In 2007, the US Centers for Disease Control and Prevention (CDC) reported that poisoning fatality surpassed both motor vehicle crashes and firearms to become the leading cause of injury related fatalities.93 This trend has continued and is largely influenced by an epidemic of prescription opioid abuse.67 Understanding the evolving trends in poisoning is essential to the development of enhanced surveillance, prevention, and education programs designed to improve medication prescribing, drug safety, and to reduce unintentional poisoning. Although data can be analyzed from numerous sources, such as death certificates, hospital discharge coding records, and poison centers, it is essential to recognize the biases that are inherent in each of these reports. Because not all significant poisonings result in either hospitalization or fatality, data from poison centers appear to offer a unique perspective.
Unfortunately, the term “poisoning” is often defined differently and therefore may be confusing. In this text, “poisoning” is used to denote any exposure to any xenobiotic (drug, toxin, chemical, or naturally occurring substance) that results in injury. Yet the data collected and disseminated by poison centers are defined by the term “exposures.”13, 14, 15, 16, 17, and 18,76,117, 118, and 119 Many exposures are of limited toxicologic consequence either because of the properties of the xenobiotic involved, the magnitude or duration of the exposure, or the uncertainties regarding whether an actual exposure has occurred. Therefore, data collected by poison centers represent a limited and ill-defined measure of poisoning.
The situation is further confounded by multiple biases that are introduced by the actual reporting process, which first and foremost is voluntary and passive. Because most calls concern self reported data from the home and are never subsequently confirmed, a significant percentage of the data generated to date may actually represent only potential or possible exposures, which can introduce large statistical errors into the database. This is highlighted by the 21 of 102 children who tested positive for a toxic alcohol discussed above.78 Despite the fact that only those 21 children were definitively exposed and potentially poisoned, all 102 were entered into the database as exposures. If these figures are representative of the rest of the data set, then they suggest that an actual ingestion does not occur in the vast majority of reported unintentional exposures in children. However, they do emphasize that in all of the cases a toxin was in an unacceptably close proximity to a child. Also, current events, hoaxes, and media awareness campaigns all may influence self-reporting rates.81 Furthermore, to report a possible exposure a caller must have a telephone, probably speak English, and have some degree of health literacy and numeracy.80,115 Although telecommunications devices for the hearing impaired and translation services exist, they are rarely used. Enhancement of technology to facilitate the accurate exchange of information between poison specialists and either hearing impaired callers or those who do not speak English is essential to the success of poison centers. While text message is an interesting option, preliminary data suggest that text messaging encounters take way too long to be productive.99 Another would be to entertain more active reporting systems automatically triggered directly by hospital laboratory values. Such a system would explore cognitive behavior of clinicians around reporting and develop a true understand of the epidemiology.
Under the present passive system, when hospitals report exposures to the poison center, a comparison of the hospital chart with the poison center record shows good agreement, demonstrating an accurate exchange of information.64 Unfortunately, a reporting bias similar to that described above is well recognized regarding professional utilization of poison centers and has been called the Pollyanna phenomenon.57 For example, in the spring of 1995, poison centers in the northeast United States began to receive numerous reports of severe psychomotor agitation and other manifestations of anticholinergic syndrome in heroin users. In the initial phase of the epidemic, most of the callers requested assistance in establishing a diagnosis, determining possible etiologies, and raised questions regarding treatment with physostigmine.58 Although the epidemic continued for many months, once the media announced that the heroin supply was tainted with scopolamine, and clinicians became familiar with the indications and administration of physostigmine, call volume decreased. Stated simply, health care professionals are less likely to call the poison center regarding issues with which they are familiar, are of little clinical consequence, or are not recognized as being related to a poison. Thus, a bias is introduced that results in a relative over reporting of new and serious events and a relative underreporting of the familiar or very common, unrecognized poisoning, and those exposures or poisonings that are apparently inconsequential. Numerous comparisons support this contention. Investigators who rely on published data from poison centers as a sole source of epidemiologic information demonstrate a failure to understand the complexity of poisoning data and the aforementioned consequential limitations of poison center–derived data.
A 4 year study compared deaths from poisoning reported to the Rhode Island medical examiner with those reported to the area poison center.79 Not surprisingly, the medical examiner reported many more deaths: 369 compared with 45 reported by the poison center. Although most of the cases not reported to the poison center were victims who died at home, were pronounced dead on arrival to the hospital, or those in whom poisoning was not suspected until the postmortem analysis, 79 patients who subsequently became unreported fatalities were actually admitted to the hospital with a suspected poisoning. In 10 of these cases, the authors concluded that a toxicology consultation might have altered the outcome. Examples of interventions that, if recommended and performed, might have resulted in a more favorable outcome included the proper use of antidotes such as naloxone for an opioid overdose, N-acetylcysteine for acetaminophen (APAP) poisoning, the cyanide antidote kit, sodium bicarbonate for a cyclic antidepressant overdose, hyperbaric oxygen for carbon monoxide poisoning, hemoperfusion for a theophylline overdose, and hemodialysis for a lithium overdose. While the xenobiotics may have changed since this study was performed, the fundamental principles highlighted remain relevant.
Likewise, when medical examiner data were analyzed in Massachusetts, more than 47% of poison fatalities had not been reported to the poison center.107 A California study evaluating 358 poisoning fatalities reported to the medical examiner showed that only 10 poison center fatalities were reported over a similar time period, demonstrating an even more consequential reporting gap.7 Once again in this study, whereas the majority of underreporting was with respect to prehospital deaths (68%), only 5 of 113 hospitalized patients who ultimately died were reported to the poison center. Additionally, a cross-sectional comparison of national mortality data with poison center data for agricultural chemical poisoning demonstrated a similar trend of underreporting to poison centers of seriously poisoned admitted patients who became fatalities.72 Furthermore, when data for an entire year from the National Center for Health Statistics were compared with the same year of data from the American Association of Poison Control Centers (AAPCC), it was apparent that the AAPCC data captured only about 5% of annual poison fatalities.63
More recent analyses have highlighted a remarkable trend. When 11 states evaluated trends in poison-related mortality from 1990 to 2001, an average increase of 145% was noted.106 A more comprehensive investigation of the National Vital Statistics System accessed via the CDC’s Web-based Injury Statistics Query and Reporting System database demonstrated a 5.5% increase in injury related mortality from 1999 to 2004. Mortality from poisoning accounted for 61.9% of the increase in unintentional injury, 28% of the increase in suicide, 81.2% of the increase in death from undetermined intent, and more than half of the total increase injury-related mortality.93 As of 2004, death from poisoning surpassed firearms and became the second most common cause of injury-related fatality. While most of the fatalities are in adults, a review of the entire 2010 AAPCC database only found 74 reported fatalities in children.23 Focusing on poison center data alone would produce the erroneous assumption that poisoning-related fatalities were not a significant public health concern. In actuality, poisoning is a significant concern in that other programs designed to reduce deaths from motor vehicle crashes and firearms have been largely successful whereas decades have gone by without a major intervention targeting poison related fatality.
It is logical to assume that similar disparities exist regarding the reporting of nonfatal poisonings. The resultant gap in public health data needs to be addressed through improved definitions, epidemiology, reporting, and analysis of poison-related data systems. This inequity has developed through a long-standing tradition of poison centers to focus attention and concentrate on the largely benign exposures in children. The emphasis needs to be redirected toward seriously ill poisoning, utilization of the intensive care unit, and other markers of actual poisoning rather than health care utilization for benign events. The necessary data for such an evaluation may already exist, but the challenge will be in linking the data sets to provide a meaningful analysis.
An outreach study in Massachusetts determined that hospitals geographically close to a poison center reported their cases almost twice as often as hospitals remotely located (46% vs. 27% of total cases).28 Additionally, the authors noted that private physicians were less likely to report cases than residents in training. A one year retrospective review demonstrated that only 26% (123 of 470) of poisoned patients who were treated in a particular ED were reported to the poison center.59 Interestingly, only 3% of inhalational exposures were reported, compared with 95% of cyclic antidepressant ingestions. The authors also noted, as suggested above, that reporting decreased when comparable exposures occurred over a short period of time. Finally, in the physician survey study cited earlier, physicians reported that they would “almost never” contact the poison center for asymptomatic exposures (62.9%), chronic toxicity (50.4%), or simply to assist in establishing a reliable database (90.2%).27 This statement is most likely accurate even in jurisdictions in which the reporting of all or select exposures is incorporated into public health laws.
Xenobiotic exposure occurs commonly in the workplace. As a result of the long-recognized association between occupational exposure and illness, several federal and state government funded agencies, such as the National Institute for Occupational Safety and Health, Occupational Safety and Health Act, and the Agency for Toxic Substances and Disease Registry (ATSDR), exist to prevent occupational illness, to educate the public, and to collect data on exposures to occupational xenobiotics. Legislation provides for mandatory reporting in some instances and offers workers job protection for voluntary reporting. Poison centers also provide information on occupational exposures and collect data. Once again, there are discrepancies between the poison information data and the data collected by governmental agencies. A six month survey in California noted that only 15.9% of the occupational cases reported to the poison center were captured by a state occupational reporting system.9 The most common occupational toxicologic illness—dermatitis—was even further underrepresented in these cases. A follow-up study by the same authors demonstrated that more than one-third of calls came directly from the individual worker, 70% of whom were unaware of the link between their occupation and their symptoms.8 Although these data suggest that poison centers can provide substantial assistance following occupational exposures, one author expressed concern, noting in a follow-up study that the poison center failed to provide an adequate epidemiologic assessment in that it did not identify an average of 12 other people per workplace who were also potentially exposed in addition to the index case.11 A 1999 survey of poison centers concluded that “responses to work-related calls are inadequate” and suggested that written protocols might be helpful.12
Adverse Drug Events and Xenobiotic Errors
Although the actual numbers are a source of controversy, data suggest that a striking number of adverse drug events (ADEs) occur each year in the United States, with many resulting in death.20,31,77 The ease of 24 hour telephone access, combined with the ability to consult with a health professional, make poison centers ideal resources for reporting of ADEs.30 Yet, more than 76% of physicians surveyed stated that they would “almost never” contact the poison center regarding adverse drug events.27 Moreover, 30 of the 56 (53.6%) poison centers surveyed stated that they had not submitted any of their ADE data to the US Food and Drug Administration’s MedWatch program.34 Many of the other centers reported only partial compliance with the MedWatch system.34 Biases may lead to disproportionate reporting of adverse events related to newer drugs skewing the interpretation of the data. For example, although bleeding may occur from both coumadin and dabigatran, it is easy to believe that reports would be more likely to be generated for the newer drug.
Prescription drug errors are another source of potential poisoning. Retrospective review of poison center data suggests that many of these errors are reported. In one report, the poison center provided valuable feedback to pharmacists and physicians about these errors. Ideally, reporting to the state board of pharmacy would assure that proper surveillance and counseling continue. The poison center would seem to be ideally suited to perform this function.104 Unfortunately, while pharmacists are ideally positioned to identify prescribing errors, data suggest that pharmacist utilization of poison centers is poor.2
Poison centers also collect data on exposures to drugs of abuse and misuse. These data consist largely of calls for information from the concerned public and reports of overdose requiring health care intervention. Although ethanol, tobacco, and caffeine are the most common xenobiotics used in society, these cases are rarely reflected in poison center data, with the exception of unintentional exposures in children. In fact, because most substance abuse does not result in immediate interactions with the health care system, other databases such as the National Institute of Drug Abuse Household Survey (now referred to as the Monitoring the Future Study) might better reflect substance abuse trends. Yet even this database has significant limitations.6,54 Because poison centers are more focused on immediate health care effects of exposures, it could be argued that only those cases in which health care interaction is required are of value in the database. Since poison centers collect data in real time, centers are ideally positioned to report on emerging trends and sentinel events. Recent examples include poison center experiences with trends opioid abuse in teenagers,51,111 bath salts,88 synthetic cannabinoid receptor agonists,89 and adulterated cocaine.112
Grossly Underreported Xenobiotics
As discussed previously, there is little doubt that ethanol and tobacco are the most common xenobiotics intentionally used in our society. Although their toxicologic manifestations can be acute and severe, chronic subclinical poisoning often goes unnoticed for many years. Similarly, more than 500,000 American children have lead concentrations above 10 μg/dL and polychlorinated biphenyls can be found in countless adults and children. We must remain cognizant of these large-scale exposures, such as bisphenol A, when we read that plants, cleaning products, and cosmetics comprise the most common exposures to xenobiotics18 are the most commonly “reported” exposures.
Integrating Epidemiologic Data as Part of the Public Health Surveillance System
With the current limitations of the poison center data, it should be clear that neither the numerator nor the denominator of the actual number of poisonings can be easily appreciated. However, analysis of these data for trends may be more useful because the inherent biases involved in poison center reporting are probably consistent over many years. Increasingly, poison center data are being used as part of surveillance and prediction models,46,123 often that extend beyond poisoning to other public health concerns.105 Rapid reporting in collaboration with the CDC highlights an essential partnership.38,55,89 Efforts should be directed to encourage reporting to poison centers by such enhanced access methods as Web-based forms for passive reporting, a direct interface between laboratory and hospital databases that actively transmits data to poison centers, and linkages to other agencies that collect reports of poisoning such as state and local health departments. Additional resources should be directed at improved case definitions (distinguishing asymptomatic exposure from poisoning) and integration with other essential databases such as MedWatch, the National Vital Statistics System, and the National Center for Health Statistics.
Despite its limitations, poison center data have significant utility. It is often an exposure rather than an actual poisoning that provides the impetus for contact with health care. For those exposures that are unlikely to be consequential, the poison center can intervene to prevent potentially harmful attempts at home decontamination and costly unnecessary visits to health care professionals. Interaction with parents at a time of perceived crisis also provides a “teachable moment” (Chap. 135) that may help prevent a more consequential exposure in the future. For those exposures that may result in poisoning, the period of time immediately following exposure is an ideal moment to initiate first aid measures designed to prevent or lessen the severity of poisoning. For both of these reasons the cost, benefits, and efficacy of poison centers especially regarding home calls must be measured in terms of exposures and not poisonings.