Characteristic features of studies attributing chronic disease to low concentrations of arsenic in drinking water

Abstract

Introduction: Arsenic in drinking water can cause multiple health effects, including cancer, cardiovascular disease and chronic respiratory disease. The epidemiological evidence comes from populations with prolonged exposure above 100µg/L of inorganic arsenic in water. However, many recent publications claim to find health effects from drinking water containing less than 50µg/L of arsenic. This paper looks to see if they have characteristic validity-related features. Methods: We identified published studies conducted in populations where drinking water arsenic concentrations had been less than 50µg/L. We scored studies based on the following points, with 1 for good quality on each point and 0 if not. We totaled the scores for each study with a possible range of 0-7. 1. Did the authors discuss scientific plausibility of their findings at such low water concentrations? 2. Did the authors present their findings in a manner that made their implausibility apparent? 3. Was there evidence of increased risks in some simple stratified analyses? 4. Were key results from analyses confined to complex non-transparent statistical modeling? 5. Did exposure data span at least 20 years of measured drinking water concentrations? 6. Was exposure data confined to one point in time? 7. Was arsenic in water measured with sufficient latency before the outcomes? Results: 21 publications attributed chronic disease to arsenic in drinking water below 50µg/L. The average score was 1.5. Only one study scored above 3, and it had serious study design problems separate from the criteria addressed in our scoring. Conclusions: The studies claiming health effects of arsenic in populations whose drinking water contained less than 50µg/L have serious flaws. Greater rigor is needed in peer review of proposed studies and of publications involving low arsenic water concentrations. Regulatory agencies need to be very cautious about using these studies for risk assessment.