Skip to main content

Healthcare utilization, costs, and productivity losses in treatment-resistant depression in Finland – a matched cohort study

Abstract

Background

Due to its relatively high prevalence and recurrent nature, depression causes a major burden on healthcare systems, societies and individuals. Our objective was to investigate healthcare resource utilization and costs associated with treatment-resistant depression (TRD) compared with non-treatment-resistant depression in Finland.

Methods

Of all patients aged 16–65 years and diagnosed with depression in Finland during 2004–2016, persons with TRD (N = 15,405) were identified from nationwide registers and matched 1:1 with comparison persons with depression who initiated antidepressant use but did not have TRD at the time of matching. TRD was defined as initiation of a third treatment trial after having failed two pharmacological treatment trials. Follow-up period covered 5 years after TRD or corresponding matching date (until end of 2018). Health care resource utilization was studied with negative binomial regression and costs of TRD (per patient per year) with generalized estimating equations, by adjusting for baseline costs, comorbidity and baseline severity of depression.

Results

Persons with TRD (mean age 38.7, SD 13.1, 60.0% women) had more health care utilization and work disability (sick leaves and disability pensions), adjusted incidence rate ratio for work disability days was 1.72 (95% CI 1.64–1.80). This resulted in 1.9-fold higher total costs for persons with TRD (15,907 versus 8335 EUR), adjusted mean difference 7572 (95% CI 7215–7929) EUR per patient per year, higher productivity losses (due to sick leaves and disability pensions, mean difference 5296, 95% CI 5042–5550), and direct healthcare costs (2003, 95% CI 1853–2151) compared with non-TRD patients. Mean difference was the highest during the first year after TRD (total costs difference 11,760, 95% CI 11,314–12,206) and the difference decreased gradually after that.

Conclusion

Treatment-resistant depression is associated with about two-fold cost burden compared with non-treatment-resistant depression.

Peer Review reports

Introduction

Depressive disorders are one of the leading causes of disability worldwide and have prevailed as a leading cause for decades [1]. Due to a relatively high prevalence, recurrent course and negative impact on work ability, depressive disorders typically result into frequent and long sick leaves, [2] culminating into vast societal and personal cost burdens in the form of permanent work disability [3]. Permanent work disability, in the form of disability pension is also associated with lower quality of life and income [4].

Treatment-resistant depression (TRD) has been suggested to be a major contributor in the societal costs of depressive disorders [5]. The definition of TRD is based on having had two consequential pharmacological treatment trials, with adequate length and dose, to which the patient has not responded or not achieved symptom remission [6]. Some definitions of TRD also include psychotherapy but detailed data on psychotherapy is lacking from many data sources and thus, cannot be included. TRD has been associated with multiple adverse outcomes as compared to regular major depressive disorder, including higher risk of suicides, [7] all-cause mortality, [8, 9] substance use disorders, [10] early labor force exit, [11] and lower quality of life and greater functional and work impairment [12]. Higher risk of disability pension [11, 13] and more frequent healthcare resource utilization in TRD [14,15,16,17, (Brenner et. al.: Health care utilisation in treatment-resistant depression: a Swedish population-based cohort study, unpublished)] lead to a major cost burden both for the society and the individuals. Previous studies on the cost burden of TRD have been conducted mainly in the US, [14, 16, 17] although also some European studies exist [18]. Cost burden and distribution of direct versus indirect costs may vary considerably between different health, work labor and social security systems and thus, more studies are needed on costs associated with TRD.

We have previously reported that the incidence of TRD in Finnish patients is 11% within 2 years of the initiation of first pharmacological treatment [19]. Here we continue to characterize these TRD patients by studying their healthcare resource utilization and costs related to TRD, including direct costs and costs related to productivity loss, in comparison to matched patients with non-treatment-resistant depression during a five-year period. We also assessed the proportion of patients having income from work, as indicative of being employed before and after TRD. This is, to our knowledge, one of the first large scale and long-term studies on the economic burden of TRD in Europe.

Methods

Data sources

Finnish nationwide register-based data was utilized for this study. Registers can be linked through personal identification codes assigned for each resident at birth or immigration (i.e. people with residence permit are included in the registers). The Care Register for the Health Care includes all visits to hospitals (since 1972) and specialized outpatient care (since 1998), including data on admission and discharge dates and recorded discharge diagnoses and operational codes. It does not cover primary care nor private health care, such as occupational health care. For this reason, we utilized also registers on sickness absences and disability pensions. Sickness absences and disability pensions are covered by the Finnish sickness insurance system and thus, recorded in the administrative registers. Sickness absences were collected from the Social Insurance Institution of Finland and disability pensions from the Social Insurance Institution and the Finnish Centre for Pensions. All sick leaves of ≥14 days are registered and paid as a part of the social insurance scheme (regardless of whether person is employed or not). All employees aged 16–67 years of age who are not able to work due to sickness are eligible for sick leave and the same applies also for persons who are not employed. The sickness allowance is on average 70% of the previous earnings and it has no upper limit or ceiling. Sickness absences are recorded regardless of the care setting, i.e. include also those prescribed by primary care and private healthcare. Similarly, disability pensions are a part of the social insurance scheme and recorded from all levels of care. A person who has been disabled for ≥300 working days can apply for disability pension (either permanent or a fixed-term disability pension, so called rehabilitation subsidy) regardless of their employment status (i.e. also unemployed or otherwise economically inactive can apply). In the registers, sickness absences and disability pensions include beginning and end dates and diagnoses. The Finnish National Prescription Register includes data on all reimbursed prescription drug purchases from Finnish pharmacies for all residents regardless of the care setting or the prescriber. Drugs provided during hospital care are not recorded but prescriptions stated by physicians working in the hospitals are record. The data includes information on date of dispensing, the Anatomic Therapeutic Chemical (ATC) code, strength, package size, the number of packages dispensed and dispensed amount in Defined Daily Doses (DDDs).”

Study population

We identified all persons diagnosed with depression (ICD-10 F32-F33) from nationwide Finnish registers, including inpatient and specialized outpatient care (The Care Register for the Health Care), sickness absence (all sick leaves > 2 weeks) and granted disability pensions (through registers of Social Insurance Institution and Finnish Centre for Pensions). Sick leaves and disability pensions included also those stated by primary care and occupational health care specialists, covering a wide range of patients treated on different levels of care. For each person, the first diagnosis of depression in the age of 16–65 years during years 2004–2016 was taken. Exclusions were made due to bipolar disorder (F30-F31), schizophrenia-spectrum disorder (F20-F29) and dementia (G30, F00-F03) when these conditions were recorded before the first depression diagnosis.

As the definition of TRD is based on medication use (supplemented with specialized health care information on neuromodulation treatments), antidepressant and other psychotropic medication use was derived from the Prescription register data, which includes reimbursed dispensing of prescribed medications, between the years 2003–2018. Antidepressants were identified as ATC (Anatomical Therapeutic Chemical)-code N06A. For each person, we identified the first antidepressant dispensing around the first depression diagnosis, and this first dispensing was assigned as the index date for TRD definition. First depression diagnosis needed to be registered within a time interval of 30 days before and up to 365 days after the index date. Inclusion of new users was conducted by inserting a 180 day washout period before the index date, meaning that persons using an antidepressant or defined potential add-on medications for depression (antipsychotics, lithium, lamotrigine, valproic acid or carbamazepine), or who were administered ECT (electroconvulsive therapy), rTMS (repetitive transcranial magnetic stimulation), tDCS (transcranial direct current stimulation) or ketamine infusion as registered in the inpatient or specialized outpatient care register during the washout period, were excluded. In Finland, these treatments are provided only in specialized health care settings. These new antidepressant users, with a registered diagnosis of depression formed the base cohort for this study (N = 177,144).

Medication dispensing data from the Prescription register was transformed to drug use periods with ‘From drug purchases to drug use periods’ (PRE2DUP) method. The method estimates when drug use started and ended based on dispensing dates, amounts dispensed and drug-specific parameters defining lower limit for daily dose in continuous use, and by taking into account on stockpiling, days spend in hospital care and personal regularity of use [20].

TRD criteria

TRD criteria were defined to be met when initiation of a third sequential treatment trial for depression occurred, after two preceding adequate trials (including the index trial and a second trial). Adequate treatment trial was defined as lasting for at least 28 days according to PRE2DUP modelled drug use periods. After first treatment trial with antidepressant, second and third treatment trials could consist of either a new antidepressant treatment, antipsychotic (quetiapine ≥100 mg, risperidone, olanzapine, aripiprazole) or mood stabilizer (lithium, valproic acid, carbamazepine, lamotrigine), or series of ECT, rTMS, tDCS or ketamine infusion (referred to as “ECT”, since only ECT was observed in our dataset before TRD criteria were fulfilled). Within drug-class changes were also considered new trials. Fulfillment of TRD criteria were followed for 2 years since index date, with maximum allowed gap (non-use) being set to 56 days.

Matching of comparators

We excluded persons who were already on disability pension (DP) at the date of TRD to observe persons in working life (and at risk of productivity loss). At the date when TRD criteria were met and within the cohort of persons with depression, each patient with TRD was matched with one comparator who did not have TRD or disability pension, had first or second antidepressant trial ongoing, and had similar age (±2 years), the same gender, calendar year of index date, and hospital district (eFig. 1). Person who later was defined as TRD could serve as a comparator until the date when TRD criteria were met. No matching comparator could be found for N = 1248 TRD patients who were excluded. These excluded TRD patients were somewhat younger, more often male and had longer follow-up time than included TRD patients (Supplementary Table 1). For the final study, N = 15,405 patients with TRD and N = 15,405 matched comparators were included.

Outcomes

Persons with TRD and their comparators were followed from TRD date (corresponding matching date for comparators) for 5 years, but censoring for death, diagnosis of bipolar disorder/ schizophrenia-spectrum disorder, fulfillment of TRD criteria for comparators and end of data linkage (December 31, 2018). Healthcare resource utilization was defined as all-cause inpatient hospital days and specialized outpatient care visits. For cost analyses, hospital day and outpatient visit costs were calculated according to Finnish healthcare system unit costs which have been derived for research purposes and adjusted for regional price differences [21]. Drug costs were calculated from costs of psychotropic medication dispensings as total costs recorded in the Prescription register data (mood stabilizers N03A, antipsychotics N05A, benzodiazepines and related drugs N05BA, N05CD and N05CF, and antidepressants N06A, other drug costs were not available). Psychotherapy costs covered costs paid by Social Insurance Institution, of state-funded psychotherapies and based on direct reimbursements paid to the healthcare provider. Hospital care, outpatient visits, medications and psychotherapies formed direct costs, whereas productivity losses covered work disability days, derived as days of sickness absence (SA) and/or disability pension (DP). Productivity losses related costs were calculated as SA and DP days multiplied by the average gross monthly salary across all sectors [22]. All costs were real-valued to Euros (€) in 2020 with the price index of public expenditure [23]. Costs were calculated for each consecutive year during 5 years, as per patient per year (PPY), by censoring persons not contributing to the study anymore.

Being employed was measured as having any income from work (yes/no), as derived from the Earnings- and Accrual register, maintained by the Finnish Centre for Pensions. The register is compiled based on calendar years, and year of TRD date was assigned as year zero. Development of the proportion of the cohort who were employed was followed for 4 years before and 4 years after year zero. The longitudinal association between TRD status and employment was assessed in two parts, during the years before and after TRD, year zero being included in the latter part.

Analyses

Descriptive statistics were calculated with proportions, means with 95% confidence intervals and compared with chi squared tests and t-tests with parametric bootstrap procedure. Healthcare resource utilization and indirect resource utilization were analyzed by negative binomial regression models and reported as adjusted Incidence Rate Ratios (IRR) with 95% confidence intervals, with matched non-TRD patients as a reference. The models took into account follow-up time and were adjusted for baseline costs (defined as total costs during 1 year before index antidepressant dispensing), index period costs (defined as total costs during time period between index antidepressant dispensing and TRD date/ corresponding matching date, varying between matched pairs from 2 months up to 2 years but being equal within the matched pair), Charlson’s comorbidity index, and baseline severity of depression. Charlson’s comorbidity score [24] was calculated based on in- and specialized outpatient care diagnoses recorded during the last 5 years before the index date. Score of zero is assigned to persons who do not have any of the comorbidities recorded. Severity of depression at first diagnosis was categorized as mild (ICD-10: F32.0, F32.8, F32.9, F33.0, F33.8, F33.9), moderate (ICD-10: F32.1, F33.1), severe (ICD-10: F32.2– F32.3, F33.2–F33.3), or unknown (for persons who had their diagnosis recorded only with three characters in the register source used and for whom severity level could not be determined).

Costs were compared with Generalized Estimating Equations (GEE) models, specifying Poisson distribution and log link, between TRD and comparators. Models were constructed separately for total costs, direct healthcare costs and productivity losses. Cost analyses were adjusted for baseline and index period total costs, Charlson’s Comorbidity Index, and baseline severity of depression. Costs are reported as adjusted means for persons with TRD, compared with non-TRD comparators, as mean costs per patient per year (PPY). Costs for years 1–5 after TRD were analyzed similarly as adjusted GEE models with year interaction and expressed as adjusted mean costs with 95% confidence intervals. Subgroup analyses were conducted among patients with severe depression recorded before TRD date/ matching date, by additionally adjusting for matching factors (as matched design was not retained, i.e. age, gender, calendar year and hospital district). Sensitivity analyses for the main analyses were conducted by censoring both members of a TRD/ non-TRD matched pair when one member of the pair was censored, resulting into equal follow-up times. Finally, we conducted sensitivity analyses by matching non-TRD comparators (1:1) to those patients who were defined as TRD during 2 years follow-up at initiation of index antidepressant and followed them 5 years beginning from index antidepressant.

GEE logistic regression model was applied to employment analyses, by adjusting for Charlson’s Comorbidity Index and baseline severity of depression. Unstructured correlation option was used due to the unbalanced nature of data, where imbalance was caused by the varying number of persons included in each timepoint due to exclusion/ censoring. Data management was conducted with SAS version 9.4 (SAS Institute, Inc., Cary, NC, USA) and statistical analyses with STATA 17.0.

Results

Persons with TRD were of mean 38.7 years old (SD 13.1) and 60.0% were women (Table 1). Comparators with non-TRD depression who had their first or second antidepressant trial were similar to persons with TRD in terms of matching factors, but had somewhat milder initial severity of depression (e.g. 11.0% had severe depression compared with 17.5% of persons with TRD). Baseline costs, measured as 1 year before index antidepressant dispensing, were relatively similar between persons with TRD and comparators, and some costs, such as hospital care costs were higher among comparators whereas persons with TRD had somewhat more SA days and associated costs. During the index period (mean duration 8 months), persons with TRD had higher costs than comparators.

Table 1 Baseline sociodemographic and clinical characteristics of the matched groups, and costs and resource utilization at baseline (one year before index antidepressant) and during index period (between index antidepressant and treatment-resistant depression, TRD or corresponding matching date for non-TRD persons). Comparators matched by age, sex, hospital district and calendar year of index date

During the 5 years follow-up, persons with TRD had higher health care resource utilization and higher risk for work disability days than comparators (Figure 1), e.g. in outpatient visits (adjusted IRR 1.66, 1.61–1.71) and work disability days (IRR 1.72, 95% CI 1.64–1.80). This resulted in significant adjusted mean cost differences, which were 7572 EUR (95% CI 7215–7929) for total costs, 2003 EUR (1853–2151) for direct costs and 5296 EUR (5042–5550) per patient per year for productivity losses (Fig. 2A). Adjusted mean costs of TRD patients were 1.9-fold, namely 15,907 EUR (95% CI 15646–16,169) per patient per year, compared with 8335 EUR (95% CI 8107–8563) in comparators. Mean cost differences were in similar range in sensitivity analyses with equal censoring for TRD pair members (Supplementary Table 2).

Fig. 1
figure 1

Health care resource utilization and indirect resource utilization measures during the 5 years follow-up. Adjusted Incidence Rate Ratios (IRR) comparing persons with treatment resistant depression (TRD) to matched non-TRD persons. Adjusted for: baseline and index period total health care costs, Charlson’s Comorbidity Index, follow-up time and baseline severity of depression

Fig. 2
figure 2

a Adjusted mean total costs, and costs divided into direct costs and productivity losses per patient per year (PPY) for treatment-resistant depression (TRD) and non-TRD comparators. Adjusted for: baseline and index period total health care costs, Charlson’s Comorbidity Index, and baseline severity of depression. b. Adjusted mean direct costs and productivity losses 1–5 years after treatment-resistant depression (TRD) for persons with TRD and non-TRD as mean of per patient per year (PPY). Adjusted for: baseline and index period total health care costs, Charlson’s Comorbidity Index, and baseline severity of depression

When stratified by years since TRD, the largest difference in adjusted mean total costs was seen during the first year (total costs difference 11,760 EUR, 95% CI 11,314–12,206) and the difference decreased gradually after that (Fig. 2B, Table 2). For the subgroup analyses among those who had been diagnosed with severe depression before matching (N = 3894 TRD vs N = 1877 non-TRD), the adjusted cost difference was 9206 EUR (8257–1055) per patient per year, mean cost 20,280 (19,680–20,880) for TRD (1.8-fold) and 11,008 EUR (10,291–11,724) for non-TRD (Supplementary Table 3). When matching and start of follow-up was conducted at index antidepressant initiation the mean cost difference between TRD and non-TRD was somewhat larger and the cost levels for both groups were higher than in the main analyses (difference 15,158 EUR, 95% CI 14,368–15,948 per patient per year) (Supplementary Table 4).

Table 2 Adjusted mean costs with 95% confidence intervals (CIs) for treatment-resistant depression (TRD) compared with non-TRD and their difference 1–5 years after TRD. Adjusted for: baseline and index period total health care costs, Charlson’s Comorbidity Index, and baseline severity of depression

Proportion being employed was rather similar 4 years before the matching date between persons with TRD and comparators (73.1% for TRD vs. 74.8% for comparators, Fig. 3). During 4 years before TRD, persons with TRD had slightly lower odds of being employed than comparators (adjusted OR 0.93, 95% CI 0.89–0.97). During the year of TRD, 67.9% of TRD and 74.1% of comparators were employed and the difference increased thereafter. Four years after TRD, 56.5% of persons with TRD and 70.0% of comparators were employed. During the 4 years after TRD, TRD was associated with 36% lower odds of employment than comparators (adjusted OR 0.64, 95% CI 0.62–0.66).

Fig. 3
figure 3

Development of proportion of persons having any income from work as a marker of being employed 4 years before and 4 years after treatment-resistant depression (TRD) in persons with TRD compared with non-TRD persons

Discussion

We found TRD to be associated with markedly increased healthcare utilization and higher productivity losses as compared with matched comparison persons without TRD. Both direct healthcare costs and productivity losses were higher in TRD, although the difference was larger for productivity losses, caused by sick leaves and disability pensions. The difference was largest during the first year after TRD and decreased during the 5 years of follow-up, still remaining significant after 5 years. TRD was also associated with decreased odds of being employed.

Higher total costs in persons with TRD are consistent with previous studies which all show a higher cost burden for TRD [14,15,16,17,18]. Some studies have reported equally large differences in direct and indirect costs, [16] others have found a higher difference in direct costs than in indirect costs [18] for TRD compared with non-treatment-resistant patients with depression. In the present study, the cost difference was larger in productivity losses (about 2.1 times higher) than in direct costs (1.8 times higher), and in both TRD and non-TRD, the majority of costs were from productivity losses. These differences between studies likely represent underlying differences in the healthcare, labor market and social security systems between different countries. For example, the direct cost estimates from the US [16] show a higher overall cost level for both TRD and comparators with MDD than in our study, but reported proportional cost differences between TRD and non-TRD are rather similar in the previous studies as in the present study. A recent Swedish study found larger differences between TRD and non-TRD for healthcare utilization, namely hospital days and outpatient visits, than found in our study (Brenner et. al.: Health care utilisation in treatment-resistant depression: a Swedish population-based cohort study, unpublished). Although healthcare systems in Finland and Sweden are rather similar, differences between the studies may stem from the study populations, as we included also persons treated outside of specialized health care (with > 2 weeks sick leave due to depression) whereas the Swedish study included only persons treated in specialized health care.

The follow-up time in our study was 5 years, which is longer than in several previous studies of TRD cost burden, restricting to 2 years [15,16,17] or less, [18] and they have also started follow-up at index antidepressant prescription whereas our follow-up started at the date when TRD definition was fulfilled (third treatment trial initiation). Our sensitivity analyses starting the follow-up from index antidepressant prescription resulted into a somewhat larger cost difference. This can be at least partly related to the fact that in this design, non-TRD group were selected based on not being defined as TRD during the 2 years definition period whereas in our main analyses, they could serve as comparators until their TRD criteria fulfilment. Both differences in direct costs and productivity losses between TRD and comparator non-TRD patients diminished in 5 years’ time, but especially the difference in productivity losses persisted, being almost two times higher for TRD than for non-TRD also after 5 years. The persisting difference is concerning and highlights the far-reaching, long-term impact of TRD.

Our study also showed a marked difference in employment as persons with TRD were associated with 36% lower odds of being employed than their comparators during 4 years follow-up after TRD. This is in line with previous studies reporting two times higher risk for disability pension, [13] and three-fold larger risk of premature workforce exit [11] and thus, showing a considerably decreased ability to attend working life. Besides high cost burden for the society and additional costs for lost tax payments and lost productivity, decreased participation in working life will also result into lifelong individual economic consequences as retirement-pension is earnings based and lower employment rate will result into a lower retirement-pension.

Subgroup analyses among persons diagnosed with severe form of depression were in line with main analyses regarding higher costs (1.8-fold) associated with TRD. Further, the level of costs associated with severe depression with TRD (total cost 20,280 EUR per patient per year) was higher as compared to the main study cohort for all TRD patients (15,907 EUR) indicating that severity of depression plays an important role in defining high costs. However, TRD was associated still with about two times higher total costs compared with non-TRD comparators which shows that association between TRD and increased costs is not explained by the higher severity of depression. In our previous study, we found that even the fifth treatment line for TRD patients is most commonly antidepressant monotherapy [19]. This is not guideline-driven optimal practice and might be one of the reasons underlying the higher costs in TRD as patients are not receiving the most efficacious treatments available and thus, will use healthcare resources longer and are at an increased risk for permanent work disability which is the main cost driver in TRD. We did not assess the impact of specific comorbidities but adjusted for Charlson comorbidity index. However, a previous study has shown that persons with TRD have more often pre-existing general medical conditions and also subsequent diseases after TRD than their comparators, [25] implying that there may be bidirectional association between TRD and comorbid conditions.

The strengths of this study include nationwide data on persons with TRD and matched comparators with depression. The study cohort included both persons treated in primary and specialized health care during a time period of over 15 years. All residents in Finland are entitled to social and health care services and data on the use of these services is routinely collected. The results of this study are generalizable to healthcare and social security systems resembling that of Finland, i.e. most closely to Nordic countries. Distribution and levels of costs likely differ between different healthcare systems, modified by different access to care and access to social security. Although we comprehensively collected information on all major depression-related health care services, such as inpatient care, specialized outpatient care, medication use and psychotherapies received, some underlying limitations exists. Primary health care utilization data was not available for the study, and some people may have received psychotherapy outside of the state-funded scheme, by covering the costs themselves. We also did not have data on possible failures of psychotherapy and thus, psychotherapies could not be evaluated as possible treatments and treatment lines. In addition, medication costs for the study cohort were only available for psychotropic medications, which limits data describing potential comorbid conditions. However, data on inpatient and specialized outpatient care treatment of these other conditions was included. Productivity losses were based on average gross monthly salary across all sectors and do not take into account possible part-time work. Only persons with residence permit were included in this study and institutionalized persons (e.g. in prisons) were missing.

Baseline severity of depression was adjusted for in all analyses and we conducted subgroup analyses among persons who had received a diagnosis of severe depression by the time of TRD or corresponding matching date. However, the severity level was not available for the entire cohort, due to lack of reporting it (possibly due to technical reasons) in some years especially in the sickness absence register. Definition of TRD as initiation of third treatment trial after two failed trials is similar as in previous studies [7, 10, 13, 26] although the clinical validity of this definition has not been assessed. Data sources did not include data on response to treatment and adherence could only be estimated through dispensing information (i.e. discontinuation of use). Different prescribers may have had different thresholds for prescribing and switching antidepressants although they are expected to follow national guidelines on treatment of depression. Sick leaves and disability pensions are regulated by the social insurance system which harmonizes the decisions between prescribers. In addition, although TRD incurred high costs as compared to non-TRD, these costs are likely underestimates, since we were not able to take into account life-years lost or the effect of work absences, disability and early pension on gross domestic product. The costs related to these phenomena are likely to be much higher than the costs presented here and make the gap between TRD and non-TRD even more pronounced.

Conclusions

Compared with non-TRD, TRD is associated with about 2-fold cost burden and increased direct costs and productivity losses up to 5 years after TRD. TRD was also associated with reduced likelihood of being employed. These results present persons with TRD as a patient group which would likely benefit from intensified monitoring and multidisciplinary efforts aiming for gaining functional and symptomatic remission and recovery.

Availability of data and materials

The data that support the findings of this study are available from the Finnish government agencies National Institute of Health and Welfare, Social Insurance Institution and Finnish Centre of Pensions, but restrictions apply to the availability of these data, which were used under license for the current study and so are not publicly available. Data are however available from the authors upon reasonable request and with permission of National Institute of Health and Welfare, Social Insurance Institution and Finnish Centre of Pensions.

References

  1. James SL, Abate D, Abate KH, Abay SM, Abbafati C, Abbasi N, et al. Global, regional, and national incidence, prevalence, and years lived with disability for 354 diseases and injuries for 195 countries and territories, 1990-2017: a systematic analysis for the global burden of disease study 2017. Lancet. 2018;392:1789–858.

    Article  Google Scholar 

  2. Blomgren J, Perhoniemi R. Increase in sickness absence due to mental disorders in Finland: trends by gender, age and diagnostic group in 2005–2019. Scand J Public Health. 2021;2020:1–5.

    Google Scholar 

  3. Amiri S, Behnezhad S. Depression and risk of disability pension: a systematic review and meta-analysis. Int J Psychiatry Med. 2019. https://doi.org/10.1177/0091217419837412.

  4. Stattin M. Retirement on grounds of ill health. Occup Environ Med. 2005;62:135–40.

    Article  CAS  Google Scholar 

  5. Greden JF. The burden of disease for treatment-resistant depression. J Clin Psychiatry. 2001;62(SUPPL. 16):26–31.

    PubMed  Google Scholar 

  6. Conway CR, George MS, Sackeim HA. Toward an evidence-based, operational definition of treatment-resistant depression: when enough is enough. JAMA Psychiatry. 2017;74:9–10.

    Article  Google Scholar 

  7. Reutfors J, Andersson TML, Brenner P, Brandt L, DiBernardo A, Li G, et al. Mortality in treatment-resistant unipolar depression: a register-based cohort study in Sweden. J Affect Disord. 2018;238:674–9.

    Article  Google Scholar 

  8. Li G, Fife D, Wang G, Sheehan JJ, Bodén R, Brandt L, et al. All-cause mortality in patients with treatment-resistant depression: a cohort study in the US population. Ann General Psychiatry. 2019;18:1–8.

    Article  Google Scholar 

  9. Madsen KB, Plana-Ripoll O, Musliner KL, Debost JCP, Petersen LV, Munk-Olsen T. Cause-specific life years lost in individuals with treatment-resistant depression: a Danish nationwide register-based cohort study. J Affect Disord. 2021;280:250–7.

    Article  Google Scholar 

  10. Brenner P, Brandt L, Li G, DiBernardo A, Bodén R, Reutfors J. Treatment-resistant depression as risk factor for substance use disorders—a nation-wide register-based cohort study. Addiction. 2019;114:1274–82.

    Article  Google Scholar 

  11. Bang Madsen K, Vogdrup Petersen L, Plana-Ripoll O, Musliner KL, Philippe Debost J-C, Hordam Gronemann F, et al. Early labor force exits in patients with treatment-resistant depression: an assessment of work years lost in a Danish nationwide register-based cohort study. Ther Adv Psychopharmacol. 2020;10:204512532097379.

    Article  Google Scholar 

  12. Dibernardo A, Lin X, Zhang Q, Xiang J, Lu L, Jamieson C, et al. Humanistic outcomes in treatment resistant depression: a secondary analysis of the STARD study. BMC Psychiatry. 2018;18:1–8.

    Article  Google Scholar 

  13. Taipale H, Reutfors J, Tanskanen A, Brandt L, Tiihonen J, Dibernardo A, et al. Risk and risk factors for disability pension among patients with treatment resistant depression- a matched cohort study. BMC Psychiatry. 2020;20:1–11.

    Article  Google Scholar 

  14. Pilon D, Joshi K, Sheehan JJ, Zichlin ML, Zuckerman P, Lefebvre P, et al. Burden of treatment-resistant depression in medicare: a retrospective claims database analysis. Plos One. 2019;14:1–16.

    Article  Google Scholar 

  15. McIntyre RS, Millson B, Power GS. Burden of treatment resistant depression (TRD) in patients with major depressive disorder in Ontario using Institute for Clinical Evaluative Sciences (ICES) databases: economic burden and healthcare resource utilization. J Affect Disord. 2020;277 December 2019:30–8.

    Article  Google Scholar 

  16. Amos TB, Tandon N, Lefebvre P, Pilon D, Kamstra RL, Pivneva I, et al. Direct and indirect cost burden and change of employment status in treatment-resistant depression. J Clin Psychiatry. 2018. https://doi.org/10.4088/jcp.17m11725.

  17. Li G, Zhang L, DiBernardo A, Wang G, Sheehan JJ, Lee K, et al. A retrospective analysis to estimate the healthcare resource utilization and cost associated with treatment-resistant depression in commercially insured US patients. Plos One. 2020;15 9 September:1–11.

    CAS  Google Scholar 

  18. Pérez-Sola V, Roca M, Alonso J, Gabilondo A, Hernando T, Sicras-Mainar A, et al. Economic impact of treatment-resistant depression: a retrospective observational study. J Affect Disord. 2021;295:578–86.

    Article  Google Scholar 

  19. Lähteenvuo M, Taipale H, Tanskanen A, Rannanpää S, Tiihonen J. Courses of treatment and risk factors for treatment-resistant depression in Finnish primary and special healthcare: a nationwide cohort study. J Affect Disord. 2022;308:236–42.

    Article  Google Scholar 

  20. Tanskanen A, Taipale H, Koponen M, Tolppanen A-M, Hartikainen S, Ahonen R, et al. From prescription drug purchases to drug use periods – a second generation method (PRE2DUP). BMC Med Inform Decis Mak. 2015;15:21.

    Article  Google Scholar 

  21. Kapiainen S, Väisänen A, Haula T. Terveyden- ja sosiaalihuollon yksikkökustannukset Suomessa vuonna; 2011. p. 2014.

    Google Scholar 

  22. StatisticsFinland. Official Statistics of Finland (OSF): Index of wage and salary earnings. http://www.stat.fi/til/ati/index_en.html. Accessed 25 Oct 2021.

  23. StatisticsFinland. Official Statistics of Finland (OSF): Price index of public expenditure. https://stat.fi/til/jmhi/index_en.html?/til/jmhi/index_en.html. Accessed 25 Oct 2021.

  24. Charlson ME, Pompei P, Ales KL, MacKenzie CR. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis. 1987;40:373–83.

    Article  CAS  Google Scholar 

  25. Madsen KB, Momen NC, Petersen LV, Plana-Ripoll O, Haarman BCM, Drexhage H, et al. Bidirectional associations between treatment-resistant depression and general medical conditions. Eur Neuropsychopharmacol. 2021;51:7–19.

    Article  CAS  Google Scholar 

  26. Hägg D, Brenner P, Reutfors J, Li G, DiBernardo A, Bodén R, et al. A register-based approach to identifying treatment-resistant depression-comparison with clinical definitions. Plos One. 2020;15:e0236434.

    Article  Google Scholar 

Download references

Acknowledgements

NA.

Funding

Open access funding provided by Karolinska Institute. This study was funded by Janssen-Cilag Finland and the Finnish Ministry of Social Affairs and Health through the developmental fund for Niuvanniemi Hospital. ML was partly funded by personal grants from the Finnish Medical Foundation and Emil Aaltonen foundation. Janssen Cilag had no role in data collection and the statistical analysis. Other funding agencies had no role in the design of the study, in data collection, analysis, and interpretation of data, or in writing the manuscript.

Author information

Authors and Affiliations

Authors

Contributions

Concept: HT, ML, AT, SR, JT. Analytic design: HT, AT, ML, JT. Data analysis: HT, AT. Data verification: HT, AT. First draft: HT. Interpretation of results: all authors. Decision to submit the manuscript: all authors. The author(s) read and approved the final manuscript.

Corresponding author

Correspondence to Heidi Taipale.

Ethics declarations

Ethics approval and consent to participate

The research project was approved by pertinent institutional authorities at the Finnish National Institute for Health and Welfare (permission 635/5.05.00/2019), the Social Insurance Institution of Finland (31/522/2019), Finnish Centre for Pensions (19023) and Statistics Finland (TK-53-569-19). Informed consent was not sought due to the registry-based nature of the study as data. According to Finnish legislation, no informed consent is required when administrative data in governmental registers already being collected is used and data is anonymized so no specific individual could be identified (the Act on the secondary use of health and social data). Study persons were not contacted and no experiments for them were made. Only researchers who were granted with administrative permission from National Institute of Health and Welfare, Social Insurance Institution and Finnish Centre of Pensions had access to raw data. All methods were carried out in accordance with Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

JT, HT and AT have participated in research projects funded by grants from Janssen Cilag and Eli Lilly to their employing institution. HT reports personal fees from Janssen Cilag and Otsuka. JT reports personal fees from Eli Lilly, Janssen Cilag, Lundbeck, and Otsuka; is a member of advisory board for Lundbeck. ML is a board member of Genomi Solutions ltd. and Nursie Health ltd., has received honoraria from Sunovion ltd., Orion Pharma ltd., Lunbdbeck ltd., Otsuka Pharma ltd., and Janssen Cilag and research funding from Janssen Cilag, the Finnish Medical Foundation, the Finnish Cultural Foundation and the Emil Aaltonen Foundation. SR and SH are employees of Janssen Cilag.

Additional information

Publisher’s Note

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

Supplementary Information

Additional file 1: eFigure 1.

Formation of matched design. AD: antidepressant (or other pharmacological treatment for depression, or ECT), TRD: treatment-resistant depression. Supplementary Table 1. Comparison between TRD cases with and without comparison persons, i.e. TRD cases included and excluded from this study. Supplementary Table 2. Adjusted mean costs per patient per year with 95% confidence intervals (CIs) for treatment-resistant depression (TRD) compared with non-TRD and their difference 1–5 years after TRD when the follow-up time for the TRD/ non-TRD matched pair was set to equal (censoring both members of the pair when the first member was censored due to any reason). Adjusted for: baseline and index period total health care costs, Charlson’s Comorbidity Index, and baseline severity of depression. Generalized Estimating Equations (GEE) model with gamma distribution and log link. Supplementary Table 3. Adjusted mean costs with 95% confidence intervals (CIs) for treatment-resistant depression (TRD) compared with non-TRD and their difference in subgroup of patients with severe depression. Adjusted for: baseline and index period total health care costs, Charlson’s Comorbidity Index, age, gender, hospital district and calendar year. Supplementary Table 4. Adjusted mean costs per patient per year with 95% confidence intervals (CIs) for treatment-resistant depression (TRD) compared with non-TRD and their difference 1–5 years after TRD in separate study design where the matching was conducted at index antidepressant initiation and 5-year follow-up started from there (instead of starting from TRD/ matching date). Adjusted for: baseline total health care costs, Charlson’s Comorbidity Index, and baseline severity of depression. Generalized Estimating Equations (GEE) model with gamma distribution and log link.

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

Taipale, H., Lähteenvuo, M., Tanskanen, A. et al. Healthcare utilization, costs, and productivity losses in treatment-resistant depression in Finland – a matched cohort study. BMC Psychiatry 22, 484 (2022). https://doi.org/10.1186/s12888-022-04115-7

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12888-022-04115-7

Keywords