Exploring Upward and Downward Provider Biases in Family Planning: The Case of Parity

Provider Bias in Family Planning Research

Over the past 4 decades, provider bias has emerged as a central topic in family planning (FP) research. In 1985, Schuler et al. were among the first researchers to describe and empirically measure bias in patient-provider interactions in FP clinics. Using mystery clients in Nepal, the team reported that “lower class” patients were less likely to receive accurate information and high-quality treatment compared to “middle class” clients, indicating bias from providers based on clients’ socioeconomic position.¹ Bruce incorporated this “bias of the provider” in her 1990 landmark paper on quality of care, with Schuler et al.’s findings part of the foundation for the interpersonal relations pillar of Bruce’s quality framework.² In 1992, Shelton et al. published an article on medical barriers to FP access, delving further into provider bias. Shelton et al. argued that provider bias is deeply intertwined with economic and political interests and can serve as a mechanism that translates ideological opposition to contraception into restricted access.³

In the years since Shelton formalized the concept of provider bias, the topic has gained increasing prominence in FP research. Yet, the construct’s definition remained imprecise until an influential 2019 review by Solo and Festin defined provider bias as “attitudes and subsequent behaviors that unnecessarily restrict client access and choice, often related to either client and/or contraceptive method characteristics.”⁴ In addition to these recent efforts to more clearly conceptualize and define provider bias, researchers have also sought to measure provider bias empirically in countries including the United States,⁵ India,⁶ Nigeria,⁷ Kenya,^8,9 Senegal,¹⁰ and Burkina Faso.¹¹ These studies use a range of data collection tools and study designs, including in-depth provider interviews,⁵ surveys, focus group discussions, mystery clients,^1,9 patient exit interviews, and discrete choice experiments¹¹ to mitigate social desirability bias and attempt to accurately capture patient and provider perspectives on clinical interactions.

The literature has shed considerable light on the scope and magnitude of provider bias, particularly the ways in which provider attitudes can restrict access to desired FP services. For example, multiple studies in Kenya have documented how providers impose restrictions on certain contraceptive methods by requiring a minimum age for provision, refusing to offer methods to nulliparous women, and requiring HIV or pregnancy testing before providing the methods.^8,9 A discrete choice experiment in Burkina Faso, Pakistan, and Tanzania reported that marital status and parity may drive provider decisions to counsel and offer contraception more so than age does, although younger women are often most impacted by biased interactions.¹¹ A study in urban Nigeria found that levels of provider bias varied depending on which contraceptive method was sought and that in-service training did not substantially reduce biases in provider-patient interactions.⁷ A qualitative study in the United States showed that providers employ a variety of strategies to mitigate potential bias, yet these strategies can exacerbate, rather than reduce, bias in interactions.¹¹ With this growing body of methodologically and geographically diverse literature, scholars now better understand the ways sociodemographic characteristics of clients—including age, marital status, and parity—impact provider behavior and reveal bias due to beliefs about who should or should not be contracepting.^{4,6,7,9,10,12}

Expanding the Definition of Provider Bias: A Conceptual Framework

Within these studies, the dominant conception of provider bias has remained focused on biases that result in restricted access to contraception. Nearly all studies empirically documenting provider bias measure the ways in which providers prevent women from accessing contraceptive methods due to factors such as young age, low parity, and/or lack of marital union. However, there has been considerably less research into the types of provider bias that lead to unwanted contraceptive use.⁵

Drawing from previous work on the construct of contraceptive coercion,¹³ we propose here a distinction between 2 broad categories of provider bias: (1) “downward” provider bias that occurs when providers hinder FP access and choice through withholding or limiting services (in line with the conventional conception of provider bias); and (2) “upward” bias that occurs when providers pressure or encourage clients to use contraception (Box). Both forms of provider bias can occur through selective framing, prodding, judgment, or other means. Similar to contraceptive coercion,¹³ both upward and downward provider biases can fall along a spectrum, ranging from subtle forms of bias in which a provider passes judgment on a patient’s sociodemographic characteristics or contraceptive desires to more overt actions of pressure or force. For example, more subtle downward provider bias might include a provider questioning a client’s method choice because, for example, the client is unmarried but still offering the client their desired method when the client insists that this method is right for them. More overt downward provider bias might include a provider refusing to provide this unmarried client with their desired method altogether.

While downward provider bias has been well explored in FP research, upward provider bias remains an underexamined and potentially important threat to reproductive health and contraceptive autonomy.¹⁴ Several qualitative studies suggest that FP providers use upward provider bias, through directive and inaccurate contraceptive counseling, to encourage clients to adopt a modern contraceptive method generally and long-acting reversible contraception in particular.^13,15,16 Studies also show that providers may hold biases against long-acting reversible contraception removal, preventing women from making autonomous decisions about contraceptive discontinuation.^13,17,18 However, few studies have attempted to quantify upward provider bias. We posit that upward provider bias may be prevalent in FP programs, especially in the counseling of multiparous women during the postpartum period and for those residing in contexts that have adopted target-based approaches to increasing contraceptive uptake.^19,20

Upward and Downward Provider Biases Due To Parity: A Case Study

Downward provider bias based on providers’ perceptions of a patient’s low parity has been well documented. Nulliparous women or those with few children may be denied a desired contraceptive method because of a provider’s belief that they should have more children before they begin to restrict their fertility or because of a provider’s beliefs about the characteristics or side effects of the client’s preferred method.^8,9,21 In contrast to the scores of studies exploring this dynamic, the ways that multiparity or grand-multiparity may lead to bias in the provision of FP services have scarcely been explored. A qualitative exploration of contraceptive coercion in an anonymized country highlighted an example of this type of upward provider bias, in which after giving birth to her tenth child, a woman was told by health workers that she needed to get the implant “by force” due to her parity.¹³ Although there has been evidence to suggest that upward provider bias is a factor in global FP programs, to our knowledge, there are no studies that attempt to quantify its scope or describe when and why it occurs.

This article is motivated by focus group discussions and in-depth interviews with reproductive-aged women in Burkina Faso who described experiences of upward provider bias based on providers’ perceptions and judgments of their parity. We sought to expand the understanding of provider bias by identifying and measuring a form of more subtle upward provider bias that participants in qualitative interviews frequently described: provider encouragement to use contraception due to perceptions of high parity. Using cross-sectional survey data from reproductive-aged women at 2 sites in Burkina Faso, in this article, we describe lifetime prevalence of experiencing provider encouragement to use contraception because a provider perceived a participant to have “too many” children. We also examine the inverse: a form of downward provider bias focused on provider discouragement away from FP because a provider perceived a participant to not have “enough” children. We explore experiences of contraceptive counseling during postpartum care to further understand when and where upward provider bias occurs. Finally, we examined sociodemographic characteristics associated with experiencing provider encouragement to use contraception based on a provider’s perceptions of parity.

Sampling and Data Collection

The data used in this analysis come from the quantitative phase of this study: a cross-sectional, population-based survey based within the Nouna and Ouagadougou Health and Demographic Surveillance Systems (HDSS). The rural Nouna HDSS is made up of the small administrative town of Nouna and 58 of its surrounding villages, and the urban Ouagadougou HDSS comprises 5 neighborhoods in the northern section of the Burkinabè capital. Full profiles of these HDSSs—including information on their periodicity, demographic makeup, and methods—are available elsewhere.^25,26

To be eligible for the survey, women had to live within the catchment area of the Nouna or Ouagadougou HDSS, be of reproductive age (aged 15–49 years, inclusive), and provide informed consent in French, Dioula, or Mooré.²⁷ In brief, random samples of reproductive-aged women were drawn in each study site using the Nouna and Ouagadougou HDSS as sampling frames. In Nouna, an initial sample of 2,700 women and 800 potential replacements was drawn, with a response rate of 96.7%. In Ouagadougou, an initial sample of 1,300 women and 700 potential replacements was drawn. Due to a sampling error, the initial sampling frame included 811 women who were “visitors” rather than “residents” in the catchment area and thus ineligible for study inclusion. Thus, a new sample of 500 additional women was drawn. Total response rate in Ouagadougou across both sample drawings was 80.4%. Inverse probability of sampling weights were created and used throughout this analysis to account for changes in the sampling approach.

In total, survey data were collected from 3,929 women between April and July 2018. All adult participants provided written informed consent. For minors, parents provided informed consent with the assent from the minor.

Trained interviewers visited women at their homes and administered the survey orally in French, Mooré, or Dioula, with data recorded on Android-based tablets. Additional detailed information on the sampling strategy of the Contraceptive Autonomy Study can be found elsewhere.²⁴

Analytic Approach

We described sociodemographic characteristics of our sample overall and stratified by site, as well as the characteristics of only women with children. We presented mode of transport as a proxy for household wealth, which is often done in the Burkinabè context.²⁷ We examined attendance at the 45th-day postpartum visit. In Burkina Faso, the 45th-day postpartum visit (sometimes called the 40th- or 42nd-day visit) is a check-up that occurs roughly 6 weeks after birth, during which providers are supposed to discuss postpartum FP with clients. Though this visit is recommended for all who deliver, attendance is variable.

Next, we examined the lifetime prevalence of experiences with provider encouragement or discouragement to use or not use contraception due to provider perceptions of parity. We examined the lifetime prevalence of 1 form of downward provider bias—discouragement from using contraception due to provider perceptions of low parity—among all reproductive-aged women using the binary (yes/no) response to the question, “Has a provider ever discouraged you from using a method because they said you don’t have a child or enough children?” The primary outcome of interest was 1 form of upward FP provider bias—encouragement to use contraception due to provider perceptions of high parity. For this outcome, we restricted our sample to women who reported having at least 1 child. Provider encouragement to use contraception due to their perception of high parity was measured via the self-reported, binary (yes/no) response to the question, “Has a provider ever encouraged you to use a method because they said you have too many children?” We did not ask participants to describe how many children they thought providers perceived to be “not enough” or “too many.”

These questions were developed based on in-depth interviews and focus group discussions; therefore, wording of these questions was designed to mirror the ways that women spoke about provider bias they have experienced in their own lives. These measures of upward and downward provider biases were therefore quantified using women’s self-reported experiences with encouragement or discouragement to use contraception based on their parity. Rather than asking women to describe specific experiences with provider bias, our person-centered measurement approach trusted respondents to serve as experts on their own lives and health. Provider encouragement to use contraception during contraceptive counseling may not in and of itself constitute evidence of inappropriately directive counseling. However, our survey items asked specifically about encouragement and discouragement driven by provider perceptions of parity rather than a woman’s own contraceptive desires. This is in line with Solo and Festin’s definition of provider bias,⁴ leading us to suggest provider bias was at play. Women’s perceptions of their interactions with providers offered valuable insights into their experiences.

Informed by our qualitative findings, we also asked those who attended a 45th-day postpartum check-up whether they were asked first if they wanted to use contraception or just told to choose a method. This was determined by the question, “At the 45th-day visit, were you told to just choose a method, or were you asked if you wanted to use a contraceptive method first?” Response options included: told to choose a method, asked to choose a contraceptive method first, asked for a contraceptive method before the provider could mention, and other.

Finally, we used logistic regression to estimate odds ratios to examine sociodemographic factors associated with provider pressure to use contraception due to high parity among women with children. We examined several factors, including age (categorical: 15–24, 25–34, 35–49 years); education level (categorical; none, some primary school, at least some secondary school); marital status (binary; yes/no); primary mode of transport (as a proxy for material wealth; categorical; foot, bicycle, motorcycle, car, other); number of children (categorical: 1–2, 3–4, 5+); site (binary; Nouna, Ouagadougou); and whether an individual had a 45th-day postpartum visit (binary; yes/no). We examined multicollinearity between explanatory variables using variance inflation factor. Because only 41 participants reported that a provider ever discouraged them from using a method because they did not have “enough” children, we did not examine factors associated with this outcome.

Ethical Approval

This research was reviewed and approved by the Institutional Review Board of the Office of Human Research Administration at the Harvard T. H. Chan School of Public Health in Boston, MA, USA (#IRB17-0511); Le Comité d’Ethique pour la Recherche en Santé du Ministère de la Santé du Burkina Faso in Ouagadougou, Burkina Faso (#2017-5-067); and Le Comité Local d’Ethique du Centre de Recherche en Santé de Nouna, in Nouna, Burkina Faso (#2017-01).