Causal Analytics for Applied Risk Analysis Louis Anthony Cox, Jr

Results

Health Outcome Distributions

Table 7.4 provides the simulated distributions for current levels of pork-attributable toxoplasmosis related health outcomes in the U.S. These estimates are very close to those obtained by Batz et al.(2012). Figure 3 is a histogram of the full distribution generated for current total QALYs lost due to toxoplasmosis.

Excess Risk Factor and QALYs Lost Due to Reduced Confinement

A shift of swine from total confinement operations by a fraction by ΔC= 0.001 (approx. 65,000 pigs) increases the excess risk factor by an average of 0.0089 (95% CI = 0.002, 0.0192). Since the risk factor distribution (Equation 7.2) is a composite of two beta distributions with long right-hand tails, it is also highly skewed as shown in Figure 7.4 (left panel). The corresponding distribution of QALYs lost due to a shift of 0.001 of pigs from total confinement operations, shown in the right panel of Figure 7.4 (and calculated via Equation 7.8 with H_k = Equation 7.7) has a mean of 35.98 quality-adjusted life-years (95% CI = 6.82, 86.92). This is equivalent to an expected loss of one human QALY for every 64,925/35.98 ≈ 1,804 pigs moved from total confinement production. From the confidence limits for QALYs lost, as many as 9,520 pigs, or as few as 747 pigs, shifted from total confinement operations, could cause the loss of approximately one human QALY. Table 7.5 shows the lower, mean, and upper estimates of the excess QALYs lost as the production shift fraction increases from 0 to 0.001.

Discussion and Conclusions

This chapter has used a simple causal model to estimate how human Toxoplasmosis rates would be affected by shifting swine production from total confinement systems to open/free range systems. It determines plausible expected values and ranges of health effects that reflect the inherent uncertainty in the input parameters. A probabilistic simulation model provided the mechanism to generate the desired results. The model predicts that every 1,804 pigs shifted to open production (95% confidence interval 747-9,520) cause an expected additional loss of a human quality-adjusted life year (QALY). As of 2011, the relatively small number of USDA certified organic swine (12,373) corresponded to an expected human health loss of about (12,373/1,783) = 6.94 human QALYs lost per year. Shifting 0.1% of total swine (approx. 65,000 pigs) from total confinement to open/free range operations would cause a loss of approximately 36 human QALYs per year, including between 1 and 2 adult deaths per year (138.40 current annual deaths x 0.0089 avg. excess risk factor). The number of pigs certified as organic increased by 3,286 pigs from 2010 to 2011 (USDA-ERS , 2012), which is equivalent to an average estimated human QALY loss of 1.82.

The results above are sensitive to several of the estimated parameters. In particular, the pork attributable fraction (proportion foodborne x proportion from pork), with an estimated mean value of approximately 0.20 (0.41 x 0.50). Equations (7.3) and (7.8) imply that the mean excess risk values are proportional to the mean pork attributable fraction. Both of the subparameter distributions are the results of aggregating expert opinions, with obvious shortcomings relative to estimates derived from experimental evidence. However, by modeling the true fractions as relatively broad probability distributions, we hope to have captured this uncertainty in our output confidence intervals. Similarly, the mean excess risk values are proportional to the difference in prevalence rates between pigs raised in confined versus open/free range systems (see equations 7.2 and 7.8). The prevalence values are based on aggregation of experimental results – better than expert opinion, but still subject to inaccuracies when applied to current-day, nationwide experiences. In this case, we also rely on having captured the uncertainty via probability distributions. But we also benefit from only needing to accurately estimate the difference in the two rates rather than their absolute values. Collecting additional data regarding the prevalence of T. gondii in open/free range swine could allow tighter estimates for the prevalence distribution (P_O in our study) and thus tighter estimates of the health impacts. Finally, the mean risk results we report are also proportional to the mean health outcome estimates of Scallan et al. (see equations 7.3-7.5 and 7.8), which have their own limitations as discussed by the authors.

The scope of this analysis is limited to human health effects attributable to Toxoplasmosis from pork for two alternative pig production systems. We do not attempt to perform a risk or cost-benefit analysis of all aspects of confined versus open/free range production of pigs. However, it may be worthwhile to outline other impacts at a high level to demonstrate that the issue is complex and multidimensional. We have not found any studies providing verified quantifiable human health benefits from consuming pork from open/free range production that could offset the risks quantified in this research. Since open/free range production is often associated with antibiotic free production, many consumers may prefer it on that basis. However, there is currently no scientific evidence that pork from pigs produced using conventional methods, which could sometimes include subtherapeutic use of antibiotics as a feed additive, or use at higher levels to treat disease under prescribed conditions, poses a greater human health risk from antibiotic-resistant organisms (Cox and Popken, 2014; Cox and Popken, 2010; Cox et al., 2009). Nor should open/free range pigs be considered free from such organisms. Weese et al. (2011) performed a longitudinal study showing that piglets raised on an antimicrobial free farm in Canada, a country even more restrictive than the U.S. regarding antibiotic use in food animals, had MRSA prevalence rates as high as 65%. O’brien et al. (2012) found no statistical difference in MRSA rates between meat from hogs raised conventionally and meat from hogs raised without antibiotics. Antibiotic free (ABF) and organic pig farms have also been reported to have higher prevalence rates than conventional farms of bacteria such as Salmonella (Gebreyes et al., 2008). Pigs raised on outdoor pastures may also be more subject to infection by parasites and certain other diseases, but perhaps less subject to respiratory diseases (Jolie et al., 1998). In Denmark, the parasite problem is considered to be a threat to outdoor pig production (Roepstorff et al., 2011)

Still, consumers may perceive health benefits from eating open/free range pork, and may also have concerns regarding animal welfare (even though pigs raised indoors are less subject to temperature extremes and sunburn) and environmental impacts. Consumers may not be aware of new and old (re-emerging) human health threats posed by open/free range production practices. Kjilstra, et al., (2009) examined policy issues related to increased use of organic livestock production in the Netherlands with reference to increased Toxplasmosis in free range pigs, and higher dioxin levels in the eggs of free range hens. They identified the need for increased communication of risks to consumers to avoid even more significant problems in the future. This could be achieved in the U.S. by labeling requirements, increased use of USDA awareness bulletins and newsletters, and training directed at farm veterinarians on how to identify and mitigate high prevalence situations. Organic farm producers need to be willing to undertake reasonable risk mitigation steps, such as reducing swine contact with cats and rodents. Consumers, especially pregnant women, need to use caution in handling raw pork, and thoroughly cook all meat before eating.

Quantitative risk assessment alone cannot determine whether the trade-offs identified in this study are desirable from a public policy point of view, but it clarifies the approximate magnitude of the plausible human health harm that might realistically be expected to be caused by greater use of open/free range production systems.