Women's Health Issues xxx-xx (2015) 1–15
www.whijournal.com
Original article
Postpartum Psychosocial and Behavioral Health: A Systematic Review of Self-Administered Scales Validated for Postpartum Women in the United States Lorraine O. Walker, RN, EdD, MPH a,*, Jin Gao, MA b, Bo Xie, PhD a,b a b
School of Nursing, University of Texas at Austin, Austin, Texas School of Information, University of Texas at Austin, Austin, Texas
Article history: Received 3 October 2014; Received in revised form 9 May 2015; Accepted 12 May 2015
a b s t r a c t Purpose: Women’s poor postpartum psychosocial and behavioral health may negatively affect them and their infants. Validated postpartum screening scales can help to identify problems early, but currently there is limited knowledge in this area. Thus, we conducted a systematic examination of self-administered psychosocial and behavioral scales validated for postpartum women in the United States in the domains of depression, body image, diet, physical activity, smoking, and alcohol use. We examined the characteristics of included scales, their validation samples, and reported psychometric properties. Method: Nine databases were searched during October 2014 through February 2015. After meeting inclusion/exclusion criteria, article information was extracted independently by two authors, compared, and differences were resolved through discussions. Results: The final sample included 23 published articles covering 19 scales. Seventeen were in the domain of depression, and one each in physical activity and dietary domains. None was found in the domains of body image, smoking, or alcohol use. The number of scale items varied from 2 to 35. The majority of scales were originally designed for postpartum women, and validated in one or two postpartum studies with samples of predominantly adult women. If reported, scale reliability coefficients were generally 0.80 or greater and validity coefficients of 0.70 or greater. Five depression scales had favorable sensitivity and specificity using standard cutpoints, but only one was tested across adolescent, low-income, and predominantly ethnic minority postpartum populations. Conclusion: No U.S.-validated postpartum scales were found for body image, smoking, or alcohol use. Most scales had limited validity testing, and validation was in predominantly advantaged samples. Further scale development and testing are recommended. Copyright Ó 2015 by the Jacobs Institute of Women’s Health. Published by Elsevier Inc.
In the early months after giving birth, women often undergo a health transition as they adapt to the joys and demands of motherhood and infant caregiving (Devine, Bove, & Olson, 2000; Walker & Wilging, 2000). During this transition, some women may confront challenges to their psychosocial health in the form of depression (Gavin et al., 2005; Vesga-Lopez et al., 2008), and a greater percentage may experience various depressive symptoms (Walker, Timmerman, Kim, & Sterling, 2002). Still others may experience body image dissatisfaction as a result of pregnancy and associated weight gain (Gjerdingen et al., 2009b; * Correspondence to: Lorraine O. Walker, RN, EdD, MPH, Luci B. Johnson Centennial Professor in Nursing, School of Nursing, University of Texas at Austin, 1710 Red River St, Austin, TX 78701. Phone: 512-232-4751; fax: 512-471-4910. E-mail address: [email protected] (L.O. Walker).
Walker et al., 2002). Behavioral changes also occur with the transition to new motherhood. Women may have suboptimal behavioral health manifested in a poor-quality diet (Durham, Lovelady, Brouwer, Krause, & Ostbye, 2011a; Fowles & Walker, 2006; George, Hanss-Nuss, Milani, & Freeland-Graves, 2005), sedentary lifestyle (Ainsworth et al., 2013; Durham et al., 2011b; Wilkinson, Huang, Walker, Sterling, & Kim, 2004), and smoking relapse (Park et al., 2009). Although postpartum women consume alcohol less than nonpregnant women (Vesga-Lopez et al., 2008), an estimated 5.6% engage in binge drinking (Laborde & Mair, 2012). These domains of new mothers’ psychosocial (depression and body image) and behavioral health (diet, physical activity, smoking, and alcohol use) are important because of their consequences for both mothers’ and infants’ health and well-being.
1049-3867/$ - see front matter Copyright Ó 2015 by the Jacobs Institute of Women’s Health. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.whi.2015.05.006
2
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Postpartum depression affects about 19% of women in the first 3 postpartum months (Gavin et al., 2005), and depressive symptoms may be even more widespread (Walker et al., 2002). Postpartum depression is associated with destabilizing symptoms (Beck & Indman, 2005; Ugarriza, 2002), which can interfere with adequate parenting (Field, 2010; Gress-Smith, Luecken, Lemery-Chalfant, & Howe, 2012; McLearn, Minkovitz, Strobino, Marks, & Hou, 2006), and infant health and development (GressSmith et al., 2012; Kahn, Zuckerman, Bauchner, Homer, & Wise, 2002). Similarly, diets high in energy and sedentary lifestyles can contribute to postpartum weight retention or weight gain (Olson, Strawderman, Hinton, & Pearson, 2003; Walker et al., 2004). Parental smoking is linked to poorer child physical health (e.g., asthma) and behavioral problems (Cook & Strachan, 1999; Kahn, et al., 2002) and increases mothers’ risk of noncommunicable diseases, such as heart disease and cancer (World Health Organization, 2011). Furthermore, many women who experience unfavorable psychosocial or behavioral health are likely to do so in two or more domains (Walker, Sterling, Guy, & Mahometa, 2013b), thereby compounding potential adverse effects on mothers and infants. Because many of these postpartum psychosocial and behavioral changes are also risk factors for chronic disease (Centers for Disease Control and Prevention, 2012; Ferketich, Schwartzbaum, Frid, & Moeschberger, 2000; Josefsson & Sydsjo, 2007; Stampfer, Hu, Manson, Rimm, & Willett, 2000), the postpartum period is a critical period for prevention-focused assessments and health services. Given the importance of psychosocial and behavioral health of new mothers, psychometrically adequate scales are paramount for research and practice. For researchers studying the important postpartum health transition, and for health professionals striving to provide holistic health services for new mothers, a comprehensive review and analysis of psychosocial and behavioral assessment scales is essential. Looking comprehensively at such scales aids in selecting from among alternative instruments and in discerning gaps across the domains of the postpartum health transition. Key psychometric aspects of measurement scales are reliability (i.e., the consistency or reproducibility of assessments) and validity (i.e., the accuracy of assessments for the phenomenon of interest; Waltz, Strickland, & Lenz, 2005). When scales are used in screening for presence of a maternal health condition or other clinical concern, such as depression or unhealthy behaviors, the scale attributes of sensitivity and specificity are additionally important (Gordis, 2000). These screening scale attributes ideally should balance detecting those with a condition or concern, while avoiding falsely identifying those without it. Other attributes that may be relevant when applying scales in research and practice include scale length, readability, and cultural suitability of scales for diverse populations (Logsdon & Hutti, 2006; Waltz et al., 2005). For example, in clinical applications brief scales reduce the burden on people completing them and increase their acceptability (e.g., see Dalrymple et al., 2013; Zimmerman & McGlinchey, 2008). These measurement properties, thus, form the core of concerns in reviewing the adequacy of scales related to maternal psychosocial and behavioral health. Existing reviews and comparisons of psychosocial or behavioral health scales for new mothers have focused almost exclusively on single domains of the postpartum transition, such as postpartum depression (e.g., Gaynes et al., 2005). No reviews to our knowledge have attempted to look more comprehensively and systematically at the scope and psychometric quality of psychosocial and behavioral health scales for postpartum
women. Still, several reviews are noteworthy. In the psychosocial realm, one set of reviews focused specifically on measurement of perinatal or postpartum depression and compared psychometric properties of four to eight scales for measuring depressive symptoms or screening for depression (Boyd, Le, & Somberg, 2005; DeRosa & Logsdon, 2006; Gaynes et al., 2005; Jolley & Betrus, 2007; King, 2012; Zubaran, Schumacher, Roxo, & Foresti, 2010). However, many of these reviews do not include recent postpartum validation studies. In the behavioral realm, one review examined scales for physical activity measurement during pregnancy, but it did not include postpartum-specific measures (Evenson, Chasan-Taber, Downs, & Pearce, 2012). Because no review articles were found that covered the larger spectrum of scales for measuring psychosocial and behavioral health among postpartum women, a review with this more comprehensive perspective would advance understanding of the current state of maternal health assessment. (Note, scales dealing with the maternal role [Barkin, Wisner, Bromberger, Beach & Wisniewski, 2010; Fowles & Horowitz, 2006; Matthey, 2011] are outside the scope of this review.) Finally, in this article we focus specifically on selfadministered scales for postpartum women in the U.S. context for several reasons. First, there is evidence that scale attributes, such as sensitivity in accurately measuring postpartum depression, may vary by country (Gibson, McKenzie-McHarg, Shakespeare, Price, & Gray, 2009). Second, there are unique features of the U.S. health care system, such as gaps in postpartum services (Walker, Murphey, & Nichols, 2015), including in some cases a lack of health care providers with whom women feel comfortable discussing postpartum issues such as depression (Walker, Im, & Tyler, 2013a). As a result, postpartum women in the United States often find themselves facing a life transition in which they have limited contact with the health care system for their own health. In response to this gap, we sought to highlight the range of validated, self-administered scales suitable for this life transition. Self-administered scales provide potential measures for assessing where needs are greatest and a means of self-assessment that may be used in clinical and community-based programs and technology-based applications designed for new mothers. Purpose The purpose of this systematic review was to determine the scope and psychometric properties of self-administered scales pertinent to the postpartum psychosocial and behavioral health of women in the United States, and, in doing so, to provide a foundation for understanding the state of the science (strengths and gaps) with regard to existing scales for assessment of psychosocial and behavioral health of new mothers. The outcome of this review contributes to identifying areas where further scale development and refinement are needed, and will advance comprehensive psychosocial and behavioral health assessment in future research and practice. This review addressed the following questions. 1. What scales have been validated for study of psychosocial or behavioral health of postpartum women, and what are the descriptive characteristics of these scales (domain, number of items, readability, time to complete, original targeted populations, e.g., low-income women or adolescents, or other scale properties that might affect applicability to postpartum women)?
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
2. For psychosocial or behavioral health scales, what are the characteristics of postpartum samples used in validation studies? 3. What are the reported psychometric properties of scales (reliability, validity, sensitivity, specificity, and other relevant findings) used to measure women’s postpartum psychosocial or behavioral health?
Methods Database Search Databases were selected from the electronic journal databases available through the University of Texas at Austin. Given the focus of this literature review, databases were selected from the following eight fields: communication, communication sciences and disorders, communication studies, consumer health, medicine, nursing, nutrition and food, and psychology. To be included, a database must: 1. Contain journals that publish peer-reviewed articles. Databases that only contain dissertations, e-books, book reviews, video/audio, encyclopedias, images, factual data, regulations/laws, citations, or directories were not included. 2. Allow keyword search in the title or abstract. 3. Be relevant to the scope of the present review. 4. Contain articles with full-text written in English. A total of 10 databases under these eight fields were selected (the year coverage of each database listed below was the coverage available through the library at the University of Texas at Austin): PubMed (1950–present), Communication & Mass Media Complete (CMMC, 1920–present), Cumulative Index to Nursing & Allied Health (CINAHL) Plus with Full Text (1937– present), PsycINFO (1887–present), Sociological Abstracts (1952– present), Cochrane Library (dates of coverage vary), Health Source: Nursing/Academic Edition (dates of coverage vary), Mental Measurements Yearbook (1978–present), and Psychology & Behavioral Sciences Collection (1965–present). Keyword Search From October 2014 to February 2015, one of the authors (J.G.) performed keywords searches in the title in the 9 selected databases by using the following combination of search terms: (measure* OR assessment* OR questionnaire* OR scale* OR index* OR instrument* OR tool*) AND (postpartum OR “new mother*” OR postnatal OR perinatal). The following inclusion/exclusion criteria were used for this round of the searches: 1. Search articles published within the year coverage of each database described above; 2. Only included articles from peer-reviewed journals; and 3. Only included articles with the full-text written in English. A total of 1,321 records were identified. Title/Abstract Screening The following predetermined inclusion/exclusion criteria were used for this round of screening:
3
1. Only included empirical studies reporting original research findings. Other types of publications, for example, literature reviews, were excluded. 2. Only included studies on populations within the United States. 3. Only included studies involving self-administered scales. Studies using instruments that only consisted of open-ended questions were excluded. Studies using researcheradministered scales were also excluded. 4. Only included studies of scales written in or translated into English. 5. Only included studies aimed to develop a scale and/or assess a scale’s psychometric properties (reliability, validity, specificity, and sensitivity). Studies that only used a scale in their studies without explicitly reporting the scale’s psychometric properties in the study samples were excluded (e.g., Georgiopoulos, Bryan, Wollan, & Yawn, 2001). 6. Only included studies of scales that measure postpartum women’s specific health-related issues, including physical activity, diet/food taking, depression, tobacco/alcohol intake and body image. 7. Only included studies of scales measuring the domains of interest in this study, but excluded studies of scales measuring predictors of these domains, for example, scale predicting the risk of depression (Records, Rice, & Beck, 2007). One of the authors (J.G.) reviewed the title/abstract of all 1,321 articles based on these predetermined inclusion/exclusion criteria, and excluded 1,278 records, leaving a total of 43 records in the sample. Full Text Screening Because some articles did not provide sufficient details (e.g., country of study population) in their titles or abstracts for us to assess their eligibility, two of the authors (J.G., L.W.) examined the full-text of these 43 articles for further assessment. Differences between these two authors were resolved through meetings of all three authors. As a result, 20 articles were excluded because they violated at least one of the seven predetermined inclusion/exclusion criteria listed. Thus, a total of 23 articles were included in the final sample for further analysis. The searching and screening process is illustrated in Figure 1. Data Extraction The 23 articles were coded independently by two of the authors (J.G., L.W.) using a coding guide developed for this study. The following information in the 23 articles was coded and synthesized into four tables: 1) description of scale characteristics, including scale name, domain, number of items, administration time, original targeted population and readability (some scale characteristics that were not reported in the selected articles were obtained either from the original reports of scales or secondary resources), 2) sample characteristics, which include author(s)/publication year, research purpose (development, validation, and comparison), sample size, postpartum year/ month/day, age, race/ethnicity, educational level, income, and married/partnered status, and 3) psychometric properties of scales, which include reliability, validity, sensitivity, and specificity. Any differences in data extraction were resolved by consensus between the first two authors or in conference with the third author. Acronyms for various scales and measures are provided in Table 1.
4
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Identification
9 electronic journal databases were identified from collections of University of Texas in Austin
1321 records ds identified iden through database searching
1 1278 records excluded, including duplicates in
Eligibility
43 full-text articles assessed for eligibility
20 full-text articles excluded for validating at least one of the predetermined inclusion/exclusion criteria for literature p sc screening.
Included
Screening
1321 titles and abstracts a ab screened
23 articles included in qualitative synthesis
Figure 1. Flow diagram of a four-phase literature search and screening process.
Results From the 23 articles included in our final sample, we identified a total of 19 distinct psychosocial and behavioral scales that were designed for self-administration and validated for use with
postpartum women in the United States. Regarding the domains covered by these scales (Table 2), 16 focused on depression symptoms, one on psychological distress with depression as a component, one on eating behaviors (two of eight subscales), and one on physical activity. Several of these scales are
Table 1 Acronyms for Scales and Measures Acronym
Scale or Measure
BDI BDI-II CES-D CES-D - adolescent version CBPDS ESI EPDS EPDS-7 EPDS-2 HRSD KSADS-PL MAQ PHQ-2 PHQ-9 PHQ 2-item dichotomous screener PDSS PDSS–SF PDM PRAMS 2 or 3 depression-item combinations PRAMS-6 PRAMS-3D SCID
Beck Depression Inventory Beck Depression Inventory-II Center for Epidemiologic Studies Depression Scale Center for Epidemiologic Studies Depression Scale - adolescent version Correa-Barrick Postpartum Depression Scale Eating Stimulus Index Edinburgh Postpartum Depression Scale Edinburgh Postpartum Depression Scale-7 Edinburgh Postpartum Depression Scale-2 Hamilton Rating Scale for Depression Schedule for Affective Disorders and Schizophrenia for School Age Children-Present and Lifetime Version Modifiable Activity Questionnaire Patient Health Questionnaire-2 Patient Health Questionnaire-9 Patient Health Questionnaire 2-item dichotomous screener Postpartum Depression Screening Scale Postpartum Depression Screening Scale–Short Form Postpartum Distress Measure Pregnancy Risk Assessment Monitoring System 2 or 3 depression-item combinations Pregnancy Risk Assessment Monitoring System - 6 item Pregnancy Risk Assessment Monitoring System - 3 item Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders
Table 2 Descriptive Characteristics of Scales Validated for Postpartum Samples Scale Name
Domain
Number of Items
Readability
Administration Time
Original Target Population
Postpartum References
Beck Depression Inventory (BDI)
Depressive symptoms
21 (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961)
NR
Psychiatric patients (Beck et al., 1961)
Ji et al., 2011
Beck Depression Inventory-II (BDI-II) Center for Epidemiologic Studies Depression Scale (CES-D) Center for Epidemiologic Studies Depression Scale (CES-D) – adolescent version Correa-Barrick Postpartum Depression Scale (CBPDS) Eating Stimulus Index (ESI)
Depressive symptoms
21 (Beck & Gable, 2001a)
NR
Depressive symptoms
20 (Radloff, 1977)
Psychiatric patients (Beck & Gable, 2001a) General population (Radloff, 1977)
Depressive symptoms
30 (Logsdon & Myers, 2010)
Sixth grade via Flesch Kincaid grade level (Berndt, Schwartz, & Kaiser, 1983) Fifth grade via Fry readability graph (Logsdon & Hutti, 2006) Second grade via Fry Readability Graph (Logsdon & Hutti, 2006) NR
Beck & Gable, 2001a; Chaudron et al., 2010; Tandon et al., 2012* Logsdon & Myers, 2010; Mosack & Shore, 2006; Tandon et al., 2012* Logsdon & Myers, 2010
Depressive symptoms
20 (Barrick et al., 2012)
NR
NR
Diet (2/8 subscales cover eating behaviors)
Sixth-grade reading level (Cahill et al., 2009)
NR
Third-grade reading level via the Fry readability graph (King, 2012)
5 min (Cox et al., 1987)
Postpartum women (Cox et al., 1987)
5–10 min (Mosack & Shore, 2006) NR
Adolescents (Logsdon & Myers, 2010)
Barrick et al., 2012 Cahill et al., 2009
Edinburgh Postpartum Depression Scale (EPDS)
Depressive symptoms
23 (total scale); 3 (emotional eating subscale; 3 (dietary restraint subscale; Cahill et al., 2009) 10 (Cox et al., 1987)
Edinburgh Postpartum Depression Scale-7 (EPDS-7) Edinburgh Postpartum Depression Scale-2 (EPDS-2) Modifiable Activity Questionnaire (MAQ)
Depressive symptoms
7 (Venkatesh et al., 2014)
NR
NR
Postpartum women (Venkatesh et al., 2014)
Depressive symptoms
2 (Venkatesh et al., 2014)
NR
NR
Postpartum women
Venkatesh et al., 2014
Physical Activity
NR
NR
Pima Indians (Bauer et al., 2010)
Bauer et al., 2010
Sixth- to ninth-grade reading level (Thibault & Steiner, 2004)
NR
Eighth grade via Flesch Kincaid Grade Level (Shumway, Sentell, Unick, & Bamberg, 2004) NR
NR
General adult population (Kroenke, Spitzer, & Williams, 2003) General adult population (Kroenke, Spitzer, & Williams, 2001) NR
Cutler et al., 2007; Gjerdingen, Crow, McGovern, Miner, & Center, 2009a Gjerdingen et al., 2009a; Hanusa et al., 2008*; Davis et al., 2013*; Flynn et al., 2011 Gjerdingen et al., 2009a Beck & Gable, 2000; Beck & Gable, 2001a; Beck & Gable, 2001b; Le et al., 2009; Chaudron et al., 2010; McCabe et al., 2012 Hanusa et al., 2008
Patient Health Questionnaire-2 (PHQ-2)
Depressive symptoms
Varies with number of activities reported by respondent (Bauer et al., 2010) 2
Patient Health Questionnaire-9 (PHQ-9)
Depressive symptoms
9
PHQ (2-item dichotomous screener) Postpartum Depression Screening Scale (PDSS)
Depressive symptoms
2
Depressive symptoms
35y (Beck & Gable, 2000)
Scored 91.9 (“very easy” to read) on the Flesch; seventh-grade reading level on the Bormuth method (Beck & Gable, 2000)
NR
Postpartum women (Beck & Gable, 2000)
Postpartum Depression Screening Scale–Short Form (PDSS–SF)
Depressive symptoms
7* (Hanusa et al., 2008)
NR
NR
Postpartum women (Hanusa et al., 2008)
NR
Beck & Gable, 2001a; Hanusa et al., 2008*; Ji et al., 2011; Kaminsky et al., 2008; Le, Perry, & Sheng, 2009; Logsdon, Usui & Nering, 2009; Logsdon & Myers, 2010; Mosack & Shore, 2006; Chaudron et al., 2010; Flynn et al., 2011; Tandon et al., 2012*; Venkatesh et al., 2014 Venkatesh et al., 2014
5
(continued on next page)
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Postpartum women (Barrick et al., 2012) Postpartum low-income women (Cahill et al., 2009)
Abbreviation: NR, not reported. * Although scales used in these studies are self-administered scales, they were administered through interviews in these studies (Davis et al., 2013; Hanusa et al., 2008; Tandon et al., 2012). y There are five items under each of the seven subscales: sleeping/eating disturbances, anxiety/insecurity, emotional labiality, cognitive impairment, loss of self, guilt/shame, and contemplating harming oneself.
Davis et al., 2013* Postpartum women (Davis et al., 2013) NR 3 Depressive symptoms
NR
Davis et al., 2013* Postpartum women (Davis et al., 2013) NR NR 6
Pregnancy Risk Assessment Monitoring System (PRAMS) 2 or 3 depression-item combinations Pregnancy Risk Assessment Monitoring System (PRAMS) 6 item (PRAMS-6) Pregnancy Risk Assessment Monitoring System (PRAMS) 3 item (PRAMS-3D)
Anxiety and depressive symptoms
O’Hara et al., 2012 NR NR Two- to 3-item sets of 16 candidate items (O’Hara et al., 2012)
Postpartum women (O’Hara et al., 2012)
NR NR
Postpartum distress (includes aspects of depression) Depressive symptoms Postpartum Distress Measure (PDM)
10 (Allison et al., 2011)
General postpartum women (Allison et al., 2011)
Original Target Population Administration Time Readability Number of Items Domain Scale Name
Table 2 (continued )
Allison et al., 2011
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Postpartum References
6
alternative or shortened versions of existing scales, such as the Center for Epidemiologic Studies Depression scale (CES-D), Patient Health Questionnaire (PHQ), Edinburgh Postnatal Depression Scale (EPDS), or Postpartum Depression Screening Scale (PDSS). No scales were found that were validated for postpartum women in the United States in the areas of tobacco use, alcohol use, or body image. Length of depression scales varied greatly, from two or three items (PHQ-2, PHQ dichotomous screening items, EPDS-2, Pregnancy Risk Assessment Monitoring System [PRAMS] item sets, and PRAMS-3D) to 35 items (e.g., PDSS). For the Modifiable Activity Questionnaire (MAQ), which measures physical activity, the number of activities scored on it depends on the number of activities that respondents report they engage in. In the dietary domain, the two eating behavior subscales of the Eating Stimulus Index (ESI) each contained 3 items. Time to administer scales was usually not reported (Table 2). Readability of scales, such as the Beck Depression Inventory (BDI), BDI-II, CES-D, ESI, EPDS, PDSS, PHQ-9, and PHQ-2, as reported in the reviewed articles (or in other sources we were able to locate) generally placed low reading demands on users. Still, Allison, Wenzel, Kleiman, and Sarwer (2011) noted that the widely studied EPDS (Cox, Holden, & Sagovsky, 1987) was not understood by some American women because of use of British idioms. The majority of scales were originally designed for postpartum populations (Correa-Barrick Postpartum Depression Scale [CBPDS], ESI, EPDS, EPDS-7, EPDS-2, PDSS, PDSS–Short Form [SF], Postpartum Distress Measure [PDM], and various versions of PRAMS scales). Table 3 presents characteristics of samples used in studies validating postpartum scales. Overall, samples sizes varied from more than 1,300 (Davis, Pearlstein, Stuart, O’Hara, & Zlotnick, 2013) to 17 (Barrick, Kent, Crusse, & Taylor, 2012). With the exception of the BDI-II, CES-D, EPDS, PHQ-9, and PDSS, most scales were validated in only one or two postpartum studies. Most depression scales were validated in adult samples except for the CES-D, CES-D adolescent version, and the EPDS, which were tested also in adolescent samples. Roughly one-half of the depression scales were only validated in postpartum samples that were predominantly White, except the BDI-II, CES-D, CES-D adolescent version, EPDS (and 2 shortened versions), PDSS, and PHQ-2. In the articles we reviewed, only the BDI-II, CES-D, EPDS, PDSS, and PHQ-2 had been validated specifically for use with low-income women. In the dietary domain, the ESI was validated in a low-income, ethnically diverse sample. The one physical activity measure we identified, MAQ, was validated as a retrospective recall of postpartum activity 6 years earlier, which was recorded in physical activity diaries (n ¼ 30). Age was the only sample descriptor reported in this physical activity study (Bauer, Pivarnik, Feltz, Paneth, & Womack, 2010). Table 4 presents reliability and validity data as reported in the studies examined in this review. For depression scales, reliability estimates (usually coefficient alpha values) were reported for the following scales and were greater than 0.80: BDI-II, CES-D (original and adolescent versions), CBPDS, EPDS, PDSS, PHQ-9, and PDM. On the ESI, the emotional eating subscale showed higher reliability than the dietary restraint subscale. With regard to validity estimates for which another depression scale was used as a parallel validity measure, Pearson correlations of 0.70 or higher were reported for the BDI-II, CES-D (both versions), CBPDS, EPDS, PHQ-9, PDSS, PDSS–SF, and PDM. Comparable Spearman correlations for validity estimates were
Table 3 Sample Characteristics of Validation Studies of Postpartum Behavioral and Psychosocial Scales Author, Year (Scale, if Multiple Versions)
Study Purpose
Scale: Beck Depression Inventory (BDI) Ji et al., 2011 V&C
Sample Size, Time Postpartum 534, 0–26 wk
Scale: Beck Depression Inventory II (BDI-II) Beck & Gable, 2001a V&C 150, 39 d
Mean Age, y (mean SD) 33.1 5.1
31 4.8
Race/Ethnicity (%)
Education (%)
Income
Married/ Partnered (%)
86 White, 10 Black/African American, 4 others
8 partial or completed high school, 57 partial or completed college, 35 partial or completed graduate school
NR
89
87 White, 8 African American, 5 other
81 college degree or higher, some college, high school completion or lower 46 less than high school graduate, 34 high school graduate or GED, 20 more than high school education 59 high school diploma/GED or greater, reminder NR
NR
92
Low
26
Low
NR
Chaudron et al., 2010
V&C
198, 2 wk-14 mo
24.6 5.6
70 Black, 17 White, 7 Hispanic, 6 Mixed
Tandon et al., 2012
V&C
63, mean ¼ 8.2 wk
24.7 5.7
100 African American
Scale: Eating Stimulus Index (ESI) Cahill et al., 2009 D&V
Validation sample: 179, 0–4 mo Test–retest sample: 31, 0–1 y
Scale: Edinburgh Postpartum Depression Scale (EPDS); EPDS-7; EPDS-2 Beck & Gable, 2001a V&C 150, 39 d (EPDS) Chaudron et al., 2010 (EPDS)
V&C
198, 2 wk-14 mo
Flynn et al., 2011 (EPDS)
V&C
104 to mean ¼ 12 wk
Hanusa et al., 2008 (EPDS) Ji et al., 2011 (EPDS)
V&C
123, 6–8 wk
V&C
534, 0–26 wk
V&C
134, 6 wk
V&C
149, 4–6 wk
Kaminsky et al., 2008 (EPDS) Logsdon, Usui & Nering, 2009 (EPDS)
Validation sample: 26.4 Test–retest sample: 26.9
31 4.8
24.6 5.6
31 6 30.1 33.1 5.1
26.5 16 1.1
44 White, 42 Black, 7 Hispanic, 7 other
Mean grade was 10th grade
NR
45
41 Hispanic, 31 African American, 25 White, 4 other
17 less than high school, 29 partial high school, 45 high school graduate, 9 college 59 high school diploma/GED or greater, reminder NR
Low
46
Low
NR
65 White, 35 African American
12 less than high school, 59 high school, 23 college, 6 graduate education
NR
NR
Validation sample: 60 Hispanic, 22 White, and 18 African American Test–retest sample: 48 White, reminder NR
Validation sample: 53 at least partial college, 20 high school education or less, reminder NR Test–retest sample: 88 high school education or more, reminder NR
Low
NR
87 White, 8 African American, 5 other
81 college degree or higher, some college, high school completion or lower 46 Less than high school graduate, 34 high school graduate or GED, 20 more than high school education NR
NR
92
Low
26
76
8 partial or completed high school, 57 partial or completed college, 35 partial or completed graduate school 34 < 12 years
Diverse based on insurance type Diverse based on insurance type NR
89
NR
NR
Mean grade was 10th grade
Low
NR
100 African American
70 Black, 17 White, 7 Hispanic, 6 Mixed 78 White, 13 African-American, and 10 other 72 White, 19 African American, 10 other 86 White, 10 Black/African American, 4 other
41 African American, 38 Hispanic, 18 White, 3 other 45 White, 42 African American, 8 Hispanic, 4 other
NR
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Scale: Center for Epidemiologic Studies Depression scale (CES-D); adolescent version of CES-D V&C 59, 4–6 wk 16.4 1.3 Logsdon & Myers, 2010 (CES-D; CES-D adolescent) 25.6 5.9 Mosack & Shore, 2006 V&C 98, aggregated sample of 19 (CES-D) prenatal and 79 up to 2 y postpartum Tandon et al., 2012 V&C 63, mean ¼ 8.2 wk 24.7 5.7 (CES-D) Scale: Correa-Barrick Postpartum Depression Scale (CBPDS) Barrick et al., 2012 V 17, mean ¼ 7 wk 20.9 4.0
82
7
(continued on next page)
8
Table 3 (continued) Author, Year (Scale, if Multiple Versions) Logsdon & Myers, 2010 (EPDS) Mosack & Shore, 2006 (EPDS)
Study Purpose
Sample Size, Time Postpartum
Mean Age, y (mean SD)
V&C
59, 4–6 wk
16.4 1.3
V&C
98, aggregated sample of 19 prenatal and 79 up to 2 y postpartum 63, mean ¼ 8.2 wk
25.6 5.9
Education (%)
Income
Married/ Partnered (%)
44 White, 42 Black, 7 Hispanic, 7 other 41 Hispanic, 31 African American, 25 White, 4 other
Mean grade was 10th grade
NR
45
17 less than high school, 29 partial high school, 45 high school graduate, 9 college 59 high school diploma/GED or greater, reminder NR 77 enrolled in school, 19 completed 12th grade, reminder NR
Low
46
Low
NR
NR
16
NR
NR
NR
57 were high school graduate, 43 were some college or college graduate NR
Low
77
NR
NR
NR
Diverse based on insurance type Diverse
76
Tandon et al., 2012 V&C 24.7 5.7 100 African American (EPDS) 53 Hispanic, 17 Black, 16 White, 14 Median was V&C 106, repeated measured at Venkatesh et al., 2014 other 3 wk, 3 mo, and 6 mo 16, ranged (EPDS;EPDS-7; from 13–18 EPDS-2) Scale: Modifiable Activity Questionnaire (MAQ) Bauer et al., 2010 V 30, 12 wk & recalled 6 y 36.2 5.0 NR later Scale: Patient Health Questionnaire (PHQ)-9; PHQ-2 and PHQ 2-item screener (with yes/no response): Cutler et al., 2007 V 94, mean ¼ 1.1 y 27.5 5.3 66 Latino, 13 African American, 6 (PHQ-2) White, 12 Asian, 3 other Davis et al., 2013 (PHQ-9)
V
1,392y, 12 mo
Flynn et al., 2011 (PHQ-9)
V&C
104; mean ¼ 12 wk
Gjerdingen, et al., 2009a (PHQ-9; PHQ-2; PHQ 2 item screener)
V
506, 0–9 mo
28.52 5.5 31 6 29.1 6.2
88 White, 5 African American, 7 other 78 White, 13 African-American, and 10 other 67 White, 18 African American, 7 Asian, 7 other
Hanusa et al., 2008 V&C 123, 6–8 wk 30.1 72 White, 19 African American, 10 (PHQ-9) other Scale: Postpartum Depression Screening Scale (PDSS); Postpartum Depression Screening Scale–Short Form (PDSS–SF) Beck & Gable, 2000 D&V Study 1: 10, 8 wk Study 1: NR Study 1: NR (PDSS) Study 2: 525, 7.8 wk Study 2: 27.6 6.9 Study 2: 79 White, 11 Black, 7 Hispanic,
www.whijournal.com
Original article
Postpartum Psychosocial and Behavioral Health: A Systematic Review of Self-Administered Scales Validated for Postpartum Women in the United States Lorraine O. Walker, RN, EdD, MPH a,*, Jin Gao, MA b, Bo Xie, PhD a,b a b
School of Nursing, University of Texas at Austin, Austin, Texas School of Information, University of Texas at Austin, Austin, Texas
Article history: Received 3 October 2014; Received in revised form 9 May 2015; Accepted 12 May 2015
a b s t r a c t Purpose: Women’s poor postpartum psychosocial and behavioral health may negatively affect them and their infants. Validated postpartum screening scales can help to identify problems early, but currently there is limited knowledge in this area. Thus, we conducted a systematic examination of self-administered psychosocial and behavioral scales validated for postpartum women in the United States in the domains of depression, body image, diet, physical activity, smoking, and alcohol use. We examined the characteristics of included scales, their validation samples, and reported psychometric properties. Method: Nine databases were searched during October 2014 through February 2015. After meeting inclusion/exclusion criteria, article information was extracted independently by two authors, compared, and differences were resolved through discussions. Results: The final sample included 23 published articles covering 19 scales. Seventeen were in the domain of depression, and one each in physical activity and dietary domains. None was found in the domains of body image, smoking, or alcohol use. The number of scale items varied from 2 to 35. The majority of scales were originally designed for postpartum women, and validated in one or two postpartum studies with samples of predominantly adult women. If reported, scale reliability coefficients were generally 0.80 or greater and validity coefficients of 0.70 or greater. Five depression scales had favorable sensitivity and specificity using standard cutpoints, but only one was tested across adolescent, low-income, and predominantly ethnic minority postpartum populations. Conclusion: No U.S.-validated postpartum scales were found for body image, smoking, or alcohol use. Most scales had limited validity testing, and validation was in predominantly advantaged samples. Further scale development and testing are recommended. Copyright Ó 2015 by the Jacobs Institute of Women’s Health. Published by Elsevier Inc.
In the early months after giving birth, women often undergo a health transition as they adapt to the joys and demands of motherhood and infant caregiving (Devine, Bove, & Olson, 2000; Walker & Wilging, 2000). During this transition, some women may confront challenges to their psychosocial health in the form of depression (Gavin et al., 2005; Vesga-Lopez et al., 2008), and a greater percentage may experience various depressive symptoms (Walker, Timmerman, Kim, & Sterling, 2002). Still others may experience body image dissatisfaction as a result of pregnancy and associated weight gain (Gjerdingen et al., 2009b; * Correspondence to: Lorraine O. Walker, RN, EdD, MPH, Luci B. Johnson Centennial Professor in Nursing, School of Nursing, University of Texas at Austin, 1710 Red River St, Austin, TX 78701. Phone: 512-232-4751; fax: 512-471-4910. E-mail address: [email protected] (L.O. Walker).
Walker et al., 2002). Behavioral changes also occur with the transition to new motherhood. Women may have suboptimal behavioral health manifested in a poor-quality diet (Durham, Lovelady, Brouwer, Krause, & Ostbye, 2011a; Fowles & Walker, 2006; George, Hanss-Nuss, Milani, & Freeland-Graves, 2005), sedentary lifestyle (Ainsworth et al., 2013; Durham et al., 2011b; Wilkinson, Huang, Walker, Sterling, & Kim, 2004), and smoking relapse (Park et al., 2009). Although postpartum women consume alcohol less than nonpregnant women (Vesga-Lopez et al., 2008), an estimated 5.6% engage in binge drinking (Laborde & Mair, 2012). These domains of new mothers’ psychosocial (depression and body image) and behavioral health (diet, physical activity, smoking, and alcohol use) are important because of their consequences for both mothers’ and infants’ health and well-being.
1049-3867/$ - see front matter Copyright Ó 2015 by the Jacobs Institute of Women’s Health. Published by Elsevier Inc. http://dx.doi.org/10.1016/j.whi.2015.05.006
2
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Postpartum depression affects about 19% of women in the first 3 postpartum months (Gavin et al., 2005), and depressive symptoms may be even more widespread (Walker et al., 2002). Postpartum depression is associated with destabilizing symptoms (Beck & Indman, 2005; Ugarriza, 2002), which can interfere with adequate parenting (Field, 2010; Gress-Smith, Luecken, Lemery-Chalfant, & Howe, 2012; McLearn, Minkovitz, Strobino, Marks, & Hou, 2006), and infant health and development (GressSmith et al., 2012; Kahn, Zuckerman, Bauchner, Homer, & Wise, 2002). Similarly, diets high in energy and sedentary lifestyles can contribute to postpartum weight retention or weight gain (Olson, Strawderman, Hinton, & Pearson, 2003; Walker et al., 2004). Parental smoking is linked to poorer child physical health (e.g., asthma) and behavioral problems (Cook & Strachan, 1999; Kahn, et al., 2002) and increases mothers’ risk of noncommunicable diseases, such as heart disease and cancer (World Health Organization, 2011). Furthermore, many women who experience unfavorable psychosocial or behavioral health are likely to do so in two or more domains (Walker, Sterling, Guy, & Mahometa, 2013b), thereby compounding potential adverse effects on mothers and infants. Because many of these postpartum psychosocial and behavioral changes are also risk factors for chronic disease (Centers for Disease Control and Prevention, 2012; Ferketich, Schwartzbaum, Frid, & Moeschberger, 2000; Josefsson & Sydsjo, 2007; Stampfer, Hu, Manson, Rimm, & Willett, 2000), the postpartum period is a critical period for prevention-focused assessments and health services. Given the importance of psychosocial and behavioral health of new mothers, psychometrically adequate scales are paramount for research and practice. For researchers studying the important postpartum health transition, and for health professionals striving to provide holistic health services for new mothers, a comprehensive review and analysis of psychosocial and behavioral assessment scales is essential. Looking comprehensively at such scales aids in selecting from among alternative instruments and in discerning gaps across the domains of the postpartum health transition. Key psychometric aspects of measurement scales are reliability (i.e., the consistency or reproducibility of assessments) and validity (i.e., the accuracy of assessments for the phenomenon of interest; Waltz, Strickland, & Lenz, 2005). When scales are used in screening for presence of a maternal health condition or other clinical concern, such as depression or unhealthy behaviors, the scale attributes of sensitivity and specificity are additionally important (Gordis, 2000). These screening scale attributes ideally should balance detecting those with a condition or concern, while avoiding falsely identifying those without it. Other attributes that may be relevant when applying scales in research and practice include scale length, readability, and cultural suitability of scales for diverse populations (Logsdon & Hutti, 2006; Waltz et al., 2005). For example, in clinical applications brief scales reduce the burden on people completing them and increase their acceptability (e.g., see Dalrymple et al., 2013; Zimmerman & McGlinchey, 2008). These measurement properties, thus, form the core of concerns in reviewing the adequacy of scales related to maternal psychosocial and behavioral health. Existing reviews and comparisons of psychosocial or behavioral health scales for new mothers have focused almost exclusively on single domains of the postpartum transition, such as postpartum depression (e.g., Gaynes et al., 2005). No reviews to our knowledge have attempted to look more comprehensively and systematically at the scope and psychometric quality of psychosocial and behavioral health scales for postpartum
women. Still, several reviews are noteworthy. In the psychosocial realm, one set of reviews focused specifically on measurement of perinatal or postpartum depression and compared psychometric properties of four to eight scales for measuring depressive symptoms or screening for depression (Boyd, Le, & Somberg, 2005; DeRosa & Logsdon, 2006; Gaynes et al., 2005; Jolley & Betrus, 2007; King, 2012; Zubaran, Schumacher, Roxo, & Foresti, 2010). However, many of these reviews do not include recent postpartum validation studies. In the behavioral realm, one review examined scales for physical activity measurement during pregnancy, but it did not include postpartum-specific measures (Evenson, Chasan-Taber, Downs, & Pearce, 2012). Because no review articles were found that covered the larger spectrum of scales for measuring psychosocial and behavioral health among postpartum women, a review with this more comprehensive perspective would advance understanding of the current state of maternal health assessment. (Note, scales dealing with the maternal role [Barkin, Wisner, Bromberger, Beach & Wisniewski, 2010; Fowles & Horowitz, 2006; Matthey, 2011] are outside the scope of this review.) Finally, in this article we focus specifically on selfadministered scales for postpartum women in the U.S. context for several reasons. First, there is evidence that scale attributes, such as sensitivity in accurately measuring postpartum depression, may vary by country (Gibson, McKenzie-McHarg, Shakespeare, Price, & Gray, 2009). Second, there are unique features of the U.S. health care system, such as gaps in postpartum services (Walker, Murphey, & Nichols, 2015), including in some cases a lack of health care providers with whom women feel comfortable discussing postpartum issues such as depression (Walker, Im, & Tyler, 2013a). As a result, postpartum women in the United States often find themselves facing a life transition in which they have limited contact with the health care system for their own health. In response to this gap, we sought to highlight the range of validated, self-administered scales suitable for this life transition. Self-administered scales provide potential measures for assessing where needs are greatest and a means of self-assessment that may be used in clinical and community-based programs and technology-based applications designed for new mothers. Purpose The purpose of this systematic review was to determine the scope and psychometric properties of self-administered scales pertinent to the postpartum psychosocial and behavioral health of women in the United States, and, in doing so, to provide a foundation for understanding the state of the science (strengths and gaps) with regard to existing scales for assessment of psychosocial and behavioral health of new mothers. The outcome of this review contributes to identifying areas where further scale development and refinement are needed, and will advance comprehensive psychosocial and behavioral health assessment in future research and practice. This review addressed the following questions. 1. What scales have been validated for study of psychosocial or behavioral health of postpartum women, and what are the descriptive characteristics of these scales (domain, number of items, readability, time to complete, original targeted populations, e.g., low-income women or adolescents, or other scale properties that might affect applicability to postpartum women)?
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
2. For psychosocial or behavioral health scales, what are the characteristics of postpartum samples used in validation studies? 3. What are the reported psychometric properties of scales (reliability, validity, sensitivity, specificity, and other relevant findings) used to measure women’s postpartum psychosocial or behavioral health?
Methods Database Search Databases were selected from the electronic journal databases available through the University of Texas at Austin. Given the focus of this literature review, databases were selected from the following eight fields: communication, communication sciences and disorders, communication studies, consumer health, medicine, nursing, nutrition and food, and psychology. To be included, a database must: 1. Contain journals that publish peer-reviewed articles. Databases that only contain dissertations, e-books, book reviews, video/audio, encyclopedias, images, factual data, regulations/laws, citations, or directories were not included. 2. Allow keyword search in the title or abstract. 3. Be relevant to the scope of the present review. 4. Contain articles with full-text written in English. A total of 10 databases under these eight fields were selected (the year coverage of each database listed below was the coverage available through the library at the University of Texas at Austin): PubMed (1950–present), Communication & Mass Media Complete (CMMC, 1920–present), Cumulative Index to Nursing & Allied Health (CINAHL) Plus with Full Text (1937– present), PsycINFO (1887–present), Sociological Abstracts (1952– present), Cochrane Library (dates of coverage vary), Health Source: Nursing/Academic Edition (dates of coverage vary), Mental Measurements Yearbook (1978–present), and Psychology & Behavioral Sciences Collection (1965–present). Keyword Search From October 2014 to February 2015, one of the authors (J.G.) performed keywords searches in the title in the 9 selected databases by using the following combination of search terms: (measure* OR assessment* OR questionnaire* OR scale* OR index* OR instrument* OR tool*) AND (postpartum OR “new mother*” OR postnatal OR perinatal). The following inclusion/exclusion criteria were used for this round of the searches: 1. Search articles published within the year coverage of each database described above; 2. Only included articles from peer-reviewed journals; and 3. Only included articles with the full-text written in English. A total of 1,321 records were identified. Title/Abstract Screening The following predetermined inclusion/exclusion criteria were used for this round of screening:
3
1. Only included empirical studies reporting original research findings. Other types of publications, for example, literature reviews, were excluded. 2. Only included studies on populations within the United States. 3. Only included studies involving self-administered scales. Studies using instruments that only consisted of open-ended questions were excluded. Studies using researcheradministered scales were also excluded. 4. Only included studies of scales written in or translated into English. 5. Only included studies aimed to develop a scale and/or assess a scale’s psychometric properties (reliability, validity, specificity, and sensitivity). Studies that only used a scale in their studies without explicitly reporting the scale’s psychometric properties in the study samples were excluded (e.g., Georgiopoulos, Bryan, Wollan, & Yawn, 2001). 6. Only included studies of scales that measure postpartum women’s specific health-related issues, including physical activity, diet/food taking, depression, tobacco/alcohol intake and body image. 7. Only included studies of scales measuring the domains of interest in this study, but excluded studies of scales measuring predictors of these domains, for example, scale predicting the risk of depression (Records, Rice, & Beck, 2007). One of the authors (J.G.) reviewed the title/abstract of all 1,321 articles based on these predetermined inclusion/exclusion criteria, and excluded 1,278 records, leaving a total of 43 records in the sample. Full Text Screening Because some articles did not provide sufficient details (e.g., country of study population) in their titles or abstracts for us to assess their eligibility, two of the authors (J.G., L.W.) examined the full-text of these 43 articles for further assessment. Differences between these two authors were resolved through meetings of all three authors. As a result, 20 articles were excluded because they violated at least one of the seven predetermined inclusion/exclusion criteria listed. Thus, a total of 23 articles were included in the final sample for further analysis. The searching and screening process is illustrated in Figure 1. Data Extraction The 23 articles were coded independently by two of the authors (J.G., L.W.) using a coding guide developed for this study. The following information in the 23 articles was coded and synthesized into four tables: 1) description of scale characteristics, including scale name, domain, number of items, administration time, original targeted population and readability (some scale characteristics that were not reported in the selected articles were obtained either from the original reports of scales or secondary resources), 2) sample characteristics, which include author(s)/publication year, research purpose (development, validation, and comparison), sample size, postpartum year/ month/day, age, race/ethnicity, educational level, income, and married/partnered status, and 3) psychometric properties of scales, which include reliability, validity, sensitivity, and specificity. Any differences in data extraction were resolved by consensus between the first two authors or in conference with the third author. Acronyms for various scales and measures are provided in Table 1.
4
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Identification
9 electronic journal databases were identified from collections of University of Texas in Austin
1321 records ds identified iden through database searching
1 1278 records excluded, including duplicates in
Eligibility
43 full-text articles assessed for eligibility
20 full-text articles excluded for validating at least one of the predetermined inclusion/exclusion criteria for literature p sc screening.
Included
Screening
1321 titles and abstracts a ab screened
23 articles included in qualitative synthesis
Figure 1. Flow diagram of a four-phase literature search and screening process.
Results From the 23 articles included in our final sample, we identified a total of 19 distinct psychosocial and behavioral scales that were designed for self-administration and validated for use with
postpartum women in the United States. Regarding the domains covered by these scales (Table 2), 16 focused on depression symptoms, one on psychological distress with depression as a component, one on eating behaviors (two of eight subscales), and one on physical activity. Several of these scales are
Table 1 Acronyms for Scales and Measures Acronym
Scale or Measure
BDI BDI-II CES-D CES-D - adolescent version CBPDS ESI EPDS EPDS-7 EPDS-2 HRSD KSADS-PL MAQ PHQ-2 PHQ-9 PHQ 2-item dichotomous screener PDSS PDSS–SF PDM PRAMS 2 or 3 depression-item combinations PRAMS-6 PRAMS-3D SCID
Beck Depression Inventory Beck Depression Inventory-II Center for Epidemiologic Studies Depression Scale Center for Epidemiologic Studies Depression Scale - adolescent version Correa-Barrick Postpartum Depression Scale Eating Stimulus Index Edinburgh Postpartum Depression Scale Edinburgh Postpartum Depression Scale-7 Edinburgh Postpartum Depression Scale-2 Hamilton Rating Scale for Depression Schedule for Affective Disorders and Schizophrenia for School Age Children-Present and Lifetime Version Modifiable Activity Questionnaire Patient Health Questionnaire-2 Patient Health Questionnaire-9 Patient Health Questionnaire 2-item dichotomous screener Postpartum Depression Screening Scale Postpartum Depression Screening Scale–Short Form Postpartum Distress Measure Pregnancy Risk Assessment Monitoring System 2 or 3 depression-item combinations Pregnancy Risk Assessment Monitoring System - 6 item Pregnancy Risk Assessment Monitoring System - 3 item Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders
Table 2 Descriptive Characteristics of Scales Validated for Postpartum Samples Scale Name
Domain
Number of Items
Readability
Administration Time
Original Target Population
Postpartum References
Beck Depression Inventory (BDI)
Depressive symptoms
21 (Beck, Ward, Mendelson, Mock, & Erbaugh, 1961)
NR
Psychiatric patients (Beck et al., 1961)
Ji et al., 2011
Beck Depression Inventory-II (BDI-II) Center for Epidemiologic Studies Depression Scale (CES-D) Center for Epidemiologic Studies Depression Scale (CES-D) – adolescent version Correa-Barrick Postpartum Depression Scale (CBPDS) Eating Stimulus Index (ESI)
Depressive symptoms
21 (Beck & Gable, 2001a)
NR
Depressive symptoms
20 (Radloff, 1977)
Psychiatric patients (Beck & Gable, 2001a) General population (Radloff, 1977)
Depressive symptoms
30 (Logsdon & Myers, 2010)
Sixth grade via Flesch Kincaid grade level (Berndt, Schwartz, & Kaiser, 1983) Fifth grade via Fry readability graph (Logsdon & Hutti, 2006) Second grade via Fry Readability Graph (Logsdon & Hutti, 2006) NR
Beck & Gable, 2001a; Chaudron et al., 2010; Tandon et al., 2012* Logsdon & Myers, 2010; Mosack & Shore, 2006; Tandon et al., 2012* Logsdon & Myers, 2010
Depressive symptoms
20 (Barrick et al., 2012)
NR
NR
Diet (2/8 subscales cover eating behaviors)
Sixth-grade reading level (Cahill et al., 2009)
NR
Third-grade reading level via the Fry readability graph (King, 2012)
5 min (Cox et al., 1987)
Postpartum women (Cox et al., 1987)
5–10 min (Mosack & Shore, 2006) NR
Adolescents (Logsdon & Myers, 2010)
Barrick et al., 2012 Cahill et al., 2009
Edinburgh Postpartum Depression Scale (EPDS)
Depressive symptoms
23 (total scale); 3 (emotional eating subscale; 3 (dietary restraint subscale; Cahill et al., 2009) 10 (Cox et al., 1987)
Edinburgh Postpartum Depression Scale-7 (EPDS-7) Edinburgh Postpartum Depression Scale-2 (EPDS-2) Modifiable Activity Questionnaire (MAQ)
Depressive symptoms
7 (Venkatesh et al., 2014)
NR
NR
Postpartum women (Venkatesh et al., 2014)
Depressive symptoms
2 (Venkatesh et al., 2014)
NR
NR
Postpartum women
Venkatesh et al., 2014
Physical Activity
NR
NR
Pima Indians (Bauer et al., 2010)
Bauer et al., 2010
Sixth- to ninth-grade reading level (Thibault & Steiner, 2004)
NR
Eighth grade via Flesch Kincaid Grade Level (Shumway, Sentell, Unick, & Bamberg, 2004) NR
NR
General adult population (Kroenke, Spitzer, & Williams, 2003) General adult population (Kroenke, Spitzer, & Williams, 2001) NR
Cutler et al., 2007; Gjerdingen, Crow, McGovern, Miner, & Center, 2009a Gjerdingen et al., 2009a; Hanusa et al., 2008*; Davis et al., 2013*; Flynn et al., 2011 Gjerdingen et al., 2009a Beck & Gable, 2000; Beck & Gable, 2001a; Beck & Gable, 2001b; Le et al., 2009; Chaudron et al., 2010; McCabe et al., 2012 Hanusa et al., 2008
Patient Health Questionnaire-2 (PHQ-2)
Depressive symptoms
Varies with number of activities reported by respondent (Bauer et al., 2010) 2
Patient Health Questionnaire-9 (PHQ-9)
Depressive symptoms
9
PHQ (2-item dichotomous screener) Postpartum Depression Screening Scale (PDSS)
Depressive symptoms
2
Depressive symptoms
35y (Beck & Gable, 2000)
Scored 91.9 (“very easy” to read) on the Flesch; seventh-grade reading level on the Bormuth method (Beck & Gable, 2000)
NR
Postpartum women (Beck & Gable, 2000)
Postpartum Depression Screening Scale–Short Form (PDSS–SF)
Depressive symptoms
7* (Hanusa et al., 2008)
NR
NR
Postpartum women (Hanusa et al., 2008)
NR
Beck & Gable, 2001a; Hanusa et al., 2008*; Ji et al., 2011; Kaminsky et al., 2008; Le, Perry, & Sheng, 2009; Logsdon, Usui & Nering, 2009; Logsdon & Myers, 2010; Mosack & Shore, 2006; Chaudron et al., 2010; Flynn et al., 2011; Tandon et al., 2012*; Venkatesh et al., 2014 Venkatesh et al., 2014
5
(continued on next page)
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Postpartum women (Barrick et al., 2012) Postpartum low-income women (Cahill et al., 2009)
Abbreviation: NR, not reported. * Although scales used in these studies are self-administered scales, they were administered through interviews in these studies (Davis et al., 2013; Hanusa et al., 2008; Tandon et al., 2012). y There are five items under each of the seven subscales: sleeping/eating disturbances, anxiety/insecurity, emotional labiality, cognitive impairment, loss of self, guilt/shame, and contemplating harming oneself.
Davis et al., 2013* Postpartum women (Davis et al., 2013) NR 3 Depressive symptoms
NR
Davis et al., 2013* Postpartum women (Davis et al., 2013) NR NR 6
Pregnancy Risk Assessment Monitoring System (PRAMS) 2 or 3 depression-item combinations Pregnancy Risk Assessment Monitoring System (PRAMS) 6 item (PRAMS-6) Pregnancy Risk Assessment Monitoring System (PRAMS) 3 item (PRAMS-3D)
Anxiety and depressive symptoms
O’Hara et al., 2012 NR NR Two- to 3-item sets of 16 candidate items (O’Hara et al., 2012)
Postpartum women (O’Hara et al., 2012)
NR NR
Postpartum distress (includes aspects of depression) Depressive symptoms Postpartum Distress Measure (PDM)
10 (Allison et al., 2011)
General postpartum women (Allison et al., 2011)
Original Target Population Administration Time Readability Number of Items Domain Scale Name
Table 2 (continued )
Allison et al., 2011
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Postpartum References
6
alternative or shortened versions of existing scales, such as the Center for Epidemiologic Studies Depression scale (CES-D), Patient Health Questionnaire (PHQ), Edinburgh Postnatal Depression Scale (EPDS), or Postpartum Depression Screening Scale (PDSS). No scales were found that were validated for postpartum women in the United States in the areas of tobacco use, alcohol use, or body image. Length of depression scales varied greatly, from two or three items (PHQ-2, PHQ dichotomous screening items, EPDS-2, Pregnancy Risk Assessment Monitoring System [PRAMS] item sets, and PRAMS-3D) to 35 items (e.g., PDSS). For the Modifiable Activity Questionnaire (MAQ), which measures physical activity, the number of activities scored on it depends on the number of activities that respondents report they engage in. In the dietary domain, the two eating behavior subscales of the Eating Stimulus Index (ESI) each contained 3 items. Time to administer scales was usually not reported (Table 2). Readability of scales, such as the Beck Depression Inventory (BDI), BDI-II, CES-D, ESI, EPDS, PDSS, PHQ-9, and PHQ-2, as reported in the reviewed articles (or in other sources we were able to locate) generally placed low reading demands on users. Still, Allison, Wenzel, Kleiman, and Sarwer (2011) noted that the widely studied EPDS (Cox, Holden, & Sagovsky, 1987) was not understood by some American women because of use of British idioms. The majority of scales were originally designed for postpartum populations (Correa-Barrick Postpartum Depression Scale [CBPDS], ESI, EPDS, EPDS-7, EPDS-2, PDSS, PDSS–Short Form [SF], Postpartum Distress Measure [PDM], and various versions of PRAMS scales). Table 3 presents characteristics of samples used in studies validating postpartum scales. Overall, samples sizes varied from more than 1,300 (Davis, Pearlstein, Stuart, O’Hara, & Zlotnick, 2013) to 17 (Barrick, Kent, Crusse, & Taylor, 2012). With the exception of the BDI-II, CES-D, EPDS, PHQ-9, and PDSS, most scales were validated in only one or two postpartum studies. Most depression scales were validated in adult samples except for the CES-D, CES-D adolescent version, and the EPDS, which were tested also in adolescent samples. Roughly one-half of the depression scales were only validated in postpartum samples that were predominantly White, except the BDI-II, CES-D, CES-D adolescent version, EPDS (and 2 shortened versions), PDSS, and PHQ-2. In the articles we reviewed, only the BDI-II, CES-D, EPDS, PDSS, and PHQ-2 had been validated specifically for use with low-income women. In the dietary domain, the ESI was validated in a low-income, ethnically diverse sample. The one physical activity measure we identified, MAQ, was validated as a retrospective recall of postpartum activity 6 years earlier, which was recorded in physical activity diaries (n ¼ 30). Age was the only sample descriptor reported in this physical activity study (Bauer, Pivarnik, Feltz, Paneth, & Womack, 2010). Table 4 presents reliability and validity data as reported in the studies examined in this review. For depression scales, reliability estimates (usually coefficient alpha values) were reported for the following scales and were greater than 0.80: BDI-II, CES-D (original and adolescent versions), CBPDS, EPDS, PDSS, PHQ-9, and PDM. On the ESI, the emotional eating subscale showed higher reliability than the dietary restraint subscale. With regard to validity estimates for which another depression scale was used as a parallel validity measure, Pearson correlations of 0.70 or higher were reported for the BDI-II, CES-D (both versions), CBPDS, EPDS, PHQ-9, PDSS, PDSS–SF, and PDM. Comparable Spearman correlations for validity estimates were
Table 3 Sample Characteristics of Validation Studies of Postpartum Behavioral and Psychosocial Scales Author, Year (Scale, if Multiple Versions)
Study Purpose
Scale: Beck Depression Inventory (BDI) Ji et al., 2011 V&C
Sample Size, Time Postpartum 534, 0–26 wk
Scale: Beck Depression Inventory II (BDI-II) Beck & Gable, 2001a V&C 150, 39 d
Mean Age, y (mean SD) 33.1 5.1
31 4.8
Race/Ethnicity (%)
Education (%)
Income
Married/ Partnered (%)
86 White, 10 Black/African American, 4 others
8 partial or completed high school, 57 partial or completed college, 35 partial or completed graduate school
NR
89
87 White, 8 African American, 5 other
81 college degree or higher, some college, high school completion or lower 46 less than high school graduate, 34 high school graduate or GED, 20 more than high school education 59 high school diploma/GED or greater, reminder NR
NR
92
Low
26
Low
NR
Chaudron et al., 2010
V&C
198, 2 wk-14 mo
24.6 5.6
70 Black, 17 White, 7 Hispanic, 6 Mixed
Tandon et al., 2012
V&C
63, mean ¼ 8.2 wk
24.7 5.7
100 African American
Scale: Eating Stimulus Index (ESI) Cahill et al., 2009 D&V
Validation sample: 179, 0–4 mo Test–retest sample: 31, 0–1 y
Scale: Edinburgh Postpartum Depression Scale (EPDS); EPDS-7; EPDS-2 Beck & Gable, 2001a V&C 150, 39 d (EPDS) Chaudron et al., 2010 (EPDS)
V&C
198, 2 wk-14 mo
Flynn et al., 2011 (EPDS)
V&C
104 to mean ¼ 12 wk
Hanusa et al., 2008 (EPDS) Ji et al., 2011 (EPDS)
V&C
123, 6–8 wk
V&C
534, 0–26 wk
V&C
134, 6 wk
V&C
149, 4–6 wk
Kaminsky et al., 2008 (EPDS) Logsdon, Usui & Nering, 2009 (EPDS)
Validation sample: 26.4 Test–retest sample: 26.9
31 4.8
24.6 5.6
31 6 30.1 33.1 5.1
26.5 16 1.1
44 White, 42 Black, 7 Hispanic, 7 other
Mean grade was 10th grade
NR
45
41 Hispanic, 31 African American, 25 White, 4 other
17 less than high school, 29 partial high school, 45 high school graduate, 9 college 59 high school diploma/GED or greater, reminder NR
Low
46
Low
NR
65 White, 35 African American
12 less than high school, 59 high school, 23 college, 6 graduate education
NR
NR
Validation sample: 60 Hispanic, 22 White, and 18 African American Test–retest sample: 48 White, reminder NR
Validation sample: 53 at least partial college, 20 high school education or less, reminder NR Test–retest sample: 88 high school education or more, reminder NR
Low
NR
87 White, 8 African American, 5 other
81 college degree or higher, some college, high school completion or lower 46 Less than high school graduate, 34 high school graduate or GED, 20 more than high school education NR
NR
92
Low
26
76
8 partial or completed high school, 57 partial or completed college, 35 partial or completed graduate school 34 < 12 years
Diverse based on insurance type Diverse based on insurance type NR
89
NR
NR
Mean grade was 10th grade
Low
NR
100 African American
70 Black, 17 White, 7 Hispanic, 6 Mixed 78 White, 13 African-American, and 10 other 72 White, 19 African American, 10 other 86 White, 10 Black/African American, 4 other
41 African American, 38 Hispanic, 18 White, 3 other 45 White, 42 African American, 8 Hispanic, 4 other
NR
L.O. Walker et al. / Women's Health Issues xxx-xx (2015) 1–15
Scale: Center for Epidemiologic Studies Depression scale (CES-D); adolescent version of CES-D V&C 59, 4–6 wk 16.4 1.3 Logsdon & Myers, 2010 (CES-D; CES-D adolescent) 25.6 5.9 Mosack & Shore, 2006 V&C 98, aggregated sample of 19 (CES-D) prenatal and 79 up to 2 y postpartum Tandon et al., 2012 V&C 63, mean ¼ 8.2 wk 24.7 5.7 (CES-D) Scale: Correa-Barrick Postpartum Depression Scale (CBPDS) Barrick et al., 2012 V 17, mean ¼ 7 wk 20.9 4.0
82
7
(continued on next page)
8
Table 3 (continued) Author, Year (Scale, if Multiple Versions) Logsdon & Myers, 2010 (EPDS) Mosack & Shore, 2006 (EPDS)
Study Purpose
Sample Size, Time Postpartum
Mean Age, y (mean SD)
V&C
59, 4–6 wk
16.4 1.3
V&C
98, aggregated sample of 19 prenatal and 79 up to 2 y postpartum 63, mean ¼ 8.2 wk
25.6 5.9
Education (%)
Income
Married/ Partnered (%)
44 White, 42 Black, 7 Hispanic, 7 other 41 Hispanic, 31 African American, 25 White, 4 other
Mean grade was 10th grade
NR
45
17 less than high school, 29 partial high school, 45 high school graduate, 9 college 59 high school diploma/GED or greater, reminder NR 77 enrolled in school, 19 completed 12th grade, reminder NR
Low
46
Low
NR
NR
16
NR
NR
NR
57 were high school graduate, 43 were some college or college graduate NR
Low
77
NR
NR
NR
Diverse based on insurance type Diverse
76
Tandon et al., 2012 V&C 24.7 5.7 100 African American (EPDS) 53 Hispanic, 17 Black, 16 White, 14 Median was V&C 106, repeated measured at Venkatesh et al., 2014 other 3 wk, 3 mo, and 6 mo 16, ranged (EPDS;EPDS-7; from 13–18 EPDS-2) Scale: Modifiable Activity Questionnaire (MAQ) Bauer et al., 2010 V 30, 12 wk & recalled 6 y 36.2 5.0 NR later Scale: Patient Health Questionnaire (PHQ)-9; PHQ-2 and PHQ 2-item screener (with yes/no response): Cutler et al., 2007 V 94, mean ¼ 1.1 y 27.5 5.3 66 Latino, 13 African American, 6 (PHQ-2) White, 12 Asian, 3 other Davis et al., 2013 (PHQ-9)
V
1,392y, 12 mo
Flynn et al., 2011 (PHQ-9)
V&C
104; mean ¼ 12 wk
Gjerdingen, et al., 2009a (PHQ-9; PHQ-2; PHQ 2 item screener)
V
506, 0–9 mo
28.52 5.5 31 6 29.1 6.2
88 White, 5 African American, 7 other 78 White, 13 African-American, and 10 other 67 White, 18 African American, 7 Asian, 7 other
Hanusa et al., 2008 V&C 123, 6–8 wk 30.1 72 White, 19 African American, 10 (PHQ-9) other Scale: Postpartum Depression Screening Scale (PDSS); Postpartum Depression Screening Scale–Short Form (PDSS–SF) Beck & Gable, 2000 D&V Study 1: 10, 8 wk Study 1: NR Study 1: NR (PDSS) Study 2: 525, 7.8 wk Study 2: 27.6 6.9 Study 2: 79 White, 11 Black, 7 Hispanic,
