European Journal of Cardiovascular Medicine

A quick and easy method of estimation of foetal weight in utero is an obvious benefit to the clinician practicing modern obstetrics as the perinatal mortality and morbidity is affected not only by foetal age but also by the foetal weight. Foetal weight has also become increasingly important in conditions like management of preterm babies, small for gestation babies, decision for delivery in growth restricted babies, mode of delivery in breech presentation, induction of labour before term in complicated pregnancies, evaluation of foeto pelvic disproportion

Assuming a crude birth rate of 25/ 10,000, there are 23 million births in India every year of which approximately 17.5 million takes place in rural India, which are under domiciliary condition. For these cases we have to search for a clinical method for foetal weight estimation, which is easy, reliable, and can be applied at PHC level by the birth attendants.

An ideal method of foetal weight estimation should have the following advantages

1. Does not need ultrasound, which is not available in most of the PHCs
2. Does not need pelvimeter
3. Considers abdominal girth in calculating baby weight which is to an extend affected by the uterine volume unlike in Johnsons or Mhaskars formula where it is not considered
4. Does not involve internal examination to determine station of the head which needs expertise that is not expected of a domiciliary birth attendant.

Thus, this study was conducted in order to estimate the foetal weight by a simple and easy method which can be taught to the medical and paramedical staff and the birth attendant under MCH teaching programme as to improve the perinatal morbidity and mortality.

Aim of study

This study was conducted over a period of one year on 195 patients who reported in third trimester of pregnancy   and delivered within one week of examination. The Aim of the study was to formulate and evaluate a simple formula for clinical estimation of foetal weight which will be useful at PHC level by the birth attendants.

Null hypothesis is symphysiofundal height (SFH) multiplied by abdominal girth (AG) equals to birth weight (BWT) in grams.

SFH x AG = BWT in grams

The actual birth weight was then compared with clinical estimation of birth weight by:

1. Clinical method
2. Johnson’s formula
3. Mhaskars formula
4. Null hypothesis

The result was compared with each other and accuracy and correlation determined to find the best method.

Sym physio fundal height (SFH) and abdominal girth (AG) was measured in 196 pregnant women attending the OPD or Maternity ward at term. The station of the head was found and various methods of calculation of baby weight was used to estimate the baby weight. The actual weight of the baby was measured at the time of birth and statistical analysis done.

MC Donalds measurement of height of fundus from the upper edge of the symphysis pubis following the curvature of the abdomen was taken in centimetres using non elastic tape with the subject in the supine position with legs extended and the bladder empty. Care was taken to ensure that the fundus was determined by gentle pressure extended in a plane right angle to the abdominal wall after the dextrorotation was corrected.

The abdominal girth (AG) was measured at the level of umbilicus in the same position of the subject with the same tape. No correction was made for breech, maternal height, weight, abdominal wall thickness, or obesity.

The measurements were taken by the same observer at each visit to the nearest .5 cm with the tape on the reverse side up for the observer not to be influenced by the values.

Station of the head:

The station of the head was found by abdominal method:

5/5 to 3/5 palpable = unengaged (13)

2/5 to 0/5 palpable = engaged (12)

Calculations

The estimation of weight was done using four methods.

1. Johnson’s formula (1957): (Mc Donalds measurements – n) x 155 = weight in grams.

• n = 13, unengaged.
• n = 12, engaged.

2. Clinical method

The clinical method included factors like the size of head, weight, and the built of the mother. The accuracy of the estimation of foetal weight depends on the experience of the clinician. Here the weight was estimated to the nearest 50 gms.

3. Formula by mhaskar (2001)

• BWT =

4. New Formula

• SFH X AG = Birth weight in grams

The actual birth weight of the babies at delivery was found at the time of delivery on the same weighing machine by the same observer to the nearest of 50 gms. The new-borns were categorised into 3 groups.

≤ 1500               = very low birth weight

1501 – 2500      = low birth weight

2501 – 3500      = normal weight

3501 – 4000      = big baby

≥ 4001         = macrosomia

Statistical analysis was used to find the following:

• Correct proportion of interpretation of the foetal weight by various methods
• Statistical difference in the interpretation rate by various methods
• Correlation between actual and estimated birth weights by different methods.
• Statistical difference in the correlation between the actual and estimated foetal weight by different methods.
• Degree of agreement in different methods of estimating foetal weight.

The standard error of mean of actual birth weight was calculated and its 95% confidence interval found out. For the predicted baby weight to be statistically correct, it should fall between the 95% confidence intervals of the actual birth weight. Such values were taken as correct.  The correct interpretation rate was calculated for each method.  It was tested if there was any statistically significant difference in the interpretation rate.

To find if there is statistically significant difference in interpretation rate by various methods, the ‘p value’ between the correct and the incorrect number of interpretations by various methods of estimated foetal weight groups were calculated and compared with the chi²value. Statistically significant difference was present if the ‘p’ value is < 0.05.

To know if there is any statistically significant correlation between various methods of estimated foetal weight and the actual birth weight; paired sample correlation was determined, using the mean of actual weight and the means of various methods of prediction of foetal weight in pairs and ‘p’ value found out. Significant correlation was present if the ‘p’ value was < 0.05.

To find if there is statistically difference in the degree of correlations between the actual birth and the various methods of estimation of birth weight, paired sample test was done.

Two observers may interpret a particular finding either in agreement or in disagreement. If they interpret similarly, they are said to be in agreement and if they interpret differently, it is disagreement.

Eg. :      clinical – correct

Agreement                

SFH x AG - correct

Clinical   - correct

Disagreement

SFH x AG   - incorrect

To find the degree of agreement between pairs of various methods of estimation of fetal weight, kappa statistics was done. kappa statistics tests whether correct estimation of by one method is correctly interpreted by another method as well. And incorrect estimation by another method is interpreted incorrectly by another method also.

The relation of maternal height, weight, parity and total weight gain during pregnancy, were not tested as these factors are already established in other studies.

The sample comprised of 195 patients. The age group of the patients were from 18 to 38 yrs. There were 38 primigravidas, 56 second gravidas, and 4 grand multiparas. Height varied from 141 cm to 168 cm; weight was between 44 kg to 72 kgs.

The babies were in the various weight groups as follows:

very low                     < 1500                - 1

low birth weight        1501 – 2500        - 59

normal group             2501 – 3500       - 124

big babies                   3501 – 4000       - 10

macrosomic babies     > 4000                - 1

of the 195 gravidae studied, the mean actual birth weight of their babies was 2781 grams. Which was lower than mean birth weight estimated by various clinical methods. All the clinical estimation of the birth weight overestimated the actual birth weight by 1 to 2 gm/100gms.

Mean birth weight estimated by various clinical methods were as follows,

Table i

The standard error of mean of actual birth weight was calculated to be 35.9 gms. (table ix). Estimation of fetal weight within 95% confidence interval should be in a range of actual birth weight ±71.8 gms. That is between 2710 – 2853gms., or the value to be statistically correct predictor of birth weight. In our study, the Clinical method of estimation was taken to the nearest 50 gms, so the predicted birth weight was taken as correct if it was with a range of ±100 gms of actual birth weight (ABW). The mean correct estimated birth weight (EBW)would therefore be between 2681 – 2981 gms. Mean foetal weight calculated by different methods were within this range except the mean of Mhaskar’s formula.this did not mean that all the individual values would be within the respective correct range so the individual values were calculated.

Number of correct estimations of foetal weight within actual weight ± 100gms were as follows:

Table ii

The above table shows that 38.46% of times clinical method interpreted the foetal weight correctly, while others were correct lesser number of times. This means that the clinical estimation of foetal weight is the best predictor of foetal weight among various methods tested (38.48%) followed by SFH x AG2 (1.53%).

Figure 1

The above figure shows that the clinical estimation of foetal weight is the best predictor of foetal weight among various methods tested, followed by SFH x AG,

 To find if there is statistically significant difference in the proportion of interpretation by various methods, the ζ² test was done.

The ζ²value for various combination of methods were found out and the following result was found:

Comparison of the ζ²value and the p value showed that the clinical method is distinct from the rest. The clinical method had a significant difference in the proportion of correct estimates from the other methods. When the clinical method is not taken into consideration, others do not have statistically significant difference between each other. This means that using any formula will give more or less the same result.

Similarly, number of cases predicted within 10% of actual weight was found so that it could be compared with the other studies. 10% of the birth weight would fall between 2781 ± 278 gms, and the correct prediction would be within a range of 2503 – 3059 gms.

Estimation of weight within 10% of actual weight is as follows:

This table shows that mean of all the methods fall within this range. It also shows that the clinical method of estimating foetal weight is the best method of all the methods tested (64.6%), followed by SFH x AG (55.8%.)

The above figure shows that the clinical method of estimating the foetal weight is the best method of the various methods tested, followed by SFH x AG.

To see if these conclusions was also true for different categories of birth weight, individual groups were tested.

The low-birth-weight babies were 59 in number. Their mean actual birth weight was 2234.75 gms. The following table shows the mean birth weight calculated by different methods, number of cases in statistically correct range and those within 10% of actual birth weight by the various methods.

Table vi

This table shows that the clinical method of estimating the weight is the best method among the various methods tested., followed by SFH x AG (23), in case of low birth group of babies when tested statistically. On an average, all the methods overestimate the actual birth weight for the low birth babies except the Mhaskar’s formula.

This figure shows that the clinical method of estimating the weight is the best method among the various methods tested., followed by SFH x AG (23), in case of low birth group of babies when tested statistically. On an average, all the methods overestimate the actual birth weight for the low birth babies except the Mhaskar’s formula.

 The normal birth weight babies were 124 in no. the mean of the actual birth weights here was 2969.35gms. The following table the means calculated by different methods, number of cases in statically correct range (actual weight ± 100 gms) and those within 10% of actual birth weight by the various methods.

Table vii

This table shows that the clinical method (47.5%) of estimating the weight is the best method among the various methods tested., followed by SFH x AG (22.2%), in case of normal birth weight group of babies when tested statistically. Here the clinical and the Mhaskars formula underestimated the birth weight, while Johnson’s formula and the SFH x AG overestimated the birth weight.

Figure 4

This figure shows that the clinical method (47.5%) of estimating the weight is the best method among the various methods tested., followed by SFH x AG (22.2%), in case of normal birth weight group of babies when tested statistically.

The big babies were 10 in number. Their mean of the actual birth weights here was 3690gms. The following table shows the means calculated by different methods, number of cases in statically correct range (actual weight ± 100 gms) and those within 10% of actual birth weight by the various methods.

Table viii

This table shows that the clinical method (60%) of estimating the weight is the best method among the various methods tested., followed by SFH x AG (50%), in case of big babies when 10% of actual weight was taken as normal; though all the methods grossly underestimated the actual birth weight statistically.

This figure shows that the clinical method (60%) of estimating the weight is the best method among the various methods tested., followed by SFH x AG (50%), in case of big babies when 10% of actual weight was taken as normal.

Comparing the above 3 tables (vi, vii, viii); it is seen that the prediction of birth weight is best among the normal birth weight, where the difference of the means is the least, the difference is towards high positive value when the actual birth weight is high. This means that we tent to overestimate the birth weight when the actual birth weight is low and underestimate when the actual birth weight is high.

Paired sample statistics showed the following results:

Table ix

To know if there is statistically significant correlation between various methods of predicting fetal weight and actual birth weight; paired sample correlation was determined, using the mean of various methods of prediction of foetal weight and the significance found. Paired sample correlation showed the following results:

Table x

All the methods of estimation of foetal weight showed good correlation with the actual birth weight, which was statistically significant. (p< 0.05)

The maximum correlation to actual birth weight was shown by the clinical method. (0. 786) compared to other methods and among the various formulas, the actual weight best correlates with SHF x AG (0.658)

Table xi

It was found that there was statistical difference in the degree of correlation between actual weight and Mhaskar’s formula ( p value<0.05) but not with other methods of calculating foetal weight and actual weight.

To find the degree of agreement between the the pairs of various methods of estimation of foetal weight, kappa statistics was done. Kappa statistics tests where correct estimation of foetal weight is correctly interpreted by another method as well and incorrect estimation by one method is interpreted as incorrect by another method also.

The following was the observation:

Table xii

This table shows that maximum agreement between two methods of estimation of foetal weight is between the clinical and SFH x AG, and there is no statistically significant agreement between Clinical and Mhaskar’s method of foetal weight estimation.

Figure 6

This figure shows that maximum agreement between two methods of estimation of foetal weight is between the clinical and SFH x AG, and there is no statistically significant agreement between Clinical and Mhaskar’s method of foetal weight estimation.

All the method of clinical estimation of foetal weight, overestimated the actual birth weight by 1 to 2 gms / 100gms (table I). Estimation of foetal weigh by different method shows that the clinical method is the best method (table ii, & table iii). 38% of babies could be estimated within 100 gms of their actual birth weight (table ii) and 64% of foetuses could be estimated within 10% of their actual birth weight (table iii). This was better than that obtained by Tivari and Sood (1989) ⁷ which was 59% by the same method, where the correct was not defined as within 10% of actual birth weight.

Devi et al (1966) ² found Johnson’s simplified method correct upto 74% but here ‘correct ‘was not defined.by Mc donald’s rule using abdominal tape measurement of the gravid uterine fundus, the diagnostic accuracy was only 37% when studied by Centrulo et al; (1977) ⁶.

Tivari and Sood (1989) ⁷ obtained 55% of cases by Johnson’s method as correct with an error of 10%. In a study conducted by Mhaskar, estimation of foetal weight by Johnson’s method overestimated the actual birth weight by 320 gms which was statistically significant. Paired t = 68, df = 99, p <0.001). in our study Johnso’s method overestimated the weight by 53.6gms.

The standard error of mean of actual birth weight was calculated to be 35.9 gms. (table ix).so the estimation of foetal weight should be in a range of actual birth weight ± 71.8 gms.ie between 2710 – 2853 gms. For the value to be statistically correct predictor of birth weight. In our study the clinical estimation of birth weight was taken to the nearest 50 gms, so the predicted birth was taken to be correct if it was within a range of ± 100 gms of actual birth weight. The mean correct estimation of foetal weight would therefore be between 2681 – 2981gms. Mean foetal weight calculated by different methods were all within this range, but this did not mean that all the individual values would be within the correct range, so the individual values were checked.it was found that the clinical estimation of foetal weight is the best predictor of birth weight among the various methods tested. (38.48%), followed by SFH x AG (21.53), though all the methods overestimated the birth weight.

To know if the above conclusion also was true when applied to various birth weight group babies, individual category wise testing was done and this was found to be true. The prediction of birth weight was closest to the actual birth weight among the normal birth weight groups, where the differences of the means is the least, the difference was towards high positive value when the actual birth weight was low and it was towards high negative side when the actual birth weight was high. This means that we tend to underestimate the birth weight when it is actually high.

This was true with all the methods of estimation of birth weight except Mhaskar’s formula where it always underestimated the birth weight compared to the normal.

 Since there were only one cases each of very low birth weight and macrosomic category. we did not include them in the category wise study.

The clinical method had a statistically significant difference (‘p’ value<0.05) when the proportion of correct estimation by different methods were compared with the other methods. When the clinical method is not taken into consideration. Others do not have statistically significant difference between each other; this means that using any formula method would give more or less the same result.

To know if there is statistically significant correlation between various methods of predicting foetal weight and the actual birth weight paired sample correlation was determined, using the mean of actual weight and the means of various methods of prediction of foetal weight and the significance found.

All the methods of estimating foetal weight shows good co relation with the actual birth weight, and gave a statistically significant (p<0.05) correlation coefficient.

The maximum correlation to the actual birth weight was shown by the clinical method (0.786) compared to the other methods and among the various formulas, the actual weight best correlates with SFH x AG (0.658)

Also, it was seen that there was no statistical difference in the degree of correlation between various methods of predicting foetal weight and the actual birth weight except Mhaskar’s method (p value =0.001) (table xi), which meant that any method was equally good statistically excluding the mhaskar’s method.

The degree of agreement between pairs of various methods of estimation of foetal weight was tested using the kappa statistics. The maximum agreement between two methods of estimation of foetal weight was found between the clinical method and SFH x AG (11%), and the least between clinical and mhaskar’s(0%).

This study proves that the -SFH x AG (in centimeters) =BTW (in grams) is a useful formula. the study also shows that there is nothing to match a clinical experience in estimating foetal weight in all groups of babies

Of the various formulas studied SFH x AG =BWT in gms is a simple and reliable method of estimating foetal weight in all the groups of baby weight expect the very low birth weight and those of macrosomic size where study was inadequate due to small sample size.

This method of foetal weight estimation does not involve expertise nor does it need a per vaginal examination. The calculation is simple and is easily done by simple multiplications. Hence this method can be inclined in the MCH teaching program to train the junior doctors and the field workers involved in the antenatal care, for a better estimation of foetal weight at the peripheral centres.

Limitations of the study:

1. The study had the following limitations:
2. The study contained only 195 patients ie the sample size was small.
3. The number of patients in very low birth weight, big baby and macrosomic groups were limited.
4. There was no correction made for gravidity, weight of mother, obesity or weight of previous babies.
5. Inter observer variation was not found out for estimating the clinical method of baby weight estimation.
6. Ultrasound estimation of bay weight was not done in our study.

1. Poulas pp and longstadt jp: am journal of obstetrics and gynaecology 65253,1953
2. Devi pk and lokadam n: obstetrics and gynaecology of india 16:539,1966
3. Mcdonald e : J.A.M.A 47: 1979,1906
4. Johnson rw &toshach ce: am journal of obstetrics and gynaecology 68:891,1954
5. Johnson RW: am journal of obstetrics and gynaecology 74:929,1957
6. Cetrulo,c.l. and freeman,r: clinical obstet. Gynec.20:4,979,1977
7. Tewari&sood-journal of obstetrics and gynaecology of india vol 39,129,1989.
8. Mhaskar&mhaskar-vol51, no.4, July-august 2001 p 73
9. Mcsweeny,Daniel j:am journal of obstetrics and gynaecology.76:12797,1958.
10. Dawn cs-journal of obstetrics and gynaecology of india 133: vol 33, February 1983
11. Donaldson,s.w,and cheney,w.d:radiology 50:666,1948
12. Campbells &Dewhurst c: diagnosis of the small for date foetus by serial ultrasound cephalometry, lancet,2:1002,1971
13. Warsof,s.l,gohari,p.berkowitz,r.l and Hobbins,j.c:the estimation of fetal weight by computer assisted analysis,am.j. obstet.gynecol.128:881,1977.
14. Manning FA: intrauterine growth retardation. In fetal medicine. Principle and practice. Norwalk, ct, Appleton&lange,1995,p 317
15. Snijders RJM,Nicolaides KJ, fetal biometery at 14 to 40 gestation. Ultrasound- obstetrics and gynaecology 4:33,1994
16. Smith GCS,smith MFS,mcnay MB,fleming JEE: the relation between abdominal circumference and birth weight: finding in 3512pregnancies-british journal of obstetrics and gynecology 104:86,1997
17. Dashe JS, mclntire DD, lucas MJ,leneno KJ: impact of asymmetric versus systemic fetal growth restriction on pregnancy outcomes-SGI abstract 96:321,2000
18. Whaston W,and seeds J. sonographic diagnosis of macrosomia.in divon M(ed): abnormal fetal growth. New York, elseveir,1991,p 237.
19. Benaceraf BR,gelman R,frigolettoRD: sonographically estimated fetal weights: accuracy and limitations am j obstet gynecol 159:1118,1988.
20. Pallock RN,divon MY: intrauterine growth retardation: definition classification and etiology. Cli obstet gynecol 87:169,1992.
21. Hadlock FP,harrist RB sharman RS,et al:estimation of fetal weight with use of head body and femurmeasurements. A prospective study. Am J obstet gynecol 1985:151:333.
22. Sherman DJ,arieli S,tovbin j,seigel,Caspi e,bukovsky I :a comparison of clinical &ultrasonic estimation of fetal weight-obstet &gynecol 91:212,1998.