Time trends in co-occurring substance use and psychiatric illness (dual diagnosis) from 2000 to 2017 – a nationwide study of Danish register data

Abstract Objective This article aims to describe the time trend in number of dual diagnosis patients treated in the psychiatric system in Denmark from 2000 to 2017. Method We calculated the share of patients with dual diagnosis, number of dual diagnosis contacts, number of unique individuals with dual diagnosis as well as number of new patients with dual diagnosis among patients in psychiatric treatment, i.e. among inpatients, outpatients and patients in emergency departments. In order to calculate this, we merged data from the National Patient Register (NPR), the National Registry of Alcohol Treatment, the National treatment registry for substance use, the National Prescription Registry and the Danish National Health Service register in the period from 2000 to 2017. Results We found an overall increase in patients with dual diagnosis in psychiatric treatment in Denmark from 2000 to 2017. We further detected an increase in the age and sex-standardized number of patients with dual diagnosis in treatment over time, however most markedly for outpatients. Further, inclusion of data from other sources of data than the NPR dramatically increased the number of patients that could be identified as dual diagnosis patients. Using these data, almost half of all male inpatients could be identified as dual diagnosis while the share was more than 40% for patients with schizophrenia, schizotypal and delusional disorders (F2) and patients with personality disorders (F6). Conclusions The increase of individual diagnosis patients necessitates action at different levels. This includes improvement of preventive measures as well as improvement of treatment for this underserved group.


Introduction
The reports from the Global Burden of Disease Study from 2017 [1] and 2022 [2] show an increasing incidence of individuals with mental illness in all diagnostic groups combined with a reduction in the age-standardized rate for the total of all psychiatric illnesses although this total hides increases or decreases in specific groups globally. The document from 2017 also reports increasing numbers of individuals with substance use disorder, as well as an increase in the total age-standardized rate. For substance use disorder, the rise is primarily driven by an increase in drug use disorder, specifically opioid-related, and is countered by a decrease in alcohol use disorder [1]. Locally in Denmark, the National Institute of Public Health [3] and the National Board of Health [4] report similar trends on the increase in mental health problems. Further, a report on the use of illegal substances in Denmark [5] shows increased use of substances in the general population from 2000 to 2017 followed by a slight decline in 2021. For individuals below 25, an overall decrease in using one or more illegal substances was seen from 2000 to 2017 and with a marked additional decline in 2021. Cannabis is still commonly used among people below 25 years with 16% having used this within the last month in 2021. The cooccurrence of substance use disorder and psychiatric illness often called dual diagnosis is well established [6]. Toftdahl et al. reported that 30.4% of all patients with a psychiatric diagnosis in Denmark also receive a substance-related diagnosis during their lifetime, with variations from 11% for an obsessive-compulsive disorder to 46% for personality disorder [7]. Further, Petersen et al. showed that patients diagnosed with schizophrenia had a more than 3-fold increased risk of subsequently being diagnosed with a substance use disorder [8]. Register-based studies from the Nordic countries report a prevalence of dual diagnosis just below one percent (0.6-1) for moderate or severe mental illness with comorbid substance use disorder [9,10]. Although these register-based estimates may not be directly comparable with data from the large American National Survey on Alcohol and Related Conditions (NESARC) due to differences in groups studied and methods used, i.e. population vs. patient, the NESARC reveals strong and persistent associations between psychiatric disorder and alcohol or substance use disorders with 16.4% of the persons with at least one psychiatric disorder also having an alcohol use disorder and 6.5% a concurrent drug use disorder [11,12]. Despite this evident pattern in combination of diseases, only very few reports exist on the development over time concerning the development in the number of patients with dual diagnosis in treatment. Hunt et al. published a systematic review and meta-analysis of the prevalence of comorbid substance use in schizophrenia from 1990 to 2017 and found an increase in the prevalence of substance use disorders of illegal substances but no changes in the prevalence of all substance use disorders over time, however, this was based on data from studies of a combination of community and clinical samples [13]. Focusing on a single population over time, Hjorthoj et al. report an increase in the incidence in combined schizophrenia and cannabis use disorder from 3/100,000 inhabitants in 1995 to 6/100,000 inhabitants in 2015 in Denmark [14]. Tracking the development in numbers of dual-diagnosis patients over time is further complicated by the persistent underdiagnosing and/or lack of reporting of substance use disorder in psychiatric patients. In 2000, Hansen et al. [15] published a study showing that only 50% of dual-diagnosis patients in two psychiatric wards were identified as such and in 2014. Bahorik et al. [16] reported that more than 50% of schizophrenia patients who tested positive for substance use did not disclose this when asked.
Especially patients with dual diagnosis experience that the treatment system is often fragmented, not existing or not tailored to their needs [17][18][19]. This is costly at both the individual and the societal level with dual-diagnosis patients incurring both higher mortality [20], morbidity [21,22] and high costs toward social services and health care compared to both patients without substance use and the general population [9]. The high cost at all levels, combined with the general reporting that health care systems irrespective of how they are funded (universal coverage, insurance-based) do not support the full recovery of this groupwarrants that the number of dual diagnosis patients in treatment should be followed in detail to guarantee that this information is available for the health care planners and funding bodies as well as preventive experts. Further, the high level of underreporting necessitates that all available information on substances and psychiatric history should be included when identifying patients with dual diagnosis. Thus, the aim of the study is to describe the development in number of dualdiagnosis patients treated in the psychiatric system in Denmark from 2000 to 2017 using all available data from the highly comprehensive Danish registers, i.e. going beyond what is available from the individual treatment contact.

Setting
Denmark is a northern European country with a population size of a little more than 5.8 million. The Danish GNP was 59,646 US$ per capita and the average health care expense was 5568 US dollars/capita ultimo 2019 according to Statistics Denmark. The Danish national health care system is universal, and primarily tax-funded, however, some with some co-payment for specific services, such as dental care, physiotherapy and some psychotherapy [23]. All Danish citizens are given a personal identification number at birth which makes it possible to link information from the extensive Danish registers.

Population and identification of dual diagnosis patients
The patients included in this study were all treated in the public Danish health care system within the specialties of adult or child psychiatry in the period from 2000 to 2017 and were above the age of 14. This age group was chosen as very few below the age of 15 had SUD diagnosis, thus leaving too few patients in these groups.
In order to identify cases with dual diagnosis, we included data from a multitude of sources in a selected sequence. Initially, we based the identification of the dual diagnosis patients on diagnostic data retrieved from the individual treatment contact as reported in the National Patient Register (NPR) where diagnosis and treatment data from all psychiatric hospitals including emergency, inpatient and outpatient care in Denmark is registered [24,25]. Thus, a person was identified with dual diagnosis if the person had a psychiatric diagnosis and a diagnosis of harmful use (F1x.1) or dependence (F1x.2) at the same contact. X denotes the specific substance in Chapter F1 (Mental and behavioral disorders due to psychoactive substance use) in the ICD10. These identified cases will be denoted dual diagnosis (NPR) moving forward.
Second, we linked the NPR information using the personal identification number with additional information on alcohol use or substance use disorder from four additional data sources: I) a previous diagnosis of harmful use F1x.1 or dependence F1x.2 from psychiatric contacts within the two years before the specific contact, II) attending alcohol or substance use disorder treatment in the municipalities within the two years before the specific contact identified in either the National treatment registry for substance use or the National treatment registry for alcohol treatment, III) receiving medication indicating alcohol or substance use disorder within the two years before the specific contact as registered in the National Prescription Registry [26]. Please refer to the appendix for information on specific ATC codes and indication text used, IIII) diagnosis from somatic hospitals indicating alcohol or substance use disorder within the two years before the specific contact.
Third, we extended the dual diagnosis definition to include patients with only a diagnosis of harmful use F1x.1 or dependence F1x.2 at the specific contact if they had indicators of previous psychiatric illness by adding information from a combination of two sources: I) previous diagnosis of psychiatric illness from other psychiatric contacts within the two years before the specific contact, II) contact with a privately practicing psychiatrist or community psychiatry identified in the National Health Insurance Service Registry. Please refer to Appendix 1 (Supplementary file) for more information on all included data and data sources. The patients identified by linking all data will be denoted dual diagnosis (all data) moving forward.

Treatment mode
Admission: All contacts defined as an admission in the NPR. To account for patients changing between wards, the admissions were combined when the dates overlapped, thus the individual admission was from the time the patient was first admitted as an inpatient to the time the patient was discharged from the psychiatric hospital. This also included transfers between different psychiatric hospitals.
Out-patient care: All patients starting out-patient psychiatric care as defined in the NPR.
Emergency contact: All patients in contact with psychiatry emergency care defined in the NPR (psychiatric emergency contact) until 2013. From 2013 onwards, the NPR was not subdivided into out-patient care and emergency care; however, we used the name tags to subdivide. If the name indicated that the ward the patient was in contact with was emergency, then the patient was grouped in the emergency contact group.

Psychiatric disorder
For admissions, all of the diagnoses given during admission were taken into consideration. For outpatients and emergency care, the diagnosis given at the specific contact was used. We used the hierarchical nature of the ICD10 F chapter to assign one main psychiatric disorder to each individual, however excluding substance use diagnosis. If a patient had more than one diagnosis registered at a specific contact, then the hierarchical approach stipulates that the diagnosis with the highest rank was given priority, e.g. a diagnosis in the chapter on Schizophrenia, schizotypal and delusional disorders (F2) trumps diagnosis in the chapter on Neurotic, stress-related and somatoform disorders (F4). Thus, we assigned each individual one primary psychiatric disorder within the ten main diagnostic groups F0-F9 (except F64 and F65.1 which were removed as an official diagnosis in Denmark in the period of interest) for each contactif one was present.

Analysis
We calculated the absolute number of dual diagnosis contacts per year, percentage of patients with dual diagnosis out of all contacts per year, number of unique individuals with dual diagnosis per year and number of new dual diagnosis cases with no previous psychiatric treatment within the five years before the contact in question for admissions, outpatient care and emergency contacts in the period from 2000 to 2017.
We further standardized the rates of patients in treatment according to gender and age distribution in the complete Danish population in the year 2008 using direct standardization.
In the main article, we show the number of dual-diagnosis patients and standardized rates for both the patients identified as dual diagnosis (NPR) and dual diagnosis (all data), however, in the appendix additional steps of inclusion are shown. The data used for the figures in the main article are highlighted in the appendix.
Finally, we computed numbers in treatment by age, sex and main diagnostic group by treatment mode. In the main article, only results for dual diagnosis (all data) on admissions are shown (all three treatment modes are shown in the appendix).
All analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).

Ethics
The approval for analyzing the data was given by the Regional data protection agency permit number RHP-2017-019. Figure 1(a-d) shows the share of patients with dual diagnosis, absolute count of dual diagnosis patients, number of unique individual patients with dual diagnosis and number of new patients with dual diagnosis among admitted, outand emergency patients in Danish psychiatry for both dual diagnosis (NPR) and dual diagnosis (all data). First and foremost, it is noticeable that the number and share of dualdiagnosis patients increase markedly irrespective of treatment mode when including additional data (Figure 1(a-d)). For example (Figure 1(a)), when only focusing on data from the specific admission (dual diagnosis NPR), patients with dual diagnosis constitute 15% in 2000 with a slight increase over time. However, when including more data (dual diagnosis all data), the numbers increase so that more than 30% of the patients in 2000 can be categorized as dual diagnosis with an increase to more than 35% in 2017. For the emergency and out-patient contacts, a decrease in the share can be seen from 2000 to 2017 when including all data; however, we still see much higher shares compared to focusing on dual diagnosis NPR only. Further, Figure 1(b,c) shows that both the absolute count of contacts as well as the number of unique patients with dual diagnosis has increased across all three treatment modes. Figure 1(d) shows that the number of new patients presenting with dual diagnosis increases over time especially in the out-patient setting, but with some fluctuation in the other settings. Figure 2 shows the development in standardized treatment rates of dual diagnosis patients for in-patient, outpatient or emergency care for both dual diagnosis NPR and dual diagnosis all data. Irrespective of treatment mode, the number of patients per 10,000 persons increases during the period, but most noticeably for out-patient care where the number of dual diagnosis patients in treatment per 10,000 persons increases from approx. 12 to almost 25/ 10,000 persons. For inpatient care, we only detect a slight increase from 12/10,000 to 14.5/10,000, however peaking in 2013 with almost 17/10,000 persons. Comparably, but not as distinct, the number of dual-diagnosis patients treated in emergency care increased from 10.5 to 11.5 per 10,000 persons during the period with a similar peak of 13.6/10,000 in 2014.

Results
When focusing on the development in dual diagnosis across sex, age and primary diagnosis groups (Figure 3(a-c)), the main findings are that the share of dual diagnosis patients among female patients increased from around 41.7% in 2000 to 47.4% in 2017, while the share of dual diagnosis patients among female admitted patients remained stable with a little  more than 20% being categorized as dual diagnosis. Similarly, the share of dual-diagnosis patients increased among the oldest and the youngest age group but remained stable in the other age groups. More markedly we see an increase in the share of dual diagnosis patients among patients with a primary F2 (Schizophrenia, schizotypal and delusional disorders) with an increase from 34.5% in 2000 to 41.7% in 2017) and especially F9 (Behavioral and emotional disorders with onset usually occurring in childhood and adolescence) diagnosis increasing from 23.5% in 2000 to 49.1% in 2017. However, we also see an increase among patients with a primary F0 (Organic, including symptomatic and mental disorders) and the combined group (F5, F7 or F8), but a decreasing F6 (Disorders of adult personality and behavior) or stable F3 (Mood (affective) disorders) and F4 (Neurotic, stress-related and somatoform disorders) development. Please refer to the appendix for details on outpatients and emergency care.

Discussion
We found a substantial under-detection of dual diagnosis patients when only including diagnosis data from the individual admission as recorded in the NPR compared to when all available data was included. Further, we saw an overall increase in patients with dual diagnosis in psychiatric health care treatment in Denmark from 2000 to 2017 both in share, absolute count and number of unique individuals. The most marked increase was detected among patients in outpatient and in-patient treatment. For emergency contacts, the share of dual-diagnosis patients decreased, but not the absolute count of contacts and unique individuals. The share of patients increased especially among males, the youngest and oldest age groups and patients with Schizophrenia, schizotypal and delusional disorders (F2) and Behavioral and emotional disorders with onset usually occurring in childhood and adolescence (F9). The results furthermore show that in 2017 almost half of all males admitted could be identified as dual-diagnosis patients. Likewise for patients with Behavioral and emotional disorders with onset usually occurring in childhood and adolescence (F9), while the share was more than 40% for patients with Schizophrenia, schizotypal and delusional disorders (F2) and patients with personality disorders (F6). We also reported an increase in number of patients with dual diagnosis in treatment over time (age and sexstandardized), however most markedly for outpatients and with a peak in 2013-2014 for admissions and emergency contacts.
Our findings are in line with the findings of dual diagnosis being very prevalent among patients with mental disorders [7,11,12]. As noted in the introduction, very few studies focus on time trends in number of dual diagnosis in treatment. Comparing our results with Hunt et al. [13] who report on no changes for prevalence of overall substance use disorders combined with an increase in illegal substance disorders in patients with a schizophrenia spectrum disorder, we see a slight increase. This difference may however be attributed to differences in design and data, where we use nationwide patient-level data and Hunt et al. numbers are based on a meta-analysis of studies from different samples. Our findings are however in line with the findings reported by Hjorthoj et al. [14] who see an increase in incidence of cannabis use disorders within schizophrenia patients. Our study is not completely comparable to the two others as we do not subdivide on specific substances and do not estimate incidence or prevalence, but instead focuses on cases in psychiatric treatment according to treatment mode (i.e. in-patient, outpatient and emergency). Further, the increase in number of patients needing treatment naturally highlights the question: 'Why do we see this development?' In this article, we will propose a few explanations, but other explanations may apply. First, an increase in use and access to substances in the general society [27] spills over to the psychiatric patients thus increasing the number of patients with dual diagnosis. This effect could impact the number of dual-diagnosis patients if more patients with psychiatric disorders use substances, but the use of substances may also cause and/or increase symptoms of psychiatric illness [28]. This argument is however is not completely in line with reported tendencies on drug use in Denmark where the use of illegal substances has declined in the general youth population [5].
Concerningly, this may not be the case among mentally vulnerable or marginalized youth, thus explaining why an increase in this group was seen in this study. In addition, we know that there has been an increase in street-level drug potency [29,30] as well as a shift in specific substances used among youth with psychiatric illness [31] and in the general population due to significant price reduction especially for cocaine. Second, these patients might always have been present in the psychiatric system but were not identified with dual diagnosis due to under registration of either the psychiatric or substance use disorder. We find a substantial underreporting when combining data from different data sources in all the included years. Thus, the numerous projects [32,33] aiming at increasing the use of diagnostic tools to identify substance use among psychiatric patients do not seem to have impacted this substantially. However, it is important to notice that our study is not a diagnostic validation study. Lastly, it has previously been reported that patients with severe mental illness are offered shorter length of stay due to the reduced number of in-patient beds compared to earlier years [34] which in turn leads to an increase in revolvingdoor patients, i.e. patients being discharged too early thus having to be readmitted or being discharged for out-patient follow-up which is in line with the increase in patients with dual diagnosis in the out-patient follow-up we have reported. We can however also see that the number of unique individuals in contact with the psychiatric system increases. Thus, the increase in number of dual-diagnosis patients cannot be fully explained by revolving-door patients or dual-diagnosis patients being referred to ambulatory care.

Strengths and limitations
The main strength of this study is the comprehensiveness of the data used which includes all public psychiatric hospitals nationwide thus including admissions, outpatients and emergency care. We further linked the data from the NPR with information sources, for example the National Prescription Registry, the two registries on substance use treatment and the Danish National Health Service register, allowing us to show the extensive underreporting of dual diagnosis if one only relies on the NPR. However, even though we have this very extensive linkage of data we could still have an underreporting of dual diagnosis. We initially defined dual diagnosis as the combination of a psychiatric illness and harmful use or dependency syndrome. However, patient contacts with other substance use diagnoses (e.g. F1x.0 Acute intoxication) or no substance-related diagnosis may also be using substances in a way that could be characterized as harmful or be dependent. The linkage of data will ensure that we find some of these patients, but this requires that the patients have received some form of treatment or at least a diagnosis of substance problems at another treatment encounter. We know that patients hide their substance use problems [16,35] and wait a long time before seeking treatment [35] and that substance diagnoses are underreported in general [15,16]. On the other hand, the linkage may also have the consequence that some patients will be categorized as dual diagnosis who are in fact not because we employ a maximum of two years between the presence of either indicator of substance use or psychiatric illness, hence both may not be present at the same time. We however believe that this is a minor problem as we have used both the least inclusive diagnostic definition of substance use (i.e. only harmful use or dependency syndrome) and information on attending substance use treatment which will only be relevant for patients with severe and persisting substance use problems.

Conclusion
The increase in patients with dual diagnosis in treatment should be of great concern as health care already struggles with supplying adequate treatment for this group, and action at different levels is necessary. At the societal level, there is a need for better understanding the drivers of both mental illness and substance use to implement more effective prevention strategies in order to stop the influx of patients as well as to prevent the harm from SUD for people with mental disorders. At the treatment level, there is a great need to ensure more intensive and coherent treatment of patients with dual diagnosis to guarantee more integrative and adequate treatment and prevent readmission for this patient group. The finding that dual-diagnosis patients constitute a large and an increasing proportion of patients also highlights the fact that staff dual-diagnosis competencies should be available at all treatment levels.

Disclosure statement
No potential conflict of interest was reported by the author(s).