Table 2. Immune-Related AEs After ICIs Among Patients Diagnosed With First Primary Melanoma During 2011-2015^a.

Immune-related AEsb	No. included in analysisc	No ICI or event prior to ICI [reference], No. of eventsd	After ICI
			No. of eventsd	HR (95% CI)e	Wald P value
Autoimmune-related AEs	3114	304	45	2.5 (1.6-4.0)	<.001
Endocrine	4420	50	38	8.8 (4.3-18.0)	<.001
Primary adrenal insufficiency	4460	39	34	9.9 (4.5-21.5)	<.001
Gastrointestinal	4219	73	20	3.5 (1.6-7.6)	.001
Regional enteritis/Crohn disease	4435	<11	<11	3.9 (0.9-17.1)	.07
Ulcerative colitis	4417	26	15	8.6 (2.8-26.3)	<.001
Miscellaneous
Asthma	3858	142	<11	0.7 (0.2-1.9)	.46
Other immune-related AEs	4489	1712	146	2.2 (1.7-2.8)	<.001
Endocrine	3022	271	52	3.3 (2.0-5.2)	<.001
Cushing syndrome	4477	<11	<11	11.8 (1.4-97.2)	.02
Thyrotoxicosis with or without goiter (hyperthyroidism)	4283	40	<11	6.3 (2.0-19.5)	.001
Hypopituitarism	4479	11	14	19.8 (5.4-72.9)	<.001
Hypothyroidism	3093	262	54	3.8 (2.4-6.1)	<.001
Other disorders of pituitary gland (includes hypophysitis)	4483	<11	<11	6.0 (1.2-30.2)	.03
Gastrointestinal	4489	501	106	3.0 (2.2-4.1)	<.001
Gastroenteritis and colitis, excluding ulcerative colitis	4489	31	<11	2.2 (0.7-6.7)	.17
Diarrhea	4489	404	95	3.5 (2.5-4.9)	<.001
Stomatitis and mucositis (including ulcerative, aphthous)	4489	37	<11	1.3 (0.4-3.8)	.66
Myalgia and myositis, not otherwise specified	4489	311	17	1.5 (0.8-2.9)	.20
Vitiligo	4477	14	<11	2.1 (0.5-8.3)	.30
Septicemia, sepsis	4489	280	59	2.2 (1.4-3.3)	<.001

Abbreviations: AEs, adverse events; HR, hazard ratio; ICI, immune checkpoint inhibitor.

^a Study population restricted to patients of White race diagnosed with American Joint Committee on Cancer stages II-IV or unknown stage cutaneous melanoma during 2011-2015 as identified in the Surveillance, Epidemiology, and End Results-Medicare linked database.

^b Outcomes for which there were at least 5 events in each group are presented. The full list of immune-related AEs is found in eTable 2 in the Supplement; HRs are not presented because counts were too small for models to be stable.

^c Immune-related AEs were designated as chronic or transient. For chronic diseases, it is assumed that an individual can have only 1 incident diagnosis per lifetime. For transient diseases, it is assumed that an individual can have multiple incident events per lifetime. Variation is seen in the analytic sample size for chronic diseases (designated in eTable 2 in the Supplement) owing to exclusion of individuals with a baseline history of the disease. For models of transient diseases, individuals with a baseline history of the diseases were included in analyses; therefore, the total sample size remains fixed.

^d Number of events captures the number of people with a claim for the event following melanoma diagnosis. The exact number is not shown if there were less than 11 events to protect patient confidentiality.

^e Hazard ratios and 95% CIs estimated from multivariable Cox proportional hazards regression with person-years as the time scale and stratified by calendar year of melanoma diagnosis. All models were adjusted for age at melanoma diagnosis (66-69, 70-74, 75-79, or 80-84), sex, stage at melanoma diagnosis, and NCI comorbidity index (0, >0-1 -<1.69, ≥1.69; cut points were derived from individuals with non-0 values), and time-dependent variables for history of autoimmune and nonautoimmune disease and for chemotherapy, radiotherapy, and other types of immunotherapy. Models of transient outcomes were further adjusted for baseline history of that disease.