Accepted Manuscript

Title: Intimate personal violence and caregiving: Influences onphysical and mental health in middle-aged women

Authors: Authors. Pablo Ferreira, Deborah Loxton, Leigh RTooth

PII: S0378-5122(17)30136-6DOI: http://dx.doi.org/doi:10.1016/j.maturitas.2017.05.001Reference: MAT 6817

To appear in: Maturitas

Received date: 16-3-2017Revised date: 2-5-2017Accepted date: 3-5-2017

Please cite this article as: Ferreira Authors Pablo, Loxton Deborah, Tooth LeighR.Intimate personal violence and caregiving: Influences on physical and mental healthin middle-aged women.Maturitas http://dx.doi.org/10.1016/j.maturitas.2017.05.001

This is a PDF file of an unedited manuscript that has been accepted for publication.As a service to our customers we are providing this early version of the manuscript.The manuscript will undergo copyediting, typesetting, and review of the resulting proofbefore it is published in its final form. Please note that during the production processerrors may be discovered which could affect the content, and all legal disclaimers thatapply to the journal pertain.

1

Title Page

Title

Intimate personal violence and caregiving: Influences on physical and mental health in

middle-aged women

Authors

Pablo Ferreiraa,c

Deborah Loxtonb

Leigh R Tootha, corresponding author; l.tooth@uq.edu.au

aSchool of Public Health, The University of Queensland, Brisbane, Queensland, Australia,

4072

bResearch Centre for Generational Health and Ageing, University of Newcastle, Australia,

2305

cCurrent affiliation: Servicio de Salud Metropolitano Norte, Santiago Chile

2

Highlights

Little research on how IPV and caregiving affect health of mid-age women.

Experiencing either associated with poorer physical and mental health.

Experiencing both associated with worse physical and mental health.

Both associated with twice the odds of depressive symptoms and perceived stress.

Possible accumulation or additive effect from experiencing both.

Abstract

Objectives: To investigate if women with a history of having experienced intimate partner

violence (IPV) who undertook caregiving would experience worse mental and physical health

compared to those without caregiving roles.

Study design and main outcome measures: IPV, caregiving history and data on covariates

were collected between 1996-2010 from 8453 participants in the Australian Longitudinal

Study on Women’s Health aged between 45 and 65 over the course of the study. Regression

analyses were used to analyse the association of IPV and caregiving (categorised as IPV+

caregiving, IPV+no caregiving, no IPV+caregiving, no IPV+no caregiving), with and without

adjustment for covariates, on mental and physical health-related quality of life (HRQOL),

depressive symptoms and perceived stress, measured in 2010.

Results: Experiencing IPV and being a caregiver was associated with poor health outcomes

on three of the four outcomes (depressive symptoms, OR 2.08, 95% CI 1.58, 2.75; stress, OR

2.11, 95% CI 1.55, 2.87; physical HRQOL β -2.39, 95% CI -3.34, -1.44; all p≤0.001, fully

adjusted) compared with not experiencing IPV or caregiving. On these outcomes, IPV and

caregiving combined had a stronger association than IPV or caregiving separately. For mental

3

HRQOL, a weaker association was found (OR 1.41 95% CI 1.02, 1.95, fully adjusted,

p=0.04).

Conclusions: This paper provides evidence for the cumulative health impact of stressful life

events, both those that are perpetrated against an individual (violence) and those undertaken

with a degree of personal agency (caregiving). The findings underscore the need to

understand the drivers of poor health, for clinicians to ask about life circumstances of patients

experiencing poor health, and for the provision of referral pathways for complex cases.

Key words: caregiving, intimate partner violence, physical health, mental health, stress,

middle-aged women

4

1. Introduction

Informal caregiving and the experience of abuse or violence have important associations with

physical and mental health in women. Informal caregiving is care provide by family members

or friends and providing this type of care for a disabled or elderly person often peaks in mid-

life for women, and can have both positive and negative impacts on their mental health,

physical health and personal wellbeing [1]. While there are positive impacts from caregiving,

such as personal growth, strength, and resiliency [2,3], many studies show that caregivers

tend to have poorer mental [4] and physical health [5] compared to non-caregivers. Women

caregivers also experience problems related to their personal wellbeing, including their

personal development and interpersonal relationships. Indeed, research has demonstrated that

female caregivers have poorer social support [6]. The importance of considering the impact

of demographic characteristics as well as other social and economic roles performed by

caregivers on their health has been advocated [1]. Yet, in order to establish the impact of

caregiving it is also important to examine the life histories of women, to take account of past

traumatic or stressful events in order to understand the personal resources (or lack of) that

women might have when entering the caregiving relationship.

Some forms of abuse, intimate partner violence (IPV) in particular, have been associated with

poorer personal resources in women, such as an increased likelihood of financial stress [7],

lower education levels [8] and poorer social support [9]. Women who have experienced abuse

in childhood or adulthood are more likely to experience mental and physical health problems

and higher stress than other women [10-12]. Although health may improve once violence

ceases, the health and personal resource deficits attributable to violence may last for many

years, including larger utilization of medical care [13].

5

Research on the consequences of informal caregiving for women tends to focus on the current

dynamics in the relationship between caregiver and care recipient, rather than on potential

past circumstances that might influence the capacity of women to undertake caregiving roles

and remain healthy [14]. No research to date has examined the impact of a history of IPV on

women caregivers’ health.

The present study aims to examine the associations between having a history of IPV, as well

as taking on caregiving, on mental and physical health-related quality of life (HRQOL) in

mid-aged women. Firstly we examined the associations between IPV and caregiving

separately on HRQOL. Secondly we examined how IPV and caregiving in combination were

associated with HRQOL. The principal hypothesis was that women with a history of IPV

who also undertake caregiving will have poorer HRQOL than women without such history.

2. Methods

2.1 Study population

Data were from the Australian Longitudinal Study on Women’s Health (ALSWH), a

population-based study of health and well-being in Australian women. Since 1996, self-

reported data on health, health service use, socioeconomic and personal information have

been collected approximately every three years from over 41,500 women in three cohorts:

those born 1973-78 (aged 18-23 in 1996); those born 1946-51 (45-50 years); and those born

1921-26 (70-75 years). The sample was randomly selected from the Medicare Australia

database, which covers all citizens and permanent residents of Australia. Informed consent

has been obtained at each survey, with ethical clearance obtained from the University of

Newcastle and the University of Queensland. Full details of recruitment and response rates

are published elsewhere [15]. This paper includes the 1946-51 cohort, who were surveyed six

times between 1996-2010 (Survey 1, 1996, N=13,714; Survey 2, 1998, N=12,338; Survey 3,

6

2001, N=11,226; Survey 4, 2004, N=10,905; Survey 5, 2007, N=10,638; Survey 6, 2010,

N=10,011). For this paper, in order to capture all experiences of IPV data from participants

who completed all six surveys (N=8453, 61.6% of baseline sample) were used.

2.2 Measures

2.2.1 Outcomes

Two mental health measures were used. General mental health was measured with the mental

health subscale from the Short Form-36 (SF-36) [16]. The mental health subscale has 5-items

measuring anxiety, depression, emotional control and psychological health, with scores ≥53

indicating good mental health [17]. Depressive symptoms were measured using the Centre

for Epidemiological Studies Depression scale (CESD-10) [18] with scores of ≥10 indicating

possible depressive illness [19].

Physical health was measured using the Physical Component Summary score (PCS) from the

SF-36 [20]. The PCS is standardised to have a mean of 50 and standard deviation of 10.

Higher PCS scores reflect better physical HRQOL. The PCS was selected over the physical

functioning subscale of the SF-36 because it was the only measure of physical health used

and reflects a broader definition of physical HRQOL, namely, both physical function and

physical wellbeing.

Stress about own health, health of family members, work, living arrangements, study, money,

and relationships was measured with the Perceived Stress Scale. Developed using the

ALSWH 1973-78 cohort [21], it has been validated with the 1946-51 cohort [22]. Scores

range from 0 (not at all stressed) to 4 (extremely stressed), and can be dichotomised to reflect

no stress/stress.

For all outcomes, scores at Survey 6 were used.

7

2.2.2 Exposure

A variable capturing the women’s experiences of IPV and caregiving was created with four

mutually exclusive categories; IPV and caregiving, IPV and no caregiving, no IPV and

caregiving, and lastly, no IPV and no caregiving. The IPV/caregiving exposure was created

using the following survey questions.

At Survey 5 (in 2007) the women answered “Have you ever been in a violent relationship

with a partner or spouse” (yes, no), followed by “If you have ever lived with a violent partner

or spouse, in which years did you experience violence” (response – I have never lived with a

violent partner or spouse, before 1996, 1996-1998, 1999-2001, 2002-2004, 2005-now).

At each survey, the women were asked “Do you regularly provide care or assistance (eg

personal care, transport) to any other person because of their long-term illness, disability, or

frailty?”. As this question was asked at multiple survey points, numerous options capturing

caregiving status and change in status were possible. In this paper we used a summary

measure of caregiving, that had been previously created using latent class analysis of

responses to caregiving questions over Surveys 3 to 6 (data from the first two surveys were

not used due to different response options (Survey 1) and data quality issues (Survey 2)) [23].

The latent class analysis produced three trajectories (latent classes) which yielded

probabilities describing: ’consistently highest’, ‘low then increasing’, and ‘consistently

lowest’ classes of caregiving. For this paper, the ‘consistently highest’ and ‘low then

increasing’ classes were combined into one class reflecting ‘caregiving’.

8

2.2.3 Covariates

A number of covariates were used to adjust all final analyses (all covariates were assessed at

Survey 1 unless specified): relationship status (partnered, unpartnered), area of geographical

residence (major city, inner regional, outer regional/remote/very remote), perception of

ability to manage on available income (it is easy, not too bad, sometimes difficult, always

difficult/impossible), highest educational qualification (no formal, up to grade 10, grades 11

or 12, trade/certificate/diploma, degree/higher degree), social support [24] (assessed at

Survey 3) and optimism/resilience [25] (assessed at Survey 3). The earliest available

measures of each outcome were included to adjust for pre-existing status on these measures.

Thus, Survey 1 measures of the SF-36 scales and stress scale, and Survey 2 measures of the

CESD-10 were used.

2.3. Statistical analysis

Chi-square analysis and one-way ANOVA were used to investigate the differences in

characteristics between those included in, and excluded from, the study, as well to examine

bivariate associations between the exposure and the outcomes. Outcome scores on the

perceived stress scale showed moderate skewness and thus the scores were recoded into

no/somewhat stressed (89.3% of the sample) versus moderate/very/extremely stressed (11.1%

of the sample). The hypothesis was tested by stepwise logistic regression for general mental

HRQOL, depressive symptoms and perceived stress, and linear regression for physical

HRQOL. In step 1 the combined IPV and caregiving exposure was entered (unadjusted); in

step 2 the model was adjusted for covariates. To enable comparisons of regression estimates

for the IPV and caregiving exposure, complete case analyses were used in unadjusted and

adjusted analyses. Separate analyses were run for each outcome. Statistical analyses were

conducted using SAS Version 9.3 (SAS Institute, Cary, NC).

9

3. Results

3.1. Participants in the ALSWH 1946-51 cohort

Of the 13,714 original women recruited in 1996, 8453 (61.2%) participated in all six surveys

and were used in this study (Fig. 1). Table 1 describes the 8453 included and 5262 excluded

women. The included women were less likely to have experienced IPV and more likely to be

a caregiver. Regarding the covariates, included women were more likely to be partnered, live

in regional areas, be able to manage on their available income, have higher educational

qualifications, and have higher social support and greater optimism/resilience. They were less

likely to have mental health problems (on both CESD-10 and SF-36) or experience stress and

had better physical HRQOL than the excluded women.

As shown in Table 1, 13% of women had experienced IPV and 32% reported caregiving.

When combined into the four level exposure, 4.5% of women reported both IPV and

caregiving, 8.6% reported IPV but no caregiving, 27.5% reported caregiving but no IPV, and

59.1% reported neither IPV or caregiving. Bivariate analyses revealed significant associations

between the combined IPV and caregiving exposure and the mental and physical HRQOL,

depressive symptoms and stress outcomes (Table 2) at both baseline and Survey 6: with

women who had experienced IPV and caregiving showing the poorest outcomes on all

measures.

The four plots in Figure 2 show the fully adjusted results of the hypothesis testing for each

outcome. The graphs present the regression estimates (odds ratios [OR] or beta coefficients

[β]) and their 95% confidence interval. In each analysis, ‘no IPV and no caregiving’ was the

reference category. The hypothesis, that women with a history of IPV who also undertake

caregiving will have poorer health outcomes than women without such history, was

supported for the depressive symptoms, physical HRQOL and perceived stress outcomes (all

p≤0.001, fully adjusted). The hypothesis was only weakly supported for the general mental

10

HRQOL measure (p=0.04, fully adjusted) (see online Table for full unadjusted and adjusted

regression estimates). The plots in Figure 2B, 2C and 2D further show that women with a

history of IPV or caregiving (but not both) showed essentially similar regression estimates on

the outcomes: Compared to women with neither IPV nor caregiving they had poorer

outcomes, however, their outcomes were not as poor as women with both IPV and

caregiving.

For all outcomes the unadjusted regression estimates were substantially attenuated by the

addition of the covariates to the models (online Table). The greatest attenuations were from

addition of the baseline measures of each of the outcomes (regression estimate attenuations of

17% for general mental HRQOL, 19% for depressive symptoms, 40% for physical HRQOL,

46% for stress) followed by manage on income (attenuation between 18-21%) and the

personal resources of social support (attenuation between 9-13%) and optimism/resilience

(attenuation between 8-16%) (data not shown).

4. Discussion

To our knowledge, this study is the first to examine the association between a history of IPV,

subsequent caregiving, and later physical and mental HRQOL and stress using data collected

over time. The principal finding was that women with a history of IPV and caregiving were

more likely to have poorer physical HRQOL, as reflected by PCS scores an average of 2.5

units lower, and twice the odds of depressive symptoms and perceived stress than women

without such history. Experiencing either IPV or caregiving was also associated with lower

PCS scores and higher odds of depressive symptoms and stress, although the effects were not

as large as when both were experienced. Of note, essentially similar regression coefficients

were found for IPV and caregiving individually.

11

Our findings of poorer health among caregivers and those who had experienced IPV may

support our earlier stated contention that IPV might, through a reduction in internal and

external resources, lead caregivers to experience even worse health outcomes. We did not

have an apriori position on whether either IPV or caregiving would individually lead to worse

health outcomes compared to the other, and there is scant literature comparing both these life

events that could inform our thinking. Our findings, as shown graphically in Figure 2, suggest

that there may be an accumulation or additive effect from experiencing IPV and caregiving

that is greater than experiencing either in isolation.

The hypothesis was strongly supported for three of the four outcomes. The outcome with the

weakest support was the SF-36 mental health subscale. This may reflect that this is a more

general measure of mental health, with the more specific measure of depressive symptoms

(CESD-10) possibly being a more relevant outcome for the exposure.

The findings contribute to the growing body of evidence that suggests women in caregiving

roles have poorer overall health and higher stress levels than women not in a caregiving role

[4,5], and that women who experience IPV have poorer mental and physical HRQOL and

higher perceived stress than women with no history of IPV [5,9,10]. In one of the few studies

to examine abuse and caregiving, Kong and Moorman [26], found frequent depressive

symptoms in people who care for previously abusive or neglectful parents. Unlike these

authors we were not able to ascertain if the care recipient was the perpetrator of the IPV, nor

the relationship between the caregiver and care recipient, and hence were not able to do

further sensitivity analyses. Wuest et al [14] suggested that women caregivers who have been

previously abused by their spouses may assume the role of carer because of a sense of duty

and obligation. This could lead to constant emotional disturbances and impact even more on

these women’s mental and physical HRQOL [14].

12

The baseline scores on all measures indicated that women who reported IPV and/or

caregiving had poorer mental and physical HRQOL and higher depressive symptoms and

perceived stress at the beginning of the study (Table 2) than women without such history. In

particular women with IPV as opposed to caregiving seemed to have poorer health. However

even after adjusting for the women’s health at baseline, women experiencing IPV and

caregiving continued to have poorer health 14 years later. We examined this further in

descriptive sensitivity analyses. The majority of women reporting IPV (977 out of 1106,

88.3%) had experienced IPV prior to the baseline survey. These 977 women had poorer

baseline health (for e.g., 34% reported depressive symptoms versus 19% of women who did

not report IPV). The prevalence of depressive symptoms at Survey 6 for these 977 women

continued to be higher than the prevalence for women with no such history (29% versus

16%); findings consistent with others showing the long-term impact of IPV on women’s

health [13]. We found a higher percentage (45%) of women reporting baseline depressive

symptoms in the 89 reporting first IPV after baseline, ie from 1996 to 2001 (Survey 3).

However, as the first or baseline measure of depressive symptoms was from Survey 2 in 1998

this may reflect that the abuse was happening around the time the women were completing

the survey. We also descriptively examined health transitions in the outcomes between

baseline and Survey 6, and showed women experiencing both IPV and caregiving had the

highest percentage reporting the poorest health at both time points (i.e., 20% of women with

IPV+caregiving had depressive symptoms at baseline and Survey 6, compared to 7% without

IPV or caregiving, 10% with caregiving but no IPV, and 17% with IPV but not caregiving).

It is also worth noting that the associations between IPV and caregiving and health outcomes

were reduced by 9-21% when personal resources including social support and

optimism/resilience were included in the model. This may imply that the provision of

13

services that augment personal resources could assist both women who have experienced IPV

and those who provide care [27].

This study had several limitations that may have contributed to our findings. The ALSWH

was designed to capture multiple aspects of women’s health and wellbeing, and not all

variables which may play an important mediating role in the relationship between IPV and

caregiving (such as caregiver-care recipient relationship, personality factors, type of

impairment of the care recipient) were measured. A meta-analysis by Pinquart and Sorensen

[5] found behaviour problems of the care recipient to be strongly associated with caregiver

health outcomes. Relatedly, the type of care recipient’s illness may greatly impact caregiver

outcomes. Differences in psychological and physical health between caregivers and non-

caregivers have been found to be considerably larger in dementia caregivers than in studies

that included a combination of caregivers [4]. Furthermore, predictors of mental HRQOL

may differ from those of physical HRQOL [5]. Regarding generalisability, the original

ALSWH sample has been compared with the 2001, 2006 and 2011 Australian Censuses and

2005 Australian National Health Survey. This reveals that while there is some

overrepresentation among these cohorts of women with higher socioeconomic status and

better health [28], the study remains broadly representative of Australian women. The sample

included in this paper were generally healthier than those who were excluded and had a lower

prevalence of reported IPV, which may have diluted the associations between IPV and the

outcomes. Misclassification in reporting IPV and caregiving status is also possible. Use of a

single, retrospective item might have resulted in underreporting of IPV, as past research

indicates that use of multiple items leads to increased identification of IPV [29], and some

women may not self-identify as ‘caregivers’. Strengths of the study were the large

community-based sample, longitudinal data collection and wide range of covariates.

14

In conclusion, we have identified that mid-age women who experience IPV or caregiving

have poorer health outcomes than women without such history, and that women who have

experienced both have worse HRQOL than women who have experienced either or none.

Pre-existing mental and physical health, socioeconomic factors and personal resources were

also important. This paper provides early evidence for the cumulative health impact of

stressful life events, both those that are perpetrated against an individual (violence) and those

that are undertaken with a degree of personal agency (caring). The findings underscore the

need to understand the drivers of poor health, for clinicians to ask about life circumstances of

patients experiencing poor health, and for the provision of referral pathways for complex

cases. Future research could examine the onset, progress and consequences of IPV over the

lifecourse, including Post Traumatic Stress Disorder, and its impact on caregiving and health

and healthcare needs including qualitative explorations of female caregivers experiences of

IPV to inform development of appropriate support and protection for those providing care.

The study from which the data comes, the Australian Longitudinal Study on Women’s

Health, is funded by the Australian Government Department of Health. The funding supports

the collection of data. The Australian Government Department of Health had no role in the

study design, analysis and interpretation of data, writing of this paper or in the decision to

submit this for publication.

Pablo Ferreira was involved in the initial conceptualisation and design of the study, data

analysis and writing. He was then involved in the writing of later drafts and approved the

final version.

Deborah Loxton contributed to the design, analysis plan, interpretation of data and writing

and approved the final version.

15

Leigh R Tooth is senior author on this paper. She supervised Pablo Ferreira and contributed

to the initial design, acquisition, analysis and interpretation of data. She conducted the data

analysis and led the writing during later stages. She approved the final version.

Acknowledgements

The research on which this paper is based was conducted as part of the Australian

Longitudinal Study on Women’s Health by the University of Newcastle and the University of

Queensland. We are grateful to the Australian Government Department of Health for funding

and to the women who provided the survey data. Pablo Ferreira was supported by a

University of Queensland 2012/2013 Summer Research Program Scholarship. The authors

thank Ms Ariel Lackoff and Ms Natalie Townsend for their assistance with literature reviews.

Pablo Ferreira was involved in the initial conceptualisation and design of the study, data

analysis and writing. He was then involved in the writing of later drafts and approved the

final version.

Deborah Loxton contributed to the design, analysis plan, interpretation of data and writing

and approved the final version.

Leigh R Tooth is senior author on this paper. She supervised Pablo Ferreira and contributed

to the initial design, acquisition, analysis and interpretation of data. She conducted the data

analysis and led the writing during later stages. She approved the final version.

16

References

1. A. Vlachantoni, M. Evandrou, J. Falkingham, J. Robards, Informal care, health and

mortality, Maturitas. 74 (2013) 114-118.

2. G. G. Fisher, M. M. Franks, B. L. Plassman, S. L. Brown, G. G. Potter, D. Llewellyn, M

A. Rogers, K M. Langa, Caring for individuals with dementia and CNID: Findings from

the Aging, Demographic and Memory Study, J Am Geriatr Soc. 59(3) (2011) 488-494.

3. W. K. Wong, J. Ussher, J. Perz, Strength through adversity: bereaved cancer carers'

accounts of rewards and personal growth from caring, Palliat Support Care. 7(2) (2009)

187-196.

4. M. Pinquart, M. Sörensen, Differences between caregivers and noncaregivers in

psychological health and physical health: a meta-analysis, Psychol Aging. 18(2) (2003)

250-267.

5. M. Pinquart, M. Sörensen, Correlates of physical health of informal caregivers: a meta-

analysis, J Gerontol B Psychol Soc Sci. 62(2) (2007) 126-137.

6. C. Lee, Health, stress and coping among women caregivers: A review, J Health Psychol.

4(1) (1999) 27-40.

7. T. Lindhorst, M. Oxford, M. R. Gillmore, Longitudinal effects of domestic violence on

employment and welfare outcomes, J Interpers Violence. 22(7) (2007) 812-828.

8. K. Nasir, A. A. Hyder, Violence against pregnant women in developing countries:

Review of evidence, Eur J Public Health. 13(2) (2003) 105-107.

17

9. C. Zlotnick, D. M. Johnson, R. Kohn, Intimate partner violence and long-term

psychosocial functioning in a national sample of American women, J Interpers Violence.

21(2) (2006) 262-275.

10. L. Bensley, J. Van Eenwyk, K. W. Simmons, Childhood family violence history and

women's risk for intimate partner violence and poor health, Am J Prev Med. 25(1) (2003)

38-44.

11. H. Ellsberg, H. A. Jansen, L. Heise, C. H. Watts, C. Garcia-Moreno, Intimate partner

violence and women's physical and mental health in the WHO multi-country study on

women's health and domestic violence: an observational study, Lancet. 371(9619) (2008)

1165-1172.

12. E.A. Greenfield. Child abuse as a life-course social determinant of adult health. Maturitas

2010, 66, 51-55.

13. F.P. Rivara, M. L. Anderson, P. Fishman, A. E. Bonomi, R. J. Reid, D. Carrell, R. S.

Thompson, Healthcare utilization and costs for women with a history of intimate partner

violence.Am J Prev Med. 32(2) (2007) 89-96.

14. J. Wuest, M. J. Hodgins, J. Malcolm, M. Merritt-Gray, P. Seaman, The effects of past

relationship and obligation on health and health promotion in women caregivers of adult

family members, Adv Nurs Sci. 30(3) (2007) 206-220.

15. A.J. Dobson, R. Hockey, W.J. Brown, J.E. Byles, D. J. Loxton, D.P. McLaughlin, L.R.

Tooth, G.D. Mishra, Cohort profile update: Australian longitudinal study on women's

health. Int J Epidemiol. 44 (2015) 1547 a-f.

16. J. Ware, K. Snow, M. Kosinski, B. Gandek, SF-36 health survey: Manual and

interpretation guide. New England Medical Centre, The Health Institute, Boston (MA),

1993.

18

17. D. Berwick, J. Murphy, P. Goldman, J. Ware, A. Barsky, M. Weinstein, Performance of a

5-item mental health screening test, Med Care. 29 (1991) 169-76.

18. R.E. Roberts, S.W. Vernon, The centre for epidemiologic studies depression scale: its use

in a community sample, Am J Psychiatry. 140 (1983) 41–46.

19. C. France, C. Lee, J. Powers, Correlates of depressive symptomatology in a representative

sample of young Australian women, Aust Psychol. 39 (2004) 228-237.

20. J. Ware, K. Snow, M. Kosinski, S.D. Keller, SF-36 physical and mental health summary

scales: A user’s manual. New England Medical Centre, The Health Institute, Boston

(MA), 1994.

21. S. Bell, C. Lee, Perceive stress revisited: the women’s health Australia project younger

cohort, Psychol Health Med. 8(3) (2003) 343-353.

22. Women's Health Australia, Technical report 16 on Australian longitudinal study on

women's health. Research Centre for Gender and Health, University of Newcastle,

Australia. http://www.alswh.org.au/publications-and-reports/technical-reports, 2001

(accessed 22 September 2016).

23. L. Tooth, G. Mishra, Socioeconomic factors associated with trajectories of caring by

young and mid-aged women: A cohort study, BMC Public Health. 14(1) (2014) 74.

24. C. D. Sherbourne, A. L. Stewart, The MOS social support survey, Soc Sci Med. 32(6)

(1991) 705-714.

25. M. F. Scheier, C. S. Carver, Optimism, coping, and health: assessment and implications

of generalized outcome expectancies, Health Psychol. 4(3) (1985) 219-247.

26. J. Kong, S. M. Moorman. Caring for my abuser: Childhood maltreatment and caregiver

depression, Gerontologist. (2013) doi: 10.1093/geront/gnt136.

19

27. S. Ahn, A. K. Hochhalter, D.K. McMaughan Moudouni, M. L. Smith, M. G. Ory, Self-

reported physical and mental health of older adults: The roles of caregiving and resources.

Maturitas. 71 (2012) 62-69.

28. Women's Health Australia, Technical report 38 on Australian longitudinal study on

women's health. Research Centre for Gender and Health, University of Newcastle,

Australia. http://www.alswh.org.au/publications-and-reports/technical-reports, 2015

(accessed 22 September 2016).

29. M. de Vries Robbé, L. March, J. Vinen, D. Horner, G. Roberts. Prevalence of domestic

violence among patients attending a hospital emergency department. Aust N Z J Public

Health. 20(4) (1996) 364–368.

20

Figure 1 Caption: Flow diagram of samples for analysis of general mental health, depressive

symptoms, physical HRQOL and perceived stress.

Explanations of abbreviations used in Figure 1

IPV Intimate Partner Violence

SF-36 MH Short Form-36 mental health subscale

Women who completed Survey 1,

n = 13714

Withdrawn n = 1192

Deceased n = 487

Uncontactable / Did not return

survey/s n = 3582

Women who completed all Surveys

1 to 6, n = 8453

Final sample for analysis, n = 8425

Missing data on IPV, n = 28

N for analysis of SF-36

MH Subscale = 8108

(Missing data on

covariates, n = 317)

N for analysis of CESD-

10 Scale = 7673

(Missing data on

covariates, n = 752)

N for analysis of SF-36

PCS = 7686

(Missing data on

covariates, n = 739)

N for analysis of Stress

Scale = 8111

(Missing data on

covariates, n = 314)

21

SF-36 PCS Short Form-36 physical component summary score

CESD-10 Centre for Epidemiological Studies Depression scale – 10-item

22

Figure 2 Caption: Regression estimates and their 95% confidence interval for the effect of

the intimate partner violence and caregiving exposure on the outcomes - 2A General mental

health; 2B Physical HRQOL; 2C Depressive symptoms; 2D Perceived stress.

Explanations of abbreviations used in Figure 2

IPV Intimate Partner Violence

SF-36 MH Short Form-36 mental health subscale

SF-36 PCS Short Form-36 physical component summary score

CESD-10 Centre for Epidemiological Studies Depression scale – 10-item

OR Odds ratio

β Regression coefficient

95% CI 95 percent confidence interval

23

Table 1. Demographic, socioeconomic and health characteristics for ALSWH women born

1946-51 who were included and excluded in the analysis, and analyses of these

characteristics by inclusion status.

Included

N=8453

(61.2%)

Excluded

N=5262

(38.4%)

Analysis of

difference

between samplesc

Exposures N (%) N (%)

Experienced IPV

Yes

No

Missing

1106 (13.1%)

7319 (86.7%)

28 (0.3%)

427 (8.1%)

1724 (32.8%)

3111 (59.1%)

χ2(1) =62.5**

Caregiving

Yes

No

Missing

2712 (32.1%)

5741 (67.9%)

0

952 (18.1%)

2877 (54.6%)

1433 (27.2%)

χ2(1) =65.6**

Combined IPV and caregiving

IPV, caregiving

IPV, no caregiving

No IPV, caregiving

No IPV, no caregiving

Missing

380 (4.5%)

726 (8.6%)

2323 (27.5%)

4996 (59.1%)

28 (0.3%)

130 (2.5%)

297 (5.6%)

492 (9.3%)

1232 (23.4%)

3111(59.1%)

χ2(3) =71.5**

Covariatesa

Relationship status

Partnered

7179 (84.9%)

4132 (78.5%)

χ2(1) =86.9**

24

Unpartnered

Missing

1242 (14.7%)

32 (0.3%)

1094 (20.8%)

36 (0.6%)

Area of residence

Major city

Inner regional area

Outer regional/remote

Missing

2967 (35.1%)

3340 (39.5%)

2145 (25.4%)

1 (0%)

2033 (38.6%)

1874 (35.6%)

1353 (25.7%)

2 (0%)

χ2(2) =24.2**

Ability to manage on income

Impossible/Difficult always

Difficult sometimes

Not too bad

Easy

Missing

1036 (12.3%)

2328 (27.5%)

3625 (42.9%)

1430 (16.9%)

34 (0.4%)

994 (18.9%)

1594 (30.3%)

2017 (38.3%)

605 (11.5%)

52 (0.9%)

χ2(3) =185.7**

Highest educational qualification

No formal

≤ 10 years

11-12 years

Trade/certificate/diploma

Degree/Higher degree

Missing

1236 (14.6%)

2675 (31.6%)

1375 (16.3%)

1767 (20.9%)

1343 (15.9%)

57 (0.7%)

1246 (23.7%)

1642 (31.2%)

912 (17.3%)

832 (15.8%)

549 (10.4%)

81 (1.5%)

χ2(4) =264.0**

Social support mean score (SD)

Missing N (%)

3.8 (1.1)

100 (1.2%)

3.7 (1.1)

2571 (48.9%)

F(1,11042)=43.3**

Optimism/resilience mean score (SD)

Missing N (%)

15.9 (3.9)

77 (0.9%)

14.9 (4.2)

2577 (48.9%)

F(1,11059)=102.7**

25

Outcomesb

SF-36 PCS, mean score (SD)

Missing N (%)

49.4 (10.1)

127 (1.5%)

47.6 (10.9)

3750 (71.2%)

F(1,9836)=41.6**

SF-36 Mental Health score, N (%)

≤ 52 (poor mental health)

≥ 53 (good mental health)

Missing N (%)

897 (10.6%)

7541 (89.2%)

15 (0.2%)

285 (5.4%)

1265 (24.0%)

3712 (70.5%)

χ2(1) =75.5**

CESD-10 score, N (%)

>10 (depressive symptoms)

≤ 10 (no depressive symptoms)

Missing N (%)

1438 (17.0%)

6795 (80.4%)

220 (2.6%)

383 (7.3%)

1103 (20.9%)

3776 (71.8%)

χ2(1) =57.1**

Perceived stress, N (%)

Not stressed

Stressed

Missing N (%)

7560 (89.8%)

857 (10.2%)

36 (0.4%)

1320 (85.4%)

225 (14.6%)

3717 (70.6%)

χ2(1) =25.9**

IPV = Intimate partner violence; SF-36 PCS = Short Form-36 physical component summary

score; SF-36 MH = Short Form-36 mental health subscale; CESD-10 = Centre for

Epidemiological Studies Depression scale – 10-item; a covariates measured at survey 1 except

for social support and optimism/resilience which were measured at Survey 3; b all outcomes

measured at Survey 6; c χ2 chi-square analyses; ** p≤0.001.

26

Table 2. Descriptive physical health, stress and mental health scores at baseline and survey 6 for women by intimate partner violence status and

caregiving exposure (N=8453)

Outcome variables: Scores at

baseline and Survey 6

Exposure

IPV, caring IPV, no caring No IPV, caring No IPV, no caring

SF-36 Mental Health, n (%)

Baseline

≤52 (poor mental health)

≥53 (good mental health)

78 (20.7%)

298 (79.3%)

139 (19.2%)

584 (80.8%)

264 (11.4%)

2047 (88.6%)

531 (10.7%)

4435 (89.3%)

χ2(3) =71.3**

Survey 6

≤52 (poor mental health)

≥53 (good mental health)

66 (17.5%)

312 (82.5%)

118 (16.3%)

607 (83.7%)

237 (10.2%)

2083 (89.8%)

469 (9.4%)

4518 (90.6%)

χ2(3) =51.4**

CESD-10, n (%)

Baseline

>10 (depressive symptoms)

130 (35.7%)

239 (34.7%)

478 (21.4%)

882 (18.3%)

χ2 (3) =144.6**

27

≤10 (no depressive symptoms) 234 (64.3%) 450 (65.3%) 1750 (78.5%) 3939 (81.7%)

Survey 6

>10 (depressive symptoms)

≤10 (no depressive symptoms)

127 (34.2%)

244 (65.7%)

195 (27.4%)

517 (72.6%)

404 (17.9%)

1859 (82.1%)

704 (14.5%)

4159 (85.5%)

χ2 (3) =151.6**

SF-36 PCS, mean (95% CI)

Baseline

49.7 (48.8, 50.5)

50.1 (49.5, 50.7)

52.7 (52.3, 52.9)

53.3 (53.1, 53.6)

F(3,8015)=49.9**

Survey 6 44.6 (43.5, 45.6) 46.3 (45.6, 47.0) 49.3 (48.9, 49.7) 50.3 (50.0, 50.6) F(3,8295)=65.8**

Perceived stress, N (%)

Baseline

Not stressed

Stressed

240 (63.3%)

139 (36.7%)

491 (67.6%)

235 (32.4%)

1902 (82.2%)

413 (17.8%)

4269 (85.8%)

708 (14.2%)

χ2 (3) =241.0**

Survey 6

Not stressed

Stressed

286 (75.9%)

91 (24.1%)

599 (83.1%)

122 (16.9%)

2046 (88.3%)

271 (11.7%)

4607 (92.6%)

368 (7.4%)

χ2 (3) =164.5**

CESD-10 = Centre for Epidemiological Studies Depression scale – 10-item; SF-36 PCS = SF-36 Short Form-36 physical component summary

score; ** p≤0.001.