1.6. Systematic Review
1.6.6. Discussion
1.6.6.2. Strengths and Limitations of Key Studies
1.6.6.2.1. Design
Of the nine studies that were reviewed, there was an even split as to whether the study design was qualitative, quantitative, or observational. Of the qualitative studies, all involved interviews with partners, using different analysis techniques (of either IPA, Grounded Theory, or Thematic Analysis) to fit their research question and design. Quantitative measures used across studies varied widely, and included: a maternal depression screening (EPDS); adjustment scales; marital satisfaction scales; temperament scales; and general health questionnaires. However, there appeared to be a lack of consistency according to when these measures were administered during the postpartum period, which reduced control for other factors which may have affected results; for example, tiredness levels immediately post-childbirth. Observations either involved parent-infant dyads or M-F-I triads, with interaction scoring measures being inconsistent across studies. Scoring measures used were:
Early Relational Assessment Scale; Ainsworth’s Sensitivity and Insensitivity Ratings; and the Mellow Parenting Coding Systems.
1.6.6.2.2. Sampling
The nine studies reviewed all involved partners of women who had experienced PND (N=9).
Within this sample, some studies also considered the family triad with mothers and infants also participating (n=5). Across studies, the average age of partners were in their thirties, but partners overall ranged from 23- to 51-years-old. A study by Beestin et al. (2014) provided detailed analysis of the main factors partners perceived as being affected by PND. Although the diversity of the sample was a strength of the study, providing varied perspectives on partners’ lived experiences, the range of ages of partners was very varied (from 25- to 50-years-old). It could be argued that such a wide age range may reflect very different experiences of partners, depending on where they were in their lives at the time of the PND episode, and thus reduce the homogeneity of the sample. Overall, sample sizes varied considerably across studies, ranging from 25 to 100 for quantitative and observational studies (including controls), and from 8 to 14 for qualitative studies.
Recruitment tended to occur indirectly through initial contact with the mothers; for example, through PND support groups, or healthcare professionals, or as a result of maternal hospital admissions. One study, by Marrs et al. (2014), stated a very low recruitment rate compared to couples approached, with only 16% of partners approached agreeing to take part in the study. Therefore, results may reflect a trait, or experience of certain partners, which makes them more likely to want to share their experiences than if a larger proportion of those approached had opted to take part. For instance, only partners who held a grievance or experienced a particular difficulty, or had a positive experience, or who were most recently or heavily involved in the service, may have volunteered in order to get their views heard.
Mothers were most often screened using EPDS, to generate a depression score rating which helped identify them as being within the PND group rather than the control group. However, the cut-off score set to identify PND on the EPDS varied in some studies, challenging the homogeneity and comparability across PND groups. At times, the cut-off was set at 10 (instead of 12 to 13) to increase the sensitivity of the measure within a community-based sample; this potentially resulted in its use lacking an appropriate level of specificity.
Goodman (2008), and Zelkowitz and Milett (1996), applied this lower cut-off score, which resulted in women scoring 10 and over as being classed with PND yet within the subclinical or borderline range on the screening. The studies do not clarify whether women scoring the highest on the EPDS had partners who scored proportionally higher on the other measures, compared to partners of women who scored within the 10 to 12 range.
Within the quantitative studies, control groups were either matched in age and by socio-demographic variables with healthy controls, or had other mental, or physical health problems. The use of control groups added strength to the quality of the studies considered.
In terms of the experimental group sample, within the research presented by Lovestone and Kumar (1993), 8 out of the 12 partners who were found to demonstrate mental health problems during their spouse’s admission had pre-existing mental health problems, which following their spouse’s admission had resulted in a relapse. The total sample in the study was 24, therefore, a high proportion of those within the experimental group had pre-existing mental health problems. It would be interesting to know if this percentage would change if a larger sample were sought.
1.6.6.2.3. Data Analysis and the Presentation of Findings
Most of the research presented within the review demonstrated a high quality rating, providing an indicator that the studies have a sound methodology and that their results can be considered with confidence. However, Goodman’s (2008) results, indicating that partners do not play a compensatory role, are to be considered with some caution given their lower quality rating compared to the other studies within the review. Goodman (2008) demonstrated a general lack of transparency in providing the complete methodology and in the general write-up of the article. Another study within the review fell short on the data management rating; although Milgrom and McCloud (1996) generally appeared to have a strong methodology, some data collected was missing as some couples had not fully completed all questionnaires presented. There is little indication within the article as to how missing data was managed, or whether there was a pattern as to which questions were skipped, or missed, when completing the questionnaires. This may have indicated a theme, or that there were certain questions that couples struggled to answer, or did not want to provide information on, or were not relevant to their experiences.
1.6.6.2.4. Interviewing Partners of Women with Postnatal Depression
The majority of studies reported interviewing the partners separately to their spouse, and it would appear from the results provided that partners provided rich and detailed responses.
However, one of the main critiques of the study by Meighan et al. (1999) would be that partners were interviewed in front of their spouses. It is clear from the article that partners still provided quite in-depth answers, but it would have been interesting to know if partners may have been more likely to raise, or discuss in more depth, certain topics if their spouse had not been present at the interview. It may be that a partner is sensitive to discussing the extent of their difficulties in front of their spouse for fear of upsetting them, or not wanting to
raise potentially traumatic memories, or feeling unable to ‘complain’ about their experiences in contrast to what their spouse may have experienced.