Heterogeneity in referral preferences of women at high risk for postpartum depression: a discrete choice experiment

Using a discrete choice experiment with 698 participants, this study reveals significant heterogeneity in referral preferences among women at high risk for postpartum depression, identifying that convenience factors like cost and waiting time, alongside personal characteristics, are key drivers for uptake and highlighting the need for flexible, personalized referral programs.

The Gist

Imagine you are a new mother, exhausted and feeling down after giving birth. You've been told you might have postpartum depression (PPD), and a doctor suggests you see a specialist. But instead of jumping at the chance, you hesitate. Why?

Maybe the clinic is too far, the wait is too long, it costs too much, or you just don't want to go alone.

This study is like a giant "Choose Your Own Adventure" game designed to figure out exactly what makes new moms say "Yes" or "No" to getting help. The researchers didn't just ask, "Do you want help?" They asked, "If help looked like this (cheap but far away) versus that (expensive but right next door), which would you pick?"

Here is the breakdown of their findings, translated into everyday language.

🕵️‍♀️ The Setup: The "Menu" of Help

The researchers created a digital survey where 698 new moms in China had to make choices between different referral options. Think of it like ordering from a restaurant menu, but instead of pizza and pasta, the options were:

• Who calls you? (Your regular pregnancy nurse or a specialist doctor?)
• Where do you go? (A general women's clinic or a mental health specialist?)
• How do you get there? (In person or via video call?)
• How long do you wait? (1 hour or 24 hours?)
• How much does it cost? (100 yuan or 700 yuan?)
• Who comes with you? (Alone or with a family member?)

🏆 The Big Winners: What Moms Actually Wanted

When the researchers crunched the numbers, they found that Time and Money were the two biggest bosses in the room.

1. The "Speed Demon" Factor: Moms hated waiting. They were willing to pay extra just to skip the line and see a doctor in one hour instead of waiting a whole day. It's like paying for "Express Lane" at the grocery store; the time saved is worth more than the extra cash.
2. The "Wallet" Factor: Obviously, cheaper was better. But the study found that even if something was free, if the wait was too long, moms still said no.
3. The "Face-to-Face" Preference: Surprisingly, despite the pandemic making video calls popular, most moms still wanted to meet a doctor in person. They wanted a real human connection, not just a screen.
4. The "Specialist" Choice: Moms didn't want to go to a general women's clinic; they wanted to go straight to a psychiatric clinic (a mental health specialist). They weren't scared of the stigma; they just wanted the expert.

🎭 The Four Types of Moms (The "Personality Clusters")

The most interesting part of the study is that not all moms are the same. The researchers used a special math trick to group the moms into four distinct personality types, like different characters in a video game:

• 1. The "Time-Sensitive" (23%): These moms are in a rush. They are like people who check their watches every 30 seconds. They want the fastest possible appointment, even if it means doing it online or paying a bit more. Speed is their only priority.
• 2. The "In-Person & Time-Sensitive" (42% - The Biggest Group!): This is the largest group. They want speed and they want to see a real doctor face-to-face. They are the "Goldilocks" group: they want it fast, personal, and convenient.
• 3. The "Frugal" (14%): These moms are the ultimate budget shoppers. They care almost exclusively about the price tag. If it's cheap, they don't mind waiting a bit longer or going to a different clinic. They are looking for the best deal.
• 4. The "Specialist & Family-Oriented" (21%): These moms want the best of the best. They want to see a top-tier psychiatrist and they want their husband or mother-in-law to walk in the room with them for support. They value quality and emotional safety over speed or cost.

💡 The "Aha!" Moment

The study found that one size does not fit all.

If a hospital creates a referral program that is "cheap but slow," they will please the "Frugal" group but lose the "Time-Sensitive" group. If they make it "fast but expensive," they lose the budget-conscious moms.

The Solution?
To get more moms to get help, hospitals need to offer a flexible menu.

• For the "Time-Sensitive" moms: Offer a "Express Lane" with quick video or in-person slots.
• For the "Frugal" moms: Offer a subsidized, low-cost option, even if the wait is slightly longer.
• For the "Family-Oriented" moms: Make sure the clinic is welcoming to family members and has top-tier specialists.

🏁 The Bottom Line

Postpartum depression is a heavy burden, but the barrier to getting help isn't always fear or shame. Sometimes, it's just bad logistics.

The study tells us that to fix the problem, we need to stop treating all new moms the same. We need to build a referral system that is as flexible as a new mom's schedule—offering speed for some, savings for others, and family support for everyone. When the "how" of getting help matches what the mom actually needs, she is much more likely to say "Yes."

Technical Summary

1. Problem Statement

Despite the high global prevalence of Postpartum Depression (PPD), referral uptake rates for diagnosed women remain critically low. In China, for instance, screening positive rates are significant (approx. 14.8%), yet referral acceptance is as low as 0.4% in some cohorts. While qualitative studies have identified barriers to referral (e.g., stigma, lack of support), there is a lack of quantitative evidence regarding the specific procedural attributes that drive or hinder a woman's decision to accept a referral. Current referral programs are often generic, failing to account for the heterogeneity in patient preferences regarding cost, time, location, and mode of care.

2. Methodology

The study employed a Discrete Choice Experiment (DCE) to quantitatively elicit stated preferences among women at high risk for PPD.

• Study Design & Sampling:

  • Location: Changsha, China (stratified cluster sampling across two districts).
  • Participants: 2,134 new mothers screened within 7 days of delivery. Those scoring $\ge$ 10 on the Edinburgh Postnatal Depression Scale (EPDS) were invited.
  • Final Sample: 617 participants (after excluding 81 who failed consistency tests) from an initial 698 completions.
  • Data Collection: Conducted via home visits by trained maternal healthcare providers using a mobile app (WeChat/So Jump).

• Experimental Design:

  • Attributes: Seven key attributes were defined based on prior qualitative research, expert panels, and focus groups:
    1. Referral Initiator: Obstetrician vs. Maternal Healthcare Officer.
    2. Referral Destination: Maternal Clinic vs. Psychiatric Clinic.
    3. Referral Route: Online vs. In-person.
    4. Waiting Time: 1, 6, 12, or 24 hours.
    5. Accompaniment: Alone vs. With family.
    6. Consultation Length: 1 hour vs. 2 hours.
    7. Expenses: ¥100, ¥300, ¥500, ¥700.
  • Task: Participants chose between two referral profiles (and an opt-out option) across 16 choice sets.

• Statistical Analysis:

  • Mixed Logit Model (MLM): Used to estimate average preferences and account for preference heterogeneity (random coefficients).
  • Latent Class Analysis (LCA): Used to identify distinct subgroups (classes) of participants with similar preference structures.
  • Willingness-to-Pay (WTP): Calculated by taking the ratio of attribute coefficients to the cost coefficient to determine the monetary value participants place on specific attributes.

3. Key Contributions

• First DCE on PPD Referrals: This is the first study to use a DCE to specifically analyze the referral process attributes (rather than just treatment outcomes) for PPD.
• Quantification of Heterogeneity: It moves beyond aggregate averages to identify distinct patient segments, demonstrating that a "one-size-fits-all" referral strategy is suboptimal.
• Contextual Insight: It challenges the assumption that obstetricians are the primary preferred initiators of referrals, highlighting the critical role of maternal healthcare officers in the Chinese context.
• Policy-Relevant Metrics: Provides specific WTP values and relative importance statistics that can guide the design of cost-effective, high-uptake referral pathways.

4. Key Results

A. Aggregate Preferences (Mixed Logit Model)

• Cost and Time are Dominant: Expenses had a significant negative coefficient ($\beta = -0.002$), indicating lower uptake with higher costs. Waiting time was the most influential non-economic factor; a 1-hour wait was highly preferred ($\beta = 0.594$).
• Mode and Destination: Participants preferred in-person consultations and referrals to psychiatric clinics over maternal clinics.
• Initiator: Referrals initiated by maternal healthcare officers were preferred over those by obstetricians.
• Accompaniment: The "With-family" attribute was generally not statistically significant in the aggregate model, though this varied by subgroup.
• Demographic Interactions: Higher education (Bachelor's+) increased referral preference. Conversely, higher EPDS scores, public sector employment, older age, and satisfaction with one's partner were associated with lower referral uptake.

B. Willingness-to-Pay (WTP)

Participants were willing to pay significant premiums for convenience:

• Waiting Time: ¥370 more for a 1-hour wait vs. 24 hours.
• Consultation Length: ¥126 more for a 1-hour consultation vs. 2 hours.
• Route: ¥65 more for in-person vs. online.
• Destination/Initiator: Marginal WTP for psychiatric clinics (¥49) and maternal officers (¥58).

C. Latent Class Analysis (Heterogeneity)

Four distinct classes were identified, accounting for the majority of the sample:

1. Time-Conscious (23.1%): Prioritized speed above all else. Highest WTP for 1-hour waiting (¥525).
2. Time-Sensitive & In-Person (42.1% - Largest Group): Valued both short waiting times and face-to-face interaction. Preferred psychiatric clinics and maternal officers.
3. Frugal (13.6%): Driven almost exclusively by cost (Relative Importance Statistic = 0.532). Other attributes had minimal impact.
4. Specialist & Family-Oriented (21.2%): Strongly preferred psychiatric clinics and family accompaniment. Highest WTP for psychiatric referral (¥349) and family presence (¥217).

5. Significance and Implications

• Tailored Interventions: The findings suggest that increasing referral uptake requires personalized strategies. For example, "Frugal" patients need cost subsidies, while "Time-Conscious" patients need rapid-access pathways.
• Resource Allocation: The strong preference for psychiatric clinics indicates a need to decentralize mental health resources or improve the perceived accessibility of these facilities, rather than relying solely on maternal clinics.
• Referral Mechanism: Empowering maternal healthcare officers (who have frequent contact with mothers) rather than obstetricians to initiate referrals could significantly improve acceptance rates.
• Social Support: The "Specialist & Family-Oriented" group highlights the importance of involving family members in the referral process, suggesting that mental health education should extend to the patient's social circle.

In conclusion, the study demonstrates that PPD referral uptake is driven by a complex interplay of economic, temporal, and social factors. To improve public health outcomes, referral programs must be flexible, offering options that cater to the specific preferences of different patient subgroups, particularly focusing on reducing wait times and costs.