About the carbon monoxide poisoning data

The following provides general information about carbon monoxide poisoning data and measures developed by the Minnesota Environmental Public Health Tracking (MN EPHT) Program. For more information about these data, contact MNPH Data Access.

  • The numbers and rates of unintentional non-fire related carbon monoxide (CO) poisoning hospitalizations, emergency department (ED) visits, or deaths by year, gender, or age group in Minnesota.
  • If a measure is going up or down over time.
  • If a segment of a population is at higher risk for hospitalization, a visit to the ED, or death resulting from unintentional non-fire related CO poisoning in Minnesota.
  • To inform the public about unintentional non-fire related CO poisoning hospitalizations, ED visits, and deaths in Minnesota.
  • State and local partners can use these data for program planning and evaluation.
  • What the source of exposure is for CO poisoning
  • The total burden of CO poisoning in a population
  • The number of people hospitalized or who visited the ED for CO poisoning. Because personal identifiers are removed from the hospital discharge data before analysis, individuals who have multiple hospitalizations cannot be identified.
  • Hospitalization and ED data are extracted from Minnesota Hospital Discharge Data (MNHDD), which is maintained by the Minnesota Hospital Association (MHA).
    • MHA represents Minnesota's hospitals and health systems. Hospitals submit inpatient discharge data to MHA using a standardized billing form. In 2010, 99.3% of all hospitals in the state report hospital discharge data to the MHA, representing 99.4% of all licensed beds in the state.
    • MHA began data-sharing agreements with several states in 2005. Minnesota residents receiving care from hospitals from the participating border states of North Dakota, South Dakota and Iowa are also included in hospitalization measures beginning in 2005. Minnesota residents receiving care from emergency departments from North Dakota are also included in emergency department measures beginning in 2005.
  • Rates are calculated using denominator counts from the US Census. Data from 2000 and 2010 are from the Decennial Census. Data from 2001-2009 and 2011 onward are from intercensal population estimates.
  • Mortality data are extracted from the Minnesota Mortality Database, which contains information on demographic and cause of death data collected from death certificates, and is maintained by the MN Center for Health Statistics at the Minnesota Department of Health.
  • Hospitalizations visits are defined as Minnesota residents who are discharged from a hospital in Minnesota or the bordering states of North Dakota, South Dakota, or Iowa. Emergency Department visits are defined as Minnesota residents who are treated and released or subsequently admitted to a facility in Minnesota or North Dakota.
  • CO poisoning hospitalizations/ED visits are defined as having a diagnosis of acute, unintentional non-fire related CO poisoning (ICD-9-CM codes: 986, E868.2, E868.3, E868.8, E868.9, E982.0, E982.1), without any intentional poisoning or injury codes (E952.0, E952.1, E950.0-E979.9, E990.0-E999) or fire related codes (E890-E899); and accompanied by any non-fire related codes (E818, E825, E838, E844, E867, E868).
  • Starting October 1st, 2015, ICD-10-CM codes were used to identify acute, unintentional non-fire related CO poisoning. ICD-10-CM codes: T58.01, T58.11,T58.2X1,T58.8X1, T58.91, excluding if any of these codes (T58.02, T58.03, T58.12, T58.13, T58.2X2, T58.2X3, T58.8X2, T58.8X3, T58.92, T58.93, X01-X08).
  • Deaths are defined as Minnesota residents with ICD-10 code T58 (toxic effect of CO) listed as a contributing cause of death on the death certificate record, and accompanied by any of the following ICD-10 codes as a contributing cause of death: V01-W99, X10-X59, or Y85-Y86.
  • Records with missing county are included in the state count but excluded from county counts. Records with missing gender are included in the combined gender count but excluded from by gender counts.

Number:

  • If you want to understand the magnitude or how big the overall burden is, then use the number.
  • The number indicates the total number of hospitalizations, ED visits, or deaths, but not the number of unique individuals hospitalized or admitted to the ED.
  • To protect an individual's privacy, hospitalization and ED visit counts from 1 to 5 are suppressed if the underlying population is less than or equal to 100,000.

Rate:

  • If you want to understand the probability or what is the underlying risk in a population, then use a rate and confidence interval. A rate is a ratio between two measures with different units. In our analysis a rate is calculated using a numerator (the number of CO poisoning hospitalizations, ED visits, or deaths during a period of time) divided by a denominator (the number of people at risk in a population during the same period of time). This fraction is then multiplied by 100,000. Mid-year Minnesota population estimates from the U.S. Census Bureau are used to calculate a rate.
  • To protect an individual's privacy, rates based on hospitalization or ED visits counts from 1 to 5 are suppressed if the underlying population is less than or equal to 100,000.
  • Rates based on counts of 20 or less are flagged as unstable and should be interpreted with caution. These rates are unstable because they can change dramatically with the addition or subtraction of one case.
  • Age-adjusted rate:
    • Age-adjusted rates are useful when comparing the rates of two population groups that have different age distributions
    • A weighted average, called the direct method, is used to adjust for age in this analysis. Age specific rates in a given population are adjusted to the age distribution in a standard population by applying a weight. The U.S. 2000 Standard population is used as the basis for weight calculations.
  • Assessing the confidence interval for the rate is one approach to determine whether there are differences over time or compared to another group. If they do not overlap, then they differ. Although it is not a true statistical test, it is a commonly accepted way to compare rates between groups.
  • A confidence interval for a rate is a measure of reliability. In this analysis, 95% confidence intervals were calculated. 95% confidence intervals is the interval within which the true value of the rate would be expected to fall 95 times out of 100. When the number of events is fewer than 100, the 95% confidence interval is calculated based on the inverse gamma distribution in this analysis. When the number of events is 100 or greater, the 95% confidence interval is calculated based on normal approximation.
  • Symptoms of CO poisoning are non-specific and can be easily misdiagnosed, resulting in uncounted hospitalizations or ED visits. CO poisoning deaths may be misclassified and attributed to other causes.
  • Since only those with the most severe or acute symptoms of CO poisoning are hospitalized, treated at emergency departments, or die, these data are not appropriate for estimating the total burden of CO poisoning in a population. The actual number of Minnesotans who experience carbon monoxide poisoning is unknown. Carbon monoxide poisoning currently is not a reportable condition in Minnesota.
  • The International Classification Codes (ICD) are developed by the World Health Organization (WHO) and used to document the type of illness or injury. ICD changes occurred in the final three months of 2015. This may impact disease rates starting with annual estimates in 2015 and will vary by disease. Rates from 2000-2014 should not be compared to rates 2015 and later.
  • Multiple hospital or emergency department admissions by the same patient can not be identified, and are not excluded.
  • Minnesota residents discharged from Wisconsin facilities are not included, so hospitalization and ED rates for counties in which residents are likely to receive care from Wisconsin may be underestimated. Rates for counties in which residents are likely to visit hospitals that do not submit data to the Minnesota Hospital Association (e.g., Veteran's Administration or Indian Health Services hospitals) may also be artificially low.
  • There is usually a two year lag period before hospitalization and ED data are available.

To learn more about the CO poisoning data and measures, contact the Data Access portal.