Associations of anticipated prescribing of long-acting reversible contraception by general practice registrars: A cross-sectional study

Long-acting reversible contraception (LARC) is defined as a contraceptive method administered less frequently than monthly,¹ and includes the copper intrauterine device (IUD), levonorgestrel IUD, subdermal etonogestrel implant and medroxyprogesterone injection.^1,2 These are the most effective methods of contraception.³ However, the uptake of these methods in Australia is low, with LARC users accounting for 12.5% of women using contraception.⁴ Over 50% of women will have an unplanned pregnancy,⁵ with 26% of pregnancies in Australia being unplanned.⁶ One in four pregnancies in Australia were terminated in 2002,⁷ although Australian termination rates appear to be declining – the figures decreased to 18% in South Australia in 2016.⁸ Increasing the uptake of LARC is proposed to be an effective strategy to reduce unplanned pregnancy and termination,⁹ and there is evidence that contraceptive counselling focusing on LARC is effective in reducing unintended pregnancy.¹⁰

Knowledge gaps and suboptimal training of healthcare providers are key barriers to increased LARC use.¹¹ If healthcare providers provide women with a comprehensive contraceptive consultation that includes discussion of LARC, there is a significant increase in LARC uptake.^12,13 General practitioner (GP) knowledge regarding the efficacy and appropriate use of LARC is limited.¹⁴ Common misconceptions are that IUDs are unsuitable for young women, nulliparous women or those with previous ectopic pregnancy.¹⁵ Additionally, insertion of contraceptive implants and IUDs requires additional training, involving expense, time and commitment to maintain skills.^9,11

The training and behaviour of general practice registrars are important indicators of future primary care provision. These authors’ previous work¹⁶ showed that the majority of general practice registrars’ prescriptions for contraception are for the oral contraceptive pill (OCP) and suggested that general practice registrars find prescription of LARC challenging when compared with prescription of non-LARC methods.

The aim of this study was to establish what training and experience general practice registrars have received in relation to LARC, and to explore their ‘theoretical’ decision making in response to a series of clinical vignettes.

Methods

Study design

This was an anonymous cross-sectional questionnaire-based online study.

Questionnaire

The questionnaire contained three clinical vignettes (Table 1), designed to elicit registrars’ anticipated contraception prescribing practice. In each vignette the first-line contraception choice (given the clinical and contextual information provided, and current evidence-based guideline recommendations) was LARC. Vignettes entailed aspects of common LARC misconceptions: LARC use, particularly IUD, in young nulliparous women (vignette 1); and IUD use in women with a history of ectopic pregnancy (vignette 2). For each vignette, participating registrars were advised the patient had no preconceived preference and was willing to proceed with any contraception method recommended, and that either the registrar or someone else in the practice was able to insert. 

The questionnaire also elicited training and experience in insertion of IUDs and implants.

Study setting

Participants were registrars training with a single regional training organisation (RTO) that is responsible for general practice training across New South Wales (NSW) and the Australian Capital Territory (ACT), which accounts for training of 32% of Australian general practice registrars.¹⁷ RTOs are government-funded, not-for profit, geographically defined GP vocational training organisations.

Data-collection methods

Registrars were invited to participate via email. They were invited to complete the questionnaire via a link to SurveyMonkey. The initial invitation was followed up by a repeat invitation three weeks later. The questionnaire did not allow registrars to skip questions. Consent was implicit in participants’ completion of the anonymous questionnaire.

Eligibility criteria

Inclusion criteria

Inclusion criteria were NSW or ACT general practice registrars currently in a general practice–based term.

Variables

Outcome variable

The outcome factors were the responses to each of the three vignette items as ‘LARC chosen’ (defined as selecting either ‘IUD’, ‘implant’ or ‘injection’) or ‘LARC not chosen’ (defined as selecting either ‘condoms’, ‘progesterone-only pill’ [POP] or ‘combined oral contraceptive pill’ [COCP]).

Clinical vignettes are valid methods that directly focus on clinicians’ process of care in actual clinical practice.¹⁸ They have been used in previous studies of Australian general practice registrars.^19,20

Independent variables

Registrar demographics included gender, age, current general practice term, part-time or full-time training, country of primary medical qualification (Australia or international) and number of full-time equivalent (FTE) GPs in the registrar’s current practice.

Variables related to reproductive health experience were work experience (having previously worked in family planning or sexual health, or having completed a hospital rotation in obstetrics and gynaecology), postgraduate qualification (either completing the Diploma of Royal Australian and New Zealand College of Obstetrics and Gynaecology [DRANZCOG] or the Family Planning Alliance Australia National Certificate in Reproductive and Sexual Health [FPAANCRSH], or overseas equivalents).

Variables related to LARC training and experience were completion of IUD or implant insertion training, IUD or implant insertion experience in general practice, and disagreement with the questionnaire item ‘I do not feel that I know enough about these methods to discuss with my patients’.

Statistical methods

For each clinical vignette, the proportion of registrars choosing LARC was calculated with 95% confidence intervals (CIs). Descriptive statistics for independent variables included frequencies for categorical variables and mean with standard deviation²¹ for continuous variables. Frequencies of categorical variables were compared between outcome categories (‘LARC chosen’ or ‘LARC not chosen’) using Chi-squared tests for all variables, except when Fisher’s exact test was appropriate. For continuous variables, means were compared using a t-test.

Univariate and multivariate logistic regressions with outcome factor ‘LARC chosen’ or ‘LARC not chosen’ were used to establish factors associated with choosing LARC for each vignette.

Univariate logistic regression analyses were conducted for each covariate. Covariates with a univariate P value <0.20 were considered for inclusion in the multiple regression model.

Once the model with all significant covariates was fitted, model reduction was assessed. Covariates that were no longer significant (at P <0.2) in the multivariate model were tested for removal from the model. If the covariate’s removal did not substantively change the resulting model, the covariate was removed from the final model. A substantive change to the model was defined as any covariate in the model having a change in the effect size (odds ratio) of greater than 10%.

The regressions modelled the log-odds that the correct response was given for each clinical vignette.

Analyses were programmed using STATA 14.0 and SAS V9.4.

Ethical approval

Ethical approval for this study was obtained from the University of Newcastle Human Research Ethics Committee (H-2018-0184).

Results

A total of 223 registrars completed the questionnaire (response rate 25%). Participating registrars’ characteristics are presented in Table 2. Regarding previous experience, 118 (53%) registrars had completed a hospital rotation in obstetrics and gynaecology, and 14 (6.3%) had ever worked in family planning or sexual health. Thirty-eight (17%) had completed the FPAANCRSH (or equivalent), and 16 (7.2%) had completed the DRANZCOG. Thirty-eight (19%) registrars had received training in IUD insertion. Eighteen registrars (9.4%) had ever inserted an IUD in general practice. A total of 128 (64%) had received training in insertion of the contraceptive implant. Ninety-six (50%) had ever inserted the implant in general practice. 

Univariate associations of choosing LARC versus non-LARC methods for each vignette are presented in Table 3. The multivariate logistic regression results for selecting LARC versus non-LARC for each vignette are presented in Table 4. 

For vignette 1 (a nulliparous young patient; Table 1), 141 (69.5%; 95% CI: 62.7, 75.5) registrars chose LARC as the best option, 127 (62.5%) selected a contraceptive implant and 14 (6.9%) selected an IUD. No registrars selected the injection. For non-LARC methods, 57 (28.1%) chose the COCP, five (2.5%) chose barrier methods and no registrars selected the POP. Choice of a LARC was significantly associated with Australian primary medical degree (OR: 3.4; 95% CI: 1.62, 7.15) and disagreement with the statement that they did not know enough about LARC to discuss with patients (OR: 2.87; 95% CI 1.21, 6.84).

For vignette 2 (a woman with a history of ectopic pregnancy; Table 1), 161 (79.3%; 95% CI: 73.1, 84.4) registrars chose a LARC method, with 142 (70%) selecting an IUD, 18 (8.9%) selecting a contraceptive implant and one (0.5%) selecting the injection. For non-LARC methods, 41 (20.2%) registrars chose the COCP, one (0.5%) chose condoms and no registrars selected the POP. The choice of a LARC method was significantly associated with the registrar being female (OR: 2.99; 95% CI: 1.44, 6.22).

For vignette 3 (no recognised obstacles to LARC use), 162 (79.8%; 95% CI: 73.7, 84.8) registrars chose a LARC method, with 126 (62.1%) selecting an IUD, 31 (15.3%) selecting a contraceptive implant and five (2.5%) selecting the injection. For non-LARC methods, 19 (9.4%) registrars selected the COCP, six (3.0%) selected the POP and 16 (7.8%) selected condoms. The choice of a LARC method was significantly associated with disagreement with the statement that they did not know enough about LARC to discuss with patients (OR: 2.99; 95% CI: 1.25, 7.15).

Discussion

For all clinical vignettes, most registrars recognised a LARC modality as the first-line option. In practice, most prescriptions issued by general practice registrars are for non-LARC methods.¹⁶ Comparing results of this ‘theoretical’ prescribing with actual prescribing from these authors’ previous research will help understand how much the modest uptake of LARC prescribing is due to knowledge deficits as opposed to alternative barriers. The findings emphasise the discrepancy between knowledge and real-world practice, and suggest barriers to LARC use beyond lack of knowledge. However, the findings suggest that, like established GPs,¹¹ general practice registrars do not recommend IUDs to nulliparous women. For an older, parous woman, most registrars chose the IUD. This suggests that registrars are aware of the efficacy of LARC methods but may lack training in the specifics of eligibility.

GPs require specific training in the insertion of the contraceptive implant and IUD, and completion of this is inadequate.^11,22 A solution could be the introduction of LARC insertion training at the general practice registrar level. Many registrars in the participating RTO received training in insertion of the contraceptive implant as part of their mandatory training. This was not the case for IUD insertion training. The present results reflect this and suggest that the majority of those who have received training in implant insertion also insert in practice. This supports the suggestion that provision of IUD insertion training within registrars’ educational programs would lead to increased IUD insertion in practice. A recent Australian trial has found that a combination of GP training on contraceptive effectiveness counselling and rapid access to LARC insertion clinics resulted in greater patient uptake of LARC.¹³ Thus, registrars’ insertion training should be supplemented with contraceptive counselling training in RTO education and training programs.

Factors associated with choosing LARC have implications for registrar training in contraception. As a quarter of all general practice registrars obtained their medical qualification outside of Australia,¹⁷ it is important that general practice registrar training addresses gaps in knowledge of LARC. This is particularly relevant, as previous research¹⁶ has shown that general practice registrars’ LARC prescribing is associated with rural, compared with major cities, practice location. Since the majority¹⁷ of international medical graduate doctors work in regional and remote regions as a result of Medicare Section 19AB restrictions, they ideally should be prepared to provide LARC in rural regions where rates of unintended pregnancy are highest.²³

The present findings also suggested that those who feel confident in their knowledge of LARC are more likely to choose LARC. This may have significant educational implications in that self-reflection of learning needs may be a valuable tool in identifying registrars that require additional training. The finding that female registrars are more likely to choose LARC in a scenario involving previous history of ectopic pregnancy suggests there may be differences in adequate reproductive health exposure between male and female trainees. This reflects previous findings of gender differences in the performing of women’s reproductive health–related procedures.²⁴

Strengths and limitations

The strength of this research is generalisability to Australian general practice vocational training.

A limitation is the relatively modest response rate of 25%. Although typical for a GP questionnaire study,²⁵ this may still have introduced volunteer bias, with registrars interested in contraception more likely to respond. If so, the data may overestimate the level of training and experience of registrars in relation to LARC, and overestimate the prevalence of LARC choice in response to the vignettes. A further limitation is that the clinical vignettes did not include patient preference and assumed that any method was available in the practice. In clinical practice, patients may prefer non-LARC contraceptive methods, even after appropriate counselling.^26,27 However, this design has allowed us to postulate that registrar knowledge alone is not the key barrier to LARC prescribing, and that the lower prescribing rates seen in these authors’ previous research¹⁶ reflect other factors such as patient factors, training in insertion, time and cost.

A further limitation of the cross-sectional study design is the establishment of association but not causality in relationships of independent variables with the outcomes.

Conclusion

The present findings suggest general practice registrars are aware that LARC is an appropriate first-line option and, taken together with these authors’ previous work, establish a seeming discrepancy between knowledge and practice. The findings further suggest implications for general practice vocational training in the need for LARC training in RTO programs, with particular consideration of training in IUD insertion, and in training of male registrars and international medical graduate registrars.