Evidence on Colorado’s LARC Program is No Lark

Usually when a public health agency makes a front-page article in the New York Times, it is not good news for the public’s health. Not true for this week’s NYT feature on the Colorado Department of Public Health and Environment and its six-year “experiment” with providing free long-acting reversible contraceptives (LARCs) to low-income teenagers. The article touts a 40% decline in the teen birth rate and a 42% reduction in the teen abortion rate since Colorado’s innovative Family Planning Initiative began in 2009. But how much of this decline can be attributed to the program?

It turns out this question is not so easy to answer because—contrary to the implication in the opening sentence of the New York Times article—a true experimental research design was not used in Colorado. Attribution is clouded by the fact that teen births have fallen across the country during this time period, partly due to the economic downturn. Fortunately, we have a new study by Jason Lindo and Analisa Packham at Texas A&M that provides strong quasi-experimental evidence about program impact, which surprisingly was overlooked in the NYT article. But this blog will give the researchers their due along with the good people in Colorado’s public health agency.

To derive estimates of causal impact, Lindo and Packham identify all the U.S. counties outside of Colorado with an operational Title X family planning clinic, and use these settings as a nonequivalent contemporaneous comparison group for the Colorado counties where the program was implemented. Colorado’s program steered resources to the state’s 37 Title X clinics to deliver LARCs free of charge to eligible clients. The researchers use difference-in-difference (DND) estimation to control for any systematic differences between the Colorado and non-Colorado settings that might otherwise confound the results. Even better, the empirical specification allows the program’s effects to vary across years while also controlling for state-specific trends in teen births, further reducing the risks of bias due to confounding.

The results suggest that Colorado’s program reduced the teen birth rate by approximately 5% in the four years following implementation. This effect size is much smaller than the 40% reduction touted in the New York Times article, but it is nevertheless a very impressive result that strongly confirms the health and economic benefits of improving financial access to long-acting contraceptives. The estimates imply a cost per teen birth avoided of around $25,000, which is a very good buy considering that the lifetime Medicaid and social services spending associated with teenage births are considerably higher. Moreover, this result probably gives us a more realistic projection of what impact we might expect from full implementation of the ACA’s mandate for first-dollar health insurance coverage of all evidence-based contraceptive options.

The disappointing news is that Colorado’s legislature so far has failed to pass a bill that would provide continued funding for the Colorado Family Planning Initiative, which has been supported to date through private philanthropy. The patchwork of federal, state, and local funding mechanisms for family planning services generally does not pay for LARCs despite their effectiveness and their beneficial economic spill-overs on other publicly-funded health and social programs. Medicaid coverage and reimbursement policies discourage many private providers from offering LARCs to their Medicaid patients, and insurer exemption and compliance issues continue to blunt the impact of the ACA’s private health insurance coverage mandate for LARCs. So for now, the sustainability and spread of Colorado’s innovation in family planning remain to be seen.

I invite you to comment on this blog, tweet at me, nudge me on linkedin, and follow my research archive.

Do Public Health Investments Crowd Out Population Health?

A new paper out by James Marton, Jaesang Sung, and Peggy Honore estimates the health effects attributable to public health spending by exploiting a unique public health financing mechanism used in the state of Georgia. The results are quite surprising, and contrast with prior studies that have examined this issue. I posted a critique of this paper on the Health Affairs blog yesterday, offering a cautionary note about the measurement and analytic strategies used to estimate the health and economic value of public health investments.

I will be hosting a roundtable discussion on these methodological issues at next week’s AcademyHealth Annual Research Meeting (ARM), as part of the Public Health Systems Research Interest Group meeting. I invite you to weigh in on this discussion by posting comments here and on the Health Affairs blog, and by joining us at the ARM meeting in Minnesota.

What the Recession and Recovery Teach Us about Public Health Delivery Systems

Parts of the U.S. economy are finally showing some signs of real rebound, but what about the nation’s public health enterprise? A new paper in the American Journal of Public Health offers a broad view of changes in public health delivery before and after the Great Recession of 2008. The results show that in communities hardest hit by the recession, public health protections remain far below their pre-recession levels. Consequently, disparities between the strongest and weakest public health delivery systems have widened considerably since the recession.

The National Longitudinal Survey of Public Health Systems (NLSPHS) tracks changes in the organization and delivery of core public health activities in a nationally representative cohort of metropolitan communities across the U.S. Our research center has conducted this survey periodically since 1998, and two recent waves of the survey – in 2006 and 2012 – span the most recent period of economic recession and the initial phases of recovery.

We reported previously that between 2006 and 2012, the average U.S. community experienced a 5% reduction in the proportion of recommended public health activities that were actually implemented in the community. However, this average decrement in public health protections masks considerable heterogeneity across the U.S. The bottom 20% of communities experienced a whopping 25% reduction in the delivery of core public health activities over this timeframe, while the top 20% of communities gained additional activities despite the economic contraction (see figure). The net result was a widening of the gap between the “haves” and the “have nots” in terms of community-wide public health protections.

Multivariate estimates confirmed what any informed model of public health production predicts: local public health delivery fell most sharply among communities that experienced the largest reductions in public health agency spending and household income and the largest increases in unemployment. Perhaps most surprising, however, were the magnitudes:

“a 10% reduction in public health agency spending per capita was associated…with a 31 percentage-point reduction in the availability of public health activities, a 36 percentage-point reduction in the intensive margin of activities contributed by local public health agencies, a 6 percentage-point reduction in the extensive margin of activities contributed by other organizations, and a 29 percentage-point reduction in the perceived effectiveness of public health activities.”

The recession also triggered some key changes in the production of public health activities, as both government and private-sector organizations adjusted their contributions to these activities. Overall, hospitals, health insurers, and community health centers showed more modest reductions in their contributions to public health activities than did government agencies – likely reflecting differences in the resiliency of the revenue sources that these different organizations tap. The net result is that many local public health delivery systems today are much more reliant on contributions from the medical care sector than they were before the recession. This growing co-dependency between public health and medical care delivery may prove to be beneficial over time as initiatives like the CMS State Innovation Models attempt to better align the prevention and treatment ends of health care delivery and financing systems in order to realize population-wide improvements in health status.

The recommended public health activities measured in the NLSPHS reflect a set of 20 programs, policies and administrative practices that national expert panels have consistently recommended to be performed in every U.S. community in order to prevent disease and injury and promote health. This recommended set includes activities to monitor community health status, investigate and control disease outbreaks, educate the public about health risks and prevention strategies, prepare for and respond to natural disasters and other large-scale health emergencies, and enforce health-related laws and regulations such as those concerning tobacco exposure, food and water safety, and air quality. These 20 activities are based on the Institute of Medicine’s Core Public Health Functions definitions and reflect high-value practices recommended by a series of expert panels convened by the U.S. Centers for Disease Control and Prevention. These activities are also closely aligned with the federal government’s Essential Public Health Services Framework and a more recently developed set of Foundational Public Health Capabilities called for by the Institute of Medicine in its 2012 consensus report on public health financing.

These results beg important questions about how public health delivery systems are evolving now as the recovery lurches forward and as key ACA coverage expansion provisions take hold. Fortunately, we are wrapping up a 5th wave of NLSPHS data collection this month, which for the first time will include a large expanded sample of small and rural communities across the U.S. Stay tuned for preliminary findings to be released at next month’s Keeneland Conference on Public Health Services & Systems Research.

The paper described in today’s post is part of an entire supplement issue of the American Journal of Public Health devoted to advances in public health services & systems research, slated for official release in April (thanks to support from the Robert Wood Johnson Foundation and CDC). As the guest editor of this issue, I will be posting highlights from some of my favorite articles on this blog over the next few weeks leading up to the official release. All of these papers are available early online so I encourage readers of this blog to take a peak and post comments here about your observations.

I invite you to comment on this blog, tweet at me, nudge me on linkedin, and follow my research archive.

National Public Health Spending: Still Waiting for Recovery

Among the many joys of a new year are the refreshed and updated full-year estimates of national health spending from the federal government’s National Health Expenditure Accounts (NHEA). We now have the federal government’s official estimates of spending from calendar 2013, including the estimates of governmental public health spending that are of particular interest for readers of this blog. Unfortunately these data show a public health sector that has yet to recover from the recessionary contractions and austerity-minded fiscal responses of recent years (see figure).

Total governmental public health spending stood at $75.4 billion in 2013, nominally up from $74.8 billion the year before but still below the high water mark of $75.5 billion in 2010. Factoring in population growth, per-capita public health spending was $239 in 2013, several dollars short of the $241 per capita expended back in 2009. Overall U.S. health spending grew at a historically low rate of 3.9% in 2013, but despite this the share of spending devoted to public health declined to 2.6% from its recent peak of 3.0% in 2009.

When these data are adjusted for inflation, public health spending trends look even more grim. Using the NHEA’s new chain-weighted price deflator (check it out), we see that total inflation-adjusted public health spending actually declined by 1.1% in 2013, continuing the negative annual rate of growth observed since 2009. Per-capita public health spending stood at $218 in inflation-adjusted dollars – a full 10% below the level observed in 2009 and a 1.6% decline from 2012.

As I posted last year, we know there are many limitations inherent in the data sources and methods used to measure public health spending in the NHEA, and there are ongoing efforts to develop enhanced estimates (including our own). But an important strength of the NHEA data are their longitudinal consistency, allowing us to track year-to-year changes in public health spending with reasonable accuracy.

These trends reveal that public health resources have yet to recover from the Great Recession of 2008, and in fact these resources are still not even keeping pace with population growth and overall levels of price inflation. Although the economy is growing again, government revenues are recovering and medical care spending remains nicely in check, public health spending has not experienced a bounce-back. The health and economic consequences of this erosion of public health resources may not be immediate, but numerous studies tell us that these trends – if left unaddressed – will yield considerable human and monetary losses over time.

The Economics of Implementing Population Health Strategies

Can economic theory and methods help us learn which mechanisms work best in scaling up and spreading evidence-based health protections and prevention strategies to the population level? Economic analysis has become an essential part of the scientific process of identifying which prevention strategies are effective and cost-effective. Two decades of research from CDC’s Prevention Effectiveness program clearly demonstrate this fact. But what does economics offer to the rapidly developing field of implementation science, particularly when applied to the task of preventing disease and injury on a population-wide basis?

I had the chance to examine this issue with some brilliant colleagues this week as more than 700 scholars convened outside Washington DC for the 7th Annual Conference on the Science of Dissemination and Implementation hosted by AcademyHealth and NIH. The field of implementation science attracts a lot of attention these days, bolstered by NIH’s focus on translational research and more recently by PCORI’s game-changing research approach and its heavy emphasis on patient-centered mechanisms for disseminating and implementing effective health interventions. Although a relatively young field of inquiry, D&I research has made impressive strides in less than a decade. A consolidated conceptual framework of D&I processes and mechanisms now exists that integrates constructs from multiple theoretical traditions. Valid and reliable measures of D&I processes are now in use. And researchers are deploying a broad array of experimental, quasi-experimental, and descriptive research designs to study D&I mechanisms in health and medicine, many of which utilize mixed-method approaches.

Even so, economic theory and economic methods are surprisingly hard to find within the current D&I research landscape. This is surprising because, after all, many of the barriers encountered in disseminating and implementing evidence-based health interventions necessarily involve resource constraints, misaligned or under-powered incentives, and asymmetric information. Economics has a lot to say about these problems and their possible solutions. As just one example, we can look to the work of Nobel prize-winning economist Eric Maskin and his game-theoretic approach to implementation theory for solutions to implementation problems that involve social decision-making in the presence of decentralized and asymmetric information – situations that characterize many complex community-level health interventions.

Economic issues are lurking in the shadows of many of the D&I studies presented at this week’s conference, and two particularly prevalent issues are worth calling out. Some D&I studies involve interventions for which no explicit financing mechanism or payment model exists, so the implementation challenge implicitly involves convincing implementation settings to reallocate resources from existing activities in order to support new interventions. For example, RAND’s randomized trials of strategies to help Boys and Girls Clubs implement evidence-based programs for pregnancy and STI prevention fall into this bucket. From an economic implementation perspective, it would seem important to study the population health trade-offs entailed in scaling back staffing for, say, physical activity programs in order to accommodate the new prevention programs.

Other D&I studies focus on interventions that do have explicit financing mechanisms – such as colorectal cancer screenings that are covered by most health insurance plans – but the implementation challenge implicitly involves uncertainties about whether available funding streams are sufficient to fully meet the resource requirements of the intervention and its associated D&I mechanisms. The Emory University/Cancer Prevention and Control Research Network study of community health centers’ use of evidence-based practices for increasing colorectal cancer screenings falls into this bucket. While centers receive insurance payments for eligible insured patients who are screened, the resources expended to implement evidence-based supports like patient outreach and education, reminders, media communication, and provider assessments and feedback may not be commensurate with screening revenue, thereby necessitating some form of cross-subsidization. The extent to which these resource flows and uncertainties influence the spread and sustainability of the intervention appear to be worthy topics for investigation as part of implementation science.

So how can we address these compelling economic research opportunities? I had the good fortune of leading a roundtable session during the second day of the conference devoted to the economics of implementing population health strategies. The goal was to raise awareness among D&I researchers about the potential utility of incorporating economic theory and methods into their implementation science studies, particularly those studies focusing on prevention and population health strategies. Our premise was fairly straightforward:

“Successful strategies to scale up and spread complex community-level interventions require an understanding of the resources required for implementation, how best to distribute them among supporting institutions, and how resource consumption and distribution varies across settings.”

To make the case, I highlighted some of the studies we now have underway through our PHSSR Center and Public Health PBRN program that explicitly examine the economics of implementation, including work involving our National Longitudinal Survey of Public Health Systems (teaser slide below), the Multi-Network Practice and Outcome Variation (MPROVE) study, the Public Health Delivery and Cost Studies (DACS), and our Costing Study of Foundational Public Health Capabilities.

The rich discussion stimulated by this roundtable session has convinced me that there is much to be gained from incorporating economic theory and methods into D&I research studies – particularly those that use a population health lens and public health orientation. The dismal science is poised to play a more active role in this solution-focused scientific endeavor to scale and spread population health.

I invite you to comment on this blog, tweet at me, nudge me on linkedin, and follow my research archive.

The Costs of Health Protection: Economic Pearls from the APHA Meetings

What’s the price of protection from disease transmission? Public health’s long-standing responsibilities in disease investigation and control have taken center stage in recent weeks in response to concerns about Ebola transmission risks in the U.S. These responsibilities remain bread-and-butter work of America’s local and state public health agencies more than 150 years after John Snow famously used them to locate and contain the source of that London cholera outbreak. Although CDC stands front and center in coordinating the U.S. Ebola response, local and state agencies shoulder most of the actual effort in investigating suspected cases, tracing and monitoring contacts, maintaining records on investigation and control activities, and disseminating guidelines to physicians and other health professionals regarding response and mitigation protocols. These agencies also play critical roles in advising state and local policymakers regarding legal interventions for disease control (no small task in the case of Ebola), and in keeping the general public up to date and informed. Ebola is a very, VERY special case, but routine disease investigation work happens every day in every community across the U.S. as agencies investigate routine but costly risks such as suspected food-borne and water-borne illnesses, vaccine-preventable diseases, and sexually-transmitted infections.

Amazingly, we know very little about the resources required to perform this work effectively in a given community or state, and about the factors that influence these resource requirements. Without evidence about what it costs to do this work, and about the community characteristics that make this work more or less resource-intensive, there is no way of knowing whether society spends too much or too little on disease investigation activities, and whether we distribute these resources optimally and equitably across the U.S. based on disease risks and prevention opportunities.

This week’s APHA meetings in New Orleans finally gave us some answers about the prices we should be willing to pay for disease investigation and control. Early results from a series of studies funded through our Public Health Delivery and Cost Studies (DACS) were featured at this year’s meetings. One of these studies, led by Adam Atherly and colleagues in the Colorado Public Health PBRN and the University of Colorado, conducted detailed time studies of disease monitoring activities carried out in that state during 2014. Results show that the cost of disease monitoring starts at about $13,000 per year in the average community, and increases by about $400 per case detected but at a decreasing rate of growth – demonstrating large economies of scale. These findings tell us quite a bit about how we might design better funding models to ensure equitable disease investigation capabilities across the U.S., and how we might pool resources and expertise across sparsely populated and low-resource communities to achieve more cost-effective protections.

Another DACS study led by Lori Bilello and colleagues from the Florida PBRN and the University of Florida-Jacksonville, focused on the cost of disease control activities for sexually-transmitted infections. The most striking finding from this study is the sheer magnitude of cost variation across Florida’s local communities. My take on this study is that much of the cost variation reflects differences in the intensity of disease control activities implemented by Florida’s county health departments and their partner organizations, begging the question of whether more intensive activities provide health returns that justify their higher levels of investment. The Florida study also identifies differences in the efficiency of STI control activities, with some agencies lagging behind in the adoption of modern screening and diagnostic technologies that do not require costly and intrusive physical examinations. Donobedian’s law of “no measurement, no improvement” is clearly on display in Florida’s study, as well as in Kim Gearin’s study of community-level variation early childhood vaccination performance in Minnesota.

APHA showcased many other interesting studies on the economics of public health delivery (see our full list of RWJF-supported PHSSR studies presented at APHA this week). This work includes:

§ Our study to estimate the effects of Medicaid expansions on public health spending and service delivery (teaser figure below);

§ Our ongoing work to estimate the costs of supporting Foundational Public Health Capabilities as recommended recently by the Institute of Medicine;

§ Our analysis of geographic variation in implementing high-value public health services for chronic disease prevention, communicable disease control, and environmental health protection using data from the MPROVE study;

§ Our analysis of the early public health implications of Kentucky’s experience in implementing key provisions of the Affordable Care Act.

§ More work from the University of Washington ‘s Betty Bekemeier and colleagues in the Public Health Activities and Services Tracking Study (PHAST), showing links between local health department spending and rates of communicable disease.

Clearly, beignets and the Big Muddy were not the only attractions on the New Orleans riverfront this week (but the beignets were very good indeed).

Stay tuned to this blog and contact the National Coordinating Center for Public Health Services & Systems Research for more information on these or other studies about the economics of public health delivery. I invite you to comment on this blog, tweet at me, nudge me on linkedin, and follow my research archive.

Learning from Variation in ACA Implementation

Political scientists, economists, and other social scientists frequently exploit variation in the implementation and timing of policy initiatives in order to estimate impact and effectiveness. The Affordable Care Act (ACA) creates abundant opportunities for these types of variation studies and natural experiments that can help us learn which implementation strategies work best, for which population groups, and under what conditions.

Numerous ACA provisions give states, local governments, health care systems, and even individual providers broad discretion in deciding what to do under ACA and when to do it. Beyond the high-visibility state decisions concerning Medicaid expansions and health insurance exchange operations, states are making hundreds of other implementation decisions about strategies like insurance outreach and enrollment, provider network adequacy, multi-payer payment models, and initiatives to expand and integrate medical, public health, and social services delivery systems to improve population-wide health status. Hospitals and physicians get to decide when and how to try their luck at ACA-supported accountable care organizations (ACOs), patient-centered medical homes (PCMHs), and various forms of bundled-payment and shared-savings models. Insurers and employers are choosing whether and how to incorporate incentives for quality improvement and wellness into their health benefits designs. Public health and community-based organizations get to decide whether and how to compete for funding to implement policy, environmental, and system (PES) changes that promote health and prevent disease and injury with support from the ACA’s Prevention and Public Health Fund. Variation due to public-private discretion, competition and entrepreneurship abound within the ACA.

How do we harvest the potential knowledge and learning from this large-scale implementation variation? A meeting this week at the Brookings Institution in Washington DC focused on developing some key research strategies. One conclusion reached early on in the meeting: it is possible to capitalize on the fact that many, many research institutions have studies underway regarding ACA implementation and impact. RAND’s Health Reform Opinion Survey and COMPARE microsimulation model, the Urban Institute’s Health Reform Monitoring Survey, the University of Chicago’s ACA Scholar Practitioner Research Network, the Princeton University/RWJF State Health Reform Assistance Network, Georgetown’s Center on Health Insurance Reforms, Kaiser Family Foundation’s Health Reform Initiative, and many other fellow travelers have big studies underway regarding key elements of ACA. (This includes our PHSSR Center’s own studies of ACA’s effects on the U.S. public health system). These many individual studies create opportunities for harmonization, triangulation, data linkage and pooled analyses, meta-analyses and research synthesis.

What’s missing from the many “big data” ACA studies are clear pictures of what’s happening on the ground in individual states and communities – an ability to characterize the patterns of variation in implementation at more granular scales, and to determine how these patterns influence health care delivery and outcomes. The research community has yet to provide a clear understanding of how and why ACA implementation strategies vary across the U.S., and how these various strategies play out in different political, institutional, socioeconomic and cultural contexts.

To fully exploit the research opportunities presented by ACA implementation variation, researchers require the ability to be many places at the same time and observe what is happening on the ground in a variety of communities and practice settings. Researchers need the stamina and staying power to observe implementation processes continuously over time. And researchers need the versatility to employ mixed-method research approaches that productively combine (1) large amounts of qualitative data on implementation strategies collected from multiple perspectives and settings with (2) the array of large quantitative data sources that offer measures of provider and consumer behaviors, service delivery patterns, and related health and economic outcomes.

This is where the Brookings Institution’s Engelberg Center for Health Care Reform and their partners at the Rockefeller Institute of Government at SUNY and the Fels Institute of Government at the University of Pennsylvania come into play. Over the past year, scholars at these institutions have convened a multidisciplinary network of researchers across the U.S. who are well-positioned geographically and institutionally to observe ACA implementation strategies at granular levels in some 36 states. The researchers that comprise the ACA Implementation Research Network have access to multiple ACA decision-makers and key informants, multiple sources of secondary data, and a variety of supporting documents and records relevant to ACA implementation at state and local levels. Already, this network has completed a series of baseline studies of ACA implementation in a broad cross-section of states (for example see our report on ACA in Kentucky and similar reports for other states).

After demonstrating proof of concept through these baseline studies in individual states, the ACA Implementation Research Network is now embarking on a series of larger-scale, cross-cutting topics that involve collection and analysis of standardized data on ACA implementation across a large number of states. This first wave of cross-cutting studies target topics such as: (1) state information technology strategies and capacities to support ACA implementation; (2) consumer support and outreach strategies for insurance enrollment; (3) provider network composition among insurers participating in the state exchanges; (4) implementation strategies used in “oppositional states” where policy leaders oppose the ACA; and (5) state strategies for reforming health care and public health delivery systems. I hope to play a leading role in this last cross-cutting study, together with other colleagues in the PHSSR enterprise. Specifically, we hope to leverage studies that we already have underway concerning public health delivery system reform, and to leverage the strength of our public health practice-based research networks (PBRNs) across the U.S.

The ACA Implementation Research Network is strongly committed to engaging knowledge users in the design and implementation of its ACA research, and to disseminating findings rapidly and continuously to decision-makers in policy and practice settings. We are excited to be a part of this collaborative experiment in large-scale policy implementation field research, and we believe it will become a valuable source of knowledge and learning over time. Fellow travelers in public health economics and delivery system research will want to keep close tabs the progress of this new effort.