Skip to main content
Download PDF
Download ePub

Cost structure in specialist mental healthcare: what are the main drivers of the most expensive episodes?

International Journal of Mental Health Systems volume 17, Article number: 37 (2023) Cite this article

Abstract

Background

Mental disorders are one of the costliest conditions to treat in Norway, and research into the costs of specialist mental healthcare are needed. The purpose of this article is to present a cost structure and to investigate the variables that have the greatest impact on high-cost episodes.

Methods

Patient-level cost data and clinic information during 2018–2021 were analyzed (N = 180,220). Cost structure was examined using two accounting approaches. A generalized linear model was used to explain major cost drivers of the 1%, 5%, and 10% most expensive episodes, adjusting for patients’ demographic characteristics [gender, age], clinical factors [length of stay (LOS), admission type, care type, diagnosis], and administrative information [number of planned consultations, first hospital visits, interval between two hospital episode].

Results

One percent of episodes utilized 57% of total resources. Labor costs accounted for 87% of total costs. The more expensive an episode was, the greater the ratio of the inpatient (ward) cost was. Among the top-10%, 5%, and 1% most expensive groups, ward costs accounted for, respectively, 89%, 93%, and 99% of the total cost, whereas the overall average was 67%. Longer LOS, ambulatory services, surgical interventions, organic disorders, and schizophrenia were identified as the major cost drivers of the total cost, in general. In particular, LOS, ambulatory services, and schizophrenia were the factors that increased costs in expensive subgroups. The “first hospital visit” and “a very short hospital re-visit” were associated with a cost increase, whereas “the number of planned consultations” was associated with a cost decrease.

Conclusions

The specialist mental healthcare division has a unique cost structure. Given that resources are utilized intensively at the early stage of care, improving the initial flow of hospital care can contribute to efficient resource utilization. Our study found empirical evidence that planned outpatient consultations may be associated with a reduced health care burden in the long-term.

Background

Mental disorders are the costliest conditions to treat in Norway [1], both in terms of health care expenditures and gross societal losses [2]. Consequently, the financial burden of mental illness has been studied from a various of angles, including societal perspective [3] and healthcare provider perspective [1]. In Norway, the Norwegian Directorate of Health provides annual health care service cost information based on data submitted by healthcare providers [4]. Since these cost data are collected and utilized primarily for management purposes in Norway, such as reimbursement and productivity comparison, they are more frequently employed for macro-level (regions and institutions) analysis. Perhaps because individual direct payments are relatively low in Norway, micro-level data (individual cost data) are relatively less frequently analyzed than in countries with high out-of-pocket expenses, where they must be analyzed for billing purposes. However, individual cost data can have major implications for systems and individuals, regardless of the direct payment sources. Moreover, cost studies are essential for providing high-value services in a sustainable manner. In order to assess whether an intervention provides good value, it is necessary to access intervention’s cost information as well as their clinical benefits and risks [5]. However, research indicates that general cost awareness is low in many countries [6,7,8,9], which necessitates enhanced access to cost information and training to be able to make more efficient and optimal decisions [10, 11]. As an example, in 2021, one Norwegian municipality proposed a 600,000 USD (“hoping to convince politicians”) as initial funding for a pilot program to introduce early care for substance-abuse patients [12]. A government document about the pilot program had reported a 33% reduction in hospital stays [13]. In a situation such this, people can make effective arguments and decisions about what to do if they are aware of the current scale of resources utilized by substance-abuse patients and the financial impact of a 33% reduction in days of hospitalization. Just a 21% reduction in the hospital stays of the one substance-abuse patient who consumed the most resources in 2021 would result in a savings of 620,000 USD, which shows that the suggested amount of initial funding would be negligible in light of its overall effect. Increased availability of cost studies and data will facilitate policy discussions based on the evidence.

Historically, cost studies were conducted in order to compare hospital reimbursements with actual expenses [14,15,16]. In the early phase of cost studies in clinical settings, hospital staff manually recorded patients’ resource utilization [17]. This method, however, had a great risk of losing precision. Due to the development of an automated hospital administration system, cost data can now be gathered directly with improved accuracy. One of the systems that has potential for the field of healthcare cost studies is the Cost Per Patient (CPP). The CPP is a patient-level cost-calculation model that is designed to collect data and demonstrate how resources are used by each patient during each hospital visit [18]. The concept of a patient-level cost-tracking system is used in various countries under different names. In England, the system is called the Patient-Level Information and Costing System (PLICS), which was implemented by the National Health Service (NHS) in 2015 [5]. Sweden calls its system Kostnad Per Pasient (KPP), which has been in operation since the early 1990s [6]. Though Germany’s healthcare system is different from the NHS model used in countries such as Sweden and England, Germany’s “Instituts für das Entgeltsystem im krankenhaus (InEK)” was established in 2001 to manage a comprehensive pricing system [7]. One of the organization’s main tasks is calculating patient-level costs [8]. One of the advantages of cost data derived from the CPP is that it is based on local “bottom–up” costing techniques where the costs of episodes reflect the actual expenditures required to provide care. Domestic cost studies can provide the most accurate information because the specific arrangement of national healthcare service provision that determines how and where costs are likely to incur vary from country to country.

This study aimed to analyze the cost structure of specialized mental healthcare and identify the main cost drivers of expensive episodes. It would be advantageous for clinicians to have an understanding of the hospital's resource utilization so that they can voice their professional opinions also on administrative decisions based on costs.

Materials and methods

Data source

The data were collected from, Helse Førde, a regional health enterprises in Norway. Helse Førde is part of Western Norway Regional Health Authority (Helse Vest). Helse Førde serves approximately 109,000 inhabitants in eighteen municipalities, as of 2023. The data were extracted from the CPP system from January 1, 2018 to December 31, 2021. The inclusion criteria were cost records of patients of any age who received inpatient or outpatient treatment in the specialized mental healthcare service during this time period. Once patients register their social security number and the name of a hospital system, the cost is calculated along with current-stay information, such as date of arrival and time of discharge. Each hospital visit is recorded as a single episode with one diagnosis-related group (DRG) code (e.g., episode 1: a one-hour outpatient session due to depressive symptoms, episode 2: a hospital stay of ten days coded with the diagnosis of schizophrenia). One hundred and twenty three hospital episodes of incorrect grouping and double records were excluded from the analyses. Twenty-three episodes of unknown admission type were also excluded. In summary, 180,220 episodes, which represented 99.9% of all actual episodes during the inclusion period, were analyzed, making the included data excellent in terms of their completeness in mental health service research.

Study design

This was a cross-sectional retrospective study on hospitals’ patient-level cost-data of the specialist mental healthcare division. This study adopted the classic two-part model for the cost analysis [19, 20]. The idea of grouping episodes by two categories (expensive and not) is that the episodes in hospitals are frequently a mixture of two types. Some patients visit the hospital for minor clinical needs, such as assessment and advice, while others require expensive treatment, such as an operation and long-term hospitalization. The former episodes are represented by values close to zero in the right-skewed cost distribution graph shown in Fig. 1, whereas the latter group is represented in the heavy right tail. We used three different cut-off points of 1%, 5%, and 10% of total costs in line with international studies [21]; thus, those in and above the 99th, 95th, and 90th percentiles of the cost distribution were defined as expensive groups. We examined the cost structure and cost drivers of these expensive groups based on this two-part model.

Fig. 1
figure 1

Illustration of the two-part model: cost distribution of 180,220 episodes

Full size image

Cost structure was analyzed from two simplified perspectives; the traditional perspective that entails the basic concepts of general accounting [22] and a relatively recent perspective (introduced in the 2000s) based on a hospital’s activity [23, 24]. The latter has been recommended for hospital cost accounting [25] as a ‘time-driven activity-based costing’ (henceforth, activity-based costing). With the traditional perspective, costs were classified as direct and indirect costs. For the activity-based perspective, costs were classified according to a hospital’s seven main activities: ambulatory service, intensive care, operations, anesthesia, radiology, and outpatient and inpatient care. Each cost category consisted of eight sub-costs elements, including direct labor costs (clinicians and other healthcare professionals), direct consumable costs (medicines, main consumables, and other consumables), and indirect costs (capital costs and other overhead costs) (see Appendix 1).

Multivariate generalized linear models (GLMs) with ten independent variables were used to identify cost drivers: two demographic variables [gender and age], four clinical variables [length of stay (LOS), admission type, care type, and diagnostic-related group (DRG)] and three administrative variables [number of planned consultations, first hospital visits, and interval between two hospital episodes (interval since last hospital episode)]. This study included basic patient information that has been reported to be associated with high healthcare costs, such as age, gender, DRG, and LOS [26,27,28]. We also analyzed new variables that were available via the CPP system to see if they affected total costs. The three new variables we included were “first hospital visits,” “interval between two episodes,” and “number of planned consultations.” The first hospital visit variable refers to the first visit registered during a follow-up period of four-years, which tended to be higher in the expensive groups than the average. The interval between two episodes was included to determine whether episodes that were too close together or too far apart may affect costs possibly because they represent early discharge or dropout. The number of planned consultations was the total number of planned psychiatric-/psychotherapeutics consultations performed by both psychologists and psychiatrists. This variable was used to determine whether planned consultations affected both total costs and average costs.

Age and gender were recorded at the time of inclusion. Seventy-three DRG codes were condensed into nineteen main categories based on professional judgment (see Appendix 2). Prior to analysis, some diagnostic categories were replaced with “Anonymous” to maintain anonymity if there were less than five patients in a diagnostic group within the same age group. The DRG category of “Not specified” refers to episodes that did not contain diagnostic information, such as “family-centered outpatient services.” Surgical intervention refers to any surgical procedure performed in a specialist mental healthcare division, as defined by the hospital, such as anesthesia involved in electroconvulsive therapy. LOS was classified into five categories: day treatment (outpatient within 1 day); less than one-week (1–7 days); less than a month (8–30 days); less than three months (31–90 days); and more than three months and less than one year (91–365 days).

Admission type and care type were used to distinguish the types of hospital visits, such as acute/planned visits and outpatient/inpatient service. The total cost of each episode was used as the dependent variable in order to identify cost drivers affecting the total consumption of resources. Cost per day (total cost/LOS) was also used as a dependent variable to eliminate the effect of accumulated hospital days.

Statistical analysis

Descriptive statistics were used the estimate the mean and 95% confidence intervals of total healthcare services costs. The GLM was used because the dependent variable “cost” was highly skewed to the right and it did not have negative values. Although there is no single optimal or dominant model for resource utilization and cost analysis in healthcare [29], the GLM with a log transformation and the Gamma distribution has been recommended for cost analysis because of its better performance in estimating population means and its realistic description of cost data [30,31,32]. We tested six different GLM models on our dataset, with log and square root transformations, and Gaussian, Poisson, and Gamma distributions. We compared the Akaike information criterion (AIC) [33] and the Bayesian information criterion (BIC) [34] of each model. The model with the lowest AIC and BIC in our dataset was the square root transformation with a Gamma distribution. However, the differences between the log transformation with the Gamma distribution model were not significant based on the AIC and BIC values, and as a log transformation is more widely used in this field, we opted to use the log distribution with a Gamma distribution model to facilitate comparisons with other studies. Independent variables that were statistically significant in the bivariate analyses were included in the multivariate model; a 5% significance level was used throughout. The average marginal and incremental effects of each variable were calculated in United States dollars (USD). Using the average exchange rate from 2021 to 2022, 10 NOK was converted to 1 USD. The data were analyzed in STATA SE version 17.

Results

Patients’ background characteristics

Table 1 displays the background characteristics of the patients. Compared to all the episodes, the expensive groups had a higher proportion of male patients. The adolescent group had the largest proportion of all episodes, whereas age was more evenly distributed across the expensive subgroups. The expensive groups had a higher proportion of “acute admissions,” “first hospital visits,” and “anonymous” categories in diagnosis.

Table 1 Patients’ background characteristics
Full size table

Cost structure of specialist mental healthcare divisions

Table 2 presents the cost structure of specialist mental healthcare divisions by the expensive and non-expensive groups. High labor (personnel) costs were observed, and labor costs were estimated to account for approximately 87% of total expenditures. Compared to the previous SAMDATA report by the Norwegian Directorate of Health, outpatient consultation and bed-day costs are comparable (see Appendix 3) [4]. From an accounting perspective, direct and indirect costs were constant across the expensive and non-expensive episodes. However, the activity-based hospital cost structures differed between the expensive and inexpensive groups. The results showed that the more expensive an episode was, the higher the proportion of inpatient care costs were (ward cost). Inpatient care costs accounted for 99% of the costs of the 1% most expensive episodes, 93% of the costs of the 5% most expensive episodes, and 88% of the costs of the 10% most expensive episodes. In contrast, inpatient care costs generally accounted, on average, for only 66% of total costs. Another feature that stood out was that a small number of episodes consumed the majority of available resources, with 57% (USD 107.7 million out of 188 million, see Table 2) of hospital resources being allocated to 1% of hospital episodes.

Table 2 Cost structure of specialist mental healthcare divisions
Full size table

General analysis: what are the main cost drivers?

The GLM results of the general analysis are shown in Table 3. Since we used a multivariate model, each variable must be interpreted in the context of controlling for the other variables to estimate its unique contribution to costs. LOS was found to have the greatest impact on the total increase in cost, among all the statistically significant variables. The total cost increased as LOS increased (USD 2,415 ~ 65,088), but the average cost (calculated by dividing the total cost by the LOS) of short episodes (LOS 1-7) had the greatest impact on the cost increase (USD 628). The variable that made the second largest contribution to the total cost was type of care. Specifically, surgical interventions had the largest impact on both the total and average costs. The effect of ambulatory service on the cost increase differed depending on whether it led to a hospitalization or not. The first hospital visit, a very short or long hospital re-visit (within 3 days and after more than 90 days), and younger age (0–9 years old) were also associated with an increase in cost.

Table 3 Results of the GLM for the general analysis
Full size table

Expensive episodes analysis: what are the main cost drivers?

Although there were inconsistencies between the subgroups, the cost drivers of expensive episodes did not differ significantly from those in the general analysis. Table 4 illustrates the results of the GLM of high-cost episode groups. LOS was found to have the greatest impact on total expenditure in all the groups, and was similar to the general analysis; i.e., the average cost per day decreased as the LOS increased (except for the 1% group). The 1% most expensive group showed a similar tendency in which a shorter LOS decreased the average cost (USD 973 → 822 → 656), but this did not apply to hospitalizations longer than three months (USD 827). As indicated by the highly skewed cost curve, 1% of the episodes were extremely expensive, even compared to the 5% and 10% episodes (Fig. 2); consequently, the effect of LOS on the 1% group appears to be quite distinct from that of some variables. The variables that were statistically significant for the 5% and 10% groups (e.g., as sex, a very short re-visit, and the number of planned consultations) were not significant for the 1% group, possibly because of condition severity in the 1% group or its smaller sample size, compared to the other groups.

Table 4 Results of GLM for the 1%, 5%, and 10% expensive groups
Full size table
Fig. 2
figure 2

Cost distribution of 180,220 episodes in the actual USD scale

Full size image

Ambulatory inpatient service was found to be an important variable for all the subgroups, as was a surgical intervention in the 1% and 5% groups. Among the DRGs, schizophrenia and substance abuse were associated with increased total costs in all three high-cost subgroups.

Discussion

Cost structure

One of the distinctive features of the cost structure of specialist mental healthcare was that only 1% episodes consumed more than half of total hospital resources. This finding is consistent with the well-known fact that a small proportion of the population consumes a disproportionately large share of healthcare spending [21, 35,36,37]. Frequent hospital visits are an expected part of the disease pathways of patients with ongoing or life-long functional impairment due to psychiatric disorders. The disproportionate use of health services in a need-based healthcare system and the high costs attributable to a small number of patients are, therefore, unavoidable and rather reasonable. Moreover, a minority subgroup of patients in specialist mental healthcare are forensic patients on court-mandated hospital stays of long duration. The reason for examining particularly expensive episodes, however, is to understand the characteristics of heavy users, to investigate how they utilize the service, and to identify improvement areas for preventing the revolving door phenomenon, when it is possible to do so. High-cost episodes may also be associated with low productivity and substandard service quality, such as multiple avoidable acute readmissions. Focused care for high-cost cases has been implemented in a variety of clinical settings because even small changes can have a significant impact on patient outcomes and healthcare costs. These efforts, however, have not always been successful. For example, the author of a longitudinal study of heavy service users in Switzerland concluded that “preventive interventions to curb excessive service use appear to be out of reach for the majority of heavy users” due to difficulty predicting hospital resource use [38]. Similarly, a study in the United Kingdom reported that a new intervention for heavy users had no discernible effect [39]. However, past studies have calculated hospital resource use based on written questionnaires (e.g., “What inpatient services have you used in the past three months?”) [40] or days of hospitalization; thus it is important to note that improved cost data, such as the data that is the basis for our study, may enhance the precision of analyses. Fortunately, contrary findings have recently demonstrated that some intensive interventions are effective among high-cost patients. For instance, assertive community treatment (ACT), an intensive type of care for people with severe mental disorders, has been demonstrated to have positive outcomes in numerous nations [41].

Other features that stood out in our analysis of the cost structure was the high ratio of labor costs, and the high ratio of inpatient costs in the expensive groups. The proportion of labor costs in specialist mental healthcare seems to have been stable for a long time. According to Finnish hospital cost research in 1980, the cost of labor accounted for 87% of the total cost [42], which is identical to the 87% found in our 2018–2021 data. This implies that the minimum number of personnel required to care for psychiatric patients has remained constant over time, and that technological investment in mental health settings has been low. A high ratio of inpatient care costs for expensive groups was also reported in other studies [27, 43].

We stratified the costs for the 1%, 5%, and 10% most expensive ‘individual patient’ (not episodes) to make further comparisons with previous studies. As mental health patients typically have multiple episodes with various diagnoses, we summed the total amount of cost spent per patient. The results were remarkably consistent with those of a US study that covered the time period from 1920 to the 1980s [44] and a systematic review that covered references from 1995 to 2012 [21] (see Appendix 4). The striking similarity of the resource-use concentration and structures across time and country suggests that the existing findings of high-cost patients and heavy user research may be useful in the Norwegian context. For instance, a systematic review of studies of high-cost patients estimated that a maximum of 10% of the total cost was considered a preventable expenditure [21].

Cost drivers

LOS had the greatest impact on hospital resource utilization. It is not surprising that the care costs rise as the number of days spent in the hospital increases. However, it was unclear whether prolonged hospital stays themselves would exacerbate the increase in total costs. The “average cost per day,” which removes the cumulative effect of LOS, tended to decrease as LOS increased. The average care cost was highest on the first seven days (see Appendix 5). Cromwell et al. reported comparable results for hospitalization costs and LOS [45]. According to that study, the first day of hospitalization was expensive due to the auxiliary services provided at the time of admission. Prolonged hospitalization was the biggest reason for the cost increase, but the rate of cost growth slowed down as hospital LOS increased, possibly reflecting the cost-reducing effect of long-term hospitalization. These findings are consistent with those of early US efforts to reduce LOS in order to contain medical expenses [46]. However, some studies have revealed the possibility that LOS reduction may not be an effective way to curb total costs due to the relatively low marginal cost of last-day-stay (the final phase of hospitalization is primarily for recuperation) [47], and a possible trade-off effect between a shorter LOS and increased readmissions [48]. Efforts to reduce LOS in Norway were accelerated by the implementation of the Norwegian Coordination Reform in 2012 [49]. The LOS In psychiatry has decreased by approximately 30% compared to 2009, which means it decreased from an average of 27 days to 18 days [50]. No research has yet been conducted in Norway on the effects of a shorter psychiatric LOS on total healthcare costs; however, a nationwide post-reform qualitative study suggests that it is unclear whether the shorter LOS resulted in a decrease in overall healthcare costs due to a marked increase in reported hospital readmissions [51]. In Norwegian context, this may suggest that a reduction in hospital bed-days may increase municipal expenditures. Based on these facts, the findings of this study suggest that a reduction in LOS in specialist mental healthcare may not always be associated with a cost reduction, even though LOS is the most influential factor that drives total costs. This result has potential implications for policymakers and clinicians collaborating with patients in discharge processes.

First hospital visit refers to the initial visit registered during the follow-up period of 2018–2021. Although this variable was not sufficient to account for extremely expensive episodes of 1% or 5%, it appears that first hospital visits use more resources than subsequent visits. Along with the results obtained from the LOS analysis, a possible implication is that enhancing the initial experience of patients, such as logistics optimization and better use of capacity, can also be economically beneficial for hospitals.

The interval between two episodes was a very interesting predictor in this data. Overall, these results indicate that costs are more likely to deviate from the mean when the intervals between visits are extremely short or extremely long. This suggests two prototypical clinical scenarios: first, a discharge that is too early while the patient remains acutely ill, leading to rapid readmission; and second, chronic and episodic pathology, which requires more serious evaluation and treatment upon readmission. The first of these may be preventable through better policy; the second may not be. Because the present data is limited to a four-year time frame, longer intervals were excluded, which could result in an underestimation of the number of people and typical time between visits. Consequently, the actual effect should be expected to be higher than the observed effect.

The current study found that one additional outpatient consultation added approximately USD 363 to the total cost (see Appendix 6). However, as the number of planned consultations increased, there was a very small but statistically significant total cost reduction. This result can be interpreted in two ways. Patients who receive frequent outpatient consultations have a lower overall cost; put differently, healthier patients utilize psychosocial consultations more often. Alternately, increased psychiatric/psychotherapeutic consultations have a long-term cost-saving effect that can offset the cost of added consultations. Many studies have demonstrated there is an economic value of psychotherapy via a reduction in other medical and/or social costs. For example, more than one-third of hospitalized patients for medical and surgical reasons have psychiatric comorbidities [52, 53], and appropriate psychiatric interventions can decrease the LOS of these patients, thereby reducing total costs [54, 55]. This is one of the most studied effects used to demonstrate the economic value of psychotherapy. Our data provide intriguing empirical evidence that psychotherapeutic consultation may already have a cost-offsetting effect in specialist mental healthcare, mitigating high costs by preventing patient deterioration over the long term. Helse Førde is known to have relatively lower outpatient consultation rate than the national average [56,57,58,59]. Therefore, generalizations require caution: cost-offsetting effect of psychotherapeutic consultations might be underestimated.

Surgery had a significant effect on increasing the total cost of all the expensive subgroups, although only 133 episodes out of 180,220 records included surgical interventions (Table 1), and the cost of the operation and anesthesia itself was negligible, accounting for only 0.1% of the total cost (Table 2). Surgical intervention is a reliable indicator of high resource consumption because it may indicate severe depression, mood disorders and self-harm. The effect of surgical intervention is reflected in the costs of other hospital activities, such as intensive care, radiology, and wards, resulting in an increase in total costs.

DRGs had a relatively minor impact on costs compared to LOS and care type, yielding inconsistent outcomes across groups. Among expensive episodes, schizophrenia and substance abuse were associated with increased total costs in all the expensive subgroups. Cost studies on mental disorders have produced similar findings for schizophrenia [1, 28, 60, 61], substance use [1, 61,62,63,64], and organic disorders [61]; (including dementia). As the treatment cost for dementia increases with symptoms [65] and disease progression [66], keeping patients at an earlier stage is beneficial for both the patient and the healthcare system. This also applies to schizophrenia. Relapse prevention has been shown to be crucial for controlling healthcare costs in many studies [67,68,69].

Our study revealed that male patients and younger patients had higher costs; however, the associations of age and gender with high-cost episodes were not consistent. Several studies have reported opposite results for females and older age groups [1, 27, 70] while others have reported the same association between young age and high hospital costs [61, 71,72,73]. Gender and age variables appear to have a range of results depending on the research data and patient episode mix.

As seen in Table 1, there was a smaller proportion of males involved in overall episodes, but a higher proportion of males involved in expensive episodes, which may indicate males exhibit a low level of help seeking behavior, which is well-documented in psychiatry [74, 75]. Male patients accounted for 44% of hospital visits, but 50–56% of the expensive episodes in our study. This might also reflect the fact that the Norwegian mental healthcare system treats significantly more men than women who have been sentenced by the court to involuntary psychiatric treatment after committing violent crimes. Such admissions are often lengthy and resource intensive.

Similarly, the elderly population (over 60 years old) accounted for only 5% of all hospital visits, but between 7 and 20% of the expensive episodes. Given that successful out-patient consultation could lead to less in-patient bed usage [76], efforts to engage these patients in regular mental healthcare services prior to the onset of severe episodes may be economically beneficial to the hospital, as well as advantageous to patients if they are satisfied with this level of care.

This is the first study in Norway to explore the individual-level cost structure of specialist mental healthcare. Our research provides a review of recent patterns of psychiatric patient resource utilization, cost-increasing factors, and differences between domestic and international findings. The current study covers a comprehensive number of episodes, including child and adolescent patients, and includes various types of disease pathways. Hence, the problem of selection bias is not very likely. In addition, by using a four-year time horizon, we analyzed the results for longer-term resource utilization patterns compared to previous LOS and cost studies.

This research has some limitations. Its biggest potential weakness is the anonymous groups in DRG. This was necessary due to the need to protect personal information, but this may have impaired the accuracy of the DRG-related analyses. Particularly, adolescence (age group 10–19) was a significant factor that increased the cost of 1% episodes, but the DRGs of nearly 47% of the episodes in the adolescence group were anonymized. Although these diagnoses are rare and affect a small number of patients, they account for a significant portion of hospital resource utilization. This suggests that appropriate research and special treatment training for clinicians can have a positive economic impact on the healthcare system. Another potential weakness is that our data were limited to costs of special mental healthcare divisions, which may underestimate the true costs associated with high-cost psychiatric patients. A growing body of research demonstrates that high-cost patients with mental illness consume more hospital resources than do non-psychiatric high-cost patients [77]. Recent meta-analyses suggest that patients with severe mental illness make greater use of non-psychiatric health services and represent a greater economic burden on hospitals [78]. Similar results have been reported for children and adolescents [79]. A comprehensive study incorporating relevant somatic costs could help estimate the true impact of heavy-use in specialist mental healthcare.

Conclusion

A specialist mental healthcare division has a unique cost structure. As 1% of episodes consumed 57% of hospital resources, any intervention that effectively prevents high-cost episodes will be likely to result in significant resource savings. LOS is the most reliable and influential factor in hospital resource consumption. However, if a shorter LOS results in frequent readmissions, it would be counterproductive in the long-run, as the costs incurred on the early days of readmission are likely to be higher than the costs incurred on the later days of previous admission. Our study revealed that first hospital visits and too short re-visits are more expensive than other visits. Therefore, improving the initial flow of hospital care, where resource utilization is intensive, could significantly enhance cost-efficiency. Our study found empirical evidence that total resource consumption is likely to decrease as the number of planned outpatient consultations increases, in terms of total costs over a relatively long period of four-years. Due to the small magnitude of the effect, however, additional research is needed. Finally, male and elderly populations in the region have been observed to make fewer hospital visits, but they account for a higher proportion of expensive subgroups. Efforts to help these patients have a stable connection with a care provider before they have expensive and serious episodes might be clinically and financially beneficial. These findings should be incorporated into future healthcare policies to improve patient care and optimize hospital resource utilization.

Availability of data and materials

In line with the requirements of the data protection officer of Førde Hospital Trust, reasonable requests for access to data are to be made in writing to the corresponding author.

Abbreviations

AIC:

Akaike information criterion

ACT:

Assertive community treatment

BIC:

Bayesian information criterion

CPP:

Cost per patient

DRG:

Diagnosis-related group

FACT:

Flexible assertive community treatment

GLM:

Generalized linear model

KPP:

Kostnad per pasient (En: Cost per patient)

LOS:

Length of stay

NOK:

Norwegian kroner

References

  1. Kinge JM, Dieleman JL, Karlstad Ø, Knudsen AK, Klitkou ST, Hay SI, et al. Disease-specific health spending by age, sex, and type of care in Norway: a national health registry study. BMC Med. 2023;21(1):201.

    Article  PubMed  PubMed Central  Google Scholar 

  2. Kinge JM, Saelensminde K, Dieleman J, Vollset SE, Norheim OF. Economic losses and burden of disease by medical conditions in Norway. Health Policy. 2017;121(6):691–8.

    Article  PubMed  Google Scholar 

  3. Christensen MK, Lim CCW, Saha S, Plana-Ripoll O, Cannon D, Presley F, et al. The cost of mental disorders: a systematic review. Epidemiol Psychiatric Sci. 2020;29: e161.

    Article  CAS  Google Scholar 

  4. Pedersen M, Lilleeng SE, Sitter M, Lehn H, Myrli TR, Mortensen S, et al. SAMDATA spesialisthelsetjeneste 2021: Status og utviklingstrekk for spesialisthelsetjenesten Kostnader. In: Helsedirektoratet, editor. Oslo2022.

  5. Owens DK, Qaseem A, Chou R, Shekelle P. High-value, cost-conscious health care: concepts for clinicians to evaluate the benefits, harms, and costs of medical interventions. Ann Intern Med. 2011;154(3):174–80.

    Article  PubMed  Google Scholar 

  6. Sorber R, Dougherty G, Stobierski D, Kang C, Hicks CW, Lum YW. Cost awareness of common supplies is severely impaired among all members of the surgical team. J Surg Res. 2020;251:281–6.

    Article  PubMed  PubMed Central  Google Scholar 

  7. van Boxel GI, van Duren BH, van Boxel E-J, Gilbert R, Gilbert P, Appleton S. Patients’ and health-care professionals’ awareness of cost: a multicentre survey. Br J Hosp Med. 2016;77(1):42–5.

    Article  Google Scholar 

  8. Tiong WHC, Shaughnessy M, Sullivan ST. Cost awareness among doctors in an irish university-affiliated teaching hospital. Clinics and Practice. 2011;1(4): e91.

    Article  PubMed  PubMed Central  Google Scholar 

  9. Schilling UM. Cost awareness among Swedish physicians working at the emergency department. Eur J Emerg Med. 2009;16(3):131–4.

    Article  PubMed  Google Scholar 

  10. Allan GM, Lexchin J, Wiebe N. Physician awareness of drug cost: a systematic review. PLoS Med. 2007;4(9): e283.

    Article  PubMed  PubMed Central  Google Scholar 

  11. Fabes J, Avşar TS, Spiro J, Fernandez T, Eilers H, Evans S, et al. Information asymmetry in hospitals: evidence of the lack of cost awareness in clinicians. Appl Health Econ Health Policy. 2022;20(5):693–706.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Louise N. Ser et stort behov for et FACT-team også i Kinn: - Det er alt jeg selv savnet. Fjordenes tidende. 2021.

  13. Landheim A, Odden S. Evaluering av FACT-team i Norge : Sluttrapport. In: HF SI, editor. 2020.

  14. Bishop CE. Nursing home cost studies and reimbursement issues. Health Care Financ Rev. 1980;1(4):47–64.

    CAS  PubMed  PubMed Central  Google Scholar 

  15. Harris BH, Bass KD, O’Brien MD. Hospital reimbursement for pediatric trauma care. J Pediatr Surg. 1996;31(1):78–81.

    Article  CAS  PubMed  Google Scholar 

  16. Skeie B, Mishra V, Vaaler S, Amlie E. A comparison of actual cost, DRG-based cost, and hospital reimbursement for liver transplant patients. Transpl Int. 2002;15(9–10):439–45.

    Article  PubMed  Google Scholar 

  17. May PR. Cost-effectiveness of mental health care. II. Sex as a parameter of cost in the treatment of schizophrenia. Am J Public Health Nations Health. 1970;60(12):2269–72.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Helsedirektoratet. Nasjonal spesifikasjon for KPP -modellering 2012 : Begreper og metoder. 2012.

  19. Diehr P, Yanez D, Ash A, Hornbrook M, Lin DY. Methods for analyzing health care utilization and costs. Annu Rev Public Health. 1999;20:125–44.

    Article  CAS  PubMed  Google Scholar 

  20. Mullahy J. Much ado about two: reconsidering retransformation and the two-part model in health econometrics. J Health Econ. 1998;17(3):247–81.

    Article  CAS  PubMed  Google Scholar 

  21. Wammes JJG, van der Wees PJ, Tanke MAC, Westert GP, Jeurissen PPT. Systematic review of high-cost patients’ characteristics and healthcare utilisation. BMJ Open. 2018;8(9): e023113.

    Article  PubMed  PubMed Central  Google Scholar 

  22. Prickett AL. General principles of cost accounting. Account Rev. 1944;19(2):169–80.

    Google Scholar 

  23. Keel G, Savage C, Rafiq M, Mazzocato P. Time-driven activity-based costing in health care: a systematic review of the literature. Health Policy. 2017;121(7):755–63.

    Article  PubMed  Google Scholar 

  24. Kaplan RS, Anderson SR. Time-driven activity-based costing. Available at SSRN 485443. 2003.

  25. Kaplan RS, Porter ME. How to solve the cost crisis in health care. Harv Bus Rev. 2011;89(9):46–52.

    PubMed  Google Scholar 

  26. Wolff J, McCrone P, Koeser L, Normann C, Patel A. Cost drivers of inpatient mental health care: a systematic review. Epidemiol Psychiatr Sci. 2015;24(1):78–89.

    Article  CAS  PubMed  Google Scholar 

  27. de Oliveira C, Mason J, Kurdyak P. Characteristics of patients with mental illness and persistent high-cost status: a population-based analysis. Can Med Assoc J. 2020;192(50):E1793–801.

    Article  Google Scholar 

  28. Anderson M, Revie CW, Quail JM, Wodchis W, de Oliveira C, Osman M, et al. The effect of socio-demographic factors on mental health and addiction high-cost use: a retrospective, population-based study in Saskatchewan. Can J Public Health. 2018;109(5):810–20.

    Article  PubMed  PubMed Central  Google Scholar 

  29. Mihaylova B, Briggs A, O’Hagan A, Thompson SG. Review of statistical methods for analysing healthcare resources and costs. Health Econ. 2011;20(8):897–916.

    Article  PubMed  Google Scholar 

  30. Dunn G, Mirandola M, Amaddeo F, Tansella M. Describing, explaining or predicting mental health care costs: a guide to regression models: methodological review. Br J Psychiatry. 2003;183(5):398–404.

    Article  PubMed  Google Scholar 

  31. Dodd S, Bassi A, Bodger K, Williamson P. A comparison of multivariable regression models to analyse cost data. J Eval Clin Pract. 2006;12(1):76–86.

    Article  PubMed  Google Scholar 

  32. Malehi AS, Pourmotahari F, Angali KA. Statistical models for the analysis of skewed healthcare cost data: a simulation study. Health Econ Rev. 2015;5:11.

    Article  PubMed  PubMed Central  Google Scholar 

  33. Akaike H. Statistical predictor identification. Ann Inst Stat Math. 1970;22(1):203–17.

    Article  Google Scholar 

  34. Schwarz G. Estimating thedimension of a model. Annals ofStatistics. 1978;6(2):461–4.

    Google Scholar 

  35. Kent S, Fogarty M, Yellowlees P. A review of studies of heavy users of psychiatric services. Psychiatr Serv. 1995;46(12):1247–53.

    Article  CAS  PubMed  Google Scholar 

  36. Christensen BJ, Gørtz M, Kallestrup-Lamb M. Medical spending in Denmark. Fisc Stud. 2016;37(3–4):461–97.

    Article  Google Scholar 

  37. French E, Kelly E. Medical spending around the developed world. Fisc Stud. 2016;37(3–4):327–44.

    Article  Google Scholar 

  38. Junghan UM, Brenner HD. Heavy use of acute in-patient psychiatric services: the challenge to translate a utilization pattern into service provision. Acta Psychiatr Scand. 2006;113(s429):24–32.

    Article  Google Scholar 

  39. Harrison-Read P, Lucas B, Tyrer P, Ray J, Shipley K, Simmonds S, et al. Heavy users of acute psychiatric beds: randomized controlled trial of enhanced community management in an outer London borough. Psychol Med. 2002;32(3):403–16.

    Article  CAS  PubMed  Google Scholar 

  40. M. K, J. B. CLIENT SERVICE RECEIPT INVENTORY (CSRI): Personal social services research unit (PSSRU); 1980 [Available from: https://www.dirum.org/assets/downloads/634462388066137028-CSRI.pdf.

  41. Bond GR, Drake RE, Mueser KT, Latimer E. Assertive community treatment for people with severe mental illness. Dis Manag Health Out. 2001;9(3):141–59.

    Article  Google Scholar 

  42. Vlnni K, Slllanpää A, Anttinen E. Cost study of mental health services. Acta Psychiatr Scand. 1980;62(S285):258–64.

    Article  Google Scholar 

  43. Pan Y-J, Kuo K-H, Yeh L-L. Healthcare cost, service use and mortality in major psychiatric disorders in Taiwan. J Affect Disord. 2019;246:112–20.

    Article  PubMed  Google Scholar 

  44. Berk ML, Monheit AC. The concentration of health care expenditures, revisited. Health Aff. 2001;20(2):9–18.

    Article  CAS  Google Scholar 

  45. Cromwell J GB, Drozd E, Maier J, Osber D, Evensen C, Richter E, Goldman H. Psychiatric inpatient routine cost analysis – final report. Massachusetts, U.S.: RTI International; 2005.

  46. Carey K. Hospital cost containment and length of stay: an econometric analysis. South Econ J. 2000;67(2):363–80.

    Google Scholar 

  47. Taheri PA, Butz DA, Greenfield LJ. Length of stay has minimal impact on the cost of hospital admission11No competing interests declared. J Am Coll Surg. 2000;191(2):123–30.

    Article  CAS  PubMed  Google Scholar 

  48. Carey K. Measuring the hospital length of stay/readmission cost trade-off under a bundled payment mechanism. Health Econ. 2015;24(7):790–802.

    Article  PubMed  Google Scholar 

  49. Services NMoHaC. The Coordination Reform (2008–2009). In: Services NMoHaC, editor. 2009.

  50. Kortere liggetid i psykisk helsevern (SSB): Statistic Norway (SSB); 2019 [Available from: https://www.ssb.no/helse/artikler-og-publikasjoner/kortere-liggetid-i-psykisk-helsevern.

  51. Gautun H, Syse A. Earlier hospital discharge: a challenge for Norwegian municipalities. Nordic J Soc Res. 2017;8(1):1–17.

    Article  Google Scholar 

  52. Wancata J, Windhaber J, Bach M, Meise U. Recognition of psychiatric disorders in nonpsychiatric hospital wards. J Psychosom Res. 2000;48(2):149–55.

    Article  CAS  PubMed  Google Scholar 

  53. Bourgeois JA, Kremen WS, Servis ME, Wegelin JA, Hales RE. The impact of psychiatric diagnosis on length of stay in a university medical center in the managed care era. Psychosomatics. 2005;46(5):431–9.

    Article  PubMed  Google Scholar 

  54. Oldham MA, Chahal K, Lee HB. A systematic review of proactive psychiatric consultation on hospital length of stay. Gen Hosp Psychiatry. 2019;60:120–6.

    Article  PubMed  Google Scholar 

  55. Yonatan-Leus R, Strauss AY, Cooper-Kazaz R. Psychodynamic psychotherapy is associated with sustained reduction in health care utilization and cost. Clin Psychol Psychother. 2021;28(3):642–55.

    Article  PubMed  Google Scholar 

  56. Helseatlas. Adults in outpatient treatment (2014–2018). 2019.

  57. Helseatlas. Children and adolescents in outpatient treatment (2014–2018). 2019.

  58. Helseatlas. Adults with substance abuse disorder outpatient treatment (2014–2018). 2019.

  59. Helseatlas. Adults with severe mental disorders outpatient treatment (2014–2018). 2019.

  60. Holohean EJ, Pulice RT, Donahue SA. Utilization of acute inpatient psychiatric services: “Heavy users” in New York State. Admin Policy Mental Health Mental Health Serv Res. 1991;18(3):173–81.

    Article  Google Scholar 

  61. Baca-Garcia E, Perez-Rodriguez MM, Basurte-Villamor I, Quintero-Gutierrez FJ, Sevilla-Vicente J, Martinez-Vigo M, et al. Patterns of mental health service utilization in a general hospital and outpatient mental health facilities: analysis of 365,262 psychiatric consultations. Eur Arch Psychiatry Clin Neurosci. 2008;258(2):117–23.

    Article  PubMed  Google Scholar 

  62. Buck JA, Teich JL, Miller K. Use of mental health and substance abuse services among high-cost medicaid enrollees. Admin Policy Mental Health Mental Health Serv Res. 2003;31(1):3–14.

    Article  Google Scholar 

  63. Labby D, Wright B, Broffman L, Holtorf M. Drivers of high-cost medical complexity in a medicaid population. Med Care. 2020;58(3):208–15.

    Article  PubMed  Google Scholar 

  64. Gobbicchi C, Verdolini N, Menculini G, Cirimbilli F, Gallucci D, Vieta E, et al. Searching for factors associated with the “Revolving Door phenomenon” in the psychiatric inpatient unit: a 5-year retrospective cohort study. Psychiatry Res. 2021;303: 114080.

    Article  PubMed  Google Scholar 

  65. Small GW, McDonnell DD, Brooks RL, Papadopoulos G. The impact of symptom severity on the cost of Alzheimer’s disease. J Am Geriatr Soc. 2002;50(2):321–7.

    Article  PubMed  Google Scholar 

  66. Quentin W, Riedel-Heller SG, Luppa M, Rudolph A, König H-H. Cost-of-illness studies of dementia: a systematic review focusing on stage dependency of costs. Acta Psychiatr Scand. 2010;121(4):243–59.

    Article  CAS  PubMed  Google Scholar 

  67. Almond S, Knapp M, Francois C, Toumi M, Brugha T. Relapse in schizophrenia: costs, clinical outcomes and quality of life. Br J Psychiatry. 2004;184(4):346–51.

    Article  PubMed  Google Scholar 

  68. Kovács G, Almási T, Millier A, Toumi M, Horváth M, Kóczián K, et al. Direct healthcare cost of schizophrenia—European overview. Eur Psychiatry. 2018;48(1):79–92.

    Article  PubMed  Google Scholar 

  69. Genduso LA, Haley JC. Cost of illness studies for schizophrenia: components, benefits, results, and implications. Am J Manag Care. 1997;3(6):873–7.

    CAS  PubMed  Google Scholar 

  70. Figueroa JF, Frakt AB, Lyon ZM, Zhou X, Jha AK. Characteristics and spending patterns of high cost, non-elderly adults in Massachusetts. Healthc. 2017;5(4):165–70.

    Article  Google Scholar 

  71. Hadley TR, Culhane DP, McGurrin MC. Identifying and tracking “heavy users” of acute psychiatric inpatient services. Admin Policy Mental Health Mental Health Serv Res. 1992;19(4):279–90.

    Article  Google Scholar 

  72. Hensel JM, Taylor VH, Fung K, de Oliveira C, Vigod SN. Unique characteristics of high-cost users of medical care with comorbid mental illness or addiction in a population-based cohort. Psychosomatics. 2018;59(2):135–43.

    Article  PubMed  Google Scholar 

  73. Gobbicchi C, Verdolini N, Menculini G, Cirimbilli F, Gallucci D, Vieta E, et al. Searching for factors associated with the “revolving door phenomenon” in the psychiatric inpatient unit: A 5-year retrospective cohort study. Psychiatry Res. 2021;303: 114080.

    Article  PubMed  Google Scholar 

  74. Addis ME, Mahalik JR. Men, masculinity, and the contexts of help seeking. Am Psychol. 2003;58(1):5.

    Article  PubMed  Google Scholar 

  75. Möller-Leimkühler AM. Barriers to help-seeking by men: a review of sociocultural and clinical literature with particular reference to depression. J Affect Disord. 2002;71(1–3):1–9.

    Article  PubMed  Google Scholar 

  76. Pearson SW, Tillett RIL, Lloyd KR. Study of psychiatric admission patterns before and after out-patient psychotherapy in a district service setting. Psychol Psychother Theory Res Pract. 2002;75(4):437–43.

    Article  Google Scholar 

  77. de Oliveira C, Cheng J, Vigod S, Rehm J, Kurdyak P. Patients with high mental health costs incur over 30 percent more costs than other high-cost patients. Health Aff. 2016;35(1):36–43.

    Article  Google Scholar 

  78. Ronaldson A, Elton L, Jayakumar S, Jieman A, Halvorsrud K, Bhui K. Severe mental illness and health service utilisation for nonpsychiatric medical disorders: a systematic review and meta-analysis. PLoS Med. 2020;17(9): e1003284.

    Article  PubMed  PubMed Central  Google Scholar 

  79. Saunders NR, Gandhi S, Chen S, Vigod S, Fung K, De Souza C, et al. Health care use and costs of children, adolescents, and young adults with somatic symptom and related disorders. JAMA Netw Open. 2020;3(7): e2011295.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

This article was supported by the research fund of Helse Vest regional hospital trust, Norway.

Author information

Authors and Affiliations

  1. Department of Research and Innovation, Helse Førde, Førde, Norway

    Yeujin Ki & Andrew Athan McAleavey

  2. Department of Psychiatry, Helse Førde, Førde, Norway

    Christian Moltu

  3. Department of Health and Caring Sciences, Western Norway University of Applied Science, Bergen, Norway

    Andrew Athan McAleavey & Christian Moltu

  4. Department of Health Management and Health Economics, University of Oslo, Oslo, Norway

    Tron Anders Moger

  5. Section of Medical Statistics, Faculty of Medicine, University of Oslo, Oslo, Norway

    Tron Anders Moger

Authors
  1. Yeujin Ki
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Andrew Athan McAleavey
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Tron Anders Moger
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Christian Moltu
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

YK: Corresponding author, Conceptualization, Data curation; Methodology; Writing—original draft; Writing—review and editing. CM: Conceptualization; Supervision; Writing—review and editing. TAM: Conceptualization; Supervision; Methodology; Formal analysis; Writing—review and editing. AAM: Conceptualization; Co-supervision, Methodology; Writing—review and editing.

Corresponding author

Correspondence to Yeujin Ki.

Ethics declarations

Ethics approval and consent to participate

This study has an approval of the exemption of consent from the Norway Regional Ethical Committee (REK reference: 255248) and has been approved by the data security and data protection officer (PVO) at Førde Hospital Trust. This study performed in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors has no conflicts of interest related with the present study.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Appendices

Appendix 1

See Table 5

Table 5 Detailed cost categories of the KPP according to traditional and activity-based perspectives
Full size table

Appendix 2

See Table 6

Table 6 DRG matching list
Full size table

Appendix 3

See Table 7

Table 7 Comparison with Norwegian national data (SAMDATA, 2021)
Full size table

Appendix 4

See Table 8

Table 8 Distribution of health expenditures for patient in the 1%, 5%, and 10% groups
Full size table

Appendix 5

See Fig. 3

Fig. 3
figure 3

Average treatment cost per LOS (USD)

Full size image

Appendix 6

See Table 9

Table 9 Average treatment cost per DRG for inpatient, outpatients, and total patients
Full size table

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Ki, Y., McAleavey, A.A., Moger, T.A. et al. Cost structure in specialist mental healthcare: what are the main drivers of the most expensive episodes?. Int J Ment Health Syst 17, 37 (2023). https://doi.org/10.1186/s13033-023-00606-6

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13033-023-00606-6

Keywords

International Journal of Mental Health Systems

ISSN: 1752-4458