@CDCMMWR reminds us that #polio eradication has a ways to go still

Wednesday, March 21, 2018

I tend to follow news related to polio outbreaks pretty closely, as the prospect of eradicating another disease (like we did smallpox) is tantalizing but seems to always be just out of reach. An article in last week's edition of MMWR serves as a sobering reminder that even if we do manage to wipe out wild poliovirus, we still have to contend with collateral in the form of vaccine-derived poliovirus:
Democratic Republic of the Congo has had cases of polio caused by vaccine-derived polioviruses (VDPVs) documented since 2004. The emergence of these VDPVs, which cause paralysis similar to wild polioviruses, can occur where population immunity to poliovirus is suboptimal.
[...]
Risk for VDPV emergence in DRC will remain unless population immunity to poliovirus is increased and maintained.
Perhaps I should bet on Guinea worm as the best candidate for the next disease to be eradicated.

#OpenData for surveillance: Flu and fires

Monday, March 19, 2018

I'm always intrigued to see analysts use traditional data (particularly government and/or open data) in novel ways, especially for monitoring or surveillance for health and human rights. Two such analyses crossed my feed last week. The first, a model built by an epidemiological team at the University of Iowa, used CDC influenza surveillance data combined with readings from an IoT-style thermometer called Kinsa to beef up predictions of flu transmission patterns:
Smart devices and mobile apps have the potential to reshape public health alerts and responses, health care industry analysts and researchers told The Bee. Over in San Francisco, for instance, the staff of smart thermometer maker Kinsa were receiving temperature readings that augured the surge of flu patients in emergency rooms there.
[...]
Computational researcher Aaron Miller worked with an epidemiological team at the University of Iowa to assess the feasibility of using Kinsa data to forecast the spread of flu. He said the team first built a model using surveillance data from the CDC and used it to forecast the spread of influenza. Then the team created a model where they integrated the data from Kinsa along with that from the CDC.

“We got predictions that were ... 10 to 50 percent better at predicting the spread of flu than when we used CDC data alone,” Miller said. “Potentially, in the future, if you had granular information from the devices and you had enough information, you could imagine doing analysis on a really local level to inform things like school closings.”
Setting aside my own unease with the "wild, wild west" nature of data collection by Internet of Things things, it was nice to see the study authors reinforce the value of CDC surveillance data (and not going the way of Google Flu):
Miller does not think, though, that these new sources of data have yet made the traditional CDC surveillance channels obsolete.

“When it comes to studying things like infectious diseases, the challenge is when we’re just getting things like temperature readings,” Miller said. “Because we’re not working with microbiological specimens, we can’t confirm that what we’re looking at is influenza. ... We went through a number of procedures in the paper to try to validate that what we were looking at influenza, but at the end of the day, we can’t necessarily confirm that.”
Unfortunately, the picture is not nearly as rosy for this analysis of arson in DRC described in the New York Times:
The clashes between the Hema and Lendu communities — on the eastern side of the Ituri province, bordering Uganda — started in December and escalated in early February.

Historically, these distant conflicts have been difficult to analyze. But new technologies allow us to investigate them in close to real time.

I immediately collected active-fire data from NASA — thermal anomalies, or hot spots, that are recorded daily. It showed dozens of fires on the densely forested mountain ridge and along the shoreline of Lake Albert, one of the African Great Lakes between Congo and Uganda.
[...]
Google and other online mapping platforms often show only blurry satellite images, or have no location names for remote areas such as the small fishing villages around Lake Albert. This makes it difficult to find places where people live. To deal with this challenge, I exported residential data from the online mapping site Openstreetmap.

I then overlaid the NASA data with this new data in Google Earth to look for recorded fires that were in or near populated places.
The author goes on to explain that they used this data in conjunction with traditional "shoe-leather reporting" to tell the story "in the most powerful and visual way." I have really appreciated that journalists have begun accompanying their reporting with pieces that explain their methodology in detail. One silver lining amid all the tempest over "fake news."

@LancetGH features a Bayesian analysis of nutrition survey data in #Afghanistan

Thursday, March 15, 2018

The current edition of Lancet Global Health features an analysis of health survey data using Bayesian modeling techniques to map nutritional status in the country:
The 2013 Afghanistan NNS is the largest, most comprehensive nutrition assessment of its kind to be done in the country. Designed to be representative at the national and provincial levels, the survey's aim was to provide estimates on various nutritional indicators, including anthropometry, infant and young child feeding, and micronutrient deficiencies, for women, children, adolescent girls, and the elderly in Afghanistan.
[...]
The specific objectives of our study were to estimate the district-level prevalence of nutritional status indicators among children (stunting, underweight, or wasting) and women of reproductive age (underweight or short stature) and to understand the major causes of undernutrition in women and children in Afghanistan through exploring a range of individual, household, community, and environmental risk factors.
What was interesting to me about this particular analysis (besides the maps) was the use of Bayesian modeling, which I now understand better after designing this visualization on county-level estimates of teen birth rates based on a Bayesian analysis. In addition to identifying socioeconomic factors correlated with the various nutritional variables of interest, the authors used the Bayesian model to populate a series of chloropleth maps on said variables:

Figure 2: Prevalence of child stunting, wasting, and underweight in districts of Afghanistan

Figure 3: Prevalence of underweight, obese, and overweight in women of reproductive age in districts of Afghanistan

Not surprisingly, nutritional indicators in children were directly related to things like household wealth, food insecurity, and maternal education and literacy. "Linear growth and weight of children younger than 5 years old were independently associated with household wealth, maternal literacy, maternal anthropometry, child age, food security, geography, and improved hygiene and sanitation conditions" - i.e., "children from richer households were taller, weighed more, and were less emaciated." Also not surprisingly, child malnutrition is at "alarming" levels in many areas, particularly those experiencing active conflict:
Districts with alarmingly high wasting prevalence (some exceeded 20%) were in central regions and regions bordering Pakistan, including the east, southeast, and south. The concurrent conflict and insecurity in these regions could be one explanation, particularly since a gradient of child emaciation across wealth status and maternal education was not evident.

Epidemiological miscellany: #PrEP, #gunviolence research, and the Great Recession

Wednesday, March 14, 2018

A couple of interesting tidbits have come across my radar over the last couple of days. The first is a handy new interactive state-level map on PrEP use by AIDSVu:
AIDSVu has released the first-ever interactive state-level maps visualizing a 73 percent increase year over year in persons using PrEP across the U.S. from 2012 to 2016, with 77,120 PrEP users in 2016. PrEP, or pre-exposure prophylaxis, is when people at high risk for HIV take HIV medicine daily to lower their chances of getting infected with HIV. AIDSVu’s maps visualize the growth in PrEP use at the state-level by year, and break down the data by age and sex. These data and maps offer important information and tools to public health officials, policymakers, and researchers to inform efforts to improve PrEP awareness and increase uptake where it is needed most.
The data comes primarily from Gilead:
De-identified, aggregate data were obtained from Source Healthcare Analytics, LLC (SHA) with the support of Gilead Sciences, Inc., and compiled by researchers at the Rollins School of Public Health at Emory University. SHA collects data from over 54,000 pharmacies, 1,500 hospitals, 800 outpatient facilities, and 80,000 physician practices across the U.S. SHA’s dataset contains prescription, medical, and hospital claims data for all payment types, including commercial plans, Medicare Part D, cash, assistance programs, and Medicaid. From this overall sample, AIDSVu presents a subset of data comprising prescriptions for TDF/FTC for PrEP.

There is also this piece in the Atlantic, which highlights a study done on health outcomes impacted by the Great Recession:
For the study, Teresa Seeman, an epidemiologist at UCLA, and her colleagues examined longitudinal data on 4,600 people between the ages of 45 and 84 collected between 2000 and 2012 to look for changes in their blood pressure and fasting blood-sugar levels.

They found that blood pressure increased significantly among all groups during the time period, and blood glucose did too, among certain groups. The authors speculate the reason for the spike was stress—potentially different stressors for different generations. The younger people in the cohort were either unemployed, or those still working were likely wondering how on Earth they would be able to retire. The older people may have owned their own homes and watched the housing market collapse. All of this, they found, likely drove up their stress levels, and blood pressure.

The authors also found that during the recession, many people stopped taking their medications—especially older homeowners, whose major sources of wealth were evaporating. “The evidence suggests that the stresses of the Great Recession took their greatest toll on those who are on medication,” they write—because they may not have been able to afford the drugs anymore.

These findings further confirm what other researchers have seen on a more cellular level: Economic hardship causes stress, and that stress can sneak under the skin, disrupting bodily systems.
The author takes a dig at economist Tyler Cowen, too, which was kind of fun.

Finally, guns are still in the news, despite The Donald having thrown in the towel. It would seem (for the moment, at least) that calls for more research into gun deaths and gun violence have managed to put down roots and put the spotlight on the Dickey Amendment:
The N.R.A. pushed Congress in 1995 to stop the C.D.C. from spending taxpayer money on research that advocated gun control. Congress then passed the Dickey Amendment in 1996, and cut funding that effectively ended the C.D.C.’s study of gun violence as a public health issue.

The result is that 22 years and more than 600,000 gunshot victims later, much of the federal government has largely abandoned efforts to learn why people shoot one another, or themselves, and what can be done to prevent gun violence.

After the Parkland school massacre in Florida last month, lawmakers and gun control experts have demanded that the agency take up the issue of studying gun violence again, arguing that the federal law doesn’t ban such research altogether but prohibits advocacy of gun control.
There is much ink to be spilled on the question of why the NRA and other conservatives feel that research on gun violence "advocates" for gun control - probably along the lines of the confirmation bias phenomenon and how people react when presented with evidence that challenges their cherished beliefs. Methodologically sound research does not "advocate" for anything. But I will leave that to more eloquent commentators. In the meantime, AJPH is working to bridge the research gap by making all of its published work on gun violence open access (it's typically behind a paywall).

@CDCMMWR article shows the #opioid epidemic is only getting worse

Tuesday, March 13, 2018

At the beginning of this month, David Greene interviewed Surgeon General Jerome Adams on NPR's Morning Edition about the opioid summit that was to be held later that day. It is always interesting to me to listen to serious public health officials do what I imagine to be a slightly painful dance in public interviews - communicate important messages while staying within partisan lines and painting the current administration with sunshine-and-daisy brush strokes. (Current CDC director Anne Schuchat is masterful at this.) I can only imagine what that was like for Dr. Adams, although he got plenty of experience managing Indiana's HIV outbreak under Pence while the latter was governing Indiana, including (by some miracle) persuading him to allow syringe exchange programs to operate. At any rate, during the interview, he pointed out something which even I did not realize. It turns out that America does not take our opioid problem seriously:
No. 1 we want America to understand this is a problem. And I'm an avid NPR listener, so I can speak from that point of view and tell you that a lot of your listeners may not believe that America doesn't see this as a problem. But the majority of the public does not see the opioid epidemic as rising to the level of an emergency. So it's important that we continue to say at the highest levels this is a problem in all communities, and it's getting worse.
I suppose I shouldn't be surprised, but every once in a while I forget that I live in an epi bubble. But in case anyone was wondering, last week's MMWR confirms that the opioid epidemic is getting worse:
From July 2016 through September 2017, a total of 142,557 ED visits (15.7 per 10,000 visits) from 52 jurisdictions in 45 states were suspected opioid-involved overdoses. This rate increased on average by 5.6% per quarter. Rates increased across demographic groups and all five U.S. regions, with largest increases in the Southwest, Midwest, and West (approximately 7%–11% per quarter). In 16 states, 119,198 ED visits (26.7 per 10,000 visits) were suspected opioid-involved overdoses. Ten states (Delaware, Illinois, Indiana, Maine, Missouri, Nevada, North Carolina, Ohio, Pennsylvania, and Wisconsin) experienced significant quarterly rate increases from third quarter 2016 to third quarter 2017, and in one state (Kentucky), rates decreased significantly. The highest rate increases occurred in large central metropolitan areas.