Stephanie J. Woods, PhD, RN Rosalie J. Hall, PhD Gayle G. Page, DNSc, RN, FAAN

Trajectories of Change and Predictors of Diurnal Cortisol Patterns in Women Who Are or Had Experienced Intimate Partner Violence

Stephanie J. Woods, PhD, RN

Rosalie J. Hall, PhDGayle G. Page, DNSc, RN, FAAN

The University of AkronJohns Hopkins University

Research supported by NINR/NICHHD (R01 009286)

Background

Alterations in the HPA system and cortisol dysregulation have been found in survivors of various types of trauma, and with depression and PTSD.

Diurnal patterns of cortisol levels may be attenuated or absent under chronic stress (Miller, Chen, & Zhou, 2007).

Purpose of Study

Describe trajectories of change in diurnal cortisol over two years in women who are or had experienced intimate partner violence

Examine predictors of cortisol change from:• Time-invariant co-variates: demographics, child

trauma, baseline intimate partner violence.

• Time-varying co-variates: BMI, post-traumatic stress symptoms, depression, co-morbid PTSD/depression.

Method

Latent growth curve models (LGCM) were used to estimate empirically-based change trajectories for AM cortisol and for PM cortisol over time, using longitudinal data collected at 7 equally spaced time periods over 2 years

Measures

Intimate partner violence• Physical, sexual and threats of violence (Severity of Violence Against

Women Scale, Marshall, 1992), Cronbach’s α = .90 -.94

• Emotional abuse (The Index of Spousal Abuse – Non Physical Subscale, Hudson & McIntosh, 1981); Cronbach’s α = .94

• Risk of homicide (The Danger Assessment Scale, DA; Campbell, 1995); alpha = .81

Childhood Maltreatment (Childhood Trauma Questionnaire, Bernstein & Fink, 1998); Cronbach’s α = .84

Post-Traumatic Stress Symptoms (PSS) (Foa et al., 1993); Cronbach’s α = .90

Depression: Trauma Symptom Inventory (TSI) Depression subscale (Briere, 1995); Cronbach’s α = .90

At Study Entry: 157 Women Currently in an Intimately Abusive Relationship

Mean Range

Age

Race/Ethnicity

Caucasian

African-American

Other

33.7 yrs (SD 9.5 yrs)

72 (46%)

74 (47%)

11 (6.9%)

18-64

Education High School 8th grade – BS

Annual income $10,000 - $14,999 <$10K to >50K

Length of abusive relationship

5.35 yrs (SD 6.7 yrs) <1 mo to 34 yrs

Childhood Trauma

Items ask about childhood abuse, neglect, and witnessing violence (6 events)

Women in this sample had a mean score of

3.5 out of 6

Only 12% - or 19 women – had a score of “0”

Intimate Partner Violence (Baseline)

Dimension Mean SD Observed Range

Physical violence (SVAWS)

(possible range: 21-84)

53.09 15.38 21-84

Emotional abuse (ISA-NP)

(possible range: 18-90)

63.03 18.56 18-90

Sexual violence (SVAWS)

(possible range: 6-24)

11.32 5.42 6-24

Threats of Violence (SVAWS)

(possible range: 19-76)

54.61 13.26 22-76

Risk of Homicide

(possible range: 0-21)

9.79 4.09 0-19

Mental Health Consequences:Percentage of Women Experiencing

PTSD and Depression

Baseline Year 1 Year 2

PTSD 92.4% 61.8% 52.3%

Depression 56.6% 21.2% 14.8%

Co-morbid PTSD/depression

57.7% 18.3% 12.5%

Time 1 (baseline): Mean AM and PM

Salivary Cortisol Values

AM-PM Cortisol Patterns Over Time

At each of the 7 measurement times,

63 to 79% of women

had atypical AM-PM cortisol patterns.

Five Consistent, Distinct AM-PM Cortisol Patterns Emerged

Mean AM* Mean PM*

Typical AM-PM pattern .44 .075

Extremely high AM pattern

1.198 .156

Blunted high AM-PM pattern

.681 .511

Blunted low AM-PM pattern

.203 .091

Reverse AM-PM pattern .205 .463*in ug/dL

Time 1 (baseline): Mean AM and PM

Salivary Cortisol Groups

Time 4 (1 year): Mean AM and PM Salivary Cortisol Groups

Time 7 (2 years): Mean AM and PM

Salivary Cortisol Groups

Trends in AM-PM Cortisol Pattern Changes Over 2 Years

Women with:• a typical AM-PM pattern tended to stay in a typical

pattern.

• blunted high cortisol pattern tended to stay in a blunted high pattern.

• blunted low cortisol pattern tended to stay in a blunted low pattern.

• reverse and extremely high AM patterns transitioned to multiple AM-PM cortisol patterns.

LGCM: Logistic Change Trajectories for AM and PM Cortisol*

LCGM results showed on average a relatively small but significant decline in AM cortisol and a small increase in PM cortisol over time.

*raw value, in ug/dL x 100

Generic Latent Growth Curve Model for AM and PM Cortisol

Both the AM and PM models included an intercept and a linear slope factor.

Predictors of log-transformed AM Cortisol Levels

Time-invariant predictors of Intercept parameterAge, B = -.011, se = .006, p = .08Child maltreatment, B = -.026, se = .015, p = .09

Time-invariant predictors of Slope parameterAge, B = .003, se = .002, p = .08Child maltreatment, B = .008, se = .004, p = .02.

No significant time-varying predictors

Predictors of log-transformed PM Cortisol Levels

No significant time-invariant predictors

Time-varying predictors (i.e., predicts each PM cortisol once intercept & slope effects have been removed, coefficients are equality constrained across time)

BMI (median is 28.3)B = -.015, se = .007, p = .04

Packs smoked (1 unit = ½ pack)B = .134, se = .054, p = .01.

Discussion/Significance

We identified several distinct atypical AM-PM cortisol patterns over time in a substantial number of study participants.

Overall, women in the sample had statistically significant declines over time in levels of AM cortisol and increases in PM cortisol. About half of these women left abusive relationships by 1 year post-baseline.

Discussion/Significance

Older women and those who had experienced more childhood maltreatment tended to have slightly lower levels of AM cortisol at baseline and less steep decreases of AM cortisol over time.

Women with higher BMI’s tended to have lower PM cortisol at each time period.

For every ½ pack of cigarettes smoked, PM cortisol values increased by about .01 ug/dL.