It is not widely known that menstrual flow contains a relatively small proportion of blood compared with the total volume.1–3 Whole blood accounts for 30–50% of the total flow in most women,2 although considerable variations are sometimes seen. The extra fluid is likely an endometrial transudate. The menstrual flow also contains tissue fragments and debris from the breakdown of superficial endometrium. There is considerable difficulty assessing the volume of menstrual blood loss in women presenting with excessive menstrual bleeding.4 A high proportion of those women with a clinically convincing complaint do not have excessive blood loss as defined in research studies,5–7 but the majority are nevertheless treated for menorrhagia without specific attempt to quantify the blood loss.
A significant reason for this is that the most widely used blood loss assay8 requires laboratory facilities, training, and safety measures suited to work with corrosive fluids (sodium hydroxide), the use of a spectrophotometer, and a fume cupboard. However, a semiquantitative technique has been developed, the pictorial blood loss assessment chart9,10 that assesses the degree of saturation of used sanitary products. However, Reid et al have recently cast doubt on the accuracy of this method.11 Pendergrass et al previously reported simple measurements of menstrual pad and tampon weights, but these authors assumed that the entire menstrual loss could be accounted for by whole blood,12 an inaccurate supposition.2 Gannon et al have developed a variation in the extraction of hemoglobin using Triton X-100 detergent and a desktop washing machine.13 However, the need for extra equipment and for the safe disposal of the washing liquid makes it unlikely that this technique will be widely adopted. An alternative, simple, and reliable method of determining menstrual blood loss volume would be of great clinical value.4
For this study, we measured menstrual blood loss and total menstrual fluid volume from 53 women; blood loss ranged from low normal to convincing menorrhagia. Our aim was to examine the relationship between total fluid volume at menstruation and its whole blood component, and to examine the relationships between, and consistency of, blood loss and total fluid volume from one menstrual period to the next. Because total fluid volume is relatively easily obtained by weighing (after prevention of loss by evaporation), a further aim of this study was to ascertain whether this more convenient measure could provide a clinically satisfactory estimate of blood loss.
MATERIALS AND METHODS
Fifty-six women were recruited, but only 53 completed full menstrual collections for two periods. Women were recruited over a 12-month period from a hospital clinic, from a previous research study, and by local advertisement. The study sample of 53 women comprised: 14 with current complaint of menorrhagia, 34 without complaint of menorrhagia, and five who had previously attended the gynecology clinic for menorrhagia but did not currently complain of menorrhagia. These varied backgrounds ensured a spread of perceived and actual menstrual losses. All women were white, none had detectable gynecologic abnormality, and all gave a history of regular menstrual cycles (24–35 days). None complained of unusual vaginal discharge. No women had been sterilized or were using any hormonal or intrauterine method of contraception. Ethical approval was obtained from the Central Sydney Area Health Service Ethics Committee. “Regular” or “super” sanitary pads (without “wings”) and tampons were provided to all subjects with careful instruction on the way they were to be used in the study. Particular care was taken with these instructions.2 Women were instructed in the meticulous collection of all menstrual loss, including “clots” that might be lost in the toilet or shower. Menstrual hygiene products were accurately pre-weighed and placed in individually labeled self-sealing polythene bags until used. After use, each pad or tampon was immediately placed in its original bag, which was then resealed. Analysis of fluid and blood content was carried out within 3 days of the end of each period. Each pad or tampon was then accurately re-weighed, and the total fluid content was calculated by subtracting the original weight. It was assumed that specific gravity equals one. Hemoglobin content of each pad or tampon was separately measured by a semi-automated modification14 of the alkaline hematin technique of Hallberg and Nilsson.8 After measuring the hemoglobin concentration in a cubital vein sample collected a few days earlier, we converted the hemoglobin content to milliliters of whole blood. The method of storage of the pads and tampons was checked for evaporation as described previously.2 Women maintained a detailed diary card throughout each menstrual period collected. For each period, the total fluid volume, the total blood volume, and the blood component percentage (100 × total blood loss ÷ total fluid volume) were calculated.
Menstrual flow contains secretions of vaginal and cervical origin, so a separate small study was designed to assess the fluid content of pads and tampons worn when women were not menstruating. Over three separate 24-hour periods, 10 women wore pads and another 10 women wore tampons. Collections were made in the mid-proliferative and mid-secretory phases, and no differences were found in the measured volumes. The women collected these samples during the normal working week and did not take part in excessive physical activity during the collections. Pads and tampons were pre-weighed, stored, used, and analyzed exactly as described above.
Analysis of variance (ANOVA) was used for statistical comparison of intermenstrual vaginal secretion volumes. For menstrual loss volumes, the marked positive skew in the distributions for blood and total fluid volume required the data to be log-transformed before the parametric analyses describing relationships between them (linear correlation and regression). Where correlations have been calculated, the correlation coefficient (r), degrees of freedom (df), and the statistical significance (P) are reported. In some cases the relationship is also interpreted in terms of the proportion of variability, using the dependent variable y, which is explained by the regression, on the independent variable x (ie, r2 × 100%). Where differences between periods in the various volumes have been calculated and correlated, a constant was added before log transformation of the raw data (because the log function is undefined for values less than or equal to zero).
For the estimation of blood loss from the total measured fluid volume, a plot of the log-transformed data showed that agreement of the observed data points with the regression line was remarkably good for periods where losses were quite high, but less satisfactory among women with the lightest losses. This difference in agreement is most likely because nonmenstrual secretions contribute a relatively greater proportion of the total fluid when overall volumes are low, and these secretions are prone to quite marked between-subject variation. Because the main clinical application of any regression equation is likely to be the prediction of blood loss from total fluid volume in women complaining of heavy periods (who are therefore likely to have at least moderate losses), we optimized the precision of the regression equation by basing the analysis only on those subjects with at least one of their two periods yielding a measured total fluid volume of 85 mL or more. This fluid volume corresponds to a blood loss of about 40 mL.
In the analysis reported, data from both periods separately were used for each subject. However, very similar regression equations for estimating blood from total fluid were obtained if just data for the first or second period, or the average of the two periods, were used (r = .94, .92, .95, respectively).
Subjects were categorized according to the type of sanitary protection used, “pads only,” “tampons only,” or “both.” In the regression analyses, there were very few women using tampons only, so the latter two groups were combined into one “tampon user” group, the majority of whom also used some pads in combination with the tampons. A binary variable “tampon user” was created for use as a potential explanatory variable in the regression model: value 0 identified women using only pads as menstrual protection, and value 1 identified women who used tampons.
Three of the 56 women recruited completed only one collection; in one case (the only subject lost from the menorrhagia complaint group) this was because the menstrual blood loss was so excessive (497 mL) that the patient wished to proceed immediately with a planned hysterectomy. Measured blood losses and total fluid volumes for the remaining 53 subjects are shown in Figure 1. Most periods had a blood volume of less than 45 mL. There was a second smaller cluster with a blood volume of 70 mL, and the remainder was dispersed from 90 mL to 242 mL, yielding a distribution with marked positive skew. Lines have been drawn at 60 mL and 100 mL to define three categories of blood loss, “normal” (<60 mL), “moderately heavy” (<100 mL), or “excessive” (≥100 mL) (n = 80, 14, and 12, respectively; 76%, 13%, and 11% of the 106 periods measured, respectively). The distribution pattern for total fluid measured was very similar to that for blood, but at higher volumes it approximately doubled.
For the whole study group, there were strong linear correlations between blood and nonblood volumes, and even stronger correlations between total fluid volume and each of its two constituent parts, blood loss and nonblood fluid volume (r = .74, .93, and .93, respectively; df = 104; P < .001 for each correlation, all three variables were log transformed).
Measurements of intermenstrual vaginal fluid secretion demonstrated that there was a small but highly significant difference between the mean fluid volume for pad wearers (1.31 mL per 24 hours) compared with tampon wearers (3.89 mL per 24 hours) (ANOVA F1,40 = 232; P < .001; standard error of the mean [SEM] = 0.6). Thus the volumes of nonmenstrual fluid collected over a 5-day period were estimated to be 6.5 mL and 19.5 mL, respectively, a difference of 13 mL. However, this is a difference at the group level, and the analysis also revealed that differences between subjects within the same group (ie, using the same hygiene product type) were very marked in comparison with day-to-day variation for each individual woman (F18,40 = 6; P < .001). Individual mean 24-hour fluid volumes for subjects wearing pads ranged from 0.47 mL to 3.2 mL and for those wearing tampons from 2.7 mL to 5.4 mL.
The data from the main study were subjected to regression analysis to explore the use of total measured fluid volume to estimate blood loss volume. The restriction, for this analysis, to women with at least one of their two periods yielding a measured total fluid volume of 85 mL or more, excluded 29 subjects, all “normal” volunteers. Excluded subjects had a mean blood volume of 18.7 mL (SE 1.2, range 4.2–44.5 mL), mean total fluid volume of 48.1 mL (SE 2.3, range 14–81.6 mL), and mean percent blood content 39.4% (SE 1.8, range 12–80%).
The regression analysis, using data from both periods for each included subject (24 subjects, 48 data points), confirmed that there is a highly significant linear relationship between log-transformed blood loss volume and log-transformed total fluid volume (F1,46 = 282; P < .001; r = .93). Thus in this sample, 86% of the variation in blood loss volume is accounted for by variation in total fluid volume (both variables were log transformed).
The regression equation for the log-transformed data is:
which, after back transformation (antilog), yields the following equation for estimating blood loss from total fluid volume:
Figure 2 shows the scatter plot of data included in the analysis, with the calculated regression equation shown as a solid line (which, once back-transformed to the original scale, is imperceptibly curved). Subjects are identified according to the type of sanitary protection used, “pads only” or “tampons,” and the 90% confidence limits for individual estimates of blood loss, for the range of total fluid volumes, are also plotted. The data show that those with moderate loss using “pads only” were more often above the line, ie, they have a higher proportion of blood to fluid than subjects using tampons. Given the finding of a significant excess in vaginal secretion content in tampons compared with pads, found in the separate nonmenstrual fluid volume study, this tendency was expected.
The contribution of type of sanitary protection used was examined by assessing the goodness-of-fit of the regression when a second variable, “tampon user” or not, was included in the analysis. This assessment resulted in a small but significant improvement in the regression estimation, with variance accounted for by the multiple regression increasing from an already high 86%, to 89% (F2,45 = 183.3; P < .001; multiple r = .94).
The resulting regression lines for tampon users and women using pads only were parallel on the log scale. After antilog transformation the equations for estimating blood loss are, for women using pads only:
and, for tampon users:
Therefore, across the range of total fluid volumes, the blood loss estimate for a woman who uses tampons is lower than that for a woman using pads only, by a factor of 0.796.
Estimates of mean blood loss calculated from the total fluid loss measurements are shown in Table 1. Four (10%) of the true values for blood loss fell outside the 90% confidence limits for the estimated blood loss. These data pertain to just two individuals, both with relatively low total fluid volumes. One of these two subjects had an exceptionally high proportion of blood in her menstrual loss (an average of 82% blood, in total fluid volumes of 72 mL and 87 mL), and the other had a very low proportion of blood (average of 27% blood, in total fluid volumes of 98 mL and 111 mL). However, for total fluid volumes between 100 mL and 200 mL the estimates were good (with the exception of one period), and for total fluid volumes over 200 mL the estimates were excellent (ie, the true values are well within the 90% confidence limits). The single menorrhagia subject excluded from analysis because she had collected for only one period had a measured total fluid volume for that period of 1041 mL, which gave an estimate of blood loss volume of 587 mL, using the simple regression equation. Despite the fact that this large total fluid volume was well outside the range of the data used to calculate the regression equation, the true (actual measured) volume of blood loss, 497 mL, was close to the estimate obtained (587 mL), and well within the 90% confidence limits for that estimate (372–927 mL). There was no correlation between low hematocrit and low blood loss volume, and the variation in fluid content was greater than the variation in hematocrit.
We calculated the intercycle variability or percentage “period-to-period change” for each woman for both blood and total fluid loss variables, as the difference between measurements for the two periods for the same woman, expressed as a percentage of the average of the two periods. For women with at least moderately heavy periods (the 24 included in the regression analysis), percentage change in blood loss is shown in Figure 3. The median (and inter-quartile range) for period-to-period “percentage change” were 18% (11–31%) for blood, and 26% (13–31%) for total fluid. So approximately one quarter of these women had period-to-period changes of 31% or more in their volume of blood loss.
For the subset of subjects complaining of menorrhagia, the correlation between changes in total fluid volume and blood loss (log-transformed) was very strong (r = .95, df = 12, P < .001), despite the low degrees of freedom. Therefore, 90% of the within-woman period-to-period change in blood loss could be predicted by change in total fluid volume. Figure 4 plots individual lines for each menorrhagic subject (ie, joining the data points for her two periods), and shows the marked similarity across subjects in slopes of blood loss related to total fluid volume. An examination of the relationship between changes from one period to the next, in the various fluid volumes, controlled to some extent for individual differences between subjects in nonmenstrual vaginal fluid secretion and for the type of sanitary protection used. For the subset of subjects with changes in total fluid volume exceeding 30 mL (n = 13), the corresponding change in blood loss, as a percent of total fluid change, averaged 52.4% (SE 7.4).
The clinical application of the regression estimation of blood loss volume was addressed by classifying all 106 collected menses as “normal” (<60 mL), “heavy” (≥60 mL), or “excessive” (≥100 mL) on the basis of actual measured blood loss (as shown by dotted lines in Figure 1). Periods were also classified on the basis of predicted blood losses, calculated from total fluid volumes using the simple regression equation and then applying the same 60 mL/100 mL criteria. This cross-classification is presented graphically in Figure 5, where it can be seen that there was only one clear discordance. This reflects the normal volunteer who had a consistently high proportion of blood (82% and 83%) in both menstrual periods; in one period, the total fluid volume was 87 mL, giving a predicted blood loss of 41 mL, whereas her actual measured blood loss was 71 mL (greater than the semi-arbitrary cut-off of 60 mL). The other discordance was marginal, and the remainder of the periods, 104 out of 106 menses (98%), would be correctly classified by predictions based on their total fluid loss.
This study has confirmed that a high proportion of fluid in the menstrual discharge cannot be accounted for by blood content,1–3 and that this applies whether the woman has light or heavy flow. It was even seen in a woman who had gross menorrhagia with a measured blood loss of 497 mL, which accounted for only 47.6% of her total flow. We have also shown that measuring total menstrual fluid volume provides an accurate estimate of blood loss in women with moderate to very heavy losses (ie, at least one total fluid greater than 85 mL). Furthermore, this estimator is not subject to behavioral distortion, as was number of menstrual hygiene products needed,5,7 or perceptual bias, which can be a problem with visual methods where subjects rate the soaking of each hygiene product when removed.9,10,11
Another clinically important fact revealed by these data is the extent of variability in loss from period to period, particularly in women complaining of menorrhagia. In our study, seven subjects had changes in blood loss of over 25 mL, and they were predominantly in the “complaint” group (n = 5). This finding conflicts with the belief that women's periods are remarkably consistent from cycle to cycle,15 but is supported by report of median coefficient of variation of 30–40% in menorrhagia.16 This variation has two important consequences for assessment of complaint of excessive bleeding in women experiencing variable losses. First, it is not the arithmetic mean of a woman's losses that is likely to be representative of her “complaint,” but rather the volume of the intermittent “heavier” periods. Second, in a short collection window such as two periods, a woman may not experience one of her “very heavy” periods.
Examination of the three equations for estimating blood volume reveals two features of the relationship between total fluid volume and blood loss. First, the proportion that estimated blood loss comprises the total fluid volume is not constant across the range of fluid volumes observed. This variation can be seen in Table 1, where the estimated percent blood content ranges from 47.6% to 52.1%. This small degree of nonlinearity results from the requirement to log transfer the raw data before undertaking linear regression. Second, for the same measured total fluid volume the blood volume for tampon users was estimated at only 79.6% of what it would be if the subject had used pads only. There are likely to be a number of factors affecting the percentage of blood content, with individual levels of nonmenstrual, non-blood fluid secretion being influential at light to moderate losses. In the study of inter-menstrually collected vaginal fluid, the content of hygiene products varied markedly between subjects, but was significantly greater for tampon users compared with pad users, amounting to ±13 mL excess over a 5-day period. We do not know how readily intermenstrual estimates of vaginal–cervical fluid may be extrapolated to the rather different setting of actual menses, but if maintained, this would tend to reduce observed percentage blood content in tampon users. When changes within subject were analyzed, effects from general level of loss, products used, changing style, and individual level of nonmenstrual secretions could be controlled for statistically, enabling a “best” estimate of percentage blood content. In this analysis, percentage blood content was found to be 52%.
Inspection of Figure 2 shows that it is at total fluid losses of less than 250 mL that the greatest separation in measured blood loss is seen between the two sanitary protection groups. However, at levels of loss over 200 mL total fluid, such divergence in the estimates is of little clinical importance, as either estimate of blood loss is well over the criterion for clinical menorrhagia. Therefore, for most clinical purposes the estimate of blood loss could be made ignoring the hygiene product type. The clinical value of this method of estimating blood loss was further shown by the fact that all but one out of 106 periods (99%) could be correctly classified as “normal,” “heavy,” or “excessive” on the basis of estimated blood loss. Using the corresponding total fluid volume “cutoffs” (ie, if a period exceeded 200 mL in total fluid volume, then blood loss was likely to exceed 100 mL, and if total fluid volume exceeded 124 mL, the blood loss was likely to exceed 60 mL), classification is conveniently made from the measured total fluid volume directly—without having to calculate the blood loss estimate. This simple method of estimation using weighed total fluid volume appears to be more accurate than the pictorial assessment chart.9 This superiority is evident both in the estimate it provides of actual blood loss, and also in the binary judgment of whether a woman's blood loss exceeds 80 mL (one recommended cut-off for menorrhagia): sensitivity was 89% using total fluid compared with 86% for the chart, specificity 98% versus 89%.
There is some concern over the invasive treatments initiated for menorrhagia, when the majority of complainers do not have abnormal blood loss.4,17 It is therefore intriguing that recent guidelines for the management of menorrhagia do not quantify the volume of blood lost.18–20 This omission is partly a recognition of the impracticability of doing so on a routine basis using the hematin method. Total fluid volume is much easier to ascertain than blood loss, because it only requires weighing sanitary products before and after use, with the difference in grams converting directly to a total fluid volume in milliliters. However, an important caveat is that women have to be carefully instructed in meticulous collection of all menstrual loss including “clots.”7 The use of individual self-sealing plastic bags for storage of individual used products also found favor with the women because it eliminated odor. This method of assessing menstrual blood loss is likely to be more acceptable to patients, and it could easily be developed into a cheap and simple commercial pack. Therefore, measurement of total fluid volume could well provide a pragmatic method of assessing menstrual blood loss in women complaining of menorrhagia.
1. Ebert R, Nold B. Gerinnungphysiologische studien am menstrualblut. Schwuez Med Wochenschr 1956;35:999–1004.
2. Fraser IS, McCarron G, Markham R, Resta T. Blood and total fluid content of menstrual discharge. Obstet Gynecol 1985;65:194–8.
3. Levin RJ, Wagner G. Absorption of menstrual discharge by tampons inserted during menstruation: Quantitative assessment of blood and total fluid content. Br J Obstet Gynaecol 1986;93:765–72.
4. Grant JM. Excessive menstruation—from diagnosis to effectiveness. (Editorial) Br J Obstet Gynaecol 2000;107:vii–vii.
5. Chimbira T, Anderson ABM, Turnbull AC. Relation between measured menstrual blood loss and patient's subjective assessment of loss, duration of bleeding, number of sanitary towels used, uterine weight and endometrial surface area. Br J Obstet Gynaecol 1980;87:603–9.
6. Fraser IS, Pearse C, Shearman RP, Elliott PM, McIlveen J, Markham R. Efficacy of mefenamic acid in patients with a complaint of menorrhagia. Obstet Gynecol 1981;58:543–51.
7. Fraser IS, McCarron G, Markham R. A preliminary study of factors influencing perception of menstrual blood loss volume. Am J Obstet Gynecol 1984;149:788–93.
8. Hallberg L, Nilsson L. Determination of menstrual blood loss. Scand J Clin Lab Invest 1964;16:244–8.
9. Higham JM, O'Brien PMS, Shaw RW. Assessment of menstrual blood loss using a pictorial chart. Br J Obstet Gynaecol 1990;97:734–9.
10. Janssen CAH, Scholten PC, Heintz APM. A simple visual assessment technique to discriminate between menorrhagia and normal menstrual blood loss. Obstet Gynecol 1995;85:977–82.
11. Reid PC, Coker A, Coltart R. Assessment of blood loss using a pictorial chart: A validation study. Br J Obstet Gynaecol 2000;107:320–2.
12. Pendergrass PB, Scott JN, Ream LJ. A rapid, non-invasive method for evaluation of total menstrual loss. Gynecol Obstet Invest 1984;17:174–8.
13. Gannon MJ, Day P, Hammadieh N, Johnson N. A new method of measuring menstrual blood loss and its use in screening women before endometrial ablation. Br J Obstet Gynaecol 1996;103:1029–33.
14. Newton J, Barnard G, Williams C. A rapid method for measuring menstrual blood loss using automatic extraction. Contraception 1977;16:269–76.
15. Hallberg L, Nilsson L. Constancy of individual menstrual blood loss. Acta Obstet Gynecol Scand 1964;43:352–9.
16. Haynes PJ, Hodgson H, Anderson ABM, Turnbull AC. Measurements of menstrual blood loss in patients complaining of menorrhagia. Br J Obstet Gynaecol 1977;84:763–8.
17. Stirrat GM. Choice of treatment for menorrhagia. Lancet 1999;353:2175–6.
18. Chadha Y, Mollison J, Howie F, Grimshaw G, Hall M, Russell I. Guidelines in gynaecology: Evaluation in menorrhagia and in urinary incontinence. Br J Obstet Gynaecol 2000;107:535–43.
19. Royal College of Obstetricians and Gynaecologists. The initial management of menorrhagia. Evidence-based clinical guidelines, No 1. London: Royal College of Obstetricians and Gynaecologists, 1998.
20. American College of Obstetricians and Gynecologists. OG Practice Bulletin. Management of anovulatory bleeding. Obstet Gynecol 2000;95:1–8.