Clinical Otolaryngoloqy 1979, 4, I 19-123
Meniere’s disease: lithium treatment (demonstration of placebo effect in a doubleblind cross-over trial) J. THOMSEN, P. BECH, S . PRYTZ, P. VENDSBORG AND K . ZILSTORFF Unicersity ENT Department, Rigskospitalet, Blegdamsziej 9, 2100 Copenhagen, Denmark
Accepted for publicutiolz 5 January 1978
THOMSEN J., BECHP., PRYTZ S., VENDSBORG P. & ZILSTORFF K.(1979) Clinical Otolaryngolocqy 4, I 19-123 Meniere’s disease: lithium treatment (demonstration of placebo effect in a double blind cross-over trial) Keywords MkniPre’s disease litlrium placebo e j k t
On the assumption that the Hydrops of the endolymphatic space present in MCniire’s disease is due to a defective transport of water and ions between the endo- and perilymphatic space, the authors in an open pilot study treated patients with this disease with lithium carbonate, since lithium, among its many biological effects, acts on the transport of ions across membranes.’ The results showed that in 70% the effect of the treatment was excellent as assessed by reduction in the frequency and severity of the attacks. The patients who responded positively in the open trial thereafter participated in a doubleblind cross-over trial with 6 months’ treatment with lithium and 6 months’ placebo-treatment.2 I n this trial it was not possible to demonstrate any difference between the effect of lithium compared with placebo. It was therefore concluded that the effect obtained in the initial open trial was in fact due to a placebo effect even if the results were not completely unequivocal. It turned out that some of the patients had such mild symptoms that their scores did not allow statistical calculations, and that the patients had a very long history of their disease, about 8-10 years. It was therefore decided to carry out a similar controlled study, this time with patients who had a shorter history of MCniitre’s disease, with more attacks allowing a more accurate statistical evaluation.
Method In order to participate in the trial the patients had to fulfil the following criteria: presence of typical attacks of fluctuating hearing loss, tinnitus and vertigo, often accompanied by nausea, o ~ o ~ - ~ ~ ~ z / ~ ~ / o$02.00 ~ o o -@ o 11979 1 9 Blackwell Scientific Publications
119
I20
J. THOMSEN E t
(41.
vomiting and pressure in the ear, with at least I attack every 2 weeks, a history not less than 6 months, but not longer than 5 years, normal renal, cardiac and thyroid function, and no obvious allergies. The patients should be considered psychologically normal, and willing to participate in the trial for the entire 12 months. They were furthermore examined in detail to exclude tumours or other pathology of the cerebellopontine angle. For a period of 3 months before treatment the patient filled in a dizziness questionnaire continuing the medical treatment initiated by their own otologist in order to ensure a reasonable amount of attacks to allow statistical evaluation, and to exclude placebo responders in the pre-test evaluation. The patients were informed that they were participating in a trial testing the efficacy of lithium, and that they would receive different concentrations of lithium in their tablets, but the number of periods at each lithium level was concealed. They were not informed that they might receive pure placebo medication. Each week all the patients filled in the dizziness questionnaire in which the frequency of the attacks as well as the duration and severity of the individual attack were recorded. All symptoms within the Mtniitre’s syndrome complex (hearing, tinnitus, dizziness, nausea, vomiting, pressure in the ears) were recorded. The patients own statement could be directly transferred to punched cards and the computer. At the end of the trial the patients’ scores could be compared for the 2 treatment periods, as well as with the pre-trial scores. The serum lithium concentration was checked every 2 weeks. All the patients were kept on serum lithium levels within the therapeutic range, for the manic-depressive psychosis (0.7-1.1 mmol/l). If the patient’s serum lithium concentration was found to be outside this range changes in the tablet intake were made; at the same time a patient in the placebo period had a similar correction of the tablet intake. This control and correction of the tablet intake was performed by a physician at the Psychochemistry Institute, who otherwise had nothing to do with the treatment of the patients. This paired correction was made to prevent the investigator from gaining insight into the patient’s current treatment period. Each month the investigator rated the patients as regards to any unwanted effect of lithium treatment.3 Furthermore, he tried to evaluate the patient’s present state of disease giving scores from 0-10depending on the severity of the disease (Table I ). After the termination of both treatment periods the patients were asked to give their opinion as to which period had been the best, i.e. the period in which they had had the fewest and lightest attacks. Audiometric examinations with pure tone audiogram, speech audiometry and recruitment test (Fowler and Metz) was performed at regular intervals.
Patients
Twenty-one patients participated in the trial: 29 men and 8 women. The average age was 5 2 years (ranging from 30 to 69 years). The duration of the disease was I year on average ranging from 6 months to 4 years. 13 patients suffered from right side disease, 16 from left side discase. Three patients dropped out: two patients because of undesirable effects of lithium
Mhniire’s disease
I21
Table I Investigators’ global rating scale of severity of hlhikre’s disease No Mknikre
0
Slight Meniire
3
I
1 Severe Meniire
2
“
Extremely severe Meniere
8 9 I0
(tremor, thirst and polyuria), the last dropped out after 2 months of placebo treatment due to lack of effect. The remaining 26 patients all completed the 12 months trial period.
Statistical analysis Non-parametric statistics were used throughout using the Wilcoxoii test (4).Two tailed levels of significance were used.
Results When assessing the patients’ scores, both total scores as well as the scores for the individual symptoms, we found no difference between lithium and placebo. However, when comparing pre-trial scoring with the scoring during the trial it appeared that 19 of 26 (73 %) had a marked reduction of both the frequency and/or severity of the attacks. This was especially evident concerning the vestibular symptoms, dizziness, nausea, vomiting, but less significant regarding hearing loss, tinnitus and pressure in the ears. Six patients were by and large free of symptoms in the entire 12 months (not including hearing loss), but 20 patients had sufficient scores to allow statistical calculation. The investigators pre-trial global evaluation of the severity of the disease (Table 2) showed a median of 8, ranging from 6-10. After 6 months of treatment (Table 2 ) the median in the placebo period was 3 (range 0-8), and in the lithium period 4 (0-10). There was thus no difference between the effect of placebo and lithium. The retrospective self-evaluation by the patient of the 2 treatment periods revealed that 15 patients preferred the placebo period, 7 the lithium period and the remaining 4 were in doubt.
I22
J . THOMSEN C t
d. Table z Investigators’ global Menikre rating of 26 patients
Median Range
Pretrial rating 8 6-10
Inter-trial rating : lithium placebo 4 3 0-10 0- 8
Twenty-one preferred the second treatment period, one the first treatment period, and 4 were in doubt. All these 5 latter cases had a treatment order placeboilithiurn. There was no difference in the severity of the disease, when assessing the scores in the different seasons of the year (Spring, Summer etc.). Audiometric testing did not reveal any difference between the lithium and placebo period, nor did it reveal any difference compared to pre-trial testing.
Discussion
An uncontrolled study can be used as a primary investigation of a drug or other method of treatment, but is worthless for a real evaluation of the therapeutic effect. This necessitates a controlled trial, which implies that in addition to the treated group there must be a control group. The investigation must be carried out doubleblind, i.e. neither the investigator nor the patient must be aware of which group the patient belongs to or what treatment he is receiving at any given time. It is probable that the bias due to the effect of the investigator on the patient is particularly prominent in the dizzy patient. The double-blind technique is therefore an absolute necessity in investigations of treatment of the dizzy patients, and any leak in the ‘blind’ must be considered disastrous. There are a number of possible designs for the controlled trial. Total randomization of the patients, so that they are randomly assigned to a treatment or control group is hardly possible for investigation of Mtnikre’s disease. The 2 groups established by this means will most probably not be comparable, especially if they are small. The use of randomized paired groups would be more effective. Here a treatment and a control-group which are comparable with regard to such factors as age, sex, duration and severity of the disease are collected. It is however, difficult to use this design in dizzy patients, in this case with MCnikre’s disease, since it is almost impossible to find completely comparable groups with respect to a number of factors which are difficult to define, such as mental constitution of the patient. The above mentioned technique of the investigation have some advantage over the cross-over technique since the 2 groups are observed simultaneously and spontaneous variations in the course of the disease are of less importance as is any failure of the blind technique. When completely comparable groups cannot be obtained, the cross-over technique is the most suitable; the patient is treated with the active drug and placebo for 2 consecutive periods
Minihe’s disease
123
in random order. The advantage of this method is that interindividual variations disappear and the treatment and control-groups are as identical as possible, since the patient serves as his own control. The disadvantage of the cross-over technique is that spontaneous variation in the course of the disease which indeed exist in MCniGre’s disease, will influence the result, because it is necessary to compare z separate periods of time. Another disadvantage of the cross over design is that both the active drug and placebo may continue to act after their withdrawal and thus have a disturbing effect in the subsequent period. In our trial, 21 patients preferred to the second treatment period. This may be due to some degree of spontaneous improvement, but could indeed also be a result of the medical interest, which is shown to the patient, which might have a curative effect in itself. We must therefore conclude, that lithium has no specific effect when compared with placebo in the treatment of MCniire’s disease. However, treating the patients did indeed improve 73% of the cases. Since we excluded all patients who had previously responded to conventional medical treatment, antihistamine, sedatives, vaso-dilators and diuretics, given by private otologists, our ‘success’ merely indicates that the application of a big hospital apparatus (elaborate examinations, frequent visits, individually prepared tablets etc) acts as a very potent placebo. The effect of most other treatment, whether medical or surgical might be caused by a placebo effect. On perusal of the literature very few controlled studies of the medical treatment of MCnitre’s disease have been published. No studies have so far demonstrated the effect L): surgery by a controlled study. No one has shown whether it is the actual shunt or decompression of the endolymphatic sac that actually gives relief to the patient, or whether it is surgery in itself that imposes an effect. We therefore feel that such controlled studies are needed, and such studies are in progress in our department.
References I
THOMSEN J., BECHP., GEISLER A., BALSLEV JORGENSEN M., RAFAELSEN 0.J., TERKILDSEN K., UDSEN J. &. ZILSTORFF K. (1974) Meniltre’s disease. Preliminary report of lithium treatment. Actu Otoluryngologicu 78, 59.
THOMSEN J., BECHP.,GEISER A., PRYTZs.,RAFAELSEN O.J. VENDSBORG P. & ZILSTORFF K. (1976) Lithium treatment of MCniltre’s disease. Results of a double-blind cross-over trial. Acta 0tolar.yngologicu 82, 294. 3 BECHP., THOMSEN J., PRYTZS., VENDSBORC P.R., ZILSTORFF K. & RAFAELSEN O.J. (1977) The profile and severity of lithium-induced side-effects in mentally healthy subjects. Neuropsychobiology in press. 4 SIEGEL S. (1956) Non-Puruniewic Statistics. McGraw-Hill, New York. 2
Meniere’s disease: lithium treatment (demonstration of placebo effect in a doubleblind cross-over trial) J. THOMSEN, P. BECH, S . PRYTZ, P. VENDSBORG AND K . ZILSTORFF Unicersity ENT Department, Rigskospitalet, Blegdamsziej 9, 2100 Copenhagen, Denmark
Accepted for publicutiolz 5 January 1978
THOMSEN J., BECHP., PRYTZ S., VENDSBORG P. & ZILSTORFF K.(1979) Clinical Otolaryngolocqy 4, I 19-123 Meniere’s disease: lithium treatment (demonstration of placebo effect in a double blind cross-over trial) Keywords MkniPre’s disease litlrium placebo e j k t
On the assumption that the Hydrops of the endolymphatic space present in MCniire’s disease is due to a defective transport of water and ions between the endo- and perilymphatic space, the authors in an open pilot study treated patients with this disease with lithium carbonate, since lithium, among its many biological effects, acts on the transport of ions across membranes.’ The results showed that in 70% the effect of the treatment was excellent as assessed by reduction in the frequency and severity of the attacks. The patients who responded positively in the open trial thereafter participated in a doubleblind cross-over trial with 6 months’ treatment with lithium and 6 months’ placebo-treatment.2 I n this trial it was not possible to demonstrate any difference between the effect of lithium compared with placebo. It was therefore concluded that the effect obtained in the initial open trial was in fact due to a placebo effect even if the results were not completely unequivocal. It turned out that some of the patients had such mild symptoms that their scores did not allow statistical calculations, and that the patients had a very long history of their disease, about 8-10 years. It was therefore decided to carry out a similar controlled study, this time with patients who had a shorter history of MCniitre’s disease, with more attacks allowing a more accurate statistical evaluation.
Method In order to participate in the trial the patients had to fulfil the following criteria: presence of typical attacks of fluctuating hearing loss, tinnitus and vertigo, often accompanied by nausea, o ~ o ~ - ~ ~ ~ z / ~ ~ / o$02.00 ~ o o -@ o 11979 1 9 Blackwell Scientific Publications
119
I20
J. THOMSEN E t
(41.
vomiting and pressure in the ear, with at least I attack every 2 weeks, a history not less than 6 months, but not longer than 5 years, normal renal, cardiac and thyroid function, and no obvious allergies. The patients should be considered psychologically normal, and willing to participate in the trial for the entire 12 months. They were furthermore examined in detail to exclude tumours or other pathology of the cerebellopontine angle. For a period of 3 months before treatment the patient filled in a dizziness questionnaire continuing the medical treatment initiated by their own otologist in order to ensure a reasonable amount of attacks to allow statistical evaluation, and to exclude placebo responders in the pre-test evaluation. The patients were informed that they were participating in a trial testing the efficacy of lithium, and that they would receive different concentrations of lithium in their tablets, but the number of periods at each lithium level was concealed. They were not informed that they might receive pure placebo medication. Each week all the patients filled in the dizziness questionnaire in which the frequency of the attacks as well as the duration and severity of the individual attack were recorded. All symptoms within the Mtniitre’s syndrome complex (hearing, tinnitus, dizziness, nausea, vomiting, pressure in the ears) were recorded. The patients own statement could be directly transferred to punched cards and the computer. At the end of the trial the patients’ scores could be compared for the 2 treatment periods, as well as with the pre-trial scores. The serum lithium concentration was checked every 2 weeks. All the patients were kept on serum lithium levels within the therapeutic range, for the manic-depressive psychosis (0.7-1.1 mmol/l). If the patient’s serum lithium concentration was found to be outside this range changes in the tablet intake were made; at the same time a patient in the placebo period had a similar correction of the tablet intake. This control and correction of the tablet intake was performed by a physician at the Psychochemistry Institute, who otherwise had nothing to do with the treatment of the patients. This paired correction was made to prevent the investigator from gaining insight into the patient’s current treatment period. Each month the investigator rated the patients as regards to any unwanted effect of lithium treatment.3 Furthermore, he tried to evaluate the patient’s present state of disease giving scores from 0-10depending on the severity of the disease (Table I ). After the termination of both treatment periods the patients were asked to give their opinion as to which period had been the best, i.e. the period in which they had had the fewest and lightest attacks. Audiometric examinations with pure tone audiogram, speech audiometry and recruitment test (Fowler and Metz) was performed at regular intervals.
Patients
Twenty-one patients participated in the trial: 29 men and 8 women. The average age was 5 2 years (ranging from 30 to 69 years). The duration of the disease was I year on average ranging from 6 months to 4 years. 13 patients suffered from right side disease, 16 from left side discase. Three patients dropped out: two patients because of undesirable effects of lithium
Mhniire’s disease
I21
Table I Investigators’ global rating scale of severity of hlhikre’s disease No Mknikre
0
Slight Meniire
3
I
1 Severe Meniire
2
“
Extremely severe Meniere
8 9 I0
(tremor, thirst and polyuria), the last dropped out after 2 months of placebo treatment due to lack of effect. The remaining 26 patients all completed the 12 months trial period.
Statistical analysis Non-parametric statistics were used throughout using the Wilcoxoii test (4).Two tailed levels of significance were used.
Results When assessing the patients’ scores, both total scores as well as the scores for the individual symptoms, we found no difference between lithium and placebo. However, when comparing pre-trial scoring with the scoring during the trial it appeared that 19 of 26 (73 %) had a marked reduction of both the frequency and/or severity of the attacks. This was especially evident concerning the vestibular symptoms, dizziness, nausea, vomiting, but less significant regarding hearing loss, tinnitus and pressure in the ears. Six patients were by and large free of symptoms in the entire 12 months (not including hearing loss), but 20 patients had sufficient scores to allow statistical calculation. The investigators pre-trial global evaluation of the severity of the disease (Table 2) showed a median of 8, ranging from 6-10. After 6 months of treatment (Table 2 ) the median in the placebo period was 3 (range 0-8), and in the lithium period 4 (0-10). There was thus no difference between the effect of placebo and lithium. The retrospective self-evaluation by the patient of the 2 treatment periods revealed that 15 patients preferred the placebo period, 7 the lithium period and the remaining 4 were in doubt.
I22
J . THOMSEN C t
d. Table z Investigators’ global Menikre rating of 26 patients
Median Range
Pretrial rating 8 6-10
Inter-trial rating : lithium placebo 4 3 0-10 0- 8
Twenty-one preferred the second treatment period, one the first treatment period, and 4 were in doubt. All these 5 latter cases had a treatment order placeboilithiurn. There was no difference in the severity of the disease, when assessing the scores in the different seasons of the year (Spring, Summer etc.). Audiometric testing did not reveal any difference between the lithium and placebo period, nor did it reveal any difference compared to pre-trial testing.
Discussion
An uncontrolled study can be used as a primary investigation of a drug or other method of treatment, but is worthless for a real evaluation of the therapeutic effect. This necessitates a controlled trial, which implies that in addition to the treated group there must be a control group. The investigation must be carried out doubleblind, i.e. neither the investigator nor the patient must be aware of which group the patient belongs to or what treatment he is receiving at any given time. It is probable that the bias due to the effect of the investigator on the patient is particularly prominent in the dizzy patient. The double-blind technique is therefore an absolute necessity in investigations of treatment of the dizzy patients, and any leak in the ‘blind’ must be considered disastrous. There are a number of possible designs for the controlled trial. Total randomization of the patients, so that they are randomly assigned to a treatment or control group is hardly possible for investigation of Mtnikre’s disease. The 2 groups established by this means will most probably not be comparable, especially if they are small. The use of randomized paired groups would be more effective. Here a treatment and a control-group which are comparable with regard to such factors as age, sex, duration and severity of the disease are collected. It is however, difficult to use this design in dizzy patients, in this case with MCnikre’s disease, since it is almost impossible to find completely comparable groups with respect to a number of factors which are difficult to define, such as mental constitution of the patient. The above mentioned technique of the investigation have some advantage over the cross-over technique since the 2 groups are observed simultaneously and spontaneous variations in the course of the disease are of less importance as is any failure of the blind technique. When completely comparable groups cannot be obtained, the cross-over technique is the most suitable; the patient is treated with the active drug and placebo for 2 consecutive periods
Minihe’s disease
123
in random order. The advantage of this method is that interindividual variations disappear and the treatment and control-groups are as identical as possible, since the patient serves as his own control. The disadvantage of the cross-over technique is that spontaneous variation in the course of the disease which indeed exist in MCniGre’s disease, will influence the result, because it is necessary to compare z separate periods of time. Another disadvantage of the cross over design is that both the active drug and placebo may continue to act after their withdrawal and thus have a disturbing effect in the subsequent period. In our trial, 21 patients preferred to the second treatment period. This may be due to some degree of spontaneous improvement, but could indeed also be a result of the medical interest, which is shown to the patient, which might have a curative effect in itself. We must therefore conclude, that lithium has no specific effect when compared with placebo in the treatment of MCniire’s disease. However, treating the patients did indeed improve 73% of the cases. Since we excluded all patients who had previously responded to conventional medical treatment, antihistamine, sedatives, vaso-dilators and diuretics, given by private otologists, our ‘success’ merely indicates that the application of a big hospital apparatus (elaborate examinations, frequent visits, individually prepared tablets etc) acts as a very potent placebo. The effect of most other treatment, whether medical or surgical might be caused by a placebo effect. On perusal of the literature very few controlled studies of the medical treatment of MCnitre’s disease have been published. No studies have so far demonstrated the effect L): surgery by a controlled study. No one has shown whether it is the actual shunt or decompression of the endolymphatic sac that actually gives relief to the patient, or whether it is surgery in itself that imposes an effect. We therefore feel that such controlled studies are needed, and such studies are in progress in our department.
References I
THOMSEN J., BECHP., GEISLER A., BALSLEV JORGENSEN M., RAFAELSEN 0.J., TERKILDSEN K., UDSEN J. &. ZILSTORFF K. (1974) Meniltre’s disease. Preliminary report of lithium treatment. Actu Otoluryngologicu 78, 59.
THOMSEN J., BECHP.,GEISER A., PRYTZs.,RAFAELSEN O.J. VENDSBORG P. & ZILSTORFF K. (1976) Lithium treatment of MCniltre’s disease. Results of a double-blind cross-over trial. Acta 0tolar.yngologicu 82, 294. 3 BECHP., THOMSEN J., PRYTZS., VENDSBORC P.R., ZILSTORFF K. & RAFAELSEN O.J. (1977) The profile and severity of lithium-induced side-effects in mentally healthy subjects. Neuropsychobiology in press. 4 SIEGEL S. (1956) Non-Puruniewic Statistics. McGraw-Hill, New York. 2
