Risk of asthma and autoimmune diseases and related conditions in patients hospitalized for obesity

Abstract

Background. Although there are putative mechanistic links between obesity and autoimmune diseases, obesity is not considered a risk factor for most autoimmune diseases.

Methods. Using the nation-wide Hospital Discharge Register we defined a cohort of 29,665 patients hospitalized for obesity since year 1964. The patients were followed for hospitalization for any of 34 autoimmune or related conditions through year 2007. Standardized incidence ratios (SIRs) were calculated for autoimmune diseases in obese individuals compared to those who had not been hospitalized for obesity.

Results. Among 22 immune diseases diagnosed after hospitalization for obesity and in at least 5 patients, the overall SIR was 2.05. Of the individual diseases studied, the risk of 16 was significantly increased; none displayed a decreased risk. Psoriasis (4.54) and Behçet's disease (4.49) exhibited the highest risks, followed by Hashimoto's disease/hypothyroidism (4.12) and asthma (3.39). Small but significant increases in SIRs were also noted for the common autoimmune diseases Graves's disease/hyperthyroidism (1.28) and rheumatoid arthritis (1.37).

Conclusions. The present population of obese individuals, subsequently diagnosed with a number of autoimmune diseases and related conditions, was hospitalized at a relatively young age. Further studies are needed to describe the morbidity in the obese population at large.

Key words::

• Asthma
• autoimmunity
• comorbidity
• overweight
• psoriasis

Key messages

• Obesity and autoimmune diseases share putative mechanistic links. However, obesity is not considered a risk factor for most autoimmune diseases.

• In the present study, the risk of 16 of the selected 34 autoimmune diseases and related conditions was increased in individuals who had previously been hospitalized for obesity.

Introduction

Obesity is a risk factor for many common diseases, including asthma, autoimmune disease psoriasis, type 2 diabetes, and cancer (1–7). Autoimmune processes are usually triggered by environmental or internal stimuli, the latter of which could include metabolic disturbances related to obesity (8). At the physiological level, obesity-related chronic metabolic and hormonal disturbances affect the IGF pathway, adipokines, and steroid hormone metabolism (6,9,10). The adipokines, leptin and adiponectin, are synthesized in the white adipose tissue at the site of energy storage, and they act as links between nutritional status, metabolism, and immunity (11,12). Leptin is considered a safety factor that regulates body-weight by suppressing appetite and stimulating energy expenditure. It is also a proinflammatory mediator that may regulate autoimmune responses (13–15). Adiponectin senses insulin levels and is down-regulated in obesity. In visceral obesity, the reduction in adiponectin levels is associated with a parallel reduction in the plasma concentration of interleukin-10, a key anti-inflammatory cytokine regulating the immune system (13). The inflammasome is a related sensor of metabolic disturbances, which regulates the release of the potent proinflammatory cytokine interleukin-1 beta (16). This cytokine is believed to be an important driver of many diseases, including type 1 and type 2 diabetes and autoinflammatory diseases (16). Interleukin-1 beta receptor antagonist, anakinra has shown efficacy in rheumatoid arthritis and type 2 diabetes (16).

The above findings suggest that immunological processes and mediators may link obesity and autoimmune diseases. Indeed, adiponectin and inflammatory cytokines have recently emerged as possible mechanistic links between these conditions (13–16). In the present study we defined a cohort of 29,665 patients who had been hospitalized for obesity since year 1964 and followed them for hospitalization for any of 34 autoimmune diseases and related conditions through year 2007. The risk of several autoimmune diseases was higher in obese individuals compared to those who had never been hospitalized for obesity. The strengths of the present study were the nation-wide recruitment of patients, the prospective design, and the diagnostic accuracy for both obesity and autoimmune and related diseases.

Material and methods

Data on obesity and autoimmune and related diseases were obtained from the Swedish Hospital Discharge Register, which contains complete data on all discharges, with dates of hospitalization and diagnoses. The register contains data for some regions of Sweden since 1964 and has had nation-wide coverage since 1986. The data were accessed via the MigMed 2 database, which was constructed by linking several national Swedish registers and which has been described previously (17). All linkages were performed using the individual national identification number that is assigned to each person in Sweden for their lifetime. This number was replaced by serial numbers in order to provide anonymity. Over 11.8 million individuals are included in the MigMed 2 database.

The International Classification of Diseases codes used for obesity as the main diagnosis were ICD7 287.00, 287.09; ICD8 277.99; ICD9 278A; and ICD10 E65–E68. The codes used for autoimmune and related diseases have been described previously (18,19). Risk of autoimmune disease was determined for persons who had previously been hospitalized for obesity any time between 1964 and 2007 and compared with that for persons who had not been hospitalized for obesity. The analyses were limited to a population aged at least 10 years because for cases (i.e. autoimmune disease hospitalization after hospitalization for obesity) hardly any young children were found. However, this age limitation had a minimal effect on the risk estimates. The 34 diseases studied comprised autoimmune and related diseases, selected based on etiological hypotheses, case numbers, or previous positive study findings (asthma selected as a proof of principle). Person-years for autoimmune and related diseases were calculated from hospitalization for obesity until first hospitalization for an autoimmune disease or related condition, death, emigration, or the study end date (31 December 2007). Standardized incidence ratios (SIRs) were calculated to compare the relative risk of hospitalization for autoimmune and related diseases in obese persons and individuals who had never been hospitalized for obesity. The SIR was calculated as the ratio of observed (O) to expected (E) number of autoimmune cases by the indirect standardization method using the following formula:

where: O = ∑O_j denotes the total observed number of autoimmune disease cases in the study group (hospitalized for obesity); E* is calculated by applying stratum-specific standard incidence rates (λ_j*) obtained from the reference group (no hospitalization for obesity) to the stratum-specific person-year (n_j) experience of the study group; o_j represents the observed number of cases that the cohort subjects contribute to the jth stratum; and J represents the strata defined by the cross-classification of various adjustment variables, including age (5-year groups), sex, socio-economic status, period (5-year group), and region (six groups) (20,21). Confidence intervals at the 5% level (95% CIs) were calculated assuming a Poisson distribution (22). All of the analyses were performed using SAS software v. 9.2 (SAS Institute Inc., Cary, NC, USA).

The study was approved by the regional ethical review board of Lund University.

Results

The number of patients hospitalized for obesity as the main diagnosis was 29,665, of whom 23.1% were men and 76.9% were women (Table I). The age-groups in which first hospitalizations were most common were 40–49 years for men and 30–39 years for women. Only about 10% of first hospitalizations occurred in subjects older than 59 years. Table I also shows the numbers of diagnoses of autoimmune diseases and related conditions (total 367,459) in the control group (individuals never hospitalized for obesity).

Table I. Number of cases of first hospitalizations for obesity in men and women and number of controls hospitalized for autoimmune and related disorders without prior hospitalization for obesity.

	First hospitalizations for obesity						First hospitalizations for autoimmune and related disorders without obesity
Age at diagnosis (years)	Men		Women		All
	n	%	n	%	n	%	n	%
10–19	816	11.9	1064	4.7	1880	6.3	37443	10.2
20–29	746	10.9	3009	13.2	3755	12.7	36241	9.9
30–39	1378	20.1	6372	27.9	7750	26.1	42321	11.5
40–49	1729	25.2	5926	26.0	7655	25.8	47021	12.8
50–59	1450	21.2	4236	18.6	5686	19.2	58111	15.8
60–69	562	8.2	1595	7.0	2157	7.3	63701	17.3
70–79	136	2.0	491	2.2	627	2.1	57307	15.6
≥80	38	0.6	117	0.5	155	0.5	25314	6.9
All	6855	100.0	22810	100.0	29665	100.0	367459	100.0

Of the autoimmune and related diseases diagnosed after obesity, only the 22 diseases diagnosed in at least five individuals are shown in Table II (none of the others had significant SIRs). The SIR for all autoimmune and related diseases was 2.05. Asthma accounted for more than one-third of all cases; the SIR without asthma was lower, 1.69. Among the individual diseases, the SIRs for 16 were significantly increased, while no SIR was decreased. The highest SIRs were found for psoriasis (4.54) and Behçet's disease (4.49), followed by Hashimoto's disease/hypothyroidism (4.12) and asthma (3.39). Small but significant increases were also noted for the common autoimmune diseases Graves's disease/hyperthyroidism (1.28) and rheumatoid arthritis (1.37). The highest risks for many diseases were found within 1 year of hospitalization for obesity. However, for almost all diseases with a significant overall excess risk, increased SIRs were observed at some point during follow-up other than the first year after hospitalization for obesity. The two exceptions were immune thrombocytopenic purpura and multiple sclerosis, although their SIRs were above unity at all follow-up times.

Table II. SIR for subsequent autoimmune and related disorders of patients with obesity.

	Follow-up interval (years)
	<1				1–4				5–9				≥10				All
Autoimmune disorders	O	SIR	95% CI		O	SIR	95% CI		O	SIR	95% CI		O	SIR	95% CI		O	SIR	95% CI
Amyotrophic lateral sclerosis	0				2	1.40	0.13	5.14	1	0.53	0.00	3.03	3	0.62	0.12	1.83	6	0.72	0.26	1.57
Ankylosing spondylitis	0				4	2.37	0.62	6.12	7	4.38	1.73	9.07	4	1.59	0.41	4.12	15	2.47	1.38	4.09
Asthma	42	10.66	7.68	14.42	74	2.74	2.15	3.44	101	3.70	3.01	4.49	142	2.98	2.51	3.51	359	3.39	3.05	3.76
Behçet's disease	2	22.22	2.09	81.72	2	3.39	0.32	12.47	3	5.45	1.03	16.15	1	1.82	0.00	10.42	8	4.49	1.92	8.90
Chronic rheumatic heart disease	0				4	1.12	0.29	2.91	5	1.39	0.44	3.28	2	0.33	0.03	1.20	11	0.80	0.40	1.43
Crohn's disease	2	1.59	0.15	5.84	9	1.16	0.53	2.21	14	1.86	1.01	3.13	17	1.36	0.79	2.18	42	1.45	1.04	1.96
Diabetes mellitus type 1	1	3.70	0.00	21.23	7	3.80	1.51	7.88	1	0.85	0.00	4.90	1	2.63	0.00	15.08	10	2.73	1.30	5.04
Graves's, hyperthyroidism	6	2.22	0.80	4.87	15	0.86	0.48	1.42	21	1.25	0.77	1.91	41	1.47	1.05	1.99	83	1.28	1.02	1.58
Hashimoto's/hypothyroidism	11	23.91	11.87	42.93	22	7.10	4.44	10.76	5	1.57	0.50	3.70	16	2.51	1.43	4.08	54	4.12	3.09	5.37
Immune thrombocytopenic purpura	1	6.67	0.00	38.22	2	2.15	0.20	7.91	3	2.97	0.56	8.79	4	1.67	0.44	4.33	10	2.23	1.06	4.12
Multiple sclerosis	3	2.80	0.53	8.30	10	1.51	0.72	2.78	9	1.43	0.65	2.73	16	1.64	0.94	2.67	38	1.60	1.13	2.20
Pernicious anemia	1	6.25	0.00	35.83	1	0.84	0.00	4.82	2	1.59	0.15	5.84	7	3.30	1.31	6.84	11	2.33	1.15	4.18
Polymyalgia rheumatica	2	3.03	0.29	11.14	10	1.97	0.94	3.64	13	2.08	1.10	3.56	23	1.35	0.85	2.02	48	1.65	1.22	2.19
Primary biliary cirrhosis	0				1	2.27	0.00	13.03	1	1.82	0.00	10.42	5	3.57	1.13	8.40	7	2.85	1.13	5.90
Psoriasis	8	10.39	4.44	20.57	28	5.48	3.64	7.93	25	4.96	3.21	7.33	28	3.23	2.14	4.67	89	4.54	3.65	5.59
Rheumatoid arthritis	5	1.79	0.57	4.22	25	1.29	0.83	1.90	29	1.41	0.94	2.03	55	1.36	1.02	1.77	114	1.37	1.13	1.65
Sarcoidosis	1	1.75	0.00	10.06	6	1.63	0.59	3.58	8	2.33	0.99	4.60	16	2.82	1.61	4.59	31	2.32	1.58	3.30
Sjögren's syndrome	0				3	3.85	0.73	11.39	4	4.44	1.16	11.49	3	1.45	0.27	4.29	10	2.59	1.23	4.78
Systemic lupus erythematosus	0				2	0.75	0.07	2.75	0				6	1.45	0.52	3.18	8	0.82	0.35	1.62
Systemic sclerosis	0				4	2.47	0.64	6.38	3	1.79	0.34	5.29	4	1.40	0.37	3.63	11	1.72	0.86	3.10
Ulcerative colitis	3	1.90	0.36	5.62	9	0.91	0.41	1.74	13	1.36	0.72	2.32	18	1.08	0.64	1.71	43	1.14	0.82	1.54
Wegener's granulomatosis	0				2	0.83	0.08	3.05	2	0.77	0.07	2.82	2	0.51	0.05	1.88	6	0.65	0.23	1.42
All of above	88	4.73	3.79	5.83	242	1.95	1.71	2.21	270	2.15	1.90	2.43	414	1.83	1.66	2.02	1014	2.05	1.93	2.18
All of above minus asthma	46	3.14	2.30	4.19	168	1.73	1.48	2.01	169	1.72	1.47	2.00	272	1.52	1.35	1.72	655	1.69	1.56	1.82

Bold type: 95% CI does not include 1.00.

O = observed number of cases; SIR = standardized incidence ratio; CI = confidence interval.

Table III shows autoimmune disease risks according to age at hospitalization for obesity when follow-up was started 1 year after hospitalization. For the diseases with high risk, such as psoriasis, Hashimoto's disease/hypothyroidism, and asthma, the age at hospitalization appeared not to be critical to disease risk. For ankylosing spondylitis, Crohn's disease, type 1 diabetes, multiple sclerosis, and sarcoidosis, young age (<30 years) at hospitalization for obesity was associated with the risk. In contrast, for Graves's disease/hyperthyroidism, Sjögren's syndrome, and ulcerative colitis, being aged over 49 years at hospitalization for obesity was associated with the highest disease risks.

Table III. SIR for subsequent autoimmune and related disorders in obese patients by age at hospitalization; follow-up was started 1 year after last hospitalization for obesity.

	Age at diagnosis of obesity (years)
	<30				30–39				40–49				50–59				≥60
Autoimmune disorders	O	SIR	95% CI		O	SIR	95% CI		O	SIR	95% CI		O	SIR	95% CI		O	SIR	95% CI
Amyotrophic lateral sclerosis	1	2.22	0.00	12.74	1	1.09	0.00	6.23	2	0.97	0.09	3.55	2	0.67	0.06	2.47	0
Ankylosing spondylitis	7	3.70	1.47	7.67	2	1.27	0.12	4.66	4	3.01	0.78	7.78	1	1.30	0.00	7.44	1	4.55	0.00	26.06
Asthma	51	2.89	2.15	3.80	70	4.10	3.20	5.18	73	3.08	2.41	3.87	90	3.44	2.76	4.23	33	1.90	1.31	2.67
Behçet's disease	1	1.75	0.00	10.06	1	3.13	0.00	17.91	1	3.33	0.00	19.11	2	7.41	0.70	27.24	1	4.76	0.00	27.30
Chronic rheumatic heart disease	0				2	1.49	0.14	5.49	1	0.34	0.00	1.97	5	1.05	0.33	2.46	3	0.87	0.16	2.59
Crohn's disease	20	1.92	1.17	2.96	10	1.55	0.74	2.87	6	1.04	0.37	2.28	4	1.05	0.27	2.71	0
Diabetes mellitus type 1	9	2.65	1.20	5.06	0				0				0				0
Graves's, hyperthyroidism	17	1.10	0.64	1.77	16	1.06	0.60	1.73	21	1.52	0.94	2.33	19	1.71	1.03	2.68	4	0.58	0.15	1.50
Hashimoto's/hypothyroidism	4	2.60	0.68	6.72	6	2.84	1.02	6.23	12	4.05	2.08	7.10	9	2.77	1.26	5.28	12	4.29	2.20	7.51
Immune thrombocytopenic purpura	3	3.09	0.58	9.16	1	1.19	0.00	6.82	2	2.04	0.19	7.51	3	3.06	0.58	9.06	0
Multiple sclerosis	14	2.29	1.25	3.85	11	1.61	0.80	2.89	7	1.20	0.48	2.48	2	0.65	0.06	2.40	1	1.27	0.00	7.26
Pernicious anemia	1	4.76	0.00	27.30	2	6.45	0.61	23.73	0				6	4.35	1.56	9.53	1	0.51	0.00	2.91
Polymyalgia rheumatica	1	0.62	0.00	3.56	5	1.87	0.59	4.41	7	1.19	0.47	2.47	18	1.91	1.13	3.02	15	1.70	0.95	2.81
Primary biliary cirrhosis	1	4.35	0.00	24.92	3	6.67	1.26	19.73	2	2.82	0.27	10.36	0				1	3.57	0.00	20.47
Psoriasis	14	3.68	2.01	6.20	18	4.75	2.81	7.52	26	5.20	3.39	7.63	13	3.04	1.61	5.22	10	5.05	2.41	9.32
Rheumatoid arthritis	11	1.40	0.69	2.51	14	1.07	0.58	1.81	35	1.59	1.11	2.21	29	1.20	0.80	1.72	20	1.50	0.92	2.33
Sarcoidosis	12	3.73	1.92	6.53	6	2.17	0.78	4.76	8	2.61	1.11	5.16	4	1.50	0.39	3.87	0
Sjögren's syndrome	1	2.17	0.00	12.46	0				1	0.89	0.00	5.12	4	4.12	1.07	10.66	4	9.52	2.48	24.63
Systemic lupus erythematosus	3	1.31	0.25	3.88	0				1	0.44	0.00	2.55	2	1.14	0.11	4.18	2	2.56	0.24	9.43
Systemic sclerosis	3	2.75	0.52	8.15	3	2.48	0.47	7.34	1	0.63	0.00	3.61	2	1.35	0.13	4.97	2	2.63	0.25	9.68
Ulcerative colitis	11	0.86	0.43	1.54	10	1.18	0.56	2.18	6	0.84	0.30	1.84	11	2.08	1.03	3.74	2	0.82	0.08	3.00
Wegener's granulomatosis	0				0				1	0.66	0.00	3.77	2	0.66	0.06	2.41	3	0.92	0.17	2.73
All of above	185	1.98	1.71	2.29	181	2.03	1.75	2.35	217	1.96	1.71	2.24	228	2.03	1.77	2.31	115	1.63	1.35	1.96
All of above minus asthma	134	1.77	1.48	2.10	111	1.54	1.27	1.86	144	1.66	1.40	1.95	138	1.60	1.35	1.89	82	1.54	1.23	1.92

Bold type: 95% CI does not include 1.00.

O = observed number of cases; SIR = standardized incidence ratio; CI = confidence interval.

Since it may be an indication of disease severity, we lastly evaluated the possible associations between number of hospitalizations for obesity and the risk of autoimmune and related diseases (Table IV). A higher number of hospitalizations correlated with an increased SIR for all autoimmune diseases as a group. Among individual diseases, an increasing trend was evident for asthma (increase from 2.89 to 4.81) and psoriasis (increase from 3.96 to 6.22). However, most patients were only hospitalized once for obesity, which resulted in the numbers of cases of some of the rarer autoimmune diseases being very small in subjects hospitalized multiple times for obesity.

Table IV. SIR for subsequent autoimmune and related disorders in obese patients by number of hospitalizations; follow-up was started 1 year after last hospitalization for obesity.

	Number of hospitalizations for obesity
	1				2				≥3
Autoimmune disorders	O	SIR	95% CI		O	SIR	95% CI		O	SIR	95% CI
Amyotrophic lateral sclerosis	3	0.51	0.10	1.51	2	1.46	0.14	5.37	1	1.06	0.00	6.10
Ankylosing spondylitis	12	2.84	1.46	4.97	2	2.08	0.20	7.66	1	1.59	0.00	9.10
Asthma	215	2.89	2.51	3.30	50	3.00	2.22	3.95	52	4.81	3.59	6.31
Behçet's disease	5	3.94	1.24	9.26	0				1	6.25	0.00	35.83
Chronic rheumatic heart disease	5	0.51	0.16	1.20	3	1.41	0.27	4.17	3	2.17	0.41	6.44
Crohn's disease	27	1.32	0.87	1.93	9	2.01	0.91	3.84	4	1.37	0.36	3.54
Diabetes mellitus type 1	5	1.75	0.55	4.11	4	10.81	2.81	27.95	0
Graves's, hyperthyroidism	53	1.17	0.88	1.53	16	1.53	0.87	2.49	8	1.23	0.52	2.43
Hashimoto's/hypothyroidism	26	2.82	1.84	4.13	11	5.16	2.56	9.27	6	4.72	1.70	10.35
Immune thrombocytopenic purpura	4	1.27	0.33	3.28	5	6.85	2.16	16.11	0
Multiple sclerosis	30	1.82	1.23	2.61	3	0.79	0.15	2.35	2	0.81	0.08	2.99
Pernicious anemia	9	2.66	1.21	5.08	1	1.30	0.00	7.44	0
Polymyalgia rheumatica	31	1.51	1.02	2.14	13	2.71	1.44	4.65	2	0.65	0.06	2.40
Primary biliary cirrhosis	4	2.35	0.61	6.08	1	2.38	0.00	13.65	2	7.41	0.70	27.24
Psoriasis	54	3.96	2.97	5.17	14	4.50	2.45	7.57	13	6.22	3.30	10.67
Rheumatoid arthritis	79	1.36	1.08	1.70	18	1.34	0.79	2.11	12	1.33	0.68	2.32
Sarcoidosis	23	2.46	1.56	3.70	4	1.89	0.49	4.88	3	2.27	0.43	6.73
Sjögren's syndrome	6	2.25	0.81	4.92	3	4.69	0.88	13.88	1	2.22	0.00	12.74
Systemic lupus erythematosus	4	0.59	0.15	1.52	3	1.95	0.37	5.77	1	0.99	0.00	5.68
Systemic sclerosis	7	1.58	0.62	3.27	2	1.96	0.18	7.21	2	2.90	0.27	10.66
Ulcerative colitis	32	1.20	0.82	1.70	3	0.52	0.10	1.53	5	1.32	0.42	3.11
Wegener's granulomatosis	5	0.75	0.24	1.77	1	0.68	0.00	3.90	0
All of above	639	1.84	1.70	1.99	168	2.14	1.83	2.49	119	2.35	1.94	2.81
All of above minus asthma	424	1.56	1.41	1.71	118	1.91	1.58	2.29	67	1.68	1.30	2.13

Bold type: 95% CI does not include 1.00.

O = observed number of cases; SIR = standardized incidence ratio; CI = confidence interval.

Discussion

As a proof of principle, we showed that psoriasis exhibited the highest obesity-associated SIR of 4.54 and asthma the fourth highest SIR of 3.39. Obesity is a known risk factor and/or co-morbidity in both of these diseases (3,7). However, for most of the 14 other autoimmune diseases with significantly increased SIRs in individuals who had been hospitalized for obesity, limited or no data are available. Some of these diseases are rare and thus not very amenable to epidemiological analysis in studies with designs other than that employed in the present study, which made use of nation-wide hospitalization data. Thus, for example, Behçet's disease, primary biliary cirrhosis, and Sjögren's syndrome, conditions whose SIRs were >2.0 in subjects who had been hospitalized for obesity but which occurred in ten or fewer such subjects, would be difficult to study by other approaches.

The use of hospital discharge data to define obesity has the advantage that most diagnoses will be correct. Wolk et al. validated the obesity data in the Swedish Hospital Discharge Register (23). In a sample of 221 reviewed records, all patients were overweight, and 85% of men and 95% of women were obese. The median BMI values for men and women were 33.7 and 35.3 kg/m², respectively. These groups included some morbidly obese individuals, but the median BMI values were not extremely high. As most hospitalizations for obesity occurred in young and middle-aged adults, it is likely that the majority of patients actively sought medical help. Obesity was clearly the reason for seeking medical advice, as very few of the concomitant diagnoses, which included type 2 diabetes, sleep disorders, heart diseases, hypertension, hernia, and diseases of the liver and gall-bladder, were related to autoimmunity ((23) and our unpublished observations). Importantly, the SIRs for many autoimmune diseases remained elevated 10 or more years after hospitalization for obesity, implying that lead time and observation biases were unlikely explanations for the findings. The reference population included all persons not hospitalized for obesity, which may have caused the true effects to have been under-estimated.

The use of the Hospital Discharge Register to identify patients with autoimmune diseases offers access to a nation-wide patient pool. Since discharge diagnoses are often made by specialists during extended examinations in their clinics, diagnostic accuracy should be high. Rates of hospitalization for the 34 autoimmune and related diseases studied probably vary considerably depending on the seriousness of the condition and the alternative forms of treatment available, as we have discussed for some of them elsewhere (17,24–26). Another issue relating to the interpretation of the results is the difference in SIRs between different follow-up intervals (Table II). For most diseases, the highest SIRs were observed in the year following hospitalization for obesity, which is likely to be due to lead time bias. However, in as far as the diagnoses are correct, the lead time bias only shifts the diagnoses earlier. In order to produce conservative risk estimates in the analyses presented in Tables III and IV, we eliminated patients diagnosed with autoimmune and related diseases during the first year following hospitalization for obesity, aware this might lead to risks being under-estimated.

A concern with the present study design is that hospitalization may involve selections, whereby persons once hospitalized may be more likely to be hospitalized again for the same or other causes. Some 25% of obese patients were hospitalized more than once for obesity, and it is possible that some would have been hospitalized for side-effects of treatment for obesity or for related causes. ‘Brittle’ asthma, type 1 diabetes, and Addison's disease have been described in the literature among patients whose disease is unstable or unpredictable, thereby leading to frequent hospitalizations (27,28). One could try and guard against this type of bias by selecting as the control group persons hospitalized for a non-obesity-related condition. However, it may be difficult, a priori, to nominate a disease that results in large numbers of hospitalizations but that is in no way related to obesity. Instead, we believe that internal comparisons among the diseases studied indicate that ‘brittle’ disease is probably not an important cause of the observed SIRs for certain autoimmune diseases being high. For example, a common disease such as ulcerative colitis showed no overall risk in Table II, and the SIRs for rheumatoid arthritis and Graves's disease/hyperthyroidism, the most common diseases after asthma, were only increased to about 1.3.

As epidemiological support for the emerging mechanistic links between obesity and immune-related diseases, we have shown herein that the SIRs for 16 of the selected 34 autoimmune and related diseases studied were increased in subjects hospitalized for obesity. In addition to the known associations with psoriasis and asthma, high risks were observed for Behçet's disease and Hashimoto's disease/ hypothyroidism. Small but significant increases were also noted for the common autoimmune disease Graves's disease/hyperthyroidism and rheumatoid arthritis. Such small effects may, however, be the result of a higher likelihood of further hospitalizations among those who have been hospitalized at least once before.

The present study was based on a population of obese individuals hospitalized at a relatively young age, and further studies are needed to describe to what extent autoimmune and related inflammatory conditions are related to obesity at large.

Declaration of interest: Supported by the Swedish Council for Working Life and Social Research, the Swedish Research Council, the Swedish Cancer Society, and Deutsche Krebshilfe. The authors declare no conflicts of interest.