OBJECTIVES: Lupus nephritis (LN) is a severe complication of systemic lupus erythematosus (SLE). While LN has no designated ICD-9 code, various approaches have been used to identify patients in administrative data. The objective was to compare health resource utilization of LN patients using different algorithms in a single data source.  METHODS: This study used the Impact Database, a commercial insurance claims database. SLE patients were identified using ICD-9 code 710.0 from ≥2 outpatient or 1 inpatient claims from  01/2004 to 06/2011. SLE patients with LN were further identified under four different algorithms: (A) ≥1 renal diagnosis, (B) ≥2 renal diagnoses, (C) ≥3 renal diagnoses, and (D) ≥3 renal diagnoses plus ≥3 nephrologist visits. Health resource utilization and expenditure of outpatient/emergency department (ED) visits, hospitalizations, and prescriptions were examined for 12 months post index date of first renal diagnosis. RESULTS: A total of 93,957 patients were diagnosed with SLE. Among them, 24,357, 11,054, 8,895, and 6,307 cases had LN using algorithms A-D. LN cases identified by algorithms A-D had similar mean age (48.3, 46.7, 46.3, and 45.7 years) and gender distribution (85.2, 83.1, 82.7, and 81.8% females). LN patients from different algorithms also had similar annual frequency of outpatient visit (35.9, 41.0, 41.1, and 42.1), ED visit (1.4, 1.4, 1.4, and 1.5), hospitalization (0.8, 0.9, 0.9, and 0.9), and prescription (6.8, 7.3, 7.2, and 7.4). The annual medical expenditures were $33,176, $36,974, $36,241, and $38,883 for algorithms A, B, C, and D, respectively.  CONCLUSIONS: Our results support that when studying health resource utilization, the results do not differ significantly based on the number of renal diagnoses codes. There is a difference in outcomes when requiring number of patients plus specialty subtype; however, in the case where specialty information is either unavailable or unreliable, using algorithms A-C proved equally reliable in an administrative claims database.