Contents - Previous - Next


In general there is much less new information concerning the energy requirements for lactation than there is for pregnancy, but this is offset by the fact that the basis for computing recommended intakes is more straightforward. The energy costs of lactation are customarily calculated as: [NPNL requirements] + [Breastmilk volume × Energy density × Conversion efficiency] [Change in body fat] [Changes in activity]. Possible changes in maternal metabolic rate and DIT can either be incorporated into the factor describing 'conversion efficiency', if this is expressed as gross efficiency, or may be dealt with as a separate category added to the above scheme. The following sections summarise the available quantitative data on each component.

Energy cost of lactation: energy output in breast milk

Volume of milk produced. Figure 15 illustrates that the volume of milk produced at peak lactation is remarkably similar in groups of women from a wide variety of nutritional and cultural circumstances.* For a previous IDECG report we demonstrated that lactational performance does not appear to be compromised by low BMI (Prentice et al, 1994). It therefore appears that even small, undernourished women produce similar amounts of milk to larger, well-nourished women. Consequently there is little difficulty in arriving at acceptable average values for milk production in early lactation. The values selected by the UK Expert Panel (COMA, 1991) were derived as the average of two of the most carefully conducted series of measurements. The data were from well-nourished women from England (Paul et al, 1988) and Sweden (Sadurskis et al, 1988) and incorporated a 4% correction to allow for insensible water losses by the infant during test-weighing. In Table 8 we have combined these with data from the Californian DARLING Study (Heinig et al, 1993a; 1993b; KG Dewey, personal communication). These values are broadly representative of other populations. They are slightly lower than the 1985 figures particularly for 3-6 months post partum. The values after 6 months of lactation in Table 8 are a rough approximation based on the 1985 figures supplemented by data from other studies of extended lactation (Prentice et al, 1986).

Figure 15

Similarity of breast-milk output in groups of women from different nutritional and cultural settings. Reproduced permission from (Prentice et al, 1986). Data are from 1118 women in 26 studies.

In later lactation the quantity of milk produced is largely dependent on cultural practices and on the mother's intentions concerning weaning. The UK Panel adopted two different sets of figures: (a) for women 'who practice exclusive or almost exclusive breast-feeding until the baby is 3-4 months old and then progressively introduce weaning foods as part of an active weaning process' and (b) for women 'who introduce only limited complementary feeds after 3-4 months and whose intention is that breast-milk should provide the primary source of nourishment for 6 months or more'. Making a distinction between partial and full breast-feeders in this way is worth considering, but would have to be made more general for a WHO document. In Table 8 the figures for partial breast-feeders have simply been set as 50% of the output of full breast-feeders.

*Note that although more modern data are available than summarised in this figure they tend to fit the same pattern and a fresh analysis was not considered necessary.

Energy content of breast milk. It is notoriously difficult to obtain a truly accurate estimate of the energy density of breast milk transferred from the mother to her infant. This is because complex diurnal, within-feed and between-breast changes in milk fat content make it virtually impossible to design a sampling protocol which will yield an integrated estimate of energy density over the whole day and in a manner which will not affect the natural pattern of milk flow. Once milk has been sampled, the energy content has usually been calculated by applying modified Atwater factors to chemical estimates of the proximate constituents.* A large government survey in the UK reported average metabolisable energy of mature milk to be 2.90 kJ/g (0.69 kcal/g) equivalent to a gross energy (GE) content of 3.06 kJ/g (0.73 kcal/g) (DHSS, 1977). This agrees exactly with estimates from the WHO multi-centre collaborative study on breast-feeding. Direct estimates of the GE content of pooled milk from well-nourished women has yielded a value of 3.10 kJ/g (0.74 kcal/g) which is also in close agreement (Garza & Butte, 1986). Many other studies yielding similar values could be quoted.

* Note that this calculation incorporates estimates of digestibility for protein, fat and carbohydrate. It therefore represents metabolisable energy available to the child, but underestimates the gross energy which must be provided by the mother (the difference appears in the infant's stools). This represents a clear error in the WHO/FAO/UNU (1985) assumptions and should be corrected even though the impact of the change is not large.

Table 8 Suggested breast-milk volumes for use as basis of calculations


Proposed volumesa

1985 valuesb

All women










Full breast-feeders










Partial breast-feedersc



Not applicable



Not applicable



Not applicable

a Derived from studies in Sweden (Sadurskis et al, 1988) England (Paul et al, 1990), USA (Heinig, 1993a,b) and the WHO Collaborative Study on Breast-Feeding (WHO, 1985). Values are g/d and include a 4% correction factor to allow for insensible water losses from the baby during test-weighing.
b Derived only from the WHO Collaborative Study on Breast Feeding. Values are g/d.
c Infants receiving breast-milk as only source of milk.

An alternative approach has been employed in an attempt to assess the energy content of breast milk indirectly from measurements of energy expenditure and from the composition of new tissue formed in breast-fed infants (Lucas et al, 1987). It is claimed that this gives a more representative picture of milk as suckled by the infant, but the method necessitates a number of assumptions and has not been independently validated. Its results are therefore viewed with some caution (COMA, 1991). Using this approach the calculated metabolisable energy in early lactation was 2.40-2.50 kJ/g (0.57 - 0.60 kcal/g). This should be increased by 5% (to about 2.55 kJ/g or 0.61 kcal/g) to obtain GE.

A consideration of the energy content of milk from under-nourished mothers is not strictly relevant to the formulation of recommended dietary allowances since these should cover optimal milk quality. Nonetheless it is worth noting that, in spite of some reports to the contrary, the consensus of opinion is that milk energy holds up very well in the face of maternal under-nutrition. Several reviews are available (Ferris & Jensen, 1984; Jensen & Neville, 1985; Neville & Neifert, 1983). Our own earlier analysis of relationships between maternal BMI and milk energy content reported that there were three studies in which there was a significant positive association between the two variables, but that most revealed no association and at least one showed a reverse trend (Prentice et al, 1994).

The choice of a figure for the energy content of milk should lie somewhere between 3.00 kJ/g or 0.72 kcal/g (average GE from proximate analysis or bomb calorimetry of pooled mature milk) and the new estimate of 2.55 kJ/g (Lucas et al, 1987). The 1985 value was 2.90 kJ/g, and the UK (COMA, 1991) adopted 2.80 kJ/g (0.67 kcal/g) as a compromise. We have used 2.80 kJ/g in the calculations below.

Efficiency of converting dietary energy to milk energy. The 1985 Report assumed a figure of 80% efficiency for milk synthesis. This was based on a study by Thompson et al (1970) and was derived as the lower 95% confidence interval of their individual estimates. The average apparent efficiency was in fact close to 100%. There are a number of serious problems with this derivation. These include the use of a cross-sectional design comparing lactating and non-lactating women; a heavy dependence on estimates of food intake in the mothers; an absence of any milk production measurements (milk energy transfer was estimated from the growth rate of the breast-fed infants and assumptions regarding the energy cost of growth); and an absence of estimates of changes in maternal fat stores. We argue below that the estimate of 80% efficiency should not be altered, but that alternative justifications for the figure should be adopted.

The efficiency with which a mother can convert dietary energy can be estimated in two ways: as biochemical efficiency or as calorimetric efficiency.. Both involve a number of assumptions and are therefore imprecise, but they provide a reassuring level of agreement.

Biochemical efficiency. It is possible to compute the biochemical efficiency of synthesising each of the proximate constituents of breast milk from the known stoichiometric equations and the obligatory heat losses associated with each step of the synthetic process. A full description is available elsewhere (Prentice & Prentice, 1988). The calculated efficiencies are 95% for lactose synthesis, 88% for protein synthesis, 73% for de novo fat synthesis; and 98% for transfer of pre-formed fat. Human milk is notable for the small proportion of fat which is synthesised (Hachey et al, 1989), even when women are consuming a low-fat diet. Most fat is assumed to be transferred from the diet, even if it is recycled (at a low biochemical cost) through adipose tissue. The overall efficiency of synthesis for all proximate constituents combined is therefore very high; averaging about 91-94%. This is much higher than in many mammals (especially ruminants) which are forced to synthesise a higher proportion of milk fat.

It is important to emphasise that these estimates of biochemical efficiency must represent the absolute maximum efficiency since they are derived from a gross simplification of the biochemical pathways involved. They make no allowances for the numerous other minor processes which are needed to support the synthetic process (Prentice & Prentice, 1988). Allowances must also be made for the further costs associated with digestion, absorption, interconversion, transport and storage of the dietary fuels and protein. These can probably best be assessed from direct measurements of DIT. Most studies report that 6-14% of the energy of a mixed meal is dissipated as heat (Jéquier & Schutz, 1985). Illingworth et al (1986) reported a 30% diminution of DIT during lactation. This figure is questioned below, but if it is accepted it would reduce estimates of thermogenesis to about 8% of any extra metabolisable energy ingested to support lactation. Adding this to the biochemical estimates for synthesis gives an overall efficiency of about 83%, which lends strong support to the assumption adopted in the 1985 report.

Calorimetric efficiency. An alternative approach to the theoretical calculations of biochemical efficiency outlined above is to perform metabolic balance studies which include calorimetric measurements of the maternal maintenance requirements. Calorimetric efficiency of milk production can then be computed as:

where the denominator is the sum of excess energy derived from the diet and by mobilisation of body fat stores, and-where the term 'maintenance needs' encompasses all components of energy expenditure including physical activity.

This method has been developed by livestock scientists and the literature is dominated by estimates from dairy cattle. Unfortunately there are few data from other species. Calorimetric efficiency in ruminants ranges from about 67-72% (Blaxter, 1962; Flatt & Moe, 1971). Some values from non-ruminants are: rat 57% (Roberts & Coward, 1982); rabbit 76% (Partridge et al, 1983); and pig 85% (Lodge, 1957). Humans are likely to lie towards the top end of this range due to the low level of fat synthesis.

Efficiency is known to vary with the state of maternal energy balance since mobilised body energy is used more efficiently than dietary energy. This is in accord with theoretical predictions based on the difference between transfer and synthesis costs for fats. The efficiencies of conversion of dietary and body energy have been calculated as 63% and 84% respectively in the dairy cow (Flats & Moe, 1971) and 74% and 94% in the rabbit (Partridge et al, 1983). In humans this difference is likely to be much less pronounced since the high fat content of the diet removes the need for substantial lipogenesis.

Recent calorimetry data from lactating women in The Gambia have been used to calculate a variant of calorimetric efficiency (Frigerio et al, 1991a). The study assumed that the extra costs of milk synthesis would be measured as part of BMR and computed efficiency as (%):

Based on cross-sectional measurements in lactating and non-lactating women, efficiency was found to be 94.2 3.5%. It must be stressed that this value is not directly comparable with the livestock definition of efficiency which incorporates all components of daily energy expenditure in the denominator. It should also be noted that this approach assumes that milk synthesis is equally active during the measurement of BMR. This assumption may not be entirely correct since, by definition, the women are measured at the greatest distance from a previous meal. If milk synthesis (and hence its effect on BMR) diminishes in the fasted state this would tend to over-estimate efficiency.. Finally it should be noted that in previous longitudinal studies in the same population (Lawrence et al, 1986) BMR was actually lower in lactating women than prior to conception. In the above method of calculation this would yield an apparent efficiency in excess of 100%. The only way that this can occur is for other components of the maternal metabolism to be down-regulated to an extent which more than compensates for the known obligatory costs associated with synthesis. The need for physiological down-regulation seems to be caused by an inadequate diet and might have detrimental consequences to the mother. Such high apparent efficiencies should not therefore be used as the basis of recommended requirements. If BMR is not suppressed in lactating women (see below) this would lead to lower estimates of efficiency if calculated using the Frigerio formula.

Blaxter (1962) noted that calorimetric efficiencies are usually 10-15% lower than biochemical efficiencies. This is entirely to be expected from the known omissions in the biochemical estimates. Applying this correction to the estimate of biochemical efficiency derived above would yield a figure of 80-85%. Once again this is in accord with the figure used in the 1985 recommendations, and suggests that no change is needed.

Energy cost of lactation: change in basal metabolic rate

Introduction. It would be anticipated that the BMR of a lactating woman would be slightly higher than in her non-lactating state if, as is generally assumed to be the case, milk synthesis is a continuous process in humans and is active during measurements of BMR. The increase would be equivalent to the 'inefficiency' of the synthetic process (i.e. 20% if we accept the 80% estimate of efficiency).. At 3 months post partum this would be about 460 kJ/day (110 kcal/day) or around 10% of BMR (derived as 820 g × 2.80 kJ/g × 0.20). Any value lower than this would be indicative of energy-sparing.

There are fewer BMR data available for lactation than for pregnancy, and there is not such a clear consensus. An earlier review (Prentice & Whitehead, 1987) tentatively concluded that there was quite good evidence that BMR was lower during lactation than in the NPNL state (see Figure 16), and that this could represent significant evidence of energy-sparing when considered in the light of the expected rise in BMR explained above. However, this analysis included data from under-nourished women and more recent studies indicate the need for a reappraisal.

Studies from developed countries. The earliest studies from well-nourished women certainly suggested that BMR was suppressed (Johnston et al, 1937; Rowe & Boyd, 1932). The study by Rowe & Boyd involving a total of 77 women in Boston seems particularly convincing and illustrates a U-shaped depression in BMR coincident with peak lactation and reaching a minimum value 15% below the Harris-Benedict and Aub-Dubois standards. By 12 months post partum BMR was only 2-4% below the standards. Similar results were later obtained by Blackburn & Calloway (1985) who made RMR measurements in 13 lactating women at 8-12 weeks post partum. On average, RMRs were 23% lower than predicted, but these data should be treated with considerable caution since it is stated that the women were allowed to sleep during the measurement if they chose. Clearly both studies are handicapped by the fact that they compare BMR in lactation with predicted values rather than with true measurements in a non lactating state.

In contrast, no differences in RMR were found between breast- and bottle-feeding women in a cross sectional study in Scotland. When RMR was remeasured post-lactation in the breast-feeding women it was also unaltered (Illingworth et al, 1986). In a longitudinal study in Cambridge (Goldberg et al, 1991), BMR measured at 4, 8 and 12 weeks lactation (5887, 5846 and 5630 kJ/day or 1406, 1396 and 1345 kcal/day) was very similar to the post-lactation value (5858 kJ/day or 1399 kcal/day). When expressed per kg fat-free mass, the values (144, 145, 141 and 142 kJ/kg FFM/day) again indicated no suppression during lactation.

A longitudinal study of 40 lactating Dutch women is a little more difficult to interpret (van Raaij et al, 1991). At 9 weeks post partum, average BMR was 5940 kJ/day (1418 kcal/day) and at 12 weeks gestation was 6380 kJ/ day (1524 kcal/day), suggesting that BMR was 440 kJ/ day lower during lactation. A cross-sectional comparison using a matched group of 16 non-lactating women measured at 9 weeks post partum revealed no difference in BMR (average 5940 kJ/day or 1419 kcal/ day). On the basis of data in Figure 16, the 12 weeks gestation figure should be adjusted downwards by about 250 kJ/day (60 kcal/day) to represent pre-pregnancy BMR. The residual difference is small. Furthermore, longitudinal measurements at 5, 9, 13, 27 and 56 weeks post partum, in a subgroup of 16 of the 40 lactating women suggested that there was no influence of decreasing milk flow or weaning (6050, 6088, 5929, 6050 and 6042 kJ/day or 1445, 1454, 1416, 1445 and 1443 kcal/day respectively). In a more recent longitudinal study of 24 Dutch women (Spaaij et al, 1994a) mean RMR was 3.80 0.35 kJ/min (0.91 0.08 kcal/min) before pregnancy and 3.98 0.40 kJ/min (0.95 1.10 kcal/min) at two months lactation. The increase was statistically significant (p < 0.05), but represents less than a 5% change.

Figure 16 Analysis of changes in BMR during pregnancy and early lactation. Reproduced with permission (Prentice & Whitehead, 1987). (Basal)

Figure 16 Analysis of changes in BMR during pregnancy and early lactation. Reproduced with permission (Prentice & Whitehead, 1987). (Resting)

Increases in RMR during lactation were also observed in a longitudinal study of 23 Swedish women (Sadurskis et al, 1988). Post-partum RMR values were 6364, 5904 and 6029 kJ/day (1520, 1410 and 1440 kal/ day) at 5-10 days, 2 months and 6 months post partum, respectively compared to 5611 kJ/day (1340 kcal/day) before pregnancy. Breast-feeding patterns at 6 months post partum were not clear from the original publication, but all women were lactating at 2 months. At this time the increase in RMR above the pre-pregnancy measurement (314 670 kJ/day or 75 160 kcal/day) was significant (p < 0.05). Differences were also significant when expressed per kg FFM since, surprisingly, FFM was found to be appreciably lower in lactation (42.3 vs 45.0 kg, p < 0.05). As in Spaaij's study the difference only amounted to about +5%.

We conclude that in well-nourished women the most robust studies involving longitudinal measurements suggest that RMR is unchanged or only very slightly elevated in lactation.

Studies from less developed countries. There are much fewer studies from women in developing countries, but with one exception they support the data from developed countries. In Guatemala (Schulz et al, 1980) RMRs were similar in 18 lactating women (3.18 0.42 kJ/min or 0.76 0.10 kcal/min) and 6 non-lactating women (3.06 0.54 kJ/min or 0.73 0.13 kcal/min). In The Gambia, (Lawrence et al, 1986) BMR was about 420 kJ/day (100 kcal/day) lower in lactating women compared to their own pre-pregnancy baseline and, furthermore, BMR was lower in unsupplemented women than in those receiving an energy-dense dietary supplement. However, cross sectional comparisons conducted in the same villages yielded different results (Singh et al, 1987). Mean RMR was 5471 776 kJ/day (1307 185 kcal/day) in 15 lactating women and 5012 130 kJ/day (1197 31 kcal/ day) in six NPNL women; the differences were not significant. ln The Philippines, Guillermo-Tuazon et al (1992) found no significant difference between BMR measured during lactation and at 13 weeks gestation in the same 32 women.

Energy cost of lactation: change in diet-induced thermogenesis

Two longitudinal studies have been published on changes in DIT in response to a standardised liquid meal during lactation. lllingworth et al (1986) observed a significant ( - 30%) reduction in DIT during lactation, compared to post-lactational measurements in 12 women (0.42 vs 0.60 kJ/min or 0.10 vs 0.14kcal/min). However, in a longitudinal study of 24 Dutch women (Spaaij et al, 1994a), DIT at 2 months lactation was the same as the pre-pregnancy baseline measurement (0.64 kJ/min or 0.15 kcal/min).

DIT in lactating and non-lactating women has been evaluated in two cross-sectional studies. In both, the energy content of the test meal was larger in the lactating than in the non-lactating women, and therefore the results cannot be compared. Motil et al (1990) observed a highly significant 4.4-fold higher DIT (kcal/kg body wt/day), in 12 lactating women compared to nine non lactating women, whereas the energy content of the test meals used was only 1.5-fold higher. This result may be influenced by quite considerable differences in body composition. In a Gambian study DIT, expressed as a percentage of the energy consumed, in 10 lactating women (at 2 months post partum) was similar to that in 12 NPNL women studied during the same (dry) season (Frigerio et al, 1992). A second group of lactating women was studied during the rainy and nutritionally unfavourable season. DIT was lower ( - 18%) than in the lactating women studied in the dry season (p < 0.05), and lower ( - 16%) than in the NPNL control group (also studied in the dry season).

The results of these four studies are conflicting. There is no clear evidence that DIT is either increased or decreased during lactation. We will assume that it remains constant when expressed as a proportion of energy ingested.

Energy cost of lactation: change in energy costs of activity

Energy cost of standardised activity. Several investigators have studied the energy cost of cycle ergometer or treadmill exercise in lactating women (Blackburn & Calloway, 1974, 1976, 1985; Edwards et al, 1981; Knuttgen & Emerson, 1974; Lotgering et al, 1991; Pernoll et al, 1975; Seitchik, 1967; Spaaij et al, 1994a; Uleland et al, 1973; van Raaij et al, 1990b). However, it was not possible to perform an analysis of any potential effects of lactation because subject groups were not clearly defined. For example, in some studies which included data from post-partum women it was not stated whether or not they were lactating. In others, lactating and non-lactating women were treated as a single group and it was not possible from the information given in the original papers to sub-divide the groups.

Changes in activity patterns (time-motion studies). In a study by van Raaij et al (1990a), eighteen Dutch women with sedentary lifestyles were studied longitudinally and kept activity diaries at 5, 9, 13 and 27 weeks lactation and at 56 weeks postpartum when NPNL.

Compared to NPNL the time spent sitting quietly or whilst carrying out light activities was increased at 5 weeks lactation (+36%, from 345 to 470 min/day). The time spent on light/moderate sitting activities was ( - 48%) lower at 5 weeks lactation. Increases were observed from 9-27 weeks but values remained lower than NPNL. The time spent on standing/light standing activities was lower at 5, 9 and 13 weeks lactation whilst at 27 weeks they were very similar to NPNL. There was very little change in the times spent moderately active, walking and cycling. The amount of time spent on infant feeding progressively decreased between 5 weeks (145 min/day) and 27 weeks lactation (75 min/day). Very little time (5 min/day) was spent on recreational activities when NPNL and at 27 weeks lactation, and none at all at 5, 9 and 13 weeks. PALs calculated from these data were 1.47, 1.50, 1.48, 1.5;0 and 1.53 at 5, 9, 13, 27 weeks and NPNL, respectively. Net costs of physical activity were calculated to be 200 kJ/day (48 kcal/day) lower in lactating compared to NPNL women.

The findings by van Raaij et al, are broadly similar to observations from British women (Goldberg et al, 1991). The lower PALs in lactation, measured by doubly-labelled water (see below) compared to NPNL were substantiated by responses to interview questionnaires. All women reported being less active and more tired, especially during the first 4 weeks when more time was spent sitting and sleeping (not necessarily as a result of suckling) and only minimal time was spent on shopping or housework. Many women did not resume discretionary activities until 8 weeks post partum or later.

In a longitudinal study of lactating Swedish women, physical activity was assessed using activity diaries kept over 24 h (Sadurskis et al, 1988). Activities were classed into nine categories and assigned a multiple of BMR (FAO/WHO/UNU, 1985). The weighted average was used as an indicator of physical activity. Values were the same at two months lactation and before pregnancy (1.59 and 1.60, respectively). The women were considered to have sedentary lifestyles and the authors stated that interviews and questions concerning sports, strenuous work, mode of transport and sleeping habits corroborated the diary data. However, when doubly-labelled water data from these subjects were analysed (Forsum et al, 1992), the subjects were re-considered to have more active lifestyles (see below).

In a cross-sectional study in The Gambia (Roberts et al, 1982), activity records were made over 3-10h and activity recalls used to study activity patterns. The data were presented for the total time spent active during the day. Activities included farmwork, housework and religious duties and were very dependent on season. The stage of lactation did not affect the total time taken or activity levels when performing household tasks or light farming work, but did influence the level of activity during heavy farming. Levels of activity were also affected by whether or not the women were accompanied by their infants. Seasonally adjusted figures for 0-11 months lactation are shown in Figure 17. On average 564/min/day was spent on activities in NPNL women and 501 min/day in [lactating women. Activity was lowest at 1 month lactation (364 min/day), but this was due to the tradition of confinement to the family compound during this time. Levels gradually increased to NPNL levels (550 min/day) by 11 months lactation.

In a longitudinal study of 40 Philippino women 5-day activity records were made during pregnancy and lactation (Tuazon et al, 1986, 1987). Observations were only made over 12 h, so overnight sleeping, Iying or resting data are unavailable. 'Baseline' data were taken at 14 weeks pregnant. Light/moderate activities were carried out in the first few months, but then returned to more habitual heavy levels. As lactation progressed resting activities decreased and other activities increased, except for light housework which remained the same. There was no suggestion that physical activity was reduced. PALs were calculated as 1.61 and 1.80 at 6 and 12 weeks Lactation, respectively.

Figure 17 Time allocation data from lactating women in The Gambia. Taken from Roberts et al (1982).

In a cross-sectional study (Panter-Brick, 1993) conducted in Nepal, NPNL and lactating women (between 0 and 35 months) were compared. Activities were recorded by observers during two non-consecutive days. PALs (averaged across seasons) averaged 1.92 in NPNL and 1.80 in lactating women. Time allocated to different activities was highly dependent on season, and activity levels were not significantly different between NPNL and lactating women in periods of intense activity. The author concluded that the seasonal constraints on the women's work prevented them from significantly curtailing physical activity at these times and effectively limited the scope of behavioural mechanisms for saving energy to support lactation.

We conclude that although there is a limited number of studies and there are differences with respect to populations and study design there are some similar trends in the data.

In developed countries women tend to decrease total physical activity, especially with respect to moderate and discretionary activities. This is sometimes, but not always, associated with childcare, for example more time spent sitting. These reductions in activity do result in energy saving which may help to support the costs of lactation. However, in women whose activity levels are already low, the potential for such savings is limited. Conversely, in affluent, sedentary women, the increased demands of childcare may serve to increase energy expenditure. An important consideration is that women have to be economically and socially able to 'afford' to reduce physical activity, for example by giving up paid employment.

In developing countries, although activity levels are generally much higher and therefore the potential for savings by reducing expenditure are greater, in practice women cannot or do not reduce their activities during lactation. Usually where decreases have been observed this is due either to cultural practices (e.g. confinement) or seasonal effects.

As for pregnancy, we propose that any new recommendations should be very cautious about making assumptions about possible energy-sparing decreases in physical activity during lactation.

Energy cost of lactation: total energy expenditure

Respiration chamber studies. The only study which has reported data from 24 h whole-body calorimetry measurements of lactating women is that by Frigerio et al (1991b) The combined 24 h CalEE of Gambian mothers and their infants measured in 16 lactating women and 16 controls was 2.3 MJ/day (549 kcal/day) higher in the lactating women. The authors attributed 66% of this to the infant's energy expenditure (1.5 MJ/day or 358 kcal/ day) and 33% (0.8 MJ/day or 191 kcal/day) to increased spontaneous activity of the mother.

Doubly-labelled water studies. Only four studies have used doubly-labelled water to measure TEE in lactation. Two of these were longitudinal in British (Goldberg et al, 1991) and Swedish (Forsum et al, 1992) women. One was cross-sectional in Gambian women (Singh et al, 1989). One other study, conducted in the USA (Lovelady et al, 1993), measured energy expenditure only once in women between 12 and 26 weeks lactating who were either exercising or not exercising, but there were no measurements of non-lactating women.

Mean** TEE: values in NPNL women ranged between 9.8 MJ/day (2340/ kcal/day) (UK) and 10.8 MJ/day (2580 kcal/day) (Sweden). Figure 18 illustrates the effects of lactation on absolute TEE for all studies. TEE in Swedish and Gambian women was higher in lactation than NPNL, whereas in the UK women, there was a marked decrease, especially at 4 weeks. Mean changes ranged from increases of 0.42 MJ/day (100 kcal/day) at 20 weeks in Gambian women to decreases of 0.94, 0.69 and 0.83 MJ/day (225, 165 and 198 kcal/day) in British women at 4, 8 and 12 weeks, respectively. In their study with no NPNL controls, Lovelady et al (1993) found a mean TEE of 10.1 MJ/day (2410 kcal/day) at 12-26 weeks; in the middle of the range from the other three studies.

** Note that DLW estimates do not include the energy transferred in milk, but do include the biochemical costs of synthesis.

Energy expended on TEE-BMR averaged 5.1 MJ/day (1218 kcal/day) in the Gambian women, higher than both Swedish (4.7 MJ/day or 1123 kcal/day) and British (2.9, 3.2, 3.3 MJ/day or 693, 764 and 788 kcal/day) women. The Gambian values were slightly (0.1 MJ/day or 24 kcal/day) higher during lactation than NPNL. In contrast, there was a marked decrease in the mean values for the Swedish (0.5 MJ/day or 119 kcal/day) and British (0.98, 0.68, 0.60 MJ/day or 234, 162 and 143 kcal/ day) women.

PALs for both non-lactating and lactating Gambian women were very high averaging 1.98 and 1.97, respectively. This reflects the women's lifestyle and the fact that measurements were made at a period of peak agricultural activity. When non-lactating, the Swedish women also had high values averaging 1.94 (the authors described the women as 'professionally active') decreasing to 1.82 and 1.79 at two and six months lactation, respectively. The British women had more moderate levels of activity with PALs averaging 1.66 when non lactating. Values during lactation were markedly lower averaging 1.50, 1.56 and 1.59 and 4, 8 and 12 weeks lactation, respectively.

Figure 18

Total energy expenditure (TEE) measured by doubly-labelled water during lactation. Squares, The Gambia (Singh et al, 1989); triangles England (Goldberg et al, 1991); circles, Sweden (Forsum et al, 1992).

With only four studies and very different population groups and lifestyles. it is impossible to draw firm conclusions about changes in TEE during lactation. In women who are obliged to continue working, possibly at very high levels of physical activity, the energy costs of lactation need to be supported by body fat mobilisation or increased energy intake. In women who can afford to, energy costs may be partially met by decreases in energy expended on physical activity.

We conclude that there is insufficient data even to make qualitative statements about 'normal' changes in TEE during lactation. There is no clear evidence that activity is habitually decreased as an energy-sparing strategy.

Energy supply during lactation: change in energy intake

Numerous assessments of EI have been made during lactation. However the majority were cross-sectional studies, or longitudinal studies which did not include a pre or post-lactational baseline measurement (see Figure 12). Only studies of longitudinal food intake measurements with a pre-pregnancy, first trimester or post-weaning baseline have been included in this review (Allen et al, 1992; Black et al, 1986; English & Hitchcock, 1968; Guillermo-Tuazon et al, 1992; Prentice et al, 1986; Sadurskis et al, 1988; Schofield et al, 1987; Spaaij, 1993; van Raaij et al, 1991; Whichelow, 1975).

Figure 19 shows the longitudinal changes in EI during lactation in women from developed and developing countries. Whilst there are large differences, the majority of studies showed an increase of between 0.3 to 1.6 MJ/day (70 to 380 kcal/day). Three studies provided multiple estimates of EI during lactation, from which it is apparent that intake was highest during the first 3 months post partum and then gradually declined. Only the study of Scottish women (Schofield et al, 1987) suggested that EI falls slightly during lactation. Whichelow's study of English women (1975) shows greater increases than any of the other studies, except that of Allen et al (1992) in Mexican women. Whichelow's paper however does not describe the methods used for data collection and it is possible that the discrepancy is a result of inadequate methodology. The two outliers (Schofield et al, 1987; Whichelow, 1975) have been excluded from the average intake figures shown by the dotted line in Figure 19. The average increase in EI 3 months post partum, the time considered to be peak lactation, was 1.35 MJ/day (322 kcal/day) and 1.01 MJ/ day (241 kcal/day) in the developed and less developed countries, respectively.

The mean changes have been ranked in order of decreasing intake in Figure 20. This highlights the outliers which have been discounted. The remaining studies reported average changes of between 0.2 MJ/day (48 kcal/day) (Schofield et al, 1987) and 2.5 MJ/day (595 kcal/day) (Allen et al, 1992).

The average increase in EI by peak lactation in the 12 longitudinal studies was 1.47 MJ/day (351 kcal/day). This is considerably lower than the figure of 1.9 MJ/day (455 kcal/day) in the review by Prentice & Whitehead (1987), who estimated the change in intake during lactation from a number of cross-sectional studies. This cross-sectional analysis was however only a crude estimate of the change in intake since the majority of the baseline NPNL data did not originate from the same studies as the lactational intake data.

The measurement of habitual food intake during lactation is beset with the same potential errors which affect all studies of EI regardless of subject group. The major errors are conscious or sub-conscious underreporting of intake by the subject and/or dieting during the study period. The latter may be a particular problem although most women are aware that dieting is discouraged during lactation. The issue of measurement fatigue raised in relation to longitudinal measurements in pregnancy is unlikely to bias the current assessment of an increment of about 1.5 MJ/day, because about half of the studies measured NPNL intake after lactation

We conclude that in affluent societies it seems clear that, as in pregnancy, observed increases in EI do not match the calculated incremental costs of full lactation. These costs can be subsidised by mobilisation of body fat and/or changes in activity pattern. It would be imprudent to assume that poorly nourished women could routinely make use of either of these strategies.

Energy supply during lactation: weight changes and mobilisation of body fat

It is often assumed that lactation is associated with weight losses and that the main purpose of the extra weight and fat which is usually deposited during pregnancy is to support the energy costs of lactation. This is implicit in current recommendations for EI of lactating women which include qualifying statements according to whether fat is gained during pregnancy and/or mobilised during lactation (see Table 2). Humans stand out from many other mammals because maternal fat loss can make a very substantial contribution to the daily and overall energy costs of lactation and, since human infants grow very slowly, human milk production is low compared to other mammals of a similar body size.

Twelve longitudinal studies of weight changes in lactation, in developed and developing countries, are reviewed in this section. These studies were conducted in The Gambia (Prentice et al, 1981), USA (Brewer et al, 1989; Butte et al, 1984; Dewey et al, 1993; Manning Dalton & Allen, 1983), Sweden (Sadurskis et al, 1988), UK (Goldberg et al, 1991), The Netherlands (van Raaij et al, 1991), Mexico (Allen et al, 1992), The Philippines (Guillermo-Tuazon et al, 1992), Egypt (Kirksey et al, 1992) and Kenya (Neumann et al, 1992).

Changes in body weight. Figure 21 illustrates the changes in body weight with no distinction made by the degree of lactation (i.e. full vs partial breastfeeders). The data are presented together since, with the exception of one study (Neumann et al, 1992), there is little to suggest that there are systematic differences between developed and developing countries. However, the Gambian data have been omitted due to the over-riding influence of the season of the year.

Table 9 Longitudinal estimates of postpartum body fat changes: skinfold thicknesses

Skinfold thickness (mm)


Stage post partum





Body fat
(kg or %)

Manning-Dalton et al (1983)

2 weeks





12 weeks





Butte et al (1984)

3 days






35 days






64 days






91 days






119 days






Brewer et al (1989)a

1-2 days BF





3 months





6 months





1-2 days FF





3 months





6 months





1-2 days CF





3 months





6 months





van Raaij et al (1991)b

5 weeks

19.3 kg

9 weeks


13 weeks


27 weeks


56 weeks


Guillermo-Tuazon et al (1992)b

6 weeks

12.4 kg

12 weeks


18 weeks


24 weeks


30 weeks


Dewey et al (1993)c

3 months BF

+ 1.3d

6 months


9 months

- 0.8d

12 months


3 months FF

+ 1.5d

6 months

+ 0.6d

9 months

- 0.2d

12 months

+ 0.2d

Allen et al (1992)

0-14 days





3 months





6 months





Kirksey et al (1992)

1 months



6 months



a BF = Exclusively breast feeding; FF = exclusively formula feeding; CF = combined breast and formula feeding.
b Calculated from S Triceps, biceps, subscapula and suprailiac skinfolds.
c BF = breast-fed for 12 months; FF = completely weaned by £ 3 months.
d Change in triceps (mm). Changes are calculated between measurement intervals.

Early weight losses in the studies by Butte et al (1984) and Brewer et al (1989) are substantially greater than in all the others, despite the corrections which were made for fluid losses. All the other data demonstrate that weight losses tend to occur throughout the first six months of lactation. In some instances the losses are very small and not different from those in matched groups of non-lactating post-partum women (Brewer et al, 1989; van Raaij et al, 1991). In the study by van Raaij et al, a small increase in body weight was observed. Dewey et al (1993) did observe significantly greater weight loss in full breastteeders compared to non-breastfeeders, especially to 3-6 months post partum. They argue that discrepancies between studies might be explained by a failure to adequately define true breastfeeding status.

Figure 19

Longitudinal changes in energy intake during lactation. Left-hand panel, Developed countries: Open circles, solid line, England (Whichelow, 1975); open squares, solid line, England (Black et al, 1986); closed circles, dotted line, The Netherlands (van Raaij et al, 1991); open triangles, solid line, England (Goldberg et al, 1991); open circles, dotted line, The Netherlands (van Raaij et al, 1991); closed triangles, solid line, Sweden (Sadurskis et al, 1988); closed circles, solid line. Australia (English et al, 1968); closed squares, dotted line, The Netherlands (Spaaij et al, 1994a); closed squares, solid line, Scotland (Schofield et al, 1987); open squares, dotted line, England (Schofield et al, 1987). Right-hand panel, Developing countries: Triangles, The Gambia (Prentice et al, 1981); circles, The Philippines (Guillermo-Tuazon et al, 1992); squares, Mexico (Allen et al. 1992).

Changes in body fat. Estimates of changes in body fat assessed by skinfold thicknesses are shown in Table 9. Assessments of changes by isotope dilution and whole body density are shown in Tables 10 and 11, respectively. Results are once again variable and certainly do not support the concept that fat loss is a biologically programmed part of normal lactation.

Figure 22 gives a clear indication of the reciprocal relationship that exists between a mother's incremental EI during lactation and her rate of fat loss. Mothers who have a large increase in EI tend not to lose weight. The least products regression (r = 0.550, p < 0.001) is illustrated and indicates that an energy increment of 2.70 MJ/day (645 kcal/day) is required for weight to remain constant. This is in excellent agreement with the computed costs and provides a reassuring cross validation. There seems to be no evidence that lactation is influenced by whether the energy to support it is derived from the diet or from fat stores (Prentice & Prentice, 1988).

Figure 20 Average changes in energy intake during lactation. References as for Figure 19

Figure 21 Longitudinal change in body weight in lactating women. Open circles, USA (Manning-Dalton et al, 1983); closed circles, The Gambia, dry season (Prentice et al, 1981); open squares, The Gambia, wet season (Prentice et al, 1981); closed squares, USA (Butte et al, 1984); open triangles, Sweden (Sadurskis et al, 1988); closed triangles, USA (Brewer et al, 1989); slashed squares, England (Goldberg et al, 1991); crosses, The Netherlands (van Raaij et al, 1991); (closed squares, The Philippines (Guillermo-Tuazon et al, 1992); crossed squares, USA (Dewey et al, 1993); asterisks, Mexico (Allen et al, 1992); closed diamonds, Kenya (Neumann et al, 1992); open diamonds, Egypt (Kirksey et al, 1992).

Table 10 Longitudinal estimates of changes in postpartum fat and lean body mass; isotope dilution


Baseline and stage post partum

Fat (kg)

Fat-free mass (kg)

Sadurskis et al (1988)a,b




5-10 days



2 months



6 months



Goldberg et al (1991)c

4 weeks



8 weeks



12 weeks



3 months post weaned



a Using 18oxygen.
b Calculated from total body water and total body potassium.
c Using deuterium and 18 oxygen.

Table 11 Longitudinal estimates of changes in post-partum body fat: whole-body density


Baseline and stage post partum

Body fat (kg)

Lean body mass (kg)

Butte et al (1984)

35 d



64 d



91 d



119 d



Figure 22 Inverse relationship between incremental energy intake during lactation and rate of post-partum weight loss. (Data from AA Paul and AM Prentice, unpublished).

We conclude that a number of recent studies have indicated that significant weight losses are not obligatory during lactation, especially in well-nourished or affluent women who can support the extra energy costs incurred by increasing EI and/or decreasing physical activity. In such women the extra weight and fat may be retained and there may be further increases during lactation eventually leading to the risk of obesity. However, on average women do tend to lose about 500 g/month and it seems reasonable to build this into the recommendations in a conditional manner. We believe that the conditional nature of this offset should be stressed more strongly than in the 1985 recommendations. Since marginally-nourished women may not have been able to build up extra stores during pregnancy we recommend that the full costs of lactation should be clearly stated as the primary recommendation.

Contents - Previous - Next