Brrtuh Journul of Ohwtriib rind Gyaecology September 1990, Vol. 97, pp 847-8.52
Vitamin B6 in the treatment of the premenstrual syndrome-a review JOS KLETJNEN,
GERBEN T E R RTET, PAUL KNIPSCHILD
Summary. A search of the literature yiclded 12 controlled trials on vitamin Bh in the treatment o f thc prcrnenstrual syndrome. These are discussed with cmphasis on rnethoclological aspects. A major drawback of the trials is the limited number of patients included. The existing cvidence of positive effects of vitamin B6 is weak, and some well-designed trials with positive results would be needed to change this view.
Many theories on the aetiology of the premenstrual syndrome (PMS) have been p r o p o d . including speculation about the effects of hormonal imbalances involving oestrogen. progesterone and prolactin, and about neuroendocrine dysfunction. Dysfunction of the pituitary gland production of endogenous opiates and/or ~:-melanocyte-stimulating hormone might explain most of the symptoms (Reid & Yen 1981). An alternative explanation might be decreased scnsitivity 01 progesterone receptors in sevcral organs. In any event, no generally accepted explanation of the aetiology currently exists. Various thcrapics have been advocated in the treatment of PMS. O n e of them is the use of megadoses of vitamin R 6 (pyridoxine). Pyridoxal 5’ phosphate, the active form of vitamin B6, serves as the coenzyme of a wide variety ol enzymes of amino acid metabolism (Friedrich 1988). It serves for instance as a cofactor i n the metabolism of tryptophan (the precursor of serotonin), and also in the metabolism of tyrosinc (leading to dopamine and noradrenalin) and glutamate (leading to gamma-aminobutyric acid). Low levels of dopamine and serotonin Department of Epidemiology/Health Care Research, University of Limburg, Postbox 616, 6200 MD Maastricht, The Netherlands J KLEIINEN
G TERRIET P KNIPSCHILD Pro/es&or Correspondeiice to J Klcijiicn
lead to high levels ol prolactin and aldosterone. explaining the fluid retention. Against this background, postulation of beneficial effects of vitamin B6 supplementation on both physical (fluid retention) and psychological (effect on neurotransmitters) symptoms related to PMS seems reasonable. However, iniplemeritation in regular practice can only be considered i1 clinical practice shows positive results. In a survey of 282 randomly chosen general practitioners in the Netherlands, 36% reported that they believed in the efficacy of vitamin Bh in PMS (Knipschild rl ul. 1990). The best way to show its efficacy is; of course, to perform randomized, placebo-controlled trials. The trials found after searching the literature are reviewed hcrc. Materials and methods
Identification ~ n srlecliori d of relevant studies Trials were only eligible if an index and a control group were includcd and if vitamin B6 was given to one of thc treated groups. Controlled studies in animal models were excluded. Experiments were found hy screening Tndex Medicus from 1966 to 1989 (keywords: pyridoxine-administration-and-dosage and pyridoxinc-thcrapeuticuse), and by screening Exccrpta Medica, section 10 (Obstetrics and Gynecology) from 19661989 (keywords: pyridoxine and vitamin B6). Further experiments were found by checking references extensively and by personal communication with authors of experiments. We also visited Hoff-
847
848
J . Kleijnen et al.
man-I,a Roche, in Hasel, in an attempt to asccrtain unpublished studies. Wc found 10 randomized double-blind trials on the effects of vitamin B6 in patients with PMS. In two other double-blind trials the effect of vitamin B6 on premenstrual mastalgia was studied. These 12 studies are listed in Table 1 and arc considercd in detail in Tables 2 and 3 . One study (Day 1979) was excluded because it is not really a cross-over study (no random allocation; all patients received vitamin B6 during cycle 2 , 3 and 4 and placebo during cycle 1 , 5 and 6). Another study was excluded because only one woman was treated (Mattes & Martin 1982). These two studies both suggcstcd positive results of vitamin B6 treatment.
Assessment of rnethoclologicrrl quality of published studies General criteria such as comparability of the treatment groups (prestratification, randomization); the interventions (adequate placebos) and the effect measurements (doublc blinding) should be met in every therapeutic trial. Special attention must be paid to some specific aspects of trials on PMS. The predominant symptoms of PMS are abdominal bloating, breast pain or tenderness, headache, oedema, fatigue, depression and irritability, starting somewhere in the second half of thc cycle, with rclicf at or soon after the onset of menstruation. A clear definition of PMS does riot exist, thus authors of trials should indicate what their inclusion criteria werc and which symptoms wcrc asscsscd in thcir investigation. An overall, global index of the complaints appears appropriate, since the impact of individual symptoms on the total discomfort may vary from patient to patient. An adequate description of the way in which these symptoms have been assessed (which questionnaire on depression, which scales for the assessment of pain etc.) is also important. Furthermore. it is advisable to extend the investigation over several cycles, especially in experiments with a limited number of patients, since there may he considerable variability of the severity of the symptoms both within patients in different cycles and between patients. To explore the possibility that an increasing likelihood of bias (with an incrcasing number of shortcomings) is reflected in the results of the trials, we used the following criteria lor a methodological assessment of the experiments:
(A) Characteristics of the women adequately described (description of symptoms. duration, severity, and previous treatments). (B) A t least 50 women in each group. (C) Prestratification (matching) on relevant prognostic factors. (D) Randomization. (E) Check of prognostic comparability. (F) Subjects blinded. ((3) Assessors blinded. (H) Effect measurements well described (reproducible). (I) Use of other concurrent treatments or vitamins has been checked and (J) presentation of the results in such a manner that the analysis can he checked by the reader. It could be argued that other criteria should be used for the methodological assessment. However, we have selected well established methodological criteria, as have been described by Meinert (1986), and our assessment can be checked by the reader (Table 2). Results
The main characteristics and results of the studies are summarized in Table 1. The dosages which were used varied from 50 mg to 500 mg pyridoxine a day (the recommended daily allowance in healthy adults is 2mgiday). No relation to the results of the experiments can he seen regarding different dosage levels. 'The duration of the experiments varied from 1 to 12 cycles: most experiments lasted 2 or 3 cycles. Individual symptoms, o r scales assessing many symptoms, e.g., 47 in a scale developed by Moos (1969), were mostly used for the measurement of the treatment effects. These were in general reasonably w d l dcscribcd. However, the numbcr of women, mostly 20-40 in each group, was limited. 'rile only exception is thc multi-centre study report(-d by Williams ef al. (1985) (200 patients in eazh group). Overall treatment effects additional to the assesment of individual symptoms were measured in only three trials (Barr 1984; Hagcn et ul. 1985 and Williams et al. 1985). Only trivial or no side cffccts wcrc reported in the studies presented. Abraham & Hargrove ( 1080), Harr (1 984) and Hallman (1987) reported positive effects: hut generally the symptoms did not disappear cornpletely. Stokes & Mendels (1972), Williams et 01. (1985), Smallwood et al. (1986), Keiidall & Schnurr (1987), and Doll et ul. (1989) fourrd ambiguous effects; overall the results of these studies were disappointing. Colin (1982), H a g m
Vitamin 86 and PMS
849
Table 1 . Characteristics of controlled trials of vitamin Bh in the treatment of the premenstrual syndromc
Study
Positive result Abraham & Hargrove (1980)
Barr (1984)
Entry critcria
Smallwood et al 1986)
Kendall & Schnurr (1987)
Negative result Colin (1982)
Significant R-6 effect in 2" 3, 19 symptoms excess of placebo effect in 21 subjects
36 Cross-over
100
2' 2, 8 symptoms
32 Cross-over
300
Premenstrual tension, depression Medication and OC were allowcd
13 Cross-over 204/230 Parallel
50
Premenstrual niastalgia, paracetamol was allowed Moderate to scvcrc symptoms, no medication and no
42 Cross-over
200
29/26 Parallcl
150
-
1
I00
32, Cross-over
50
17115 Parallel, random? No medication and no 34, Cross-ovcr OC
500
No medication and no 19, Cross-over, OC random? No medication and no 34142 Parallel OC
300
Moderate to severe symptoms, n o medication, OC were allowed
-
Mastalgia
Hagen et al. (1985)
Malmgrcn et al. (1987) Van den Berg et ul. (1989)
Outcome data
500
OC Doll et al. (1989)
No. of cycles,* effcct measure&
25 Cross-over
Modcrate to severe symptoms, no Medication and no OC
(most patients wcrc not on any OC) Hallman RC Orcland Severe symptoms, no hormonal therapy (1987) Amhiguout r e d t Stoke\ & Mendels (1972) Williams era/. (1965)
Number and design
Daily dosagc (mg)
100
120
Improvement for 83% on B-6 and 28% o n placebo 2" 4, Significant 14 symptoms improvements found for 8 symptoms
8-12, Not clear, 47 symptoms 'disapppointing' 3. lmprovement for 82% 9 symptoms on 8-6 and 70% on placebo 2" 2. No cignificant pain, improvemcnt: tenderness positive trend for tenderness on B-6 2, Significant amount of 47 symptoms symptomatology remained (positive for behaviour, dizziness and nausea) 2" 3, Significant heneficial 9 symptoms effect on cmotional symptoms, no effect on other symptoms
2 1 months, pain, temperature 2* 2. scale of 6 symptoms 2* I, many symptoms 4. 47 cymptoms
Improvcment for 59% on B-6 and 60% on placebo lmprovement for 4% on B-6 and 21 % on placebo using visual analogue scale N o significant improvement of symptoms No diffcrcncc
2* K = two periods of K cycles (cross-ovcr). OC = oral contraceptives.
et al. (1985), Malmgren et al. (1987), and Van den Berg et al. (1989) could not find any positive effects. Table 2 shows that all trials had at least two methodological shortcomings. 'The mean 'score'
for the trials with positive, ambiguous and negative results is 4-5,6.2 and 4.1 respectively (maximum is lo), indicating that there is no trend in any direction according to the likelihood of bias (number of methodological shortcomings).
Table 2. Mcthodological score assigned to the 12 trials Methodological criteria Stlldy
Positive result Abraham & Hargrovc (1980) Barr (1984) Hallman & Oreland (1c)87) Ambigu:uou\ rcmlt Stohes 6r Mendels (1972) Willidms ef al. (1985) Smallwood el at. (1986) Kcnddll6r Schnurr ( 1987) Doll cl ul (1989) Ncgative result Colin (1 982) Hagen et al. (1985) Malmgren er al. (1987) Van dcn Bcrg e t a ! . (1989)
A
B
C
D
i
-
-
+
~
* - -
_
_
? + -
+ -
&
-
-
_ t -
+ +
E
+
+
-- + - +
+ +
F
-
+ +
+
+ +
+
-
+
_
-
+
-
+
+
-
-
-
+
-
+
+
+
-
-
+
-
+
+
-
-
-
+
+
+
+
-
+
.
5
-
-
t
-
6
+ -
TOT
J
+
-
-
+ +
+
-
?
+ + +
-
+
+ + +
7
+ + +
+
I
+ -
+
+ + +
+ +
H
ti
3 4
3
+
8 6.5
k
7
+
6.5
-
-
2.5
t
-
-
s 3.5 5.5
Explanation of scores: +, item is scored; -, item is not scored; t . item is partly scored; TOT, number of items which were scored. A, Characteristics of subjects adcquately described (description of symptoms. duration, sevcrity, and pi-cvious treatments); B. at leas150 women per group; C. prestratification (matching) on relevant prognosticfactors: I > , randomization: E. check of prognostic comparability; F, subjects blinded; G , assessor blinticd; H, cffcct measurement well described (reproducible); I. use of other concurrent treatments or vitamins has been checked; J , prcscntation of the results in such a manner that thc analysis can he checked by the reader.
Table 3 shows which symptoms improved and which symptoms were not influenced by vitamin B6 treatment in the trials reporting on individual symptoms. The variation between trials turns out to he large and no trends are visible.
Discussion Although PMS is prevalent among women o f child bearing age. the number of participants in most studies was disappointingly low. Furthermore, compliance. blinding and prognostic coniparability m7as checked in only 3 , 0 and 1 trials respectively. It could be argued that in cross-over studies fewer subjects need t~ be admitted because each subject serves as her own control. In these designs, howwer, the basic assumption that the women arc prognostically comparable at the cross-over to the alternativc treatment may no( be met (inadequate washout, carry-over effects). In the eight cross-ovcr studics considered in this rcvicw only Hallman & Oreland (19S7) and Hagen rt al. (1985) included wash-
out periods, unless one assumes that the post menstrual phase of 2 weeks is sufficient to assure prognostic comparability of the groups at the beginning of the second treatment phase. Estimation of the bias introduced by carry-over effects is very difficult. Interestingly of the four studies with a parallel group design, two showed negative results and two ambiguous results. All positive assessments rcsulted from cross-over ~tudies. We did not pool the results of the trials. The methodological quality must obviously be high before statistical pooling makes sense. Morcover, there werc differences in the characteristics of the women, the dosage of vitamiii Bh, the duration of the intervention and the outcome measuremcntb. Publication bias is important in every review article. Trials with negative results are less likely to be submitted and published than trials with positive results, especially if limited numhers of subjects have participated. 'I'hk might well be the case for ti-ials o f vitamin €56 in PMS. Thc magnitude of this bias is difficult to estimate.
Vitamin B6 and PMS
851
Table 3. The effects of vitamin B6 on symptoms of thc premenstrual syndrome
Study
Positive effccts
Hagen et al. (1085)
Williams et al. (1985)
Smallwood er ul. (1986) Kendall & Schnurr (1987) IIallmann & Oreland (1987)
Breast tenderness Dizziness, nausea, poor performance. social activities Depression, irritability, swollenness, headache, hyperactivity. clumsiness, skidhair changes
Van dcn Bcrg e f al. (1989)
Doll et al. (198Y)
Depression, irritability, tiredness
No positive or ambiguous effects of vitamin B6 have been observed in controlled trials on related subjects, such as the treatment ofdepression related to the use of oral contraceptives (Adams et al. 1973, Adams el al. 1974), inhibition of lactation (Foukas 1972, Foukas 1973, Macdonald et al. 1976, Canales et ul. 1976), and in nausea and vomiting of pregnancy (Hesseltine 1946, General Practitioner Research Group 1963. Hillman et al. 1963). In future studies, with larger groups of wdmen, more attention must be paid to assuring prognostic comparability of the treatment groups, and to checking of compliance and blinding. A question remains about which dose of the vitamin should be used, what the optimal duration of treatment is, and whether perhaps only special subgroups (e.g., with relative deficiencies of vitamin B6) might benefit from vitamin R6 supplementation. At the moment there is no evidence that vitamin B6 is efficacious in the treatment of pdtients with PMS.
Acknowledgments This work was supported by a grant from the Dutch Ministry of Welfare, Public Health and Cultural Affairs.
No difference
Depression, irritability, bloatedness, weight gain. headache, breast tenderness, fatigue Depression, irritahility, swelling/ bloating. headache, brcast tenderness, tension, violcnt feelings, coordination Breast pain Many including: depression, anxiety Tension, anxiety, apathy, sleeplessness. tiredness, appetite Many including: depression, irritability, bloating, weight gain, hcadache, mastalgia, tension, anxiety Swollcnness, headache, breast discomfort, stomach cramps, backache
References Abraham G. E. & Hargrove J. T. (1980) Effect of vitamin B6 on premenstrual tension syndromes: a double blind crossover study. Infertirity 3,155-165. Adams P. W., Wynn V.. Kosc D . P., Seed M.. Folkard J . & Strong R. (1973) Effect of pyridoxine hydrochloride (vitamin Bh) upon depression associated with oral contraception. Lancet i, 897-904. Adams P. W., Wynn V.. Sccd M. & Folkard J . (1974) Vitamin B6, depression, and oral contraception. Lancer ii, 516-517. Barr W. (1984) Pyridoxinc supplcments in the premenstrual syndrome. Practitioner 228, 425-427. Canales E. S., Soria J . , Zirate A , , Mason M. & Molina M. (1 976) The influencc of pyridoxine on prolactin secretion and milk production in women. Rr .I Obsret Gynaecol83,387-388. Colin C. (1982) Etudes contr61Ces de I’administration orale de progcstagenes, d’un antioestrogkne et de vitamine B6 dans le traitement des mastodynies. Rev Med Brux 3, 605-609. Day J. B. (1979) Clinicaltrkals in the premenstrual syndromc. Curr Med Res Opin 6 , suppl. 5,40-45. Doll H., Brown S., Thurston A. & Vesscy M. (1989) Pyridoxine (vitamin B6) and the prcmcnstrual syndrome: a randomized crossover trial. J R Coll Gen Pract 39, 364-368. Foukas M. (1972) Laktationsheinmende Wirkung dcs Pyridoxins. Dtsch Med Wochenschr 97, 396-397.
852
J . Kleijncn ct al.
Foukas M. D. (1973) An antilactogenic effect of pyridoxine. J Ubstet Gyriuecol Br Cornmonvtmltli 80, 7 18-720. E'ricdrich W. (19x8) Vitamin H6. In Vitamins, dc Gruyter. Berlin, New York. pp 543-618. General Practitioner Research Group (1963) Meclozine and pyridoxine in pregnancy sickness. F'riictirioner 190, 25 1-253. Hagen 1.: Nesheim B-I. & Tuntland T. (1985) N o effect of vitamin B6 against premcnstrual tension. A controlled clinical study. Actu Ohstet Gynecol Scuntl64, 667-670. Hallman J. & Oreland L. (1Y87) Therapeutic cffect of vitamin B6 i i i the treatment of prcnienstrual syndrome. A double-blind cross-over study. Comprehensive Sunitnuries of IJppsalu Di.ssertaiion.s from the Faculty of Medicine 88, 1-15. Hcsseltine H. C. (1946) Pyridoxine failure in nausea and vomiting of pregnancy. Am J Ohstet Gynecal 51, 82-86. Hillman R. W., Cabaud P. G., Nilsson D. E.. Arpin P. D. & Tufano R. J . (1963) Pyridoxine supplcmcntation during pregnancy. Ani J C h i Nutv 12, 427-330. Kendall K. F.. & Schnurr P. P. (1987) The effects of vitamin B6 supplementation o n premenstrual symptoms. Ohstet Gynecol70, 145-149. Knipschild P., Klcijnen J. & ter Riet G. (1900) Belief in the efficacy of alternative medicine among gcncral practitioners in the Netherlands. Soc Sci Med, in press. Macdonald H. N . , Collins Y. D.: Tobin M. J. W. &
Wijayaratne D. N.(1976)'l'hc failure of pyi-idoxinc in suppression of pucrperal lactation. Rr .I Ohstet C;l'nnecol83, 54-55. Malmgren K.,Collins A . & Nilcson C-G. (1987) Platelet serotonin uptake and effects o f vitamin B6treatment in premenstrual tension. Neuropsychubiology 18, 83-88. Mattes J. A . & Marlin D. (1982) Pyridoxine in premrnstrual deprcssion. Hum Nutr Appl Nutr 36A, 131-133. Meinert C. L. (1986) Clinicul Tricrls. Design. Cunt-lucf and Ariulysis. Oxford University Press. New York; Oxford, pp 65-70, Moos K. H . (1 969) Typology of menstrual cyclc symptoms. A m J Obstei Gynecol 103, 300-402. Reid R . L. & Yen S. S. C. (1991) Premenstrual syndrome. A m J Ohstet Gynecol 139, 85-104. Smallwood J . . Ah-Kye D. & Taylor I . (1986) Vitamin R6 in the treatment of prc-menstrual mastalgia. Br J Clin Pruct 40, 532-533. Stokes J . & Mcndcls J. (1 972) Pyridoxine and prerncnstrual tcni;ion. Lancet i, 1177-1178. Van den Berg H., Schrijver J.. Rruinsc H. W. & Van der Plocg H. M. (1989) Vitamin R6 and premenstrual syndrome (PMS). Voeding 50, 58-62. WilliamsM. .J.,tIarrisR.I. & DeanB. C. (1985)Controlled trial o f pyridoxine in the premenstrual syndrome. .JZnl Med R e s 13, 174-179.
Received 2 Ocivber 1989 Accepted 14 Muy 1990
Vitamin B6 in the treatment of the premenstrual syndrome-a review JOS KLETJNEN,
GERBEN T E R RTET, PAUL KNIPSCHILD
Summary. A search of the literature yiclded 12 controlled trials on vitamin Bh in the treatment o f thc prcrnenstrual syndrome. These are discussed with cmphasis on rnethoclological aspects. A major drawback of the trials is the limited number of patients included. The existing cvidence of positive effects of vitamin B6 is weak, and some well-designed trials with positive results would be needed to change this view.
Many theories on the aetiology of the premenstrual syndrome (PMS) have been p r o p o d . including speculation about the effects of hormonal imbalances involving oestrogen. progesterone and prolactin, and about neuroendocrine dysfunction. Dysfunction of the pituitary gland production of endogenous opiates and/or ~:-melanocyte-stimulating hormone might explain most of the symptoms (Reid & Yen 1981). An alternative explanation might be decreased scnsitivity 01 progesterone receptors in sevcral organs. In any event, no generally accepted explanation of the aetiology currently exists. Various thcrapics have been advocated in the treatment of PMS. O n e of them is the use of megadoses of vitamin R 6 (pyridoxine). Pyridoxal 5’ phosphate, the active form of vitamin B6, serves as the coenzyme of a wide variety ol enzymes of amino acid metabolism (Friedrich 1988). It serves for instance as a cofactor i n the metabolism of tryptophan (the precursor of serotonin), and also in the metabolism of tyrosinc (leading to dopamine and noradrenalin) and glutamate (leading to gamma-aminobutyric acid). Low levels of dopamine and serotonin Department of Epidemiology/Health Care Research, University of Limburg, Postbox 616, 6200 MD Maastricht, The Netherlands J KLEIINEN
G TERRIET P KNIPSCHILD Pro/es&or Correspondeiice to J Klcijiicn
lead to high levels ol prolactin and aldosterone. explaining the fluid retention. Against this background, postulation of beneficial effects of vitamin B6 supplementation on both physical (fluid retention) and psychological (effect on neurotransmitters) symptoms related to PMS seems reasonable. However, iniplemeritation in regular practice can only be considered i1 clinical practice shows positive results. In a survey of 282 randomly chosen general practitioners in the Netherlands, 36% reported that they believed in the efficacy of vitamin Bh in PMS (Knipschild rl ul. 1990). The best way to show its efficacy is; of course, to perform randomized, placebo-controlled trials. The trials found after searching the literature are reviewed hcrc. Materials and methods
Identification ~ n srlecliori d of relevant studies Trials were only eligible if an index and a control group were includcd and if vitamin B6 was given to one of thc treated groups. Controlled studies in animal models were excluded. Experiments were found hy screening Tndex Medicus from 1966 to 1989 (keywords: pyridoxine-administration-and-dosage and pyridoxinc-thcrapeuticuse), and by screening Exccrpta Medica, section 10 (Obstetrics and Gynecology) from 19661989 (keywords: pyridoxine and vitamin B6). Further experiments were found by checking references extensively and by personal communication with authors of experiments. We also visited Hoff-
847
848
J . Kleijnen et al.
man-I,a Roche, in Hasel, in an attempt to asccrtain unpublished studies. Wc found 10 randomized double-blind trials on the effects of vitamin B6 in patients with PMS. In two other double-blind trials the effect of vitamin B6 on premenstrual mastalgia was studied. These 12 studies are listed in Table 1 and arc considercd in detail in Tables 2 and 3 . One study (Day 1979) was excluded because it is not really a cross-over study (no random allocation; all patients received vitamin B6 during cycle 2 , 3 and 4 and placebo during cycle 1 , 5 and 6). Another study was excluded because only one woman was treated (Mattes & Martin 1982). These two studies both suggcstcd positive results of vitamin B6 treatment.
Assessment of rnethoclologicrrl quality of published studies General criteria such as comparability of the treatment groups (prestratification, randomization); the interventions (adequate placebos) and the effect measurements (doublc blinding) should be met in every therapeutic trial. Special attention must be paid to some specific aspects of trials on PMS. The predominant symptoms of PMS are abdominal bloating, breast pain or tenderness, headache, oedema, fatigue, depression and irritability, starting somewhere in the second half of thc cycle, with rclicf at or soon after the onset of menstruation. A clear definition of PMS does riot exist, thus authors of trials should indicate what their inclusion criteria werc and which symptoms wcrc asscsscd in thcir investigation. An overall, global index of the complaints appears appropriate, since the impact of individual symptoms on the total discomfort may vary from patient to patient. An adequate description of the way in which these symptoms have been assessed (which questionnaire on depression, which scales for the assessment of pain etc.) is also important. Furthermore. it is advisable to extend the investigation over several cycles, especially in experiments with a limited number of patients, since there may he considerable variability of the severity of the symptoms both within patients in different cycles and between patients. To explore the possibility that an increasing likelihood of bias (with an incrcasing number of shortcomings) is reflected in the results of the trials, we used the following criteria lor a methodological assessment of the experiments:
(A) Characteristics of the women adequately described (description of symptoms. duration, severity, and previous treatments). (B) A t least 50 women in each group. (C) Prestratification (matching) on relevant prognostic factors. (D) Randomization. (E) Check of prognostic comparability. (F) Subjects blinded. ((3) Assessors blinded. (H) Effect measurements well described (reproducible). (I) Use of other concurrent treatments or vitamins has been checked and (J) presentation of the results in such a manner that the analysis can he checked by the reader. It could be argued that other criteria should be used for the methodological assessment. However, we have selected well established methodological criteria, as have been described by Meinert (1986), and our assessment can be checked by the reader (Table 2). Results
The main characteristics and results of the studies are summarized in Table 1. The dosages which were used varied from 50 mg to 500 mg pyridoxine a day (the recommended daily allowance in healthy adults is 2mgiday). No relation to the results of the experiments can he seen regarding different dosage levels. 'The duration of the experiments varied from 1 to 12 cycles: most experiments lasted 2 or 3 cycles. Individual symptoms, o r scales assessing many symptoms, e.g., 47 in a scale developed by Moos (1969), were mostly used for the measurement of the treatment effects. These were in general reasonably w d l dcscribcd. However, the numbcr of women, mostly 20-40 in each group, was limited. 'rile only exception is thc multi-centre study report(-d by Williams ef al. (1985) (200 patients in eazh group). Overall treatment effects additional to the assesment of individual symptoms were measured in only three trials (Barr 1984; Hagcn et ul. 1985 and Williams et al. 1985). Only trivial or no side cffccts wcrc reported in the studies presented. Abraham & Hargrove ( 1080), Harr (1 984) and Hallman (1987) reported positive effects: hut generally the symptoms did not disappear cornpletely. Stokes & Mendels (1972), Williams et 01. (1985), Smallwood et al. (1986), Keiidall & Schnurr (1987), and Doll et ul. (1989) fourrd ambiguous effects; overall the results of these studies were disappointing. Colin (1982), H a g m
Vitamin 86 and PMS
849
Table 1 . Characteristics of controlled trials of vitamin Bh in the treatment of the premenstrual syndromc
Study
Positive result Abraham & Hargrove (1980)
Barr (1984)
Entry critcria
Smallwood et al 1986)
Kendall & Schnurr (1987)
Negative result Colin (1982)
Significant R-6 effect in 2" 3, 19 symptoms excess of placebo effect in 21 subjects
36 Cross-over
100
2' 2, 8 symptoms
32 Cross-over
300
Premenstrual tension, depression Medication and OC were allowcd
13 Cross-over 204/230 Parallel
50
Premenstrual niastalgia, paracetamol was allowed Moderate to scvcrc symptoms, no medication and no
42 Cross-over
200
29/26 Parallcl
150
-
1
I00
32, Cross-over
50
17115 Parallel, random? No medication and no 34, Cross-ovcr OC
500
No medication and no 19, Cross-over, OC random? No medication and no 34142 Parallel OC
300
Moderate to severe symptoms, n o medication, OC were allowed
-
Mastalgia
Hagen et al. (1985)
Malmgrcn et al. (1987) Van den Berg et ul. (1989)
Outcome data
500
OC Doll et al. (1989)
No. of cycles,* effcct measure&
25 Cross-over
Modcrate to severe symptoms, no Medication and no OC
(most patients wcrc not on any OC) Hallman RC Orcland Severe symptoms, no hormonal therapy (1987) Amhiguout r e d t Stoke\ & Mendels (1972) Williams era/. (1965)
Number and design
Daily dosagc (mg)
100
120
Improvement for 83% on B-6 and 28% o n placebo 2" 4, Significant 14 symptoms improvements found for 8 symptoms
8-12, Not clear, 47 symptoms 'disapppointing' 3. lmprovement for 82% 9 symptoms on 8-6 and 70% on placebo 2" 2. No cignificant pain, improvemcnt: tenderness positive trend for tenderness on B-6 2, Significant amount of 47 symptoms symptomatology remained (positive for behaviour, dizziness and nausea) 2" 3, Significant heneficial 9 symptoms effect on cmotional symptoms, no effect on other symptoms
2 1 months, pain, temperature 2* 2. scale of 6 symptoms 2* I, many symptoms 4. 47 cymptoms
Improvcment for 59% on B-6 and 60% on placebo lmprovement for 4% on B-6 and 21 % on placebo using visual analogue scale N o significant improvement of symptoms No diffcrcncc
2* K = two periods of K cycles (cross-ovcr). OC = oral contraceptives.
et al. (1985), Malmgren et al. (1987), and Van den Berg et al. (1989) could not find any positive effects. Table 2 shows that all trials had at least two methodological shortcomings. 'The mean 'score'
for the trials with positive, ambiguous and negative results is 4-5,6.2 and 4.1 respectively (maximum is lo), indicating that there is no trend in any direction according to the likelihood of bias (number of methodological shortcomings).
Table 2. Mcthodological score assigned to the 12 trials Methodological criteria Stlldy
Positive result Abraham & Hargrovc (1980) Barr (1984) Hallman & Oreland (1c)87) Ambigu:uou\ rcmlt Stohes 6r Mendels (1972) Willidms ef al. (1985) Smallwood el at. (1986) Kcnddll6r Schnurr ( 1987) Doll cl ul (1989) Ncgative result Colin (1 982) Hagen et al. (1985) Malmgren er al. (1987) Van dcn Bcrg e t a ! . (1989)
A
B
C
D
i
-
-
+
~
* - -
_
_
? + -
+ -
&
-
-
_ t -
+ +
E
+
+
-- + - +
+ +
F
-
+ +
+
+ +
+
-
+
_
-
+
-
+
+
-
-
-
+
-
+
+
+
-
-
+
-
+
+
-
-
-
+
+
+
+
-
+
.
5
-
-
t
-
6
+ -
TOT
J
+
-
-
+ +
+
-
?
+ + +
-
+
+ + +
7
+ + +
+
I
+ -
+
+ + +
+ +
H
ti
3 4
3
+
8 6.5
k
7
+
6.5
-
-
2.5
t
-
-
s 3.5 5.5
Explanation of scores: +, item is scored; -, item is not scored; t . item is partly scored; TOT, number of items which were scored. A, Characteristics of subjects adcquately described (description of symptoms. duration, sevcrity, and pi-cvious treatments); B. at leas150 women per group; C. prestratification (matching) on relevant prognosticfactors: I > , randomization: E. check of prognostic comparability; F, subjects blinded; G , assessor blinticd; H, cffcct measurement well described (reproducible); I. use of other concurrent treatments or vitamins has been checked; J , prcscntation of the results in such a manner that thc analysis can he checked by the reader.
Table 3 shows which symptoms improved and which symptoms were not influenced by vitamin B6 treatment in the trials reporting on individual symptoms. The variation between trials turns out to he large and no trends are visible.
Discussion Although PMS is prevalent among women o f child bearing age. the number of participants in most studies was disappointingly low. Furthermore, compliance. blinding and prognostic coniparability m7as checked in only 3 , 0 and 1 trials respectively. It could be argued that in cross-over studies fewer subjects need t~ be admitted because each subject serves as her own control. In these designs, howwer, the basic assumption that the women arc prognostically comparable at the cross-over to the alternativc treatment may no( be met (inadequate washout, carry-over effects). In the eight cross-ovcr studics considered in this rcvicw only Hallman & Oreland (19S7) and Hagen rt al. (1985) included wash-
out periods, unless one assumes that the post menstrual phase of 2 weeks is sufficient to assure prognostic comparability of the groups at the beginning of the second treatment phase. Estimation of the bias introduced by carry-over effects is very difficult. Interestingly of the four studies with a parallel group design, two showed negative results and two ambiguous results. All positive assessments rcsulted from cross-over ~tudies. We did not pool the results of the trials. The methodological quality must obviously be high before statistical pooling makes sense. Morcover, there werc differences in the characteristics of the women, the dosage of vitamiii Bh, the duration of the intervention and the outcome measuremcntb. Publication bias is important in every review article. Trials with negative results are less likely to be submitted and published than trials with positive results, especially if limited numhers of subjects have participated. 'I'hk might well be the case for ti-ials o f vitamin €56 in PMS. Thc magnitude of this bias is difficult to estimate.
Vitamin B6 and PMS
851
Table 3. The effects of vitamin B6 on symptoms of thc premenstrual syndrome
Study
Positive effccts
Hagen et al. (1085)
Williams et al. (1985)
Smallwood er ul. (1986) Kendall & Schnurr (1987) IIallmann & Oreland (1987)
Breast tenderness Dizziness, nausea, poor performance. social activities Depression, irritability, swollenness, headache, hyperactivity. clumsiness, skidhair changes
Van dcn Bcrg e f al. (1989)
Doll et al. (198Y)
Depression, irritability, tiredness
No positive or ambiguous effects of vitamin B6 have been observed in controlled trials on related subjects, such as the treatment ofdepression related to the use of oral contraceptives (Adams et al. 1973, Adams el al. 1974), inhibition of lactation (Foukas 1972, Foukas 1973, Macdonald et al. 1976, Canales et ul. 1976), and in nausea and vomiting of pregnancy (Hesseltine 1946, General Practitioner Research Group 1963. Hillman et al. 1963). In future studies, with larger groups of wdmen, more attention must be paid to assuring prognostic comparability of the treatment groups, and to checking of compliance and blinding. A question remains about which dose of the vitamin should be used, what the optimal duration of treatment is, and whether perhaps only special subgroups (e.g., with relative deficiencies of vitamin B6) might benefit from vitamin R6 supplementation. At the moment there is no evidence that vitamin B6 is efficacious in the treatment of pdtients with PMS.
Acknowledgments This work was supported by a grant from the Dutch Ministry of Welfare, Public Health and Cultural Affairs.
No difference
Depression, irritability, bloatedness, weight gain. headache, breast tenderness, fatigue Depression, irritahility, swelling/ bloating. headache, brcast tenderness, tension, violcnt feelings, coordination Breast pain Many including: depression, anxiety Tension, anxiety, apathy, sleeplessness. tiredness, appetite Many including: depression, irritability, bloating, weight gain, hcadache, mastalgia, tension, anxiety Swollcnness, headache, breast discomfort, stomach cramps, backache
References Abraham G. E. & Hargrove J. T. (1980) Effect of vitamin B6 on premenstrual tension syndromes: a double blind crossover study. Infertirity 3,155-165. Adams P. W., Wynn V.. Kosc D . P., Seed M.. Folkard J . & Strong R. (1973) Effect of pyridoxine hydrochloride (vitamin Bh) upon depression associated with oral contraception. Lancet i, 897-904. Adams P. W., Wynn V.. Sccd M. & Folkard J . (1974) Vitamin B6, depression, and oral contraception. Lancer ii, 516-517. Barr W. (1984) Pyridoxinc supplcments in the premenstrual syndrome. Practitioner 228, 425-427. Canales E. S., Soria J . , Zirate A , , Mason M. & Molina M. (1 976) The influencc of pyridoxine on prolactin secretion and milk production in women. Rr .I Obsret Gynaecol83,387-388. Colin C. (1982) Etudes contr61Ces de I’administration orale de progcstagenes, d’un antioestrogkne et de vitamine B6 dans le traitement des mastodynies. Rev Med Brux 3, 605-609. Day J. B. (1979) Clinicaltrkals in the premenstrual syndromc. Curr Med Res Opin 6 , suppl. 5,40-45. Doll H., Brown S., Thurston A. & Vesscy M. (1989) Pyridoxine (vitamin B6) and the prcmcnstrual syndrome: a randomized crossover trial. J R Coll Gen Pract 39, 364-368. Foukas M. (1972) Laktationsheinmende Wirkung dcs Pyridoxins. Dtsch Med Wochenschr 97, 396-397.
852
J . Kleijncn ct al.
Foukas M. D. (1973) An antilactogenic effect of pyridoxine. J Ubstet Gyriuecol Br Cornmonvtmltli 80, 7 18-720. E'ricdrich W. (19x8) Vitamin H6. In Vitamins, dc Gruyter. Berlin, New York. pp 543-618. General Practitioner Research Group (1963) Meclozine and pyridoxine in pregnancy sickness. F'riictirioner 190, 25 1-253. Hagen 1.: Nesheim B-I. & Tuntland T. (1985) N o effect of vitamin B6 against premcnstrual tension. A controlled clinical study. Actu Ohstet Gynecol Scuntl64, 667-670. Hallman J. & Oreland L. (1Y87) Therapeutic cffect of vitamin B6 i i i the treatment of prcnienstrual syndrome. A double-blind cross-over study. Comprehensive Sunitnuries of IJppsalu Di.ssertaiion.s from the Faculty of Medicine 88, 1-15. Hcsseltine H. C. (1946) Pyridoxine failure in nausea and vomiting of pregnancy. Am J Ohstet Gynecal 51, 82-86. Hillman R. W., Cabaud P. G., Nilsson D. E.. Arpin P. D. & Tufano R. J . (1963) Pyridoxine supplcmcntation during pregnancy. Ani J C h i Nutv 12, 427-330. Kendall K. F.. & Schnurr P. P. (1987) The effects of vitamin B6 supplementation o n premenstrual symptoms. Ohstet Gynecol70, 145-149. Knipschild P., Klcijnen J. & ter Riet G. (1900) Belief in the efficacy of alternative medicine among gcncral practitioners in the Netherlands. Soc Sci Med, in press. Macdonald H. N . , Collins Y. D.: Tobin M. J. W. &
Wijayaratne D. N.(1976)'l'hc failure of pyi-idoxinc in suppression of pucrperal lactation. Rr .I Ohstet C;l'nnecol83, 54-55. Malmgren K.,Collins A . & Nilcson C-G. (1987) Platelet serotonin uptake and effects o f vitamin B6treatment in premenstrual tension. Neuropsychubiology 18, 83-88. Mattes J. A . & Marlin D. (1982) Pyridoxine in premrnstrual deprcssion. Hum Nutr Appl Nutr 36A, 131-133. Meinert C. L. (1986) Clinicul Tricrls. Design. Cunt-lucf and Ariulysis. Oxford University Press. New York; Oxford, pp 65-70, Moos K. H . (1 969) Typology of menstrual cyclc symptoms. A m J Obstei Gynecol 103, 300-402. Reid R . L. & Yen S. S. C. (1991) Premenstrual syndrome. A m J Ohstet Gynecol 139, 85-104. Smallwood J . . Ah-Kye D. & Taylor I . (1986) Vitamin R6 in the treatment of prc-menstrual mastalgia. Br J Clin Pruct 40, 532-533. Stokes J . & Mcndcls J. (1 972) Pyridoxine and prerncnstrual tcni;ion. Lancet i, 1177-1178. Van den Berg H., Schrijver J.. Rruinsc H. W. & Van der Plocg H. M. (1989) Vitamin R6 and premenstrual syndrome (PMS). Voeding 50, 58-62. WilliamsM. .J.,tIarrisR.I. & DeanB. C. (1985)Controlled trial o f pyridoxine in the premenstrual syndrome. .JZnl Med R e s 13, 174-179.
Received 2 Ocivber 1989 Accepted 14 Muy 1990
