Duke Environmental Economics Working Paper EE 10-06, October, Lori Bennear, Alessandro Tarozzi, Alexander Pfaff, H.B. Soumya, Kazi Matin Ahmed, and Alexander van Geen (2010)

Interventions aiming to change health behavior through risk communication are gaining ground in the developing world.  This randomized evaluation of a program communicating the health risks of arsenic in Bangladesh suggests that the format of the information conveyed is an important determinant of how individuals respond.  Unfortunately, risk communication may be the least effective for high risk individuals.

In recent years, health information campaigns have become an important public health drive in developed countries.  These campaigns range from nutritional facts on packaged foods, laws that require calorie counts on fast food menu boards, to advisories to beware of mercury levels in fish.  Fewer such campaigns have taken place in developing countries.  This paper evaluates one such campaign: an information intervention to make citizens aware of the risks of arsenic in drinking water in Bangladesh. 

Arsenic naturally occurs in groundwater in some locales, such as the United States, India, and China, but levels are particularly high in Bangladesh.  Geographically, arsenic concentrations are very heterogeneous in Bangladesh, with some unsafe wells only 1-200 meters away from safe wells.  As of 1999, 30% of Bangladeshies were consuming shallow tubewell water that exceeded national guidelines for arsenic, and 45% were consuming water that exceeded World Health Organization guidelines. 

The health risks of long term arsenic exposure are pronounced, including skin lesions, cardiovascular disease, and cancer.  Cardiovascular problems may emerge after a latency period of 5-15 years and cancer after 20 or more years.

Over the past 15 years, the Bangladeshi government has launched policies to raise arsenic awareness, including testing wells and painting the spout of those that meet the national standard green and those that do not red.  As of 2005, approximately 5 million tube wells have been tested.  These programs have been successful in leading citizens to shift to safe well usage, but the overall level of arsenic exposure in Bangladesh remains high.  The national program focuses on reducing high levels of arsenic exposure, but health gains can still be made by reducing low levels of exposure.

The authors evaluate a randomized intervention carried out in 43 villages in Araihazar district that provided information on well-water arsenic.  Two types of messages were conveyed: a “bright-line” message that emphasized where arsenic levels stood relative to the national guideline and a “gradient” message that emphasized that lower levels of arsenic exposure are always better.  The bright-lines message was orally delivered to 266 households in 23 villages and the gradient message was orally delivered to 267 households in 20 villages.  On average, households in the sample had 5.3 members and PPP US$262 in monthly expenditure.  80% used a sanitary latrine on a regular basis, and 11% had good quality (“pukka”) dwellings.  33% of household heads were literate, and 21% of them held a secondary school diploma.  Child school enrolment was relatively high at 76% for 6-14 year olds.  Of all wells used by sample households, arsenic concentrations stood at 116 parts per billion on average, over twice the national threshold of 50 parts per billion.

The findings are surprising and suggest that the message delivered strongly affects responses in different contexts.  When exposed to moderately unsafe levels of arsenic (50-100 parts per billion), households receiving the gradient message were 50% more likely to switch than those receiving the bright-line message.  However, when exposed to high levels of arsenic (typical of most unsafe wells in the area), households receiving the gradient message were 40% less likely to switch than those receiving the bright-line message.

Thus, the authors summarize that “In the context of our intervention, providing richer risk information yielded mixed results leading to better health choices among users of moderately unsafe wells and reducing mitigating behavior among well users with very high levels of arsenic” (4).  They conclude that “conveying information about the continuous nature of arsenic risk may not lead to increases in health-improving behaviors and may actually have the opposite effect for the most at-risk population, at least in the short term (less than one year)” (26).

What to make of this?  This study plays into past findings—e.g. in genetic counselling—that the format of interventions is very important in influencing health behavior.  It also supports past findings that risk communication may be least effective for high risk individuals.

This finding is important because many policy interventions in the developing world aim to change behavior through information provision—from those designed to influence sexual practices to those seeking to educate participants about returns to education or child nutrition.  The authors argue that “further research is clearly needed in order to develop more effective health risk communication strategies” (26).