71
Clinica Chimica Acta, 89 (1978) 71-78 @ Elsevier/North-Holland Biomedical Press
CCA 9669
INTERPRETATION FOR MONITORING PREGNANCY
OF URINARY 0ESTROGEN:CREATININE FETOPLACENTAL FUNCTION IN LATE
RATIO
SUSAN D. PHILLIPS 8, J.G. SALWAY ***, R.B. PAYNE a+** and H.N. MACDONALD a Department of Chemical Pathology Hospital, Leeds LS9 7TF (U.K.) (Received
and b Department
of Obstetrics,
b
St. James’s University
April 13th, 1978)
Summary In pregnancies with a normal outcome the oestrogen:creatinine ratio in early morning samples of urine showed a smaller day-to-day variation than the ratio in 24-h urine or the 24-h total oestrogen excretion, and a significant fall could be detected more easily. Patients admitted to hospital who eventually delivered a healthy baby provided a reference range which, after logarithmic transformation, increased linearly with period of gestation. A fall in log,,, oestrogen: creatinine ratio exceeding 40% of this range is unusual in pregnancies with a healthy outcome, and suggests impaired fetoplacental function.
Introduction There are three biochemical methods for monitoring fetoplacental function in late pregnancy which are based on the analysis of urine samples: the 24-h oestrogen excretion, the ratio of oestrogen to creatinine concentration in 24-h urine and the oestrogen:creatinine ratio in random urine. Of these, the last has practical advantages for patients, doctors and laboratory staff because the difficulties, errors and time delays of 24-h urine collections are avoided. However, the best time of day for the collection of urine is not known, little information is available about reference ranges, and the magnitude of a significant fall in oestrogen:creatinine ratio has not been established. There is considerable day to day variation both in 24-h oestrogen excretion and in the oestrogen:creatinine ratio [l] and a major problem in the detection of fetoplacental dysfunction is to distinguish a genuine fall in oestrogen excre-
* Present address: Department of Biochemistry, University of Surrey, Guildford. ** To whom correspondence should be addreseed.
72
tion from random day to day variation. The smaller the day to day variation, the more readily can a real fall be recognised. The purpose of the present investigation was to determine the time of day when random urines gave the most consistent values for the oestrogen:creatinine ratio, and then to establish which of the three biochemical indices of fetoplacental function showed the least day to day variation in relation to its reference range in pregnancies which resulted in the delivery of a normal healthy baby. The logarithm of the oestrogen:creatinine ratio in early morning specimens of urine proved to be the best test in this respect. We were able to define an objective criterion by which a significant fall in oestrogen:creatinine ratio could be distinguished from random day to day variation. Materials and methods Patients and specimens
We studied patients in hospital on whom 24-h oestrogen determinations were being performed routinely but who eventually delivered a healthy baby weighing at least 2.5 kg. Approximately 5 ml of an early morning specimen of urine was transferred to a tube and sent to the laboratory with the completed 24-h urine collection. Request forms noted drugs and other factors, such as glycosuria and ketonuria, which might affect the interpretation of the oestrogen or creatinine results, and such specimens were excluded. Analytical
methods
Urine total oestrogens (pmol/l) and creatinine (mmol/l) were measured by combining an all-aqueous automated, continuous-flow method for oestrogen [2] with the simultaneous measurement of creatinine. The analytical time by this method is less than 20 min. Between-batch coefficients of variation in 30 measurements of commercial control urines (International Diagnostic Aids, Ltd., Gloucester) for oestrogen were 3.5% at 63 pmol/l (18 mg/l) and 7.3% at 23 pmol/l (6.6 mg/l); and for creatinine were 2.6% at 6.5 mmol/l(O.74 g/l) and 4.8% at 2.3 mmol/l (0.26 g/l). The within-batch standard deviation calculated from 103 pairs of observations on patient’s urines was 2.92 pmol/l (0.84 mg/l) for oestrogen and 0.18 mmol/l (0.02 g/l) for creatinine. The within-batch standard deviation for the oestrogen:creatinine ratio was 0.37 ~mol/mmol (0.94 mg/g) at a mean of about 10 pmol/mmol (25.5 mg/g), so that the within-batch coefficient of variation was less than 4%. Reference
ranges
The data were collected over a period of one year. When several results were available for an individual within any week of gestation only the first result in a given week was used for the reference range, a total of 490 results during weeks 34 to 40. Logarithmic
transformation
of data
Probit analysis of the reference data for the 24-h oestrogen excretion, the oestrogen:creatinine ratio in 24-h urine and the oestrogen:creatinine ratio in early morning urine confirmed their logarithmic distribution within each week
73
of gestation. Data were therefore transformed to log,, for subsequent calculations. The upper and lower limits of the reference ranges for each week, for each test, were calculated as mean + 2 standard deviations of the log&ransformed data. Between-day variation in relation to time of day
The between-day variation of the oestrogen:creatinine ratio in random urine was investigated in urines voided at different times of day by eight highlymotivated pregnant patients who were in hospital for the man~ement of problems arising from complications other than suspected impairment of fetoplacental function. The patients collected each micturition into a separate bottle and carefully noted the time on each. Specimens from each patient were analysed in the same batch. For each individu~ log,, oestrogen:~reat~ine ratio was plotted against time.‘ We looked for the time of day when the ratio on consecutive days showed the least change by calculating at each hour for all patients the standard deviation of the values at times t and t + 24 h interpolated from their graphs, using the equation S.D. = Jm where d = difference and N = number of differences. Within-week variation
The within-week standard deviation of each test was calculated from the first and last result for each patient in each week after logarithmic transformation using S.D. = v/w where d = difference and iV = number of differences. The standard deviations were then expressed as a percentage of the log-normalised reference range at that week to make it possible to compare the different expressions of oestrogen excretion. Between-week variation in individual inpatients
Longitudinal studies were made in 21 patients who had a minimum of eight urines measured at weekly intervals. The week-to-week variation was estimated by calculating the standard deviation from successive duplicates [3] to minimise the effect on the standard deviation of the upward trend in values week by week. Results Reference ranges The reference values for the oestrogen:creatinine
ratio in early morning urine are shown in Fig. 1 and after logarithmic transformation in Fig. 2. The 24-h oestrogen excretion, the oestrogen:creatinine ratio in 24-h urine and the ratio in early morning urine showed significant correlations with each other (P < 0.001).
Between-day variation in relation to time of day in inpatients
The .standard deviations of logIO’oestrogen:creatinine ratio in urine passed at different times of day calculated from inte~ola~d values 24 h apart showed
-34
35
36
Maturity
37
38
39
40
(weeks)
Fig. 1. 0estrogen:creatinine ratios (wmol/mmol) in early morning specimens of urine from 490 inpatients who delivered a healthy baby. The dashed line is drawn by eye through the geometric mean for each week of gestation and the continuous lines through the 2 S.D. limits of the logarithmic distributions. Values can be converted to mg/g by multiplying by 2.55.
that the lowest between-day variation was in the early morning, with the variation increasing markedly after 11.00 h (Fig. 3). Within-week variation The weighted mean of the within-week standard deviations for log,, oestrogen:creatinine ratio in early morning urine from weeks 34 to 39 was 14% of the reference range (Table I). This within-week variation was significantly less (P < 0.01) than that for 24-h oestrogen excretion (19%) and oestrogen:creatinine ratio in 24-h urine (18%). When the random variations of measurements have a Gaussian distribution there is only a 1 in 20 chance (P < 0.05) that two values will differ by more than the standard deviation multiplied by 1.96 d/2 [4]. The magnitude of this significant change is expressed in terms of the reference range in the bottom line of Table I.
75
_W: t. . -/
I
I?/
34
35
36
Maturity
38
37
39
40
(weeks)
Fig. 2. The oestrogenxreatinine ratios from inpatients shown in Fig. 1 plotted as logarithmslo. The daabed line is the line of best fit through the geometric mean for each week of gestation from 34 to 39 weeks and the upper and lower continuous lines are drawn through the weighted 2 SD. limits. The inter mediate continuous lines divide the reference range into 20 percentiles. 0.00
._
e
0.07
.Z ; t
0.06
51
0*05
Bf km “0 w 0.04 08 -
*J P .j
‘ii
003
e 0.02
0 0.01 L
0
1
I
04.00
I
I
0000
I
I
12.00
I
I
1600
I
I
2090
I
I
24.00
Time of day Fig. 3. The standard deviations of the logarithms of the oestrogen:ereatinine ratio in urine passed at different times of day by eight patients in hospital for problems other than suspected impairment of fetoplacental function calculated from interpolated values 24 h apart.
76 TABLE
I
WITHIN-WEEK VARIATION OF THREE CIES WITH A HEALTHY OUTCOME
MEASURES
OF OESTROGEN
EXCRETION
Data transformed to logarithms, the standard deviation calculated from paired individuals and expressed as a percentage of the appropriate reference range Week of gestation
34 35 36 37 38 39 Weighted mean
Number patients
of
16 49 35 94 59 30 303
Magnitude of a significant change in value (at P < 0.05; see text for details)
Between-week
Standard
deviation
as a percentage
of reference
IN PREGNAN-
within-week
data from
range Oestrogen 24 h (%)
Oestrogen: creatinine ratio in early morning urine (%)
Oestrogen: creatinine ratio in 24-h urine
10.8 13.7 14.1 12.7 14.4 17.9 13.9
18.6 18.7 16.6 16.9 20.6 18.0 18.0
16.4 20.4 21.4 17.7 18.6 17.2 18.9
38.5
49.9
52.4
excretion
in
(%)
variation in individual patients
We studied log,, oestrogen:creatinine ratio in early morning urine over at least eight weeks in 21 patients. The standard deviation of the ratios (expressed as a percentage of the reference range for the appropriate week) was 15% close to the value of 14% in the within-week study. When log-transformed data was used there was no significant correlation between individual patients’ standard deviations and their oestrogen:creatinine ratios at, for example, 36 weeks (r = -0.232; P > 0.3). However, with arithmetic data there was a strong positive correlation (r = +0.569; P < O.Ol), the standard deviation being greater the higher the value. Discussion The oestrogen excretion in 24 h and the oestrogen:creatinine ratio both in 24 h and in early morning urine showed a logarithmic distribution within each week of late pregnancy, and there were highly significant correlations between them (P < 0.001). Because random urine had several practical advantages, we evaluated it for the routine monitoring of fetoplacental function. There were two reasons for expressing the within-week variation in individuals as a percent of the normalised reference ranges derived from patients who were in hospital but who delivered a healthy baby. First, we used the inpatient ranges for each of the three biochemical measurements rather than the somewhat higher values from outpatients because the great majority of inpatients deliver a healthy baby, and this is the control group from which we wished to distinguish those with failing fetoplacental function. Secondly, we expressed the variation as a percent of these ranges because the biochemical test with the smallest percent value must have the range with the smallest contribution from
77
variation within individuals, and would thus allow a fall in a patient’s value to be detected more easily. The period of least variation of the logarithm of the oestrogen:creatinine ratio was in the early rno~~g (Fig. 3), and the logarithm of the ratio in early morning urine showed a smaller variation within each week than the logarithm of the ratio in 24-h urine or the logarithm of the total oestrogen excretion in 24 h (Table I). Logarithmic transformation of the oestrogen:creatinine reference data gave a range which showed a linear increase from week 34 to 39 (Fig. 2). An advantage of transfo~ation was that ~dividu~ patients showed random variations which were of the same order no matter where their values lay within the reference range, whereas with untransformed data there was a greater variation the higher the value. All tests currently available show considerable overlap between reference values and those found in pregnancies with fetoplacen~ dysfunction [l] so obstetricians look not only for low values but also for a significant fall of oestrogen excretion within the reference range. Hitherto, this has been a subjective judgment. We have shown that a significant within-week fall in the logarithm of the oestrogen:creatinine ratio, suggesting impaired fetoplacental function, is about 40% of the normalised reference range, whereas a change in total 24-h oestrogen excretion would have to exceed 50% of its reference range to reach significance (Table I). The oestrogen:creatinine ratio change has been conveniently represented in a graphical form which is used in the patients’ clinical notes, and consists of the lines (but not the data points) in Fig. 2. If, when plotted serially, the values of an individual should fall within a period of one week to an extent greater than 40% of the reference range, equivalent to two spaces between the 20 percentile lines, then fetoplacen~l ~sufficiency should be considered. Earlier studies on the daily variation in oestrogen excretion in timed urine, on the variation in the oestrogen:creatinine ratio in random urine, and on the relation between them have been reviewed by Wilde and Oakey [ 11. None of them has taken account of the log~thmic dist~bution of the data, so that it is not possible to make direct comparisons with the present work. We are not aware of any previous studies which define an objective criterion by which a small but significant fall in oestrogen excretion may be distinguished from random day to day variation. Rao [ 5 3 has recently demonstrated the usefulness of the oestrogen:creatinine ratio in predicting fetal death. In 5429 prelacies there were 188 patients with values below his lower limit of normal. Among these there were 57 fetal deaths, and in 51 the ratio had fallen to less than 3 pmol/mmol (8 mg/g). In contrast, there were 59 patients in whom the ratio fell from normal to subnormal during weeks 34 to 40 of pregnancy who were delivered after clinical intervention before the ratio fell below 3 ~mol/mmol (8 mg/g). In every case a live infant was delivered. Rao’s “critical value” of 3 ~mol/mmol corresponds to a value of 0.5 for the logarithm of the oestrogen:creatinine ratio. In summary, the oestrogen:creatinine ratio in early morning urine shows a smaller day to day variation than the 24-h oestrogen excretion, and a significant fall is more readily recognised. It has the advantage that it obviates the need for a 24-h urine collection which delays the result, may be unreliable, and is inconvenient for patients, ward and laboratory staff alike.
78
Acknowledgement We are indebted to nursing staff at St, Mary’s Hospital and St. James’ University Hospital for their invaluable help with this investigation, and to clinical colleagues for allowing us to study their patients. References 1 2 3 4 5
Wilde, C.E. and Oakey. R.E. (1975) Ann. Clin, 3iochem. 12.83-118 Little, A.J., Au&on. K. and Payne, R.B. (1975) Clin. Chim. Acts 65.167-173 Payne, R.B. (1972) Clin. Chim. Acta 42, 263-266 Campbell, D,G. and Owen, J.A. (1967) Clin. Biochem. 1.3-11 Rao. L.G.S. (1977) Br. Med. J. 2.874-876
Clinica Chimica Acta, 89 (1978) 71-78 @ Elsevier/North-Holland Biomedical Press
CCA 9669
INTERPRETATION FOR MONITORING PREGNANCY
OF URINARY 0ESTROGEN:CREATININE FETOPLACENTAL FUNCTION IN LATE
RATIO
SUSAN D. PHILLIPS 8, J.G. SALWAY ***, R.B. PAYNE a+** and H.N. MACDONALD a Department of Chemical Pathology Hospital, Leeds LS9 7TF (U.K.) (Received
and b Department
of Obstetrics,
b
St. James’s University
April 13th, 1978)
Summary In pregnancies with a normal outcome the oestrogen:creatinine ratio in early morning samples of urine showed a smaller day-to-day variation than the ratio in 24-h urine or the 24-h total oestrogen excretion, and a significant fall could be detected more easily. Patients admitted to hospital who eventually delivered a healthy baby provided a reference range which, after logarithmic transformation, increased linearly with period of gestation. A fall in log,,, oestrogen: creatinine ratio exceeding 40% of this range is unusual in pregnancies with a healthy outcome, and suggests impaired fetoplacental function.
Introduction There are three biochemical methods for monitoring fetoplacental function in late pregnancy which are based on the analysis of urine samples: the 24-h oestrogen excretion, the ratio of oestrogen to creatinine concentration in 24-h urine and the oestrogen:creatinine ratio in random urine. Of these, the last has practical advantages for patients, doctors and laboratory staff because the difficulties, errors and time delays of 24-h urine collections are avoided. However, the best time of day for the collection of urine is not known, little information is available about reference ranges, and the magnitude of a significant fall in oestrogen:creatinine ratio has not been established. There is considerable day to day variation both in 24-h oestrogen excretion and in the oestrogen:creatinine ratio [l] and a major problem in the detection of fetoplacental dysfunction is to distinguish a genuine fall in oestrogen excre-
* Present address: Department of Biochemistry, University of Surrey, Guildford. ** To whom correspondence should be addreseed.
72
tion from random day to day variation. The smaller the day to day variation, the more readily can a real fall be recognised. The purpose of the present investigation was to determine the time of day when random urines gave the most consistent values for the oestrogen:creatinine ratio, and then to establish which of the three biochemical indices of fetoplacental function showed the least day to day variation in relation to its reference range in pregnancies which resulted in the delivery of a normal healthy baby. The logarithm of the oestrogen:creatinine ratio in early morning specimens of urine proved to be the best test in this respect. We were able to define an objective criterion by which a significant fall in oestrogen:creatinine ratio could be distinguished from random day to day variation. Materials and methods Patients and specimens
We studied patients in hospital on whom 24-h oestrogen determinations were being performed routinely but who eventually delivered a healthy baby weighing at least 2.5 kg. Approximately 5 ml of an early morning specimen of urine was transferred to a tube and sent to the laboratory with the completed 24-h urine collection. Request forms noted drugs and other factors, such as glycosuria and ketonuria, which might affect the interpretation of the oestrogen or creatinine results, and such specimens were excluded. Analytical
methods
Urine total oestrogens (pmol/l) and creatinine (mmol/l) were measured by combining an all-aqueous automated, continuous-flow method for oestrogen [2] with the simultaneous measurement of creatinine. The analytical time by this method is less than 20 min. Between-batch coefficients of variation in 30 measurements of commercial control urines (International Diagnostic Aids, Ltd., Gloucester) for oestrogen were 3.5% at 63 pmol/l (18 mg/l) and 7.3% at 23 pmol/l (6.6 mg/l); and for creatinine were 2.6% at 6.5 mmol/l(O.74 g/l) and 4.8% at 2.3 mmol/l (0.26 g/l). The within-batch standard deviation calculated from 103 pairs of observations on patient’s urines was 2.92 pmol/l (0.84 mg/l) for oestrogen and 0.18 mmol/l (0.02 g/l) for creatinine. The within-batch standard deviation for the oestrogen:creatinine ratio was 0.37 ~mol/mmol (0.94 mg/g) at a mean of about 10 pmol/mmol (25.5 mg/g), so that the within-batch coefficient of variation was less than 4%. Reference
ranges
The data were collected over a period of one year. When several results were available for an individual within any week of gestation only the first result in a given week was used for the reference range, a total of 490 results during weeks 34 to 40. Logarithmic
transformation
of data
Probit analysis of the reference data for the 24-h oestrogen excretion, the oestrogen:creatinine ratio in 24-h urine and the oestrogen:creatinine ratio in early morning urine confirmed their logarithmic distribution within each week
73
of gestation. Data were therefore transformed to log,, for subsequent calculations. The upper and lower limits of the reference ranges for each week, for each test, were calculated as mean + 2 standard deviations of the log&ransformed data. Between-day variation in relation to time of day
The between-day variation of the oestrogen:creatinine ratio in random urine was investigated in urines voided at different times of day by eight highlymotivated pregnant patients who were in hospital for the man~ement of problems arising from complications other than suspected impairment of fetoplacental function. The patients collected each micturition into a separate bottle and carefully noted the time on each. Specimens from each patient were analysed in the same batch. For each individu~ log,, oestrogen:~reat~ine ratio was plotted against time.‘ We looked for the time of day when the ratio on consecutive days showed the least change by calculating at each hour for all patients the standard deviation of the values at times t and t + 24 h interpolated from their graphs, using the equation S.D. = Jm where d = difference and N = number of differences. Within-week variation
The within-week standard deviation of each test was calculated from the first and last result for each patient in each week after logarithmic transformation using S.D. = v/w where d = difference and iV = number of differences. The standard deviations were then expressed as a percentage of the log-normalised reference range at that week to make it possible to compare the different expressions of oestrogen excretion. Between-week variation in individual inpatients
Longitudinal studies were made in 21 patients who had a minimum of eight urines measured at weekly intervals. The week-to-week variation was estimated by calculating the standard deviation from successive duplicates [3] to minimise the effect on the standard deviation of the upward trend in values week by week. Results Reference ranges The reference values for the oestrogen:creatinine
ratio in early morning urine are shown in Fig. 1 and after logarithmic transformation in Fig. 2. The 24-h oestrogen excretion, the oestrogen:creatinine ratio in 24-h urine and the ratio in early morning urine showed significant correlations with each other (P < 0.001).
Between-day variation in relation to time of day in inpatients
The .standard deviations of logIO’oestrogen:creatinine ratio in urine passed at different times of day calculated from inte~ola~d values 24 h apart showed
-34
35
36
Maturity
37
38
39
40
(weeks)
Fig. 1. 0estrogen:creatinine ratios (wmol/mmol) in early morning specimens of urine from 490 inpatients who delivered a healthy baby. The dashed line is drawn by eye through the geometric mean for each week of gestation and the continuous lines through the 2 S.D. limits of the logarithmic distributions. Values can be converted to mg/g by multiplying by 2.55.
that the lowest between-day variation was in the early morning, with the variation increasing markedly after 11.00 h (Fig. 3). Within-week variation The weighted mean of the within-week standard deviations for log,, oestrogen:creatinine ratio in early morning urine from weeks 34 to 39 was 14% of the reference range (Table I). This within-week variation was significantly less (P < 0.01) than that for 24-h oestrogen excretion (19%) and oestrogen:creatinine ratio in 24-h urine (18%). When the random variations of measurements have a Gaussian distribution there is only a 1 in 20 chance (P < 0.05) that two values will differ by more than the standard deviation multiplied by 1.96 d/2 [4]. The magnitude of this significant change is expressed in terms of the reference range in the bottom line of Table I.
75
_W: t. . -/
I
I?/
34
35
36
Maturity
38
37
39
40
(weeks)
Fig. 2. The oestrogenxreatinine ratios from inpatients shown in Fig. 1 plotted as logarithmslo. The daabed line is the line of best fit through the geometric mean for each week of gestation from 34 to 39 weeks and the upper and lower continuous lines are drawn through the weighted 2 SD. limits. The inter mediate continuous lines divide the reference range into 20 percentiles. 0.00
._
e
0.07
.Z ; t
0.06
51
0*05
Bf km “0 w 0.04 08 -
*J P .j
‘ii
003
e 0.02
0 0.01 L
0
1
I
04.00
I
I
0000
I
I
12.00
I
I
1600
I
I
2090
I
I
24.00
Time of day Fig. 3. The standard deviations of the logarithms of the oestrogen:ereatinine ratio in urine passed at different times of day by eight patients in hospital for problems other than suspected impairment of fetoplacental function calculated from interpolated values 24 h apart.
76 TABLE
I
WITHIN-WEEK VARIATION OF THREE CIES WITH A HEALTHY OUTCOME
MEASURES
OF OESTROGEN
EXCRETION
Data transformed to logarithms, the standard deviation calculated from paired individuals and expressed as a percentage of the appropriate reference range Week of gestation
34 35 36 37 38 39 Weighted mean
Number patients
of
16 49 35 94 59 30 303
Magnitude of a significant change in value (at P < 0.05; see text for details)
Between-week
Standard
deviation
as a percentage
of reference
IN PREGNAN-
within-week
data from
range Oestrogen 24 h (%)
Oestrogen: creatinine ratio in early morning urine (%)
Oestrogen: creatinine ratio in 24-h urine
10.8 13.7 14.1 12.7 14.4 17.9 13.9
18.6 18.7 16.6 16.9 20.6 18.0 18.0
16.4 20.4 21.4 17.7 18.6 17.2 18.9
38.5
49.9
52.4
excretion
in
(%)
variation in individual patients
We studied log,, oestrogen:creatinine ratio in early morning urine over at least eight weeks in 21 patients. The standard deviation of the ratios (expressed as a percentage of the reference range for the appropriate week) was 15% close to the value of 14% in the within-week study. When log-transformed data was used there was no significant correlation between individual patients’ standard deviations and their oestrogen:creatinine ratios at, for example, 36 weeks (r = -0.232; P > 0.3). However, with arithmetic data there was a strong positive correlation (r = +0.569; P < O.Ol), the standard deviation being greater the higher the value. Discussion The oestrogen excretion in 24 h and the oestrogen:creatinine ratio both in 24 h and in early morning urine showed a logarithmic distribution within each week of late pregnancy, and there were highly significant correlations between them (P < 0.001). Because random urine had several practical advantages, we evaluated it for the routine monitoring of fetoplacental function. There were two reasons for expressing the within-week variation in individuals as a percent of the normalised reference ranges derived from patients who were in hospital but who delivered a healthy baby. First, we used the inpatient ranges for each of the three biochemical measurements rather than the somewhat higher values from outpatients because the great majority of inpatients deliver a healthy baby, and this is the control group from which we wished to distinguish those with failing fetoplacental function. Secondly, we expressed the variation as a percent of these ranges because the biochemical test with the smallest percent value must have the range with the smallest contribution from
77
variation within individuals, and would thus allow a fall in a patient’s value to be detected more easily. The period of least variation of the logarithm of the oestrogen:creatinine ratio was in the early rno~~g (Fig. 3), and the logarithm of the ratio in early morning urine showed a smaller variation within each week than the logarithm of the ratio in 24-h urine or the logarithm of the total oestrogen excretion in 24 h (Table I). Logarithmic transformation of the oestrogen:creatinine reference data gave a range which showed a linear increase from week 34 to 39 (Fig. 2). An advantage of transfo~ation was that ~dividu~ patients showed random variations which were of the same order no matter where their values lay within the reference range, whereas with untransformed data there was a greater variation the higher the value. All tests currently available show considerable overlap between reference values and those found in pregnancies with fetoplacen~ dysfunction [l] so obstetricians look not only for low values but also for a significant fall of oestrogen excretion within the reference range. Hitherto, this has been a subjective judgment. We have shown that a significant within-week fall in the logarithm of the oestrogen:creatinine ratio, suggesting impaired fetoplacental function, is about 40% of the normalised reference range, whereas a change in total 24-h oestrogen excretion would have to exceed 50% of its reference range to reach significance (Table I). The oestrogen:creatinine ratio change has been conveniently represented in a graphical form which is used in the patients’ clinical notes, and consists of the lines (but not the data points) in Fig. 2. If, when plotted serially, the values of an individual should fall within a period of one week to an extent greater than 40% of the reference range, equivalent to two spaces between the 20 percentile lines, then fetoplacen~l ~sufficiency should be considered. Earlier studies on the daily variation in oestrogen excretion in timed urine, on the variation in the oestrogen:creatinine ratio in random urine, and on the relation between them have been reviewed by Wilde and Oakey [ 11. None of them has taken account of the log~thmic dist~bution of the data, so that it is not possible to make direct comparisons with the present work. We are not aware of any previous studies which define an objective criterion by which a small but significant fall in oestrogen excretion may be distinguished from random day to day variation. Rao [ 5 3 has recently demonstrated the usefulness of the oestrogen:creatinine ratio in predicting fetal death. In 5429 prelacies there were 188 patients with values below his lower limit of normal. Among these there were 57 fetal deaths, and in 51 the ratio had fallen to less than 3 pmol/mmol (8 mg/g). In contrast, there were 59 patients in whom the ratio fell from normal to subnormal during weeks 34 to 40 of pregnancy who were delivered after clinical intervention before the ratio fell below 3 ~mol/mmol (8 mg/g). In every case a live infant was delivered. Rao’s “critical value” of 3 ~mol/mmol corresponds to a value of 0.5 for the logarithm of the oestrogen:creatinine ratio. In summary, the oestrogen:creatinine ratio in early morning urine shows a smaller day to day variation than the 24-h oestrogen excretion, and a significant fall is more readily recognised. It has the advantage that it obviates the need for a 24-h urine collection which delays the result, may be unreliable, and is inconvenient for patients, ward and laboratory staff alike.
78
Acknowledgement We are indebted to nursing staff at St, Mary’s Hospital and St. James’ University Hospital for their invaluable help with this investigation, and to clinical colleagues for allowing us to study their patients. References 1 2 3 4 5
Wilde, C.E. and Oakey. R.E. (1975) Ann. Clin, 3iochem. 12.83-118 Little, A.J., Au&on. K. and Payne, R.B. (1975) Clin. Chim. Acts 65.167-173 Payne, R.B. (1972) Clin. Chim. Acta 42, 263-266 Campbell, D,G. and Owen, J.A. (1967) Clin. Biochem. 1.3-11 Rao. L.G.S. (1977) Br. Med. J. 2.874-876
