The risk of infection from polychlorinated biphenyl exposure in the harbor porpoise (Phocoena phocoena): a case-control approach

Abstract

The objective of this study was to determine whether the risk of mortality from infectious disease in harbor porpoise in U.K. waters increased with high exposure to polychlorinated biphenyls (PCBs), using a case-control study design. This is the first time that data from a long-term marine mammal strandings scheme have been used to estimate any increase in risk. The exposure odds ratio (OR) from a logistic regression model with infectious disease deaths as cases and physical trauma deaths as controls, after controlling for the effect of confounding factors, was 1.048 [95% confidence interval (CI), 1.02-1.07]. To further adjust for the difference in energetic status between cases and controls and account for the negative relationship between PCBs (sum of 25 chlorobiphenyl congeners) and blubber mass, we also "standardized" the blubber PCBs to an optimal blubber mass. This lowered the OR to 1.02 (95% CI, 1.00-1.03). Thus, for each 1 mg/kg increase in blubber PCBs, the average increase in risk of infectious disease mortality was 2%. A doubling of risk occurred at approximately 45 mg/kg lipid. In this study, we have endeavored to avoid selection bias by using controls that died of physical trauma as representative of the exposure prevalence in the population that gave rise to the cases. In addition, we controlled for the effect of variation in energetic status among the cases and controls. However, as with case-control studies in human and veterinary epidemiology, unforeseen misclassification errors may result in biased risk estimates in either direction.

Figure 1

Frequency distribution of ∑25PCBs in the blubber of harbor porpoises selected as cases or controls.

Figure 2

Geometric mean ∑25PCBs (geometric 95% CI) in the blubber of harbor porpoises for cases and controls.

Figure 3

Relative body weight of harbor porpoises [residuals around the best-fit linear regression between ln(body mass) and ln(body length)] among the cases and controls. Values shown are median, 25th–75th percentiles, and minimum–maximum.

Figure 4

Frequency distribution of estimated blubber mass in harbor porpoises as a percentage of total body mass.

Figure 5

Relationship between ln(body mass) and ln(body length) in harbor porpoises. Line shows best-fit, least-squares linear regression model [ln(mass, kg) = −9.0 + 2.57 × ln(length, cm)].