Skip to main content

The allocation of resources in the care for patients with panic disorder in Germany: an excess cost analysis informing policy and science

Abstract

Background

Panic disorder is a mental disorder of high prevalence, which frequently co-occurs with agoraphobia. There is a lack of studies measuring excess costs of panic disorder patients with and without agoraphobia. We compared costs of panic disorder patients with or without agoraphobia with costs of the anxiety-free population in Germany.

Methods

Primary data from a cluster-randomized trial of adults suffering from panic disorder (n = 419) and from a representative survey of the German general population (N = 5005) were collected between 2012 and 2014. Missing data from the cluster-randomized trial were first imputed by multiple imputation using chained equations and subsequently balanced with the data from the survey by Entropy Balancing. The societal perspective was chosen. Excess costs were calculated by generalized linear models and two-part-models.

Results

Entropy Balancing led to an exact match between the groups. We found 6-month total excess costs of 3220€ (95% CI 1917€–4522€) for panic disorder patients without agoraphobia and of 3943€ (95% CI 2950€–4936€) for patient with agoraphobia. Panic disorder patients with or without agoraphobia had significantly higher costs for psychotherapy, general practitioners, general hospital stays and informal care Indirect costs accounted for approximately 60% of the total excess costs.

Conclusions

Panic disorder with or without agoraphobia is associated with significant excess costs. Agoraphobia changes the pattern of resource utilization. Especially indirect costs are relevant. Agoraphobia influences resource utilization in the inpatient sector.

Trial registration ISRCTN64669297

Introduction

Panic disorder, defined as recurrent, unpredictable panic attacks associated with feelings of unreality, chest pain, palpitations, choking sensations, and dizziness [1], poses a major challenge to the health care system. According to the representative DEGS1-MH study, in Germany the 12-month prevalence of panic disorder (with or without agoraphobia) is approximately 2.0% in the general adult population, with women being affected twice as often as men (prevalence rate of 2.8% vs. 1.2%) [2]. This study considered the German adult population aged 18 to 79. Based on data of the German Statistical Office, this age group consisted of around 64 million people in 2016 [3]. This means that around 1.28 million people in Germany suffered from panic disorder (with or without agoraphobia). Comorbidity among anxiety disorders is very common, with the highest rates found in panic disorder [4]. Agoraphobia is a frequent comorbidity in panic disorder patients. There is evidence that approximately 35% to 65% of the panic disorder patients are affected by agoraphobia [5].

The evidence highlights the challenge posed by frequency. Further evidence underlines the challenge posed by health burden. A study by Saarni et al. [6] showed that health-related quality of life was decreased for panic disorder patients as well as for patients with agoraphobia compared to individuals without a mental disorder [based on a composite international diagnostic interview (CIDI)]. Additionally, Beard et al. [7] reported that panic disorder patients with agoraphobia had lower physical and mental component scores in the Short-Form 36 compared to panic disorder patients without agoraphobia. Both studies indicate that the health burden of panic disorder with and without agoraphobia is high.

In addition, several studies emphasized the challenges to health care delivery and financing. Prospective longitudinal studies have shown that misinterpretation of bodily sensations and health anxiety are important factors which might contribute to elevated help-seeking in the health care system [8, 9]. Furthermore, the economic burden of panic disorder has been addressed by several international studies [10]. However, only one study from the Netherlands analysed the economic burden of panic disorder with a special view on agoraphobia [11]. The authors of this study differentiated between full-blown and subthreshold panic disorder with or without agoraphobia. They found that in case of a panic disorder agoraphobia raises the annual costs from around €12,000 to €15,000, while in case of subthreshold panic disorder the increase is even more pronounced, from around €6700 to nearly €17,000. Based on these results, we expect that in our study patients with a comorbid agoraphobia will have higher costs, too. However, due to methodological differences, we abstain from deriving assumptions about the amount of the difference and will not be able to compare the findings of Batelaan et al. to our results.

The special need to address these challenges has been recognized by German researchers and disease-specific interventions have been developed, implemented and assessed, e.g. the Jena-Paradise study (ISRCTN64669297). However, there is no scientific evidence regarding the impact of panic disorder (with or without agoraphobia) on health care delivery and costs in Germany. This information is pivotal to answer different questions, like “In which sectors are resource mainly utilized?”, “Which sectors need to be addressed by political and medical interventions?” and “Has an intervention been successful in addressing these sectors?”.

In this study, we will present this pivotal information by performing an excess cost analysis of patients with panic disorder (with or without agoraphobia) in comparison to a representative sample of the German adult population. By doing this, we will inform decision-makers on the allocation of health care resources, give them information to govern this allocation, support researchers in the development of future interventions and enable the interpretation of the impact of these interventions on the delivery of health care services.

Methods

Study population

Two samples are needed to estimate excess costs. The first sample consists of patients who suffer from the disorder of interest. The second sample is a population not affected by the disorder. In any other characteristic, the samples must be comparable. For this reason, we balanced the data of patients diagnosed with PD or PDA and data of individuals without anxiety disorders.

Patient data were collected in a cluster-randomized controlled trial including patients with panic disorder (PD) or panic disorder and agoraphobia (PDA) [12] (Current Controlled Trials: ISCRTN64669297). Patients had to be at least 18 years, diagnosed with PD (ICD-10; F41.0) or PDA (ICD-10; F40.01) by a GP-led clinical interview, scored at least 8 points on the Overall Anxiety Severity and Impairment Scale (OASIS) [13], and had a minimum of two positive answers on the panic module of the ‘Patient Health Questionnaire’ (PHQ) [14, 15]. Patients were excluded if they had one of the following conditions: suicidal tendencies, psychotic or substance-related disorders, severe physical illness, pregnancy, or if they were currently under psychotherapeutic treatment. In order to estimate excess costs, we used data from the baseline assessment, which included information about resource utilisation in the last 6 months before the assessment, sociodemographic characteristics and information on comorbidity. 419 patients were included in the study, 315 (75%) were diagnosed with PDA and 104 (25%) with PD. To handle missing values (44.6% of all patients had at least one missing value), we applied Multiple Imputation by Chained Equations [16]. In total 40 datasets (m = 40) [17] were generated by predicted mean matching [18] using sociodemographic characteristics, comorbidities, disease-specific measures, and health care utilisation as covariates in the imputation models.

The dataset used for comparison came from a telephone survey [19]. The sample is representative of the total German-speaking population aged 18 years and older. The interview included questions regarding health care utilisation in the last 6 months, the disease history (see Table 2), the PHQ-4 [20, 21] as a screening tool for anxiety disorders or depression, the ISR-S [21, 22] as a screening tool for somatoform disorders as well as sociodemographic questions. In total, 5005 individuals completed the telephone interview. A description of the dataset can be found elsewhere [19]. As individuals in this dataset should serve as an anxiety free control group in our analysis, we excluded individuals with anxiety disorders (n = 265) or with a score of at least 3 on the PHQ-4 anxiety module [21] (n = 1448). Furthermore, 78 individuals with missing values were removed from the dataset, as it was not possible to impute both datasets. Overall, 3214 individuals remained in the dataset as “anxiety free group (AF)” for further analysis.

Health care utilisation and costs

Both studies were conducted by researchers from the same academic department by using approaches to data collection standardized in the department. The description of the single services as well as the units of measurement (contacts, days, hours) were comparable in both studies. In both datasets, information on health care resource use in three different health care sectors was collected. In the outpatient sector, visits to a psychiatrist/neurologist, psychologist/psychotherapist, general practitioner and other medical specialists were considered. In the inpatient sector, days in general, psychiatric and rehabilitation hospitals were assessed. In the home care sector, received hours of professional and informal caregiving (unpaid support and care provided by family members and friends) were measured. Additionally, disease-related days away from work (absenteeism) were recorded.

We applied the societal perspective [23]. Direct health care costs, i.e. costs directly related to treatments of diseases, were determined by valuing resource use data with German unit costs [24]. Indirect costs, i.e. productivity losses due to disease-related absence from work, were determined applying the human capital approach [25], valuing sick leave days with an average daily wage rate (including full- and part-time work, corrected for employees’ share of social contributions) [26]. All applied unit costs are reported in Table 1. Informal care has been monetarily valued by the replacement cost approach. The reference year of the cost calculation was 2012. If unit costs were not available for the base year, values were inflated to the year 2012 using the German consumer price index [27]. To avoid influence of outliners with high costs on the results we excluded individuals who had total healthcare costs above the 99th percentile (33 individuals in the AF group, 8 individuals in the PDA and 2 individuals in the PD group).

Table 1 Unit costs for different resource uses in 2012 Euros

Statistical analysis

In order to estimate excess costs of patients with PD and PDA in comparison to AF individuals, we applied a two-stage approach, comprising the pre-processing and the estimation stage. The main analyses of these study were performed based on the imputed dataset. However, to assess the influence of the imputation method (MICE) on the results we re-ran sensitivity analyses based on the original dataset, which is the dataset without imputation. Additionally, we performed sensitivity analyses based on a sample including the cost outliers to assess the influence of the approach in our main analysis. All analyses were conducted in Stata 14 (StataCorp, Texas, USA).

The pre-processing stage was necessary because the three groups differed in terms of their covariates and were therefore not directly comparable. Hence, we reweighted the AF group on the PD and PDA group separately, applying the reweighting method of Entropy Balancing [28]. In Entropy Balancing covariates of a target group will be made comparable to those of a comparison group in reference to predefined moments. In case of our study, all available covariates (Table 2) were included and the predefined moments were the mean and the variance of the target group. The process is successful if the value of the adjusted moments of the covariates in the target group are similar to the corresponding moments of the comparison group, which means that the adjusted value of all included covariates is in a predefined area of tolerable deviation. In case of our study this tolerance level was set to 0.005, a rigorous level of tolerance (the default level set in STATA is 0.015). Entropy balancing works in such a way that every observation in the target group receives a balancing weight. This balancing weight represents the number of times the specific observation is considered in the process of adjusting the included covariates in terms of the predefined moments. The algorithm varies the balancing weights until the moments of all included covariates of the target group are comparable to the moments of the comparison group [28]. A reweighted dataset is similar to a randomized trial, where both groups are expected to have no systematic difference in observed baseline characteristics. We chose this approach as it does not exclude members of the target group, reaches a higher degree of balance in the covariates than other approaches, and does not require manual adjustments by the scientist, which is a source of computational delay and human error. Entropy Balancing was performed for each of the imputed dataset (m = 40) separately. To perform the method we used a user written Stata program (Ebalance) by Hainmueller et al. [29].

Table 2 Pre- and post-balanced covariates

In the estimation stage, we estimated mean cost differences (excess costs) between the groups for each cost category (Table 3). As costs cannot reach negative values, are often skewed, and sometimes have a large number of ‘zero’ values [30], the choice of an appropriate statistical model is an important step in modelling costs. For cost categories with a small number of ‘zeros’, costs were modelled using a generalized linear model (GLM) with gamma family and log-link function [31]. For cost categories with a large number of ‘zeros’, a two-part (TP) model approach was applied [32]. In the first part of the TP-model we used a logit model in order to estimate the probability of positive costs, in the second part we used a GLM with gamma family and log-link function to estimate costs in patients with positive costs [32]. The TP-model was estimated using the user-written Stata command twopm by Belotti et al. [32]. Mean differences in costs were obtained by calculating margins [32], which means we predicted the cost difference between the groups based on the results of the previously run model under consideration of the specific cost distribution. These results are referred to as excess costs in the further course of this article. As the twopm and margin command is not supported under Stata’s “mi estimate option”, we estimated 40 separate regression models (TPMs and GLMs) and combined the results using Rubin’s pooling rules for multiple imputed datasets [33]. All regressions were conducted as weighted regressions using the balancing weights from the pre-processing stage [28]. The only independent variable in all regressions was the group variable [PD or PDA (yes/no)]. We did not include further covariates into regression analyses as all observed covariates were already included in the Entropy balancing and would not have any further effect on averages of cost estimates [34].

Table 3 Average per capita 6-month costs of AF, PD and PDA for different cost categories in 2012 Euros

Results

The pre- and post-balancing results of the demographic and clinical characteristics for individuals from the AF population and patients with PDor PDA are shown in Table 2. By using entropy balancing we reached a high degree of concordance between the groups, in all included covariates.

The results of the cost-analysis of the AF group in comparison to the PD group are shown in Tables 3 and 4. As a main result, we found average (SD) 6-month total costs of 969€ (110€) in the AF group and 4189€ (658€) in the PD group, resulting in significant 6-month total excess costs of 3220€ (95% CI 1917€–4522€) for patients with PD in comparison to AF individuals. About 1953€ (95% CI 909–2996€) of the 6-month total excess costs were indirect costs due to absenteeism. This corresponds to 61% of the total excess costs. The largest share of the direct excess costs of 1267€ (95% CI 565€–1969€) was incurred by statistically significant general hospital excess costs of 520€ (95% CI 124€–916€) and excess costs due to informal care of 367€ (95% CI 31€–703€). In the outpatient sector we found significant excess costs of 130€ (95% CI 104€–155€) caused by visits to general practitioner and excess costs of 89€ (95% CI 19€–159€) for psychologist or psychotherapist visits. We found no statistically significant excess costs for treatments by other medical specialists, psychiatrists/neurologists, and for treatments in psychiatric or rehabilitation hospitals.

Table 4 6-month per capita excess cost of PD and PDA for different cost categories in 2012 Euros

As shown in Tables 3 and 4, patients with PDA had average (SD) 6-month total costs of 5072€ (473€) compared to 1129€ (184€) in AF individuals, resulting in statistically significant total excess costs of 3943€ (95% CI 2950€–4936€). The share of indirect cost (2331€ (95% CI 1617€–3045€)) was 59%. The largest share of the direct excess costs (1612€ (95% CI 994€–2229€)) in the PDA group were excess costs of 579€ (95% CI 281€–876€) for informal care. In the inpatient sector, statistically significant excess costs incurred for psychiatric (331€ (95% CI 53€–610€)), general (275€ (95% CI 14€–535€)) and rehabilitation hospital stays (187€ (95% CI 6€–368€)). In the outpatient sector, excess costs incurred for treatments by general practitioners (130€ (95% CI 112€–148€) and psychologists or psychotherapists (119€ (95% CI 78€–159€)). Low but statistically significant excess costs were found for treatments by psychologists or psychotherapists with 15€ (95% CI 2€–27€). Excess costs for treatment by other medical specialists were not statistically significant.

The sensitivity analyses on the imputation of missing values led to results comparable to the results of the main analyses (Appendix Table 5). The amount of the excess costs decreased in several cost categories and the CI of the costs for inpatient rehabilitation services crossed the 0€ threshold, leading to an insignificant group difference. The sensitivity analyses considering the cost outliers showed no noticeable deviating results in the comparison of AF and PD. However, in the comparison of AF and PDA, we observed that in the inpatient sector the cost differences in general and psychiatric hospital services were not statistically significant anymore. The small sample size of the PDA group and the high cost in the inpatient sector could be an explanation for this.

Discussion

Our analysis showed that PD and PDA raised health care costs (direct costs) and productivity losses (indirect costs) substantially. Around 60% of total costs were caused by productivity losses due to absenteeism. In the first instance, productivity losses are a loss to the economy and society and not necessarily influenced by resource allocation. However, scrutinizing the composition of excess cost for health care resources, our results indicate that there could be an interconnection between resource allocation and absenteeism. The utilization of general hospital services by PD patients is the resource category associated with the largest difference in costs. For PDA patients only the difference in informal care was larger. It is obvious that a higher utilization of hospital services goes along with a higher degree of absenteeism. This is an important aim for policy and medical interventions. Measures structured to shift the emphasis of treatment from the inpatient to the outpatient sector could not only result in lower health care costs but also in smaller losses of economic productivity. However, in this context it is indicated to differentiate between PD and PDA. While PDA showed an increase of inpatient service utilization in all subtypes of hospitals—and the largest increase for utilization of psychiatric hospitals—, PD showed a pronounced increase for general hospital services exclusively.

This pronounced excess utilization of general hospital services by PD patients can be interpreted as a direct result of a panic attack, which manifests with distinct somatic symptoms like chest pain, dizziness and breathing difficulties [1]. These symptoms can lead to the assumption of a severe physical and maybe even life-threatening disease. In this situation, people will be admitted to a hospital, maybe as emergency, and will first underwent a thorough evaluation in the inpatient setting. This process utilized resources in an unproductive way. There are different explanations for this. For example, (a) the patient was not diagnosed with a panic disorder, (b) the patient is not informed about the nature of his disorder, (c) the clinical staff is not qualified enough to differentiate between a heart attack and a panic attack with comparable symptoms or (d) clinical pathways demand that a patient showing these symptoms has to be hospitalized for a rigorous examination. Although some of these problems have been addressed before [35], the reasons for the hospitalization of PD patients in a general hospital has to be evaluated more thoroughly in further research.

In case of PDA, the situation is different. General hospital and psychiatric hospital services are both utilized to a high degree. This is not necessarily a miss-allocation as patients with PDA receive professional services in accordance with their disorder. Yet, the question arises whether these services must be delivered in an inpatient setting. It could be worthwhile to develop outpatient-based intervention for PDA to reduce the length of hospital stays [12, 36].

In the outpatient sector, the utilization of mental health services is more pronounce for PDA patients in comparison to PD patients. This in combination with the utilization of inpatient mental health care services could be regarded as indicator for a greater health burden of PDA and hence an increased need for services or a greater awareness of need for support. However, the excess costs for general practitioner (GP) services were comparable for PD and PDA. This and the relatively high costs for GP services (Table 3) could be an indicator for the important part the GP has in the care for PD and PDA patients. This important role of the GP poses a risk and a chance. The risk is that the GP might be not qualified enough to provide or has no access to PD(A)-targeted services. The chance is that the GP as access point for patients into the health care system could be a valuable stakeholder in the care process by providing identification, education and early treatment [12]. Consequently, interventions to empower the GP to identify PD and PDA, to educate the patient and to offer first-line support could be a valuable addition to the services offered by mental health specialists, especially in the German setting, where waiting periods for psychotherapy are quite long [37].

The elevated costs for informal care, can be considered as manifestation of a high degree of need for support. This transfers a part of health care provision into the informal and hence unpaid sector. This relieves the health care budget and promotes the flexibility of service delivery, as family members, friends and neighbours can offer support in a variety of small tasks on short notice. However, there is evidence that the provision of informal care, can lead to worse health and a loss of income on the part of the provider of informal care [38, 39]. This in turn affects the health care budget and societal productivity.

Strengths of our study include the used data sources and the analytical approach. In our study we combined data from a representative population survey [19] with baseline data from a high quality clinical trial with patients diagnosed with panic disorder (with and without agoraphobia) [12]. The combination of these two datasets enabled a valid comparison of panic disorder patients with the general population. In order to reach comparability of the datasets, it is essential to apply elaborated matching or balancing methods. For this reason we applied an innovative reweighting method, Entropy Balancing [28]. In comparison to other matching methods (e.g. propensity score matching), Entropy Balancing is favourable and advantageous [28]. In validation analyses, Entropy Balancing showed a higher degree of balance in covariates in comparison to other common processing adjustments (matching methods), e.g. propensity score or genetic matching [28].

Some limitations should be considered when interpreting our results. As two independent samples where used in this analysis, some data adjustments had to be performed. The assessment of visits to other medical specialist and comorbidities was differently conducted in the datasets. Therefore, data had to be aggregated. Furthermore, the mode of the assessment of comorbidities was different. In the clinical trial general practitioners coded diagnosis, while in the general population sample patients reported their diagnosed comorbidities.

In the main analysis of this study, AF individuals from the general population dataset were compared to individuals with PD and PDA. To determine the population of AF individuals from the general population dataset, individuals with a diagnosed anxiety disorder or with a score of ≥ 3 in the anxiety module of the PHQ-4 [20] were excluded. All other mental comorbidity variables of this AF group were used in the Entropy Balancing in order to create a comparable comparison group to the PD and PDA group in term of mental comorbidities. As anxiety disorders are often diagnosed with other severe mental disorders [40,41,42,43,44], it is possible that the balanced individuals in the AF group had relative mild mental comorbidities and therefore low health care costs in the psychiatric cost categories.

Finally, we want to declare that despite the efforts of mitigating the effects of missing values and differences in sample characteristics by using MICE and entropy balancing, we cannot warrant that the results of our analyses are not influenced by any kind of bias. Entropy balancing led to a high degree of comparability between the groups in measured variables and missing values were imputed by an approach considering uncertainty. However, as both approaches are based exclusively on measured variables, they have no direct influence on unobserved heterogeneity.

Conclusion

Panic disorder is associated with high excess costs and constitutes a major economic challenge to health care system and society. Resources are mainly utilized in the inpatient sector. The development of measures to identify, educate and treat patients in the outpatient sector could optimize resource allocation. This approach could also lead to a reduction of absenteeism. Further research is needed. Additionally, panic disorder not only affects the patient but also his or her personal/familial network. This could lead to further health and occupational consequences.

Abbreviations

AF:

anxiety free population

GLM:

generalized linear model

PD:

panic disorder

PDA:

panic disorder with agoraphobia

PHQ:

Patient Health Questionnaire

PPP:

purchasing power parities

TP:

two part model

References

  1. International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10)-WHO Version for 2016. http://apps.who.int/classifications/icd10/browse/2016/en#/F40-F48. Accessed 17 Jan 2018.

  2. Jacobi F, Höfler M, Siegert J, Mack S, Gerschler A, Scholl L, Busch MA, Hapke U, Maske U, Seiffert I. Twelve-month prevalence, comorbidity and correlates of mental disorders in Germany: the mental health module of the German health interview and examination survey for adults (DEGS1-MH). Int J Methods Psychiatr Res. 2014;23:304–19.

    Article  Google Scholar 

  3. GENESIS-Online: population: Germany, age in years. https://www-genesis.destatis.de/genesis/online/data;jsessionid = 80068337C390C68BC488BE42BB13F725.tomcat_GO_1_3?operation = abruftabelleBearbeiten&levelindex = 2&levelid = 1528701706985&auswahloperation = abruftabelleAuspraegungAuswaehlen&auswahlverzeichnis = ordnungsstruktur&auswahlziel = werteabruf&selectionname = 12411-0005&auswahltext = %23SALT013-ALT063%2CALT062%2CALT061%2CALT060%2CALT022%2CALT055%2CALT021%2CALT054%2CALT020%2CALT019%2CALT059%2CALT058%2CALT057%2CALT023%2CALT056%2CALT018%2CALT052%2CALT051%2CALT050%2CALT049%2CALT053%2CALT044%2CALT048%2CALT047%2CALT046%2CALT045%2CALT077%2CALT043%2CALT076%2CALT042%2CALT075%2CALT041%2CALT074%2CALT079%2CALT078%2CALT036%2CALT035%2CALT034%2CALT040%2CALT039%2CALT038%2CALT037%2CALT073%2CALT072%2CALT071%2CALT033%2CALT066%2CALT032%2CALT065%2CALT031%2CALT064%2CALT030%2CALT070%2CALT069%2CALT068%2CALT067%2CALT025%2CALT024%2CALT029%2CALT028%2CALT027%2CALT026&werteabruf = Werteabruf. Accessed 10 June 2018.

  4. Michael T, Zetsche U, Margraf J. Epidemiology of anxiety disorders. Psychiatry. 2007;6:136–42.

    Article  Google Scholar 

  5. Wittchen HU, Gloster AT, Beesdo-Baum K, Fava GA, Craske MG. Agoraphobia: a review of the diagnostic classificatory position and criteria. Depress Anxiety. 2010;27:113–33.

    Article  Google Scholar 

  6. Saarni SI, Suvisaari J, Sintonen H, Pirkola S, Koskinen S, Aromaa A, Lönnqvist J. Impact of psychiatric disorders on health-related quality of life: general population survey. Br J Psychiatry. 2007;190:326–32.

    Article  Google Scholar 

  7. Beard C, Weisberg RB, Keller MB. Health-related Quality of Life across the anxiety disorders: findings from a sample of primary care patients. J Anxiety Disord. 2010;24:559–64.

    Article  Google Scholar 

  8. Rudaz M, Craske M, Becker E, Margraf J. Health anxiety and fear of fear in panic disorder, agoraphobia, and social phobia: a prospective longitudinal study. Depress Anxiety. 2010;27:404–11.

    Article  Google Scholar 

  9. Woud ML, Zhang XC, Becker E, McNally RJ, Margraf J. Don’t panic: interpretation bias is predictive of new onsets of panic disorder. J Anxiety Disord. 2014;28:83–7.

    Article  Google Scholar 

  10. Konnopka A, Leichsenring F, Leibing E, König H-H. Cost-of-illness studies and cost-effectiveness analyses in anxiety disorders: a systematic review. J Affect Disord. 2009;114:14–31.

    Article  Google Scholar 

  11. Batelaan N, Smit F, de Graaf R, van Balkom A, Vollebergh W, Beekman A. Economic costs of full-blown and subthreshold panic disorder. J Affect Disord. 2007;104:127–36.

    Article  Google Scholar 

  12. Gensichen J, Hiller TS, Breitbart J, Teismann T, Brettschneider C, Schumacher U, Piwtorak A, König H-H, Hoyer H, Schneider N. Evaluation of a practice team-supported exposure training for patients with panic disorder with or without agoraphobia in primary care-study protocol of a cluster randomised controlled superiority trial. Trials. 2014;15:112.

    Article  Google Scholar 

  13. Norman SB, Campbell-Sills L, Hitchcock CA, Sullivan S, Rochlin A, Wilkins KC, Stein MB. Psychometrics of a brief measure of anxiety to detect severity and impairment: the Overall Anxiety Severity and Impairment Scale (OASIS). J Psychiatr Res. 2011;45:262–8.

    Article  Google Scholar 

  14. Löwe B, Gräfe K, Zipfel S, Spitzer RL, Herrmann-Lingen C, Witte S, Herzog W. Detecting panic disorder in medical and psychosomatic outpatients: comparative validation of the Hospital Anxiety and Depression Scale, the Patient Health Questionnaire, a screening question, and physicians’ diagnosis. J Psychosom Res. 2003;55:515–9.

    Article  Google Scholar 

  15. Spitzer RL, Kroenke K, Williams JB, Patient Health Questionnaire Primary Care Study Group. Validation and utility of a self-report version of PRIME-MD: the PHQ primary care study. JAMA. 1999;282:1737–44.

    Article  CAS  Google Scholar 

  16. Azur MJ, Stuart EA, Frangakis C, Leaf PJ. Multiple imputation by chained equations: what is it and how does it work? Int J Methods Psychiatr Res. 2011;20:40–9.

    Article  Google Scholar 

  17. Graham JW, Olchowski AE, Gilreath TD. How many imputations are really needed? Some practical clarifications of multiple imputation theory. Prev Sci. 2007;8:206–13.

    Article  Google Scholar 

  18. van Buuren S. Flexible imputation of missing data. Boca Raton: Taylor & Francis; 2012.

    Book  Google Scholar 

  19. Grupp H, Konig HH, Konnopka A. Health care utilisation and costs in the general population in Germany. Health Policy. 2016;120:159–69.

    Article  Google Scholar 

  20. Lowe B, Wahl I, Rose M, Spitzer C, Glaesmer H, Wingenfeld K, Schneider A, Brahler E. A 4-item measure of depression and anxiety: validation and standardization of the Patient Health Questionnaire-4 (PHQ-4) in the general population. J Affect Disord. 2010;122:86–95.

    Article  Google Scholar 

  21. Kemper CJ, Brähler E, Zenger M. Psychologische und sozialwissenschaftliche Kurzskalen: Standardisierte Erhebungsinstrumente für Wissenschaft und Praxis. MWV Medizinisch Wissenschaftliche Verlagsgesellschaft mbH & Company KG; 2014.

  22. Tritt K, von Heymann F, Zaudig M, Zacharias I, Söllner W, Loew T. Entwicklung des Fragebogens» ICD-10-Symptom-Rating «(ISR). Z Psychosom Med Psychother. 2008;54:409–18.

    PubMed  Google Scholar 

  23. Drummond M, Manca A, Sculpher M. Increasing the generalizability of economic evaluations: recommendations for the design, analysis, and reporting of studies. Int J Technol Assess Health Care. 2005;21:165–71.

    Article  Google Scholar 

  24. Bock JO, Brettschneider C, Seidl H, Bowles D, Holle R, Greiner W, Konig HH. Calculation of standardised unit costs from a societal perspective for health economic evaluation. Gesundheitswesen. 2015;77:53–61.

    PubMed  Google Scholar 

  25. Rice DP, Cooper BS. The economic value of human life. Am J Public Health Nation’s Health. 1967;57:1954–66.

    Article  CAS  Google Scholar 

  26. Federal Statistical Office. Statistical year book 2013. Wiesbaden: Federal Statistical Office; 2013 (in German).

    Google Scholar 

  27. Verbraucherpreise (Consumer Prices). https://www.destatis.de/DE/ZahlenFakten/GesamtwirtschaftUmwelt/Preise/Verbraucherpreisindizes/Tabellen_/VerbraucherpreiseKategorien.html;jsessionid=014AA97F7BFDDE. Accessed 17 Jan 2018.

  28. Hainmueller J. Entropy balancing for causal effects: a multivariate reweighting method to produce balanced samples in observational studies. Polit Anal. 2012;20(1):25–46.

    Article  Google Scholar 

  29. Hainmueller J, Xu Y. Ebalance: a Stata package for entropy balancing. J Stat Softw. 2013. https://0-doi-org.brum.beds.ac.uk/10.18637/jss.v054.i07.

    Article  Google Scholar 

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

    Article  Google Scholar 

  31. Hardin JW, Hilbe JM. Generalized linear models and extensions. 2nd ed. College Station: Stata Press; 2007.

    Google Scholar 

  32. Belotti F, Deb P, Manning WG, Norton EC. twopm: two-part models. Stata J. 2015;15:3–20.

    Article  Google Scholar 

  33. Rubin DB. Multiple imputation for nonresponse in surveys. New York: Wiley; 2004.

    Google Scholar 

  34. Marcus J. The effect of unemployment on the mental health of spouses—evidence from plant closures in Germany. J Health Econ. 2013;32:546–58.

    Article  Google Scholar 

  35. Goodwin RD, Faravelli C, Rosi S, Cosci F, Truglia E, de Graaf R, Wittchen HU. The epidemiology of panic disorder and agoraphobia in Europe. Eur Neuropsychopharmacol. 2005;15:435–43.

    Article  CAS  Google Scholar 

  36. Curth NK, Brinck-Claussen UØ, Davidsen AS, Lau ME, Lundsteen M, Mikkelsen JH, Csillag C, Hjorthøj C, Nordentoft M, Eplov LF. Collaborative care for panic disorder, generalised anxiety disorder and social phobia in general practice: study protocol for three cluster-randomised, superiority trials. Trials. 2017;18:382.

    Article  Google Scholar 

  37. BPtK Study: waiting times for outpatient psychotherapeutic services. http://www.npce.eu/mediapool/113/1137650/data/NPCI_Info_waitinglists_EN.pdf. Accessed 17 Jan 2018.

  38. Chiao C-Y, Wu H-S, Hsiao C-Y. Caregiver burden for informal caregivers of patients with dementia: a systematic review. Int Nurs Rev. 2015;62:340–50.

    Article  Google Scholar 

  39. Lilly MB, Laporte A, Coyte PC. Labor market work and home care’s unpaid caregivers: a systematic review of labor force participation rates, predictors of labor market withdrawal, and hours of work. Milbank Q. 2007;85:641–90.

    Article  Google Scholar 

  40. Nabavi B, Mitchell AJ, Nutt D. A lifetime prevalence of comorbidity between bipolar affective disorder and anxiety disorders: a meta-analysis of 52 interview-based studies of psychiatric population. EBioMedicine. 2015;2:1405–19.

    Article  Google Scholar 

  41. Thibodeau MA, Welch PG, Sareen J, Asmundson GJ. Anxiety disorders are independently associated with suicide ideation and attempts: propensity score matching in two epidemiological samples. Depress Anxiety. 2013;30:947–54.

    Article  Google Scholar 

  42. Gadermann AM, Alonso J, Vilagut G, Zaslavsky AM, Kessler RC. Comorbidity and disease burden in the National Comorbidity Survey Replication (NCS-R). Depress Anxiety. 2012;29:797–806.

    Article  Google Scholar 

  43. McEvoy PM, Grove R, Slade T. Epidemiology of anxiety disorders in the Australian general population: findings of the 2007 Australian National Survey of Mental Health and Wellbeing. Aust N Z J Psychiatry. 2011;45:957–67.

    Article  Google Scholar 

  44. Greene AL, Eaton NR. Panic disorder and agoraphobia: a direct comparison of their multivariate comorbidity patterns. J Affect Disord. 2016;190:75–83.

    Article  Google Scholar 

Download references

Authors’ contributions

TH, JB, JM, JG and HHK, contributed to the conception and design of the CRCT. AK and HHK contributed to the conception of the telephone survey. All authors contributed to the acquisition of the data. FB and CB performed the analysis. All authors contributed to the interpretation of the results. CB and FB drafted the manuscript. TH, AK, JB, JM, JG and HHK, revisited the manuscript critically. All authors read and approved the final manuscript.

Acknowledgements

We thank Dirk Heider by providing statistical support in applying Rubin’s pooling rules to Two-Part Models.

The Jena-PARADISE study group comprises the following contributors/collaborators (in alphabetical order): Wolfgang Blank, Florian Bleibler, Jörg Breitbart, Christian Brettschneider, Antje Freytag, Jochen Gensichen, Thomas S. Hiller, Heike Hoyer, Bert Huenges, Hans-Helmut König, Armin Mainz, Jürgen Margraf, Sylvia Sänger, Mercedes Schelle, Konrad Schmidt, Nico Schneider, Elisabeth Schöne, Ulrike Schumacher, Michael Sommer, Tobias Teismann, Paul Thiel, Michel Wensing.

Competing interests

The authors declare that they have no competing interests.

Availability of data and materials

The datasets used and analyzed are available from the corresponding author on reasonable request.

Consent for publication

Not applicable.

Ethics approval and consent to participate

The cluster-randomized trial was approved by the ethics committee of the Friedrich-Schiller-University at the Medical Faculty in Jena, Germany (reference 3484-06/12). The telephone survey was approved by the ethics committee of the Medical Chamber Hamburg (reference WF-011/14). In both studies written consent to participate was obtained.

Funding

This study was supported by the German Federal Ministry of Education and Research (Grant Numbers: 01GY1146 and 01EH1101B). The sponsor had no role in study design, collection, analysis, and interpretation of the data, in writing the manuscript or in the decision to submit the manuscript for publication.

Publisher’s Note

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

Author information

Authors and Affiliations

Authors

Consortia

Corresponding author

Correspondence to Christian Brettschneider.

Appendix

Appendix

See Tables 5, 6.

Table 5 6-month per capita excess cost of PD and PDA for different cost categories in 2012 Euros (complete case analysis)
Table 6 6-month per capita excess cost of PD and PDA for different cost categories in 2012 Euros (Sensitivity analysis: outlier included)

Rights and permissions

Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. 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.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Brettschneider, C., Bleibler, F., Hiller, T.S. et al. The allocation of resources in the care for patients with panic disorder in Germany: an excess cost analysis informing policy and science. Cost Eff Resour Alloc 17, 9 (2019). https://0-doi-org.brum.beds.ac.uk/10.1186/s12962-019-0177-4

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://0-doi-org.brum.beds.ac.uk/10.1186/s12962-019-0177-4

Keywords