Carbon Monoxide Poisoning in the Veterans Health Administration, 2010 - 2016

Gina Oda, Russell Ryono, Cynthia A. Lucero-Obusan, Patricia Schirmer, Mark Holodniy


ObjectiveTo describe characteristics of Veterans Health Administration(VHA) patients with ICD 9/10 CM inpatient discharge and/oremergency department (ED)/urgent care outpatient encounter codesfor carbon monoxide (CO) poisoning.IntroductionIt is estimated that in the United States (US), unintentional non-firerelated CO poisoning causes an average of 439 deaths annually, and in2007 confirmed CO poisoning cases resulted in 21,304 ED visits and2,302 hospitalizations (71 per million and 8 per million population,respectively)1. Despite the significant risk of morbidity and mortalityassociated with CO poisoning, existing surveillance systems in theUnited States are limited. This study is the first to focus specificallyon CO poisoning trends within the VHA population.MethodsQueries were performed in VA PraedicoTMPublic HealthSurveillance System for inpatient discharges and emergency roomand urgent care outpatient visits with ICD 9/10 CM codes for COpoisoning from 1/1/2010 – 6/30/2016. A dataset of unique patientencounters with CO poisoning was compiled and further classified asaccidental, self-harm or unspecified. Patients with carboxyhemoglobin(COHb) blood level measurements≥10%2for the same timeframewere extracted and merged with the CO poisoning dataset.We analyzed for demographic, geographic and seasonal variables.Rates were calculated using total unique users of VHA care formatching time frame and geographic area as denominators.ResultsThere were a total of 671 unique VHA patients identified with COpoisoning. Of these, 298 (44%) were classified as accidental, 104(15%) self-harm, and 269 (40%) unspecified. A total of 6 patientsdied within 30 days of their coded diagnosis, however only 1 ofthese was directly attributable to CO poisoning. The overall rate ofCO poisoning over the study time frame was 18 per million uniqueusers of VHA care. CO poisoning diagnoses were obtained from396 (59%) outpatients, 216 (32%) inpatients, and 59 (9%) patientswith both and outpatient visit and inpatient admission. Patientswith self-harm classification were less likely to be seen in the ED(only 24 (6%) unique patients compared to 190 (48%) accidental and182 (46%) unspecified classifications). Of patients seen in the ED andsubsequently admitted, patients with the classification of accidentalpoisoning made up the largest percentage with 36 unique patients(61%). There were 71 (11%) females compared to 600 (89%) males.The highest represented age group was 45-64 with 342 unique patients(51%). Rates by US Census Region were highest in the Midwestand Northeast (27 and 23 per million unique users, respectively)compared to the West and South (15 and 13 per million uniqueusers, respectively) (Figure 1). Accidental CO poisonings showed aseasonal pattern with peaks occurring in late fall, winter, and earlyspring months (Figure 2). CO poisonings classified as unspecifiedhad a similar but less pronounced pattern, while those classified asself-harm were too few to observe any pattern over time. COHb bloodlevels≥10% were present in 111 (17%) of patients with CO poisoningcodes. Of patients with COHb measures≥10%, those with self-harmclassification were least represented with only 7 unique patients (6%).Accidental and unspecified classifications were equally representedwith 53 (48%) and 51 (46%) unique patients, respectively.ConclusionsThe impact of CO poisoning on the VHA patient population hasnot been well studied. The geographic distribution of the majorityof cases in the Midwest and Northeast, and the seasonal distributionof accidental cases in colder months seems to be appropriate withrespect to what is known of unintentional CO poisoning as oftenassociated with heat-generating sources3. Opportunities for furtherinvestigation include how potential CO poisoning cases are evaluatedin VHA given the low percentage of cases with COHb blood levelmeasurements.

Full Text:



Online Journal of Public Health Informatics * ISSN 1947-2579 *