Low birth weight

5. Fetal outcomes Low birth weight Definition. Low birth weight (LBW) is a birth weight of <2500 grams. A summary of the odds ratios for each indi- cator appears in Table 10. The OR values for the in- dicators applied to all mothers are given in the first column. The same indicators applied only to sub- groups of mothers with a pre-existing height or weight deficit are presented in the next two columns respectively. The relationship between maternal nutritional status and low birth weight has been extensively studied using standard maternal anthropometric indi- cators based on height, weight, arm circumference, and combinations of these. The relationship is less direct than for IUGR (see below) since LBW can result from two separate and probably independent conditions: (i) the infant may be small-for dates (IUGR), or (ii) it may be normal-for-dates, but deliv- ered preterm (PTB). The IUGR component of LBW is conceptually related to maternal nutritional status and hence maternal anthropometry serves as a useful Table 10: Summary of combined odds ratios for each indicator of LBW. ORs refer to the relative risk for LBW for the lowest quartile versus the highest quartile of the indicator's distribution. Odds ratio for combined profiles With With pre- maternal pregnancy Indicator LBW heighta weightb Maternal height 1.7 Mid-upper-arm circumference 1.9 Pre-pregnancy weight 2.3c 2.6 Attained weight by month 5 2.4c 2.5 2.4 Attained weight by month 7 2.4c 2.6 2.7 Attained weight by month 9 2.5c 2.9 2.8 Pre-pregnancy BMI 1.8 BMI month 5 1.6 BMI month 7 1.9 BMI month 9 1.8 Weight gain: Pre-pregnancy to month 5 1.5 1.9 2.6 Pre-pregnancy to month 7 1.5 2.0 3.4 Pre-pregnancy to month 9 1.6 2.2 3.2 Month 5 to month 7 1.6 2.6 2.0 Month 5 to month 9 1.7 2.7 1.6 Month 7 to month 9 1.2 1.8 1.7 a In mothers with below average maternal height. b In mothers with below average pre-pregnancy weight. c Highest ORs for the single indicator are given in bold type. marker, but the route through which nutritional stat- us influences the onset of labour and hence preterm birth is less obvious (4). The magnitude of effect for any given anthropometric indicator of LBW will therefore be reduced compared with those reported for IUGR. Nevertheless, as LBW and its frequency are measures employed almost universally for both clinical and public health purposes, it is useful to establish the performance of this set of indicators with respect to it. Furthermore, in conditions of severe economic and environmental constraint with relatively high LBW rates, a substantial proportion of these will be attributable to IUGR. Thus in the early stages of programme development when LBW rates are relatively high, this outcome is useful for evaluating the response to intervention directed pri- marily at IUGR. Krasovec & Anderson (1) provide a summary of the experience of previous investigators in connection with LBW. The simple, stable and potentially useful indica- tors of matemal height (HT) and arm circumference (MUAC) show relatively low ORs for predicting low birth weight. The highest ORs for single indicators, highlighted in the first column, are for pre-pregnancy weight and attained weights at 5, 7, and 9 months. These range from 2.3 to 2.5, indicating relatively high associated risks. When these are applied to selected subgroups with height and weight deficits their power is increased moderately by 0.3 to 0.4 OR units, but only by introducing a two-step procedure involving pre-selection of the appropriate subgroups. It will be seen that the largest change in ORs in the subgroups is that associated with the weight gain indicators which increase by between 0.3 and 1.0 unit. Before concluding that these may be the best predictors for this outcome, the reliability of these OR estimates must be considered across the varying genetic, environmental and service conditions repre- sented by the 25 different study sites. Such variation is expressed graphically in Fig. 5 as a confidence profile. Weight attained. For these weight indicators there is more stability in the estimate at month 7 than is seen at either month 5 or month 9 and this is noticeable in the slightly tighter profile for the combined OR (Fig. 5). The profile for the pre-pregnancy period is narrower still, although the differences overall are moderate. There is no clear progression in the mean of the profiles across the periods as they display a narrow range from 2.3 to 2.5. Weight attained with maternal height. A comparison of Fig. 5 and Fig. 6 shows the effect of incorpor- WHO Bulletin OMS: Supplement Vol. 73 1995 21 Chapter 5 Fig. 5. The combined OR profiles for attained weight indicators and LBW are shown. Means (with 95% confidence interval) are estimated as: WT pp: 2.3 (2.1-2.5); WT/5: 2.4 (2.0-2.8); WT/7: 2.4 (2.1-2.7); and WT/9: 2.5 (2.2-2.9). The ORs are for weight below the lowest quartile versus weight above the highest quartile. Fig. 6. The combined OR profiles for weight attained indicators and LBW are shown. Means (with 95% confi- dence interval) are estimated as: WTpp: 2.6 (2.3-2.9); WT/5: 2.5 (2.0-3.2); WT/7: 2.6 (2.2-3.1); and WT/9: 2.9 (2.5-3.4). The ORs are for mothers with low weight (below the lowest quartile) and low height (below the median) ver- sus weight above the highest quartile and above median height. 1.0 1.75 2.5 3.25 4.0 0 1.0 2.0 3.0 4.0 Odds ratio Odds ratio ating maternal height into the equation. Here we see the profile for attained weight in the subgroup of mothers of below average height. The various pro- files are shifted to the right, although the size of the shift is rather small. Again there is relatively little difference in the mean ORs for these periods. The highest subgroup ORs predictably tend to be for indicators requiring measurement relatively late in pregnancy such as the 7th and 9th lunar months, although in this instance there is relatively little separation between the profile peaks of these indicators. In biological terms, it is to be expected that indicators operating late in pregnancy and close to the time of outcome will be relatively more powerful predictors of adverse outcome, in this case LBW. However, this also implies that identifi- cation of the need for intervention, such as food sup- plementation, is delayed and that the response to such special care will be correspondingly reduced. Given such operational requirements, WTpp and attained weight by lunar month 5 (16 to <20 weeks) in mothers of low height provide useful compromises, giving reasonable ORs (effect size), with early warn- ing of the need for intervention. Such choices may, however, be further modified when the specificity and sensitivity of the indicator are examined, as discussed below. Sensitivity and specificity Plots of the ranges for sensitivity (Se) and specificity (Sp) across data sets for all LBW indicators are pre- sented in Fig. 7. The indicators are numbered 1 to 35 and follow the sequence shown in Table 11. The horizontal lines in Fig. 7 mark the chosen minimum levels of Se and Sp. The trade-off between Se and Sp in individual indicators is obvious from these figures; indicators with low Se tend to have high Sp and vice versa, e.g., for weight gain from pre-pregnancy to 5 months (indicator 30). This indicator would successfully exclude nearly all mothers likely to have normal weight babies (high specificity) - and therefore will not be needing special care - but would fail to iden- tify many mothers who are due to deliver low-birth- weight babies (low sensitivity) and who would ben- efit from intervention. Maximum spread for Se is shown by attained weight at 9 lunar months in moth- ers with low pre-pregnancy weight (indicator 29), with values ranging from 0.28 to 0.94. Clearly this indicator has poor consistency in predicting LBW across the 25 data sets. Minimum spread for Se is shown by the weight gain from pre-pregnancy to 9 lunar months in mothers with a low pre-pregnancy weight (indicator 35) with values of 0.12 to 0.31. The latter has the second highest estimated OR at 3.2; however, this low Se range means that this indi- cator would not meet the criterion for an acceptable minimum Se (>0.35) with an Sp greater than 0.7. The same problem arises for indicator number 34 (weight gain from pre-pregnancy to 7 lunar months) which has the highest estimated OR for LBW of the 35 measures. WHO Bulletin OMS: Supplement Vol. 73 199522 Fetal outcomes Table 11: Performance of indicators for LBW. The asterisked ones, having met the selection cri- teria, are considered candidates for 'best' indicator. Low birth weighta Indicator No. Proportion Min Se Max Se 1. Maternal height 24 0.17 0.37 0.40 2. Pre-pregnancy weight 21 0.62* 0.35 0.47 3. Attained weight by month 5 15 0.53* 0.36 0.56 4. Attained weight by month 7 14 0.50* 0.38 0.52 5. Attained weight by month 9 17 0.47* 0.38 0.47 6. Mid-upper-arm circumference 14 0.14 0.36 0.39 7. Weight gain: months 5 to 7 12 0.08 0.37 0.37 8. Weight gain: months 5 to 9 13 0.00 9. Weightgain: months 7 to 9 13 0.15 0.36 0.45 10. Weight gain: pp to month 5 8 0.25 0.36 0.36 11. Weight gain: pp to month 7 8 0.25 0.35 0.35 12. Weight gain: pp to month 9 8 0.13 0.36 0.36 13. Pre-pregnancy BMI 21 0.38 0.35 0.40 14. BMI month 5 14 0.07 0.37 0.37 15. BMI month 7 13 0.08 0.36 0.36 16. BMI month 9 15 0.13 0.35 0.41 In mothers with low matemal height: 17. Pre-pregnancy weight 19 0.11 0.49 0.56 18. Attained weight by month 5 13 0.15 0.42 0.50 19. Attained weight by month 7 13 0.15 0.39 0.42 20. Attained weight by month 9 13 0.08 0.67 0.67 21. Weight gain: months 5 to 7 11 0.09 0.40 0.40 22. Weight gain: months 5 to 9 12 0.08 0.38 0.38 23. Weight gain: months 7 to 9 12 0.25 0.36 0.44 24. Weight gain: pp to month 5 8 0.13 0.36 0.36 25. Weight gain: pp to month 7 8 0.13 0.36 0.36 26. Weight gain: pp to month 9 8 0.25 0.37 0.41 In mothers with low pre- pregnancy weight: 27. Attained weight by month 5 8 0.13 0.47 0.47 28. Attained weight by month 7 8 0.13 0.43 0.43 29. Attained weight by month 9 8 0.00 30. Weight gain: months 5 to 7 6 0.00 31. Weight gain: months 5 to 9 6 0.17 0.39 0.39 32. Weight gain: months 7 to 9 7 0.29 0.38 0.40 33. Weight gain: pp to month 5 8 0.00 34. Weight gain: pp to month 7 8 0.00 35. Weight gain: pp to month 9 8 0.00 a No. is the number of data sets for which the indicator could be estimated. Proportion gives the proportion of data sets in which the Se/Sp criterion was met. The asterisk denotes that the indicator met the criterion (Sp >0.7 and Se >0.35) in over 40% of the studies. The observed maximum and minimum sensitivities across studies that meet the Se/Sp criterion are recorded in the final two columns. WHO Bulletin OMS: Supplement Vol. 73 1995 23 Chapter 5 Fig. 7. Ranges in computed sensitivity (left side) and specificity (right side) across all relevant data sets for all indicators for LBW. The indicators are numbered 1 to 35 and follow the sequence shown in Table 11. 1.0 _ 1.00.9r,,,,,,,,,,,,,,,,,,.. - ; l l 0.sLl90 9--8 ---- ---- ----- ------- --------------- ------------- -- -+--------- ---- -f0C - - ---E--r-W --t----- ----- - t -F--------- --t -l - E 0.9 0.8I- ------- .8 _ 0.7 - ---07--- ----- ----- - 0;7 c 0.5 ----------------I---- - -----t-F-t ---------I -----r-t - -- -----T - --- -----L ------ ----------------- -------l-t--- --E- ------------ 0.5 cf 0.4t;- - ---- Ef --ip - W --- T ---- ---------- --' ----------------- ------ ----------- t- ----------- 104 COi.~~~~~~~j 1 p 0.2 L -- .0------ ---------;-----;------------------;----------;----------; - --------- .2 o o 0 0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35 Indicators Indicators WHO 95 In summary, the four indicators that meet the full study criteria for consistency are pre-pregnancy weight and attained weight by months 5, 7 and 9 (see Table 11). In this group the maximum Se is 0.56 and minimum, 0.35 (specificity >0.7 in all cases). BMIpp falls just short of the final selection requirements at 38%. The acceptable indicators showed relatively high ORs in Table 10, ranging from 2.3 to 2.5. While ORs for these indicators increased when applied to mothers of low height and low pre-pregnancy weight, the Se/Sp values were not sufficiently con- sistent across studies to permit their further consider- ation. The profiles for the marked indicators confirm that these have reasonable reliability across the stud- ies. In service terms, pre-pregnancy or early preg- nancy weights are difficult to obtain as registration is often delayed until around the 20th week, particular- ly in the low-income groups with the highest rates of LBW. This leaves the attained weights at 5 or 7 lunar months as the most useful service indicators for LBW, reflecting current practice in many set- tings. Intrauterine growth retardation Definition: Intrauterine growth retardation (IUGR) has occurred when a birth weight less than the 10th centile of weight-for-gestational age is determined during the analysis by comparing birth weight with reference values derived from a high-income popula- tion (5). This group includes infants who are full- term (.37 weeks) growth retarded as well as preterm (<37 weeks) growth retarded. In view of the com- mon 'working definition' of a growth retarded infant, i.e., full-term infant with birth weight less than 2500 g, the analysis was also undertaken for this group of infants. A summary of the ORs for each of the basic and subgroup indicators, laid out in the same format as for LBW, is presented in Table 12. In the final col- umn, the odds ratios for low-birth-weight infants delivered at full term (FT-LBW) are added. The biological relationship between maternal nutrition status and IUGR is more precise than for LBW. The papers by Kramer (4) and Krasovek & Anderson (1) discuss the mechanisms underlying this relationship and also review the past experience of investigators in this area. As stated in the previous section, LBW contains a variable component of pre- term births that are not growth retarded but which have low weight at birth. The size of this component will vary under different socioeconomic and health care conditions. Determining the size of the compo- nent is often difficult as service operations do not necessarily record a diagnosis of IUGR for lack of suitable reference tables and/or because the appropri- ate service response does not require such discrimi- nation. For the future, the routine recording of IUGR as a component of LBW rates will greatly enhance the value of service statistics to reproductive epidem- iology and subsequent programme planning and evaluation. ORs for simple indicators range from 1.4 (WTg/7-9) to 3.0 (WT/7 and WT/9). Maternal height and arm circumference show relatively low ORs (1.9 and 1.6) which are similar to those found 24 WHO Bulletin OMS: Supplement Vol. 73 1995