Should Your Newborn Spend An Extra Day in the Hospital?

With so much emphasis on efficiency, it is no wonder that cutting down on expensive hospital stays deemed unnecessary has been a staple of American health policy for the last 30 years. Figuring out whether (and when) hospital stays are unnecessary is difficult. Ideally, one would want to randomize patients to differing length of stays and compare outcomes across these different groups. After all, a negative correlation between hospital stays and outcomes (for example) in observational studies could simply be driven by the fact that sicker people spend more time in the hospital, rather than any negative causal effect of a longer stay. Needless to say, however, a randomized experiment like this is difficult to do and probably impossible to pass by an ethics board. As such, researchers are forced to find random sources of variation in hospital stays so as to get at causal estimates.

How does one go about recovering causal estimates then? One option is to look at the correlation between stays and outcomes conditioning on the initial health/diagnosis of the individual. One could argue that, when comparing individuals with the same health status or diagnosis, any differences in lengths of stay are essentially "random." However, even with all these controls, it is entirely possible that unmeasured attributes that are related to sickness and health also determine hospital stays.

In looking at the impact of an extra day in the hospital on newborns, Douglas Almond and Joseph Doyle, Jr, employ a different strategy. Here it is in their own words:

This paper compares the costs and benefits of extending the length of hospital stay following delivery using a discontinuity in stay length for infants born close to midnight. Third-party reimbursement rules in California entitle newborns to a minimum number of hospital "days," counted as the number of midnights in care. A newborn delivered at 12:05 a.m. will have an extra night of reimbursable care compared to an infant born minutes earlier. We use a dataset of all California births from 1991-2002, including nearly 100,000 births within 20 minutes of midnight, and find that children born just prior to midnight have significantly shorter lengths of stay than those born just after midnight, despite similar observable characteristics.

The authors find that an extra hospital day has no effect on infant and maternal mortality.

The strategy used in this paper is called "regression discontinuity" and involves using sharp cutoffs or thresholds in policies and comparing individuals that fall right around that cutoff. The main assumption that needs to hold for this to work is that those falling just around the threshold are essentially identical. In this paper, the main assumption is that newborns born at 12:03 are similar in observed and unobserved health attributes to infants born at 11:57. Where thresholds are respected and there is little scope for behavioral manipulation on the part of individuals being assigned to different policies, these assumptions are generally reasonable.

In such cases, this technique can be a powerful tool towards informing health policy. For more on regression discontinuity, see this excellent resource by Guido Imbens and Thomas Lemieux.

The Sinking Ship?

Some new data to back up my earlier assertion that the Democrats are destroying themselves on their way to the nomination:

"Twenty-eight percent of current Hillary Clinton supporters say they would vote for John McCain over Barack Obama in the general election. Nineteen percent of Obama supporters would vote for McCain over Clinton."

More here. There is some talk that Senator Clinton is using this data to support her contention that she should receive the Democratic nomination over Obama. That's kind of interesting (and a bit perverse) since the Clinton Campaign is probably culpable for most of the acrimony in this race. (After all, Clinton is the one who went around saying that John McCain is a better choice than Senator Obama.)

Whether or not these latest poll results have implications for the general election is an open question. One the one hand, those polled may simply be reacting to the bitter infighting among Democrats, with the questions framed in such a way to induce them to take sides. Once a nominee is decided and the convention completed, rank and file Dems will happily be rank and file Dems. On the other hand, independents and moderates, who have tended to gravitate towards McCain in past years, might be turned off by the infighting and board the straight-talk express for good.

It will certainly be interesting to see how this pans out.

New Blog

A few of us (some economics and sociology graduate students and myself) have started a new blog. The subject of the blog is economic development and public policy. It's a space to pose and reflect on the big questions. The current dialog focuses on foreign aid, with specific references to the Sachs/Easterly debate.

Some of you might have noticed that I've added a few new links. For those of you interested in economic development, Chris Blattman's blog is an excellent resource and chock full of thought provoking post. If you want your daily dose of behavioral economics, check out "Predictably Irrational" by Dan Ariely, in addition to stalwart blogger (and Yale econ PhD student) Santosh Anagol over at the "Brown Man's Burden." Finally, Tim Harford's "Undercover Economist" applies neoclassical economic theory (i.e. rational choice) to a wide range of phenomena.

Random Wednesdays

I have a bunch of random things in my head today. Here goes:

1) Is it just me or does the Democratic Party always find innovative ways to lose elections? This year infighting might be their undoing. Indeed, the ongoing Clinton-Obama spat should certainly make John McCain happy.

I've tried to stay away from expressing my political opinions in this space, but I've been so frustrated with the current goings on that I have to get this out: if the Democrats lose the general election, you can blame Hillary Clinton. I find it unbelievable how dirty the Clinton campaign has been in this election. Here's a great illustration: they've somehow made Sen. Obama responsible for other people's incendiary comments. Why should Obama be responsible for each and every view or rant presented by his friends? (By the way, Sen. Obama handled the whole ruckus about his pastor with the usual thoughtfulness and aplomb.)

Check out my aunt's recent post on the subject for more on this troubling race to the nomination.

2) I realized recently that I take way too long to make simple decisions. I guess I'm too deliberate for my own good. As such, I've decided to follow the Blink philosophy and allow my intuition/gut-instincts to inform my opinions and decisions. My plan is to start with minor decisions and, over time, work this approach into more important ones.

My first application of this thinking? TV shows. Every day someone recommends a new show to me. I don't have the time to watch full episodes, so I thought it would be interesting to watch a five minute sample of a show and see if I like it enough to invest more time in it.

While waiting for someone to show up, I decided to start with One Tree Hill (on the totally useless Soap Network). In the five minute span, I witnessed cheerleaders and basketball players fighting during some Midnight Madness Celebration; a creepy guy urinating in his wife's pool before she goes for a swim and, later, announcing his intention to run for mayor; Chad Michael Murray's girlfriend flipping out because he was hanging out with another girl even though she made him do so; some girl named Peyton accusing her newly introduced biological mother of not having cancer, etc.

I'm happy to report that I was able to make a quick, firm decision: I will never watch this show again.

3) I'm not a big fan of going to the dentist, but it turns out there might be some economic benefits to having nice teeth, especially for women. Check out this recent paper by Sherry Glied and Matthew Neidel. Here is the abstract:

Healthy teeth are a vital and visible component of general well-being, but there is little systematic evidence to demonstrate their economic value. In this paper, we examine one element of that value, the effect of oral health on labor market outcomes, by exploiting variation in access to fluoridated water during childhood. The politics surrounding the adoption of water fluoridation by local water districts suggests exposure to fluoride during childhood is exogenous to other factors affecting earnings. We find that women who resided in communities with fluoridated water during childhood earn approximately 4% more than women who did not, but we find no effect of fluoridation for men. Furthermore, the effect is almost exclusively concentrated amongst women from families of low socioeconomic status. We find little evidence to support occupational sorting, statistical discrimination, and productivity as potential channels of these effects, suggesting consumer and employer discrimination are the likely driving factors whereby oral health affects earnings.

What if Lay People Don't Think Much of Your Dissertation?

Here's a conversation I've been having a lot recenetly:

Other Person: What are you doing your thesis on?

Me: I'm looking at intergenerational poverty and health

Other Person: Oh fantastic! What are you doing specifically?

Me: (Thrilled that someone is interested) Well, in my first two papers, I'm studying intergenerational associations between parent and child height, which are strongly associated with an assortment of health and economic outcomes in adulthood, and trying to tease apart the mechanisms enabling these transfers. The main idea is that a notable portion of this appears to be driven by pathways that are amenable to public policy intervention. In the second paper I address this aspect, and study the extent to which these correlations are modulated by public health programs and enviromental shocks.

If the person is still interested, I go on and talk about my nascent model and empirical methodology, which actually is intuitively simple but still geeky and technical. Usually though, the person is not interested: "Oh...ummm...great...hey, I gotta run, talk to you later." (And I haven't even gotten my third paper!)

Recently, I realized that these conversations get to me. In general, I find my project exciting and interesting - otherwise I wouldn't be pursuing it. Every now and then, though, I have these doubts: what if it's NOT that exciting or interesting or (gasp) that important? At those times, some external validation is nice So I chat up my advisors and they seem genuinely excited about my work. So it's all good...right?

To some extent, yes. It's your professors and peers that know your field well, and who have a good sense of what is important and what is not. However, these days health economics (and econ in general) seems to have transcended the realm of nerd-dom into the public conscience. People are avid consumers of economics (see the comments in any Freakonomics post for proof), and I suppose one would want the public to see the value in what he/she does.

I'm not just referring to validation of the sort where one writes a paper on something fun and has it picked up in the New York Times. It's also that many health policy makers are not economists and probably don't read much academic health economics. Many of us came into this field with an eye on figuring stuff out that's helpful to people. So, if outsiders don't see the value in your work or find it disinteresting, perhaps non-economist policy makers might not either. Also, doing a PhD involves life sacrifices and it's a bit of a monastic existence. If what you are spending night and day on is not "important," then what the hell are you doing?

In the event that outsiders are lukewarm about your work, where does that leave you? After countless conversations such as those outlined at the beginning, this is the question I am asking myself. My advisors, other health economists, and I see some value in my work. Few other people I have talked to share that excitement. How am I supposed to feel about that? I think there are four rationalizations:

1) My research might not be policy relevant, but advances the science - I've used this one before, but I don't like it. While any one paper need not be policy relevant, it should help provide information on some theory or behavior where that understanding would be useful in informing policy. I think my dissertation might touch on this, but I'm not sure.

2) My research is designed to show of my technical skills and get me a good job. All policy relevant work will commence after procuring said job - Not bad. However, I am part of the Department of Health POLICY and Administration. What am I projecting if my dissertation has no "policy" in it? I might be good at stats but does my thesis send the signal that I am not imaginative when it comes to policy evaluation?

3) People are slow to come around - Many times, there is a lag between evidence and the recognition of the value of that evidence. So, a nice, perhaps delusional, thought is that my research is important, and others will come around to seeing that.

4) I am doing what I want to do - Perhaps the best of the set. See number 7 in this (excellent) list.

Smoking Cessation, Behavioral Economics Style

Steven Levitt at the Freakonomics blog on a new study by Yale economist Dean Karlan and others on using pre-commitment devices to promote smoking cessation. The paper is available here. And here's the abstract:

We designed and tested a commitment product to help smokers quit smoking in the Philippines. Individuals who sign a Committed Action to Reduce and End Smoking (CARES) contract deposit money into a savings account and agree to let the bank forfeit their entire balance to charity if they fail a urine test six months later. Bank marketers offered the product by approaching smokers in public places. Marketers administered a short survey, provided a standard pamphlet with information on smoking’s harmful effects and how to quit, and then made one of three randomly assigned offers: (i) CARES; (ii) aversive “cues”: graphic, pocket-sized pictures of the negative health effects of smoking, modeled on Canada’s cigarette packaging mandate; (iii) nothing. 11 percent of individuals offered CARES accepted. Six months after marketing, the bank marketing team returned and administered urine tests to participants from all three groups. Subjects offered CARES were 3.1 percentage points more likely to pass the test than the control group (a 38.8 percent increase); this intent-to-treat effect rises to 4.3 percentage points for those who reported in the baseline survey that they wanted to quit smoking at some point in their lives. Treatment-on-the-treated estimates suggest that those who signed a CARES commitment were 29 and 33 percentage points more likely to pass the test, respectively.

As you might recall from an earlier post, Karlan (and Yale economist Ian Eyres) were responsible for starting stickK, a company which allows clients to put down money in the present and recover this money in the future conditional on meeting their prespecified goals.

Public Sector Health Spending in India to See a Boost

The recently announced 2008/2009 Budget for India decrees a 15% increase in Central Government spending towards health and health care. This, along with the 20% increase in spending on education, has been welcomed by many as a positive step towards both acheiving poverty alleviation and sustained future economic growth.

What can we expect from this spending hike? Put a bit differently, what does a 15% increase in health spending actually buy you? Clearly, this is not an easy question to answer statistically: current health care spending is likely correlated with a variety of other characteristics associated with population health. Furthermore, it is not clear that the spending-health relationship is even contemporaneous. After all, investments may take some time to yield dividends.

A recent paper in the journal Health Economics has explored this question in the Indian context for child health (non-gated version here). The author, Sonia Bhalotra, finds the following:

Existing research presents little evidence of an impact [of public expenditures on health] on childhood mortality. Using specifications similar to those in the existing literature, this paper finds a similar result for India, which is that state health spending saves no lives. However, upon allowing lagged effects, controlling in a flexible way for trended unobservables and restricting the sample to rural households, a significant effect of health expenditure on infant mortality emerges, the long run elasticity being about −0.24. There are striking differences in the impact by social group. Slicing the data by gender, birth order, religion, maternal and paternal education and maternal age at birth, I find the weakest effects in the most vulnerable groups (with the exception of a large effect for scheduled tribes).

The identifying assumption in this paper is that, conditional on state fixed effects, state-specific time trends, and controls for total state income, rainfall and other covariates, health expenditures (current and lagged) are uncorrelated with the regression error. A tough sell perhaps, but the paper presents some reasonable robustness checks and has some interesting findings that should be explored further.

Obviously, the question of whether additional health care expenditure buys better health is extremely relevant for the U.S. context. I spoke about the conventional wisdom (i.e., increased spending is wasteful) and recent evidence that casts some doubt on this conclusion in an earlier post.

Behavioral Economics of Pain and SSRIs

More health economics in the popular press:

1) Check out this interesting article in the NYT about a study in which patients were given placebos for pain relief. A randomly selected group of these patients were told that the placebos were (relatively) costly while another group was told that these were cheap. Those in the former group were much more likely to report reduced pain post-treatment.

More behavioral economics: a recent New Yorker book review with a great introduction to the field. (Props to Jeremy and Maheer, respectively, for the links).

2) This week's Economist has a great article on the efficacy of selective serotinin reuptake inhibitors (SSRIs) in treating depression. Apparently, a meta-analysis in PLoS Medicine suggests that previous studies have overstated the SSRI treatment effect. The discrepancy appears to be due to the treatment of publication bias: whereas earlier analyses looked exclusively at published studies, this paper was careful to track down unpublished clinical trials, as well. (The Economist piece cites supporting evidence that selective publication is rampant in the anti-depressant literature. As you might guess from previous posts, I am not too surprised by this.)

However, as the article notes, all may not be lost for SSRIs: a year old National Bureau of Economic Research working paper finds large effects of anti-depressant use on reducing completed suicides. Using cross-country data, this study relies on an instrumental variable strategy where sales of other drugs are used as IVs for SSRI sales conditional on a variety of observable socioeconomic characteristics, and country and time fixed effects. The argument is that, conditional on observables, the IV is uncorrelated with population mental health and other unmeasurable that determine the suicide rate. I'm a bit skeptical, but the authors do a decent job presenting various robustness and falsification tests.

A few questions that come to mind:

1) Is it possible that both conclusions on SSRIs to be correct? That SSRIs might be not be helpful in treating depression but may be prevent depressed people from taking their own lives? If you have some thoughts on this, please comment. I'm extremely curious.

2) Why don't more meta-analyses make the attempt to gather unpublished study results? According to the Economist piece, unpublished results are required to be submitted to the FDA by law. So the data is out there and (obviously) not impossible to retrieve.

Forensic Economics

I was reading an article about the John Ritter wrongful-death trial, when I came upon the following sentence (emphasis mine):

"In other testimony, forensic economist Tamara Hunt said that if Ritter had lived and his show had continued for seven years he could have earned nearly $41.9 million."

Wait...what? I'd never heard of forensic economics and its' practitioners before, so I decided to google the term. I guess I'm pretty oblivious to what is going on around me, because I found that the field has it's own association and peer-reviewed journal.

What does a forensic economist do exactly? According to the previously linked website:

"Forensic economics is the scientific discipline that applies economic theories and methods to the issue of pecuniary damages as specified by case law and legislative codes. Topics within forensic economics include (1) the analysis of claims involving persons, workers, firms, or markets for evidence concerning damage liability; (2) the calculation of damages in personal and commercial litigation; and, (3) the development and use of generally accepted forensic economic methodologies and principles. [Definition adopted by NAFE Board of Directors, 11/6/2006]"

Who chooses to become a forensic economist? I always hear people talk about how they want to become labor or development or IO or applied micro economists. I've never heard anyone say that they wanted to become a forensic economist. Given the setting in which I learned about all this, my first thought was that the field was populated by opportunistic hacks (insert your behavioral economics insight here). However, this is probably wrong: according to the wikipedia page on the topic, "a doctorate degree in economics is the usual qualification of forensic economists." (Though I suppose it's possible to be an opportunistic hack with a PhD...).

So, which doctoral students choose to become forensic economists? I have no idea, but will let you know if I find out. In the meantime, here is yet another area where there is demand for economists (or people with economics training) to come in and help make sense of the world.