of all poisoning episodes, based on the 24 percent of estimated NHIS episodes that were treated solely by a telephone call.
A subset of 188 records (out of a total of 120,464), including an ICD-9-CM diagnosis or E-code for poisoning, was extracted from the National Ambulatory Care Medical Survey 1997–2001 data files, resulting in an annual estimate of approximately 1,582,000 visits (Table 3-4). This represents approximately 0.2 percent of all doctor’s office visits annually estimated through this survey. This estimate is nearly four times higher than what might be expected given the estimates from the NHIS (as shown in Table 3-3). Patterns of rates among the various demographic groups demonstrated similarities and differences compared with the NHIS data. Male and female patients in NAMCS had similar rates of poisoning-related visits, as opposed to the lower rates for males in the NHIS.
Rates for the various ethnic groups are presented in Table 3-4, but are too sparse in the nonwhite categories to be estimated with precision. In addition, 18 percent of the patient visits were associated with unknown race/ethnicity. Visit rates were highest in the Midwest and West and were lowest in the South. Although the Midwest was also highest in the NHIS (see Table 3-2), the other regions appear to differ in their rank order based on the NAMCS data.
Two-thirds of these visits were associated with an ICD-9 external cause of injury E-code of poisoning; a slightly smaller proportion (56 percent) was associated with an ICD-9 diagnosis code of poisoning (multiple codes possible for the same event). The most common ICD-9 codes were “toxic effects of other substances (venom; bites of venomous snakes, lizards, and spiders; tick paralysis)” (989.5), followed by “accidental poisoning by unspecified substance” (E866.8) and “accidental poisoning by unspecified drug” (E858.9). Observations are too sparse to generate reliable incidence estimates by category of specific ICD-9 code.
The NAMCS survey is one in which “patient reason for visit” data could be present, coded (not by the ICD-9 scheme) from an open-ended “chief complaint” or main symptom from the patient’s perspective. Despite this option, in practice concomitant “patient reason” poisoning codes were relatively infrequent. It should also be noted again that cases were not selected for inclusion or exclusion in the principal analysis based on patient reason codes (see Methods in Appendix 3-A). Had this been a basis for inclusion (e.g., not confirmed by a concomitant ICD-9 diagnosis or E-code for poisoning), only 12 observations would have been added, an increase of 6 percent (total estimate of 1,689,000 visits, rather than 1,575,000). The relative rank of “patient reason for visit” responses will be