Resource utilization in mental illness – evaluation of an instrument for measuring direct costs of treatment for patients with severe mental illness (SMI)

Abstract

Purpose

Serious mental illnesses (SMIs) exert a considerable financial burden on health-care systems. In this study, the resource utilization in mental illness (RUMI) tool was developed and employed to evaluate resource utilization in patients with SMI.

Materials and methods

Data from 107 patients with SMI treated in four psychiatric outpatient clinics in Sweden were collected. The relationships between costs for physical and psychiatric care, social services, and the justice system, to self-reported health and quality of life, educational level, Global Assessment of Functioning (GAF), the Clinical Global Impressions scale score (CGI), and body mass index (BMI) were studied.

Results

Sixteen out of 107 patients accounted for 74% of the total costs. The mean and median cost for 6 months included in the survey for social services, family and social welfare and healthcare, psychiatric and physical treatment interventions, mean 8349 EUR, median 2599 EUR per patient (currency value for 2021). Education and psychosocial function (GAF) were both negatively correlated with costs for the social services (education, r=–0.207, p < 0.014; GAF, r=–0.258, p < 0.001). CGI was correlated with costs for physical and psychiatric healthcare (r = 0.161, p < 0.027), social services support (r = 0.245, p = 0.002) and total cost (r = 0.198, p = 0.007). BMI correlated with costs for psychiatric and physical health settings interventions (r = 0.155, p < 0.019) and for somatic and psychiatric medicines (r = 0.154, p < 0.019).

Conclusion

The RUMI scale was acceptable and enabled estimation of resource utilization in a comparable manner across different care settings. Such comparable data have potential to provide a basis for budgeting and resource allocation.

Keywords:

• Health economics
• psychiatric care
• mental health
• psychosis

Introduction

Serious mental illnesses (SMIs), such as schizophrenia, bipolar disorder and schizoaffective disorder, can not only pose a major health challenge for patients and their families but also convey a significant cost [1,2]. These costs differ substantially between countries. Factors associated with higher reported individual costs include patient demographics, the severity of the illness and any co-morbidities and the structure of the healthcare system in question [3]. The costs are also influenced by the definitions, assumptions and methods used to calculate the costs of productivity losses [3]. Health economic studies can provide useful information on the financial support needed for care. Instruments to measure cost consistently at both group and individual levels are necessary to inform comparative economic evaluations of health care interventions, including cost-effectiveness and cost-utility studies.

In Sweden and the Nordic countries, assessments of costs are complicated by the fact that treatment and support is provided through a number of different funding streams. Medical treatments for physical and mental illnesses are funded by the health system, through physical and psychiatric clinics. Patients with functional disability can separately apply and receive help for the impact of that disability from the social services administration in the area where they live, e.g. financial compensation, support at home, support outside the home and assisted living. Additionally, in Sweden, the Social Insurance Agency is responsible for certain rehabilitation services and for financial compensation.

The total annual economic cost in Sweden for bipolar disorder, depression, anxiety, schizophrenia and generalized anxiety disorder amounts to 75 billion SEK (7.5 billion €). Of this, the greater proportion, 57 billion SEK (5.7 billion €), derives from indirect costs [4]. The average cost for a person diagnosed with schizophrenia in 2008 was SEK 509,000 (49,535 €). The indirect costs of absence from work account for as much as 75% of the total societal cost. All costs were valid for one year (2008) and were stated in the 2009 monetary value, SEK [4].

In a 2018 systematic review of indirect costs of schizophrenia in Europe, average indirect costs were 44%, rising to 58% when the cost of productivity loss by patients and caregivers was included. Factors associated with higher indirect costs included patient characteristics such as age, sex and severity of symptoms. The indirect costs varied greatly between studies, from 119 to 62,034 Euros [5]. Over time, costs vary for individuals, subgroups and the patient group as a whole, with hospital stay a major driver. Patients with schizophrenia who experience readmission have significantly higher mean costs of care and greater medical costs than those who are not readmitted and matched controls [6].

Adequate treatment at the right time has the potential to reduce both inpatient length of stay and associated costs [7,8]. Observational studies of whole populations have shown that the risks of rehospitalization, premature mortality and costs are related to medication type and mode of administration [9,10]. Improvements in level of functioning, ability to work and health have substantial positive economic effects from both an individual and societal perspective [4]. Despite this, at present there are no standardized instruments in Sweden to document direct costs across the various health and care providers for patients with SMI.

The instrument resource utilization in mental illness (RUMI) was adapted from the ‘Resource Utilization in Dementia’ (RUD) tool [11], and has a structure similar to the ‘The Client Service Receipt Inventory’ (CSRI) [12]. The RUD was developed for cost-effectiveness studies in dementia care [13]. RUMI was designed to be a brief, time efficient questionnaire to allow providers of social service and healthcare to capture the direct costs associated with SMI over a period of 6 months. The instrument takes the national social structures in Sweden and the Nordic countries into account, where treatments, psychosocial interventions and financial compensation are handled by different authorities, e.g. social services, health care, Swedish social insurance agency and the justice system, a complexity less well served by other tools such as the European version CSSRI-EU [14]. This information is not always documented in the patient's psychiatric records, so RUMI is designed to be completed in conjunction with staff familiar with the patient's history and having access to the necessary sources of information.

Direct and indirect costs interact and are affected, among other things, by severity of symptoms [5]. Therefore, in this project, we have chosen to focus on the costs of the direct interventions.

Study aims

• To describe and calculate the direct costs of patients with SMI using the RUMI instrument.

• To evaluate the relationship between clinical features and resource utilization in patients with a diagnosis of SMI.

Material and methods

Study design and setting

The utility of RUMI was evaluated by applying it to resource utilization data collected as part of the MINT study, ‘Outcome of a psychosocial health promotion intervention aimed at improving physical health and reducing alcohol use in patients with schizophrenia and psychotic disorders’ [15]. MINT was a naturalistic study with patients enrolled sequentially at specialized departments with catchment area responsibility for outpatients with SMI in four Swedish cities (Gothenburg, Kristianstad, Hässleholm, and Malmö). The cost of MINT was covered by the Swedish Universal Health Care system. MINT was designed to assess the effectiveness of a psychosocial health promotion intervention in improving cardiometabolic risk, quality of life and severity of psychotic illness in patients with psychotic disorders compared to usual treatment. MINT received ethical approval as such from the Stockholm Regional Ethics Review Board (dnr. 2011/849-32). All included patients gave informed consent for participation.

Participants

Patients were eligible to participate if they were 18 years or older with a severe mental illness (SMI) (ICD 10 diagnoses F20-29: Schizophrenia, schizotypal and delusional disorders; F31.2: bipolar affective disorder, current episode manic with psychotic symptoms; and F31.5: bipolar affective disorder, current episode severe depression with psychotic symptoms). The participants spoke Swedish. Exclusion criteria were (a) a primary diagnosis of learning disability; (b) a co-existing physical health problem that would in the opinion of a medical doctor independently impact cardiometabolic measures and/or substance use habits; (c) current pregnancy or ≤6 months post-partum; or (d) a life-threatening or terminal medical condition for which the person already received extensive care. Psychiatric diagnoses were confirmed by the treating psychiatrist in accordance with ICD-10 diagnostic criteria [15].

The RUMI instrument

RUMI was adapted from the most recent validated version of the RUD Instrument [16] to measure the direct costs associated with SMI.

Data collection takes place via an interview supplemented with information from other sources, including the medical records. The patient is asked questions about the extent to which they, during the previous 6 months, have had: interventions from psychiatric and physical healthcare; social services interventions; justice system contacts (see Supplementary Appendix 1). A RUMI assessment takes approximately 20 min to administer to the patient. In MINT, the RUMI interview was carried out by staff involved in the patient's care who, where necessary, retrieved additional information from caregivers, clinical team members or social workers, and/or the medical records.

The cost components in RUMI were classified according to the following definitions:

• Direct healthcare costs: Physical and psychiatric care settings (hospitalization, number of care days and number of visits); primary health care (number of visits to doctor and nurse, paramedical contacts); type of medications doses and formulations; outpatient psychiatric care (visit to doctor, nurse, psychologist or paramedic staff).

• Direct non-healthcare costs: Social services (visit to social worker, support in housing, sheltered accommodation, housing at an institution).

• Justice system costs: Judicial system and its societal costs such as contact with justice and court costs, police, correctional treatment, probation (prison, tagging).

The cost for each area of RUMI has been collected from operations managers, financial controllers for the social services, administrators at hospitals and legal authorities, respectively (see Supplementary Appendix 2). Cost for visiting a psychiatrist or a physician, social worker, has been calculated based on their salaries, including social fees, for a one hour visit including documentation in the case records. Legal costs refers to the most common legal proceedings such as cases at court including judge, prosecutor, lawyer, and lay assessors.¹

All costs in this study apply to the six months prior to data collection and were measured at baseline.

Clinical outcome measures

Self-reported health and quality of life

There were three measures of self-reported health and quality of life. The first asked people to rate their global health on a scale from one (‘very good’) to five (‘very bad’). The second was the position on the visual analogue scale (EQ VAS) of the EuroQol five dimensions questionnaire (EQ5D). The EQ VAS asks participants to rate their health on a scale from ‘the best health you can imagine’ to ‘the worst health you can imagine.’ The third was the index assessment of the EQ5D.

Psychological functioning and severity of illness

Measures of psychological functioning and severity of illness included the Global Assessment of Functioning (GAF) scale score [17], which represents a psychiatrist's assessment of a patient's current social, occupational, and psychological functioning and the Clinical Global Impressions – Severity scale (CGI-S) scale score [18].

Body mass index (BMI) has strong associations with overall physical health and increased costs [19,20] in the general population and was therefore selected to assess the costs associated with risk for physical illness.

Data collection

Measures in the MINT study were assessed 6 months before the designated start of the intervention (pre-baseline) in the intervention group, with assessments repeated at baseline and six months later, post-intervention. The RUMI data presented in this paper are from pre-baseline assessments (2021 currency value) and were completed by the patient’s therapist/case-manager at the same time as collecting the other MINT data. The therapist/case-manager role is to regularly meet with the patient and participate in coordinated meetings between the patient and various care and social services providers.

Statistical analysis

Statistical analyses were conducted with SPSS (version 25.0, Chicago, IL). All variables were summarized with standard descriptive statistics. A cost analysis was carried out with respect to the full cost per patient for 6 months prior to data collection and were measured at baseline. Relationships with the clinical ratings were quantified using a non-parametric Kendall’s rank correlation. Categorical variables, e.g. group and gender, were analyzed with Pearson's chi-squared test or Fisher’s exact test, if the expected cell frequency was 5 or less.

The significance level in all analyses was set at 5% (two-tailed).

Results

Measurements and costs

RUMI data were collected for 107 (90%) of the 119 subjects included in the MINT study, all data in this study apply to the six months prior to data collection and were measured at baseline. The therapists in 12 cases did not conduct the RUMI interview. No systematic loss of data was noted, with no significant differences between the total sample and the participants assessed with RUMI. The mean age of participants was 45.94 years (SD 11.45), 51% were female, and 27% had college or higher education (Table 1). The most common psychiatric diagnoses were schizophrenia (48%) and schizoaffective disorder (17%). In the total sample, 62% of participants had a CGI score >3, indicating moderate to severe illness.

Table 1. Patient characteristics of patients included in the project (n = 107^a).

Variables		Patients
Age (years)	Mean	45.9
	Range	21–82
Participants, males	% (n)	49 (52)
Education college or higher	% (n)	27 (29)
EQ5D^b
General health (VAS)	M (SD)	60.6 (20.1)
Index	M (SD)	0.73 (0.221)
Main psychiatric diagnosis^c
Schizophrenia	% (n)	48 (51)
Schizoaffective disorder	% (n)	17 (18)
Delusion disorder	% (n)	13 (14)
Psychosis UNS	% (n)	8 (8)
Bipolar disorder	% (n)	5 (5)
Other	% (n)	9 (10)
GAF^d >50	% (n)	41(41)
CGI^e >3 (moderate or worse)	% (n)	62 (63)

^a Sporadic missing values might occur.

^b EuroQol.

^c According to DSM-IV.

^d Global Assessment of Function.

^e Clinical Global Impression Scale.

The average total treatment and care costs (2021) from mental and physical health services for 6 months, including the cost of medication/drugs and social services interventions, was 8349 EUR (SD 17,923 EUR). This included mean costs for social services interventions for six months of 2948 EUR (SD 4323 EUR) and mean costs for physical and psychiatric care of 5349 EUR (SD 16,122 EUR). Mean drug costs for mental and physical health intervention per patient were 894 EUR (SD 1161 EUR), Table 2. Sixteen out of the 107 patients accounted for 74% (661,046 EUR) of the total costs (893,306 EUR).

Table 2. The average costs 2021 six months prior to data collection for treatment and care of psychotic patients in Sweden; all cost are given in Euro (€); patient characteristics of patients included in the project (n = 107^a).

National level	Median	M	SD	Minimum	Maximum
Social services	1.907	2.948	4.323	103	41.064
Healthcare, psychiatric and physical treatment, incl. pharm. drugs	89	5.349	16.122	0	63.718
Pharmaceutical drugs only	478	894	1.161	0	4.538
Justice system	0	51	327	0	3.253
Total	2.599	8.349	17.923	103	97.488

Md: median; M: mean; SD: standard deviation.

^a Sporadic missing values might occur.

Higher levels of education (r=–0.207, p < 0.014) and psychosocial function, GAF (r=–0.258, p < 0.001) were negatively correlated with costs of social services. Severity of illness (CGI) was positively correlated with costs for physical and psychiatric treatment (r = 0.161, p < 0.027), costs for social services (r = 0.245, p < 0.002) and total costs (physical and psychiatric treatment, social services and justice system) (r = 0.198, p < 0.007). Body mass index was positively correlated with the sum of costs for physical and psychiatric treatment (r = 0.155, p < 0.019) and the costs for all medicines prescribed from physical and psychiatric treatment services (r = 0.154, p < 0.019). There was no significant correlation between costs and age, global health, or quality of life (EQ5D) (Table 3).

Table 3. Relationships between patient characteristics and cost per year, 2021.

	Physical and psychiatric treatment, sum of costs per year incl. medicines, baseline	Social services, sum of costs per year, baseline	Justice system, sum of costs per year, baseline	Total costs, per year, baseline	Sum of costs for all medicines, baseline, in SEK
Kendall's tau_b
Age, baseline
Correlation coefficient	–0.067	–0.088	–0.016	–0.120	0.037
Sig. (two-tailed)	0.309	0.208	0.840	0.070	0.569
n	108	109	111	107	112
Global health baseline
Correlation coefficient	0.098	–0.045	0.095	0.064	0.063
Sig. (two-tailed)	0.180	0.565	0.277	0.385	0.386
n	107	108	110	106	111
EQ5D, index, baseline
Correlation coefficient	–0.075	–0.076	–0.050	–0.098	–0.120
Sig. (two-tailed)	0.277	0.297	0.542	0.156	0.078
n	106	107	109	105	110
Higher education, college or higher, baseline
Correlation coefficient	–0.048	–0.207*	–0.142	–0.135	0.059
Sig. (two-tailed)	0.544	0.014	0.136	0.091	0.456
n	108	109	111	107	112
GAF, minimum of symptom and function, baseline
Correlation coefficient	–0.034	–0.258**	0.097	–0.108	–0.036
Sig. (two-tailed)	0.624	0.001	0.243	0.121	0.601
n	101	101	103	100	104
Clinical Global Impressions (CGIs), baseline
Correlation coefficient	0.161*	0.245**	0.017	0.198**	0.093
Sig. (two-tailed)	0.027	0.002	0.849	0.007	0.203
n	102	103	105	101	105
Body mass index (BMI), baseline
Correlation coefficient	0.155*	0.026	–0.189*	0.063	0.154*
Sig. (two-tailed)	0.019	0.716	0.018	0.340	0.019
n	105	105	107	104	108

* Correlation is significant at the 0.05 level (two-tailed).

** Correlation is significant at the 0.01 level (two-tailed).

Discussion

In this study, quantifying costs in patients with SMI, RUMI achieved high levels of acceptability, being completed for 90% of participants. It appears to capture costs in the areas of health care, social services intervention and the judiciary across four geographically dispersed healthcare services. Standardized tools, such as RUMI, make it possible to monitor and analyze costs to mental and physical health services, social services, and the judiciary, across different parts of the country.

More severe mental illness (CGI) was associated with a higher total cost while higher social service costs were associated with a lower educational level and poorer global function. In line with the literature for the general population, a higher BMI coincided with higher costs for the physical health and a higher medicine cost for people with SMI.

Comparison with other studies

There are few systematic studies of the costs of schizophrenia and psychotic disorders in Sweden. Ekman reported the average annual cost per patient with schizophrenia for psychiatric services in 2008 was 42,700 EUR. To this was added costs for community mental health care of EUR 12,400 per patient, giving a total cost of 55,100 EUR per patient. The two largest cost items were indirect costs due to lost productivity 28,548 EUR (60%) and community mental health care (22% of the total cost). Psychiatric costs were negatively correlated with function (p < 0.001). Direct costs were calculated as 19,032 EUR [21]. Lindström et al. [22] measured direct and indirect costs of 225 risperidone-treated patients over 5 years. Direct costs upon inclusion were a total of 36,938 EUR and the indirect cost of lost productivity were 25,265 EUR.

Overall, the direct costs for this population over 6 months (8349 EUR) (extrapolated to 12 months as 16,698 EUR) are slightly lower than the costs presented by Ekman et al. [21] (19,032 EUR) and clearly lower than those presented by Lindström et al. [22] (39,328 EUR). Notably, recruitment for MINT was from community settings and only a minority of participants spent time in hospital during the period of interest. More seriously ill and psychosocially vulnerable patients are less likely to participate in a health intervention study [23], which can partially account for a lower cost per patient. RUMI focuses on direct costs identifiable by the clinical team, and so does not measure, for example, loss of income for patients or relatives, although it does include direct costs for contacts with the judiciary.

RUMI

This study introduces a new standardized tool to calculate economic costs for patients with SMI across different locations, services and interventions. The correlations between clinical characteristics and direct healthcare costs as calculated by RUMI are in keeping with findings from other research [16] and suggest validity of the RUMI measures, though further validation work is needed. RUMI was able to identify characteristics of patients with high treatment needs, potentially making it possible to target interventions towards better outcomes thus lowering the costs in the future. The completion results in this study indicate high acceptability for use by clinician-raters.

Strengths and limitations

This study has several strengths.

We used a manual based assessment instrument developed from a well validated instrument in Sweden, RUD [16]. The study was done in a real-life clinical setting with broad inclusion criteria. The RUMI assessment of all cost variables/data collection was made by staff who knew the patients and their living arrangements well. The examination took a relatively short time and the staff had access to the patient medical records.

This assured good quality data with clinical face validity, and a time-efficient process, as the staff member and patient together had access to the information required. It also suggests that the tool has potential to be incorporated into routine clinical practice.

A limitation of the tool is that no consideration is taken of indirect costs such as absence from work for patient and relatives.

Also, this was a relatively small sample size of 107 patients from four different clinics in Sweden. However, the sample in this study is comparable with regard to age and gender with large nation-wide cohort studies of antipsychotics and mortality. Also, there is comparability with regard to age and gender with large nation-wide cohort studies of antipsychotics and mortality [9]. Nevertheless, this sample was recruited from a study of the effectiveness of a community-based physical health intervention, which meant that patients with more challenging psychiatric symptoms or those in longer-term in-patient care were less likely to participate, and their health and care costs were likely to be relatively low.

Generalizability and clinical implications

RUMI was completed in 90% of assessments and enabled clinicians/researchers to estimate resource utilization across different patient groups and care settings. This suggests that RUMI could be a time-effective, clinically based tool to inform cost assessments and planning, and if validated further, for potential use in budgeting and resource allocation.

Further research

An assessment against CSRI and the European version CSSRI-EU, should be considered as part of future research, assessing comparability of outcome measures, accuracy of findings and time for completion.

In addition, there are the hidden costs carried by the families and informal caregivers. This burden deserves to be explored further and to be taken into account when evaluating the cost of illness. A recent report by EUFAMI (European Federation of Associations of Families affected by mental illness) and LSE (The London School of Economics and Political Science/Policy and Evaluation Centre) found that that informal caregivers are spending an average of 43 h a week on caregiving activities [24].

Conclusion

The cost over 6 months for social services and psychiatric and physical treatment interventions measured using the novel RUMI tool was mean 8349 EUR, median 2599 EUR per patient (currency value for 2021), with 16 out of 107 patients accounting for three-quarters of total costs.

RUMI describes the direct costs of mental and physical healthcare along with direct non-healthcare, psychosocial interventions and judicial system costs. It showed high rates of acceptability, being completed in 90% of assessments, and identified clinical characteristics associated with high resource utilization. The wider use of the RUMI tool should be further evaluated to confirm its potential utility in quantifying SMI-related direct costs in Sweden.

Acknowledgements

The authors are grateful to the participating patients and psychosis departments in Gothenburg, Malmö, Kristianstad and Hässleholm, including the care managers and IMPaCT supervisors in Sweden.

Disclosure statement

RS, JE, JW, GE, SVE, UO and EJ declare that they have no conflicts of interest. FG has received support or honoraria from, Lundbeck, Otsuka and Sunovion, and has a family member with previous professional links to Lilly and GSK.

Correction Statement

This article has been republished with minor changes. These changes do not impact the academic content of the article.

Additional information

Funding

MINT was affiliated to the IMPACT programme which was funded by the National Institute for Health Research (NIHR) Programme Grant (RP-PG-0606-1049). FG is in part supported by the National Institute for Health Research’s (NIHR) Biomedical Research Centre at South London and Maudsley NHS Foundation Trust and King’s College London, the Maudsley Charity and the National Institute for Health Research (NIHR) Applied Research Collaboration South London (NIHR ARC South London) at King’s College Hospital NHS Foundation Trust.

Notes on contributors

Richard Stenmark

Richard Stenmark, M.Sc. Researcher at the Severe Mental Illness research unit, Dept of Clinical Sciences Lund, Lund University. Licensed psychologist and specialist in clinical treatment. Supervisor in psychotherapy at the Department of Psychology and Social Work, Mid Sweden University, Sweden.

Jonas Eberhard

Jonas Eberhard, MD PhD, Associate Professor & Senior lecturer at the Dept of Clinical Sciences in Helsingborg, Lund University, Psychiatry research Skåne, Sweden. Focusing on Epidemiology and clinical trials in Severe Mental Illness.

Fiona Gaughran

Fiona Gaughran, MD, Professor of Physical Health and Clinical Therapeutics in Psychiatry, Institute of Psychiatry, Psychology and Neuroscience, King’s College London. Lead Consultant National Psychosis Unit and Director of R&D, South London and Maudsley NHS Foundation Trust.

Erik Jedenius

Erik Jedenius, PhD, Researcher at the Severe Mental Illness research unit, Dept of Clinical Sciences Lund, Lund University, Psychiatry research Skåne, Sweden and Area of activity Psychiatry Psychosis at Sahlgrenska University Hospital. Research; health economy in severe mental illnesses, dementia and scale evaluation and development.

Notes

1 A more detailed description of the procedures can be obtained after contact with the corresponding author.