The economic impact of galantamine vs placebo: An analysis based on functional capacity in a Swedish cohort study

Abstract

Objective:

To analyse the economic impact of galantamine, based on basic activities of daily living (ADL).

Methods:

Data were derived from Swedish patients enrolled in a 6-month placebo-controlled trial of galantamine (GAL-INT-1; n = 80), and from the Kungsholmen–Nordanstig Project, a longitudinal study of 919 elderly persons in Sweden. Basic ADL were assessed using the Katz’ Index of Independence in Activities of Daily Living (ADL) (number of ADL lost [dependency in 0, 1–2, 3–4, or 5–6 ADL]). Costs were appraised based on regression analysis and on costs directly linked to ADL. Six-month costs for galantamine and placebo were calculated.

Results:

In the regression analyses, each increase in a Katz stage was associated with an annual cost increase of SEK 81,415–83,683 (∼€8000). Results were similar using stage-specific costs. Overall, there was a small, non-significant numerical cost benefit for galantamine indicating cost neutrality.

Limitations:

The small number of Swedish patients in the GAL-INT-1 study, which was not powered for economic outcomes, limits the statistical power of the analysis. In addition, long-term outcomes are difficult to assess in persons with dementia because of practical and logistical problems.

Conclusions:

The benefits of galantamine in patients with AD can be achieved with no increase in cost. Combined with positive effects in terms of outcome, treatment with galantamine can be regarded as cost-effective using a cost–consequence approach.

Keywords::

• Economic impact
• Functional capacity
• Galantamine

Introduction

The increasing worldwide prevalence and incidence of dementia, particularly Alzheimer’s disease (AD), is a great challenge for social and healthcare systems1. The worldwide prevalence of dementia in 2010 was estimated at 35.6 million people, with the numbers expected to double every 20 years2, and the global costs associated with dementia in 2009 were estimated at US$422 billion (€300 billion)3. The prevalence of dementia in Western Europe is estimated at 7.2% of persons aged >60 years, equivalent to 7 million individuals, and is predicted to increase to 13.4 million by 20502. The increase in the number of persons with dementia is largely due to an overall increase in the number of elderly individuals, particularly the ‘oldest-old’2; the size of the population aged ≥85 years in the coming years is expected to increase considerably relative to the population aged ≥65 years and to the working-age population (20–64 years)4.

In 2005, the total cost-of-illness of dementia in Sweden was estimated at SEK50.1 billion (€5.4 billion)5,6, an increase from SEK49.6 billion (€5.3 billion) in 20006,7. Given this large economic burden, any intervention that can potentially reduce healthcare costs and/or improve the quality-of-life of patients and their relatives is of great value. Several pharmacological interventions for the treatment of AD are currently available and a number of pharmacoeconomic evaluations have been published (reviewed recently by Cappell et al.8). Although comparisons of costs across studies are limited because of the different methods used, the results of these studies show a clear trend towards cost neutrality or small cost benefits. It is therefore of great interest to investigate the links between costs and specific outcomes such as cognition, functional capacity, and quality-of-life9,10. For example, the Swedish population-based Kungsholmen Project showed that institutionalized patients with dementia required considerably more assistance with activities of daily living (ADL) and overall care assistance compared with institutionalized elderly individuals without dementia11.

The cholinergic system of neurotransmission in the brain plays an important role in memory and learning and is impaired in AD12. One of the available treatments for patients with AD is the cholinesterase inhibitor galantamine, which enhances cholinergic function by competitively and reversibly inhibiting acetylcholinesterase13. Unlike other drugs in the cholinesterase inhibitor class, galantamine also potentiates cholinergic nicotinic neurotransmission by allosteric modulation of nicotinic acetylcholine receptors14. Given its dual effects on the cholinergic system, galantamine is an effective treatment for patients with AD15. In addition to significant benefits over placebo on cognitive function16–21, galantamine is significantly more effective than placebo with regard to patients’ activities of daily living (ADL)16–19,21, as well as their global condition16–18,20 and behavioural symptoms16,18.

It has also been shown that functional capacity in terms of ADL is a stronger predictor of costs than cognition22. Thus, the aim of the present study was to analyse the economic impact of galantamine in Sweden, based on basic ADL.

Methods

Data for the analysis were derived from two sources: Swedish patients enrolled in the placebo-controlled GAL-INT-1 study (n = 80)21, and patients aged ≥75 years with dementia in the Kungsholmen–Nordanstig Project (KNP).

GAL-INT-1 was a 6-month randomized, double-blind, placebo-controlled study in patients with mild-to-moderate AD21. Basic and instrumental ADL were measured using the Disability Assessment in Dementia (DAD), a 46-item questionnaire with a score range of 0–100²³.

The KNP’s resource utilization and cost database24, which has been used in several Swedish cost of illness studies of dementia7,25,26, is the most comprehensive cost database of dementia in Sweden. It includes resource use data within the framework of the Resource Utilization in Dementia instrument (RUD)27, such as living situation (home or institutionalized, home care services, home nursing care, hospital care, and informal care. Through the use of unit costs (Table 1), these items can be incorporated into overall costs. Thus, a great advantage is that it has a societal perspective including, among other aspects, costs of informal care. This cross-sectional database consisted of 919 persons 75 years and older, of whom 182 were classified as having dementia. Cost data were available for 162 persons with dementia, with only the cost drivers included. The costs reflect the situation in 2005, which was the reference year when the database was used in a Swedish cost-of-illness study26.

Table 1.  Unit costs.

Resource item	Unit cost (SEK^a)
Informal care	91/hour
Home care^b	350/hour
Institutional care^c	1185/day
Hospital care	3000/day

^aDivide by 10 to obtain approximate costs in €.

^bTotal costs includes all salary costs (wage, pension, holiday) and transportation during work.

^c‘Hotel costs’ (for accommodation and food) are excluded.

ADL were evaluated using the Katz’ Index of Independence in Activities of Daily Living, which includes six basic elements of ADL28. The DAD instrument was not included in the KNP but, as some of the components in DAD are included in Katz’ (basic ADL), it was possible to convert basic ADL items in DAD to items in Katz’, giving each person a converted Katz-stage for each visit. Thus, although the GAL-INT-1 study lacks resource utilization and cost data, it was possible to link costs of dementia care from the KNP to the corresponding states of ADL in GAL-INT-1.

The Katz’ instrument is widely used29–34, but methods of analysis vary. In its basic form, Katz’ is a hierarchic scale in which the loss of six ADL functions follows a specific order, leading to assignment of a cumulative Katz’ stage from A to G (A = no loss; B = bathing, C = dressing, D = toileting, E = transferring, F = continence, G = feeding). Some the patients, however, do not follow this progression, resulting in an ‘others’ category. Furthermore, there are two methods of categorizing each item, with two or three levels, where the latter can be transformed to the dichotomous version according to a simple schedule. Thus, an alternative way of analysing the Katz’ items is to group the capacity according to the number of basic ADL abilities lost, irrespective of order, most frequently by grouping the items into four categories: dependent on 0, 1–2, 3–4, and 5–6 activities. This latter approach has been used in large population-based studies, such as the National Long Term Care Survey in the USA35, and is more appropriate for the present analysis.

Statistically, the present analysis includes two approaches, the first based on regression (the cost that is linked to a shift of one level in basic ADL capacity according to Katz’), labelled as the ‘regression approach’, and the second based on the costs that are directly linked to the specific stage of ADL capacity in the KNP, labelled as ‘stage approach’. As the KNP cohort also consists of institutionalized persons, we present data on the whole sample of people with dementia and a sub-group of KNP patients living at home, the latter to better reflect the GAL-INT–1 study population. In total, therefore, four costing models are used.

Results on the economic impact of galantamine are presented as costs for galantamine and placebo over a 6-month period based on GAL-INT-1. At each study visit, each patient had a specific ADL-status assigned in terms of number of ADL abilities lost. Specific costs according to each of the four approaches described above were then linked to the patient- and visit-specific basic ADL-statuses, allowing a 6-month cost for each patient to be calculated based on one-way analysis of variance (ANOVA). Analyses included a cost for galantamine in Sweden of SEK3170.50 per 98 days.

Results

The distribution of the dependency in basic ADL of people with dementia in the KNP and in the Swedish sub-sample of GAL-INT-1 (after transformation from basic ADL items measured scored on DAD) are shown in Table 2. Functional capacity was, as expected for a selected study population, better in GAL-INT-1 than in the overall KNP population, although the sub-group of KNP patients living at home had similar functional capacity to those in GAL-INT-1.

Table 2.  Distribution in dependency in ADL (Katz’ items) (percentages) in the KNP and the Swedish sub-sample of GAL-INT-1 (the latter after transformation from DAD).

ADL lost	All patients in KNP with dementia	Patients in KNP withdementia living at home	GAL-INT-1
0	41.2	73.1	58.0
1–2	17.6	16.4	32.5
3–4	15.8	6.0	8.0
5–6	25.5	4.5	1.5
Total	100	100	100

Costs associated with stages of ADL capacity

In the regression analysis of number of ADL lost (adjusted R²= 0.35), the regression equation for the full KNP dementia population was:

For each increase in dependency category (dependency in 0, 1–2, 3–4, or 5–6 ADL), the annual increase in costs was SEK 83,683 (∼€8000) from a basic cost of SEK 221,906 (∼€21,000) (Table 3).

Table 3.  Annual costs (SEK)* of dementia in the KNP, based on regression analysis of number of ADL abilities lost.

ADL lost	All patients in KNP with dementia	Patients in KNP with dementia living at home
0	221,906	110,649
1–2	305,589	192,064
3–4	389,272	273,479
5–6	472,955	354,894

*Divide by 10 to obtain approximate costs in €.

For the KNP patients living at home, the equation resulted in a different distribution (Table 3):

Stage-related costs directly from the KNP database are shown in Table 4; the differences between 3–4 and 5–6 lost basic ADL abilities are smaller with these data than with the regression approach. The main reason for the somewhat higher cost associated with 3–4 vs 5–6 lost basic ADL abilities is a high cost for informal care for those with 3–4 lost abilities, while those with 5–6 lost abilities are more often bedridden.

Table 4.  Annual costs of dementia (SEK*) based on stage-specific costs according to number of ADL abilities lost and data from the KNP.

ADL lost	Direct medical costs	Direct non-medical costs	Informal care	Total costs (SD)
All patients
0	12,944	151,078	36,405	200,426 (184,101)
1–2	13,623	331,091	24,727	369,441 (133,816)
3–4	15,823	378,705	30,660	425,188 (76,135)
5–6	9,408	408,221	26,126	443,756 (56,844)
0–6	12,617	282,700	30,935	326,252 (175,487)
Patients living at home
0	13,446	43,728	48,487	105,660 (120,009)
1–2	27,218	162,571	33,547	223,336 (105,389)
3–4	18,293	116,070	166,075	300,438 (145,342)
5–6	21,460	141,863	132,860	296,183 (14,036)
0–6	16,355	70,415	57,185	143,955 (133,438)

* Divide by 10 to obtain approximate costs in €.

Economic impact of galantamine

Based on the regression analysis, there was a non-significant trend to lower costs in patients receiving galantamine compared with placebo, regardless of whether the ‘living at home’ or the ‘all patients’ costing model was used (Table 5). With the stage approach, the cost difference between galantamine and placebo was almost zero, again regardless of which costing model was used (Table 6).

Table 5.  Six-month costs per patient with AD with and without galantamine, based on the regression analysis model.

	n	Costs (SEK)*	95% CI
Living at home
Galantamine	48	79,774	51,605–107,943
Placebo	32	110,809	76,310–145,309
All patients
Galantamine	48	135,930	108,277–163,582
Placebo	32	166,111	132,243–199,978

* Divide by 10 to obtain approximate costs in €.

Table 6.  Six-month costs per patient with AD with and without galantamine, based on the stage-based costing model.

	n	Costs (SEK)*	95% CI
Living at home
Galantamine	48	129,273	114,747–143,799
Placebo	32	129,381	111,591–147,172
All patients
Galantamine	48	185,387	168,248–203,427
Placebo	32	185,525	164,983–208,068

*Divide by 10 to obtain approximate costs in €.

Discussion

Results of these analyses show that costs associated with galantamine treatment of AD are similar using four different methods, and that galantamine treatment is cost neutral. Thus, the clinical benefits associated with galantamine on cognition, global condition, behaviour, and ADL can be achieved with no increase in overall costs. In addition galantamine has been shown to have beneficial effects for caregivers, such as reduced time spent in caring36, that are not accounted for in this study. Based on a health economic decisions rule matrix (Figure 1) with a cost-consequence approach, and given the range of clinically relevant outcomes for galantamine that have been demonstrated in randomized trials, there is potential for galantamine to be a cost-effective treatment for AD (Figure 1; Cell 4).

Figure 1.  Health economic decision matrix. T, new treatment; C, comparator (current standard of care, placebo, no treatment, etc.).

In most pharmacoeconomic evaluations, cognition—usually by the use of Mini Mental State Examination (MMSE)³⁷—has been used as the ‘vehicle’ of costs, such as in the construction of progression models38. While MMSE progression is indeed strongly correlated with increasing costs33,38 and decreasing health utilities39 in population-based studies, other measures, such as ADL40 and institutionalization41, have been suggested and used in trials to reflect the daily clinical situation more closely. This issue is, however, complicated. If the purpose is limited to analysis of costs in a pharmacoeconomic evaluation, the main criterion must be that the chosen vehicle has a strong correlation with costs, while in cost-effectiveness analyses it is important that the outcome is clinically relevant10. As the basic design of the present study was a cost analysis, we used ADL as the vehicle of costs, as recent studies have shown that ADL is a better predictor of costs than cognitive function22,42. From a cost-consequences viewpoint, considering the results of the present cost analysis in the context of the available clinical data for galantamine16–21, there is indirect evidence for cost-effectiveness of galantamine. Whether the differences in ADL that were found in the GAL-INT-1 study are meaningful in daily practice and care is, however, a complicated issue, and a cost analysis such as the present study cannot answer this question.

The outcomes of the present study, showing that galantamine is cost-neutral, are broadly consistent with other economic analyses of galantamine and alternative cholinesterase inhibitors in Swedish patients with mild-to-moderate AD43–47. For example, the AHEAD study looked at the economic impact of galantamine in Sweden in terms of delaying the need for full-time care42. The results of the model predicted that galantamine could reduce the time spent in full-time care by 10%, leading to overall cost savings. Similarly, reduced time to nursing home placement was shown in patients who responded to the cholinesterase inhibitor tacrine in a 5-year study of Swedish outpatients with AD44. Economic analysis of donepezil in Sweden based on disease progression (Progressive Deterioration Scale) and ADL (Instrumental ADL Scale) showed that mean annual healthcare costs were SEK137,752 (∼€15,300) per patient for the donepezil group and SEK135,314 (∼€15,100) for the placebo group⁴⁵. When caregiver time and healthcare costs were included, donepezil was associated with an annual saving of SEK9190 (∼€1000) over placebo.

Given the small number of Swedish patients in the GAL-INT-1 study, the statistical power of the analysis of the impact of galantamine was limited. Our study also shares the limitations of similar studies where within-trial data are used to analyse costs and cost-effectiveness. Even with greater samples, as in the donepezil study45, it was concluded that the study population was too small to detect significant economic differences, as the study was powered for efficacy rather than economic outcomes. Indeed, to our knowledge, none of the small number of published studies using empirical clinical data has demonstrated a statistically significant economic difference. Another problem is that it is difficult to accomplish long-term studies (e.g., 3–5 years) on persons with dementia because of practical and logistic problems. To detect long-term effects, analytic approaches such as economic modelling can be used. Given that small sample size and the short duration of the GAL-INT-1 study, the present analysis must be judged with care when attempting to generalize the results to wider populations.

As in many clinical trials, the GAL-INT-1 study did not include within-trial data on resource utilization and costs. Thus, the frequently used method of adopting external sources for the cost estimates was employed. On one hand, external data on costs from population-based studies, such as the KNP, probably better represent the societal state-related costs than trial study populations, which are more selected as a result of inclusion and exclusion criteria. On the other hand, if the discrepancy between the source population for the cost estimates (KNP) and the clinical trial population is large, the validity of the link between these two populations may be questioned. In this study we regard the differences between the two populations as acceptable, particularly between GAL-INT-1 and the KNP ‘living at home’ population.

The differences between the regression and direct stage-related approaches are probably a result of the rather small (‘all patients’ option) or almost zero (‘living at home’ option) cost differences between the 3–4 and 5–6 basic ADL abilities lost levels with the stage approach. This implicitly indicates that there may be intervention effects of galantamine between these two levels of basic ADL capacity that were better detected by the regression model.

In conclusion, based on four separate approaches, galantamine treatment over a 6-month period is cost-neutral, and may be cost-effective in the context of clinically relevant outcomes. The results are consistent with those for other available acetylcholinesterase inhibitors42–46.

Transparency

Declaration of funding

The GAL-INT-1 study was sponsored by Janssen Research Foundation, Beerse, Belgium.

Declaration of financial/other relationships

Anders Wimo has acted as consultant to drug companies that are purchasing or developing drugs for treatment of Alzheimer’s disease or other dementias (Janssen–Cilag, Pfizer, Astra–Zeneca, Novartis, Merz, Lundbeck, Forest, GSK, Wyeth, Sanofi, Elan, Neurochem). Maren Gaudig and Barbara Schäuble are employees of Janssen–Cilag EMEA. Erik Jedenius is an employee of Janssen–Cilag AB.

Acknowledgements

Medical writing assistance with this manuscript was provided by Daniel Booth (BioScript Stirling Ltd, London, UK) and funded by Janssen-Cilag EMEA. This analysis was made possible by unrestricted support from Swedish brainpower.