CO Poisoning surveillance data MMWR Surveillance Summaries articleDRAFT

1. Problem/Condition

Steve Macdonald with APRHB comments

Carbon monoxide (CO) is a colorless, odorless, nonirritating gas that is produced through the incomplete combustion of hydrocarbons. Sources of CO include combustion devices (e.g., boilers and furnaces), motor-vehicle exhaust, generators and other gasoline or diesel-powered engines, gas space heaters, woodstoves, gas stoves, fireplaces, tobacco smoke, and various occupational exposures. CO poisoning is CO poisoning is preventable; nonetheless it is a leading cause of unintentional poisoning deaths in the United States. Unintentional, non-fire related CO poisoning is responsible for approximately 450 deaths and 21,000 emergency department (ED) visits each year. (1-2) [ref: 1. Centers for Disease Control and Prevention (CDC). Nonfatal, unintentional, non--fire-related carbon monoxide exposures--United States, 2004-2006. MMWR Morb Mortal Wkly Rep. 2008 Aug 22;57(33):896-9. 2. Centers for Disease Control and Prevention (CDC). Carbon monoxide--related deaths--United States, 1999-2004. MMWR Morb Mortal Wkly Rep. 2007 Dec 21;56(50):1309-12. ]

CO poisoning may be misdiagnosed due to its nonspecific symptoms, ranging from minor flu-like symptoms (e.g., headache, dizziness, nausea, vomiting, fatigue, and confusion) to more severe effects (e.g., impaired memory, collapse, cardiac irregularities, coma, and even death). (3-4) [CU2][ref: 3. Wright J. Chronic and occult carbon monoxide poisoning: we don't know what we're missing. Emerg Med J. 2002 Sep;19(5):386-90. 4.Hampson NB, Piantadosi CA, Thom SR, Weaver LK. Practice recommendations in the diagnosis, management, and prevention of carbon monoxide poisoning. Am J RespirCrit Care Med. 2012 Dec 1;186(11):1095-101. ] Diagnosis of carbon monoxide exposure can be confirmed by measuring carboxyhemoglobin levels in the blood. CO can be removed from the body gradually by administering oxygen therapy, or rapidly in hyperbaric treatment facilities for the most severe cases. [CU3]Exposures[S4] to CO may be prevented by not using fuel based products in poorly ventilated areas, the proper installation of CO alarms and routine maintenance of home heating systems, together with adequately venting cooking and fuel-burning appliances [cite CDC guidance for individuals: Laws and ordinances requiring installation and use of CO alarms, enforcement of workplace standards as well as targeted public health messages aimed at susceptible populations, can reduce the number of CO exposures. Ongoing surveillance of CO exposure will continue to aid public health practitioners in the development of prevention measures and targeted interventions.

2. Reporting Period

NPDS: 2010-2012

State data: 2009-2011

3. Description of System

(3A)Surveillance data system methods for reporting and national notification

Steve Macdonald

The Council of State and Territorial Epidemiologists (CSTE) adopted the position statement “Public Health Reporting and National Notification for Carbon Monoxide Poisoning” (13-EH-01) in June 2013, in order to promote and standardize methods used for nationwide surveillance of carbon monoxide (CO) poisoning. (5) [ref: 5. CSTE. Public Health Reporting and National Notification for Carbon Monoxide Poisoning (13-EH-01). June 2013. ] This position statement describes methods for inclusion of CO poisoning in standard public health reporting, based on use of CO exposure and CO poisoning case data available from Poison Control Centers (PCC) as the core case-ascertainment source, and case notification to CDC by means of the American Association of Poison Control Centers (AAPCC) National Poison Data System (NPDS). (6) [ref: 6. Centers for Disease Control and Prevention (CDC). Carbon monoxide exposures--United States, 2000-2009. MMWR Morb Mortal Wkly Rep. 2011 Aug 5;60(30):1014-7. ]

The position statement describes four tiers of surveillance activities, which can vary depending upon the resources available: PCC only; case-finding using multiple data sources, including PCC; case-finding using multiple data sources with matching and record linkage; and, case-finding using multiple data sources with individual case investigation. Data sources used for CO poisoning case ascertainment include: reporting from clinicians, laboratories, hospitals and other entities; death certificates; and hospital discharge, or outpatient records. The position statement standardizes:

•Criteria for case identification, including both criteria to determine whether a case should be reported to public health authorities, and criteria for case-finding from administrative data

•Criteria for case classification for both reporting systems and administrative data, including definitions for a confirmed case, a probable case, and a suspected case

•Criteria to distinguish a new case of CO poisoning from reports or notifications which should not be enumerated as a new case for surveillance

Add text here on AAPCC mechanisms to share data (include ref to AAPCC-CSTE 2012 survey results update to include description of web services. [Steve, Jay]

From Jay:

There are different mechanisms utilized to access poison center data. All require contact and discussion with, and permission for access by, the poison center covering the geographic area for which the data is requested. Depending upon the capabilities of the public health entity and the specific poison center, the data can be obtained from the national NPDS system directly or locally through the poison center itself. The methodology available at the local poison center level may be quite variable.

Access to the NPDS web portal is granted through the local poison center and provides the user with an array of enterprise reports that can run on the poison center’s data. This access can also include the capability to set up volume and clinical effects anomaly alerts as well as case based alerts. If the public health entity has the resources, access to the NPDS’ Web Services allow the user to query the Center’s data within NPDS directly with query results returned by the web service system.

Access to local Poison Center data can encompass a more granular and complete record of the exposure as no actual PHI (other than zip code) nor actual case progress notes/comments are available in the NPDS data uploaded by poison centers. Hence, many public health entities who require more specific details of the exposure may opt for local access versus those looking for a more aggregate format gravitating to the NPDS data access model. Local poison center capabilities to provide data vary widely and can range from simple static report (aggregate or individual case) propagation either scheduled or upon request; to desktop access of local poison center data through established VPNs; to sophisticated models where poison center data is uploaded in real time for access within public health-based systems.

(3B) Surveillance data system protocols in various states

Kathy, Prakash

Include: description of:

  • how surveillance practices differ by hierarchical tier;
  • mandatory vs. voluntary case reporting;
  • criteria for case investigation;
  • utility of case data beyond “unintentional, non-fire-related carbon monoxide”;
  • definition of “cluster” variation by state [from conf call minutes: MI: >1 (death? Case?); MO: ≥2 Deaths?, ≥2 with 15% or higher carboxyhemoglobin; WI: >1 case? during one exposure event].
  • how data system protocols describe data dissemination to agency program staff for use in prevention policy development and program planning and implementation [this will bridge to section 5].

Consider inclusion of actual protocols in Appendix [which we may be able to include as “Supplemental Online Material”]

From Missouri:

“Missouri electronically collects information on all residents who die with carbon monoxide (CO) poisoning as an underlying or contributing cause. We also collect data on all residents who are reported to have a carboxyhemoglobin (COHb) laboratory test. Deaths and COHb results >=15% are considered cases and are investigated. Investigation includes requesting medical examiner/coroner reports and hospital medical records to determine intent, source of carbon monoxide exposure, work-relatedness, smoking status, and other CO cases.”

(3C) “Active” vs. “passive” case-finding: contacting reporters to remind them of reporting responsibility; assessments of completeness of reporting (“passive surveillance system with followup”)

Kathy

Cases of CO poisoning are typically identified by either ‘active’ or ‘passive’ case finding. In active case finding surveillance staff engages in finding and verifying cases of CO poisoning (e.g. real time reporting and follow-up). Passive case finding involves the receipt of case reports from reporting sources. Passive case finding may also include a follow-up component to verify the report and assure that enough information is collected to accurately classify the case. The type of case finding affects the degree to which data are considered accurate and complete.

The design of the surveillance system as well as the reporting source both determine whether a report is actively or passively collected. For example, in Maine reports are passively received from clinical laboratories, healthcare providers, and the New England Poison Control Center. Reports from these sources are then followed-up to determine the accuracy of the report and collect additional risk factor information. At the same time, Maine’s electronic reporting sources, Syndromic Surveillance and National Electronic Disease Surveillance Systems, are actively followed to identify cases that are then followed-up[DK5].

Aside from the mechanics of the underlying surveillance system, reporting sources must know and understand who, what, when and how to report cases of CO poisoning. Initial and continued education and outreach to reporting sources that describe the parameters of reporting and whether reporting is mandatory are key to completeness of reporting. For example, in Maine, an evaluation of the system identified under-reporting from hospitals. This information was used to inform through outreach with hospitals to improve case reporting.

(3D) Data sources: obtaining datasets retrospectively to confirm completeness of reporting

Kathy

Case reports of CO poisoning typically come from multiple reporting sources, such as, administrative billing data, poison control center reports, healthcare providers, and clinical laboratories. For newly established surveillance systems it is important to understand the degree to which cases are reported to the surveillance system, i.e. completeness of reporting.

To understand the completeness of a particular reporting source there must be a secondary source for comparison. A surveillance systems ability to detect under-reporting of cases varies based on the level of information collected by the system. For surveillance systems that do not collect individual level data, completeness may be assessed by comparing frequencies of cases reported in a time period against secondary data sources for the same time period. In systems where individual level data are collected more accurate means can be employed, such as, matching surveillance data collected against secondary data sources to assess under-reporting of cases. Three secondary sources are typically available for case matched surveillance data; administrative billing data, poison center control data, vital statistics data (mortality).

Administrative billing data are a good secondary source to assess completeness of hospitalizations and/or ED visit reports for CO poisoning. Primary or secondary discharge code(e.g. 986) and where available external cause of injury codes (e.g. E8682,E8683, E8688, E8689, E9820, E9821)assigned to administrative data are used to identify unintentional non-fire CO poisonings. These data may then be matched by date of birth, sex, admission date to surveillance data to identify potentially unreported cases.

Poison center control data are collected routinely and may be available for direct matching against surveillance data…..I need some statement about how available these data are across the country to state programs. I am not sure of this…. In Maine, we used weekly case counts and compare those to our data collected to see if we missed a report. We do not use matching.

Vital statistics data is another secondary source used to identify missed cases of CO poisoning deaths. Deaths having any contributing or underlying cause of death coded as T58, toxic effect of carbon monoxide from all sources, may also be matched by date of birth, sex, decedent name to surveillance data.

(3E) Obtaining datasets retrospectively to supplement case reports

Kathryn Lane

In addition to tracking hospital visits, deaths, and Poison Control Center data, health departments may obtain other sources of data to supplement CO health surveillance and case investigation. For instance, environmental data from fire departments or utility companies– which typically are the first responders to CO incidents triggered by an alarm or an adverse health event – may be obtained retrospectively. In New York City, data from the fire department is obtained on a yearly basis. Data are classified as incidents where CO was not detected at the scene upon arrival, detected at low levels (1-9 ppm), or detected at emergency levels (> 9 ppm[S6]). Data also include information on whether a CO alarm activated or not. While this type of data may not include any information on numbers of people affected, health symptoms, or health outcomes, it can allow for assessment of the burden of response to CO incidents (including “false alarms” or those that do not result in contact with the healthcare system), tracking of CO levels found in the environment, and measurement of changes in the detection of indoor CO emission hazards over time. For instance, this data can be used to track the number of serious incidents detected by a CO alarms and how this number may change over time.

(3F) “interstate reciprocal notification”: non-resident data exchange and/or dual data display by location of residence and location of exposure; location of diagnosis issues

Steve Macdonald

For the most part, state and municipal CO poisoning surveillance data systems do not use the methods developed for use in infectious disease (ID) surveillance termed “interstate reciprocal notification” (IRN). In ID surveillance, non-resident case information is sent to the jurisdiction of residence, because cases are assumed to have acquired their disease at their place of residence, and the jurisdiction of residence is then used in official case counts for statistical purposes. Standard forms are used for this purpose [cite CDC form and Florida form ]. In chronic disease surveillance and environmental disease surveillance, exposure pathways may be less certain, and diagnosis may occur at some distance from the location of residence. In the absence of standardized IRN, some jurisdictions have established informal means of non-resident data exchange, but most have not. For this reason, the surveillance data may be best understood when there is dual data display by location of residence and location of exposure, in separate tables. This practice is not yet widely used.

Prakash: please edit/expand as needed. Also, please confirm or correct the Florida IRN form URL.

(3G) NPDS scenario and substance-based fields

Prakash

The primary core functions of Poison Control Centers (PCCs) includes patient exposure assessment and management, accurate data collection and coding through the use of an electronic medical record, and response to the continuing need for poison-related public and professional education. Calls received at US PCCs are managed by healthcare professionals who have received specialized training in toxicology and the assessment, triage, management and monitoring of toxic exposure emergencies. Fifty-six participating centers cover all states and territories of US and submit their data to the American Association of Poison Control Centers’ (AAPCC) National Poison Data System (NPDS). NPDS is the data repository for all poisoning and information calls received by all poison centers across the country. All centers upload their call data near-real time through an automated upload process. The web-based NPDS software has the capability for data analysis through enterprise reporting, volume and clinical effects anomaly detection, and the recognition and reporting of NPDS surveillance anomalies. PCCs can also elect to share NPDS surveillance technology with external organizations such as their state and local health departments or other regulatory agencies. Few states in Tier 4 have direct access to all the collected PCC data fields through an automated data transmission.

PCCs collect additional fields that are not uploaded to NPDS and are only available to their local or state partners. These fields may include personal identifiers, full case note information, specific labs or other non-NPDS fields. Likewise, some systems (e.g. Florida) have elected to also deploy “rule-based coding” which allows for PCCs using this methodology to code additional field based on the substance and some specifics of the exposure. This added information helps to collect additional information useful to conduct surveillance and identify areas for prevention. Adoption of rule-based code by the AAPCC and including additional substance specific fields would enhance NPDS ability to conduct CO surveillance.

(3I) Case definition for unintentional non-fire CO poisoning (APRHB, Tracking Network)

Jackie/Kanta, Steve

Describe the differences between APRHB (which focuses on unintentional non-fire) and Tracking (which also includes fire-related and unknown intent/mechanism). Describe differences (if any) between unintentional non-fire case ascertainment and classification used by Tracking vs. APRHB.