What role can policy play in reducing bias among healthcare providers?
Editor’s note: This episode of VoxDevTalks is also available on Spotify, Apple Podcasts, and YouTube.
In this episode of VoxDevTalks, host Tim Phillips speaks with Zachary Wagner and Manisha Shah about Beyond Bias: an ambitious, multi-country project to tackle prejudices held by healthcare providers delivering family planning services. Conducted in Burkina Faso, Pakistan, and Tanzania, the initiative aimed to measure, understand, and reduce, provider bias—especially against young, unmarried, or childless women seeking contraception.
The conversation explores why these biases exist, how they manifest in contraceptive care, the design of the intervention, and the measurable impact it had on attitudes, counselling quality, and client experience.
Understanding the causes and consequences of provider bias
Healthcare providers, despite their medical training, often bring their own social norms and cultural beliefs to work. In the context of family planning, these biases typically relate to age, marital status, and parity (whether a woman has given birth).
For example, young unmarried women are frequently excluded from outreach programmes; recently married but childless women may be discouraged from using contraception. This leads to poor quality care and health outcomes, discouraging many women from visiting clinics altogether.
“It can lead to poor quality care, poor health outcomes, it can exacerbate existing disparities that we already observe in health settings… it can also just discourage people from seeking care.” Shah
A multi-component intervention to tackle provider bias
Beyond Bias targeted young women aged 15–24 and was implemented in 227 clinics across three countries. The intervention had three components:
- One-day training: Introducing concepts of provider bias and its impact on family planning.
- Ongoing peer support: In-person and via WhatsApp groups, to reinforce learning and encourage discussion.
- Non-financial awards: Performance-based recognition for clinics improving services for clients under 25, including certificates signed by the Minister of Health.
“The idea was to foster competition amongst the service providers to provide better care to young people.” Shah
The intervention was informed by extensive human-centred design and qualitative research with providers and clients to ensure cultural and operational fit.
Measuring impact using surveys, mystery clients, and indices
The research team used three main data sources:
- Provider surveys: Roughly four per clinic, capturing self-reported attitudes and behaviours across 47 bias-related characteristics.
- Over 70,000 patient exit surveys: Assessing counselling quality and client experience.
- Mystery client visits: Trained actors presented with different combinations of age, marital status, and parity to detect differential treatment.
This mix of self-reporting, direct observation, and client feedback helped reduce bias in measurement.
Results: Significant attitude shifts, nuanced behavioural change
The intervention produced a two-standard-deviation reduction in self-reported provider bias—a result consistent across all three countries. Providers became more likely to counsel clients on the full range of contraceptive methods, and women felt they were treated more respectfully.
However, there was no significant change in the actual methods received by real clients, largely because 90% were already using contraception before the intervention. The mystery clients—who represented women more likely to face bias—reported they could access a wider range of methods in treatment clinics.
The results varied by subgroup. For nulliparous women (i.e. those who had never given birth before), the disparity in counselling on hormonal methods was “pretty much wiped out”, indicating a substantial reduction in bias against women who had not given birth. Among young women, there remained a persistent gap in being offered any method, although there was a slight improvement observed in Tanzania. In contrast, unmarried women continued to report worse treatment than their married counterparts, with no measurable improvement detected.
Leveraging qualitative insights in empirical work
The qualitative work at the start shaped the intervention, while follow-up interviews helped explain the findings. Providers reported structural barriers to fully implementing what they had learned, such as lack of private spaces for youth consultations.
“We do see big attitudinal changes coming from the providers that didn’t necessarily manifest into big changes in service delivery… part of this could have been the result of some of these structural issues that we learned about in the qualitative data.” Shah
This revealed a gap between changing attitudes and practice, highlighting areas for policy and infrastructure support.
Lessons for future family planning interventions
Although the intervention delivered greater shifts in provider attitudes than anticipated, researchers emphasise several areas for improvement:
- Financial incentives may have more direct impact than non-financial awards.
- Component testing: Untangling which of the three intervention components drive change could improve cost-effectiveness.
- Encouraging youth attendance: Attracting more young women to clinics remains a key challenge.
Importantly, the consistency of results across three very different cultural and religious contexts suggests that Beyond Bias is broadly adaptable.
“There are some aspects of bias that we won’t be able to move with this intervention… For example, unmarried women in Pakistan, that was really just a non-starter… that’s not really an element of provider bias that we’d be able to change in that context.” Wagner
Policymaker engagement will depend on context. For instance, in sub-Saharan Africa, high rates of unintended youth pregnancy and HIV among young people may make this a higher priority than in South Asia. Collaboration with large non-government organisations has already opened doors for scaling discussions with health agencies in developing countries.