Patterns of contraceptive use through later reproductive years: A cohort study of Australian women with chronic disease.

Unintended pregnancy is highest among young women, but there is also a growing consensus that it is a significant public health issue for women of older reproductive age [ 1 , 2 ]. In addition, the prevalence of chronic disease among women of childbearing age is on the rise. In Australia it is estimated that around 30% of women will be diagnosed with at least one chronic health condition during their reproductive years [ 3 ]. This trend is predicted to increase over the coming decade, with chronic disease prevalence increasing substantially across successive generations. For women with chronic disease, unintended pregnancies are associated with serious adverse maternal and perinatal outcomes such as congenital abnormalities, early pregnancy loss, and stillbirth [ 4 – 7 ]. Use of effective contraception is recommended as a key strategy to allow these women to plan pregnancies for times of optimal health, or to provide sufficient time to discontinue potentially teratogenic medications and switch to medications which have greater compatibility with pregnancy. However, contraceptive counselling among chronic disease populations remains low [ 8 , 9 ]. Given chronic disease is on the rise and these women report unintended pregnancy rates at similar or higher rates than the general population, understanding how they use or don’t use contraception is critical to tailoring contraceptive counselling interventions for women with chronic disease as they move through their childbearing years [ 8 , 10 , 11 ]. Despite this, there is limited population-level evidence regarding the contraceptive practices of women with chronic disease, and no studies have examined contraceptive patterns using nationally representative longitudinal data. Of the few available studies, the findings have been equivocal, driven by a reliance on retrospective cross-sectional study designs, differences in contraceptive methods examined and a focus on small single disease samples [ 10 , 12 , 13 ]. As such, the prevalence of contraceptive use and types of methods employed have varied widely across chronic disease populations with contraceptive use found to range from around 30% to as high as 99% [ 14 – 16 ]. Only one study has longitudinally examined contraceptive use among women with chronic disease of childbearing age using state-based insurance claims data [ 17 ]. While they found only one-third of women with chronic disease were prescription contraceptive users (compared to 40% of women without a chronic condition), they were unable to examine a range of contraceptive options and they failed to account for women not at risk of pregnancy. Further, while previous studies have involved women with chronic disease across the reproductive life span, recent Australian research suggests that contraceptive patterns differ markedly by life stage and over time [ 18 , 19 ]. It is therefore important to take a lifecourse approach to contraceptive use among women with chronic disease to understand contraceptive practices as women transition through their childbearing years. This study aimed to establish an evidence-base regarding the contraceptive practices of women with chronic disease by examining patterns of contraceptive use over time among an Australian cohort of women born 1973–78 who have been prospectively followed for over 20 years.

In 2006, 25.8% of women (aged 28–33 years) reported at least one chronic disease. This increased to 35.5% by 2018 when aged 40–45 years ( Table 1 ). The most common chronic condition was asthma (17.9% in 2006). Other conditions, including diabetes, cardiac disease, autoinflammatory arthropathies and connective tissue disease, IBD, multiple sclerosis, and thyroid disease each had prevalence rates of less than 5% in 2006. However, the prevalence of each of these conditions had approximately doubled by 2018 when the women were aged 40–45 years. Given the low frequencies of autoinflammatory conditions (AICs; i.e., autoinflammatory arthropathies and connective tissue disease, IBD, multiple sclerosis, and thyroid disease) these were combined into a single autoinflammatory disease category for subsequent modelling. a Includes autoinflammatory arthropathies, inflammatory bowel disease, thyroid disease and multiple sclerosis; no missing data for chronic disease variables as these were derived from a range of data sources. Comparing women with and without chronic disease in this sample, there were few differences across sociodemographic, health behaviour and reproductive health characteristics at Survey 4 in 2006 ( Table 2 ). Women with chronic disease were less likely to have a university degree (42.5% vs. 48.0%), and more likely to report their ability to manage on available income as impossible or difficult always (16.8% vs. 11.7%). Similar differences were also observed in 2018 when the women were aged 40–45 years. *PCOS = Polycystic ovary syndrome; missing data for each variable was less than 0.5%, except for history of pregnancy (0.52%), menstrual symptoms (0.68%) and body mass index (2.52%). In 2006, 91.5% of women with chronic disease were using some form of contraception compared to 93.2% of women without chronic disease ( Table 3 ). At this time point, 43.8% of women with chronic disease were using short-acting hormonal contraception compared to 46.2% of women without chronic disease. By 2018, these proportions had reduced, but were similar (15.1% vs 15.6%). In 2006, 6.5% and 5.3% of women with and without chronic disease were using LARC, respectively. By 2018, LARC use had increased substantially with 23.7% of women with chronic disease using this method compared to 20.7% of women without chronic disease. A similar increase was noted for sterilisation methods, with relatively low use in 2006 (11.3% vs 9.0%, respectively) and substantially higher use at age 40–45 years in 2018 (31.6% vs 33.0%, respectively). A The short-acting category was composed of the pill (91.5%), the minipill (5.8%), injection (2.5%) and vaginal ring (0.7%). B The long-acting reversible contraception (LARC) category was composed of progestogen-only IUD (64.5%), implant (32.2%) and copper IUD (4.0%). C The natural methods category was composed of withdrawal method (88.9%) and fertility awareness methods (20.1%). D The sterilisation category was composed of vasectomy (79.1%) and tubal ligation (21.6%). Note: Types of contraception do not add to 100% as respondents were able to select multiple methods. LARC = long-acting reversible contraception. Among women with chronic disease, use of one contraceptive method only was similar over time (70.0% and 71.9% in 2006 and 2018 respectively). Among women with chronic disease, use of two or more contraceptive methods declined from 21.5% in 2006 to 14.5% in 2018. Contraceptive non-use increased over the observation period with 13.6% of women with chronic disease and 12.7% without chronic disease not using any contraception in 2018 (when aged 40–45 years). Emergency contraception was reported with very low frequencies across time for women with and without chronic disease (1.6% vs 1.1% in 2006 and 0.5% vs 0.2% in 2018, respectively). Given these low frequencies and the purpose of emergency contraception being to prevent pregnancy where contraception has not been used, misused, or has failed, it was excluded from further analysis. A five-status LTA model was selected as the optimal model, given its reasonable clinical interpretability and goodness-of-fit ( S3 Table ). Status 1 ( Table 4 ), “short-acting and condom” was characterised by high use of short-acting methods (100% probability) with some supplementation with condoms (15% probability). Status 2, “condom and natural” captured high use of condoms (69% probability) with some supplementation of non-hormonal natural methods (38% probability) and other methods (8% probability). Status 3, “sterilisation and other” was dominated by vasectomy or tubal sterilisation (100% probability) but included supplementation of other methods for some women (16% probability). Status 4, “LARC” included the use of long-acting methods (100% probability), with a small amount of supplementation such as condoms (3% probability). The “no contraception” status (Status 5) captured the absence of contraceptive use (100% probability). Status 1, “short-acting and condom”, was selected as the reference status as it was comprised of two of the most popular contraceptive methods among reproductive-aged women, the pill and condom. This was reflected in the data, with 44% of women in 2006 belonging to Status 1 in 2006 ( S4 Table ). As such, Status 1 was appropriate as a baseline contraceptive against which to make comparisons in the subsequent modelling. Note: Dashed cells have item-response probabilities <0.01. Shaded status (#1, short-acting and condom) was selected as the reference class for subsequent analysis. Natural methods = withdrawal and fertility-awareness methods. LARC = Long-acting reversible contraception. Women were most likely to remain in the same latent status between time points ( S2 Table ), with women in Status 3 (“sterilisation and other”) in 2006 the most likely to continue using the same contraception in 2012 (probability = 0.86). Women in Status 5 in 2006 (“no contraception”) were most likely to remain in Status 5 when measured again in 2012 (probability = 0.32) but had a moderate probability of transitioning to Status 2 (“condom and natural”, probability = 0.26) or Status 3 (“sterilisation and other”, probability = 0.18). Women in Status 1 (“short-acting and condom) in 2012 had similar probabilities of transitioning to “condom and natural” (probability = 0.14), “sterilisation and other” (probability = 0.17), “LARC” (probability = 0.15)” and “no contraception” (probability = 0.13). There was no evidence to support a difference in the patterns of contraceptive use for women with any chronic disease compared to women without chronic disease ( Table 5 ). When each of the chronic conditions were examined separately, a difference in contraceptive patterns was observed only for women with autoinflammatory disease. Women with autoinflammatory disease had increased odds of using condom and natural methods (OR = 1.20, 95% CI = 1.00, 1.44), and sterilisation and other methods (OR = 1.61, 95% CI = 1.08, 2.39) or no contraception (OR = 1.32, 95% CI = 1.04, 1.66), compared to women without chronic disease using short-acting methods and condoms (full model results are presented in S5 – S9 Tables). Reference status = short-acting and condom; reference level for disease = disease not present. Each model controlled for age, country of birth, area of residence, highest educational qualification, relationship status, health care card holder status, smoking status, body mass index, history of pregnancy, history of termination, history of miscarriages, menstrual symptoms, history of polycystic ovary syndrome, history of endometriosis, and survey wave. LARC = Long-acting reversible contraception. Full model results can be found in S5 – S9 Tables.

Across all women, the use of highly effective contraception such as LARC increased over time, although rates were still relatively low and contraceptive non-use also increased. This is problematic and highlights the general need for contraceptive care and counselling to reduce the risks of unintended pregnancy for all women, including those at later reproductive life. We also demonstrated that women with chronic disease take up contraception at similar rates to their same aged peers in the community. However, among women with chronic disease, women diagnosed with AICs were more likely to engage in low efficacy methods of contraception or did not use contraception. Our study therefore highlights the potential gaps in the provision of appropriate contraceptive access and care for women with chronic disease, particularly around the suitability of estrogen-containing contraceptives, including those diagnosed with AICs. The need for the development of national guidelines and a clear contraceptive strategy from adolescence through to menopause are required. Guidelines should encourage regular contraceptive review during routine care as well as training and education for medical professionals to increase support for, and agency among, women with chronic disease. This, in turn, will reduce the occurrence of high-risk unintended pregnancies and facilitate optimal outcomes for planned pregnancies.

Data were obtained from the 1973–78 cohort of the Australian Longitudinal Study on Women’s Health (ALSWH), a national population-based study examining health and wellbeing among Australian women. Women were randomly sampled through the national health insurer’s database (Medicare). This cohort has been found to be largely representative of the population of women in this age group [ 20 ]. These women have completed surveys in 1996, 2000 and then on a three-yearly schedule thereafter. This retrospective analysis focused on women who completed Surveys 4, 6 or 8 conducted in 2006 (aged 28–33 years), 2012 (aged 34–39 years) and 2018 (aged 40–45 years). These time points provided measurements across women’s main reproductive years. Of the 14,247 women who completed the baseline survey in 1996, 9,604 women were eligible for linked data analysis and completed the questions related to contraceptive use at the selected analysis time points ( Fig 1 ). At each survey, women were considered not at risk of an unintended pregnancy if they reported any of the following: no male partner, hysterectomy, currently pregnant, trying to become pregnant, infertile partner, or partner with low or zero sperm count. Excluding women who were not at risk at all time points (N = 1,574), the final sample for analysis included 8,030 women. Women included in this analysis had a similar demographic profile to the full 1973–78 cohort at baseline in 1996 ( S1 Table ). *Percentage of women at each survey who completed contraceptive questions but were excluded due to being not at risk of an unintended pregnancy: 2006 = 36%, 2012 = 31%, 2018 = 28%. Contraceptive use was measured at each survey. Contraceptive patterns were derived from the question “What forms of contraception do you use now?” At each survey, contraception was measured with 14 response options (participants were able to select more than one option). Women were also asked if they had had a tubal sterilisation or if their partner had had a vasectomy. For this analysis, responses were collapsed into eight groups based on contraceptive efficacy: sterilisation (tubal sterilisation or vasectomy); long-acting reversible contraception (LARC; the progestogen intrauterine system, the copper intrauterine device and the progestogen-only implant); short-acting hormonal contraception (progestogen-only contraceptive pill, combined oral contraceptive pill, oral contraception of unknown type, vaginal ring and depot injection); condoms; natural methods (withdrawal and fertility-based awareness methods); emergency oral contraception; other contraception; and no contraception [ 19 ]. Chronic disease was measured at each survey by the presence or absence of seven physical chronic diseases that have been associated with poor maternal and fetal outcomes. These included diabetes, cardiac disease (including hypertension), asthma, autoinflammatory arthropathies and connective tissue disease (e.g., rheumatoid arthritis [RA] and systemic lupus erythematosus [SLE]), inflammatory bowel disease (IBD), multiple sclerosis, and thyroid disease). Chronic diseases were ascertained using multiple data sources (self-report survey, state-based hospital admissions data, pharmaceutical prescriptions and disease-specific medical claims from general practitioners and specialist care) and employing disease-specific algorithms (developed in concert with clinical experts) to increase chronic disease accuracy in this population. Detailed information on the methods employed are described elsewhere [ 21 ]. Time-varying covariates were measured at each survey. Sociodemographic variables included age, highest educational qualification (no formal qualifications, school/higher school certificate; trade/certificate/diploma; university/higher degree), area of residence (categorised according to the Accessibility/Remoteness Index of Australia (ARIA+) classification system as: major cities; inner regional; outer regional/remote/very remote), relationship status (partnered; unpartnered), and country of birth (Australia; other English speaking; other). Health care card holder status (a concession card provided for government-subsidised health care) was also included as a surrogate for socioeconomic status (yes; no). Health-related factors included smoking status (current smoker; ex-smoker; non-smoker) and body mass index (underweight [<18.5 kg/m 2 ]; healthy [≥18.5 and <25 kg/m 2 ]; overweight [≥25 and <30 kg/m 2 ]; obese [≥30 kg/m 2 ]) [ 22 ]. Reproductive health factors included history of pregnancy (yes; no) and history of pregnancy termination (yes; no). Given that contraception is often used for non-contraceptive reasons, the presence or absence of self-reported gynaecological conditions such as polycystic ovarian syndrome and endometriosis (yes; no) as well as the frequent experience of menstrual symptoms such as irregular periods, heavy period or severe period pain were also included (yes; no). All data for this project were obtained from the ALSWH (see www.alswh.org.au/ for further details), approved under Expression of Interest process (EOI A696) and provided in de-identified form. This project has ongoing ethical clearance from the University of Newcastle and University of Queensland’s Human Research Ethics Committees. Ethical approval for linkage of ALSWH survey data to the Admitted Patients Data Collections (APDC) was received from the NSW Population and Health Services Research Ethics Committee and other equivalent state and territory-based committees. Linkage to the National Death Index (NDI) was approved by the Australian Institute of Health and Welfare Ethics Committee. Women provided explicit written consent to participate in the ALSWH as well as for linkage to de-identified administrative health records prior to 2005. From 2005, an ‘opt-out’ consent process was approved by the data custodians and relevant ethics committees for data linkage, with participants regularly reminded of this process. ALSWH participants who decline health record linkage are excluded from data linkage requests. Contraceptive data were entered into separate latent transition models evaluating three to eight latent statuses. Establishing the optimal LTA model was conducted in a similar fashion to Harris et al. [ 19 ] and was based on clinical interpretability, latent class separation and goodness-of-fit statistics. Women were assigned a latent status at each survey they completed where they were at risk of an unintended pregnancy, up to a maximum of three time points. These latent statuses could be the same or vary over time. The probabilities of women transitioning being latent statuses over time are presented in S2 Table . Each latent status described one or more contraceptives being used concurrently. For example, Status 1 (“short-acting and condom”) refers to women who used short-acting contraceptive methods and condoms simultaneously. A classify-analyse approach was then used to assign each participant to a latent status at each time point, according to the latent status with the greatest posterior probability. Latent transition analysis was performed using PROC LTA procedure (The Methodology Centre, Penn State) in SAS 9.4 software (SAS Institute Inc). The number of time points with latent statuses with low probabilities (<2%) were minimised as this would contribute to numerical estimation issues in the subsequent regression models. Multinomial mixed-effect logistic regression models were developed, with the assigned latent status as the multinomial outcome, participant ID as a random effect to account for correlation between observations on the same participant over time, while controlling for time-varying covariates as fixed effects. We performed a complete case analysis with participants with missing data omitted from the analysis. Models were developed for the presence of any chronic condition and separately for each chronic disease.

(DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file. (DOCX) Click here for additional data file.