Postnatal depression affects around 1 in 10 women and not only impacts on the wellbeing of the mother, but can also have long term impacts on the mental and physical health of the infant (Stein et al, 2014).

The term ‘postnatal depression’ (PND) is often used as an umbrella term to describe psychological distress after birth. This can be unhelpful as it doesn’t take into account the range of experiences that women have and can lead to treatments being offered that are not appropriate. This is particularly the case for anxiety, which may be more prevalent than depression in the postnatal period (e.g. Fairbrother et al., 2016) and requires different kinds of interventions.

The authors of the current paper are from the Postpartum Depression: Action Towards Causes and Treatment (PACT) consortium. This is an international group of perinatal psychiatrists from 19 institutions in seven countries who aim to gather information about PND to explore a number of questions, including whether there are distinguishable subtypes of PND which might be relevant for treatment and prognosis, in particular taking into account comorbid anxiety.

In a previous paper (PACT, 2015) they identified factors which seemed to differentiate between types of PND, namely, severity, timing of onset, the presence of suicidal ideation and comorbid anxiety. The current paper (Putman et al, 2017) aims to extend that work to see if the patterns of symptoms differ depending on timing of onset, and to look more specifically at the role of anxiety. In this way, the authors want to “identify and describe clinically relevant subtypes of perinatal depression”.

As a clinical psychologist, I have received many referrals with the description ‘postnatal depression’. This term does little to capture the complexity or variety of emotions which are present at the time of having a baby. Despite this, there are often recognisable patterns of difficulties across different women and so it’s an intriguing idea that we might be able to identify specific subtypes and use this to help guide treatment decisions.

Methods

Participants included women from seven of the 19 international sites in the PACT consortium who had data fitting the following criteria:

• Reported depression in the perinatal period
• Between 19 and 40 years old
• Had information about the onset of symptoms
• Had complete data on the Edinburgh Postnatal Depression Scale (EPDS).

The EPDS is a 10-item self-report questionnaire which is widely used in research and clinical settings to identify women with symptoms of PND. 663 women met these criteria and the median EPDS measurement was taken at 4.5 months postpartum.

Based on their sample, EPDS scores were categorised into four severity levels:

1. No depression 0-9
2. Mild-moderate 10-16
3. Moderate to severe 17-21
4. Very severe 22-30

The authors also used the scores to divide the EPDS into three subscales using principal components analysis. They identified subscales related to depressed mood, anxiety and anhedonia (lack of pleasure).

Using women’s scores on each subscale as well as their overall severity score, the authors used k-means clustering to identify 5 subtypes of perinatal depression which differed in severity and type of symptoms.

In addition, by dividing the perinatal period into 6 time periods (the three trimesters of pregnancy and three postnatal periods, 0 to <4 weeks, 4 to <8 weeks and over 8 weeks) they explored whether women’s reported time of symptom onset was different in the five subtypes of PND.

Results

Timing of onset was associated with overall EPDS score; two thirds of women who reported that their depression started in the first trimester had ongoing moderate to severe or very severe symptoms when they completed the EPDS postnatally. In contrast, women who reported that their depression started in the second or third trimester were more likely to be in the mild-moderate or remitted category at the time of EPDS assessment. This suggests that a later onset during pregnancy is associated with better outcomes postnatally.

Where women reported postnatal onset this seemed to be associated with more severe symptoms at EPDS assessment.

The five subtypes of PND were:

1. Severe anxious depression, characterised by comorbid anxiety and depressive symptoms with a high overall score (32% of the sample)
2. Moderate anxious depression, characterised by comorbid anxiety and depressive symptoms but with a lower overall score (19%)
3. Anxious anhedonia, characterised by high scores on the anxiety and anhedonia items and with a high overall score (12%).
4. Pure anhedonia, characterised by high scores on the anhedonia symptoms with a mild-moderate overall score (11%).
5. Resolved depression, characterised by low scores across all subscales and an overall score below the cut-off for depression (26%).

The subtypes indicate that many women had both anxiety and depression symptoms, and that this comorbidity is associated with a higher overall score.

When looking at time of onset for the different subtypes, type 1 and 2 were more likely to have onset either in the first trimester or more than 8 weeks after the birth; type 3 was more likely to have onset in the first 8 weeks after the birth; type 4 was quite evenly distributed across the time periods.

Conclusions

The authors describe the results as: “an important hypothesis-generating foundation for future work”, rightly indicating that the results need to be replicated in other samples and with more robust measures.

Given some of the differences in time of onset, they suggest that: “future clinical and research efforts should focus on the potential phenomenological and biological differences characterising onset of depression during pregnancy versus the postpartum period”.

They also mention possible clinical implications: “…development of effective screening strategies that allow for the delivery of targeted therapies to women with different clinical phenotypes and severity of perinatal depression is imperative”.

Strengths and limitations

The authors identify several limitations in the research.

Missing data. Only 663 out of 17,912 cases (3.7%) across the 19 sites met inclusion criteria, mainly because they did not have data on time of onset of symptoms. Within this sample, many women did not have data on demographic characteristics and so, although some data is presented about the characteristics of women in the different subtypes, this is sometimes based on very small numbers. This is problematic as there is likely to be bias in the results and 663 is a small sample to identify subtypes which are intended to be clinically relevant.

Most women (82%) were white. The lack of diversity in the sample makes any conclusions very difficult, as women from different backgrounds may express distress in very different ways and show different patterns of symptoms.

The one-off postpartum EPDS measure. The authors report a median of 4.5 months postpartum as the time when women completed this, but there is very wide variation, with many women completing the EPDS less than 2 months after giving birth and others completing it over 7 months postnatally. The enormous changes that take place across the perinatal period mean that mood and symptoms can fluctuate over time, particularly postnatally. Additionally, a 10-item self-report measure can’t capture the full range of experiences that women have. Can we really draw conclusions about subtypes of PND based on a one-off EPDS assessment at such varied time points?

The lack of data on other psychosocial variables. While there is some data on variables such as marital status, education and obstetric complications, there are many missing values and discussion of these variables is left to the supplementary material. Factors such the quality of the partner relationship or housing instability, which would be very hard to measure in this type of sample, have a huge impact on women’s mood and their ability to recover from episodes of mental illness. It might be that the more severe, long term subtypes of PND are present in those women who are experiencing significant social stressors such as partner violence or facing eviction. This then isn’t so much a subtype of depression as an understandable response to very stressful circumstances, which may remit quickly if those circumstances change. It would be hard to capture this kind of complexity in data aggregated across so many sites, and yet these factors are extremely important when making decisions about the support that women need to recover from PND.

The unknown depression status of women pre-pregnancy. It’s likely that some women in the sample had ongoing depression which began prior to pregnancy, but this is not captured in the data. This may be a very different experience of depression compared to those women whose onset is during the perinatal period.

Comments

While identifying subtypes of PND is an interesting question and the concept seems to fit with clinical experience, I believe the limitations above mean it’s not possible to draw conclusions from this research. A study of this type, where data is aggregated across many different sites and there is little data about historical or background variables, means that some of the key factors contributing to the onset and maintenance of mental illness do not contribute to the analysis. Reducing symptoms of distress without context runs the risk of increasing the stigma and burden of women who are need of support.

What this study highlights is that all women presenting with distress during the perinatal period should be assessed and asked about a range of factors before any treatment decisions are made. Ideally this assessment would include a psychosocial history alongside questions about current circumstances and symptoms, backed up with data from validated measures. This remains the best way to ensure that women receive appropriate treatments that match their needs.

Links

Primary paper

Putman K et al. (2017) Clinical phenotypes of perinatal depression and time of symptom onset: analysis of data from an International Consortium. Lancet Psychiatry 4(6) 477-485.  http://dx.doi.org/10.1016/S2215-0366(17)30136-0

Other references

Fairbrother N, Janssen P, Antony M, Tucker E, Young A. (2016) Perinatal anxiety disorder prevalence and incidence. Journal of Affective Disorders. 200, 148-155. https://doi.org/10.1016/j.jad.2015.12.082

Postpartum Depression: Action Towards Causes and Treatment (PACT) Consortium (2015). Heterogeneity of postpartum depression: a latent class analysis. Lancet Psychiatry 2: 59-67  http://dx.doi.org/10.1016/S2215-0366(14)00055-8

Stein, Pearson, Goodman, Rapa, Rahman, McCallum, Pariente C. (2014) Effects of perinatal mental health disorders on the fetus and child. Lancet Psychiatry 384(9956), 1800-1819. http://dx.doi.org/10.1016/S0140-6736(14)61277-0