Michael Freeman

Assistant Professor of Technology and Operations Management @ INSEAD

Curriculum vitae


1 Ayer Rajah Avenue
Singapore 138676

Gatekeepers at work: An empirical analysis of a maternity unit

Journal article

Michael Freeman, Nicos Savva, Stefan Scholtes
Management Science, vol. 63(10), 2017, pp. 3147-3167

View PDF


APA   Click to copy
Freeman, M., Savva, N., & Scholtes, S. (2017). Gatekeepers at work: An empirical analysis of a maternity unit. Management Science, 63(10), 3147–3167. https://doi.org/10.1287/mnsc.2016.2512

Chicago/Turabian   Click to copy
Freeman, Michael, Nicos Savva, and Stefan Scholtes. “Gatekeepers at Work: An Empirical Analysis of a Maternity Unit.” Management Science 63, no. 10 (2017): 3147–3167.

MLA   Click to copy
Freeman, Michael, et al. “Gatekeepers at Work: An Empirical Analysis of a Maternity Unit.” Management Science, vol. 63, no. 10, 2017, pp. 3147–67, doi:10.1287/mnsc.2016.2512.

BibTeX   Click to copy

  title = {Gatekeepers at work: An empirical analysis of a maternity unit},
  year = {2017},
  issue = {10},
  journal = {Management Science},
  pages = {3147-3167},
  volume = {63},
  doi = {10.1287/mnsc.2016.2512},
  author = {Freeman, Michael and Savva, Nicos and Scholtes, Stefan}


We use a detailed operational and clinical data set from a maternity hospital to investigate how workload affects decisions in gatekeeper-provider systems, where the servers act as gatekeepers to specialists but may also attempt to serve customers themselves, albeit with a probability of success that is decreasing in the complexity of the customers’ needs. We study the effect of workload during a service episode on gatekeepers’ service configuration decisions and the rate at which gatekeepers refer customers to a specialist. We find that gatekeeper-providers (midwives in our context) make substantial use of two levers to manage their workload (measured as patients per midwife): they ration resource-intensive discretionary services (epidural analgesia) for customers with noncomplex service needs (mothers with spontaneous onset of labor) and, at the same time, increase the rate of specialist referral (physician-led delivery) for customers with complex needs (mothers with pharmacologically induced labor). The workload effect in the study unit is surprisingly large and comparable in size to those for leading clinical risk factors: when workload increases from two standard deviations below to two standard deviations above the mean, noncomplex cases are 28.8% less likely to receive an epidural, leading to a cost reduction of 8.7%, while complex cases are 14.2% more likely to be referred for a physician-led delivery, leading to a cost increase of 2.6%. These observations are consistent with overtreatment at both high and low workload levels, albeit for different types of patients, and suggest that smoothing gatekeeper workload would reduce variability in customer service experience.


Follow this website

You need to create an Owlstown account to follow this website.

Sign up

Already an Owlstown member?

Log in