Postnatal depression: mapping the territory and revealing styles

Postnatal depression (PND) has been described and defined in many ways over the past 200 years (Gruen,1993), with both convergent and divergent accounts of incidence, characteristic symptoms and features, onset and course, as well as aetiology reported. The lack of any clear and comprehensive definition has influenced and been affected by methods of assessment and quantification, thus making comparative study and evaluation problematic, with consequential roll-over effects into identification and intervention.

The current study was designed to integrate the varying descriptive features of postnatal depression into a coherent, empirically supported framework or 'map' as a base from which to understand the ways in which both the noted similarities and differences in experience relate to the overall concept.

The literature reviewed as a base for this research incorporated readings from academic, research, clinical, and popular accounts of postnatal depression, so as to include as many perspectives as possible into the framework.

Current definitions in the literature frequently focus on a continuum of postnatal mood disorders, from the mild "blues" through varying forms of severity to post-partum depression with psychotic features (Affonso et al. 1990; Nonacs,1998; Pritchard & Harris, 1996). For the purposes of the current study, the terms "postpartum depression" and "postnatal depression" both relate to the one descriptor, abbreviated to "PND" for ease of reading, and defined here as: a depression characterised by significant disturbance in mood, cognition, and behaviour, occurring any time from 2 weeks to a year postpartum and lasting up to 2 years, of sufficient magnitude to consistently impact on a mother's enjoyment of, and interest and participation in the tasks related to her life roles.

There has been much debate about the specificity of postnatal depression, with a range of contradictory evidence and opinion supporting both its unique character (Affonso et al. 1990; Pitt, 1968; Cohen, 1998) and homogeneity with depression occurring at other times (Murray 1991; Nonacs, 1998). The DSM-IV (American Psychiatric Association, 1994) says that PND has both typical, and atypical features such as anxiety and variable mood, but is not essentially different to other major depressions. Paykel (2000) asserts that the whole range of postpartum disorders is largely "ignored" ' as a sub-group both by the DSM-IV, and ICD-10 diagnostic classification systems, and proposes increased recognition of PND in particular via the incorporation of an obligatory post-partum specifier, a broader timeframe for onset, and a category for subthreshold depression common in the postpartum period. Commonly cited co-morbid features include anxiety and panic disorder (DSM-IV), obsessive-compulsive disorder, and phobias (Cox, 1986). Other atypical features include 'masking'/outward "normality" (Pritchard & Harris 1996) as well as irritability and anger (Affonso et al., 1990).

Of interest is the trend towards continuing differentiation of PND into distinct 'subtypes', linked to special features in onset, duration, and severity (Ballard, Davis et al., 1994; Cooper & Murray, 1995; Affonso et al., 1990) and symptom profile (Campbell et al., 1992), frequently on the basis of proposed aetiology. Examples of this include Kumar's (1997) co-morbid maternal bonding disorder; Cooper and Murray (1995) and Harris' (1994) biological/hormonal sub-group; Fisher et al. (1997) and Ballard, Stanley, and Brockington's (1995) post-traumatic sub-group; and the agitated melancholic depression noted by Boyce et al (2000). Rosenbloom, Mazet, and Benony (1997) identified two types of PND related to infant attachment style: one "dull and slow" and the other "stressed and irritable". Also noted in the literature is debate about the normal (when contextualised) versus pathological view of PND (Nicholson, 1990; Campbell et al., 1992), linked to the sociocultural genesis proposed by authors such as Oakley (1980, cited by Elliot et al., 1988), and the bereavement and loss model proposed by Nicholson (1990).

Although Cox et al. (1987) designed a screening tool specifically to detect PND (the Edinburgh Postnatal Depression Scale, or EPDS), researchers more frequently modify existing assessment instruments to accommodate the noted variations in presentation, especially the prominence of anxiety, and potential confounds with the somatic symptoms which are a common feature of the puerperium (Affonso et al., 1990; Okun et al., 1996; Boyce et al. 1993). General rating scales such as the Beck Depression Inventory (BDI) are frequently criticised because of these potential confounds (Harris et al., 1989, I) when used to assess PND.

In addition to capturing the distinctive features of symptom profile compared to depressions occurring at other times, authors such as O'Hara (1995) emphasise the need to differentiate between depressive symptomatology (such as negative cognitions) and clinical depression in relation to PND. This necessitates the careful evaluation of both symptom profiles and their relationship to the broader concept or syndrome noted by Rosenblum et al. (1997), and Snaith (1993).

One study went beyond elementary symptom/syndrome description and sought to identify a more experiential "underlying structure" of PND. Beck (1992) extracted 11 core themes from accounts of women's 'lived experience', which she believed should be incorporated into any measures which seek to quantify PND. Although her study was limited by a time consuming and "dense" interviewing/data collection method that severely limited sample size, it yielded rich qualitative data which she has since incorporated into a comprehensive new screening scale (the Postpartum Depression Screening Scale, Beck, 2000).

The present study extended this line of research. Its overall objective was to examine the relationships among the diverse features associated with PND, and to identify any distinct constellations of these features. This involved three phases, with separate sub-goals:

1. showing how a set of PND "items" may be mapped

2. identifying psychological points of reference within this map

3. demonstrating how aggregated profiles may be grouped into styles

PHASE 1: MAPPING

Method

Participants. Thirty-two adult, opportunistically selected, volunteers participated in the initial sorting task.

Stimuli. The same 73 item-statements, each printed onto a 35mm x 75mm card, were used in both phases (see Appendix 1). Each item-statement described a characteristic feature of PND, identified from a comprehensive review of various sources, including academic, research, clinical, and popular literature, and women's personal accounts. A number of these items were adapted from pre-existing scales and/or categorisation systems including: Cox et al's (1987) EPDS, the DSM-IV (APA, 1994), Montgomery and Asberg's (1979) Maternal Anxiety & Depression Rating Scale (MADRS), the Beck Depression Inventory (BDI) (Beck et al., 1960), and the Hamilton Rating Scale (Hamilton 1960), Affonso et al.'s (1990) eight primary symptoms, Beck's (1992) core phenomenological "themes", and Snaith's (1993) "master-list", drawn from his comprehensive review of depression rating scales.

Procedure

Each participant engaged in four separate sorting activities, summarised by the acronym GOPA (Bimler & Kirkland, 2001a). In brief, individual participants each shuffled the entire 73-item deck and arranged them into clusters of items that seemed to "belong together", i.e. were similar in content, based on their face-values. This is the G (grouping) task. Typically between 10 - 16 similarity clusters were formed, with single-item clusters acceptable. Next, participants were requested to identify clusters that appeared to be "opposites" (the O-task). Thirdly, each original cluster was partitioned (P-task) into sub-groups, each participant then applying a higher threshold of "similarity". Finally, the original clusters were merged into larger ones by relaxing the similarity threshold (additive or A-task). In summary, each person created a hierarchical tree structure to reflect his or her view of how closely the different items and groups of items were related to one another. A full sort took about an hour to complete.

Analyses

A custom-designed suite of multidimensional scaling (MDS) algorithms which have been documented elsewhere (Bimler & Kirkland, 2001b) was applied to these data.

Results

One outcome from this 'mapping' phase portrayed the similarities among items as a dendrogram, which is an hierarchical taxonomy showing groups of conceptually-related items and the ways in which they successively integrate. Essentially it is a consensus across the individual, incomplete hierarchical trees provided by the informants. Although caution should be exercised when interpreting a dendrogram (since it is a one-dimensional representation of multi-dimensional space) the major sub-groups of items fell into the following themes: "classical" depression; cognitive clouding; false-forced manner; anxiety - from worry through to panic; trauma (intrusive thoughts/conflict); loss of "self"; guilty inadequacy; despair; loss of sense of control; anger; not feeling valued/cared for; and social/interpersonal disconnection.

A second outcome was a spatial model or "map" in which points represented the items, to apply a geographical analogy. Items perceived as similar (frequently grouped together) were shown as neighbouring points. A three-dimensional model appeared to be optimal.

Discussion

From other research (e.g. Bimler & Kirkland, 2001b) it is known that 30 complete sorts are sufficient to fix items into a stable structure and additional data make little difference to this configuration. This inter-item map displayed some interesting features, such as the consensual clustering of many items into recognisably familiar sub-sets, or "themes" as mentioned above.

The map is like a backdrop for interpreting subjective data. In the next phase we document how these subjective data were collected and reflected within this map.

PHASE 2: POINTS OF REFERENCE

Method

Participants. Seventy-six participants were drawn from throughout New Zealand. All self-identified as having had experience of PND within the last 2 years. These volunteers were recruited via a letter placed in a magazine for mothers, and via the Wellington Post and Ante-Natal Distress Support Group network. Each respondent was given detailed verbal and written information about the nature and purpose of the research, their rights within it, and the procedures involved prior to completing the sorting task.

Procedure

Participants completed the Method Of Successive Sorting (MOSS) task either face-to-face with the senior author or independently, returning completed forms by mail. In this procedure items were ranked according to their applicability to a "target" person, in this case how closely each one described the participant's personal experience of PND. Items from a shuffled item-deck were first allocated into one of two piles, then reallocated twice more into successive sub-groups, according to variations on: "tends to apply to my situation" versus "tends not to apply to my situation". The outcome was that each person created a personally meaningful set of eight rank-ordered piles with no constraints about the number of items placed into each one, unlike the usual Q-sort approach. Each pile was assigned a corresponding rank-value, ranging from +3.5 (most-applicable) through to -3.5 (least applicable), and this value was assigned to all the items within that pile. A line-drawn template showing the relationships between levels and piles was provided as a guide to record participants' decision-making. This procedure was readily understood and took about 20 minutes to complete.

Analysis. The three-dimensional model obtained earlier provided a tool to summarise the rank-values assigned to items, in each participant's description of her experience of PND. The 73 points representing items within the model were augmented by a small number (F) of abstract 'hot-spot' points, each corresponding to a broad category or theme. The sum of the rank-values, weighted by the items' proximity to a given hot-spot, yielded an overall score on that theme. The weights have maximal value for items closest to the hot-spot (i.e. most similar to it, capturing the theme's core meaning), and fall off towards zero with increasing item / hot-spot distance (i.e. greater dissimilarity). By repeating the weighted summation for each hotspot, the 73 pileplacement values comprising a description of PND were reduced to F aggregate scores, easily presented as a profile. In practice this is performed by a spreadsheet, containing F columns (hotspots) and 73 rows (items).

Results

In the present study 10 hot-spots were identified. Note that the locations of the hot-spots are not arbitrary. They are positioned, via an iterative computation (Bimler & Kirkland, 2001b), in areas of the spatial model where the range of rank-values assigned to items is greatest, so that the resulting summaries capture as much as possible of the variation among informants. Unlike the item clusters described in the dendrogram (which were "face-value" rational groups) the hot-spots are nodes of psychological meaning - they represent what is personally meaningful within an interpretative map.

The brief summary provided for each hot-spot below gives a sense of their general "flavour" with the most important (top three) items suggesting sub-titles. Common themes derived from the first 10 items were considered by the senior author in an effort to capture the complex semantic and thematic "flavour" of each hotspot profile. The following list sequence is arbitrary.

A. Core depression (Not Coping/Exhausted/Dysphoric): equates to classic primary symptoms of depression (American Psychiatric Association, 1994). This forms a core constellation that features in all of the 8 subsequently identified "styles".

B. Emotionally overwhelmed (Overwhelmed/Trapped/Deconstructed): Characterised by negative phenomenological / emotional statements - a sense of falling apart, powerlessness, and incompetence, and feeling negative, moody, and pessimistic.

C. Isolative (Avoidant/Dissatisfied/Alone): Characterised by a sense of social isolation and disconnection, items in this cluster describe avoidance of intimacy, and a sense of alienation, preoccupation and diminished self-confidence.

D. False facade (Forced/Robotic/Preoccupied): Characterised by items consonant with dissociation/depersonalisation e.g. a falsely bright manner, feeling robotic, preoccupied, confused, forgetful, and indecisive.

E. Cognitive confusion (Disorganised/Distracted/Indecisive): Includes difficulty planning and organising, concentrating, making decisions, thinking and remembering. This set also has strong elements of psychomotor retardation, whereas "D" is somewhat more "agitated".

F. Psychic and somatic anxiety (Anxious/Panicky/Worried): Characterised by features on the spectrum from generalised/pervasive anxiety and worry through to panic attacks and specific phobic responses.

G. Negative thoughts (Guilty/Hopeless/Grieving): This constellation has a theme of "negative thoughts" related to a sense of loss, blame, hopelessness, pessimism, and anger, unresolved "issues", and perhaps not surprisingly, incorporates thoughts of self-harm.

H. Unsupported (Unsupported Practically/Emotionally/Uncared For): This set reflects the instrumental and emotional components of social support noted by others as a key associated factor of PND (O'Hara et al., 1983), and also includes social withdrawal and isolation, whether as cause or consequence.

I. Controlled (Compelled/Propelled/Rigid): This set of items has a theme of both over- and undercontrol, including obsessive-compulsive thoughts and actions, feeling sped-up and guarded, being troubled by unwanted thoughts, dreams, images, and unable to discuss these with anybody, which overlap with symptoms describing post-traumatic responses (Fisher et al., 1997). It contains one item "sleeps much more than usual" (6.), which at first glance contradicts the psychomotor agitation described by other items in the set. Anecdotal evidence however, reveals that for some women at least, the intensity of waking hours does co-exist with exhaustion, and sleeping more as an escape/coping mechanism or consequence of this.

J. Harmed/Traumatised (Self-harming/troubled/traumatised): This set shares 6 items with the previous "hotspot" (I). It is characterised by thoughts of harm/trauma, self-harming actions, thoughts, and plans, intrusive thoughts and sense of trauma associated with labour and delivery and other fears/anxieties manifested by obsessive-compulsive features of experience. It may be that the overlapping although relatively inverted item-set represented in "I" and "J" reflect differing aetiology, as with "D" and "E".

Discussion

Overall, these points of reference do relate to studies and classification systems describing symptom profiles, sub-scales, and co-morbid features outlined previously, although as with the similarity data, are more complex. Of note, is that although they capture similar themes, they also incorporate a broader phenomenological viewpoint, highlighting the ways in which more clearly defined taxonomic groupings (eg. those formed by similarity data in this study) may fail to adequately capture the complexity of lived experience.

PHASE 3: PROFILES TO STYLES

Method

Data. The 73 profiles from phase 2, each summarising an individual's MOSS description of her own experience of PND, were used once more.

Analysis. The hot-spot profiles were arranged into clusters, using k-means analysis. Essentially this allots profiles to a specified number of clusters so as to maximise the distances between the clusters' average or centroid profiles. The results can be interpreted as distinct 'styles'.

Results

Various cluster solutions were examined for a range of numbers of clusters. On the basis of parsimony, common sense, and optimal utility, it was decided that eight distinct styles emerged - fewer clusters were too coarse and additional refinements too fused. These eight styles are summarised as follows:

Style 1: 13% of participants generated this style, featuring a core of depressive symptoms (hot-spot A), feelings of being overwhelmed and trapped (B), the "anxious" subset (F), and cognitive confusion (E) in descending order, with some elements of the 'G' (negative thoughts) constellation at moderate to low levels from 4 of 10 respondents.

Style 2: was generated by data from 8% of participants, characterised by a close combination of depressive and cognitive symptoms, with a lesser presence of the social isolation/disconnection embodied in hotspot "C". This is the only style which did not incorporate the 'B' (emotionally overwhelmed) constellation, aside from 2 of 6 respondents, perhaps representing less emotional "chaos"/energy. It also includes moderate to low "D" (false facade) and "F" (anxious).

Style 3: is a "smorgasbord" of 6 of the 7 hotspots which emerged as significant for 13% of participants - depressive symptoms, social isolation, and feeling overwhelmed, a close but lesser contribution of depersonalisation (false/forced) and cognitive symptoms, and a lack of social support.

Style 4: features 4 hotspots, characterised by core primary depressive symptoms, feeling overwhelmed, socially isolative, and also lacking in social support, with 8% of participants.

Style 5: in addition to being the most common (18%), this style is also relatively simple - there is a very close association of core symptoms of depression (A) and feeling overwhelmed (B), with lesser contribution of feeling socially isolated (C). It is similar in structure to style 2, but with B interchanged with E, the cognitive set.

Style 6: the same hotspots as style 1, but with primacy of the A set (depression), plus cognitive clouding (E), with the sense of being overwhelmed and anxious present to a lesser degree.

Style 7: primacy of core depressive symptoms, with lesser contribution from overwhelmed, isolative disconnection and cognitive clouding closely connected, and "false-forced" (D) set less so. Similar to "3", but without H, the "unsupported" group, and with B more dominant than C, 13% of participant profiles composed this group.

Style 8: represents the second largest constellation, with 15% of participants. It is similar to the most common constellation, Style 5, but with different weightings related to the primacy of A and B, and much less influence of C (isolative). The primary difference is the inclusion of F, the anxiety set.

Discussion

The distinctions between each style were less clear than expected, with several overlapping features between them. Seven of the previously identified hot-spots featured as positively represented in the identified styles. (As "positively represented" or "significant" we mean contributing to the styles in a meaningful way.) Hotspots A and B, together incorporating the classic core symptoms of depression, were the most frequently occurring, featuring in all but one style set (2). The isolative symptoms embodied in "C" also appear frequently in 6 of the 8 styles, and E (cognitive) in 5.

Of note is the confirmation of the frequency of anxiety (F) when viewing individual profiles, which, although only significant in 3 styles, appears in at least moderate-low levels in a further two.

Although G, the "guilty" subset, does put in an appearance at moderate to low levels in Styles 1 and 8, and low to absent in 4 and 5, hotspots G, I, and J did not emerge as 'positively significant' (that is, signalled above the mid-point) in any of the subsequently identified "styles" in this study. This is at odds with clinical and anecdotal accounts of women's experience, and may be an artifact of this particular sample, or an indication that the wording of the items related to these categories needs to be reconsidered, to see if they clearly capture the concepts intended. For J, it may be that although suicidal ideation is a feature of some women's experience, it is either under-reported or tends to take the form of wishing it was not so/wishing self dead (passive) rather than a more active intent to die. Clinical experience confirms that for those women who do consider suicide as a way out, mothers of young children will often discount this option because of the baby - and this in itself for some compounds the sense of lack of control/freedom within their lives - so it may be that different expressions of suicidal inclinations/wishes to remove self from life are present e.g. shutting down/off emotionally, cognitively, socially, physically, and functionally. Coble et al. (1994) in a study of psychiatric symptomatology in women during pregnancy and early puerperium, similarly found that thoughts of death/suicide amongst this group are rarely/never represented symptoms.

Psychotic symptoms were not included in the item set, although 4 respondents did self-identify as having a diagnosis of puerperal psychosis, and 1 with bipolar affective disorder. A further 5 respondents identified as having an anxiety disorder. These data sets were identified individually to ascertain if any significant differences in patterns of depressive symptomatology were present, but each was scattered through various 'style' groupings, with no unique features detected amongst these groups.

GENERAL DISCUSSION

Kumar and Robson (1984) suggest self-report questionnaires, the presence or absence of selected signs and symptoms of depression, and the necessity of an outpatient referral or prescription of psychotropic medication are all valid indicators of the presence of PND for research purposes. Other authors have found a significant correlation between self-report and other quantifiable indicators of depression (Small et al., 1994; Affonso & Domino, 1984; Richters, 1992). In the present study, the MOSS sorts themselves yielded information about symptom profile, and cover-sheet data elicited further details/corroborative circumstantial evidence of participants' experience of PND, including diagnosis and intervention. Of the 76 participants, only 6 reported they received neither diagnosis nor formal intervention (e.g. medication, counselling, support group attendance) for PND. Self-report was thus considered a valid indicator of the presence of PND for the purposes of the study.

Much of the face and construct validity of the proposed study rested on the selected items being sufficiently representative of the whole "territory" to create a legitimate map. Data from the study in the form of identified essential items constituting the "territory" and various subgroupings within it (similarity, hotspots, styles) were validated by concordance with documented literature, clinical and phenomenological experience, and research.

In terms of the initial construction of the map, although not essential for validity, 12 GOPA participants were mental health professionals/others with professional experience of PND. Many were thus familiar with standard categorisation systems related to diagnosis and classification, so it is likely that this influenced sorts, although there were also other items which did not fit these standard classification systems. Regardless of experience, most participants similarly took great care to get it "right" - with some lengthy explanations to the senior author about the rationale for their sorting decisions.

The item set could benefit from further revision to improve the specificity of some items, and eliminate others that are essentially repetitive. Difficulties may have arisen in the wording of items, and as a consequence of trying to incorporate too many different features of a concept into the one item. Participants were asked about "missing" items from the current set in a cover sheet that accompanied MOSS instructions. Several comments exemplified feelings of anger/hostility towards the baby and other children (e.g. "wishing the baby would go away so I could get better"; "feel as if I really hate my child at times"; "feel as if I'm about to hurt the baby") even though the researchers believed this had been encapsulated in item 60. Although commonly noted anecdotally as well as in the literature (Affonso et al., 1990; Harris et al. 1989, II) as a "hallmark" of postnatal depression, the item statement relating to anger (60.) only appears as 7th on the list within an otherwise less important hotspot (G). The related concept of irritability did emerge as contributing to hotspots A and B, the latter of which also captured other related elements (e.g. 61. moodiness, and 63. negativity and pessimism), but not as a central feature of their scores. Subsequent refinements should thus include more specific examples of anger/stress/irritability both in relation to the baby and within the self (e.g. decreased noise tolerance).

This study was relatively free of problematic ethical issues, although care was taken to adhere to basic principles throughout. Of note is the principle about not doing harm - although the procedure itself may not appear overtly harmful, the item-cards describe aspects of distress that have been very real for all participants. Several participants commented on how accurately these items described and summarised their own experiences. Two participants were unable to complete the task because it highlighted the extent of their as-yet unresolved depression. The senior author made contact with them to ensure they were linked in with some form of personal support. Other recipients commented that they found the task therapeutic, primarily because it reinforced progress, and many wrote that they felt good about being able to contribute in some way to the advancement of knowledge that may benefit other women. The main point to bear in mind is that operationalising a psychological construct such as depression in this way makes it real - and the more accurately an item-set "captures" the construct, the more potential power it has to do both good and harm for participants.

Many incidental inquiries were prompted by the initial letter requesting participants - both requests for and supply of information about resources, NZ-wide groups and initiatives, and issues of interest. Overall, the project felt like "good" research - interactive, and of benefit to both researcher and participants.

No external objective or succinct measure of PND, or corroborative data from sources other than the mother were sought. One participant wrote back to say her husband had completed a MOSS with her as the target, and placed the same items into the same piles as her own self-assessment. Further studies may incorporate corroboration of self-report via such parallel sorts, or measures such as the EPDS.

There are acknowledged limitations in cultural specificity of the current study, especially related to ethnicity and socio-economic diversity, so no claims are made for generalisation beyond the scope of this sample. Of the 76 participants, all but two self-identified as European. In addition, it is likely that both the magazine and Postnatal Depression Support Network used to recruit MOSS participants only sampled a relatively limited section of childbearing women.

Although not yet fully explored, there is likely to be some variability of cultural experience and manifestation of PND. Culbertson (1997), in an international review on depressive disorders and gender relationships for WHO, asserts that symptom similarities for depression across nations appear to be fairly reliable, but in New Zealand although there is as yet little information on variations in cultural expressions of PND, slightly different presentation and an increasing incidence of PND have been noted in the Samoan community (National Health Committee of New Zealand, 1996).

CONCLUSION

Affonso et al. (1990) criticise the failure of both traditional diagnostic criteria and assessment methods to take account of the unique features of the puerperium, and hence the symptoms and complex nature of PND. Effective identification, assessment, and intervention require a comprehensive, consistent, and inclusive framework, which acknowledges both reported core features and variations in experience, and which enables both individual symptom profiles and constellations of these to be viewed in relation to each other.

Although the current study was able to "map" these various features of PND, and identify patterns/"styles" within it, it has also highlighted the complexity which occurs when transferring conceptual frameworks into the realm of lived experience. Even though the "hotspots" fell into relatively clear categories which match current views of classification of symptoms, the lack of clear categorisation and overlap evident in the "style" patterns identified by participants may at least partly account for the documented difficulties to date in clearly defining, evaluating, and accounting for PND.

There are many possibilities for future work in relation to this study and findings. The item set has potential value as a tool to more clearly describe both individual and group profiles, and thus contribute to more comprehensive detection, and the design of intervention programmes tailored to individual needs and circumstance. Now that a weighted-spreadsheet is available any individual's new rank-pile data may be entered from these same item-cards and a personal profile generated. There is also work to be done linking "styles" to other variables such as risk and associated factors, proposed aetiology and sub-groups, or other related constructs such as parenting sensitivity (Nichols & Kirkland, 1996), infant temperament, attachment styles and cultural variations in experience.

Appendix I: PND Study Item List

1. feels extreme and persisting tiredness/lack of energy/exhaustion

2. feels unusually suspicious/mistrustful of or persecuted by others

3. doesn't enjoy/laugh at/look forward to/feel pleasure or enthusiasm for things she used to

4. doesn't feel interested in or motivated to do usual activities

5. has difficulty getting off to sleep/wakes early due to own thoughts and worries rather than the baby's patterns of behaviour

6. sleeps much more than usual

7. has nightmares

8. often has physical symptoms of anxiety (e.g. sweaty palms / upset stomach / dizziness / rapid heartbeat), with no apparent physical cause

9. feels anxious for no apparent reason

10. worries excessively about things which would not normally bother her

11. feels afraid/insecure/overly sensitive and vulnerable

12. is consumed by feelings of guilt/shame/self-blame

13. feels unusually afraid/panicky about specific situations e.g. being alone, going outside, mixing with people

14. has difficulty concentrating on familiar everyday tasks/activities such as following a conversation/reading more than a few lines at a time/following instructions

15. feels criticised/negatively evaluated by significant others

16. finds it hard to make everyday decisions

17. thinking feels confused/"hazy"/unclear

18. has difficulty remembering in the course of normal daily tasks/activities e.g. things to be done that day or week, where the car was parked

19. often feels afraid that some harm will come to the baby

20. has a lot of worries about the baby's health and well-being which aren't relieved by reassurance from others

21. is troubled by intrusive and unwelcome thoughts and images of the birth

22. can't seem to get going in the morning

23. has a negative and overly critical view of self as a mother

24. feels sad/low/depressed most of the time

25. feels dull/numb/detached/empty

26. feels tearful/cries more easily than usual, or for no reason

27. has lost hope for the future

28. has anxiety/panic attacks

29. doesn't feel satisfied/fulfilled

30. has lost self-confidence

31. is socially withdrawn/isolates self

32. feels alone/socially isolated

33. doesn't feel valued/significant

34. feels powerless and out of control/controlled by circumstances

35. feels alienated/disconnected from others/as if no-one else understands

36. thinks about/plans self-harm, suicide attempts/death

37. has acted on self-harming thoughts and plans

38. doesn't feel like her 'real' self anymore/feels she has 'lost' herself

39. feels frustrated/restricted/trapped

40. feels unsupported emotionally (no 'shoulder to cry on')

41. feels unsupported practically e.g. assistance with child care and household tasks

42. felt traumatised by labour and delivery (out of control/shocked/robbed of a vital experience/tortured)

43. feels a sense of loss/failure related to pregnancy and birth/the baby/being a mother

44. feels overwhelmed by the responsibilities/demands/tasks/consequences of motherhood and baby care

45. has a lot of unexplained low-grade physical complaints/physical ill health

46. looks 'normal' and 'together' on the outside, but it feels false/as if this is a 'mask'

47. feels physically and emotionally rigid and over-controlled

48. doesn't feel adequately cared for/looked after

49. feels as if her world has been turned upside down/is falling apart

50. feels as if she is in a black hole/dark tunnel with no escape

51. feels as if old/unresolved emotional conflicts/issues have resurfaced

52. feels well supported and/or has a 'good' baby - as if she has nothing to be depressed about - but still feels awful

53. doesn't feel able to relate normally to people she is usually close to/significant relationships have deteriorated

54. feels pre-occupied with self and present circumstances

55. can't see past the present (day-to-day is as much as she can manage)

56. has appetite and/or weight changes greater/more persistent than would normally be expected

57. avoids intimacy with partner (e.g. marked decrease in sexual interest/responsiveness unrelated to normal postnatal changes) and which is persistent/prolonged

58. feels uptight/restless/nervous/unable to rest and relax

59. feels extremely irritable

60. feels angry/resentful/hostile/aggressive towards others (partner/the baby/family/friends)

61. feels moody/subject to mood changes with no apparent trigger

62. thoughts and actions feel slowed down

63. thinking is negative and pessimistic (e.g. about people, events, the future)

64. is troubled by 'odd'/inappropriate/disturbing thoughts/fantasies

65. feels compelled to perform certain actions (e.g. cleaning, checking, washing hands, particular sequences)

66. doesn't feel she is coping as well as she should be able to/used to be able to

67. feels inadequate/a failure/incompetent/unable to cope with the demands of child care/household/usual tasks and roles

68. is unable to talk to anyone about how she feels

69. is afraid she is going crazy (can't comprehend/make sense of what is happening)

70. feels on a 'high'/sped up with thoughts racing

71. has difficulty planning and organising usual daily activities

72. actions feels automatic/'robotic'

73. has a forced/falsely bright manner

REFERENCES

Affonso, D., Domino, G. (1984). Postpartum depression: a review. Birth, 11, 231- 235.

Affonso, D., Lovett, S., Paul, S., Sheptak, S. (1990). A standardised interview that differentiates pregnancy and postpartum symptoms from perinatal clinical depression. Birth, 17, 121-130.

American Psychiatric Association. (1994). Diagnostic and statistical manual of mental disorders 4th edn., Washington, DC: Author.

Ballard, C. G., Davis, R., Cullen, P. C., Mohan, R. N., Dean C. (1994). Prevalence of postnatal psychiatric morbidity in mothers and fathers. British Journal of Psychiatry, 164, 782-788.

Ballard, C. G., Stanley, A. K., Brockington, I. F. (1995). Post-traumatic stress disorder (PTSD) after childbirth. British Journal of Psychiatry, 166, 525-528.

Beck, A. T., Ward, C. H., Mendelson, M., Mock, J., Erbaugh, J. (1960) An inventory for measuring depression. Archives of General Psychiatry, 4, 561-571.

Beck, C. T. (1992). The lived experience of postpartum depression: a phenomenological study. Nursing Research, 41, 166-70.

Beck, C. T. (2000). The Postpartum Depression Screening Scale. Paper presented at the Marce Society Biennial Conference, September 6, Manchester, U.K.

Bimler, D., Kirkland, J. (2001a). Fables of the reconstruction: Characterizing business structures with online data collection. In John D. Haynes (Ed.), Internet Management Issues: A Global Perspective, Chapter 16. London: Idea Publishing.

Bimler, D., Kirkland, J. (2001b). School truants and truancy motivation sorted out with multidimensional scaling. Journal of Adolescent Research, 16, 75-102.

Boyce, P., Stubbs, J., Todd, A. (1993). The Edinburgh Postnatal Depression Scale: Validation for an Australian sample. Australian and New Zealand Journal of Psychiatry, 27, 472-476.

Boyce, P., Stanek, S., Gilchrist, C. (2000) Postpartum melancholia: clinical descriptions of a distinct type of postnatal depression. Poster presented at the Marce Society Conference, September, Manchester, England.

Campbell, S., Cohn, J. F., Flanagan, C., Popper, S., Meyers, T. (1992). Course and correlates of postpartum depression during the transition to parenthood. Development & Psychopathology, 4, 29-47.

Coble, P. A., Reynolds, C. F., Kupfer, D. J., Houck, P. R., Day, N. L., Giles, D. E. (1994). Childbearing in women with and without a history of affective disorder. I. Psychiatric symptomatology. Comprehensive Psychiatry, 35, 205-214.

Cohen, L. (1998). Heterogeneity of postpartum depressive illness: implications for treatment. Paper presented at the Marce Society Conference, September 6, London, England.

Cooper, P. J., Murray, L. (1995). Course and recurrence of postnatal depression: evidence for the specificity of the diagnostic concept. British Journal of Psychiatry, 166, 191-195.

Cox, J. L. (1986). Postnatal depression: a guide for health professionals. Edinburgh: Churchill Livingstone.

Cox, J. L., Holden, J. M., Sagovsky R. (1987). Detection of postnatal depression: development of the 10-item Edinburgh Postnatal Depression Scale. British Journal of Psychiatry, 150, 782-786.

Culbertson, F. M. (1997). Depression and gender - an international review. American Psychologist, 52, 25-31.

Elliot, S. A., Sanjack, M., Leverton, T. J. (1988). Parents groups in pregnancy: A preventive intervention for postnatal depression? In B. H. Gottlieb (Ed.) Marshaling Social Support - Formats, Processes, and Effects, Chapter 3. Newbury Park, Ca: Sage.

Fisher, J., Astbury, J., Smith, A. (1997). Adverse psychological impact of operative obstetric interventions: a prospective longitudinal study. Australian and New Zealand Journal of Psychiatry, 31, 728-738.

Gruen, D. S. (1993). A group psychotherapy approach to postpartum depression. International Journal of Group Psychotherapy, 43, 191-203.

Hamilton Depression Rating Scale 1960. (1966). In National Health Committee Guidelines for the treatment and management of depression by primary healthcare professionals, Appendix III. Wellington: National Advisory Committee on Health and Disability.

Harris, B. (1994). Biological and hormonal aspects of postpartum depressed mood. Working towards strategies for prophylaxis and treatment. British Journal of Psychiatry, 164, 288-292.

Harris, B., Huckle, B., Thomas, R., Johns, S., Fung, H. (1989). The use of rating scales to identify postnatal depression. British Journal of Psychiatry, 154, 813-817.

Harris, B., Johns, S., Fung, H., Thomas, R., Walker, R., Read, G., Riad-Fahmy, D. (II). (1989). The hormonal environment of postnatal depression. British Journal of Psychiatry, 154, 660-667.

Kumar, R., Robson, R. M. (1984). A prospective study of emotional disorders in childbering women. British Journal of Psychiatry, 144, 35-47.

Kumar, R. C. (1997). "Anybody's child": severe disorders of mother-infant bonding. British Journal of Psychiatry, 171, 175-181.

Montgomery, S. A., Asberg, M. (1979). A new depression scale designed to be sensitive to change. British Journal of Psychiatry, 134, 382-389.

Murray, L. (1991). Intersubjectivity, object relations theory, and empirical evidence from mother-infant interactions. Infant Mental Health Journal, 12, 219-232.

National Health Committee Guidelines for the treatment and management of depression by primary healthcare professionals. (1996). Wellington: National Advisory Committee on Health and Disability.

Nicholls, A. & Kirkland, J. (1996). Maternal sensitivity: A review of attachment literature definitions. Early Child Development and Care, 120, 55-65.

Nicholson, P. (1990). Understanding postnatal depression: A mother-centred approach. Journal of Advanced Nursing, 15, 689-695.

Nonacs, R., Cohen, L. S. (1998). Postpartum mood disorders: Diagnosis and treatment guidelines. Journal of Clinical Psychiatry, 59, 34-40.

O'Hara, M. W. Neunaber, D. J., Zekoski, E. M. (1984) A prospective study of postnatal depression: Prevalence, course, and predictive factors. Journal of Abnormal Psychology, 93, 158-171.

O'Hara, M. W. (1994). Postpartum depression: Causes and consequences. New York, NY: Springer-Verlag.

Okun, A., Stein, R., Bauman, L., Silver, E. (1966). Content validity of the Psychiatric Symptom Index, CEC-Depression Scale, and State-Trait Anxiety Inventory from the perspective of DSM-IV. Psychological Reports, 1059-1069.

Paykel, E. (2000). Classification of postnatal disorders. Paper presented at the Marce Society Conference, September 6, Manchester, England.

Pitt, B. (1968). 'Atypical' depression following childbirth. British Journal of Psychiatry, 114, 1325-1335.

Pritchard, D.B., Harris, B. (1996). Aspects of perinatal psychiatric illness. British Journal of Psychiatry, 169, 555-562.

Rosenblum, O., Mazet, P., Benony, H. (1997). Mother and infant affective involvement states and maternal depression. Infant Mental Health Journal, 18, 350-363.

Small, R., Brown, S., Lumley, J., Astbury, J. (1994). Missing voices: What women say and do about depression after childbirth. Journal of Reproductive and Infant Psychology, 12, 89-103.

Snaith, R. P. (1993). What do depression rating scales measure? British Journal of Psychiatry, 163, 293-298.

Richters, J. E. (1992). Depressed mothers as informants about their children: a critical review of the evidence for distortion. Psychological Bulletin, 112, 485-499.

ACKNOWLEDGEMENT:

We appreciate the assistance of Kurt Wendelborn and anonymous reviewers for their helpful comments and suggestions.

FOOTNOTE

Consideration of space required the omission of the dendrogram, and figures showing the MDS solution and the profiles, but these are available from the corresponding author.