residency, comfort discussing IUC, belief about patients' receptivity to learning about IUC, belief that IUC is effective, time required to discuss IUC, likelihood to recommend IUC and knowledge score. b
patients about IUC. Reported adjusted odds ratios are from the logistic model containing all predictors.
that inserters were statistically more likely than non-inserters to recommend an IUC (2.72 vs. 2.27) (Table 2).
3.1.1. Demographics and practice factors In bivariate analyses, many variables were correlated with inserting IUC, including female gender, inserting IUC during residency, completing residency more recently, caring for more women of reproductive age, performing Pap smears and discussing contraception more frequently. However, in multivariate analysis, only inserting IUC during residency remained a significant predictor: an increase of one category in number of IUCs inserted during residency (0, 1–9, 10+) corresponded to a 1.44 increase in the likelihood of being an inserter (Table 1).
3.1.3. Knowledge On the four-item summed knowledge score, inserters scored .5 point higher than noninserters (3.31 vs. 2.82); this difference is significant in both bivariate and multivariate analyses. Table 3 illustrates responses to individual questions. 3.1.4. Attitudes and beliefs As Table 2 illustrates, while both inserters and noninserters had a positive attitude toward IUC safety and efficacy, inserters rated them as safer and more effective. In our unadjusted model, the absolute difference between inserters and noninserters' belief that currently available IUCs are safe is 10.9%; the difference between groups in feeling very/somewhat comfortable discussing IUC with patients was 16.6%. Eighty-nine point three percent (89.3%) of inserters as compared to 54.8% of noninserters believe
3.1.2. Likelihood of recommending IUC Comparison of bivariate mean composite scores (measured on a five-point Likert scale) from the eight separate patient scenarios, none of which precluded IUC use, showed Table 3 Knowledge about IUCa
Inserter
Knowledge sum scale (all four items)
Non-inserter
Mean
S.D.
Mean
S.D.
3.31
0.74
2.82
0.88
Correct answer:
n
(%)
n
(%)
IUDs do not increase the long-term risk of cervical dysplasia IUDs do not have to be inserted when a woman is menstruating IUDs do not increase the long-term risk of PID IUDs can be used in women with no history of previous pregnancy
172 170 113 162
88.7 86.7 58.2 84.8
430 377 172 441
69.7 60.5 27.6 71.6
p
Adjusted OR (CI)b
b.001
1.85 (1.32–2.60)
PID, pelvic inflammatory disease. a Percent calculated for number of respondents to each individual question. b Adjusted odds ratios reported are from regression model containing all predictors - gender, years since completing residency, % of practice that includes women of reproductive age, % of patients on whom perform Pap smears, % of patients with whom discussed contraception, number of IUDs inserted during residency, comfort discussing IUC, belief about patients' receptivity to learning about IUC, belief that IUC is effective, time required to discuss IUC, likelihood to recommend IUC and knowledge score.
476
S.E. Rubin et al. / Contraception 83 (2011) 472–478
Table 4 Practice patterns of clinicians who insert intrauterine contraception (n=199)a
# IUDs inserted in past 12 months (range 1–50)
Type(s) of IUC inserted
Frequency of antibiotic prescription at time of IUC insertion Maximum amount time from last STI screen
If no prior STI screen, would you… If last normal Pap smear were 2 years ago, would you… To rule out pregnancy prior to inserting, do you check…
CuT 380A only Levonorgestrel IUS only Both Always Sometimes Rarely Never 1 month 3 months 6 months 9 months N9 months Same day screen and insert Screen, follow-up for insertion Other Same day screen and insert Screen, follow-up for insertion Other Urine hCG only Serum hCG only Both urine and serum hCG Other
Mean
S.D.
9.36
9.7
n 41 59 85 6 4 31 146 30 46 28 6 71 56 104 20 81 90 16 147 20 9 9
(%) 22.2 31.9 45.9 3.2 2.1 16.6 78.1 16.6 25.4 15.5 3.3 39.2 31.1 57.8 11.1 43.3 48.1 8.5 79.5 10.8 4.9 4.9
STI, sexually transmitted infection; hCG, human chorionic gonadotropin. a Percent calculated for number of respondents to each individual question.
their patients are receptive to discussing IUC. Significant predictors in the multivariate model included greater odds of feeling very/somewhat comfortable discussing IUC (OR 2.35, 1.30–4.27) and an increasing belief that their patients are receptive to IUC (OR 2.96, CI 2.03–4.32). Belief that IUC is safe was not included in the logistic model since it was not a significant predictor of inserter status in the reduced sample. 3.2. Inserters practice patterns Inserters reported that 41.1% of their patients were women of reproductive age; they performed Pap smears in 67.5% and discussed contraception with 68.2% of these patients. Despite discussing contraception with a median of 1140 female patients annually, they inserted a mean of just nine IUC in the prior 12 months. Only 22.6% (n=45) placed more than 10 devices in the previous 12 months. The inserters were generally up-to-date with their IUC insertion practices. For example, 31.1% would screen for STIs and insert IUC at the same visit (Table 4). 4. Discussion US FPs have training and knowledge gaps, as well as attitudes, that result in missed opportunities to discuss and provide IUC to all eligible patients. FPs who insert IUC are using up-to-date guidelines, yet they insert few devices per
year. Further work needs to be done to understand why FPs do not insert more IUC and to establish effective programs to increase insertion rates. It is encouraging that the factors we found differentiating inserters from noninserters are potentially modifiable. Our results support previous findings that an important predictor of being an inserter is IUC insertion experience during residency [29]. However, FPs lack adequate training in IUC insertion [30,31]. In 1995, 66% of family medicine chief residents had no clinical experience with IUC insertion and only 6% had managed more than six cases during residency [32]. In order to increase IUC use in the US, professional organizations have called for an approach based on the European model where IUC insertion training and education is routinely offered as part of the medical curriculum and for established practitioners [33]. In our sample, the more recent graduates are more likely to insert, indicating that FP residencies may be improving IUC training. It would be informative to have updated data from family medicine residency programs. The other factors associated with inserting in our multivariate model — knowledge, comfort discussing and believing patients are receptive to IUC — are influenced throughout a FPs career. Therefore, these factors are more complex to assess and modify than gaps in residency training. Since we collected data at one point in time, it is uncertain whether, for example, increased knowledge increases the likelihood that a FP inserts IUC, or if inserting IUC leads to better knowledge. Yet, this association between inserting IUC and contraception knowledge is consistent with prior studies [34]. It is also unclear how different clinical practice settings affect these factors. While we found that FPs with certain personal practice characteristics were more likely to insert, many practice setting and systems factors we did not measure including access to other providers for insertion, cost of the device, reimbursement, insurance coverage and limited scope of practice may also affect the ability of a FP to discuss and insert IUC. The most effective methods to change practicing physicians' knowledge, attitudes and practice have not yet been determined and are influenced by a myriad of practice and personal factors [35]. Increasing the proportion of FPs who insert IUC is particularly challenging because in addition to changing knowledge and attitudes, it also involves integrating a new procedure into practice [36]. It may be prudent to focus future efforts on FPs who already provide reproductive health care, including Pap smears and endometrial biopsies. Over half (52.5%) of FPs perform endometrial biopsy [37], the diagnostic procedure most akin to IUC insertion. Further studies of FPs who already provide reproductive health care could clarify details of why some are doing IUC insertions and others are not. Training initiatives could establish how easily FPs with endometrial biopsy skills can learn IUC insertions. Finally, competency in insertion and believing that IUC is safe and effective does not automatically equate with
S.E. Rubin et al. / Contraception 83 (2011) 472–478
inserting IUC. The vast majority of respondents from a national survey of obstetrician-gynecologists agree IUC is safe and effective, yet respondents are unduly restrictive in their eligibility criteria and insert a mean of seven devices annually [23]. A 2007 study in California found that obstetrician-gynecologists are familiar with, have a positive attitude towards and have good knowledge of IUC, but insertion rates are relatively low [38]. Physicians trained to insert IUC may provide it infrequently because they ascribe to out-of-date eligibility criteria [21,39,40]. For example, despite the fact that IUC is an option for women with a history of ectopic pregnancy [19,41], for nulligravid women and adolescents [19,42,43] and for women with a history pelvic inflammatory disease (PID) or STI more than 3 months prior to insertion [14,15,44], many of our respondents were neutral or unlikely to recommend IUC for a women with any one of these issues in their medical history. Our survey has a number of limitations. First, our response rate is low, though it does fall within the range of published physician survey studies [45–47]. Since a topic's perceived relevance affects response rate [48], our response rate may reflect nonresponders' perception that IUC does not pertain to their practice. Thus, our result showing that 24% of FPs insert IUC is likely higher than the actual percentage of FPs who insert. Second, by conducting the logistic regression as a complete-case analysis, there was a substantial drop in the sample size. This reduced the statistical power of the model, and yielded a potentially biased sample. Support for the complete-case approach can be found in the consistency of bivariate associations in the full and subgroup sample; all but one of the bivariate associations found to be significant in the full sample were also significant in the reduced sample. Therefore, despite potential bias in the sample, results from the multivariate analysis provide important information about identifying differences between inserters and non-inserters when controlling for other variables in the model. Third, as with all survey studies, a response bias is of concern. However, due to the relative homogeneity of physicians, it has been suggested that response bias is less concerning as compared to layperson surveys [28]. Finally, we analyzed our data comparing inserters and non-inserters. The non-inserters are likely heterogeneous in the frequency with which they counsel about IUC. We did not measure knowledge, attitudes and beliefs that affect whether a non-inserter is more or less likely to discuss and recommend IUC. Studies examining barriers and facilitators of IUC counseling, not simply insertion, at both the individual provider and larger systems level would be informative. Despite these limitations, our study demonstrates that significant knowledge and training gaps, as well as attitudes, prevent FPs from offering IUC to all eligible patients. More IUC education and insertion training in residency, as well as directing interventions at FPs who already provide reproductive health services, could increase the proportion of women who have access to IUC. Increasing use of this
477
effective, user-independent contraceptive method could ultimately reduce the rate of unintended pregnancy, bringing us closer to the objectives of Healthy People 2010.
References [1] Mosher W, Jones J. Use of contraception in the United States: 19822008. Natl Cent Health Stat Vital Health Stat 2010;23:84. [2] Bensyl DM, Iuliano DA, Carter M, et al. Contraceptive use—United States and territories, Behavioral Risk Factor Surveillance System, 2002. [erratum appears in MMWR Morb Mortal Wkly Rep. 2005;54:1214] Morb Mortal Wkly Rep Surveill Summ 2005;54:1–72. [3] Trussell J. Contraceptive failure in the United States. Contraception 2004;70:89–96. [4] Hatcher R, Trussell J, Nelson A, et al. Contraceptive Technology. 19th edition. New York: Ardent Media; 2008. [5] Healthy People 2010. http://www.healthypeople.gov/Document/html/ volume1/09Familyhtm [cited 2010 March 1]; Available from: http:// www.healthypeople.gov/Document/html/volume1/09Family.htm. [6] Jones RK, Singh S, Finer LB. Repeat abortion in the United States. Occasional Report. New York: Guttmacher Institute; 2006. [7] Jones RK, Darroch JE, Henshaw SK, et al. Contraceptive use among U.S. women having abortions in 2000-2001. Perspect Sex Reprod Health 2002;34:294–303. [8] Landry DJ, Wei J, Frost JJ. Public and private providers' involvement in improving their patients' contraceptive use. Contraception 2008;78:42–51. [9] Moreau C, Cleland K, Trussell J. Contraceptive discontinuation attributed to method dissatisfaction in the United States. Contraception 2007;76:267–72. [10] Frost JJ, Darroch JE, Remez L. Improving contraceptive use in the United States. Issues Brief (Alan Guttmacher Inst) 2008:1–8. [11] Nelson AL, Westhoff C, Schnare SM. Real-world patterns of prescription refills for branded hormonal contraceptives: a reflection of contraceptive discontinuation. Obstet Gynecol 2008;112:782–7. [12] Smith JD, Oakley D. Why do women miss oral contraceptive pills? An analysis of women's self-described reasons for missed pills. J Midwifery Women's Health 2005;50:380–3. [13] Rosenberg MJ, Waugh MS, Meehan TE. Use and misuse of oral contraceptives: risk indicators for poor pill taking and discontinuation. Contraception 1995;51:283–8. [14] Shelton JD. Risk of clinical pelvic inflammatory disease attributable to an intrauterine device. Lancet 2001;357:443. [15] Grimes DA. Intrauterine device and upper-genital-tract infection. Lancet 2000;356:1013–9. [16] Grimes DA, Lopez LM, Manion C, et al. Cochrane systematic reviews of IUD trials: lessons learned. Contraception 2007;75(6 Suppl):S55–9. [17] Trussell J, Lalla AM, Doan QV, et al. Cost effectiveness of contraceptives in the United States. Contraception 2009;79:5–14. [18] Chiou CF, Trussell J, Reyes E, et al. Economic analysis of contraceptives for women. Contraception 2003;68:3–10. [19] World Health Organization. Medical eligibility criteria for contraceptive use - 3rd edition. Intrauterine devices (IUDs); 2004. with 2008 update [cited 2010 June 22]; Available from: http://www.who.int/ reproductivehealth/publications/family_planning/9241562668index/ en/index.html. [20] Population Reference Bureau. Family Planning Worldwide, 2008 Data Sheet; 2008 [cited 2010 February 10]; Available from: http://www.prb. org/pdf08/fpds08.pdf. [21] Speidel JJ, Cynthia CH, Wayne CS. The potential of long-acting reversible contraception to decrease unintended pregnancy. Contraception 2008;78:197–200. [22] Espey E, Ogburn T, Espey D, et al. IUD-related knowledge, attitudes and practices among Navajo Area Indian Health Service providers. Perspect Sex Reprod Health 2003;35:169–73.
478
S.E. Rubin et al. / Contraception 83 (2011) 472–478
[23] Stanwood NL, Garrett JM, Konrad TR. Obstetrician-gynecologists and the intrauterine device: a survey of attitudes and practice. Obstet Gynecol 2002;99:275–80. [24] Konrad TR, Slifkin RT, Stevens C, et al. Using the American Medical Association physician masterfile to measure physician supply in small towns. J Rural Health 2000;16:162–7. [25] Williams PT, Whitcomb M, Kessler J. Quality of the family physician component of AMA Masterfile. J Am Board Fam Pract 1996;9:94–9. [26] Paragard FDA package insert. 2009 [cited 2009 June 13]; Available from: http://www.paragard.com/health_care_professional/global/pdf/ Prescribing-Info.pdf. [27] Edwards P, Roberts I, Clarke M, et al. Increasing response rates to postal questionnaires: systematic review. Bmj 2002;324(7347):1183. [28] Kellerman SE, Herold J. Physician response to surveys. A review of the literature. Am J Prev Med 2001;20:61–7. [29] Harper CC, Blum M, de Bocanegra HT, et al. Challenges in translating evidence to practice: the provision of intrauterine contraception. Obstet Gynecol 2008;111:1359–69. [30] Cheng D. Family planning training in Maryland Family Practice and Obstetrics/Gynecology residency programs. J Am Med Womens Assoc 1999;54(4):208–10. [31] Schreiber CA, Harwood BJ, Switzer GE, et al. Training and attitudes about contraceptive management across primary care specialties: a survey of graduating residents. Contraception 2006;73:618–22. [32] Steinauer JE, DePineres T, Robert AM, et al. Training family practice residents in abortion and other reproductive health care: a nationwide survey. [see comment]Fam Plann Perspect 1997;29:222–7. [33] Sonfield A. Popularity disparity: attitudes about the IUD in Europe and the United States. Guttmacher Policy Rev 2007;10:19–24. [34] Dehlendorf C, Levy K, Ruskin R, et al. Health care providers' knowledge about contraceptive evidence: a barrier to quality family planning care? Contraception 2010;81:292–8. [35] Cabana MD, Rand CS, Powe NR, et al. Why don't physicians follow clinical practice guidelines? A framework for improvement. JAMA 1999;282:1458–65. [36] Litaker D, Ruhe M, Flocke S. Making sense of primary care practices' capacity for change. Transl Res 2008;152:245–53.
[37] American Academy of Family Physicians, Practice Profile II Survey. Performance of diagnostic procedures in family physicians. Table #64. American Academy of Family Physicians, Practice Profile II Survey. July 2008 [cited 2010 January 9]; Available from www.aafp.org/ online/en/home/aboutus/specialty/facts/64.html. [38] Postlethwaite D, Shaber R, Mancuso V, et al. Intrauterine contraception: evaluation of clinician practice patterns in Kaiser Permanente Northern California. Contraception 2007;75:177–84. [39] Madden T, Allsworth JE, Hladky KJ, et al. Intrauterine contraception in Saint Louis: a survey of obstetrician and gynecologists' knowledge and attitudes. Contraception 2010;81:112–6. [40] Rubin SE, Fletcher J, Stein T, et al. Underuse of the IUD in contraceptive care and training. Fam Med 2010;42:387–8. [41] Skjeldestad FE. How effectively do copper intrauterine devices prevent ectopic pregnancy? [Article] Acta Obstet Gynecol Scand 1997;76: 684–90. [42] Hubacher D, Lara-Ricalde R, Taylor DJ, et al. Use of copper intrauterine devices and the risk of tubal infertility among nulligravid women. N Engl J Med 2001;345:7. [43] National Institute for Health and Clinical Excellence. Long-acting reversible contraception. 2005 [cited 2010 March 5]; Available from http://www.nice.org.uk/Guidance/CG30/NiceGuidance/pdf/English. [44] Steen R, Shapiro K, Steen R, et al. Intrauterine contraceptive devices and risk of pelvic inflammatory disease: standard of care in high STI prevalence settings. Reprod Health Matters 2004;12:136–43. [45] Cummings SM, Savitz LA, Konrad TR. Reported response rates to mailed physician questionnaires. Health Serv Res 2001;35: 1347–55. [46] Moeller MP, White KR, Shisler L. Primary care physicians' knowledge, attitudes, and practices related to newborn hearing screening. Pediatrics 2006;118:1357–70. [47] Astin JA, Soeken K, Sierpina VS, et al. Barriers to the integration of psychosocial factors in medicine: results of a national survey of physicians. J Am Board Fam Med 2006;19:557–65. [48] Dillman D. Constructing the questionnaire. Mail and Internet Surveys: The Tailored Design Method, 2nd ed. 2000. p. 87.