点击显示 收起
1 From the University of Melbourne, Department of Medicine (Royal Melbourne Hospital) Parkville, Australia (LMP, JLA, CM, MGF, BK, and JDW), and the School of Health Sciences, Deakin University, Burwood, Australia (CAN).
2 Supported by the National Health and Medical Research Council of Australia and the Victorian Health Promotion Foundation, Victoria, Australia. 3 Reprints not available. Address correspondence to JD Wark, The University of Melbourne, Department of Medicine (Royal Melbourne Hospital), Parkville, Victoria 3050, Australia. E-mail: jdwark{at}unimelb.edu.au.
ABSTRACT
Background: The long-term effects of pregnancy and lactation on measures of bone mineral in women remain unclear.
Objective: We studied whether pregnancy or lactation has deleterious long-term effects on bone mineral in healthy women.
Design: We measured bone mineral density (BMD; g/cm2) in women aged
Conclusion: We found no long-term detrimental effect of pregnancy or breast-feeding on bone mineral measures.
Key Words: Pregnancy • lactation • bone mineral density • twins • bone mineral content • women
INTRODUCTION
Both pregnancy and lactation place significant stress on maternal calcium homeostasis, potentially resulting in substantial changes in bone mineral. Whereas bone loss in the first 6 mo of lactation has been documented (1–7), with an estimated 4–6% loss, there is no clear consensus regarding the recovery of bone mineral from the effects of either pregnancy or lactation. If the lost bone is not completely restored, pregnancy or lactation may increase the risk of osteoporosis in later life. Previous studies have provided conflicting findings on the long-term effects of pregnancy and lactation on bone mineral density (BMD, g/cm2).
Parra-Cabrera et al (8) retrospectively reported that the number of pregnancies had a deleterious effect on BMD (in 313 women aged 26–83 y), particularly at the lumbar spine. Berehi et al (9) found no significant influence on BMD of the number of children (in 159 women aged 20–70 y). The potentially positive effects of parity were supported by Fox et al (10), who found a 1.4% increase in distal forearm bone density with each additional birth in a study of 2230 women aged 65 y. Tuppurainen et al (11), in a study of 1605 peri- and postmenopausal women, detected a protective effect of parity against low BMD with increased number of pregnancies. In a small prospective study, Black et al (12) found a significant decrease of 3.5% in spinal BMD over the 9 mo of pregnancy. In another prospective study, Sowers et al (13) found no detectable loss in proximal femur bone mineral after pregnancy.
Most studies investigating the effects of lactation have focused on the first 6–12 mo postpartum. Several reports suggest a full recovery from the effect of lactation, even when there was a subsequent pregnancy (14). Pollatti et al (6) found a significant decrease in bone mineral at 6 mo postpartum followed by an increase in BMD (1.1–1.9% compared with baseline) at 18 mo postpartum. Yasumizu et al (15) compared 13 women who breast-fed for < 3 mo with 14 women who breast-fed for 6 mo. The short-term breast-feeders recovered to puerperal levels after 9 mo, but the long-term breast-feeders did not. Sowers et al (2) found mean losses of 5.1% at the lumbar spine and of 4.8% at the femoral neck in 98 women who had given birth and lactated for a period of 6 mo compared with a matched group who did not breast-feed. Tuppurainen et al (11) found no correlation between breast-feeding and BMD.
The objective of the present study was to investigate whether pregnancy and lactation have long-term effects on BMD by studying female twins, including twin pairs discordant in their history of pregnancy and lactation. Our co-twin design matched for age, genetic factors, and some environmental factors, thereby enhancing the statistical power and reducing the risk of confounding.
SUBJECTS AND METHODS
Selection of subjects
All subjects in this analysis were identified retrospectively from the Bone Research Program in Twins database. Subjects were selected to allow 3 types of data analysis to be performed, as described below. (Note that the twin pairs included in studies 1 and 2 were drawn from the individual twins included in study 3.)
Study 1: Eighty-three twin pairs (21 monozygotic and 62 dizygotic) aged 18 y and discordant for ever having been pregnant beyond 20 wk were included in this analysis.
Study 2: Four hundred ninety-eight female twin pairs aged 18 y were incorporated in this analysis (including all 83 pairs from study 1).
Study 3: This investigation of 1354 individual females aged 18 y consisted of individual members of twin pairs, sister pairs, and their female relatives. Subjects were divided into 3 groups: 1) those having had no pregnancies > 20 wk, 2) those having had 1 or 2 pregnancies beyond 20 wk, and 3) those with 3 pregnancies beyond 20 wk.
Studies 1 and 2 used the powerful co-twin design to assess the associations of within-pair discordance in history of pregnancy or lactation with within-pair difference in measures of bone mineral. Bone mineral measures in nulliparous and parous twins were compared in study 1. The effect of increasing within-pair discordance for number of pregnancies was investigated in study 2. The cross-sectional analysis in study 3 allowed inclusion of the maximum number of available subjects and permitted a direct comparison with other cross-sectional studies reported in the literature.
Twins were recruited through the Australian Twin Registry. The study was approved by The Royal Melbourne Hospital Clinical Research and Ethics Committee and the Australian Twin Registry. All subjects gave written informed consent.
Bone mineral density and body composition
Bone mineral content (BMC) and areal BMD were measured by dual-energy X-ray absorptiometry with the use of the same Hologic QDR 1000W instrument (Hologic Inc, Waltham, MA). Measurements were made at the lumbar spine (L2–L4), total hip, and the total body. The CV was 0.5% for a spine phantom and 0.8% for a total-hip phantom. The BMD of each subject was measured within 2 mo of her twin. All measurements were performed 6 mo after pregnancy or weaning. Trained research assistants measured each subject’s weight and height. Fat mass and lean mass were determined by using Hologic software version 4.76. z Scores (the number of SDs above or below the age-matched mean bone mineral measure) were determined by using Hologic reference data.
Measurement of anthropometric and lifestyle factors
Height was measured to the nearest 0.1 cm on a wall-mounted stadiometer, and weight was measured on a balance scale to the nearest 0.1 kg. Medical and reproductive history was assessed by self-administered questionnaires (16) that were reviewed subsequently by trained research assistants. Information on the number of pregnancies beyond 20 wk, the number of children, and the number of months spent breast-feeding was assessed. Current dietary calcium intake was determined from a food-frequency questionnaire (16) and did not include calcium from supplements. Use of calcium supplements was obtained by questionnaire. A regular smoker was defined as a woman having smoked on average 7 cigarettes/wk for 1 y, and cigarette use was expressed as pack-years (average number of cigarettes smoked per day divided by 20 and multiplied by the number of years of use). Body mass index (BMI) was calculated as weight (kg) divided by height2 (m2).
Statistical analysis
All analyses were performed by using the statistical package SPSS version 11 (SPSS Inc, Chicago). Differences were assessed by t test or analysis of variance with a Tukey correction when > 2 groups were compared. Assessing the differences in one variable within a twin pair and comparing these to within-twin pair differences in another variable can provide evidence of an association (16). Bone density results were expressed as both the absolute within-pair difference and as the average within-pair percentage difference with the SEM. The within-pair percentage difference was calculated as the value for the parous twin minus that for the nulliparous twin (study 1) and as the value for the twin pair member with the greater number of pregnancies minus that for the twin pair member with the lesser number of pregnancies (study 2), both expressed as a percentage of the mean value of the pair for any given variable. Associations between variables were determined by linear regression. Linear regression residuals, ß coefficients, and regression constants were used to calculate the adjusted mean values for any given variable. For example, lumbar spine BMD adjusted for age was calculated as follows: Adjusted lumbar spine BMD = (ß coefficient for age) x (mean age of the population) + (residual for lumbar spine BMD) + lean regression constant. Both unadjusted and adjusted data are presented when appropriate because we considered it important to show the effects of the adjustments on the data and that the analysis be as transparent as possible.
RESULTS
Study 1: within-pair discordance for parity and lactation
Eighty-three female twin pairs (21 monozygotic and 62 dizygotic) discordant for ever having been pregnant were studied (Table 1). The mean (± SD) age of the cohort was 42.2 ± 15.7 y. The parous twins had a mean (± SD) of 2.3 ± 0.13 pregnancies beyond 20 wk and those who breast-fed did so for 8.4 ± 1.67 mo per child. There were no significant differences between the nulliparous and parous groups in height, weight, BMI, total-hip BMD, lumbar spine BMD, total-body BMC, lean mass, and fat mass or in total-hip BMD, lumbar spine BMD, and total-body BMC adjusted for age, height, and fat mass. Current dietary calcium intake, use of calcium supplements, and smoking history were also not significantly different between the groups (Table 1).
View this table:
TABLE 1 . Study 1: characteristics of the nulliparous and parous twin pairs discordant for ever having been pregnant beyond 20 wk1
There was no within-pair percentage difference in total-hip BMD, lumbar spine BMD, and total-body BMC (Figure 1) or in height, weight, and BMI. Adjustment for age, height, and fat mass did not alter these results. Twin pairs were grouped according to the degree of discordance in the number of pregnancies within the pair into those different by 1, 2, or 3 pregnancies. There were no significant within-pair or between-group differences in measures of total-hip BMD, lumbar spine BMD, and total-body BMC for any group (Figure 1). No association was evident between age and the within-pair difference in total-hip BMD, lumbar spine BMD, and total-body BMC (Figure 1).
FIGURE 1. . Study 1. Within-pair percentage difference in total-hip bone mineral density (A), lumbar spine bone mineral density (B), and total-body bone mineral content (C) according to the degree of discordance for pregnancy: discordant for 1 pregnancy (), discordant for 2 pregnancies (+), and discordant for 3 pregnancies ().
Lactation
Of the parous women, 71% had breast-fed for > 1 mo. There were no detectable within-pair differences in lumbar spine and total-hip BMD, total-body BMC [unadjusted and adjusted for age, height (m), and fat mass (g)], lean mass, or fat mass for twin pairs in whom the parous twin breast-fed (n = 58 pairs) or for those pairs in whom the parous twin did not breast-feed (n = 25 pairs). The adjusted mean (± SEM) bone mineral measures for the parous breast-feeders compared with the nulliparous women were as follows: lumbar spine, 1.09 ± 0.02 and 1.07 ± 0.03 g/cm2; total hip, 0.95 ± 0.02 and 0.96 ± 0.03 g/cm2; and total-body BMC, 2.27 ± 0.03 and 2.29 ± 0.04 kg, respectively, all P > 0.05. The adjusted mean values for the parous non-breast-feeders compared with the nulliparous women were as follows: lumbar spine, 1.07 ± 0.03 and 1.02 ± 0.02 g/cm2; total hip, 0.96 ± 0.03 and 0.97 ± 0.03 g/cm2; and total-body, 2.24 ± 0.07 and 2.16 ± 0.06 kg, respectively, all P > 0.05.
Study 2: within-pair difference in the number of pregnancies
Four-hundred ninety-eight pairs of female twins aged 42.2 y (
View this table:
TABLE 2 . Study 2: cross-sectional comparison of the group means for women with no pregnancies (group 1), 1–2 pregnancies (group 2), or 3 pregnancies (group 3)1
There was a significant difference in the average unadjusted within-pair percentage difference in total-body BMC for group 1 (-0.4%) compared with the average within-pair percentage difference for group 3 (5.1%; P = 0.03). There was no significant difference in total-body BMC adjusted for age, height, and fat mass. There was no significant difference in lumbar spine or total-hip BMD, total fat mass, or lean mass between groups with or without adjustment for age, height, and fat mass (Figure 2). The above analysis was repeated by using the absolute difference in the number of pregnancies within a pair. Pairs had differences of 0–7 pregnancies. There was no significant within-pair difference for lumbar spine BMD, total-hip BMD, total-body BMC, lean mass, fat mass, or adjusted measures of lumbar spine BMD, total-hip BMD, or total-body BMC, regardless of the within-pair difference in the number of pregnancies.
To investigate the effect of pregnancy independently of the potential effect of breast-feeding, we compared pairs with a within-pair difference in the number of pregnancies. We compared separately pairs concordant for breast-feeding (n = 131) and pairs concordant for non-breast-feeding (n = 7). There was no within-pair difference in unadjusted or adjusted total-hip BMD, lumbar spine BMD, total-body BMC, or body composition.
Study 3: cross-sectional analysis
A total of 1354 individual women aged > 18 y were included in study 3. The group consisted of individual twins, their female siblings, and other female family members. Subjects were divided into groups on the basis of the number of pregnancies beyond 20 wk duration. Group 1 consisted of women with no pregnancies beyond 20 wk (n = 426), group 2 consisted of women with 1–2 pregnancies (n = 455), and group 3 consisted of those with 3 pregnancies (n = 473). The women with no pregnancies (group 1) were significantly younger (by 12.0 compared with 16.7 y; P < 0.05), had a lower BMI (by 6.9% compared with 9.1%; P < 0.05), and were taller (by 0.7 compared with 1.8 cm; P < 0.05) than the women in groups 2 and 3, respectively (Table 3). The women in group 2 were significantly taller than the women in group 3 (by 1.1 cm; P < 0.05).
View this table:
TABLE 3 . Study 3: cross-sectional comparison of the group means for women with no pregnancies (group 1), 1–2 pregnancies (group 2), and 3 pregnancies (group 3)
Because of the significant differences in age and body composition between the 3 groups, measurements of bone mass were adjusted for age (y), height (m), and fat mass (g) (Table 4). Women in groups 2 and 3 had higher adjusted lumbar spine BMD (2.9% compared with 3.8%; both P = 0.001) and adjusted total-body BMC (2.7% compared with 3.1%; both P < 0.001) than did those in group 1. Adjusted total-hip BMD was greater in group 3 than in group 1 (2.1%; P = 0.02). There was no significant difference between groups 2 and 3 in body composition or in adjusted and unadjusted measures of regional BMD.
View this table:
TABLE 4 . Study 3: comparison of the group means for women with no pregnancies (group 1), 1–2 pregnancies (group 2), and 3 pregnancies (group 3) for measures of bone mass adjusted for age (y), height (m), and fat mass (g)1
z Scores were calculated by using the Hologic normal reference range. There were significantly higher lumbar spine BMD (L2–L4) z scores in groups 2 and 3 than in group 1 (Table 3). There was no significant difference in the total-hip z score between the pregnancy groups.
There was no significant difference in the number of lifetime fractures (not taking into account the level of trauma associated with the fracture). Group 1 subjects reported 143 fractures; group 2, 150 fractures; and group 3, 161 fractures.
In separate analyses for study 3, we compared the women in groups 2 and 3, separating breast-feeders and non-breast-feeders. There was no significant difference in lumbar spine BMD, total-hip BMD, total-body BMC, lean mass, or fat mass or in adjusted lumbar spine BMD, total-hip BMD, and total-body BMC between parous breast-feeders in groups 2 and 3 or between parous non-breast-feeders in groups 2 and 3.
Associations with lactation
Of the 928 parous women, 82% breast-fed for > 1 mo. There was no significant difference in age, weight, BMI, or current dietary calcium intake between those parous women who breast-fed and those who did not. The parous women who breast-fed were taller than those who did not (1.62 ± 0.002 compared with 1.60 ± 0.02 m, respectively; P = 0.03). Of the parous women who breast-fed, 37% had smoked, 20% had used hormone replacement therapy, and 74% had used the oral contraceptive pill. Of the parous women who did not breast-feed, 46% smoked, 18% had used hormone replacement therapy, and 84% had used the oral contraceptive pill. Neither the average time spent breast-feeding per child nor the number of women who breast-fed was significantly different between groups 2 and 3 (Table 3). Total fat mass was lower in those parous women who breast-fed than in those who did not (20.5 ± 0.34 compared with 23.02 ± 1.1 kg; P = 0.01; Figure 3). Adjusted total-body BMC and total-hip BMD were significantly higher in those who breast-fed than in those who did not [2.31 ± 0.01 compared with 2.24 ± 0.02 kg (P = 0.005) and 0.97 ± 0.004 compared with 0.94 ± 0.01 g/cm2 (P = 0.04), respectively; Figure 3]. There was no significant difference in adjusted lumbar spine BMD.
FIGURE 3. . Study 3. Comparison of mean (± SEM) fat mass, adjusted total-body bone mineral content (TB BMC), and adjusted total-hip bone mineral density (TH BMD) in parous women who had ever breast-fed and those who had never breast-fed. TB BMC and TH BMD were adjusted for age, height, and fat mass.
Calcium supplementation during pregnancy and breast-feeding
Of the 928 parous women, 31% took calcium supplements during pregnancy, 7% were unsure, and 6% gave no response. On average, women took calcium supplements for 80% of their pregnancies. Of the 817 women who breast-fed, 12% took a calcium supplement while breast-feeding, 75% did not while breast-feeding, 11% were unsure, and 2% provided no response.
DISCUSSION
As shown in study 1, there was no detectable long-term effect of either pregnancy or breast-feeding on regional areal BMD or total-body BMC in twin pairs discordant for ever having been pregnant. As shown in study 2, there were no within-pair differences in bone mass, either unadjusted or adjusted for age, height, and fat mass. After grouping by the discordance in the number of pregnancies, total-body BMC was significantly higher in those women who had 3 pregnancies more than their twins. After adjustment for age, height, and fat mass, however, there was no significant difference. Cross-sectionally (study 3), after adjustment for age, lean mass, and fat mass, total-hip BMD and lumbar spine BMD were greater in those who had had either 1 or 2 pregnancies or 3 pregnancies than in nulliparous women. There was no significant difference between groups 2 and 3 in body composition or in adjusted and unadjusted measures of regional BMD. Total-body BMC was greater in those women with 3 pregnancies. Total body fat was lower in those parous women who breast-fed than in those parous women who had not breast-fed. After adjustment for age, lean mass, and fat mass, those parous women who had breast-fed had a higher total-body BMC and total-hip BMD than those who did not.
An important consideration with respect to the long-term effect of pregnancy and lactation on bone is the potential influence of pregnancy and lactation on the body composition of the mother. There are many anecdotal reports of increasing body weight with increasing number of pregnancies. Higher body weight is associated with higher bone density. Lean mass has been found to be a major predictor of bone mass in adolescents (16), and fat mass becomes an important predictor later in life (17). However, we could find no within-pair evidence that pregnancy history was associated with increased body weight. A review of studies that examined weight changes related to pregnancy or parity indicated that the average weight increments with pregnancy are generally < 1.5 kg during a single reproductive cycle (before pregnancy to 1 y postpartum), and that only a small number of women greatly increase their weight during the reproductive cycle (18). Cross-sectionally, after adjustment for age and current body composition, bone density was higher in those who had been pregnant at least once, indicating that pregnancy may have a positive effect on bone density independent of body composition. Alternatively, underlying hormonal factors in nulliparous women may adversely affect bone. The findings in study 1, however, argue against this interpretation of the results of study 3.
Breast-feeding may also confer some benefits on bone density. Cross-sectionally, when those parous women who had breast-fed were compared with those who had not, those who had breast-fed had a higher adjusted total-body BMC and a lower fat mass. Although these findings are interesting, they were not supported by the within-twin pair analysis. Note that there was no evidence within twin pairs for a detrimental effect of either pregnancy or breast-feeding on bone density. Generally, however, the mean calcium intake of this group was high and 31% reported taking calcium supplements in pregnancy and 12% while breast-feeding.
Few substantive prospective studies have looked at the effect of pregnancy on bone mass. In a small study of 10 women aged 23–40 y, Black et al (12) found a significant decline in lumbar spine BMD (3.5%), total-hip BMD, and femoral neck BMD over the 9 mo of pregnancy. Forearm BMD measured at the midradius, one-third distal, and ultradistal radius also decreased, but not significantly. The findings suggested that the skeleton may be in negative calcium balance during pregnancy. Although a sound study design was used, the sample size was small and so the conclusions must be tentative. Sowers et al (13) also measured maternal bone mineral before and after pregnancy. They found no detectable loss at the proximal femur. These authors postulated that "compensatory calcium may have been made available by increasing the absorptive efficiency for calcium from the gut." Laskey and Prentice (19) found no significant difference in bone mineral status in 12 women measured before conception, immediately postpartum, and after a subsequent pregnancy.
Our results also agree with the findings of Berehi et al (9). The population studied by Berehi et al comprised Omani women with a high average number of children (range: 0–14, with an average of 5.1), thus giving greater power to find a correlation between BMD and the number of pregnancies. They found no significant relation. This is in contrast with the findings of Parra-Cabrera et al (8), who found that increasing number of pregnancies in a population of Hispanic women was associated with a deleterious effect on BMD. However, a large retrospective study of 2230 women aged 65 y by Fox et al (10) showed that each additional birth conferred a 1.4% increase in distal radial bone density. These investigators used single-photon absorptiometry and measured only radial density; this is a site of high cortical bone content, which gives limited information relevant to BMD in the spine and proximal femur.
The lack of a discernible adverse effect of breast-feeding on BMD agrees with some, but not all, of the existing literature. Whereas Pollatti et al (6) found a significant decrease in bone mineral in the 6-mo postpartum period and an increase in lumbar spine and radial BMD (1.1–1.9% compared with baseline) at 18 mo postpartum, they found no significant difference between the group who breast-fed for 6 mo and a control group who did not breast-feed. Sowers et al (14) found that women recovered fully from the effects of breast-feeding even when there was a subsequent pregnancy. In contrast, Yasumizu et al (15) compared 13 women who breast-fed for < 3 mo with 14 women who breast-fed for 6 mo. The short-term breast-feeders recovered to puerperal levels after 9 mo, but lumbar spine bone density in the long-term breast-feeders was significantly lower than puerperal levels. Jones and Scott (20) found that smokers who breast-fed at least one child had an additional deficit in bone mass. However, the relation between weight and BMD was not taken fully into account, nor was smoking quantified.
It seems pertinent that to assess bone recovery after pregnancy and lactation, we should wait for the resumption of ovarian function. The return of menses varies by individual and is influenced greatly by the length of breast-feeding. Sowers et al (14) found that the return of ovarian function (as indicated by a subsequent pregnancy) correlated with recovery of bone mineral. Laskey and Prentice (19) found that the changes that occur in BMD were reversible and "do not persist after a subsequent pregnancy, even when conception occurs during lactation at a time when bone loss is still evident." Pollatti et al (6) found that the earlier menses were resumed, the less bone was lost. In contrast, those women who experienced a longer delay in the return of menses experienced a greater gain in bone to prepregnancy levels. Yasumizu et al (15) found that the BMD lost during pregnancy was restored 4–5 mo after weaning of the infant. We did not have detailed information regarding the time since pregnancy nor the time between births. However, the lack of these data does not detract from our conclusion that there were no long-term deleterious effects of pregnancy on bone mineral measures. If postpartum recovery of bone mineral was incomplete in an individual at the time of measurement, any adverse long-term effect of pregnancy would be overestimated, not underestimated.
A major strength of the present study design is the use of a twin population. The co-twin study design enhances statistical power and intrinsically adjusts for age, for some or all genetic variation, and for several environmental factors. This is especially true in early life when much of the twins’ environmental exposures were shared. Twins allow for the discrimination of genetic and environmental determinants of traits, and we can select for discordant factors (such as pregnancy and lactation) to study their effects.
Care needs to be taken in interpreting these findings, however. Our study population was well-nourished with respect to both energy and calcium needs. Because they were volunteers, there is a potential bias that this population may be more health-conscious than the general population. The study was also retrospective, with reproduction and breast-feeding information being collected over several years, and in some cases decades after the pregnancies had taken place. Because of the limitations of the cross-sectional analysis, when an age difference existed, age was taken into consideration. Consequently, data were adjusted for known confounders of bone mass, in this case age, height and fat mass. Results of cross-sectional data analysis with these adjustments were consistent with the finding of the within-pair modeling of the data, suggesting that we can draw reasonable conclusions from both the cross-sectional and the paired analyses. We do not consider that there was any intrinsic bias toward a null finding in our analyses.
In this analysis of a large population of healthy Australian women, no consistent, long-term detrimental effects of pregnancy or breast-feeding on bone mineral mass were observed. After adjustment for body size, there was some evidence from the cross-sectional analysis that pregnancy may increase total bone mineral mass, although no within-pair difference in bone was observed in twin pairs discordant for ever having been pregnant. Breast-feeding may lead to lower total body fat. Pregnancy and lactation appeared to have little residual effect on the skeleton.
FIGURE 2. . Study 2. Mean (± SEM) within-pair percentage difference in adjusted total-hip bone mineral density (A), adjusted lumbar spine bone mineral density (B), and adjusted total-body bone mineral content (C) by the within-pair difference in the absolute number of pregnancies: 1, no difference; 2, different by 1 or 2; and 3, different by 3. Bone mineral density was adjusted for age (y), height (m), and fat mass (g).
ACKNOWLEDGMENTS
JDW designed the study, with contributions from CAN, JLA, and LMP. JLA, CM, MGF, and LMP collected the data. BK assessed bone density. LMP performed the statistical analysis. LMP and JLA wrote the manuscript, with contributions from JDW and CAN. No author had any financial or personal interest in any company or organization involved with sponsorship or funding of this work
REFERENCES