Insurance and Hospital Stays: Q&A with Professor Douglas Almond

Professor Douglas Almond recently published a new article in the American Economic Journal: Economic Policy, titled “After Midnight: A Regression Discontinuity Design in Length of Postpartum Hospital Stays.”

In his paper, Professor Almond, with Professor Joseph Doyle from MIT, looked at how insurance can cause overuse of health care, due to economic concepts like moral hazard and adverse selection. They found that health insurance allowing longer stays in the hospital after childbirth does not lead to additional health benefits. 

What led you to the topic, and why did you choose to cover it via the lens of economics, in an economics journal, rather than health or medical policy?

Visiting a hospital in Boston when a friend’s first son was born was the impetus for the paper. For some reason, I read the discharge policies, which drew a sharp distinction between births before versus after 8 p.m.  

My co-author Joe Doyle and I originally tried to get requisite natality and discharge data from the State of Massachusetts, but they repeatedly denied our request. I think the reason our request was denied was because the particular econometric approach proposed (regression discontinuity design) was unfamiliar in the biomedicine and public health community, and more popular in economics.

This remains the case, and that leads to the second part of your question. In general it can be an uphill fight for an economist to publish in medical journals, and in our case I think this was exacerbated by the novelty of the methodology.  

Why is your study important?

I think our study is important because it assesses the impact of relatively clean and exogenous variation in length of stay in a context where stay lengths are already quite short on average (when compared to other countries, or in the U.S. prior to significant managed care penetration). This “short stay” context suggests exogenous changes to stay length might have important health impacts, so we were surprised to find they didn’t have detectable ones.

Further, our study lingers on the question of what the affected population (“compliers”) look like, which I think helps speak to the generalizability of natural experiments like ours.

How did you conduct your research, using data from hospital admissions and releases? Physicians’ reports?

My co-author at MIT Sloan, Joe Doyle, obtained the restricted access data from the State of California that reported exact time of birth. These California data merged together information from the birth certificate with that from the hospital discharge record.

Do you have a conclusion, based on your specific study of additional significant costs for no health benefits in the case of hospital care for childbirth, that could apply to the health care system at large? 

We found that hospitals seem to do a good job of providing requisite medical care within a surprisingly short time period following routine births.  Thus, laws requiring a minimum stay length for routine childbirths appear to generate unnecessary costs.

In your opinion, how can these unnecessary costs be removed from the system? Do you have an estimate of how much these costs are, in the case of your study for example?

I think that by including the vast majority of deliveries that are routine, mandates for minimum hospital stay length were unnecessarily broad.  We estimated that the mandates have a price tag of about $1.1 billion per year (over $10 billion since 1997).

That said, there may be some important benefits of increased stay length, such as parental peace of mind, that go unobserved in our data and therefore uncounted by our analysis.

Michelle Chahine, October 28, 2011