A novel methodology to measure waiting times for community-based specialist care in a public healthcare system

Rachel Wilf‐Miron, Ilya Novikov, Arnona Ziv, Avishai Mandelbaum, Ya’acov Ritov, Osnat Luxenburg

TL;DR

Researchers have developed a novel, low-cost methodology to measure national healthcare waiting times by mining data from existing computerized scheduling systems. By distinguishing between patients seeking a "specific" doctor versus "any" available specialist, the study reveals that choosing a specific physician can increase wait times by up to 50%, while highlighting significant geographic disparities in access to care.

What Problem Does This Paper Address?

In public healthcare systems, "timeliness" is a pillar of quality, yet it is notoriously difficult to measure accurately. Long waiting times (WT) lead to poorer health outcomes, lower patient satisfaction, and a shift toward expensive private financing.

Current measurement methods are often flawed:

  • Surveys: Expensive, infrequent, and subject to recall bias.
  • "Secret Shoppers": Provide only a snapshot and are hard to scale.
  • Administrative "First Available" Metrics: Often underestimate wait times because they don't account for the volume of patients competing for those slots or the impact of last-minute cancellations.

The authors set out to create a standardized, national-level metric for Israel’s public healthcare system that utilizes existing data to reflect the actual experience of a patient trying to book a non-urgent specialist appointment.

Methods

The researchers introduced a metric called Offered Wait Time (OWT). Instead of just looking at the single "next available" slot, their algorithm utilizes the first 50 available appointments from every physician’s book.

The core of their approach relies on two key components:

  1. The "Steady State" Assumption: They assume that in a stable system, supply equals demand. They estimated "demand" (the number of patients looking for an appointment daily) by looking at the "supply" (the average number of actual visits that occurred at that practice).
  2. Two Patient Scenarios:
    • Specific Physician: The patient wants a "named" doctor (prioritizing continuity of care).
  3. Any Physician: The patient is willing to see any specialist within their geographical region (prioritizing speed).

By matching the daily demand to the corresponding number of available slots in the scheduling system, the algorithm calculates a distribution of wait times that is more representative of the "real-world" queue than a simple "next available" data point.

Datasets

The study utilized a massive administrative dataset from all four Health Maintenance Organizations (HMOs) in Israel:

  • Appointments: 1,017,478 available appointments extracted daily over 31 days in December 2018.
  • Practices: Data from 6,040 community-based physician practices.
  • Specialties: Five high-volume areas: orthopedics, ophthalmology, gynecology, dermatology, and otolaryngology (ENT).
  • Geography: Analysis was conducted at the town, "natural area" (52 regions), district, and national levels.

Key Insights and Ideas

  • The Price of Preference: There is a clear trade-off between continuity of care and speed. Patients who insist on a specific doctor wait significantly longer. This "voluntary delay" is a critical factor that previous administrative metrics often ignored.
  • Supply as a Proxy for Demand: In the absence of "intent-to-book" data (which many IT systems don't track), using the average number of actual visits to estimate daily demand provides a clever workaround for calculating wait distributions.
  • Overcoming IT Fragmentation: The methodology provides a "common denominator" that works across different HMOs with varying IT infrastructures, allowing for the first-ever unified national comparison in Israel.
  • Geographic Inequity: The data exposed "healthcare deserts." For example, the 75th percentile wait time for an ophthalmologist in the Southern district was 33 days, compared to just 11 days in Tel Aviv.

Key Results

  • National Medians: For a specific physician, median wait times ranged from 6 days (ophthalmology) to 13 days (dermatology).
  • The "Any Physician" Advantage: Wait times for the "any physician" scenario were 28% to 50% shorter than for a specific physician. For example, the median wait for gynecology dropped from 8 days (specific) to 4 days (any).
  • Mean vs. Median: Mean wait times were significantly higher than medians (e.g., 21.7 days for gynecology), indicating a "long tail" where some patients wait a very long time for high-demand experts.
  • Validation: The results closely mirrored independent consumer surveys conducted during the same period, suggesting the algorithm is a highly accurate, lower-cost alternative to traditional polling.

Limitations and Future Work

The authors acknowledge several limitations:

  • Static Demand: The model assumes demand is identical every day of the month, ignoring spikes on Sundays or lulls during holidays.
  • Walk-ins and "Give-ups": The "steady state" assumption doesn't perfectly account for "walk-in" patients (who have 0 wait time) or the 19% of patients who give up on booking entirely due to long waits. The authors suggest these factors might partially cancel each other out.
  • Snapshot vs. Stream: Data was extracted once daily. A more "ideal" system would continuously stream data to capture cancellations and re-bookings in real-time.

Future iterations of this national measurement tool aim to incorporate real-time data and expand to primary care and other medical specialties.

Summary updated

Abstract

Abstract Background Monitoring of waiting time (WT) in healthcare systems is essential, since long WT are associated with adverse health outcomes, reduced patient satisfaction and increased private financing. Objective To develop a methodology for routine national monitoring of WT for community-based non-urgent specialist appointments, in a public healthcare system. Design Observational historical study using data from computerized appointment scheduling systems of all Health Maintenance Organizations (HMOs) in Israel. Data sources Data included available appointments for community-based specialists and actual number of visits. The first 50 available appointments from each specialist appointment book were collected throughout December 2018. Five most frequently visited specialties - orthopedics, ophthalmology, gynecology, dermatology and otolaryngology - were included. Data collection/extraction approach WT offered to HMO members (OWT) were calculated for two scenarios: “specific” physician and “any” physician in the clinic’s geographical region. Distribution of OWT was calculated separately for each specialty and geographical region, combined to create the nationwide distribution, and expressed as mean, standard deviation and percentiles. Principal findings Estimated national median OWT for “specific” physician, ranged from 6 days (ophthalmology) to 13 days (dermatology), with large variation between geographic regions. OWT for “any” physician were 28-50% shorter than for “specific” physician. Conclusions This novel method offers a solution for ongoing national WT measurement, using computerized scheduling systems. It integrates patient preferences for physician choice and allows identification of differences between specialties and regions, setting the ground for interventions to strengthen public healthcare systems.
BibTeX
@inproceedings{wilfmiron2020novel,
  author = {Rachel Wilf‐Miron and Ilya Novikov and Arnona Ziv and Avishai Mandelbaum and Ya’acov Ritov and Osnat Luxenburg},
  title = {{A novel methodology to measure waiting times for community-based specialist care in a public healthcare system}},
  booktitle = {Health},
  year = {2020},
  doi = {10.1016/j.healthpol.2020.06.006},
}

Loading related content...

Figure