Direct and Indirect Costs of Refractory Epilepsy in a Tertiary Epilepsy Center in Germany


Address correspondence and reprint requests to Dr. Hajo Hamer at the Department of Neurology, Philipps-University Marburg, Rudolf-Bultmann-Str. 8, 35039 Marburg, Germany. E-mail:


Summary: Purpose: There are only few studies on the costs of epilepsy in Germany. Therefore, we performed a pilot study to estimate the direct and indirect costs of refractory epilepsy in a German epilepsy center.

Methods: A “prevalence-based,” cross-sectional convenience sample of adults with active epilepsy attending the outpatient clinic of our tertiary epilepsy center was evaluated. Seizure-free patients and patients presenting with their first seizure were excluded. Direct and indirect costs were prospectively recorded over a three-month period using questionnaires and a patient diary. Cost driving factors were identified.

Results: One hundred one patients were included (40.7 ± 15.2 years; disease duration: 18.1 ± 15.3 years; 6 patients had focal epilepsy with simple partial seizures only, 28 with complex partial seizures, 43 with secondarily generalized tonic–clonic seizures; 20 had idiopathic generalized epilepsy with generalized tonic–clonic seizures). The total costs of epilepsy per patient were in average € 2610 ± 4200 over the three-month period. Direct cost contributed 39% to the total costs. Costs of anticonvulsant medication were the main contributor to the direct costs while indirect costs were caused mainly by losses due to early retirement. Cost driving factors included higher seizure frequency, longer disease duration, ictal falls, and situationally inappropriate complex behavior during or after the seizure.

Conclusions: Indirect costs were higher than direct costs in adult patients with active epilepsy attending a German epilepsy center. Medication contributed the most to the direct costs and early retirement was the main factor for the indirect costs. The costs of poorly controlled epilepsy in this German study were above average of the European costs of epilepsy.

Epilepsy imposes a substantial burden on individuals and society as a whole. Assuming a prevalence of 0.52% (Forsgren et al., 2005), more than 400,000 persons suffer from epilepsy in Germany. The resource implications of treating and caring for people with epilepsy are likely to be significant but especially in Germany, only very few studies have examined the costs of this condition (Dodel et al., 1996; van Hout et al., 1997). The need for such studies increased in recent years with the marketing of the vagal nerve stimulator, the intensifying of surgical epilepsy treatment, and the advent of several new antiepileptic drugs (AED) whose costs exceed by far those of the older AED. This occurred in an era when growth in health care spending has slowed and concerns about rising costs and limited information about long-term patient outcome enforced awareness of cost containment and accountability.

Therefore, we performed a pilot study to estimate the direct and indirect costs of active epilepsy in Germany. For the purpose of the study, the broad perspective of society was adopted. We evaluated a “prevalence-based,” cross-sectional convenience sample of patients attending the outpatient clinic of our tertiary epilepsy center. We cannot exclude that the sample was largely composed of those patients with more severe disease or who had other associated problems making their costs proportionally higher than those of the general epilepsy population (Begley et al., 1999b). The costs of epilepsy in the three months following the visit were prospectively recorded. In addition, factors were identified which influenced the financial burden of the illness.


In 2003, we studied the first 100 consecutive patients who had an appointment in the outpatient clinic of our epilepsy center and were willing to participate in the study. In addition, they had to fulfill the following criteria of eligibility.

Inclusion criteria

  • 1Only patients who suffered from epilepsy were included. At least two unprovoked seizures were required for the diagnosis of epilepsy (Blume et al., 2001).
  • 2All patients had to be 18 years of age or older.

All patients provided written informed consent after a detailed explanation of the study. The study had the approval of the local Ethics Committee.

Exclusion criteria

  • 1Patients were excluded when the diagnosis of epilepsy could not be determined without doubts.
  • 2Patients with one seizure only were excluded.
  • 3Patients were excluded if they were more than one year seizure-free at the time of study entry. This was done because the focus of this study was on the costs of active epilepsy.

Cost assessment

Costs were assessed based on a patient diary and via a questionnaire administered at baseline and at three months. The following costs were evaluated: direct costs (including drug, medical, and nonmedical costs) and indirect costs. The data were calculated from the perspective of “Gesetzliche Krankenversicherung” (GKV, statutory health insurance) and the societal perspective. All costs were calculated for a three-month period in 2003 Euros (€).

The goal of this article was to calculate the genuine costs due to epilepsy and, not costs that may be triggered by comorbidities. Therefore, the patient was asked in detail in the questionnaire whether or not the patient used a service or a resource especially due to epilepsy. In respect to medication, the same was asked. The medication was checked by the treating physician during the interview together with the patient, whether it was used in relation to epilepsy.

Direct medical costs from the perspective of the “GKV”

Drug costs were obtained from the official German price list of drugs (“Rote Liste,” 2003) and incorporated a 6% deduction according to the GKV guidelines. Expenditures for inpatient care and rehabilitation during the observation period were calculated by multiplying the total number of inpatient days by the mean costs of one day of inpatient care (basic daily charge + specific ward charge). The charges have been courtesy of the different hospital administrations. Costs for outpatient care were calculated by multiplying the number of visits times the provider-specific charge listed in the “Einheitlichen Bewertungsmaßstab”[doctor's fee scale, (EBM)]. Hospital outpatient ambulatory visits were calculated by multiplying the number of visits times the hospital-specific charge. Ancillary treatment services, including physiotherapy and speech therapy, were taken from the patient diary and calculated by multiplying the number of services times the EBM service-specific charges. The number of ambulatory diagnostic procedures at office-based neurologists or general practitioners was extracted from patient diaries, with costs calculated by multiplying the number visits times the EBM procedure-specific charges.

Direct nonmedical costs

People with epilepsy often require services not traditionally available in medical care databases. These include items such as assistive or protective equipment, or dental care for side effects of AED. Costs for home equipment were calculated using the official German price lists.

Indirect costs from the societal perspective

The evaluation of indirect costs followed the German recommendations for performing health economic evaluations (Schöffski and von Schulenburg, 2000). A more detailed description of our evaluation of the indirect costs has been given in a previous study (Spottke et al., 2005).

Costs due to lost productivity

Indirect costs due to a patient's earlier retirement were calculated using the following formula based on the human capital calculation: calendar days of the remaining study period prior to the 65th birthday *€90 [the mean gross income in 2003 was €32,609, given by Federal Statistics (, divided by 365 calendar days =€90].

For productivity losses attributable to epilepsy, also €90 per day were considered and multiplied with the days of temporary sickness. The costs were calculated for the evaluation period.

Cost driving factors

Potential cost driving factors were recorded. These were epilepsy diagnosis, seizure frequency, duration of epilepsy, age of patient, gender, ictal fall, ictal loss of consciousness, and “situationally inappropriate, complex behavior” during or after the seizure. “Inappropriate, complex behavior” was included because the occupational medicine concept of the German employer's liability insurance association attaches the highest risk for injuries at work to seizures associated with “situationally inappropriate, complex behavior.” This is defined as uncontrolled complex, situationally inappropriate behavior, such as manual automatisms or uncontrolled moving around, during or after the seizure.

Data entry and statistics

Data entry was performed using a Microsoft Access database (Microsoft, 2000, Redmond, WA, U.S.A.). Statistical analyses were performed using SAS Version 8 (SAS Institute, Inc., Cary, NC, U.S.A.). Cost data are presented as mean ± SD, median, maximum and minimum, and nonparametric 95% confidence intervals. As most cost variables had right-skewed distributions, data were transformed by the natural log function (cost +€1) to achieve normally distributed variables before using nonparametric (rank) methods for further statistical analyses. Parameter correlation was calculated using Spearman's rank correlation coefficient. To test for differences between groups defined by potential cost predictors, we used the Mann–Whitney U test for comparisons between two groups and the Kruskal–Wallis test for the comparison of multiple groups.

As our analyses were exploratory and involved multiple tests, all p-values were interpreted as descriptive measures rather than results of hypothesis testing. No adjustments for multiple testing were performed.

Literature search

To identify economic studies on epilepsy for comparison to our results, we performed a systematic literature search in electronic databases using a combined search strategy including the following terms: “cost,”“cost of illness,”“healthcare evaluation,”“cost-effectiveness,”“cost utility,”“cost benefit,” and “epilepsy” and “seizure.” The search was first performed on the databases Medline and PreMedline (1990 until June 2006), and the Cochrane's database of systematic reviews (until first quarter 2006). A more broadly specified search was performed on the databases DARE (database of abstracts of reviews of effectiveness), NHS EED (economic evaluation database), and HTA (health technology assessment) of the United Kingdom's NHS Centre for reviews and dissemination ( We examined the reference lists of all identified studies and previous reviews for additional studies and asked clinical experts for additional published studies. Articles were also identified through searches of the extensive files of the authors. There was no restriction on language.


One hundred and one patients were enrolled in the study. Of these, 47 were male and 54 were female, with a mean age of 40.7 ± 15.7 years (range: 18–78 years) and an average disease duration of 18.1 ± 15.4 years (range: 0.1–52 years). No patient died during the study period. Table 1 shows patients' characteristics depending on the epileptic syndromes.

Table 1. Epilepsy syndromes and patients' characteristics
 NMean age (in yr)Duration of disease (in yr)Seizure frequency (per mo)c
  1. aIncluding 2 patients with multifocal epilepsy.

  2. bIncluding 1 patient with multifocal epilepsy.

  3. cMedian.

Focal epilepsy with simple partial seizures only  635.9 ± 15.013.1 ± 8.8 1.7
Focal epilepsy with complex partial seizuresa 2845.5 ± 16.619.1 ± 15.72.0
Focal epilepsy with secondarily generalized tonic–clonic seizuresb 4338.7 ± 13.316.0 ± 13.82.1
Idiopathic generalized epilepsy with generalized tonic–clonic seizures 2039.0 ± 15.724.8 ± 18.70.7
Unclassified syndrome  437.4 ± 11.95.4 ± 2.40.4
Total patient group10140.7 ± 15.718.1 ± 15.42

Cost assessment

Total costs of epilepsy amounted to a mean of € 2610 ± 4200 per patient over the study period of three months (Table 2).

Table 2. Costs of epilepsy in Germany over three-month observation period (in 2003 €)
 MeanStandard deviationMinimumMedianMaximum
  1. N = 101.

  2. “GKV”: “Gesetzliche Krankenversicherung” (statutory health insurance).

Direct costs (“GKV”)
 Anticonvulsant drugs 600 61000 2460
 Hospital admission 280115000 7910
 Rehab  90 89000 8870
 Diagnostics  20  5000  240
 Physician/outpatient clinics  10  5000  440
 Physical therapy  10  4000  290
 Special equipment   3  1000   80
 Total1010160000 9600
Indirect costs
 Working days lost due to early retirement 780269000 9900
 Productivity loss due to part-time work 420136000 4950
 Off-work days due to seizures 41018200012760
Total (direct and indirect) costs of epilepsy26104200085020400

Direct medical costs

Inpatient care

A total of seven patients were hospitalized during the observation period due to epilepsy. Five of the seven patients were admitted for 4–42 days because of acute seizures in order to improve the anticonvulsant medication. One patient was treated in the hospital for one day because of a strained ankle and one patient was admitted for prolonged video-EEG monitoring for 19 days. The mean hospital costs for the study population were € 280 ± 1150 over the three-month period. When excluding the patient who underwent video-EEG monitoring we calculated costs of € 200 ± 850.

Drug costs

Drug costs were calculated separately for anticonvulsant drugs and drugs not used for epilepsy treatment. Of the 101 patients who were enrolled in the study, all but 4 (4%) patients received AED, 40 patients (40%) were on monotherapy, 34 patients (34%) received two, 20 patients (20%) three, and three patients (3%) took four different AED. Total anticonvulsant drug costs were € 600 ± 610 per patient for the study period of three-month. The costs by different drugs are listed in Table 3. The highest medication costs were seen in patients treated with topiramate, lamotrigine, levetiracetam, and gabapentin.

Table 3. Drug costs (in 2003 €)
DrugNDosage (in mg)a Mean ± SDCostsb/mo Mean ± SD
  1. aMonthly dosage considering only patients taking the particular AED.

  2. bMonthly costs considering only patients taking the particular AED.

  3. cIncluding clobazam, clonazepam, and diazepam.

  4. n.a., Not applicable; SD, standard deviation.

Carbamazepine401037.5 ± 471.027.6 ± 12.9
Valproate371509.5 ± 587.731.6 ± 11.9
Lamotrigine32 357.0 ± 180.0248.5 ± 120.3
Levetiracetam22 2022.7 ± 1115.3249.4 ± 137.9
Topiramate11 354.4 ± 236.2275.4 ± 179.8
Primidone11 590.9 ± 324.610.4 ± 5.73
Oxcarbazepine 71564.3 ± 524.9106.0 ± 35.5 
Phenytoin 6 207.1 ± 148.43.17 ± 2.0 
Phenobarbitol 4168.8 ± 32.5 2.5 ± 0.48
Gabapentin 42800.0 ± 490.0230.0 ± 42.4 
Benzodiazepinesc 4n.a.5.96 ± 3.39
Vigabatrin 22250.0 ± 750.0159.8 ± 53.3 
Tiagabine 130161.5


Nursing (either at home or institutional care) is reimbursed primarily by the “Pflegeversicherung” (PV) in Germany. Specialized or additional services may require patient copayment. Of the 101 patients enrolled, none of the patients requested and received PV support.

Cost-driving factors

Considering complex partial and generalized tonic–clonic seizures only, increasing seizure frequency was associated with higher direct costs (Table 4). Further significant cost-driving factors included longer disease duration as well as ictal falls and “inappropriate, complex behavior.” The costs did not vary significantly among the different epilepsy syndromes. Neither age nor gender of the adult patients significantly influenced the direct costs of the disease.

Table 4. Direct costs stratified by potential cost driving factors (in 2003 €)
  1. SPS, simple partial seizures; CPS, complex partial seizures; GTC, generalized tonic–clonic seizures; SD, standard deviation.

Epilepsy diagnosis
 Focal epilepsy with SPS 6 810 650 50 51 16900.85 
 Focal epilepsy with CPS2810101390 30 851 7510 
 Focal epilepsy with GTC43 9201510  0 401 8640 
 Generalised epilepsy with GTC2016903080  0 36112960 
 Unclassified syndrome 4 430 410100 301  880 
Seizure frequency (including only CPS and GTC)
 <1/ mo40 380 470  0 101 15000.007
 >1/mo2915302080 20 611 8640 
 >1/week24 960 550 70 911 2410 
 >1/day 821002480 30169113000 
Duration of epilepsy
 <1 yr10 210 340  0  71  9900.001
 1–5 yr20 8001310 30 201 4520 
 >5 yr7112702060  0 84112960 
Age of patients
 <20 yr 5  80  10 65  81  1000.12 
 21–40 yr4712502230  0 70612960 
 41–60 yr3411801760 30 561 8640 
 >60 yr15 610 340 80 611 1210 
Ictal falls
 Yes5113001820 20 861 86400.04 
 No49 8401880  0 27112960 
Ictal loss of consciousness
 Yes8111001900  0 611129600.49 
 No19 9701710  0 361 7510 
Inappropriate, complex behavior during or after seizure
 Yes3317602630  71041129600.001
 No67 7401210  0 301 7510 
 Male4711502200  0 551129600.87 
 Female5410001500  0 641 8640 

Indirect costs

The total indirect cost amounted to an average of € 1610 ± 3460 (Table 2). Of the 101 patients enrolled in the study, 8% (n = 8) reported a premature end of employment or profession due to epilepsy. Losses due to early retirement for patients <65 years were in average € 780 ± 2680 and lost working days attributed to seizures in 10 patients (10%) amounted to an average of € 410 ± 1820 (Table 2).


The present report is one of the few economic studies on epilepsy performed in Germany. The previous German studies focused on certain aspects of the costs of epilepsy, such as seizure frequency (van Hout et al., 1997) or medication costs (Dodel et al., 1996). We studied the direct and indirect costs of refractory epilepsy in a “convenience sample” of adults recruited from the outpatient clinic of our tertiary epilepsy center in which patients with poorly controlled epilepsy were likely to be over represented. Attention was paid to include genuine costs due to epilepsy and not the costs that may be triggered by comorbidities. Direct medical costs due to epilepsy were in average €1010 per patient over the three-months observation period. Medication more than hospital admissions contributed to the direct costs confirming recent studies from Italy (Tetto et al., 2002; Beghi et al., 2004). The medication costs found in the present study (mean €600) exceeded the ones recorded in a retrospective German evaluation from 1994 (mean €220) (Dodel et al., 1996). As expected (Jacoby et al., 1998), the costs of the newer AED were well above the costs of older AED in our study group. However, direct costs constituted only about 40% to the total costs while indirect costs were responsible for the remainder. We adopted the human capital approach in deriving the indirect costs, which focused on job-related indirect costs and did not capture other possible indirect costs associated with household and leisure activities. Early retirement had the greatest impact on the indirect costs. The present study evaluated premature retirement, part-time work, and days of work lost due to seizures, but no underemployment, excess mortality, or relative/care giver's loss of productivity. Indirect costs may be overestimated in an economy with less than full employment (Begley et al., 1999a). Moreover, it is often difficult to distinguish the epilepsy-specific costs from costs for comorbidities. Data tend to show the mean costs of patients with epilepsy rather than the epilepsy-specific costs.

In the present study, cost driving factors included higher seizure frequency as has been reported earlier (van Hout et al., 1997; Jacoby et al., 1998; Berto et al., 2000) and longer disease duration possibly due to higher medication costs. Specific seizure semiology such as falls or “inappropriate, complex behavior,” but not the epilepsy syndrome correlated with the direct costs. Age and gender did not influence the costs of the disease in the present study on adults.

Direct medical costs of epilepsy depend on the time period considered after diagnosis (Begley et al., 1994; Cockerell et al., 1994). We excluded seizure free patients and patients who had experienced their first seizure. Inclusion of patients presenting with their first seizure will likely come to higher costs of epilepsy because of the costs of the initial mostly hospital based evaluation. About half of these people, however, will not develop epilepsy. On the other hand, exclusion of these cases will disregard the costs of the first evaluation of patients who will subsequently develop epilepsy.

Despite the detailed study protocol and the prospective nature of our study, it has some limitations, which at least in part are related to the bottom-up approach. First, we did not use an epidemiological approach to recruit the patients involved in the study. This study represents a selected population of adults with intractable, mostly focal epilepsy suffering from a median of two seizures per month. Overall, 20%–30% of a prevalence population with epilepsy had more than monthly seizures while 40%–50% were seizure free during the last year in a systematic review on the epidemiology of epilepsy in Europe (Forsgren et al., 2005). Therefore, the cost estimates of the present study cannot easily be extrapolated to all epilepsies, and it also may explain why the costs are higher than in some other studies (Table 5). Second, due to the billing modalities of the German health care system, it is difficult to evaluate the real costs due to physician's services. This may lead to an underestimation of the costs. As these costs—similar to others studies—are small compared to the total costs, we did not perform a field study to evaluate these costs in more detail. Third, we calculated indirect costs using the human capital approach. In the current German situation of underemployment with a high number of unemployed subjects, the costs may not exactly reflect the burden to society. However, due to the limitations inherent to the friction cost approach, we decided to calculate the indirect costs using the human capital approach, which is in accordance with the national and international recommendations for conducting economic evaluations (Schöffski and von Schulenburg, 2000). Fourth, no patient was included in this study who required at-home or institutional care. Consideration of this group of patients would have increased the direct costs. Fifth, in our study we did not calculate the costs due to surgical treatment options (Boon et al., 2002). Sixth, another limitation of the study is the relatively short study period of three-month which may have led to a large variability in cost estimates.

Table 5. Cost-of-illness studies after 1990 evaluating adults with epilepsy
AuthorCountryaYrStudy populationStudy designMethodMedication (% of direct cost)Hospitalization (% of direct cost) cost)Direct annual cost (mean) cost)Direct annual costs in PPP 2003gIndirect costs (% of total annual costs)
  1. BU, bottom-up; yrs, years; n.a, not available.

  2. aUK, United Kingdom; US, United States of America; I, Italy; F, France; NL, The Netherlands.

  3. bDirect medical costs only.

  4. cExchange rate (01/07/1991): UKp = 1.614 US$.

  5. dExchange rate (01/03/1994): UKp = 1.476 US$.

  6. eIn purchasing power parity.

  7. fAnnual costs extrapolated from a three-month observation period.

  8. gPurchasing power parity was calculated according to OECD data 2003. Inflation rates ware adapted accordingly based on the data of the health sector from the German Federal Department of Statistics (Bundesamt für Statistik).

Cockerell et al. 1994UK19901,628RetrospectiveBU2059US$ 2073154069
602, all agesProspectiveBUn.a.n.a.917b (1 yr) 
n.a.n.a.254b (8 yr) 
Gessner et al., 1993Switzerland199037,000, all agesRetrospectiveBUn.a.16US$ 4005233035
Begley et al., 1994US1990n.a., all agesModelBU3814n.a.n.a.63
Swingler et al., 1994UK1991303, older than 7 yrRetrospectiveBU4930US$ 2252c1780n.a.
Jacoby et al., 1998UK1993789, all agesRetrospectiveBU2358US$ 2314d1950n.a.
Murray et al., 1996US1994n.a.ModelBU2055US$ 2885286066
Halpern et al., 2000US1995n.a., all agesRetrospectiveBU24n.a.US$ 14901490n.a.
Begley et al., 2000US1995608/1168, all agesRetrospectiveBU31n.a.US$ 6429645085
Mak et al., 1999Hong Kong1992–1996745RetrospectiveBU2435US$ 295n.a.57
Berto et al., 2000I19962,307, all agesRetrospectiveBU1164US$ 1547147012
de Zelicourt et al., 2000F19981,942, all ages; newly diagnosedProspectiveBU8 (1 yr)68 (1 yr)US$ 2432 (1 yr)2450n.a.
40 (2 yr)40 (2 yr)US$ 640 (2 yr) n.a.
Kostopoulos et al., 2003NL1999116, all agesRetrospective/prospectiveBUn.a.n.a.€ 18732190n.a.
Al-Zakwani et al., 2003Oman2000486 older than 13 yrProspectiveBU2352US$ 1432n.a.n.a.
Tetto et al., 2002I2000525, all agesProspectiveBU65n.a.€ 19202340n.a.
Krishnan et al., 2004India2001182RetrospectiveBU43n.a.US$ 6391064
Beghi et al., 2004In.a.631, adultsProspectiveBU6019US$ 13021360n.a.
Andlin-Sobocki et al., 2005Europe20042,690,608, all agesModelBUn.a.n.a.PPPe 2598260055
Present studyGermany2003101ProspectiveBU5928€ 4040f402562

Comparing the results of this study with previously reported data is difficult because of the varying level and methods of data collection and patient recruitment. The costs per case differed significantly among countries (Begley et al., 1999a; Heaney et al., 2001; Kotsopoulos et al., 2001) and were highest in countries with the highest national health care expenditure per capita, such as Germany (Andlin-Sobocki et al., 2005) (Table 5). Moreover, samples drawn from specialist clinics, as was done in the present study, will be in danger to overestimate population costs because specialists are likely to see more complex cases (Langfitt, 2000). Unbiased, representative sampling from the population of interest would provide the most valid cost estimates. Considering the cost-of-illness studies after 1990 including adults, the European direct costs of epilepsy did not appear to have risen dramatically in this period of time as judged by the power purchasing parity given in Table 5. Indirect costs remained consistently the main contributor to the total costs of epilepsy across most of the studies in Europe and the USA (Table 5). This is contrasted by an Italian study (Berto et al., 2000) which reported indirect costs being responsible for only 15% of the total costs but only the indirect costs of working days lost due to hospitalizations and visits for epilepsy were estimated. This approach, however, did not consider the indirect costs of productivity losses due to part-time work and early retirement, which constituted the greatest part of the indirect costs in the present study. Similarly to the Italian study (Berto et al., 2000), lost workdays due to seizures were responsible for 16% of the total costs recorded in our patient group. Therefore, an underestimation of the indirect cost has to be assumed.

In conclusion, the direct costs of refractory epilepsy in adults attending the outpatient clinic of our German tertiary epilepsy center were below the indirect costs. Medication contributed the most to the direct costs and early retirement to the indirect costs. The costs of epilepsy were above average of the European costs of epilepsy and were influenced by seizure frequency, disease duration and seizure semiology. Being aware of the criticism of cost-of-illness studies (Drummond, 1992), the data of this study, nevertheless, add to the few data available about the differential costs associated with epilepsy in Germany. This pilot study was meant to project further studies comparing the German economic situation of patients with varying epilepsies treated in different settings (outpatient clinics vs. private practices; public vs. private hospitals) and to provide economic data for a decision analysis model to evaluate further costs and outcomes in the treatment of epilepsies. In addition, future studies are warranted to answer questions, such as whether the costs of newer AED are offset by fewer side effects leading to improvements in functioning as well as to reduction in early retirement rates and service use.