The story about unwashed grocery bags causing an increase in deaths in San Francisco recently made the rounds. The ban on plastic bags leading to the use of reusable bags is blamed for the increase in deaths. But, what does the science say? The study from Klick and Wright came from an economics institute, and used economic research tools. The public heath people have responded with an analysis based upon knowledge of the topic. First, they critique the methods used in the study.
The Klick & Wright study is classified as an ecological study; that is, if factor A (reusable bags) increased in a location, and then factor B (gastrointestinal bacterial infections) increase in the same location, therefore, factor A caused factor B. Drawing causal conclusions from this type of study is called an “ecological fallacy.”2 The basic study flaw is that persons that use reusable bags frequently may not be the same persons that were diagnosed with gastrointestinal bacterial infections in their study. This is the reason epidemiologists will not use ecological studies to test causal hypotheses. At best, ecologic studies raise epidemiologic causal hypotheses but cannot test them.
• In testing causal hypotheses, it is necessary to measure the outcome (gastrointestinal infec- tions) and exposure to the putative cause (reusable bags) in the same persons. Because of their study design, this was not possible.
• In testing causal hypotheses, it is necessary to “control for” alternative causal explanations (called “confounders”). Because of their study design, this was not possible. For example, gastrointestinal bacterial infections are not only caused from contaminated food, but also from contaminated water, improper food handling or preparation, or from person-to-person spread (such as sexual activity, especially in men who have sex with men). In any causal study, investigators always adjust for the “usual suspects.”
The original authors never established a causal link. In a real public health study, one would look at the bags used by the people who died and compare any bacteria found with the ones that caused the deaths. That was not done. There is an association, but no causal link. However, there are problems with the association.
From our surveillance data, there has been an increase in campylobacteriosis, no increase in salmonellosis, and no increase in enterotoxigenic Escherichia coli (Appendix: Figures 1 and 2). Interpreting these changes is not straightforward. The epidemiology of enteric pathogens in San Francisco differs compared to surrounding counties because we are an urban center with a larger population of ethnic immigrants and men who have sex with men (MSM).3 Research studies need to adjust for these population differences.
• The authors analyze deaths due to ICD-10 cause of death codes A00-A09 (intestinal infectious diseases). From 2001 through 2010 San Francisco had a total of 140 deaths from these causes. However, 111 of them (79%) were for code A047 (Enterocolitis due to Clostridium difficile). These infections have indeed increased in San Francisco since 2005 (before the ban) (Table 1). Toxin-producing C. difficile causes enterocolitis through overgrowth when exposed to antibiotics, most commonly in hospitalized patients. However, in recent years we have seen an unexplained increase of C. difficile enterocolitis in the United States, Europe, and Canada. The increase in San Francisco probably reflects this international increase. Foodborne exposures is not yet an established cause of C. difficile enterocolitis, but is an active area of research.4 5
For these reasons, the authors should not have included C. difficile deaths in their analysis. Without C. difficile, there were a total of 29 deaths in these codes over 10 years through 2010. So their analysis of deaths, and costs due to deaths, is completely invalid as evidence
The increase in deaths is largely due to C. difficile. There had been an international increase in deaths related to C. difficult preceding the ban on plastic bags. The cause of this increase is unknown. That other bacterial infections are not also increasing suggests the unwashed bags are probably not the source. Should you still wash your bags? Sure. Even if they are proven to not be causing our current problem, why would we not wash items that that come into contact with our food. Just don’t, based on current evidence, expect it to protect you from these infections.