In our nationwide cohort study, we found that individuals with type 1 diabetes had more than a twofold higher incidence of eating disorders than diabetes-free control individuals. Among the 250 eating disorder cases detected, over 60% received newly prescribed psychotropic medications, with no differences observed between those with vs without type 1 diabetes. However, those with type 1 diabetes received less outpatient hospital treatment for their eating disorders than those without diabetes.

To our knowledge, this is the first nationwide study to assess differences in hospital treatment for eating disorders between people with and without type 1 diabetes. Our findings indicating that eating disorders in patients with type 1 diabetes are undertreated relative to those without diabetes are in line with previous small, single-centred studies, which have also reported less intensive treatment for those eating disorder patients with type 1 diabetes [25, 26]. These findings are concerning, as eating disorders are associated with increased morbidity and mortality rates among type 1 diabetes patients [12,13,14,15,16,17,18]. Thus, eating disorders among people with type 1 diabetes should be treated more intensively, not less intensively, than eating disorders in general. Treatment outcomes are also poorer among those eating disorder patients who also have type 1 diabetes when compared with those who do not [24,25,26], which further highlights the need for rigorous treatment. Future studies should study associations between eating disorder treatment and diabetes-related outcomes, such as glycaemic control and complications.

Reasons for the undertreatment of eating disorders among those with type 1 diabetes have been assessed previously. Adherence issues are common [25], and dropout rates are high [26]. High dropout rates have been associated with the patients’ lower motivation and the perceived intensity of the eating disorder. It is possible that the eating disorders among people with type 1 diabetes could be less severe than the eating disorders of those with no diabetes. Although the difference was not statistically significant, in our study a slightly higher proportion of eating disorders seemed to be diagnosed as ‘other eating disorder’ among type 1 diabetes patients than among diabetes-free control patients. However, the diagnosis of ‘other eating disorder’ is not necessarily less severe than anorexia nervosa or bulimia nervosa among those with type 1 diabetes; it could also reflect atypical features of eating disorders, such as insulin omission. We could not directly assess the severity of eating disorders, but earlier evidence indicates that the psychiatric and medical characteristics of eating disorders do not appear to differ between those with or without type 1 diabetes [26]. It is also possible that some eating disorder-related outpatient visits of those with type 1 diabetes could have been recorded with only a diabetes diagnosis. However, as hospital treatment for eating disorders is given mostly in highly specialised psychiatric eating disorder units in Finland, it is unlikely that the clinicians would not have recorded any eating disorder diagnosis for these hospital visits. Other possible reasons for undertreatment of eating disorders among type 1 diabetes patients include the lack of evidence-based treatments available for this group [31], the complexity of treating co-occurring eating disorders and type 1 diabetes and lack of healthcare personnel with adequate knowledge about both eating disorders and type 1 diabetes. Qualitative research has shed light on the care experiences of patients with co-occurring eating disorders and type 1 diabetes and indicated a need for multidisciplinary collaboration and professionals who understand both disorders [41].

We are not aware of previous studies assessing the use of psychotropic medications among eating disorder patients with type 1 diabetes, let alone comparisons with eating disorder patients without diabetes. Our findings of a substantial use of these medications, with no differences between type 1 diabetes and diabetes-free eating disorder patients, is noteworthy. Despite eating disorder patients with type 1 diabetes responding less well to eating disorder treatment [25, 26], they appear not to be treated differently with psychotropic medications compared with diabetes-free patients, even though type 1 diabetes comes with an elevated risk for anxiety and depression [3, 4]. We suspect that, because of the timeframe for consultation, there are limits to how psychiatric problems can be addressed when there are additional complicating factors such as co-occurring type 1 diabetes. In addition, clinicians might be in general more cautious in prescribing psychotropic medications to those with type 1 diabetes due to their adverse metabolic effects. Further, despite these effects, which might occur even independent of weight gain [33], 15% of type 1 diabetes patients with an eating disorder had used antipsychotics in the present study, which might contribute to worse metabolic balance and increased morbidity and mortality rates [42]. Future studies should assess the outcomes and safety of antipsychotic medication use among eating disorder patients with type 1 diabetes.

Our results of over twofold higher incidence of eating disorders among people with type 1 diabetes compared to diabetes-free individuals complement previous similar findings [3, 4]. Dybdal et al used hospital registers to assess eating disorder incidence with somewhat similar findings to ours (aIRR 2.35 [95% CI 1.80, 3.09] in our study vs HR 2.02 [95% CI 1.54, 2.64] in girls and 3.73 [95% CI 1.71, 8.11] in boys in that of Dybdal et al) [3]. Cooper et al used registers from both primary and secondary healthcare. Although they found a high HR for eating disorders (5.06; 95% CI 2.3, 10.9) in those with type 1 diabetes compared with those without diabetes, the confidence intervals were compatible with our findings as our upper limit of 3.09 is higher than their lower limit of 2.3 [4]. Due to our register-based setting focused solely on specialised healthcare, underdiagnosis of eating disorders is evident. Because records from primary healthcare were not included, it is likely that only the most severe eating disorders were detected in this study. As binge–purge spectrum symptoms are the most reported eating disorder symptoms in patients with type 1 diabetes [11], some eating disorder patients may not have been detected due to limitations of ICD-10 in detecting binge eating disorder. However, in Finland there has been a local agreement to diagnose binge eating disorder using ICD-10 codes F50.3, F50.8 or F50.9, which likely alleviates this problem to some degree. Our approach also missed any eating disorders that had not been detected in the healthcare system [43]. We are not aware of interview-based studies that would have estimated the incidence of eating disorders specifically among individuals with type 1 diabetes, but in general interview-based studies have found much higher estimates for the incidence of eating disorders than our register-based study [1, 44]. Nonetheless, despite underestimation of absolute incidence, the relative 2.35-fold higher incidence of eating disorders among those with type 1 diabetes compared with diabetes-free controls found in our study is supported by a large interview-based study, which found a 2.4-fold higher odds for eating disorders among people with type 1 diabetes compared with healthy control individuals [5].

We found that the higher incidence of eating disorders among individuals with type 1 diabetes vs diabetes-free control individuals was not explained by age, sex, socioeconomic status or place of residence. Previous studies have adjusted their models only for age and sex [3, 4]. Smaller studies have found that physical, psychological and family factors might increase the risk for disturbed eating among type 1 diabetes patients, yet they lacked comparison to diabetes-free participants [6,7,8,9,10]. To confirm these findings, replication in larger settings with comparison to diabetes-free control participants is needed. Furthermore, it has been unclear whether sex interacts with diabetes on the risk of incident eating disorders [3, 4]. Our finding of no interaction between type 1 diabetes and sex on the incidence of eating disorders indicates that type 1 diabetes comes with an elevated risk for eating disorders regardless of sex.

Our study comes with major strengths. Our nationwide, longitudinal setting with a diabetes-free control group and a large sample enabled us to expand earlier findings. Furthermore, our method of ascertaining psychotropic medication use is robust because we utilised registers that included only prescribed medications that had been purchased, increasing the probability that they had indeed been used. Finally, we had a sufficient number of male participants in our sample to assess sex differences.

Our study also has some weaknesses. First, we lacked information of clinically confirmed type 1 diabetes diagnoses. To identify type 1 diabetes patients, we used a literature-based method choosing insulin users who did not use any other glucose-lowering medications and were aged under 30 years [37, 38]. Based on earlier studies, this approach selects people with type 1 diabetes while excluding virtually all those with type 2 diabetes [37, 38], although there might be some severely ill type 2 diabetes patients whose poor renal function limits their diabetes medication to insulin only. Further, the few type 1 diabetes patients who were excluded by our methodology because their diabetes started after age 30 are not likely to have a major impact on our findings, because most eating disorders start in adolescence or early adulthood [45]. Second, we included both primary and secondary diagnoses of eating disorders in the analyses. This approach maximised the sensitivity of eating disorder ascertainment but might have biased our findings on eating disorder treatment. In some cases, eating disorder diagnoses might have been recorded as secondary diagnoses also in inpatient stays and outpatient visits where eating disorders were not the target of the treatment. However, our findings of eating disorder undertreatment are in line with previous studies, supporting the validity of our methods [25, 26].

In conclusion, we found over twofold higher incidence of eating disorders in individuals with type 1 diabetes compared with diabetes-free control individuals. The use of psychotropic medications among eating disorder patients was substantial and did not differ among those with vs without type 1 diabetes. However, those with type 1 diabetes received less outpatient hospital treatment for their eating disorders than those without diabetes. Our findings emphasise the need to reduce this discrepancy in treatment intensity in order to alleviate the burden of this severe dual condition of co-occurring eating disorder and type 1 diabetes.