Skip to main content
Download PDF

Development and internal validation of a non-invasive clinical tool to predict sufficient omega-3 levels in early pregnancy

BMC Pregnancy and Childbirth volume 23, Article number: 442 (2023) Cite this article

Abstract

Background

Complications from preterm birth (PTB) are the leading cause of death and disability in those under five years. Whilst the role of omega-3 (n-3) supplementation in reducing PTB is well-established, growing evidence suggests supplementation use in those replete may increase the risk of early PTB. Aim: To develop a non-invasive tool to identify individuals with total n-3 serum levels above 4.3% of total fatty acids in early pregnancy.

Methods

We conducted a prospective observational study recruiting 331 participants from three clinical sites in Newcastle, Australia. Eligible participants (n = 307) had a singleton pregnancy between 8 and 20 weeks’ gestation at recruitment. Data on factors associated with n-3 serum levels were collected using an electronic questionnaire; these included estimated intake of n-3 (including food type, portion size, frequency of consumption), n-3 supplementation, and sociodemographic factors. The optimal cut-point of estimated n-3 intake that predicted mothers with total serum n-3 levels likely above 4.3% was developed using multivariate logistic regression, adjusting for maternal age, body mass index, socioeconomic status, and n-3 supplementation use. Total serum n-3 levels above 4.3% was selected as previous research has demonstrated that mothers with these levels are at increased risk of early PTB if they take additional n-3 supplementation during pregnancy. Models were evaluated using various performance metrics including sensitivity, specificity, area under receiver operator characteristic (AUROC) curve, true positive rate (TPR) at 10% false positive rate (FPR), Youden Index, Closest to (0,1) Criteria, Concordance Probability, and Index of Union. Internal validation was performed using 1000-bootstraps to generate 95% confidence intervals for performance metrics generated.

Results

Of 307 eligible participants included for analysis, 58.6% had total n-3 serum levels above 4.3%. The optimal model had a moderate discriminative ability (AUROC 0.744, 95% CI 0.742–0.746) with 84.7% sensitivity, 54.7% specificity and 37.6% TPR at 10% FPR.

Conclusions

Our non-invasive tool was a moderate predictor of pregnant women with total serum n-3 levels above 4.3%; however, its performance is not yet adequate for clinical use.

Trial registration

This trial was approved by the Hunter New England Human Research Ethics Committee of the Hunter New England Local Health District (Reference 2020/ETH00498 on 07/05/2020 and 2020/ETH02881 on 08/12/2020).

Peer Review reports

Background

Preterm birth (PTB), defined as delivery before 37 completed weeks’ gestation, is the leading cause of death and disability in those under five years of age [1, 2]. Globally, PTB is estimated to affect 10.6% of births, equating to nearly fifteen million babies annually [2], and this rate is rising [2, 3]. The acute complications of PTB coupled with potential ongoing neurodevelopmental impairments have significant socioeconomic costs for families and health care systems [3]. Of babies born preterm, 20–30% are defined as early PTBs (i.e., delivery before 34 completed weeks’ gestation). This subgroup has the highest rates of morbidity and mortality [2, 4].

For most pregnancies, PTB is an idiopathic disorder with multifactorial causes, posing challenges for PTB prevention [5]. Current interventions with an established evidence base for most pregnancies include midwifery-led continuity care and screening for lower genital tract infections, with suggestive evidence for pharmacological smoking cessation to reduce rate of PTB [6].

The role of omega-3 (n-3) in the prevention of PTB has been studied for more than three decades in hundreds of thousands of women [7,8,9]. Omega-3s are polyunsaturated fatty acids (PUFAs) commonly obtained from marine, animal, and plant-based sources. After consumption, these essential fatty acids can be metabolised into more biologically potent lipid mediators known as long chain PUFAs (LC-PUFAs) [10]. The parent n-3 PUFA, α-linolenic acid (ALA) can be metabolised into eicosapentaenoic acid (EPA), docosapentaenoic acid (DPA) and docosahexaenoic acid (DHA) [10, 11] via alternating steps of desaturation and elongation by Δ6-desaturase and Δ5-desaturase, and elongase respectively [10, 12]. Whilst the exact mechanism(s) of action remains unclear, it has been proposed that increased dietary n-3 LCPUFAs may increase the production of n-3 prostaglandin derivatives which have less potent pro-inflammatory action compared to n-6 prostaglandin derivatives [13]. This interferes with the production of n-6 prostaglandin (E2 and F2α) which may play a role in premature uterine contractions and cervical ripening [14]. Further, possible anti-arrhythmic effects of n-3 on cardiac activity may translate to the myometrium, reducing electrical and contractile activity, ultimately delaying the onset of labour [15]. Overall, an individual’s n-3 level is associated with their dietary and supplementary intake of n-3 [16, 17] but may also be influenced by a number of non-dietary factors including age, body mass index (BMI), socioeconomic status (SES), smoking status, alcohol use, and presence of genetic variants involved in metabolism of PUFAs [16, 18, 19].

Studies evaluating the role of n-3 PUFA interventions on PTB rates have demonstrated mixed results. Taken together, the majority of studies suggest that n-3 supplementation may be valuable in PTB risk reduction [7,8,9], particularly in those with low levels of n-3 [20, 21]. Epidemiological evaluation of multiple large cohort studies has reported associations between increased n-3 intake and decreased PTB risk [22, 23]. Further, associations between decreased n-3 intake [24] or serum levels [25], and an increased risk of PTB have also been reported. Recent meta-analyses of more than 36 randomised controlled trials (RCTs) comparing n-3 supplementation versus controls or placebo have reported a 12% reduction in PTB (risk ratio (RR), 0.88; 95% CI (confidence intervals) 0.81–0.95). Regarding early PTB, meta-analyses have reported a 35% risk reduction (RR, 0.65; 95% CI 0.46–0.92) [9].

It has been proposed that the mixed results from the RCTs evaluating n-3 supplementation to prevent PTB may be due to variations in baseline n-3 levels in the populations studied [20, 26]. A recent cross-sectional analysis of n-3 intake in 184 countries demonstrated that increasing n-3 intake by one standard deviation in countries with low baseline n-3 intake (< 600 mg/day) was associated with a reduction in PTB [27]. In contrast, a similar increase in n-3 intake in countries with high baseline n-3 intake had no effect on PTB rate [27]. The results from the Omega-3 to Reduce the Incidence of Preterm Birth (ORIP) trial have taken this observation further; it reported that n-3 supplementation in those with low baseline levels of n-3 was associated with a 77% reduction in early PTB (RR, 0.23; 95%CI 0.07–0.79), whereas supplementation in those with high baseline levels of n-3 was associated with a more than two-fold increased risk of early PTB (RR, 2.27; 95%CI 1.13–4.58) [20, 28]. Taken together, these data highlight the potential importance of measuring serum n-3 levels prior to commencement of n-3 supplementation to reduce the rate of early PTB. Blood sampling for fatty acid analysis is an invasive procedure associated with pain, risk of infection and rarely, haemorrhage. This is currently the only way to accurately estimate baseline levels of n-3 at the start of pregnancy. It, however, is not part of routine antenatal screening and hence involves an expensive out-of-pocket cost, potentially precluding access for many patients for an important assessment. To our knowledge, no non-invasive tools have been developed to predict total n-3 serum levels above specific thresholds in early pregnancy. This resource gap highlights the need for the development of a non-invasive tool to determine baseline levels of n-3 early in pregnancy.

Methods

The aim of this study was to develop and internally validate a non-invasive clinical tool to predict sufficient serum total n-3 levels (> 4.3% of total fatty acids) in early pregnancy. This study followed the Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis (TRIPOD) checklist (See Additional file 1).

Ethics

This observational study was approved by the Hunter New England Human Research Ethics Committee of the Hunter New England Local Health District (Reference 2020/ETH00498 and 2020/ETH02881). All participants provided informed written consent.

Participants

A total of 331 participants were recruited in this prospective observational study from three sites (two pathology collection centres and one antenatal clinic) in Newcastle, Australia between March 2021, and March 2022. Participants were eligible if they had a singleton pregnancy between eight- and 20-weeks’ gestation at the time of recruitment. Pregnancy dating was calculated using a reliable menstrual period if available or confirmed by a first trimester ultrasound before 14 weeks gestation. If the two parameters differed by more than five days, dates from the ultrasound measurements were used. Participants were excluded if they had a multi-fetal pregnancy or did not provide complete dietary data or a blood sample.

Data sources

For participants recruited at pathology collection centres, a blood sample was collected at recruitment. For participants recruited at the antenatal clinic, a blood sample was collected at their first trimester appointment. Participants ideally completed the questionnaire at the time of blood collection or where this was not possible due to time constraints, as soon as possible after blood collection.

Estimated n-3 intake was assessed using an electronic questionnaire based on the previously validated Australian Clinical Omega-3 Dietary Survey (CODS), a food frequency questionnaire used to assess intake of foods high in n-3 LFCPUFAs with consideration to serving size and frequency of consumption over the previous three months [29]. Servings sizes were estimated using a photo aid and/or a stated standardised portion size (i.e., one portion is approximately the size of a deck of cards). The frequency of consumption was quantified using open ended frequency categories (i.e., how many times per month, week, or day). Data on type, amount and frequency of n-3 supplementation use was also collected using this electronic questionnaire. Estimated weekly intake of n-3 PUFAs was quantified by conversion of questionnaire data using the Australian Food and Nutrient Database (AUSNUT 2011–2013) [30]. Estimated intake of n-3 PUFAs included ALA, EPA, DPA, DHA, total n-3 LCPUFAs (sum of EPA, DPA and DHA) and total n-3 PUFAs (sum of ALA, EPA, DPA and DHA) presented as milligrams per week (mg/week).

Maternal self-reported demographic data (including age, weight, height, residential postcode, highest level of completed education, multivitamin or supplementation use, smoking status, and alcohol intake) were collected using the electronic questionnaire. Body mass index was calculated by dividing the maternal weight (in kilograms) by height (in metres) squared. The maternal residential postcode was used to derive an Index of Relative Socioeconomic Advantage and Disadvantage (IRSAD). The index, ranged from 1 to 10 with a higher score indicating a relative lack of disadvantage and greater advantage, was a proxy for socioeconomic status.

Measured n-3 levels were evaluated from a single blood sample via venepuncture. Five millilitres (mL) of venous blood were collected in a serum separating gel tube. Samples were placed on ice for a maximum of four hours before being centrifuged at 3800 g for ten minutes at four degrees Celsius (°C). Serum samples were extracted, de-identified and stored at -80 °C until analysis. Fatty acid analysis was performed using gas chromatography at the South Australian Health and Medical Research Institute (SAHMRI) using methods published elsewhere [31].

Outcomes

The primary outcome of this study was total serum n-3 PUFA levels above 4.3% of total fatty acids. This level was selected based on the dried blood spot data converted to a serum n-3 level using data supplied by SAHMRI [32].

Statistical analysis

Model development

A full model approach was used for development of our prediction model. Factors that may influence serum n-3 levels were identified from the literature and were considered as candidate predictors; these included dietary and supplementary n-3 intake, maternal age, BMI, SES, smoking status, and alcohol use [16, 18, 19].

A proportion of ALA may be metabolised into n-3 LCPUFAs including EPA, DHA and DPA; however, the specific rate of conversion has not been well-established in the literature. Our prediction models accounted for conversions between 0 and 25% at 2.5% increments based on the evidence available [33]. All analyses performed were based on complete case analysis for dietary data. The association between predictors and the primary outcome was analysed by univariate regression models. Multivariate logistic regression, adjusting for variables including maternal age, BMI, IRSAD, n-3 supplementation use was used to develop multiple prediction models for measured n-3 levels. Smoking status and alcohol use were not included as there were too few events for these predictors for analysis.

Model internal validation

Internal validation of models was performed using 1000 bootstraps to generate 95% confidence intervals for performance metrics reported.

Model performance

Model performance was assessed by reporting sensitivity, specificity, accuracy, area under the receiver operating characteristic (AUROC) curve, maximum (Emax) and average (Eavg) difference in predicted versus less-calibrated probabilities, net reclassification improvement (NRI), Brier score, and scaled Brier score. Net benefit (NB) at threshold 50%, true positive rate (TPR) at 10% false positive rate (FPR) and TPR at 10% false negative rate (FNR), positive and negative predictive values (PPV and NPV) were evaluated to contextualise the model’s clinical value. Additionally, metrics that examined the model’s balance between specificity and sensitivity were also assessed. These metrics included Youden’s Index, Concordance Probability, Closest to (0,1) Corner and Index of Union [34].

Power analysis

Based on the ORIP data, we assumed the incidence of total serum n-3 levels above 4.3% would be 55% of the study population [20, 28] which gives a maximum R2CS (max Cox-Snell R squared statistic), a pseudo measure of variance explained for models developed using logistic regression, of 0.75. Parameters for sample size calculation included adjusting for five predictors and using a conservative signal to noise ratio to give an adjR2CS = 0.2*maxR2CS [35, 36]). A subsequent step was also applied ensure the estimated sample size would not result in a model that would exceed the pre-specified shrinkage factor (s) of 0.90 and to ensure a small absolute difference of less than 0.05 between the developed model’s maxR2CS and adjR2CS. A minimum of 276 women will be required to minimise variability in the prediction model to limit shrinkage to less than 10%.

All analyses were performed in R (v4.1.0) and its associate libraries [37].

Results

Participants

A total of 331 participants were recruited from three clinical sites in Newcastle. After withdrawal of consent, and exclusion due to incomplete dietary data, outcome data was available for 307 participants (92.7%) (Fig. 1).

Fig. 1
figure 1

Participant flow through study

Full size image

Baseline maternal characteristics of the total cohort are as summarised in Table 1. The median maternal age in study participants was 32.5 years (interquartile range [IQR] 29.5–35.0). Most participants were Caucasian (90.7%) and the median BMI was 24.4 kg/m2 (IQR 22.2–29.0). The median gestational age at sample collection was 14.1weeks (IQR 11.8–15.3). Most study participants consumed pregnancy multivitamin supplements (n = 271, 88.3%) but only a small proportion took separate n-3 supplements (n = 18, 5.9%). Supplementation profiles of study participants are summarised in Table 2.

Table 1 Maternal baseline characteristics. All data are shown as median (IQR) or as number/total number (%)
Full size table
Table 2 Supplementation status in total cohort (n = 307). All data are shown as number (%)
Full size table

Based on the modified CODS questionnaire, the median estimated weekly intake of total n-3 PUFAs was 4648.0 mg (IQR 6443.0). The median estimated weekly intake of ALA, and total combined DHA and EPA was 2487.0 mg and 1431.0 mg respectively. These intakes were significantly below the Australian National Heart Foundation dietary guidelines which recommend a daily intake of 1000 mg of ALA and 250-500 mg of DHA and/or EPA. This equates to a weekly intake of 7000 mg of ALA and 1750-3500 mg of DHA and/or EPA. Table 3 summarises the estimated weekly intake of n-3 PUFAs. The measured serum levels of n-3 PUFAs are also presented in Table 3.

Table 3 Estimated and measured levels of n-3 PUFAs. All data are shown as median (IQR).
Full size table

Outcome

Of 307 participants included in the model development, 180 (58.6%) had a total n-3 serum level above 4.3% of total fatty acids.

Model development

A total of 1370 models were examined. Predictors in the optimal model included dietary intake of total n-3 LCPUFAs (with no conversion of ALA to n-3 LCPUFAs), maternal age, BMI, IRSAD, and n-3 supplementation. The univariate analyses between predictors in the optimal model and outcome are presented in Table 4. Regression coefficients for predictors in the optimal prediction model are presented in Table 5.

Table 4 Predictors included in the optimal prediction model
Full size table
Table 5 Univariate analyses between predictors and outcome
Full size table

Model performance

To optimise clinical relevance of identifying pregnant women with total serum n-3 levels above 4.3%, the top five models were selected by maximisation of sensitivity. The top five models are presented in Tables 6, 7, and 8. From these top five models, the optimal model was selected with consideration to the Youden’s Index, Concordance Probability, Closest to (0,1) Corner and Index of Union (Table 8). Further, the assessment of metrics in additional domains including overall performance (Brier Score, Scaled Brier Score in Table 6), calibration (Emax, Eavg in Table 6), NRI (Table 6) and NB (Table 7) demonstrated consistency in the predictive performance of optimal model selected for this study.

Table 6 Overall performance measures (95% confidence interval) for top five models
Full size table
Table 7 Performance measures (95% confidence interval) assessing clinical utility
Full size table
Table 8 Performance metrics (95% confidence interval) balancing sensitivity and specificity
Full size table

An estimated n-3 intake above 767.8 mg/week with adjustment for maternal sociodemographic factors was predicted to give a total serum n-3 level above 4.3%. The optimal prediction model had an AUROC of 0.744 (95% CI 0.742–0.746), a sensitivity of 84.7%, a specificity of 54.7%, TPR at 10% FPR of 37.6%, and TNR at 10% FNR of 36.0%. These data are presented graphically in Fig. 2. Distribution data for the measured n-3 levels for those predicted to be above and below 4.3% of total fatty acids are presented in Fig. 3. In those predicted to have measured n-3 levels greater than 4.3%, 72.5% had measured n-3 levels greater than 4.3% (i.e., the PPV was 72.5%). Conversely, in those predicted to have measured levels less than or equal to 4.3%, 71.9% had concordant measured n-3 levels (i.e., NPV was 71.9%).

Fig. 2
figure 2

Receiver operating characteristic curve of the optimal prediction model

n-3 = omega-3; AUC = area under the curve

Full size image
Fig. 3
figure 3

Distribution of measured total serum n-3 levels based on predicted levels of n-3

n-3 = omega-3

Full size image

Discussion

In this study, we developed a non-invasive dietary prediction tool to identify a sub-population of pregnant women who are likely to have a total serum n-3 level above 4.3% of total fatty acids based on their dietary and demographic characteristics. This tool took an average of six minutes to administer and had moderate test performance with 84.7% sensitivity and 54.7% specificity. To our knowledge, this is the first study to develop a model to predict n-3 serum levels above or below a given threshold in pregnancy.

A model has previously been developed in a non-pregnant population that aimed to predict the effect of n-3 supplementation on erythrocyte EPA and DHA levels [38]. In contrast to our study, their primary outcome was a continuous measure. The non-pregnant model observed that EPA and DHA levels had a modest correlation with predicted EPA and DHA levels, with a R2 of 0.69 [38]. Similar to our study, their prediction model included the usage of n-3 supplementation. Other studies in pregnant populations have examined the associations between dietary and demographic factors on n-3 markers [39, 40], with negative associations detected between smoking status and alcohol use with n-3 levels [40]. We could not evaluate the role of smoking or alcohol use in this study as there were very few smokers or mothers consuming alcohol in our study population.

This study demonstrates the potential to predict pregnant women with total serum n-3 levels above 4.3% of total fatty acids using basic sociodemographic data and a short nutrition questionnaire. Using our optimal prediction model, we had a sensitivity of 84.7%. For clinical utility, a sensitivity of 90% would be optimal to minimise any potential harm from n-3 supplementation in pregnancy to women with high baseline n-3 levels. In general, an AUROC curve of 0.5 suggests no discrimination, 0.7 to 0.8 is considered acceptable, 0.8 to 0.9 is considered excellent, and more than 0.9 is considered outstanding [41]. The performance measures of our optimal model demonstrate acceptable discrimination, however, in its current form, our prediction model is not appropriate for routine antenatal care. Further, this tool has been developed to predict serum n-3 levels associated with an increased risk of early PTB with additional n-3 supplementation and consequently is not applicable to other perinatal outcomes including small for gestational age or intrauterine growth restriction, Currently, there are no well-defined serum n-3 levels predictive for other perinatal outcomes.

To improve model performance, adjustment for factors including alcohol use and smoking status may be useful [40]. Further, adjusting for total energy intake may be valuable as it has been shown to improve the correlation between estimated dietary intake and measured nutrient concentrations in blood [42]. However, additional assessment of total energy intake to our predictive model would substantially increase the time required to complete the questionnaire. This may be prohibitive when designing a tool to screen all pregnant women. Further, future prediction models may need to be more complex and capable of predicting more than one threshold. This need is highlighted by the recent South Australian recommendations based on ORIP trial data that recommends three specific n-3 supplementation guidelines dependent on baseline serum n-3 levels. Pregnant women with total serum n-3 levels below 3.7% of total fatty acids are recommended to begin n-3 supplementation, with evidence suggesting the potential to reduce the risk of early PTB by 77% [20]. For those with levels between 3.7% and 4.3%, no change in n-3 supplementation status is required. Women with levels above 4.3% who are taking high dose supplementation (> 1 g/day) are recommended to cease supplementation if it has already been commenced or refrain from initiating additional high dose n-3 supplementation (> 1 g/day) as it may increase the risk of early PTB by 2.27-fold [43]. Ideally, development of a non-invasive multinomial model to predict membership to one of the three groups (< 3.7%, 3.7 to 4.3% and > 4.3%) may be valuable in optimising which pregnant women should be offered n-3 supplementation. However, development of such a model presents significant statistical challenges where careful consideration is required in the rate of various events in each group and the number of predictors accounted for in the model [44]. Based on the distribution of women from this study and the ORIP trial, the rate of each event (i.e., total n-3 serum levels < 3.7%, 3.7–4.3% and > 4.3%) is estimated to be 15–20%, 20–40% and 55–60%, respectively. Assuming adjustment for up to eight predictors (i.e., dietary intake of n-3, supplementary n-3 use, total energy intake, maternal age, BMI, socioeconomic status, alcohol use and smoking status), an estimated 600–1200 participants will be required to ensure ten to twenty events occur per variable to perform a sufficiently powered study [44].

One potential use for our prediction tool may be used to reduce the number of pregnant women requiring expensive serum testing (Fig. 4.) In this scenario, women with singleton pregnancies with predicted total n-3 serum levels equal to or below 4.3% of total fatty acids should undergo definitive serum fatty acid testing and receive n-3 supplementation as appropriate. In contrast, women with singleton pregnancies with predicted total n-3 serum levels above 4.3% of total fatty acids, do not undergo further serum fatty acid testing or n-3 supplementation. This reduces the test burden to 30% of our study population. Our predictive tool was able to correctly classify 72.7% of the population to this higher group. However, due to the limited predictive ability of our developed tool, 19% of the total study population were incorrectly classified and may miss out on potential benefits of n-3 supplementation. By utilising Australian birth data [45] and n-3 fatty acid results of Australian pregnant women [32], using the predictive tool as a screening tool to reduce serum testing, resulted in an estimated total saving of AUD$33.9 million when the financial burden of preterm birth was considered. In Fig. 4, this scenario (3. Screen all/test low) is compared to two other scenarios with consideration to the percentage of the population undergoing n-3 screening, testing and supplementation and the costs associated. In these scenarios, definitive serum fatty acid testing cost $200 which is congruent with costs in NSW [46].

Fig. 4
figure 4

Scenarios for n-3 supplementation using screening and/or testing of maternal n-3 baseline levels

~ = approximately; M = million; n-3 = omega-3; supp. = supplementation

Full size image

Overall, this study has several strengths. First, the sample size is adequate to minimise variability in the prediction model to ideally minimise shrinkage to less than 10%. Second, the inclusion criteria for this study were broad increasing its generalisability to an Australian population. Third, we followed guidelines and recommendations for development of prediction models supported by current literature [47,48,49,50]. For example, internal validation was performed with bootstrapping to assess for potential overfitting and optimism. Bootstrapping is considered a superior approach compared to other methods as it provides stable estimates with less bias [51]. Fourth, the prediction tool developed was non-invasive, inexpensive, and quick to administer.

There are several important limitations that should be considered when interpreting this study. First, n-3 dietary intake was evaluated using a dietary questionnaire which has an inherent risk of recall bias and misreporting. Our modified questionnaire assessed three months of intake retrospectively which may increase recall bias. To minimise misreporting of food portions, we included visual aids where possible. Second, adjustment for total energy intake was not performed in this study which may attenuate the correlations between dietary intake and biomarkers of n-3. Third, other factors that may influence n-3 levels including physical activity levels, alcohol intake, smoking status and genetic differences between individuals were also not considered in our analysis. Fourth, no external validation has yet been performed for our optimal prediction model. Fifth, due to wide variation in population based nutritional profiles, the generalisability of this tool is likely to be limited to an Australian Caucasian population. Our dietary questionnaire assessed consumption of n-3 food items common to an Australian diet and the proportion of n-3 in food items was defined by an Australian national nutrient database. For applicability in other populations, this prediction tool may need to be customised to population-based nutritional profiles.

Conclusion

In conclusion, we developed a simple tool to predict the subgroup of pregnant women who have total serum n-3 levels above 4.3% based on their dietary intake of n-3 and sociodemographic factors. Although our optimal prediction model only had moderate test performance, this study demonstrates the potential for inexpensive and non-invasive tools to identify baseline maternal n-3 status in early pregnancy which may play a valuable role in optimising n-3 supplementation. Before this tool can be applied to the clinical setting, further enhancement of model performance and model validation is required.

Availability of data and materials

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Abbreviations

ALA:

α-linolenic acid

AUROC:

Area under the receiver operating characteristic

CI:

Confidence interval

DHA:

Docosahexaenoic acid

DPA:

Docosapentaenoic acid

EPA:

Eicosapentaenoic acid

FNR:

False negative rate

FPR:

False positive rate

LCPUFA:

Long chain polyunsaturated fatty acid

n-3:

Omega-3

NB:

Net benefit

NPV:

Negative predictive value

NRI:

Net reclassification improvement

ORIP:

Omega-3 to Reduce the Incidence of Preterm Birth

PPV:

Positive predictive value

PTB:

Preterm birth

PUFA:

Polyunsaturated fatty acid

RCT:

Randomised controlled trial

RR:

Risk ratio

TRIPOD:

Transparent Reporting of a multivariable prediction model for Individual Prognosis or Diagnosis

TPR:

True positive rate

TNR:

True negative rate

References

  1. Liu L, Oza S, Hogan D, Chu Y, Perin J, Zhu J, et al. Global, regional, and national causes of under-5 mortality in 2000–15: an updated systematic analysis with implications for the Sustainable Development Goals. The Lancet. 2016;388(10063):3027–35.

    Article  Google Scholar 

  2. Chawanpaiboon S, Vogel JP, Moller A-B, Lumbiganon P, Petzold M, Hogan D, et al. Global, regional, and national estimates of levels of preterm birth in 2014: a systematic review and modelling analysis. The Lancet Global Health. 2019;7(1):e37–e46.

    Article  PubMed  Google Scholar 

  3. Blencowe H, Cousens S, Chou D, Oestergaard M, Say L, Moller A-B, et al. Born too soon: the global epidemiology of 15 million preterm births. Reproductive Health. 2013;10(1):2.

    Article  Google Scholar 

  4. Martin JA, Osterman MJK. Describing the Increase in Preterm Births in the United States, 2014–2016. NCHS Data Brief. 2018(312):1–8.

  5. Romero R, Dey SK, Fisher SJ. Preterm labor: one syndrome, many causes. Science. 2014;345(6198):760–5.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  6. Medley N, Vogel JP, Care A, Alfirevic Z. Interventions during pregnancy to prevent preterm birth: an overview of Cochrane systematic reviews. Cochrane Database of Systematic Reviews. 2018(11).

  7. Middleton P, Gomersall JC, Gould JF, Shepherd E, Olsen SF, Makrides M. Omega-3 fatty acid addition during pregnancy. Cochrane Database of Systematic Reviews. 2018(11).

  8. Serra R, Peñailillo R, Monteiro LJ, Monckeberg M, Peña M, Moyano L et al. Supplementation of Omega 3 during Pregnancy and the Risk of Preterm Birth: A Systematic Review and Meta-Analysis. Nutrients. 2021;13(5).

  9. Best KP, Gibson RA, Makrides M. ISSFAL statement number 7 – Omega-3 fatty acids during pregnancy to reduce preterm birth. Prostaglandins Leukot Essent Fatty Acids. 2022;186:102495.

    Article  CAS  PubMed  Google Scholar 

  10. Calder PC, Yaqoob P. Understanding omega-3 polyunsaturated fatty acids. Postgrad Med. 2009;121(6):148–57.

    Article  PubMed  Google Scholar 

  11. Saini RK, Keum YS. Omega-3 and omega-6 polyunsaturated fatty acids: Dietary sources, metabolism, and significance - A review. Life Sci. 2018;203:255–67.

    Article  CAS  PubMed  Google Scholar 

  12. Calder PC. Docosahexaenoic acid. Ann Nutr Metab. 2016;69(Suppl 1):7–21.

    PubMed  Google Scholar 

  13. Calder PC. N – 3 Polyunsaturated fatty acids, inflammation, and inflammatory diseases. Am J Clin Nutr. 2006;83(6):1505S–19S.

    Article  CAS  PubMed  Google Scholar 

  14. Olsen S, Sørensen TA, Secher N, Hansen H, Jensen B, Sommer S, et al. Intake of marine fat, rich in (n-3)-polyunsaturated fatty acids, may increase birthweight by prolonging gestation. The Lancet. 1986;328(8503):367–9.

    Article  Google Scholar 

  15. Olsen SF, Secher NJ, Bjornsson S, Weber T, Atke A. The potential benefits of using fish oil in relation to preterm labor: the case for a randomized controlled trial? Acta Obstet Gynecol Scand. 2003;82(11):978–82.

    Article  PubMed  Google Scholar 

  16. de Groot RHM, Emmett R, Meyer BJ. Non-dietary factors associated with n-3 long-chain PUFA levels in humans – a systematic literature review. Br J Nutr. 2019;121(7):793–808.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Calder PC. Omega-3 fatty acids and inflammatory processes. Nutrients. 2010;2(3):355–74.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  18. Block RC, Harris WS, Pottala JV. Determinants of blood cell Omega-3 fatty acid content. Open Biomark J. 2008;1:1–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Nordgren TM, Lyden E, Anderson-Berry A, Hanson C. Omega-3 fatty acid intake of pregnant women and women of Childbearing Age in the United States: potential for Deficiency? Nutrients. 2017;9(3):197.

    Article  PubMed  PubMed Central  Google Scholar 

  20. Simmonds LA, Sullivan TR, Skubisz M, Middleton PF, Best KP, Yelland LN, et al. Omega-3 fatty acid supplementation in pregnancy-baseline omega-3 status and early preterm birth: exploratory analysis of a randomised controlled trial. BJOG. 2020;127(8):975–81.

    Article  CAS  PubMed  Google Scholar 

  21. Carlson SE, Gajewski BJ, Valentine CJ, Kerling EH, Weiner CP, Cackovic M et al. Higher dose docosahexaenoic acid supplementation during pregnancy and early preterm birth: A randomised, double-blind, adaptive-design superiority trial. EClinicalMedicine. 2021;36.

  22. Leventakou V, Roumeliotaki T, Martinez D, Barros H, Brantsaeter AL, Casas M, et al. Fish intake during pregnancy, fetal growth, and gestational length in 19 european birth cohort studies. Am J Clin Nutr. 2014;99(3):506–16.

    Article  CAS  PubMed  Google Scholar 

  23. Brantsæter AL, Englund-Ögge L, Haugen M, Birgisdottir BE, Knutsen HK, Sengpiel V, et al. Maternal intake of seafood and supplementary long chain n-3 poly-unsaturated fatty acids and preterm delivery. BMC Pregnancy Childbirth. 2017;17(1):41.

    Article  PubMed  PubMed Central  Google Scholar 

  24. Olsen SF, Secher NJr. Low consumption of seafood in early pregnancy as a risk factor for preterm delivery: prospective cohort study. BMJ. 2002;324(7335):447.

    Article  PubMed  PubMed Central  Google Scholar 

  25. Olsen SF, Halldorsson TI, Thorne-Lyman AL, Strøm M, Gørtz S, Granstrøm C, et al. Plasma concentrations of long chain N-3 fatty acids in early and mid-pregnancy and risk of early Preterm Birth. EBioMedicine. 2018;35:325–33.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Klebanoff MA, Harper M, Lai Y, Thorp J Jr, Sorokin Y, Varner MW, et al. Fish consumption, erythrocyte fatty acids, and preterm birth. Obstet Gynecol. 2011;117(5):1071–7.

    Article  PubMed  PubMed Central  Google Scholar 

  27. Ciesielski TH, Bartlett J, Williams SM. Omega-3 polyunsaturated fatty acid intake norms and preterm birth rate: a cross-sectional analysis of 184 countries. BMJ Open. 2019;9(4):e027249.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Makrides M, Best K, Yelland L, McPhee A, Zhou S, Quinlivan J, et al. A randomized trial of prenatal n – 3 fatty acid supplementation and Preterm Delivery. N Engl J Med. 2019;381(11):1035–45.

    Article  CAS  PubMed  Google Scholar 

  29. Zhang AC, Downie LE. Clinical Omega-3 Dietary Survey. Nutrients. 2019.

  30. Food Standards Australia New Zealand. AUSNUT 2011–13—Australian Food Composition Database Canberra, Australia2014 [Available from: www.foodstandards.gov.au.

  31. Liu G, Makrides M, Coates P, Lam K, Ranieri E, Mas E, et al. A rapid method for the screening of fatty acids in lipids in plasma or serum without prior extraction. Prostaglandins Leukot Essent Fatty Acids. 2022;178:102416.

    Article  CAS  PubMed  Google Scholar 

  32. Simmonds LA, Yelland LN, Best KP, Liu G, Gibson RA, Makrides M. Translating n-3 polyunsaturated fatty acid status from whole blood to plasma and red blood cells during pregnancy: translating n-3 status across blood fractions in pregnancy. Prostaglandins Leukot Essent Fatty Acids. 2022;176:102367.

    Article  CAS  PubMed  Google Scholar 

  33. Burdge GC, Wootton SA. Conversion of alpha-linolenic acid to eicosapentaenoic, docosapentaenoic and docosahexaenoic acids in young women. Br J Nutr. 2002;88(4):411–20.

    Article  CAS  PubMed  Google Scholar 

  34. Unal I. Defining an optimal cut-point value in ROC analysis: an Alternative Approach. Comput Math Methods Med. 2017;2017:3762651.

    Article  PubMed  PubMed Central  Google Scholar 

  35. Riley RD, Snell KI, Ensor J, Burke DL, Harrell FE Jr, Moons KG, et al. Minimum sample size for developing a multivariable prediction model: PART II - binary and time-to-event outcomes. Stat Med. 2019;38(7):1276–96.

    Article  PubMed  Google Scholar 

  36. Cox DR, Snell EJ. Analysis of binary data:Routledge; 2018.

  37. R Core Team. R: A language and environment for statistical computing. Vienna, Austria2018 [Available from: https://www.R-project.org/.

  38. Walker RE, Jackson KH, Tintle NL, Shearer GC, Bernasconi A, Masson S, et al. Predicting the effects of supplemental EPA and DHA on the omega-3 index. Am J Clin Nutr. 2019;110(4):1034–40.

    Article  PubMed  Google Scholar 

  39. Crawford SA, Christifano DN, Kerling EH, Gajewski BJ, Valentine CJ, Gustafson KM et al. Validation of an abbreviated food frequency questionnaire for estimating DHA intake of pregnant women in the United States. Prostaglandins, leukotrienes and essential fatty acids. 2022;177:102398-.

  40. Wilson NA, Mantzioris E, Middleton PF, Muhlhausler BS. Influence of sociodemographic, lifestyle and genetic characteristics on maternal DHA and other polyunsaturated fatty acid status in pregnancy: a systematic review. Prostaglandins Leukot Essent Fatty Acids. 2020;152:102037.

    Article  CAS  PubMed  Google Scholar 

  41. Mandrekar JN. Receiver operating characteristic curve in diagnostic test assessment. J Thorac Oncol. 2010;5(9):1315–6.

    Article  PubMed  Google Scholar 

  42. Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake in epidemiologic studies. Am J Clin Nutr. 1997;65(4 Suppl):1220S-8S; discussion 9S-31S.

  43. Omega-3 status test for prematurity risk South Australian Health and Medical Research. Institute2022 [Available from: https://sahmri.org.au/research/themes/women-and-kids/programs/child-nutrition-research-centre/projects/omega-3-status-test-for-prematurity-risk.

  44. de Jong VMT, Eijkemans MJC, van Calster B, Timmerman D, Moons KGM, Steyerberg EW, et al. Sample size considerations and predictive performance of multinomial logistic prediction models. Stat Med. 2019;38(9):1601–19.

    Article  PubMed  PubMed Central  Google Scholar 

  45. Newnham JP, Schilling C, Petrou S, Morris JM, Wallace EM, Brown K et al. The health and educational costs of preterm birth to 18 years of age in Australia. Aust N Z J Obstet Gynaecol. 2021.

  46. NSW Health Pathology. Fatty acids - Essential (Serum) [Available from: http://www.palms.com.au/php/labinfo/info_index.php?tab=1.

  47. Collins GS, Reitsma JB, Altman DG, Moons KGM. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ. 2015;350(jan07 4):g7594–g.

    Article  PubMed  Google Scholar 

  48. Leisman DE, Harhay MO, Lederer DJ, Abramson M, Adjei AA, Bakker J et al. Development and Reporting of Prediction Models: Guidance for Authors From Editors of Respiratory, Sleep, and Critical Care Journals. Crit Care Med. 2020;48(5).

  49. Steyerberg EW. Validation in prediction research: the waste by data splitting. J Clin Epidemiol. 2018;103:131–3.

    Article  PubMed  Google Scholar 

  50. Steyerberg EW, Harrell FE Jr. Prediction models need appropriate internal, internal-external, and external validation. J Clin Epidemiol. 2016;69:245–7.

    Article  PubMed  Google Scholar 

  51. Steyerberg EW, Harrell FE, Borsboom GJJM, Eijkemans MJC, Vergouwe Y, Habbema JDF. Internal validation of predictive models: efficiency of some procedures for logistic regression analysis. J Clin Epidemiol. 2001;54(8):774–81.

    Article  CAS  PubMed  Google Scholar 

Download references

Acknowledgements

The authors would like to thank all the women for their participation in this study and all members involved in development of the electronic questionnaire, and recruitment and processing of samples. We would also like to extend our gratitude to Professor Maria Makrides and Dr Ge Liu from South Australian Health and Medical Research Institute for their ongoing support and expertise.

Funding

Funding for the fatty acid analyses was provided by the John Hunter Hospital Maternity and Gynaecology Department Research Trust Fund.

Author information

Authors and Affiliations

  1. School of Medicine and Public Health, University of Newcastle, Callaghan, NSW, 2300, Australia

    Joanna YX Fu, Carol A Wang, Elyse Mead, Jason Phung & Craig E Pennell

  2. Hunter Medical Research Institute, New Lambton Heights, NSW, 2305, Australia

    Carol A Wang, Jason Phung & Craig E Pennell

  3. South Australian Health and Medical Research Institute, Adelaide, SA, 5000, Australia

    Ge Liu & Maria Makrides

  4. John Hunter Hospital, New Lambton Heights, NSW, 2305, Australia

    Jason Phung & Craig E Pennell

Authors
  1. Joanna YX Fu
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Carol A Wang
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Ge Liu
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Elyse Mead
    View author publications

    You can also search for this author in PubMed Google Scholar

  5. Jason Phung
    View author publications

    You can also search for this author in PubMed Google Scholar

  6. Maria Makrides
    View author publications

    You can also search for this author in PubMed Google Scholar

  7. Craig E Pennell
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

JYXF wrote the first draft; CAW performed statistical modelling; GL conducted serum fatty acid analysis; JYXF, CAW and CEP extensively edited the manuscript; all other authors provided input on the final version and approved the final manuscript.

Corresponding author

Correspondence to Craig E Pennell.

Ethics declarations

Ethics approval and consent to participate

This study was approved by the Hunter New England Human Research Ethics Committee of the Hunter New England Local Health District (Reference 2020/ETH00498 and 2020/ETH02881). All methods were carried out in accordance with relevant guidelines and regulations. All participants were consented by a research assistant not directly involved in the provision of clinical care. Informed written consent was obtained from all study participants and/or their legal guardian(s) including for collection of human tissue samples. All clinical data collected was de-identified. In the case that n-3 levels were abnormal, the participant and their general practitioners were notified so the appropriate follow up could be provided.

Consent for publication

Not applicable.

Competing interest

The authors declare they have no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Additional file 1. Supplementary Table 1

. Transparent Reporting of Individual Prognosis and Diagnosis (TRIPOD) Checklist

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Fu, J.Y., Wang, C.A., Liu, G. et al. Development and internal validation of a non-invasive clinical tool to predict sufficient omega-3 levels in early pregnancy. BMC Pregnancy Childbirth 23, 442 (2023). https://doi.org/10.1186/s12884-023-05687-2

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s12884-023-05687-2

Keywords

BMC Pregnancy and Childbirth

ISSN: 1471-2393

Contact us