Abstract
Background/Aims Stevens-Johnson syndrome (SJS) and toxic epidermal necrolyssis (TEN) are life-threatening widespread serious skin rashes (SSR), often accompanied by extensive epidermal detachment and/or mucous membrane involvement. Recent genetic studies have shown a strong link in Han Chinese between HLA-B*1502 and risk for SJS with use of Carbamazepine; however, more research is needed to identify similar markers among other ethnic groups. With support from The International Serious Adverse Event Consortium, we evaluated the potential for using electronic clinical and administrative data from Kaiser Permanente Hawaii and Kaiser Permanente Georgia to identify provisional SSR cases.
Methods We searched inpatient encounters, problem lists and allergy tables for codes specific to SJS/TEN and for related non-specific codes for erythema multiforme, drug dermatitis, and drug allergies. All records from 1/1/00–8/1/09 that were identified as potential cases were reviewed manually for confirmation and abstraction of suspected implicated drugs. In addition, we reviewed medical records of patients with two additional SSR: angioedema on ACE inhibitors, and drug hypersensitivity syndrome (DHSS).
Results Of 29 patients with codes specific for SJS/TEN, 23 (79%) were confirmed on chart review. Of 140 patients with non-specific SJS/TEN codes, 13 (9%) were confirmed. The most commonly implicated medications included amoxicillin, dilantin, trimethoprim and sulfamethoxazole, and isoniazid. Of 56 subjects with angioedema while on ACE inhibitors, 20 (36%) were confirmed. Of patients with DHSS, 1 of 29 (3%) that had inpatient codes plus abnormal lab values were confirmed on record review, while only 1 of 51 (2%) with inpatient DHSS codes and normal lab values were confirmed.
Conclusions We found that large linked electronic HMORN datasets efficiently identified highly probable cases of SJS/TEN. Searches for angioedema following ACE inhibitor exposure were also relatively successful. However, searches for DHSS were unsuccessful and only yielded 2 potential cases. These findings will be useful for planning the next study phase where patients will be recruited for genomic studies from a larger number of HMO sites.
