|
Introduction The
growing acceptance of non-conventional therapies (NCT) among the German population
conflicts with the practice of medical insurance companies to refund these treatments only
in individual cases and under special circumstances [1]. This is mainly due to the fact
that, generally, non-conventional therapies are not scientifically approved, yet. Between
representatives of conventional and unconventional therapies, there is a controversy
regarding the value of NCTs and a dispute over how their efficacy has to be shown.
The mounting cost problems of a health care system that is focused on conventional
medicine have increasingly led to a sober attitude towards the assessment of
non-conventional therapies. Thus, non-conventional medicine attracts the attention of
health insurance companies primarily when costs can be reduced by these treatments.
Against this background the Centre for Documentation of Non-Conventional Therapies
(CDN) initiated a special regulation, a so-called "Erprobungsregelung" according
to §§ 63-68 SGB V, in which nine BKK´s from North-Rhine Westphalia participate. Under
this agreement, unconventional diagnostic procedures and treatments are provided to
members of the nine BKK´s over a period of five years.
In this contribution, we report preliminary results of the scientific evaluation, which is
part of the agreement.
Purpose
A chronic and, in particular, a chronic progressive disease is a serious problem for
the affected person which may lead to a severe impairment of quality of life. The emphasis
of the NCTs applied in this study is on the improvement of quality of life through the
application of particular therapies. However, these NCTs claim to relieve symptoms as well
as to work at discovering and removing the underlying causes of disease. In addition, the
supporters of NCT argue that their holistic treatment approach is economical. However, up
to now there is little evidence that a nonconventional medicine is more economical than
the established therapies in general [3-6].
The summary goals of this study are to investigate the postulated cost reduction by
nonconventional therapies, and the therapeutic effects of these treatments in patients
whose state of disease cannot be improved by conventional medical therapies. We conducted
a prospective observational study to evaluate 1) the costs of selected non-conventional
therapies according to a cost-outcome description [7], and 2) the long-term effects on the
patients’ health.
Because of practical and ethical restrictions a randomized study would not have been
feasible in this study population. Futhermore, for practical reasons we have not included
a control group up to now. So, in our context, a pre-/post-comparison within the
individual patients seemed to be the only way of obtaining answers to the questions
formulated above. Note that a major bias from spontaneous regression of disease is
unlikely given the selection criteria for the study population (long-term chronic
condition).
The number of sick-leave days, stays in hospital or in health resorts, and costs of
treatment (doctors, dentists, drugs, material) were chosen as the primary outcome
variables. Secondly, we examined the changes in frequency and severity of the symptoms of
disease according to the assessments of the treating physicians and the patients. The
variables were ascertained retrospectively (up to 5 years before the beginning of the
study), as well as prospectively (up to 5 years after the beginning of the study).
Methods
The prospective observational study described here is divided into two phases. A first
phase of orientation, from July 1994 to July 1996, was conceived as a feasibility study.
Its purpose was to test the instruments of assessment and to obtain information regarding
their validity and reliability. The second phase consists of the main study.
At the moment 40 general practitioners, one internist, one ear, nose and throat
(ENT)-specialist, and nine dentists are participating in the study.
Inclusion criteria for the patients are the adherence to one of the involved company
health insurance funds and the presence of a chronic disease that has not been cured or
improved with conventional therapies, or for which, according to medical experience, a
cure or further improvement cannot be expected.
The diagnostic and therapeutic non-conventional methods were selected by a scientific
committee and authorized by the German Federal Insurance Office (BVA).
Sources of data were case report forms obtained from physicians, and three questionnaires
filled in by the patients (before and after the treatment, and a follow up 2 years later).
Routine data were received from the BKK´s, the invoice data from a particular accounting
organization for physicians (Privatärztliche Verrechnungsstelle, Mülheim, Germany).
A detailed and close monitoring of the participating practitioners was undertaken twice a
year.
The flow of information is explained in Figure 1. The data of the National Health Survey
(1991) in the framework of the Deutsche Herz-Kreislauf-Präventionsstudie (DHP) is
used for comparison [8].
Figure 1: Flow of information between the parties of the
study
Biometrical Methods of Evaluation
Data of all patients participating in the study were included in the analysis, even if
the treatment was broken off (intention to treat-analysis). Routine data from the
company health insurance funds for the drop-outs were continuously documented, in the same
manner as the data of patients who finished the study. For these cases, the last
observation carried forward-procedure (LOCF) was applied to the symptom score
completed by the patients. Patients below the age of 16 years were evaluated separately as
a special population. The costs and the sick-leave days were compared intra-individually
(before and after the inclusion into the study) and were tested for differences with the
Wilcoxon test for paired observations. The data of the questionnaires were compared
intra-individually at the time before and after the treatment as well as two years after
leaving the study (follow up). Differences in the continuous variables and sum scores from
the questionnaires, as well as the patients’ subjective assessment of severity and
duration of disease before, during and after the treatment, were tested with the Wilcoxon
test for paired observations.
Tests for intra-individual differences in qualitative data were done using the sign test.
Results
Given that this was a pilot study investigation with a short observation time and a
relatively small number of cases, the preliminary results described here must be
interpreted cautiously. At best they can show tendencies regarding the development of the
target variables under treatment.
Until now 477 patients have been included in the study and have been treated with NCT. The
following presentation of the results will focus on the analysis of the sick-leave days,
stays in the hospital, overall costs, and course of symptoms.
Distribution of age and sex
Mean age of the study group was 42 (SD = 18) years. With respect to the age and sex
distribution, the study group differed from the general German population. Thus, the
percentage of 30 - 60 year-old persons was higher than the average, while the age groups
<25 years and >70 years were underrepresented. Furthermore, women were somewhat
overrepresented in the study group (59%).
Social status and state of health
Here, we will discuss the results for the variables education, profession, the
assessment of health status, and the description of complaints, as surrogate markers for
social status and true state of health of the participants. The data of the DHP (last
ascertainment in 1991) were used for comparison. However, this comparison was restricted
to 25 - 69 year-old patients, this age group being the one investigated in the DHP.
After standardizing the study data by age and sex, the study group had, on average, a
higher social status than the DHP population. This is seen in a higher proportion of
qualified employees and a lower proportion of unqualified workers.
As expected, the largest differences to the DHP were found for health status. In
comparison with the DHP population, our patients participated in more sports, cared more
for their health, and were more convinced that one can influence one’s own state of
health. On the other hand, they considered their current state of health as much worse,
and symtoms such as backache, pain in neck and shoulders, feeling of weakness, weariness,
and being restless were more frequent than in the DHP population.
Main Diagnosis
For the majority of the patients (54%), backache (ICD-9 720-724), rheumatoid arthritis
(RA, ICD-9 714), migraine (ICD-9 346) and asthma (ICD-9 493) were documented as the main
diagnosis. With the exception of cardiovascular diseases, which were rare, the
distribution of diseases was similar to that found in the frequency table of diseases of
the German Federal Association of the company health insurance funds [9]. Allergies to
which no ICD code could be assigned were summarized as "allergies, otherwise not
categorizable". 50 patients were registered whose symptoms, according to the
diagnosis, were of allergic origin.
As for the chronicity of the disease, the mean duration of the disease at recruitment was
10 (SD = 9.8) years, the number of prior physician consultations generally varied between
20 and 100. However, in extreme cases (11 patients) more than 1000 contacts were noted. On
the average, six different therapies for the main diagnosis had been recorded before entry
in the study.
Sick-leave days
For the analysis of sick-leave, patients (employees) who were observed for less than
half a year before or after the admission to the study were excluded from the evaluation.
Using this criterion, 162 cases were evaluable. For each patient, sick-leave days during
the study were calculated and compared to the sick-leave days in the same time interval
prior to the entry into the study. The average observation time of the 162 evaluable
patients was 1.2 years each before and after the admission to the study.
As Table 1 shows, the average sick-leave was 27.6 days per year before the beginning of
the study. By comparison, the average number of sick-leaves days reported for the
individuals insured by German (west) company health insurance funds, was 21 [9]. There is
no apparent difference in the number of sick-leave days before and after the beginning of
the study.
Table 1: Changing pattern of sick leave days per patient and
year before and after admission to the study
| Sick-leave/Year |
n |
 ± Sem |
Q1-  - Q3 |
range |
| Retrospective |
162 |
27.6 ± 4.0 |
0 / 8 / 28 |
0 / 276 |
| Prospective |
162 |
28.2 ± 4.7 |
0 / 7 / 26 |
0 / 365 |
| Difference |
162 |
0.5 ± 4.5 |
-8 / 0 / 9 |
-252 / +289 |
Figure 2 shows the development of the average annual number of sick-leave days per
patient up to 5 years before and 2 years after the beginning of the treatment with NCT.
While there was a marked increase from 19 to 37 days before the beginning of the study,
this tendency was then reversed. Two years after the admission into the study the average
annual number of sick-leave days per person had dropped to 24.
However, the sick-leave data must be interpreted cautiously. First, the observation period
was fairly short. Also, the observed decline of the sick-leaves could be partly due to regression
to the mean. Furthermore, a causal attribution of a change in the number of sick-leave
days to the influence of therapy is unjustified given the complexity of the factors
involved. Thus, the fitness for work may be influenced by working conditions and workload.
Also, if the chronic disease is severe even a marked improvement of health does not
automatically mean that the patient is able to work. Apart from this, for one third of the
participants no sick-leave days were reported before the admission to the study. Clearly,
in this group a decrease in sick-leave days could not be observed.
Figure 2: Course of sick-leave days before and after the
beginning of the study per patient and year (mean ± sem;
the dashed line represents sick-leave days of 22 patients with a complete observation time
of 5 years before and 2 years after the admission to the study)
Longer periods of observation, a larger sample and more detailed analyses of sick-leave
days will be necessary to obtain valid results.
Stays in Hospital
Data on stays in the hospital were available for 414 participants. The following
results are based on 304 participants with an observation time of at least half a year
after inclusion in the study. As with the analysis of sick-leave days, for each patient
the period within the study was compared to the period (of identical length) before
admission to the study.
The average costs of stays in hospital before the beginning of the study (1989 -
1993) was 2 - 3.5 times higher than, e.g., the average cost per policy holder of the BKK
Krupp (one of the participating BKK´s). During the study, the costs recorded for the
study group remained higher than those documented by the BKK Krupp. However, in contrast
to the strong nation-wide increase in expenses for stays in hospital observed between
1989-1993 and 1994-1995 (at the BKK Krupp this increase was 38%), the cost for stays in
hospitals effectively declined in the study group. Thus, the average annual costs
per patient went down by DM 79, or 5.9%, compared to the reference period before the
admission to the study, see Table 2.
Similar statements can be made for the duration of stays in the hospital, though
the slight decline (-0.47 days/patient/year; Table 2) may be attributable to the effect of
the 2nd phase of the "Gesundheitsreform" (put into operation on Jan. 1, 1995),
after which a decline in the period of stays in hospitals per stationary case was to be
expected.
Likewise, the average annual number of stays in hospital per patient decreased by
15 compared to the period before the admission to the study.
Table 2: Duration and costs of
the stays in hospital before and after admission to the study
| |
± Sem |
range |
| Retrospective observation [N=304] |
| Hospital costs / year / patient [DM] |
1340 ± 222 |
0 - 42.766 |
| Duration of stay / year / patient [days] |
3.4 ± 0.6 |
0 - 108 |
| Prospective observation[N=304] |
| Hospital costs / year / patient [DM] |
1261 ± 289 |
0 - 49.182 |
| Duration of stay / year / patient [days] |
2.9 ± 0.6 |
0 - 121 |
| Difference between prospective and
retrospective observation, calculated for one year [N=304] |
| Hospital costs / year / patient [DM] |
-79 ± 245 |
-29751 / +30072 |
| Duration of stay / year / patient [days] |
-0.47 ± 0.58 |
-69 / +66 |
Analysis of total expenditures
The total expenditures consist of costs for hospital, ambulatory care, dentists (DMD),
material and medication. Data for 279 patients were evaluated.
During the first year of the treatment with nonconventional therapies the total costs
increased from just under DM 3000,- to DM 7500,- per patient and year (Fig. 3). This
upward trend was reversed during the second year, with a decline to DM 5000,-. A more
detailed analysis of the total costs is needed to obtain a better understanding of this
development.
As seen in Fig. 3, the expenditures for hospital stays (black columns) and medications
(white columns) already started dropping during the first year of treatment with NCT, a
trend which was amplified during the second year of the study. The total expenditures were
almost exclusively determined by the development of costs for ambulatory care, i.e., for
the physicians. In the first study year, the average annual costs for ambulatory care
(excluding dentists) amounted to DM 5022,-. Thus, these costs increased more than ten
times per patient in the first year of the study. However, these observations do not
permit a straightforward interpretation. A direct pre/post comparison of the ambulatory
costs is problematic because ambulatory costs were paid by different clearing systems.
Furthermore, a special payment system for the NCT was created to ensure a higher valuation
of the work of the practitioners (e.g. the anamnestic conversation is paid substantially
better within the regulation than within the normal clearing system.)
Another problem is that in many cases expensive conventional diagnostic tests and
therapies that were undoubtedly applied in the beginning of the disease, lay several years
in the past. A direct evaluation of these costs is not possible since expenditures of
ambulatory costs are stored only for two years by insurance companies. The comparison of
costs, therefore, has partly to rely on estimates and/or incomplete ascertainment of data.
This may explain why the ambulatory costs before admission to the study were even
considerably lower than the average costs per insured person of the BKK´s. (The
difference was 30% for insured persons and family members, and up to 50% for retired
persons, respectively).
Figure 3: Comparison of average total expenditures per
patient and year before and after the beginning of the study (n = 304)
Furthermore, it is common that the introduction of new therapies into the health care
system is accompanied by a cost increase. It is, therefore, of interest to examine whether
the total expenditures under NCT persist at this high level . Our first data show that the
sharp increase mostly occurred within the first months of the treatment, and that a cost
decrease (of about 50 %) can already be observed after this period, independent of
sex and age.
In the future, our investigations will focus on the long-term development of the
expenditures during and after the treatment with NCT. A direct evaluation of the
expenditures is only possible if a control group (treated by conventional medicine) is
included. If this is unfeasible, a comparison between the expenditures after the end of
the initial phase of the study and the follow up expenditures up to five years thereafter
might be conceivable. The question will then be if there is a further decline of the total
expense rate of NCT to or below the level of costs of conventional therapies.
Analysis of Symptoms
Data sets from 280 patients were analysed. The LOCF (last observation carried forward)
procedure was applied to cases off-study. The symptom score was formed by summing up the
frequency of reported individual symptoms weighted with the degree of severity of all
recorded symptoms.
Figure 4: Percentage course of the symptom score assessed by
the patients (the open circles indicate the median of n=56 patients with a complete
observation time of 18 months; the closed circles indicate the median ± standard error of the median of n=280 patient analyzed by
LOCF)
The greatest effect on the symptom score occurred during the first months. The average
decrease of the symptom score observed within the first six months was 56 %
(95 % C.I.: 50-62 %), see Fig. 4. Moreover, 18 months after the beginning of the
treatment with NCT the score still was below 50 % of the initial value. A similar
development of the symptom score was observed in a subsample of 56 patients with a
follow-up of at least 18 months (these were mainly cases with poor symptom scores).
The observed decrease of the symptom score is an indication that NCT may lead to a
sustainable (possibly only subjective) improvement of the state of health. However, this
explanation needs a confirmation by further follow-up and larger samples.
Conclusions
- Despite initial doubts and problems, practitioners involved supported the scientific
participation. They showed themselves open-minded and cooperative. Thus, it can be stated
that scientific investigations with practitioners in the field of unconventional medicine
are feasible.
- There was an excellent cooperation with the BKK´s participating in the study. Data from
the BKK´s (especially data for sick-leave days and costs for stay in hospital) were
readily available.
- The study instruments have proved suitable: The structure of the case report forms, the
questionnaires for patients, and the process of the procedure were optimized during the
test phase and approved by the scientific committee.
- As for the evaluation of expenditures, a direct pre-/post-comparison of costs is not
possible at the moment. A comparative evaluation of costs would require an internal
control group (such as is planned for the main study). Whenever, due to technical problems
and practical reasons, a control group is unavailable, the examination of the long-term
development of the expenditures during and after the treatment with NCT may be a
reasonable alternative method for measuring the effects of unconventional medicine. This
feasibility study has shown that intra-individual retrospective and prospective
comparisons may give some idea of the potential future role of unconventional therapies in
the health care system.
- The preliminary results and experience of this feasibility study suggest that the
continuation of the main study may be worthwhile, even if the commonly accepted
methodological principles of comparative studies (control group, randomization) cannot be
implemented. With an observation time of at least 5 years and a planned sample size of
5000, it can be expected that this study will yield detailed information and well-founded
results. We believe that such an observational study is overdue. Its results may help
reduce costs of treatment and thus contribute to better patient care, and it may
constitute a factual basis for a rational discussion about unconventional medicine.
References
- [1]
- Schneider-Danwitz A. & Glaeske G. (1996): Unkonventionelle
Heilmethoden in der Krankenversicherung. Die schillernde Welt der Wunderheiler und
Scharlatane. Ersk. 10/96: 353-8
- [2]
- Nagel, G.A. (1995): Zur wissenschaftlichen Überprüfung
unkonventioneller Mittel in der Onkologie: Warum und wie? In: Biometrie und
unkonventionelle Medizin. Antes G., Edler L., Holle R., Köpcke W., Lorenz R. &
Windeler J. (Hrsg). Biometrische Berichte, Bd. 3, Geidel H. (Hrg.). Schriftenreihe der dt.
Region der Internationalen Biometrischen Gesellschaft
- [3]
- Perger, F. (1977, ): Herdbelastung und Krankenstände in einem
Industriebetrieb. Arbeitsmed, Sozialmed, Präventivmed 12, 254-7
- [4]
- Groh, H-P. (1991): Können Naturheilkunde und Homöopathie einen Beitrag
zur Kostendämpfung bei der ambulanten kassenärztlichen Versorgung leisten? Dissertation,
Medizinische Fakultät Ulm
- [5]
- Berg W (1992): Ganzheitsmedizin. Die Betriebskrankenkasse 10/92: 564-8
- [6]
- Wiesenauer M (1992): Naturheilverfahren - Möglichkeiten und Grenzen. Die
Betriebskrankenkasse 5/92: 281-6
- [7]
- Drummond, G., Teeling Smith, N. Wells (1989): Wirtschaftlichkeitsanalyse
bei der Entwicklung von Arzneimitteln. OHE/MPS, Med.-Pharmazeut. Studienges. e.V., Bonn.
- [8]
- Deutsche Herz-Kreislauf-Präventionsstudie, Nationaler
Untersuchungssurvey (3. Erhebungsrunde 1995), gefördert durch d. Bundesminister f.
Forschung und Technologie, d. Bundesminister f. Jugend, Familie, Frauen und Gesundheit und
d. Bundesminister f. Arbeit und Sozialordnung.
- [9]
- BKK Bundesverband (1996): Krankheitsarten 1994. BKK Bundesverband, Essen
|
