Advertisement

Journal of Behavioral Medicine

, Volume 41, Issue 5, pp 690–702 | Cite as

Shortening day length: a potential risk factor for perinatal depression

  • Deepika Goyal
  • Caryl Gay
  • Rosamar Torres
  • Kathryn Lee
Article

Abstract

The aim of this secondary analysis was to determine whether seasonal light exposure, categorized by type of day length, is associated with or predictive of depressive symptoms in late pregnancy and the first 3 months postpartum. Women (n = 279) expecting their first child were recruited from prenatal clinics and childbirth education classes. Depressive symptoms were assessed with the Center for Epidemiologic Studies Depression Scale. Day lengths were categorized into short, lengthening, long and shortening. Data analysis included linear mixed models and multiple linear regression. When days were shortening (August to first 4 days of November) in late third trimester, depressive symptom scores were highest (35%) and continued to be higher at each postpartum assessment compared to other day length categories. Implications for clinical practice include increased vigilance for depressive symptoms, particularly if late pregnancy and birth occurs during the 3 months around the Autumn equinox when day length is shortening. Strategies that increase light exposure in late pregnancy and postpartum should also be considered.

Keywords

Day length Season Autumn Winter Pregnancy Postpartum Depressive symptoms Sleep Actigraphy Mood 

Introduction

Defined as the onset of an affective mood disorder within the first 12 months after childbirth, postpartum depression (PPD) affects up to 20% of all women (American College of Obstetricians and Gynecologists [AGOG], 2016; Sit et al., 2015). Multiple biological, psychological, and social factors are well documented in the development of PPD. A previous history of depression, depression during pregnancy (Kettunen & Hintikka, 2017), and a history of intimate partner violence (Howard et al., 2013) are among the psychological factors associated with PPD. Biological factors related to PPD include sleep disturbance (Goyal et al., 2009; Stremler et al., 2017), low vitamin D levels (Fu et al., 2015; Gur et al., 2014; Robinson et al., 2014) and melatonin hormonal changes (Werner et al., 2015). Social factors include low socioeconomic status (Goyal et al., 2010), low level of social support (Kettunen & Hintikka, 2017), and adolescence or advanced maternal age (Muraca & Joseph, 2014; Torres et al., 2017).

Although reduced exposure to natural light has been associated with depression among adults in the general population (Marqueze et al., 2015), conflicting research exists regarding the effect of light exposure or seasonality on depressive symptoms during pregnancy and postpartum. Light exposure influences levels of vitamin D (National Institutes of Health, 2016) as well as melatonin secretion, and research suggests that light therapy may be effective for adults who experience seasonal affective disorder, also known as “winter depression” (Golden et al., 2005; Nussbaumer et al., 2015). Using actigraphy monitoring for light exposure, Wang et al. (2003) found no difference in mood or light exposure over 72 h when they compared 15 postpartum women with 22 non-pregnant women during a 2-month period (July–August) in San Diego CA. Moreover, they concluded that infant care responsibilities did not limit postpartum women’s light exposure (Wang et al., 2003). In a sample of 15 women with PPD randomized to 6 weeks of bright light or dim red light, both groups improved (Corral et al., 2007). In their review of five studies on efficacy of light therapy for antenatal depression and PPD, Crowley and Youngstedt (2012) concluded that bright light therapy may be beneficial for pregnant/new mothers by offsetting insufficient low light levels. To promote sleep among mothers with infants in the intensive care unit, researchers tested the effectiveness of bright light on sleep and depressive symptoms among women randomized into a control (n = 14) or treatment (n = 16) group (Lee et al., 2013). Women in the treatment group self-administered blue–green bright light 30 min per day for 3 weeks using a light visor. No statistically significant difference was noted between the two groups in this small sample, however, there was improvement in depression scores for the treatment group (Cohen’s d effect size = 0.40).

With inconclusive research findings regarding light exposure and light therapy for PPD, the aim of this secondary analysis was to determine if seasonal light exposure is associated with depression symptoms after controlling for known risk factors such as depression history, age and low socioeconomic status. During months with short day lengths, less daylight may influence levels of vitamin D and melatonin, placing women at further risk for depression than pregnancy and other known bio-psycho-social risk factors. Therefore, we aimed to test three hypotheses: (1) seasonal day length will be associated with the trajectory of postpartum depressive symptoms over time, (2) third trimester symptoms of depression will be associated with day length after controlling for bio-psycho-social factors, and (3) day length at third trimester will be predictive of depressive symptoms at 3 months postpartum after controlling for current day length, history of depression, and third trimester depressive symptoms.

Methods

Participants

The data for this secondary analysis were drawn from two longitudinal clinical trials designed to improve maternal sleep in the postpartum period [removed for blind review]. Parity was controlled by recruiting only first-time mothers, and health status was controlled by recruiting only low risk pregnant women in the third trimester with no indications of pregnancy complications. Both studies were approved by the Committee on Human Research at [removed for blind review] and all participants provided written consent prior to any data collection. Participants in the original studies were recruited during their third trimester between 2001 to 2003 and 2004 to 2008 using convenience sampling from childbirth education classes and antenatal clinics in Northern California. Participants for the 2001–2003 study were recruited from paid childbirth preparation classes with both the pregnant women and her partner enrolled, yielding an older, more affluent, and largely Caucasian sample. In 2004–2008, participants were recruited from clinics primarily serving low-income women, which yielded a younger, less affluent, and racially and ethnically diverse sample. Women were included in the study if they were ≥ 18 years of age, expecting their first child, and able to read and write in English. Detailed eligibility criteria for the two trials have been previously reported [removed for blind review]. The first sample included women with a past history or episode of depression or other mental health problem (n = 17) and the second sample excluded women with any history of mental health problems. At enrollment into the study, all women received an information sheet about PPD that included a contact list of available local resources.

Measures

Demographic and personal characteristics

All participants completed demographic questions (age, race/ethnicity, marital status, monthly income, educational level, employment status) and pregnancy history information in the third trimester. During the first 2 weeks postpartum, participants provided information by telephone that included type of birth (vaginal, cesarean), length of hospital stay for mother and infant, and infant’s gender and birth weight.

Seasonal day length

Due to the earth’s changing position with respect to the sun, seasons are characterized by temperature and weather patterns defined by specific calendar start and end dates. In the northern hemisphere, spring is from March equinox to June solstice, summer is from June solstice to September equinox, autumn is from September equinox to December solstice, and winter is from December solstice to March equinox. While the exact day depends on the year and time zone, twice each year at equinox, day and night lengths are equal (Sept 22 and March 20). Each year, the sun reaches its highest point at noon (summer solstice; longest day) and its lowest point at noon (winter solstice; shortest day) (National Weather Service, 2017).

For this analysis, we used equal distance from the middle of each season, rather than the start of each season, to represent day length in order to maximize the difference between the four types of seasonal day lengths. Thus, the category for “short” day length was from November 5 to February 4, the “lengthening” days category was February 5 to May 4, the “long” days category was May 6 to August 4, and the “shortening” days category was from August 5 to November 4. Given the potential importance for early clinical intervention, seasonal day length category at the time of the late third trimester measures was of primary interest. Because time from the late third trimester measures to the date of birth varied between 1 and 4 weeks, seasonal day length at birth was also examined for its relationship to postpartum depressive symptom scores.

Depressive symptoms

The 20-item, self-report, Center for Epidemiologic Studies-Depression (CES-D) 20-item scale (Radloff, 1977) was used to assess frequency of depressive symptoms in the third trimester and at 1, 2, and 3 months postpartum. The CES-D is well validated and has been used to assess prenatal and postpartum depressive symptoms (Beeghly et al., 2003; Marcus, 2009). Participants were asked to rate their symptoms during the prior week by indicating how often they experienced each symptom on a 4-point Likert scale ranging from 0 (rarely/none or less than 1 day) to 3 (most/all of the time or 5–7 days). Total scores can range from 0 to 60, with higher scores indicating more frequent depression symptoms. A score of ≥ 16 on this depression screening tool indicates high risk for PPD and the need for further assessment of clinical depression (Easterbrooks et al., 2016; Marcus et al., 2003). To estimate internal consistency reliability for the CES-D, Cronbach alpha coefficients were calculated and acceptable at .85 during the third trimester assessment, and above .86 at each postpartum visit.

Perceived stress

The 10-item version of the Perceived Stress Scale (PSS) was used to assess level of stress during the past month (Cohen et al., 1983). Items assess specific stress domains of unpredictability, lack of control, burden overload and stressful life circumstances; responses are based on a 5-point Likert scale (0 “never” to 4 “experienced very often”). Total scores can range from 0 to 40, with higher scores indicative of greater perceived stress. The PSS provides a global assessment of perceived stress, which has been shown to contribute to PPD (Shaw et al., 2000). The Cronbach alpha coefficient in this sample was ≥ .80 at each time point.

Relationship satisfaction

The 13-item Relationship Satisfaction Scale (RSAT) (Burns et al., 1994) was used to measure the participant’s current satisfaction with her relationship with her spouse or significant other Items are rated on a 7-point scale from not at all satisfied (0) to very satisfied (6) and higher scores indicate higher relationship satisfaction. The Cronbach alpha coefficient was > .90 at the third trimester and at 3 months postpartum.

Sleep duration

To obtain an objective measure of total sleep time, each participant wore a Mini Motionlogger wrist actigraph (Ambulatory Monitoring Inc., Ardsley, NY) for 48–72 consecutive weekday hours at each time point. A minimum of 48 consecutive hours was required for actigraphy data analysis. Wrist actigraphy provides continuous motion data using a wrist-watch size, battery operated microprocessor that senses motion with a piezo-electric linear accelerometer. Each 30-s epoch is scored as sleep or wake using an established algorithm from Action4 software (Ambulatory Monitoring Inc., Ardsley, NY) without researcher bias to obtain total minutes of sleep during the night. Although polysomnography is the gold standard for measuring duration of sleep stages, actigraphy is valid and reliable for estimating sleep and wake times in healthy young adults (Ancoli-Israel et al., 2003). In the present sample, there was no statistically significant difference between night 1 and night 2 data. In addition, both nights were significantly correlated (r > 0.50), indicating that 48-h sleep estimates were stable and reliable.

Sleep disturbance

The 21-item self-report General Sleep Disturbance Scale (GSDS) (Lee, 1992) was used to assess subjective sleep disturbance over the past week at each time point. Participants were asked to rate the frequency of specific sleep problems during the past week from 0 (not at all) to 7 (every day). The GSDS addresses problems such as sleep quality, difficulty falling asleep, daytime sleepiness and medication use. Scores can range between 0 and 147. A higher score is indicative of increased frequency of sleep disturbance. The GSDS had good internal consistency, with a Cronbach alpha coefficient of .88 in samples of employed women and childbearing women (Lee, 1992; Lee & DeJoseph, 1992).

Data analysis

Measures of central tendency [mean ± standard deviation (SD)], medians, and frequencies were used to describe the third trimester socio-demographic and clinical characteristics of participants in the combined studies. Variables were examined for normal distributions, and square root transformation of CES-D scores was sufficient to meet the assumptions for linear regression models. To address the first hypothesis regarding third trimester depression, Pearson correlations were calculated to determine relationships between depression symptom scores and continuous variables in the third trimester prior to randomization. Chi square (X2) tests were used to test for associations between categorical variables. Group differences on continuous variables were evaluated using one-way analysis of variance (ANOVA) with Sheffé post hoc tests.

To test the first hypothesis, linear mixed models with post hoc pairwise comparisons were used to evaluate changes in depression symptoms scores from third trimester through the first 3 months postpartum by day length category in late third trimester and by day length category at birth. To test the second hypothesis, the eleven bio-psycho-social variables from Table 1 along with the type of day length at the time of the third trimester assessment were entered into a multiple linear regression model to account for the variance in CES-D scores at the third trimester time point. Categorical data were dummy coded when appropriate (i.e., race/ethnicity, age groups, and type of day length). Intervention group assignment was not included in the model predicting third trimester CES-D scores as the assessment was conducted prior to the behavioral intervention. To test the third hypothesis regarding the extent to which day length category at third trimester predicts postpartum CES-D scores, we controlled for intervention group assignment, type of birth (vaginal or cesarean) and infant’s sex. We also included third trimester CES-D scores along with other relevant third trimester variables and current 3-month postpartum measures in the multiple regression model.
Table 1

Sample characteristics (n = 293)

Characteristic

Mean ± SD

No. (%)

Third Trimester Depressive Symptom Score (CES-D)

Total sample

  

13.2 ± 8.16

Age (years)

29.5 ± 6.3

 

F[2,288] = 4.1, p = .018

 18–24 years

 

78 (27%)

14.8 ± 8.8

 25–34 years

 

152 (52%)

11.9 ± 7.7*

 35 + years

 

63 (21%)

14.2 ± 8.0

Marital status

  

F[1,287] = 2.2, p = .144

 Single/unmarried

 

24 (8%)

15.5 ± 8.1

 Partnered or Married

 

268 (92%)

12.9 ± 8.2

Annual income

  

F[2,288] = 5.8, p = .004

 < $60,000

 

143 (49%)

14.8 ± 9.0

 > $60,000

 

121 (41%)

11.5 ± 6.9

 Not reported

 

29 (10%)

12.0 ± 7.4

Race/ethnicity

  

F[4,284] = 2.1, p = .08

 Caucasian

 

130 (45%)

12.1 ± 6.8

 Asian

 

70 (24%)

13.3 ± 9.4

 Hispanic

 

46 (16%)

14.5 ± 9.6

 African American

 

22 (8%)

17.0 ± 9.3

 Other

 

21 (7%)

12.2 ± 5.7

Education

  

F[1,288] = 14.5, p < .001

 Completed college

 

171 (58%)

11.7 ± 7.1

 Did not complete college

 

122 (42%)

15.3 ± 9.1

Employment status

  

F[2,288] = 1.3, p = .286

 Not employed or working

 

121 (41.3%)

14.0 ± 8.9

 Employed, not currently working

 

70 (23.9%)

13.0 ± 8.2

 Employed, currently working

 

98 (33.4%)

12.2 ± 7.1

 Not reported

 

4 (1.4%)

 

History of mental health problem

  

F[1,292] = 0.13, p = .723

 No

 

277 (95%)

13.2 ± 8.3

 Yes

 

16 (5%)

13.9 ± 7.0

Perceived stress (PSS)

15.1 ± 6.39

 

r = .650, p < .001

Relationship satisfaction (RSAT)

59.1 ± 13.9

 

r =  −.469, p < .001

Sleep disturbance (GSDS)

45.6 ± 16.4

 

r = .474, p < .001

Sleep duration (actigraphy hrs) (N = 277)

7.0 ± 1.3

 

r = −.107, p = .077

F[1,272] = 10.2, p = .002

 < 7 h

 

130 (47%)

15.0 ± 9.1

 > 7 h

 

147 (53%)

11.8 ± 7.1

GSDS general sleep disturbance scale, PSS perceived stress scale, RSAT relationship satisfaction scale

*Post hoc Sheffé: 25–34 years old CES-D scores significantly lower than 18–24 years old (p = .036)

Data were analyzed using SPSS version 22.0 (IBM, Armonk, NY). Statistical significance was set at p < .05, and standardized effect sizes (Cohen’s d) were calculated (mean difference/pooled SD) to describe the magnitude of group differences in seasonal day length when relevant.

Results

Of the combined sample of 304 participants with complete data during the third trimester, 11 cases were lost to follow up and had no birth data. They were excluded from this sample due to unknown duration of gestation at their pregnancy assessment. There were 273 (90%) cases with data for all four time points, and there was no attrition for any of the 17 women with a prior mental health problem. There were 16 women with incomplete actigraphy data at the third trimester assessment and 24 women with incomplete actigraphy data at the 3-month postpartum assessment.

Sample characteristics

The 293 women in this analysis were anticipating the birth of their first child and sample characteristics at third trimester are presented in Table 1. Participants were 29.5 (SD = 6.3) years of age with 42% unemployed or not working at the time of data collection. There was no significant difference in the number of births by season, most (73%) of the women gave birth vaginally, and 56% of the infants were males.

Depressive symptoms over time and seasonal day length

Mean CES-D scores and risk for PPD decreased significantly (p < .001) over time from the third trimester to 3 months postpartum regardless of seasonal day length at third trimester or birth (Fig. 1). A sensitivity analysis of women with complete data for both the third trimester and 3 months postpartum also demonstrated decreasing scores over time, indicating that attrition was not likely to be responsible for the decrease in CES-D scores over time.
Fig. 1

Depressive symptom scores by seasonal day length category at third trimester. Legend: There was a significant (F[3,248] = 20.95; p < .001) decrease in CES-D (square root transformed) scores over time from the third trimester to the 3-month postpartum assessment for the overall sample regardless of day length group. Linear fixed effects omnibus model for third trimester day length with Bonferroni adjustment for multiple comparisons was significant (F[3,1105] = 3.155, p = .024). Note: black square, shortening day length group (n = 58); white circle, lengthening day length group (n = 68); black circle, long day length group (n = 60); white square, short day length group (n = 63)

The linear mixed models omnibus analysis indicated a main effect of day length at the late third trimester assessment on the trajectory of CES-D scores (square root transformed) over time (F[3,1105] = 3.155, p = .024). There were significant post hoc pairwise comparisons for shortening and short day lengths, but no main effect of lengthening or long day light. The post hoc linear mixed model comparing shortening third trimester day length against the other types of day length indicated that the CES-D scores were significantly higher (p = .005) over time (95% CI: 12.0, 14.0) compared to the other three types of day lengths (95% CI: 10.9, 11.9). The post hoc linear mixed model comparing short third trimester day length against the other types of day length indicated that the CES-D scores were significantly lower (p = .042) over time (95% CI: 10.1, 12.3) compared with the other three types of day lengths (95% CI: 11.8, 13.0).

Results from the mixed model and post hoc comparisons were similar when day length at birth was examined (F[3,1105] = 3.44, p = .016); women who gave birth during the shortening day length time frame had higher CES-D scores over time (95% CI: 11.8, 14.1) than women who gave birth during the other three types of day length (95% CI: 10.8, 11.9). Table 2 summarizes CES-D scores and risk for PPD at each assessment by day length at birth. Effect sizes for mean differences in CES-D depressive symptom scores between shortening day length and the other types of day lengths at the time of birth were small to moderate at 1 month (d = .40) and 2 months (d = .27) postpartum (see Table 3). Effect sizes for mean differences between shortening day length and the other three types of day lengths at the time of birth were small at both the third trimester (d = .24) and third month postpartum (d = .19). Women who gave birth during the shortening day length time frame were also at greater risk for depression (CES-D scores ≥ 16) compared to women who gave birth during the other three types of day length periods. Their risks for depression were statistically significant at 1 month (p = .017) and 2 months postpartum (p = .044) but there was no significant higher risk for depression at the 3-month postpartum time point (Table 2).
Table 2

Depression scores by day length category at birth

CES-D scores

Total sample

Short day length at birth

Lengthening day length at birth

Long day length at birth

Shortening day length at birth*

3rd trimester

Mean ± SD

13.2 ± 8.16

12.7 ± 7.87

12.7 ± 7.06

12.9 ± 8.81

14.7 ± 9.18a

Median

11

11

11

12

12

Risk for depression (scores > 15), n (%)

88/289 (30%)

23/80 (29%)

24/83 (29%)

18/63 (29%)

23/63 (37%)b

1 month postpartum

Mean ± SD

13.2 ± 8.61

11.6 ± 8.00

13.9 ± 8.56

12.6 ± 9.16

14.9 ± 8.79c

Median

12

10.5

12

10

14

Risk for depression (scores > 15), n (%)

88/281 (31%)

17/78 (22%)

28/80 (35%)

18/61 (30%)

25/62 (40%)d

2 months postpartum

Mean ± SD

10.6 ± 7.48

10.0 ± 7.24

10.8 ± 7.33

9.7 ± 6.95

11.8 ± 8.50e

Median

9

9

9

9

9

Risk for depression (scores > 15), n (%)

61/271 (23%)

13/71 (18%)

22/81 (27%)

9/62 (15%)

17/57 (30%)f

3 months postpartum

Mean ± SD

10.5 ± 7.94

10.2 ± 8.27

10.0 ± 7.91

10.4 ± 7.75

11.5 ± 7.88g

Median

8

7.5

8

8

11

Risk for depression (scores > 15), n (%)

59/273 (22%)

15/72 (21%)

13/80 (16%)

16/62 (26%)

15/59 (25%)h

*Fixed effects for day length category at birth with Bonferroni adjustment for multiple comparisons was significant (F[3,1105] = 3.44, p = .016); post hoc pairwise comparisons: shortening vs short (CES-D square root transformation mean difference = .294, p = .029); shortening versus long (CES-D square root transformation mean difference = .294, p = .041)

aCohen’s d = 0.237 (shortening vs. short)

bX [1] 2  = 0.97, p = .321 NS

cCohen’s d = 0.395 (shortening vs. short)

dX [1] 2  = 5.6, p = .017

eCohen’s d = 0.273 (shortening vs. long)

fX [1] 2  = 4.1, p = .044

gCohen’s d = 0.190 (shortening vs. lengthening)

hX [1] 2  = 1.8, p = .183 NS

Table 3

Depressive symptoms by seasonal day length category in late third trimester

Third trimester depressive symptom scores (CES-D)

Total N = 289

Short day length 9.9 ± 0.25 h (autumn into winter) (n = 76)

Lengthening day length 12.3 ± 1.06 h (winter into spring) (n = 77)

Long day length 14.5 ± 0.28 h (spring into summer) (n = 68)

Shortening day length 12.0 ± 1.80 h (summer into autumn) (n = 68)

Mean ± SD

13.1 ± 8.2

11.9 ± 6.7

13.0 ± 7.6

13.5 ± 9.5

14.4 ± 8.8*

Median

11

11

12

11.5

12

Risk for depression (score ≥ 16)

88 (30%)

20 (26%) lowest rate

23 (30%)

21 (31%)

24 (35%) highest rate

*Fixed effects for third trimester day length with Bonferroni adjustment for multiple comparisons was significant (F[3,1105] = 3.155, p = .024); post hoc pairwise comparisons: shortening vs short (CES-D square root transformation mean difference = .315, p = .014), d = 0.325

Third trimester depressive symptom severity and risk of depression

The third trimester CES-D depressive symptom scores for all participants (n = 289) was 13.1 (SD = 8.2). As shown in Table 3, the lowest mean score was during the ‘short’ day length period (late fall into early winter) when day length averaged about 10 h. The highest mean score was during the ‘shortening’ day length period (late summer into early fall) when day length averaged about 12 h. However, one-way ANOVA showed no significant difference in mean scores by the four categories of day length. The difference in mean scores between ‘short’ day length and ‘shortening’ day length approached significance (t = 1.95, p = .053) with a small to medium effect size (d = 0.325). See Fig. 2.
Fig. 2

Depressive symptom scores by seasonal day length category in late third trimester. Legend: Third trimester CES-D means (black square) with 95% confidence intervals are depicted on the Y axis according to the four seasonal day length categories described on the X axis. There was no significant difference in CES-D scores (square root transformed) by third trimester day length category. The difference in CES-D scores between ‘short’ day length and ‘shortening’ day length approached significance (t = 1.95, p = .053), with a small to medium effect size (d = 0.325)

The overall rate for risk of depression (CES-D ≥ 16) in the third trimester was 30% (n = 88). As shown in Table 3, the lowest risk for depression (26%) occurred during the ‘short’ day length period and the risk for depression was highest (35%) during the ‘shortening’ day length period. The difference in prevalence rates between these two types of day length periods approached statistical significance (p = .068).

Multivariate analysis for third trimester depressive symptom severity

Participant characteristics (maternal age groups, education, marital status, income, race/ethnicity, work status, mental health history), relationship satisfaction score, perceived stress score, general sleep disturbance score, and actigraphy sleep duration were entered into a linear regression model to account for the variance in CES-D scores (square root transformed) during the third trimester. In addition to these salient bio-psychosocial variables, shortening day length at third trimester was also specified in the model based on the omnibus main effects for CES-D scores over time. The model was significant, accounting for 57% of the variance in CES-D scores (R2 = .568; F[13, 235] = 23.5, p < .001). Controlling for all 13 variables, significant variables included older maternal age, higher income, higher perceived stress, lower relationship satisfaction, higher sleep disturbance scores, and shortening day length (See Table 4). Prior history of mental health issues, marital status, employment, education, and sleep duration were not significant in the model, and being non-Hispanic White race/ethnicity approached significance (p = .07).
Table 4

Linear regression model accounting for the variance in third trimester depressive symptom scores

Model

Unstandardized coefficients

Standardized coefficients

t

p value

B

SE

Beta

(Constant)

2.386

.407

 

5.869

< .001

Age (< 25 years = 1; other = 0)

− .155

.151

− .058

− 1.026

.306

Age (> 34 years = 1; other = 0)

.307

.132

.109

2.324

.021

Mental health history (yes = 1; no = 0)

− .021

.246

− .004

− .087

.931

Married (yes = 1; no = 0)

.056

.133

.024

.425

.671

Ethnicity (white = 1; other = 0)

.222

.116

.096

1.918

.056

College graduate (yes = 1; no = 0)

− .160

.137

− .068

− 1.169

.244

Currently working (yes = 1; no = 0)

.105

.121

.044

.865

.388

Income (> $60,000 = 1; other = 0)

.279

.128

.120

2.184

.030

3rd trimester perceived stress (PSS)

.080

.009

.458

9.280

< .001

3rd trimester relationship satisfaction (RSAT)

− .020

.004

− .247

− 4.800

< .001

3rd trimester sleep disturbance (GSDS)

.018

.003

.263

5.478

< .001

3rd trimester sleep duration (actigraphy hours)

− .060

.106

− .026

− .569

.570

3rd trimester day length (shortening = 1; other = 0)

.068

.030

.099

2.245

.026

Dependent variable: 3rd trimester CES-D score (square root transformation); R2 = .568, F[13,245] = 23.5, p < .001

Multivariate analysis for depressive symptom severity at 3 months postpartum

The same maternal characteristics (age groups, education level, marital status, income, race/ethnicity, work status, mental health history), along with group assignment and birth and postpartum variables that included baby’s sex, vaginal or cesarean birth, and exclusive breastfeeding or bottle feeding were included in the linear regression model to predict depressive symptom severity (CES-D transformed scores) at 3 months postpartum. Scores for current perceived stress, relationship satisfaction, sleep disturbance, and sleep duration at 3 months postpartum were also included. In addition, third trimester CES-D scores and third trimester day length (shortening vs. other) were entered. Because of varying time intervals between the third trimester day length assessment and the final assessment at 3-months postpartum, types of day length (short vs other shortening, vs other; and long vs other) at 3 months postpartum were also included. This model with 21 variables was significant, accounting for 67% of the variance in CES-D scores at 3 months postpartum (R2 = .667; F[21, 188] = 16.7 p < .001). Results were similar in a linear regression model that also included day length at birth (R2 = .66; F[22, 188] = 19.2), but day length at birth was not a significant predictor and day length at third trimester remained significant (data not shown).

In this multivariate analysis with prior history of mental health problems and third trimester CES-D scores in the model, older maternal age group and higher income were no longer significant predictors of postpartum depressive symptom severity. As in the third trimester model, current higher perceived stress, lower relationship satisfaction and higher sleep disturbance scores were significant. Unlike the third trimester model, sleep duration was also significant, with shorter postpartum sleep duration associated with higher CES-D scores (See Table 5). The third trimester shortening day length remained significant (p = .029) even after controlling for day lengths at the 3-month postpartum assessment.
Table 5

Regression model accounting for the variance in depressive symptoms at 3 months postpartum

Model

Unstandardized coefficients

Standardized coefficients

t

p value

B

SE

Beta

(Constant)

1.775

.594

 

2.986

.003

Randomized group assignment

− .101

.119

− .040

− .847

.398

Mental health history (yes = 1; no = 0)

.955

.312

.152

3.063

.003

3rd trimester CES-D score (transformed)

.200

.056

.176

3.582

< .001

Age (< 25 years = 1; other = 0)

.334

.183

.112

1.830

.069

Age (> 34 years = 1; other = 0)

.136

.148

.045

.914

.362

Married (yes = 1; no = 0)

.057

.157

.022

.360

.719

Ethnicity (white = 1; other = 0)

.089

.136

.035

.658

.511

College graduate (yes = 1; no = 0)

− .116

.150

− .044

− .777

.438

Currently working (yes = 1; no = 0)

.050

.153

.015

.329

.742

Income (> $60,000 = 1; other = 0)

− .075

.155

− .030

− .484

.629

Gender of baby (girl = 1; boy = 2)

− .115

.120

− .045

− .960

.338

Type of birth (vaginal = 0; cesarean = 1)

− .208

.138

− .072

− 1.507

.134

Exclusive breastfeeding (yes = 1)

.066

.125

.026

.530

.587

Current perceived stress score (PSS)

.098

.011

.492

8.899

< .001

Current relationship satisfaction (RSAT)

− .012

.004

− .136

− 2.597

.010

Current sleep disturbance (GSDS)

.016

.004

.209

3.974

< .001

Current sleep duration (actigraphy hours)

− .002

.001

− .104

− 2.179

.031

3rd trimester day length (shortening = 1; other = 0)

.109

.049

.139

2.200

.029

Current (3 months postpartum) day length:

     

 (Short = 1; other = 0)

− .388

.206

− .118

− 1.883

.061

 (Shortening = 1; other = 0)

.032

.174

.010

.183

.855

 (Long = 1; other = 0)

− .201

.155

− .074

− 1.298

.196

Dependent variable: CES-D total score (square root transformation) at 3 months postpartum; R2 = .667, F[21,188] = 16.7, p < .001

Discussion

Study findings support our first hypothesis that the trajectory of postpartum depressive symptoms over time would differ by seasonal day length at third trimester. In fact, after adjusting for multiple comparisons with linear mixed models, both models with shortening day length in the third trimester and shortening day length at birth were significant for higher depressive symptom scores over time. This is also clinically important with implications for earlier intervention during late pregnancy, but if that time point is missed, results suggest that increasing light exposure would also be effective if birth occurs in Autumn when day length is shortening. Future clinical trials research should consider the potential for achieving larger effect sizes for light therapy if it is introduced when day length is shortening at the time of birth for at-risk women.

Our second hypothesis was supported. In this sample of ethnically diverse, first-time mothers, the type of day length in the third trimester, specifically day lengths that are shortening compared to day lengths that are short, long or lengthening, were associated with concurrent depressive symptom severity. Our rate of high risk for depression was 30% (range 26–35%) in the third trimester and 22% (range 16–26%) at 3 months postpartum. The highest rate of clinically elevated depression scores (35%) was noted when the third trimester coincided with shortening day lengths (late summer into early autumn). Our findings support other research findings suggesting reduced natural light exposure due to shortening day length is associated with depression (Marqueze et al., 2015), particularly for women regardless of age (Lyall et al., 2018). Moreover, our findings support Crowley and Youngstedt (2012) who concluded that offsetting insufficient low light levels with bright light therapy may be beneficial for women at high risk of perinatal depression. Our findings go further to suggest that initiating light treatment in the late third trimester when seasonal day length is shortening may serve to minimize postpartum depressive symptoms in high-risk mothers during the first 3 months postpartum.

Our final hypothesis was also supported. Type of day length in late third trimester was predictive of postpartum depressive symptoms at 3 months even when controlling for other bio-psycho-social variables that included mental health history and third trimester depressive symptom scores. In the multivariate regression analyses, type of day length at third trimester also remained significant in explaining the variance in postpartum depressive symptoms at 3 months while also considering their current 3-month postpartum day length. Findings for type of day length at birth were similar to findings for day length at late third trimester, likely because of the short time frame between these two time points.

Given the well-documented role of sunlight in suppression of melatonin (Lewy, 2007) and in the synthesis of vitamin D (National Institutes of Health, 2016), our findings support Robinson et al. (2014) who concluded that a low vitamin D level during pregnancy was a risk factor for developing postpartum depressive symptoms. During the first 2 months postpartum, new mothers are more likely to be on maternity leave from employment, protecting their infants in the home without much light exposure. A shortening day length just before or at birth can further reduce their sunlight exposure during postpartum recovery. The lack of sunlight can result in increased melatonin levels and higher melatonin levels in the postpartum period have been associated with depression (Parry et al., 2008). By 3 months postpartum, new mothers are more likely to be returning to usual routines and employment schedules, getting more light exposure regardless of seasonal day length, and more likely to be outdoors with their infant.

Implications for clinical practice

Our study findings have several implications for clinical practice, including encouraging frequent exposure to daylight throughout the pregnancy to enhance vitamin D levels and suppress melatonin. Ensuring daylight exposure is particularly important for women living in higher latitudes where day light exposure is even shorter than in northern California. Additionally, clinicians should encourage women to get more exercise outdoors when weather and safety permit. Daily walks during daylight hours may be more effective in improving mood than walking inside a shopping mall or using a treadmill in a gym. Likewise, early morning or late evening walks may be relaxing but would be less effective in increasing vitamin D exposure or suppressing melatonin. The use of light therapy strategies (e.g. delivered by light boxes) to assure adequate light exposure may be especially important for women with a previous history of mental health problems, for women already experiencing depressive symptoms in the third trimester, for at-risk women living at high latitudes who are in the third trimester during shortening day lengths, or for women without easily accessible outdoor activity during shortening day lengths.

Although ACOG (2015) recommends that pregnant women consume vitamin D enriched foods to promote growth of fetal teeth and bone, routine prenatal 25-hydroxyvitamin D [25(OH)D] screening is not recommended (ACOG, 2011). This lack of screening is unfortunate when considering altered vitamin D metabolism in African American and Asian women with darker skin pigmentation and increased risk of vitamin D deficiency (Awumey et al., 1998; Harris, 2006; Mitchell et al., 2012). Given the association between PPD and vitamin D deficiency, clinicians should consider 25(OH)D screening and vitamin D supplementation for all women regardless of light exposure. In addition to the shortening day length, the significant contribution from other bio-psycho-social variables in the multiple regression models should not be overlooked for their clinical implications. Thus, along with strategies for enhancing light exposure, clinicians should continue to carefully evaluate new mothers for known risk factors that are present regardless of seasonal day length, such as level of stress, sleep disturbances, satisfaction in their relationships, and lack of resources that accompany low socioeconomic status. What may be challenging for clinicians to address is the new mother’s short sleep duration; more research is needed to better understand why short sleep duration had a significant role in the experience of postpartum depressive symptoms in our sample, but was less of a factor in late pregnancy. It may be that women need varying amounts of sleep during pregnancy compared to postpartum in order to feel rested and experience a more positive mood.

Strengths and limitations

Findings of this study should be interpreted with caution given the secondary analysis approach that involved combining two ethnically and socioeconomically diverse samples in longitudinal consecutive studies with similar designs and methods. The secondary analysis approach also dictated the sample size and variables available for analysis. Prior research and our analyses support the role of day length on depressive symptoms. While it would be logical to consider the body of research on seasonal affect disorder (winter depression) and hypothesize that short day length would be associated with depression, it was surprising to find that symptoms were associated with shortening day length rather than short day length. These findings need to be replicated in larger samples where the seasonal time frame could be narrowed to a more precise month when risk of PPD is at its highest. We were sufficiently powered to detect a medium effect size (d = 0.5) with 64 participants per group. However, based on our small effect size estimates (d = 0.2 to 0.4), we would likely have needed 100–200 women in each of the four types of day light groups in order to have sufficient power to detect differences in type of day length with statistical significance at p < .05 (Cohen, 1988). Our large effect sizes for the multivariate analyses (R2 > .55) were sufficient for statistical power. However, there was collinearity between depressive symptom scores and perceived stress scores that allowed perceived stress to account for a substantial portion of the variance in CES-D scores.

In our relatively large diverse sample of first-time new mothers living in northern California, we had few cases that could be categorized as high risk for PPD, but the range of 16–26% of the sample with scores ≥ 16 on the CES-D at 3 months postpartum reflects the prevalence ranges reported in prior studies, and season or latitude of those studies may partially explain their widely varying prevalence rates. While the rate of PPD is within the expected norms previously reported, it may also be lower than expected due to a potential Hawthorne effect from participants feeling special and contributing to women’s health research.

It was interesting that older age was a significant factor in the CES-D scores at third trimester, but younger age was a factor in PPD. These age group differences should be examined in future studies, and the ages of women in our sample should be noted. They averaged 29.5 years of age, while first time mothers in the United States are typically younger (26.3 years) (Mathews & Hamilton, 2016), and this age difference may limit generalizability of our findings to all first-time mothers.

Another potential limitation was the short monitoring period of 48 h to objectively estimate women’s sleep duration in addition to their sleep disturbance scores for the past week. Although both objective and self-report sleep measures were significant in the regression models, longer actigraphy monitoring periods would be desirable for estimating sleep patterns. Nevertheless, the short monitoring periods at each of the 4 study assessments captured objective sleep duration during rapidly changing maternal–infant time points, reduced participant burden, reduced the amount of missing data, and avoided variability between weekdays and weekends.

Our study was one of associations without causality or physiological mechanism inference. It is well known that African Americans and Asians synthesize vitamin D differently than other populations (Awumey et al., 1998; Harris, 2006), and our findings are limited by the lack of vitamin D measures.

While others have hypothesized that the season itself, such as winter or low light exposure, would be associated with depressive symptoms, we hypothesized that PPD would be related to the change in the pattern of day light exposure, specifically shortening day length. In most research that examines seasonality in relation to depressive symptoms, investigators either have very short assessment periods, or include winter months with shortening (early winter), short (winter), and lengthening (late winter) periods of day light. Thus, findings of an association between depressive symptom severity and shortening day length between August and early November need to be replicated at other latitudes and extended to include measures of vitamin D and melatonin in order to uncover potential mechanisms for perinatal depression.

As seen in our multivariate regression analyses, perinatal depression is a complex health problem that is influenced by many bio-psycho-social factors. Our regression models account for over half of the variance in depressive symptom scores, with shortening day length prior to birth, high perceived stress, less satisfying marital relationship, and perception of poor sleep among the most significant factors that need to be replicated and considered in future studies.

Notes

Funding

Both randomized controlled trials reported in this paper were funded by: NIH/NINR Grant #: R01 NR45345.

Compliance with ethical standards

Conflict of interest

Deepika Goyal, Caryl Gay, Rosamar Torres, Kathryn Lee, declares that they have no conflict of interest.

Human and animal rights and Informed consent

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study.

References

  1. American College of Obstetricians and Gynecologists. (2011). Committee opinion #495: Vitamin D screening and supplementation during pregnancy. Retrieved from https://www.acog.org/Clinical-Guidance-and-Publications/Committee-Opinions/Committee-on-Obstetric-Practice/Vitamin-D-Screening-and-Supplementation-During-Pregnancy. Accessed 4 April 2018
  2. American College of Obstetricians and Gynecologists. (2015). Nutrition during pregnancy: Why is vitamin D important during pregnancy and how much do I need daily? Retrieved from https://www.acog.org/Patients/FAQs/Nutrition-During-Pregnancy#vitamin. Accessed 4 April 2018
  3. American College of Obstetricians and Gynecologists. (2016). Committee opinion number 630: Screening for perinatal depression. Retrieved from https://www.acog.org/-/media/Committee-Opinions/Committee-on-Obstetric-Practice/co630.pdf?dmc=1. Accessed 4 April 2018
  4. Ancoli-Israel, S., Cole, R., Alessi, C., Chambers, M., Moorcroft, W., & Pollak, C. P. (2003). The role of actigraphy in the study of sleep and circadian rhythms. Sleep, 26, 342–392.CrossRefGoogle Scholar
  5. Awumey, E. M. K., Mitra, D. A., Hollis, B. W., Kumar, R., & Bell, N. H. (1998). Vitamin D metabolism is altered in Asian Indians in the southern United States: A clinical research center study. The Journal of Clinical Endocrinology and Metabolism, 83, 169–173.  https://doi.org/10.1210/jcem.83.1.4514 CrossRefPubMedGoogle Scholar
  6. Beeghly, M., Olson, K. L., Weinberg, M. K., Pierre, S. C., Downey, N., & Tronick, E. Z. (2003). Prevalence, stability, and socio-demographic correlates of depressive symptoms in Black mothers during the first 18 months postpartum. Maternal and Child Health Journal, 7, 157–168.  https://doi.org/10.1023/A:1025132320321 CrossRefPubMedGoogle Scholar
  7. Burns, D. D., Sayers, S. L., & Moras, K. (1994). Intimate relationships and depression: Is there a causal connection? Journal of Consulting and Clinical Psychology, 62, 1033–1043.CrossRefGoogle Scholar
  8. Cohen, J. (1988). Statistical power for the behavioral sciences (2nd ed.). New York: Lawrence Erlbaum Associates.Google Scholar
  9. Cohen, S., Kamarck, T., & Mermelstein, R. (1983). A global measure of perceived stress. Journal of Health and Social Behavior, 24, 385–396.CrossRefGoogle Scholar
  10. Corral, M., Wardrop, A. A., Zhang, H., Grewal, A. K., & Patton, S. (2007). Morning light therapy for postpartum depression. Archives of Women’s Mental Health, 10, 221–224.  https://doi.org/10.1007/s00737-007-0200-1 CrossRefPubMedGoogle Scholar
  11. Crowley, S. K., & Youngstedt, S. D. (2012). Efficacy of light therapy for perinatal depression: A review. Journal of Physiolgical Anthropology, 31, 15.  https://doi.org/10.1186/1880-6805-31-15 CrossRefGoogle Scholar
  12. Easterbrooks, M. A., Kotake, C., Raskin, M., & Bumgarner, E. (2016). Patterns of depression among adolescent mothers: Resilience related to father support and home visiting program. American Journal of Orthopscychiatry, 86, 61–68.CrossRefGoogle Scholar
  13. Fu, C. W., Liu, J. T., Tu, W. J., Yang, J. Q., & Cao, Y. (2015). Association between serum 25-hydroxyvitamin D levels measured 24 hours after delivery and postpartum depression. British Journal of Obstetrics and Gynaecology, 122, 1688–1694.  https://doi.org/10.1111/1471-0528.13111 CrossRefPubMedGoogle Scholar
  14. Golden, R. N., Gaynes, B. N., Ekstrom, R. D., Hamer, R. M., Jacobsen, F. M., Suppes, T., et al. (2005). The efficacy of light therapy in the treatment of mood disorders: A review and meta-analysis of the evidence. American Journal of Psychiatry, 162, 656–662.  https://doi.org/10.1176/appi.ajp.162.4.656 CrossRefPubMedGoogle Scholar
  15. Goyal, D., Gay, C., & Lee, K. (2009). Fragmented maternal sleep is more strongly correlated with depressive symptoms than infant temperament at three months postpartum. Archives of Women’s Mental Health, 12, 229–237.  https://doi.org/10.1007/s00737-009-0070-9 CrossRefPubMedPubMedCentralGoogle Scholar
  16. Goyal, D., Gay, C., & Lee, K. A. (2010). How much does low socioeconomic status increase the risk of prenatal and postpartum depressive symptoms in first-time mothers? Womens Health Issues, 20, 96–104.  https://doi.org/10.1016/j.whi.2009.11.003 CrossRefPubMedPubMedCentralGoogle Scholar
  17. Gur, E. B., Gokduman, A., Turan, G. A., Tatar, S., Hepyilmaz, I., Zengin, E. B., et al. (2014). Mid-pregnancy vitamin D levels and postpartum depression. European Journal of Obstetrics, Gynecology, and Reproductive Biology, 179, 110–116.  https://doi.org/10.1016/j.ejogrb.2014.05.017 CrossRefGoogle Scholar
  18. Harris, S. S. (2006). Vitamin D and African Americans. The Journal of Nutrition, 136, 1126–1129.CrossRefGoogle Scholar
  19. Howard, L. M., Oram, S., Galley, H., Trevillion, K., & Feder, G. (2013). Domestic violence and perinatal mental disorders: A systematic review and meta-analysis. PLoS Medicine, 10, e1001452.  https://doi.org/10.1371/journal.pmed.1001452 CrossRefPubMedPubMedCentralGoogle Scholar
  20. Kettunen, P., & Hintikka, J. (2017). Psychosocial risk factors and treatment of new onset and recurrent depression during the post-partum period. Nordic Journal of Psychiatry, 71, 355–361.  https://doi.org/10.1080/08039488.2017.1300324 CrossRefPubMedGoogle Scholar
  21. Lee, K. A. (1992). Self-reported sleep disturbances in employed women. Sleep, 15, 493–498.  https://doi.org/10.1093/sleep/15.6.493 CrossRefPubMedGoogle Scholar
  22. Lee, S. Y., Aycock, D. M., & Moloney, M. F. (2013). Bright light therapy to promote sleep in mothers of low-birth-weight infants: A pilot study. Biological Research for Nursing, 15, 398–406.  https://doi.org/10.1177/1099800412445612 CrossRefPubMedGoogle Scholar
  23. Lee, K. A., & DeJoseph, J. F. (1992). Sleep disturbances, vitality, and fatigue among a select group of employed childbearing women. Birth, 19, 208–213.  https://doi.org/10.1111/j.1523-536X.1992.tb00404.x CrossRefPubMedGoogle Scholar
  24. Lewy, A. J. (2007). Melatonin and human chronobiology. Cold Spring Harbor Symposia on Quantitative Biology, 72, 623–636.  https://doi.org/10.1101/sqb.2007.72.055 CrossRefPubMedGoogle Scholar
  25. Lyall, L. M., Wyse, C. A., Celis-Morales, C. A., Lyall, D. M., Cullen, B., Mackay, D., et al. (2018). Seasonality of depressive symptoms in women but not in men: A cross-sectional study in the UK Biobank cohort. Journal of Affective Disorders, 229, 296–305.  https://doi.org/10.1016/j.jad.2017.12.106 CrossRefPubMedGoogle Scholar
  26. Marcus, S. M. (2009). Depression during pregnancy: Rates, risks and consequences—Motherisk update 2008. Canadian Journal of Clinical Pharmacology, 16, e15–e22.PubMedGoogle Scholar
  27. Marcus, S. M., Flynn, H. A., Blow, F. C., & Barry, K. L. (2003). Depressive symptoms among pregnant women screened in obstetrics settings. Journal of Women’s Health (Larchmt), 12, 373–380.  https://doi.org/10.1089/154099903765448880 CrossRefGoogle Scholar
  28. Marqueze, E. C., Vasconcelos, S., Garefelt, J., Skene, D. J., Moreno, C. R., & Lowden, A. (2015). Natural light exposure, sleep and depression among day workers and shiftworkers at arctic and equatorial latitudes. PLoS ONE, 10, e0122078.  https://doi.org/10.1371/journal.pone.0122078 CrossRefPubMedPubMedCentralGoogle Scholar
  29. Mathews, T. J., & Hamilton, B. E. (2016). Mean age of mothers is on the rise: United States, 20002014. (232). National Center for Health Statistics, Division of Vital Statistics, Reproductive Statistics Branch Retrieved from https://www.cdc.gov/nchs/data/databriefs/db232.pdf. Accessed 4 April 2018
  30. Mitchell, D. M., Henao, M. P., Finkelstein, J. S., & Burnett-Bowie, S.-A. M. (2012). Prevalence and predictors of vitamin D deficiency in healthy adults. Endocrine practice: Official Journal of the American College of Endocrinology and the American Association of Clinical Endocrinologists, 18, 914–923.  https://doi.org/10.4158/EP12072.OR CrossRefGoogle Scholar
  31. Muraca, G. M., & Joseph, K. S. (2014). The association between maternal age and depression. Journal of Obstetrics and Gynaecology Canada, 36, 803–810.  https://doi.org/10.1016/S1701-2163(15)30482-5 CrossRefPubMedGoogle Scholar
  32. National Institutes of Health. (2016). Vitamin D fact sheet for health professionals. Retrieved from https://ods.od.nih.gov/factsheets/VitaminD-HealthProfessional/. Accessed 4 April 2018
  33. National Weather Service. (2017). The seasons, the equinox, and the solstices. Retrieved from http://www.weather.gov/cle/Seasons. Accessed 4 April 2018
  34. Nussbaumer, B., Kaminski-Hartenthaler, A., Forneris, C. A., Morgan, L. C., Sonis, J. H., Gaynes, B. N., et al. (2015). Light therapy for preventing seasonal affective disorder. Cochrane Database Syst Rev.  https://doi.org/10.1002/14651858.cd011269.pub2 CrossRefPubMedGoogle Scholar
  35. Parry, B. L., Meliska, C. J., Sorenson, D. L., Lopez, A. M., Martinez, L. F., Nowakowski, S., et al. (2008). Plasma melatonin circadian rhythm disturbances during pregnancy and postpartum in depressed women and women with personal or family histories of depression. American Journal of Psychiatry, 165, 1551–1558.  https://doi.org/10.1176/appi.ajp.2008.08050709 CrossRefPubMedGoogle Scholar
  36. Radloff, L. S. (1977). The ces-d scale: A self-report depression scale for research in the general population. Applied Psychological Measurement, 1, 385–401.  https://doi.org/10.1177/014662167700100306 CrossRefGoogle Scholar
  37. Robinson, M., Whitehouse, A. J., Newnham, J. P., Gorman, S., Jacoby, P., Holt, B. J., et al. (2014). Low maternal serum vitamin D during pregnancy and the risk for postpartum depression symptoms. Archives of Women’s Mental Health, 17, 213–219.  https://doi.org/10.1007/s00737-014-0422-y CrossRefGoogle Scholar
  38. Shaw, W., Dimsdale, J., & Patterson, T. (2000). Stress and life events measures. In A. J. Rush (Ed.), Handbook of psychiatric measures (1st ed., pp. 221–237). Washington, DC: American Psychiatric Association.Google Scholar
  39. Sit, D., Luther, J., Buysse, D., Dills, J. L., Eng, H., Okun, M., et al. (2015). Suicidal ideation in depressed postpartum women: Associations with childhood trauma, sleep disturbance and anxiety. Journal of Psychiatric Research, 66–67, 95–104.  https://doi.org/10.1016/j.jpsychires.2015.04.021 CrossRefPubMedGoogle Scholar
  40. Stremler, R., Sharkey, K., & Wolfson, A. (2017). Postpartum period and early motherhood. In M. Kryger, T. Roth, & W. C. Dement (Eds.), Principles and practice of sleep medicine (6th ed., pp. 1547–1552). Philadelphia, PA: Elsevier.CrossRefGoogle Scholar
  41. Torres, R., Goyal, D., Burke-Aaronson, A. C., Gay, C. L., & Lee, K. A. (2017). Patterns of symptoms of perinatal depression and stress in late adolescent and young adult mothers. Journal of Obstetric, Gynecologic, and Neonatal Nursing, 46, 814–823.  https://doi.org/10.1016/j.jogn.2017.08.002 CrossRefPubMedGoogle Scholar
  42. Wang, E. J., Kripke, D. F., Stein, M. T., & Parry, B. L. (2003). Measurement of illumination exposure in postpartum women. BMC Psychiatry, 3, 5.  https://doi.org/10.1186/1471-244x-3-5 CrossRefPubMedPubMedCentralGoogle Scholar
  43. Werner, E., Miller, M., Osborne, L. M., Kuzava, S., & Monk, C. (2015). Preventing postpartum depression: Review and recommendations. Archives of Women’s Mental Health, 18, 41–60.  https://doi.org/10.1007/s00737-014-0475-y CrossRefPubMedGoogle Scholar

Copyright information

© Springer Science+Business Media, LLC, part of Springer Nature 2018

Authors and Affiliations

  • Deepika Goyal
    • 1
  • Caryl Gay
    • 2
  • Rosamar Torres
    • 3
  • Kathryn Lee
    • 2
  1. 1.The Valley Foundation School of NursingSan Jose State UniversitySan JoseUSA
  2. 2.Family Health Care NursingUniversity of San CaliforniaSan FranciscoUSA
  3. 3.UCLA School of NursingLos AngelesUSA

Personalised recommendations