Barriers To Accessing Contraception and Cervical and Breast Cancer Screening During COVID-19: A Prospective Cohort Study

Abstract

Objective

This study sought to examine how access to contraception and cervical and breast cancer screening in British Columbia, Canada, has been affected by the COVID-19 pandemic.

Methods

From August 2020 to March 2021, 3691 female residents of British Columbia (age 25–69 y) participated in this study. We used generalized estimating equations to analyze the proportion of females accessing contraception and the proportion having difficulty accessing contraception across the different phases of pandemic control measures, and logistic regression to analyze attendance at cervical and breast cancer screening. We added sociodemographic and biological variables individually into the models. Self-reported barriers to accessing contraception and attending screening were summarized.

Results

During phases with the highest pandemic controls, self-reported access to contraception was lower (OR 0.94; 95% CI 0.90–0.98; P = 0.005) and difficulty with access was higher (OR 2.74; 95% CI 1.54–4.88; P = 0.001). A higher proportion of adults aged 25–34 years reported difficulty accessing contraception than those aged 35–39 years (P < 0.0001), and participants identifying as Indigenous had higher odds of access difficulties (OR 5.56; 95% CI 2.44–12.50; P < 0.001). Of those who required screening during the COVID-19 pandemic, 62% and 54.5% did not attend at least one of their cervical or breast screening appointments, respectively. Those with a history of breast cancer had significantly higher odds of self-reporting having attended their mammogram appointment compared with those without a history of breast cancer (OR 5.62; 95% CI 2.69–13.72; P < 0.001). The most common barriers to screening were difficulty getting an appointment and appointments being considered non-urgent.

Conclusions

The COVID-19 pandemic has uniquely affected access to contraception and cancer screening participation for various subgroups. Self-reported data present potential avenues for mitigating barriers.

A. Baaske

Introduction

Internationally, women’s access to contraception and to cervical and breast cancer screening has been disrupted1, 2, 3 as a result of the pandemic-associated public health measures designed to limit the spread of SARS-CoV-2.^{4 , 5} Interrupted access to contraception can result in consequences such as unintended pregnancy.⁶ Delays in cancer screening can result in later state disease detection and excess morbidity and mortality. For instance, the 5-year survival rate for cervical cancer decreases from 93% (stage 1A) to 80% (stage 1B) and further to 15% (stage 4B) when diagnosis and treatment starts at later stages.⁷

Though there are international data, the effect of the pandemic on contraception and cancer screening access has not been well characterized for women in Canada. Other topics currently underexplored include whether access to contraception and cancer screening has been differentially impacted based on sociodemographic factors, as physical and mental health outcomes have during the pandemic,^{8 , 9} and barriers to health-seeking behaviours. Furthermore, across jurisdictions, the pandemic public health response has been tailored to prevent hospitalization and infection rates,¹⁰ resulting in a series of lockdowns followed by periods of relaxed restrictions. Yet little research has explored the impact of these different “phases” of control measures on access to reproductive care and screening services.

To address these gaps, we examined access to birth control and emergency contraception across different phases of the pandemic, and attendance at the Pap test for cervical screening and mammography for breast screening across a 1-year period in British Columbia (BC). Based on the published literature,1, 2, 3 we hypothesized decreases in access and attendance, specifically during the phases of heightened pandemic controls. We hypothesized that factors such as younger age, non-White ethnicity, rural living, lower household income, lower education level, immigrant status, Indigenous ancestry, nonheterosexual orientation, and nonbinary/trans gender identity,^{1 , 2 ,} 11, 12, 13 as well as no history of breast or cervical cancer, would be predictors of access difficulties. Finally, we expected that difficulty getting an appointment and fears about exposure to SARS-CoV-2 would be commonly reported as barriers to access.¹⁴

Methods

Setting and Participants

The current study was a part of a larger study from Women's Health Research Institute (full study methods reported elsewhere^{15 , 16}), which invited residents of BC aged 25 to 69 years to participate in an online survey on the impact of the COVID-19 pandemic between mid-August 2020 and March 1, 2021.¹⁷ The participants were stratified into 9 5-year age strata and recruitment continued until a target of n = 750 was reached for each stratum. Public recruitment occurred through social media and research platforms. The recruitment ended for individuals aged 45 to 64 years in mid-November 2020 and individuals aged 40 to 44 years in mid-December after targets were reached for these strata. Ethical approval was received from BC Children’s and Women’s Research Ethics Board (approval number H20-01421).

Survey Design and Measures

We asked the participants to respond retrospectively across 5 phases of the pandemic, delineated by pandemic control measures, and prepandemic for specific items. The phases were as follows: pre-COVID (December 2019 to mid-March 2020), phase 1 (mid-March 2020 to mid-May 2020), phase 2 (mid-May 2020 to November 30, 2020), phase 3 (mid-May 2020 to August 31, 2020), phase 4 (September 1, 2020 to October 31, 2020), and phase 5 (November 1, 2020 to March 1, 2021).¹⁰ The participants who completed the survey before November 30, 2020, were asked to respond based on pre-COVID outcomes as well as phase 1 and phase 2. On November 30, 2020, phase 2 was removed and phases 3 to 5 were added.

For each phase, the participants responded to 4 contraception outcome measures: (1) regularly accessed birth control methods (e.g., birth control pill), (2) accessed emergency contraception (e.g., Plan B), (3) had difficulty accessing birth control methods, and (4) had difficulty accessing emergency contraception. Those who indicated that they had difficulty accessing contraception at any point during phases 1 through 5 were asked “What made accessing birth control or emergency contraceptives difficult during COVID-19?” and selected all that applied from a list of options ( Table S1). The participants were eligible for contraception outcomes analysis if they responded to ≥1 outcome measure across all phases and were of female sex, not postmenopausal, and not pregnant. Although it is possible for females >55 to still be menstruating, we also restricted contraception analyses to those aged ≤55 years.

We asked all survey participants to report on their attendance (attended all, some, or none) at cervical or breast screenings during the pandemic. The participants who reported that they only attended some or none of their appointments were asked “Why did you not attend the appointment(s)” and were able to select from a list of reasons (options were mutually exclusive). The listed reasons included: my doctor or clinic was not accepting in-person appointments, my appointment was considered “nonurgent,” worried about visiting doctors or doctors’ office, and other. The participants who selected “other” were able to provide a free-text response. For the analyses of cervical (Pap) and breast (mammogram) cancer screening outcomes, the eligibility was restricted to female sex and those who responded that during the COVID-19 pandemic (mid-March to now) they were due for, or in need of, a Pap test/mammogram.

We collected sociodemographic information including sex assigned at birth, postmenopausal status (have not had a period in ≥1 year), age, ethnicity (White and non-White), Indigenous ancestry, and other demographics (online Appendix). Indigenous ancestry was self-reported and assessed separately from ethnicity.

Data Analysis

All analyses were carried out in R v. 4.0.3.¹⁸ The significance threshold was set at P < 0.05. Missing data were excluded from analyses.

Access to birth control/emergency contraception was pooled due to a small proportion of respondents indicating accessing emergency contraception (1.7% pre-COVID, dropping to <1% in the later phases). By using generalized estimating equations with a first order autoregressive /cicorrelation structure using the “geepack” package^{19 , 20} to account for the repeated measures and including age as a covariate, we estimated both the proportion accessing contraception and the proportion having difficulty with access across the phases. We investigated the relationships between access, having difficulty accessing contraceptives, and sociodemographic variables by adding them individually to the models.

Attendance at required cervical or breast screening was dichotomized into attendance at none/some versus all appointments. The proportion attending all visits was analyzed using logistic regression. The reasons for nonattendance were reported as the percent of those who attended some/none of their appointments. We conducted thematic analyses on the open-ended responses for reasons for nonattendance at breast and cervical screening. We used a deductive, essentialist approach and constructed themes around semantic content by analyzing the surface meaning of each response.²¹

Results

In total, 3691 participants who provided informed consent for our larger survey were eligible for this study, and among these, we had sample sizes n = 2542 for contraception outcomes, n = 1077 for Pap test outcomes, and n = 1226 for mammogram outcomes (online Appendix). A participant could fall in more than 1 of these 3 cohorts.

Access to Contraception During COVID-19

After controlling for age, there was a significant relationship between the pandemic phase and both the proportion of females that accessed (P < 0.02) and that had difficulty accessing (P < 0.001) birth control/emergency contraception. During phase 1, fewer people accessed contraception (P = 0.005), and during phases 1 and 2 more people had difficulty accessing contraception (both P = 0.001) than during pre-COVID. Across all phases, we found a significant effect of age, with younger participants (25–34 years) proportionately more likely to report that they were accessing contraception (P < 0.0001) and had difficulty accessing contraception (P < 0.0001). Across the phases, <3% of those aged 35 to 55 years reported difficulties with access (Table 1 , Supplemental Figures S1 and S2).

Table 1.

Impact of age and COVID-19 phase on contraception access

Predictor	Using contraception		Difficulty accessing contraception
	Adjusted odds ratio (95% CI)	P value	Adjusted odds ratio (95% CI)	P value
Age group, y
25–29	Reference		Reference
30–34	0.70 (0.47–1.02)	0.065	0.63 (0.25–1.58)	0.325
35–39	0.37 (0.26–0.53)	< 0.001	0.40 (0.18–0.89)	0.025
40–44	0.31 (0.22–0.44)	< 0.001	0.24 (0.10–0.59)	0.002
45–49	0.26 (0.18–0.38)	< 0.001	0.18 (0.06–0.52)	0.002
50–55	0.14 (0.08–0.22)	< 0.001	0.00 (0.00–0.00)	< 0.001
Phase
Pre-COVID	Reference		Reference
Phase 1	0.94 (0.90–0.98)	0.005	2.74 (1.54–4.88)	0.001
Phase 2	0.96 (0.90–1.02)	0.218	3.20 (1.66–6.18)	0.001
Phase 3	0.99 (0.90–1.08)	0.761	1.75 (0.68–4.48)	0.243
Phase 4	1.09 (0.96–1.24)	0.206	1.59 (0.59–4.27)	0.359
Phase 5	1.18 (1.02–1.36)	0.025	2.14 (0.89–5.15)	0.090

After controlling for age and phase, we found a significant relationship between Indigenous status and the proportion who had difficulty accessing contraception, with self-identified Indigenous participants having significantly higher odds of experiencing difficulty than non-Indigenous participants (P < 0.001). Further, we found that nonheterosexual participants had a significantly lower prevalence of accessing contraception than heterosexual respondents across all phases (P = 0.001), but there was no impact of sexual orientation on difficulties accessing (Table 2 ).

Table 2.

Bivariable results of sociodemographic factors assessed individually on the proportion using/with difficulty accessing contraception

Predictor	Using contraception		Difficulty accessing contraception
	Adjusted odds ratio (95% CI)	P value	Adjusted odds ratio (95% CI)	P value
Geographic region
Fraser	Reference		Reference
Interior	1.33 (0.85–2.08)	0.213	1.09 (0.40–2.93)	0.868
Northern	0.75 (0.36–1.57)	0.444	2.14 (0.47–9.70)	0.324
Vancouver Coastal	1.32 (1.03–1.70)	0.028	0.60 (0.29–1.25)	0.172
Vancouver Island	1.13 (0.83–1.53)	0.433	0.68 (0.29–1.60)	0.374
Ethnicity
White	Reference		Reference
Non-White	1.32 (0.97–1.79)	0.073	0.78 (0.37–1.64)	0.506
Household income, CAD$
<10 000–20 000	Reference		Reference
20 000–40 000	0.7 (0.37–1.31)	0.265	0.64 (0.19–2.14)	0.465
40 000–60 000	1.09 (0.61–1.97)	0.764	0.73 (0.23–2.31)	0.594
60 000–80 000	0.98 (0.55–1.75)	0.938	0.38 (0.09–1.60)	0.186
80 000–100 000	0.57 (0.32–1.01)	0.055	0.18 (0.03–0.98)	0.047
100 000–150 000	1.06 (0.62–1.82)	0.828	0.50 (0.15–1.67)	0.260
>150 000	0.94 (0.55–1.61)	0.835	0.36 (0.11–1.20)	0.096
Education
More than high school	Reference		Reference
High school or less	0.82 (0.58–1.17)	0.278	0.97 (0.34–2.75)	0.955
Gender identity
Cis-gender	Reference		Reference
Non–cis-gender	0.53 (0.22–1.29)	0.160	0.51 (0.06–4.03)	0.524
Sexual orientation
Heterosexual	Reference		Reference
Nonheterosexual	0.62 (0.46–0.82)	0.001	0.79 (0.39–1.61)	0.519
Immigrant status
Immigrated <5 y ago	Reference		Reference
Immigrated ≥5 y ago	1.24 (0.52–2.93)	0.632	1.47 (0.18–12.32)	0.721
Nonimmigrant	1.24 (0.64–2.39)	0.521	1.38 (0.25–7.66)	0.713
Indigenous status
Non-Indigenous	Reference		Reference
Indigenous	0.95 (0.58–1.56)	0.849	5.56 (2.44–12.50)	< 0.001

Note: All sociodemographic variables were adjusted for phase and age.

For those who had difficulty accessing contraception at any point (78 respondents), difficulty getting an appointment (37%) and worries about COVID-19 exposure (28%) were the top barriers to access. Other barriers to access endorsed by respondents are reported in, Table S1 in online Appendix.

Cervical Screening Attendance Throughout the Pandemic: Factors and Barriers

Of the sample who self-reported requiring a cervical screening (Pap test), 37.0% attended all, 9.9% attended some, and 52.1% attended none of their appointments. None of the sociodemographic factors nor history of cervical cancer affected the attendance at the cervix screenings (all P > 0.05).

For the respondents who indicated that they attended some or none of their cervical screening appointments (668 respondents), primary endorsed reasons were as follows: my doctor or clinic was not accepting in-person appointments (32.5%), my appointment was considered nonurgent (25.3%), worried about visiting doctors or doctors’ office (24.6%), other (17.7%). For the 17.7% of respondents (118) who endorsed “other” and provided an open-ended response, thematic analysis revealed 2 overarching theme categories of “personal barriers” and “pandemic control measures,” each with several themes and subthemes (Figure 1).

Figure 1.

Thematic map of free-text responses to the question “Why did you not attend the (Pap test) appointment(s)?” Note: n values represent the number of participant responses within subthemes.

Breast Screening Attendance Throughout the Pandemic: Factors and Barriers

Of the sample who self-reported requiring breast screening (mammogram), 44.8% attended all, 10.7% attended some, and 43.8% attended none of their appointments.

Those with a history of breast cancer had significantly higher odds of self-reporting attending their mammogram during COVID-19 compared with those who did not (odds ratio 5.62; 95% CI 2.69–13.72; P < 0.001). There was no significant relationship between the mammogram attendance and any sociodemographic factor analysed (all P > 0.05). Appointments being considered “nonurgent” was the most endorsed primary reason for nonattendance at mammogram (30.1%), followed by: other (27.4%), worried about visiting doctors or doctors’ office (23.1%), my doctor or clinic was not accepting in-person appointments (19.3%).

For those that selected “other” and provided an open-ended response (183 respondents), thematic analysis revealed overarching theme categories of “personal barriers” and “pandemic control measures,” each with several themes and subthemes (Figure 2).

Figure 2.

Thematic map for free-text responses to the question “Why did you not attend the (mammogram) appointment(s)?” Note: n values represent the number of participant responses within subthemes. Responses were excluded (cervical analysis: n = 11, breast analysis: n = 7) if respondents used the free-text space to discuss something other than barriers to attendance.

Discussion

Our study showed a correlation between high levels of pandemic controls and a higher prevalence of difficulty accessing contraception, which continued even once the pandemic control measures loosened. This finding is unlike past findings¹ and may have important implications for the longer-term effects of strict controls on female public health. Our findings did not support the notion that nonheterosexual woman are more likely to face barriers to contraception access during COVID-19 than heterosexual women.¹ Rather, we found that nonheterosexual females reported accessing birth control less, consistent with the generally lower prevalence of contraception use among LGBTQ women.²²

Our study also described the low rates of breast and cervical screening. Before the pandemic, between 2017 to 2019, participation in the BC Cancer Breast Screening Program was approximately 50% and cervical screening participation was approximately 68% (corrected for hysterectomy rate).^{23 , 24} Therefore, our data suggested 9% and 93.75% increases in nonparticipation in breast and cervical cancer screening during COVID-19, respectively. To clear the large queue of missed screenings and support participation during future disruptions, clear communication around scheduling, messaging on the importance of timely screening, and guidance on screening participation for individuals without a family doctor are points to consider. The barriers to cervical screening attendance identified in this analysis highlighted an opportunity for innovative approaches that minimize in-person contact, such as telehealth with self-collected screening options.²⁵ Those without a history of breast cancer may have been less likely to seek breast screening during the COVID-19 pandemic due to less motivation among cancer-free populations to be screened for cancer.²⁶ Mammography participation may have also been disrupted due to deferrals of radiology appointments perceived as “less urgent,”⁷ which is consistent with other findings that show larger volume drops for screening mammography compared with diagnostic mammography during COVID-19.²⁷

Others have documented both within and outside of the COVID-19 pandemic that other sociodemographic factors, such as ethnicity, are predictors of health care access.^{1 ,} 11, 12, 13 Future research should assess the impact of the pandemic on groups previously demonstrated to face inequitable health care access using representative samples to the overall target population.

The limitations of our study were the retrospective nature of the questions, which could have introduced recall error, and the confinement of this study to the BC population. This study included a population with a higher percentage of respondents who identified as White, with more than a high school education, and who were more likely to live in the southern part of the province compared to the general population of BC (online Appendix),²⁸ limiting the generalizability to similar populations. As well, the level of contraception access was not specified, which could have limited interpretability. For example, we do not know whether “difficulty getting an appointment” occurred at the level of initial contact with a clinic or due to clinics being over capacity. It is also unclear whether respondents may have changed the type of contraception they used, such as turning from contraception methods that require insertion (e.g., intrauterine devices) or hormonal methods that require a prescription to nonhormonal contraception (e.g., barrier methods). Because the question about birth control was in brackets (“e.g., birth control pill”), respondents may have assumed that it was referring to hormonal methods as opposed to all methods. Of note, it would still be an issue if females turned to less effective contraception such as nonhormonal methods, but this was not captured by our survey. The use of self-reported data was another limitation. For example, respondents may have self-reported requiring screening when in fact they were not due for an appointment as per BC guidelines.^{29 , 30}

Conclusion

This study provided critical data on the barriers to females’ access to contraception and cancer screening in Canada during a year of unprecedented health care disruptions. Self-reported barriers to screening attendance presented in this study offer potential avenues for increasing cervical and breast cancer screening participation. These data can inform health leadership on the impacts of restricting health service delivery and of demographics warranting thoughtful consideration as they navigate pandemic recovery and future planning.