The use of the Edinburgh Postnatal Depression Scale (EPDS) for screening followed by health visitor interventions for postnatal depression have increased under the recent evidence-based practice emphasis in the NHS. There is consistent research support for this approach (Cox et al., 1987; Holden et al., 1989; Gerrard et al., 1993; Seeley et al., 1996; Wickberg & Hwang, 1996a). Unfortunately, there is little evidence that the other components of evidence-based practice have been appropriately addressed. In order to implement research findings it is vital to be clear what the research does not show (Beutler et al., 1993) and about the importance of traditional sources of knowledge such as clinical experience and user preference.
The authors of the original research developing the EPDS (Cox et al., 1987) and the health visitor intervention (Holden et al., 1989) themselves recognized that health professionals require more guidance than is provided by research reports in journals. They produced a book designed to promote appropriate use of the EPDS (Cox & Holden, 1994). This included the warning that 'far from being a magic wand, the EPDS by itself is just a piece of paper'(Elliott, 1994, p. 224). Unfortunately, this appears to have been ignored by some purchasers and managers responsible for determining health care systems. Barker (1998,1999) outlined his concerns about such purchaser- or manager-imposed systems as represented by policies for the use of the EPDS and other 'tick-boxes and checklists'. He claims that use of instruments to assess the developmental levels of children is declining so the era of pop psychology instruments is drawing to a close. But with the demise of one form of deskilling, another — the Edinburgh Postnatal Depression Scale — has arrived. [...] Sensitive health visitors (the majority) tell me they have always relied on their own assessment of the mother to help form a judgement [...]. Now we are beginning to hear of health visitors who have identified mildly or seriously depressed mothers who originally 'passed' the EPDS, in one case by as much as three points below the cut-off point (scoring only 9). Whatever, the merits of the EPDS as a backup in cases of uncertainty, imposing its use on all new parents can be counter-productive and damaging to some mothers. (Barker, 1998, p. 305)
Barker (1999) argues for health visitor intuition, described as 'forming judgements in the light of their own training and experience' (p. 32). He provided an example of this alternative approach attributed to 'one perceptive health visitor':
To pick up the antecedents of postnatal depression, to observe the interaction between the mother and baby, to discuss the mother's mental well-being with the mother, and finally to listen to what the mother is saying. (Barker, 1998, p. 305) In a challenge to this, Taylor (1998) argues that: Health visitors need to remain committed to developing a really sound and strong knowledge-base if the profession is to survive. Relying on Walter Barker's somewhat sentimental ideal of intuition merely assists the deskilling process that he says he is trying to prevent. (p. 427)
So while researchers view the issue of screening as resolved, with the EPDS the preferred methodology and requiring no further research, confusion reigns amongst practitioners. The implication of Barker's (1998) paper is that the health visitors have acquired this perspective from their local purchasers or managers and have received neither training nor references relevant to screening or intervention. Various misunderstandings that seem to underlie the views of health visitors working in imposed systems can be extracted from Barker's paper and from concerns (regarding the respective roles of health visitors and community psychiatric nurses (CPNs) and the use of the EPDS in other cultures) reported during the discussion of papers at the Society for Reproductive and Infant Psychology 1999 annual conference (Almond, 1999; Finch, 1999; Seeley, 1999). These are listed below. Each is considered in the light of available evidence. In most instances there is very little research available. The authors, therefore, respond from experience of health visitor training and clinical practice. Ideally this article will prompt research as well as provoke debate on practices being adopted in the name of evidence-based practice.
'The EPDS identifies 'depression', meaning that it is diagnostic'
The EPDS does not have a perfect relationship with diagnosis obtained at psychiatric interview so cannot be used as a substitute for a diagnostic interview. It does not make a differential diagnosis. If a woman scores over 12, then probably she has clinically significant levels of depression, but the primary diagnosis may not be 'Clinical Depression' or 'Postnatal Depression' (Leverton & Elliott, 2000). The depression could be part of another psychiatric problem. It is worth noting that the EPDS score correlates with Crown Crisp Experiential Index (Crown & Crisp, 1979) anxiety and somatic scales as well as depression (see Table 1, using data from a prediction and prevention study, Elliott et al., 1988; Elliott et al., 2000), positive moods (Self Report Questionnaire, Elliott et al., 1983) and Spielberger State Anxiety (Green, 1998). This is despite the deliberate omission in the EPDS of items which may be influenced by the physical aspects of pregnancy or the puerperium. This is probably explained by the fact that comorbidity is common in non-psychotic problems in primary care (Meltzer et al., 1994).
'The EPDS can be used as a 'back up in cases of uncertainty', again implying confirming a diagnosis'
The EPDS has higher predictive value when used only with those whom the health visitor earlier suspected had problems. However, this may be a statistical artefact of such samples having a higher base rate of depression or due to selecting those willing to communicate their distress to health professionals by any means. Even in these samples the relationship of EPDS to diagnosis of depression in a psychiatric interview is imperfect, so it cannot be used diagnostically. Particularly important here is the fact that the EPDS does not, and has never been claimed to, make a differential diagnosis. In this context it means that clinical levels of depression may be present as part of a primary diagnosis of depression, or as a secondary problem accompanying another diagnosis. If the health visitor has a sense of problems then she knows already that she should explore the issues further in order to reach a decision about extra help. If she refers to a GP then the GP may make a diagnosis as a preliminary step in treatment decisions. The EPDS cannot do this for her. It is possible, however, that health visitors lacking confidence in their own judgement will feel more comfortable referring to the GP if an EPDS score can be included in the referral, to 'back up' her request for a GP assessment. This use of the EPDS to provide additional information about the level of dysphoria is, of course, acceptable. Problems arise, however, if lack of confidence in health visitors becomes institutionalized so that postnatal referrals to the GP or secondary mental health services must include a high EPDS score. This would preclude referral of those for whom EPDS completion is judged insensitive and those scoring below cut-off whom the health visitor believes would benefit from further help. In areas where health visitors make a high number of inappropriate referrals, improvement can be made by training and facilitated peer supervision by mental health professionals (Gerrard et al., 1993), with or without the introduction of the EPDS. There is no evidence for improvement in the appropriateness of health visitor referrals as a result of the introduction of the EPDS without training. Perhaps the most worrying aspect of the notion that the EPDS should be introduced in an area as a 'back up in cases of uncertainty' is that it misses the point of the exercise — primary screening to avoid missed cases. As with brief health visitor contacts afforded by clinic attendance, the EPDS is a rough and ready means of separating those who may require further help from those who probably do not. As yet, the evidence is not available as to whether postnatal depression detection rate increases because of the EPDS per se or because of the requirement to see everyone at 6 weeks, plus possibly the health visitor training and improved primary/secondary care liaison which accompanies it (Leverton & Elliott, 2000).
'A score below cut-off on the EPDS confirms a diagnosis of no disorder'
It is unlikely that someone scoring below 10 on the EPDS in a research trial would have clinically significant levels of depression. However, if the 12/13 cut-off is chosen some women who are clinically depressed or on the path to clinical levels of disorder will be missed. Also a low EPDS score does not rule out symptomatology of other psychiatric disorders, or, indeed, other problems of concern to health visitors. Finally there is the question of what elements of service delivery influence the proportion of deliberate false negatives (Elliott, 1994; Hunn, 1999).
'Women can "pass" or "fail" the EPDS, implying both a clear dichotomy and a value judgement'
The graph of EPDS scores shows the distribution is not bi-modal but a normal distribution without the lower tail due to the 'floor' effect of zero (see Figure 1).
This data is consistent with other data (Green, 1998; Green & Kafetsiosis, 1997) which make clear that the EPDS provides scores on a continuum of dysphoria, with no clear dichotomy. The choice of cut-off score depends on the purpose of the screening (Cox & Holden, 1994). The terminology used for designating those above cut-off should be chosen with sensitivity. The term 'fail' is probably not helpful.
'The EPDS makes the decision to treat, so a score above 12 means referral to the CPN, and a score below 12 means no help is required'
Level of depressive symptomatology in any given week is only one factor to be considered when reaching a decision on when, whether, where and by whom a therapeutic intervention should be provided. The decision should be a collaborative one between the woman and her health visitor or other primary care professional taking a holistic approach (Appleby et al., 1997).
'CPNs should work only with clients with "Severe Mental Illness"; health visitors can provide the services for non-psychotic problems in parents'
In this system only those postnatal women with puerperal psychoses or chronic non-psychotic disorders would reach the mental health services. If there is no primary care psychology service, or other direct referral therapy service delivering uniprofessional psychotherapy services, health visitors are left to deal with all non-psychotic mental health problems identified as a result of EPDS screening with 'active listening' visits, although most of them have no mental health training or experience.
'Health visitors should be trained in making diagnoses, since the EPDS does not make the decision to treat'
The first problem with this solution is that diagnoses per se do not dictate treatments. (Bebbington et al., 1997, 1999). Secondly, diagnostic criteria were not designed to determine appropriateness for health visitor listening visits. They were designed to form subgroups within populations in psychiatric hospitals. They did come to have a function in filtering to ensure that only those with diagnosable psychiatric disorders received expensive psychiatric services. The philosophy of health visitor interventions following EPDS screening is consistent with their role in health promotion and in universal non-stigmatizing screening programmes (Elliott, 1994). Listening visits with women scoring highly on the EPDS aim for secondary prevention, thanks to early detection of postnatal problems. The object is to prevent depression descending the spiral far enough to reach diagnosable severity. This cannot happen if health visitors are instructed to undertake listening visits only when depression has already reached diagnosable criteria.
'The EPDS can detect suicide risk and prevent deaths — failure to use the EPDS could make a health visitor responsible for the death of a mother and her baby'
There has been no research on the relationship of EPDS total score or score on question 10 ('The thought of harming myself has occurred to me') and suicidal intent or action. It is known that suicide is lower in pregnant and postnatal women despite high rates of psychiatric morbidity (Appleby, 1996). The implication is that depression is a less strong predictor in this group. There is a rumour that question 10 assesses suicidal intent and/or indicates severity. In fact, question 10 can be scored not only by those contemplating deliberate self-harm but also by those with obsessional fears of harm and by those with fears of accidents, though this is possibly more frequent in pregnancy (Green et al., 1991; Green & Murray, 1994). Whilst this item is the least frequently endorsed, which would be compatible with the assumption that it is only scored by those with severe depression, it has actually the lowest correlation with total score which serves as a proxy for severity of depression (see Table 2).
'The EPDS administered on one occasion can determine postnatal mental health, that is, mental health over the 1st postnatal year'
The EPDS explicitly refers to the past week, so cannot be expected to convey information about the other 51 weeks of the year. High scores may remain high for weeks or months without help but some remit spontaneously and low scores can become high at any time. Recipients of the EPDS should be made aware that they can approach their health visitor at any time should they subsequently recognize that they are experiencing the problems indicated in the questionnaire.
'The EPDS system was designed to "assess the well-being of [health visitors'] clients"'
The EPDS was developed as a screening tool to sift out those in greater need of scarce resources. It was not designed to describe well-being, in all its complexity, though it transpires that it does provide an indication of position on a continuum of negative affect (Green, 1998).
'Adoption of the EPDS screening system requires "imposing its use on all new parents"'
The training programme for the introduction of EPDS systems is explicit on the person-centred philosophy within which it was developed and within which it should be delivered (Gerrard et al., 1993; Holden et al., 1989). The use of the system requires that all women are screened but not that the EPDS is used in all cases.
'The EPDS indicates the replacement of health visitors who "called to inspect bathrooms" by those using "pop psychology" '
The EPDS was not developed by 'pop psychologists' but by a reputable psychiatrist who is currently the president of the Royal College of Psychiatrists and his health professional colleagues (Cox et al., 1987). Its use is entirely compatible with sensitive use in a caring system guided by experience and insight, which Barker claims is the case with the instruments used by GPs and psychologists (Barker, 1999), but not those provided for use by health visitors.
'The EPDS should replace the judgement of a trained professional'
None of the validation studies have suggested this. Indeed, the decision to treat was made by the psychiatrist, not the EPDS, in the original validation and treatment study (Cox et al., 1987; Holden et al., 1989). The second screen by interview of high-scoring women is a central feature of training programmes (Gerrard et al., 1993) and, along with termination issues and referral decisions, is the rationale for the inclusion of mental health professionals in the trainer team.
'Routine screening with the EPDS is the right system for everyone so it should be translated into all languages'
The EPDS was, as the name implies, developed in and for Scotland. Misery is expressed in different ways in different cultures (Pilgrim & Bentall, 1999). Misery is responded to in different ways in different cultures. Some languages do not have the word 'depression' or would not perceive the health service as the appropriate source of help for emotional problems. Even if it is established that depression is expressed and managed in similar ways, and that paper and pencil screening tests are acceptable, translated versions of the EPDS must be submitted to validation studies (Barnett et al., 1999a, b; Cox & Holden, 1994; Pen et al., 1994; Wickberg & Hwang, 1996b; Ghubash et al., 1997; Matthey et al., 1997; Guedeney, 1998; Lee et al., 1998; Benvenuti et al., 1999).
'EPDS screening does not need to be followed by a second screen process'
Clearly, as the EPDS cannot make a diagnosis, and neither the EPDS nor a diagnosis can make the decision to treat, an interview would be preferred as a second screen process. Research is required to confirm that this can be done by an appropriately trained and supervised health visitor, not just by a psychiatrist as in the original Edinburgh study (Holden et al., 1989).
'The EPDS can, and should, be introduced without training in its use or on postnatal disorders and health visitor interventions'
No research trials have attempted to do this. Areas that adopt this practice are therefore not following evidence-based practice. Anecdotal reports suggest that health visitors in imposed programmes become stressed and avoid or abandon use of the EPDS or obtain their own training by other routes. Research is required to determine the quantity and quality of training necessary for each type of intervention used by health visitors, notably 'active listening' based on non-directive counselling (Holden et al., 1989; Gerrard et al., 1993), CREST postnatal cognitive behavioural counselling (Appleby et al., 1997; Whitton et al., 1996), and techniques from cognitive behaviour therapy (Crocket, 1999).
'The introduction of systems using the EPDS is associated with deskilling rather than skilling of health visitors'
Systems using the EPDS are traditionally introduced with training (Elliott et al., 1998). Health visitor questionnaires completed pre- and post-training and implementation of the EPDS screening indicate increases in skill and knowledge (Whitton et al., 1996), though changes in practice have yet to be systematically evaluated.
The EPDS is clearly not a magic wand to be distributed for compulsory use without training (Elliott, 1994). Alone it is just a piece of paper, a checklist, as Barker (1998) claims. Combined with training in prevention, detection and treatment, however, it becomes an important part of an effective programme (Cox & Holden, 1994; Gerrard et al., 1993; Seeley et al., 1996).
- Elliott, S. A., Leverton, T. J.; Is the EPDS a magic wand?: 2. 'Myths' and the evidence base; Journal of Reproductive & Infant Psychology; Nov 2000; Vol. 18, Issue 4
Reflection Exercise #8
The preceding section contained information
regarding the use of the Edinburgh Postnatal Depression Scale only as an aid for diagnosis of PPD. Write three
case study examples regarding how you might use the content of this section in
Online Continuing Education QUESTION 22
Does Elliott suggest that the EDPS is able to help diagnose females at risk for PPD when the health-care provider (health-visitor) is untrained in its usage? Record the letter of the correct answer the .