Prevalence and determinants of perinatal mental disorders in women with gestational diabetes in New Zealand: Findings from a national longitudinal study.

INTRODUCTION

Gestational diabetes mellitus (GDM) refers to hyperglycemia (blood glucose concentrations which are above normal but below those diagnostic of diabetes mellitus) detected during pregnancy, 1 and is the most common metabolic disease experienced in pregnancy. 2 In New Zealand, the prevalence of GDM has been reported as approximately 6%, 3 with an annual 14% increase in national prevalence from 2001 to 2012. 4

GDM can result in adverse perinatal and long‐term effects for mother and child, including high rates of maternal perinatal mental disorders. A recent systematic review that included 62 studies reported a twofold increase in the risk of antenatal depression and a 60% increase in the risk of postnatal depression in women with GDM. 5 Other mental health outcomes including anxiety and stress have been reported to be higher among women with GDM than women without GDM. 6 , 7 Varying mechanisms, including biological and psychosocial factors, have been proposed to underlie the association between GDM and poor mental well‐being. Biologically, depression and GDM are both associated with inflammation, oxidative stress, neurodegenerative processes and hormonal dysregulation. 8 , 9 In addition, psychosocial factors such as perception of a high‐risk pregnancy, difficulties in accessing and adhering to treatment, and concerns about perinatal complications are suggested to contribute to the association between GDM and poor mental health outcomes. 9 , 10

However, high‐quality evidence on predictors of poor mental health outcomes in women with GDM is limited. Two longitudinal cohort studies assessing predictors of perinatal depressive symptoms among women with GDM in China suggested that hyperglycemia may be associated with perinatal depression. 11 , 12 A small cross‐sectional study in Japan also reported that symptoms of depression among pregnant women with GDM may be associated with diet. 13 Considering that concurrent diagnosis of GDM with antenatal depression is associated with higher rates of poor perinatal outcomes including preeclampsia, preterm birth and neonatal respiratory distress syndrome, 14 , 15 it is imperative that women who are most at risk of perinatal mental disorders are identified early and managed effectively. In the New Zealand context, there are no data on predictors of perinatal mental disorders in women with GDM.

We, therefore, aimed to assess the sociodemographic, pregnancy‐related and clinical predictors of perinatal mental disorders, namely anxiety, depression and poor mental health‐related quality of life (HRQoL) among a cohort of women with GDM in New Zealand.

MATERIAL AND METHODS

This was a prospective cohort study of a subgroup of women with GDM who took part in the TARGET Trial. 16 The trial was a nationally representative multicenter stepped‐wedge cluster randomized trial in 10 hospitals in New Zealand conducted between 2015 and 2018 to assess optimal glycemic treatment targets for GDM care. All 10 hospitals initially used less‐tight targets 17 (fasting plasma glucose <5.5 mmol/L [<99 mg/dL]; 1‐hour postprandial <8.0 mmol/L [<144 mg/dL[; 2‐hour postprandial <7.0 mmol/L [<126 mg/dL]), and were sequentially randomized in clusters of two to using tighter targets 4 (fasting plasma glucose ≤5.0 mmol/L [≤90 mg/dL], 1‐hour postprandial ≤7.4 mmol/L [≤133 mg/dL]; 2‐hour postprandial ≤6.7 mmol/L [≤120 mg/dL]) at 4‐month intervals. Women diagnosed with GDM were treated based on the targets being used by the hospital at the time they received their antenatal care. Women between 22 and 34 weeks’ gestation with GDM diagnosed by a 75‐g oral glucose tolerance test (OGTT) who consented to receiving questionnaires screening for anxiety, depression and HRQoL were eligible for inclusion in our study (Figure 1). The questionnaires were administered at time of diagnosis of GDM (trial entry), 36 weeks’ gestation and 6 months after the birth. Exclusion criteria for the TARGET study included women with a known major fetal malformation.

Anxiety was assessed using the 6‐item Spielberger State–Trait Anxiety Inventory (STAI) 18 with scores >15 indicating presence of symptoms of anxiety. 19 Depression was assessed by the Edinburgh Postnatal Depression Scale (EPDS), with a score >12 indicating significant vulnerability to depression. 19 , 20 Poor mental HRQoL was assessed using the mental component summary (MCS) of the 36‐Item Short‐Form General Health Survey (SF‐36). Women were deemed to have a poor mental HRQoL if they had MCS <40 (>1 SD below the New Zealand standardized mean of 50). This cut‐off has good positive predictive value for poor mental health outcomes compared with other similar validated psychological instruments. 21

Potential predictors of interest were chosen on the basis of expert clinical knowledge so as not to overcapitalize on chance associations, and were grouped into sociodemographic, pregnancy‐related and clinical factors. The sociodemographic factors included age, ethnicity, body mass index (BMI) and smoking status. Pregnancy‐related factors were parity, previous history of GDM, gestational age at trial entry and gestational weight gain. The clinical factors assessed included dietary adherence score, 22 75‐g oral glucose tolerance test (OGTT) diagnostic values and blood pressure measurements. Maternal age was assessed as both continuous and categorical (categorized as <30, 30–34, 35–39 and ≥40 years).

Ethnicity was self‐reported and prioritized using the New Zealand Ministry of Health categories: Māori, Pacific Peoples, Asian, Other and European. 23 Body mass index (BMI) was calculated from self‐reported pre‐pregnancy or booking weight and height, and categorized as 18.5–24.9 kg/m2 (normal), 25.0–29.9 kg/m2 (overweight) and ≥30.0 kg/m2 (obese). Smoking status was collected from the hospital record at the first antenatal visit and coded as “yes” or “no or unknown”. Parity was defined as the number of previous births at or after 20 weeks’ gestation, and categorized as 0, 1–3 and ≥4. Previous GDM was assessed among women with a previous pregnancy of ≥20 weeks’ gestation e and categorized as “yes” or “no or unknown”. Gestational age at oral glucose tolerance test was calculated in weeks and assessed as a continuous variable. Gestational weight gain was calculated as the difference in weight at 36 weeks’ gestation or at term minus the pre‐pregnancy or booking visit weight. The values were assessed as a continuous variable. Dietary adherence score was developed using data from the TARGET Trial, which measure the level of adherence to all 10 New Zealand food‐related guidelines for women with GDM at 36 weeks. 22 The continuous scores range from 0–10, where 0 indicates no adherence and 10 indicates full adherence to the food‐related dietary recommendations. OGTT result at trial entry, fasting plasma glucose and 2‐hour postprandial results at time of GDM diagnosis were assessed as continuous variables. Blood pressure, systolic and diastolic measurements at booking, 36 weeks’ gestation and 6 months postpartum were assessed as continuous variables.

Baseline characteristics of all women were summarized using descriptive analyses and SAS software version 9.4 (SAS Institute, Cary, NC, USA). Mental health outcomes were analyzed as categorical variables (eg vulnerable to depression = 1, not vulnerable = 0) to enhance clinical interpretation, and the independent variables assessed were chosen based on biological plausibility and expert clinical evidence. It was anticipated that each of these planned analyses would have adequate power to explore relationships between one independent variable per 10 outcomes.

First, univariable logistic regression analyses were conducted to de‐termine the independent relationships of each of the potential predictors with each of the outcome measures at each time point (baseline, 36 weeks’ gestation and 6 months after birth). A statistical significance of P < 0.15 was used as the cut‐off to include predictors into the multivariable logistic regression models. Potential predictors were then examined and eliminated according to their least significant contribution in explaining the logarithm of the odds of having poor mental health outcomes until only statistically significant predictors remained (ie P < 0.05). Adjusted odds ratios (ORs), 95% confidence intervals (CIs), and P‐values are provided for the final multivariable models.

To compare the change in prevalence of the poor mental health outcomes over the three time points, risk differences (with their 95% CI and P‐values) were calculated between trial entry and 36 weeks’ gestation, trial entry to 6 months after birth, and 36 weeks’ gestation to 6 months after birth.

RESULTS

Of the 455 eligible participants, 414 women completed the screening questionnaires at trial entry (91%), 341 (75%) at 36 weeks’ gestation and 340 (75%) at 6 months after birth (Figure 1).

The mean age in the study was 33 years and 44% of women were of European ethnicity. Most of the women in the study were overweight or obese (89%) and 80% had no history of previous GDM. The median gestational age at trial entry was 27 weeks. At this time, the median fasting plasma glucose concentration was 5.0 mmol/L and the 2‐hour postprandial concentration was 9.4 mmol/L (Table 1).

Characteristics of women at trial entry, 36 weeks’ and 6 months after birth.

Note: All variables presented as mean ± standard deviation, number (percentage) or median (interquartile range) unless otherwise indicated.

Abbreviations: BMI, body mass index; EPDS, Edinburgh Postnatal Depression Scale; GDM, gestational diabetes mellitus; HRQoL, health‐related quality of life; OGTT, oral glucose tolerance test; MCS, Mental Component Summary; STAI, Spielberger State–Trait Anxiety Inventory.

Among women with previous pregnancy ≥20 weeks’ gestation.

The prevalence of anxiety was 17% (71/408) at trial entry, 15% (50/327) at 36 weeks’ gestation and 13% (42/333) at 6 months after birth (Table 2, Figure 2). The prevalence of vulnerability to depression was 16% (66/405) at trial entry and decreased to approximately 8% at 36 weeks’ gestation (27/334) and 6 months after birth (25/332). A similar decrease in prevalence was noted for poor mental HRQoL from 27% (105/396) at trial entry to 20% at 36 weeks’ gestation (65/321) and 6 months after birth (67/325). Overall, 29% of women in our study reported at least one poor mental health outcome at trial entry and 12% reported all three.

Prevalence and changes in mental health measures during pregnancy and 6 months after birth.

Graph of mental health measures at trial entry, 36 weeks’ gestation and 6 months after birth.

At trial entry, univariable logistic regression analyses identified ethnicity, BMI and fasting plasma glucose concentration as factors associated with anxiety (Table S1). At 36 weeks’ gestation, BMI and fasting plasma glucose concentration remained associated, and age and gestational weight gain were additionally associated. At 6 months after the birth, gestational age at OGTT, 2‐hour postprandial glucose concentration and diastolic blood pressure were associated with anxiety.

On multivariable analysis, including the factors confirmed from the univariable analysis, there were no statistically significant independent predictors of anxiety at trial entry. However, at 36 weeks, a 1‐year increase in age was associated with a small decrease in the odds of having anxiety (OR 0.93; 95% CI 0.87–0.99), whereas a 1‐kg increase in gestational weight gain was associated with a small increase in the odds of anxiety (OR 1.05; 95% CI 1.00–1.10). At 6 months after the birth, a 1‐week increase in the gestational age at OGTT was associated with a 27% increase in the odds of anxiety, and an increase of 1 mmol/L in the 2‐hour postprandial glucose concentration was associated with a 31% increase. An increase of 1 mmHg in the diastolic blood pressure was associated with a 4% decrease in the odds of anxiety (Table 3).

Results from the multivariable analysis*.

Abbreviations: GDM, gestational diabetes mellitus; HRQoL, health‐related quality of life; OGTT, oral glucose tolerance test.

Only factors associated with anxiety, depression and poor mental HRQoL at trial entry, 36 weeks'gestation and 6 months postpartum with a P‐value <0.15 from univariable analyses were included in the multivariable analysis and are shown in the table.

At trial entry, univariable logistic regression analyses identified age and previous history of GDM as factors associated with depression (Table S2). Ethnicity was associated with depression at 36 weeks’ gestation, and gestational age at OGTT and diastolic blood pressure were associated at 6 months after the birth.

Multivariable regression results showed that at trial entry, women who had a previous history of GDM had twice the odds of depression compared with those with no or unknown history (OR 2.07; 95% CI 1.11–3.87) (Table 3). At 36 weeks’ gestation, women of Pacific ethnicity had three times the odds of depression (OR 3.38; 95% CI 1.12–10.23) compared with those of European and Other ethnicities. There were no statistically significant predictors of depression at 6 months after the birth from the multivariable models.

At 36 weeks’ gestation, univariable analyses identified age, ethnicity and previous history of GDM were associated with poor mental HRQoL, and at 6 months after birth, smoking, gestational age at OGTT, fasting and 2‐hour postprandial glucose concentration were associated with poor mental HRQoL (Table S3).

Results from the multivariable regression found that women of Pacific ethnicity compared with women of European/Other ethnicities had double the odds of having poor mental HRQoL (OR 2.54; 95% CI 1.24–5.20) at trial entry. At 36 weeks’ gestation, a 1‐year increase in age was associated with a 6% decrease in the odds of having poor mental HRQoL (OR 0.94; 95% CI 0.88–0.99), and a 1‐week increase in gestational age at OGTT was associated with a 21% increase in the odds of poor mental HRQoL (OR 1.21; 95% CI 1.04–1.41). Women with a previous history of GDM had twice the odds of poor mental HRQoL compared with those with no or unknown history (OR 2.20; 95% CI 1.07–4.52). Being of Asian ethnicity was also associated with almost twice the odds of poor mental HRQoL at 36 weeks’ gestation (OR 1.93; 95% CI 1.00–3.71). At 6 months after the birth, a 1‐week increase in the gestational age at OGTT was associated with a 15% increase in the odds of poor mental HRQoL (Table 3).

DISCUSSION

At time of GDM diagnosis, around one in four women (29%; 95% CI 25–33%) in our study reported at least one mental health disorder (any of anxiety, vulnerability to depression or poor mental HRQoL), and one in eight women (12%; 95% CI 9–16%) reported coexisting anxiety, depression and poor mental HRQoL. This comorbid prevalence is higher than that previously reported in healthy pregnant women. A meta‐analysis of 44 studies, including studies from New Zealand and Australia, estimated a 9% prevalence of coexisting antenatal anxiety and depression in healthy pregnant women 24 The higher prevalence observed in our study is consistent with findings that pregnancy complications, including GDM, increase the risk for perinatal mental disorders. 25

Around one in six women (17%) reported symptoms of anxiety at time of GDM diagnosis. High rates of anxiety in women with GDM at time of diagnosis have been associated with increased fear of diabetes‐related treatments and procedures (eg insulin injections) and fear for mother's own health and infant's health. 26

One in six women (16%) reported depressive symptoms at time of GDM diagnosis, a lower estimate than has previously been reported in women with GDM, and healthy pregnant women. A systematic review and meta‐analysis of 62 studies reported that 28% of women experienced depressive symptoms at time of GDM diagnosis. 5 This meta‐analysis, however, used a lower EPDS cut‐off of ≥10, which may explain the higher prevalence. Generally, women with GDM have been reported to have higher rates of depression compared with pregnant women without GDM due to biological and psychosocial risk factors associated with both pregnancy and diabetes. 5 , 27 However, in our study, vulnerability to depression at time of GDM diagnosis was less common compared with the healthy pregnant population estimates (18% in high‐income countries). 28 It is possible that the lower prevalence of depression found in our cohort may be due in part to women who suffer from poor mental health outcomes not consenting to be a part of the study. Since we were unable to collect information from women who did not consent to receiving questionnaires, we cannot explore this possibility further.

By late pregnancy (36 weeks’ gestation), the women in our study reported a lower prevalence of vulnerability to depression (8%) and poor mental HRQoL (20%) compared with trial entry. This could be because all women in our study received treatment after diagnosis with GDM. In an Australian trial, women randomized to receive treatment for GDM showed an improvement in their HRQoL and reduced incidence of depression compared with those randomized to receive routine antenatal care. 19 We found no significant change in prevalence of anxiety from time of diagnosis to 36 weeks’ gestation. This is consistent with evidence that treatment of GDM does not raise anxiety levels in women with GDM; rather, efforts to achieve glycemic treatment targets may reassure women. 29

At 6 months after the birth, the prevalence of anxiety, vulnerability to depression and poor mental HRQoL did not differ significantly from late pregnancy levels, and women remained less likely to be depressed and had a better mental HRQoL than at time of GDM diagnosis. This improvement in mental health status in the postnatal period among women with GDM has been reported elsewhere, attributed to high‐quality clinical care, receipt of relevant information and support for women to navigate the postpartum period. 30

Among the sociodemographic factors we assessed, younger maternal age was associated with more anxiety and poorer mental HRQoL during pregnancy. Being of Pacific ethnicity vs European/Other ethnicity was associated with more vulnerability to depression and poorer mental HRQoL. Younger maternal age may reflect psychosocial risk factors such as inadequate social support and socioeconomic deprivation. 31 Similar to our findings, another New Zealand study identified ethnicity as a risk factor for antenatal depression in healthy pregnant women. 32 In that study, both Pacific and Asian ethnicity were associated with increased odds of antenatal depression, which persisted after adjusting for socioeconomic status. The authors suggested that this could be due to underlying high levels of perinatal depression among different ethnic groups. However, other studies have attributed the link between minority ethnicity and antenatal depression to socioeconomic deprivation and inequitable healthcare access. 31 , 33 A report on Pacific health in New Zealand reported late antenatal booking and infrequent maternity care as barriers to improving perinatal health outcomes in Pacific mothers. 34 These reasons could contribute to the increased odds for poor mental health outcomes among Pacific mothers in our study.

Among the pregnancy‐related factors we examined, previous history of GDM compared with no/unknown history was associated with increased risk of vulnerability to depression at time of GDM diagnosis and poor mental HRQoL at 36 weeks’ gestation. A qualitative study assessing the experiences of women with previous diagnoses of GDM reported that some women felt their pregnancies were “over‐monitored” and experienced stress in trying to comply with treatment regimens. 35 A subsequent diagnosis of GDM may result in poorer mental health outcomes in such women. Older gestational age at time of GDM diagnosis was also associated with poor mental HRQoL in both the antenatal and postnatal period, as well as postnatal anxiety. Although the reason for this finding is not clear, we hypothesize that women who do not receive their OGTT within the recommended 24–28 weeks may be facing difficulties with accessing antenatal care which may be a source of mental distress.

In the postpartum period, elevated 2‐hour postprandial glucose concentration during the diagnostic OGTT screening was associated with postnatal anxiety, whereas a high diastolic blood pressure was associated with marginally reduced odds of postnatal anxiety. Hyperglycemia has been shown to be associated with risk of depression in the general diabetes population via inflammation and oxidative stress caused by hyperglycemia which can result in cerebral microvascular dysfunction and subsequent depression. 36 Unexpectedly, a higher diastolic blood pressure was associated with a 4% decrease in the odds of postnatal anxiety in our study. The evidence surrounding the association between gestational hypertension and perinatal mental disorders is inconclusive, with a systematic review of six studies reporting higher anxiety scores in women with gestational hypertension; however, this was not statistically significant. 37 In women with GDM, a concurrent diagnosis of depression and GDM has been associated with increased rates of gestational hypertension, 14 but the extent to which high blood pressure in women with GDM predisposes to perinatal mental disorders has not been assessed. In our study the median blood pressure measured across all time points was within the normal range, presumably because women with GDM are also treated for other risk factors such as high blood pressure.

This study has several strengths. First, the study design was a prospective cohort looking at timepoints during pregnancy and the postnatal period allowing us to assess mental health outcomes at different timepoints. Secondly, we included the measurement of mental HRQoL as well as the well‐known measures of anxiety and depression in women with GDM to provide a broader perspective of mental health status in women with GDM. Additionally, the data used in the study were nationally representative, hence our findings are generalizable to the New Zealand healthcare context. However, it may not be generalizable to other settings, especially those that are more resource‐constrained.

The main limitation of our study was that our sample was made up of women who consented to further assessments in the TARGET Trial, and this process of self‐selection may have biased our sample towards healthier women. Another limitation is our lack of data on prenatal and early pregnancy mental health status of women in our study, factors which can predict perinatal mental disorders among women with GDM. 38 , 39 With regard to our assessment tools, although the SF‐36 questionnaire has been validated for use in the New Zealand population, the EPDS and 6‐item STAI have not been validated for use across all ethnicities in the New Zealand population to determine the optimal cut‐off points. We assigned a cut‐off for the SF‐36 MCS using an arbitrary cut‐off of <1 SD deviation below the New Zealand population mean. Additionally, the study had no control group, hence comparisons with the general population should be interpreted with care.

CONCLUSION

Our study found that around one in four women experienced poor mental health during pregnancy, with the number reducing to one in five in the postnatal period. We found a relatively high prevalence of comorbid antenatal mental disorders, highlighting the need to raise awareness of the burden of poor mental health in women with GDM. Reassuringly, the prevalence of these disorders does not appear to persist into late pregnancy and after birth. Our findings highlight the need to screen for other perinatal mental disorders (anxiety and poor mental HRQoL) in addition to depression, with younger mothers and women with Pacific ethnicity at particular risk, to ensure adequate mental health care is available to all mothers during pregnancy and after birth.

All the authors were involved in the conceptualization of the paper. PO wrote the main paper and JH and CC critically reviewed and revised it. PO and GG developed the methodology and PO carried out the statistical analysis. All authors read and approved the final article.

The TARGET Trial was funded by the Health Research Council of New Zealand (Grant Number 14/499). The funders had no role in study design, data collection and analysis, decision to publish or preparation of the article.

The authors have stated explicitly that there are no conflicts of interest in connection with this article.

The TARGET trial was approved by the Northern A Health and Disability Ethics Committee in New Zealand (14/NTA/163/AMO1) on October 23, 2014.