Article

The Family Journal: Counseling and

Therapy for Couples and Families
Distress in Caregivers of a Family 2015, Vol. 23(3) 286-294
ª The Author(s) 2015
Reprints and permission:
Member With Serious Mental Illness sagepub.com/journalsPermissions.nav
DOI: 10.1177/1066480715572967
tfj.sagepub.com

Allison Crowe1 and Jason Brinkley2

Abstract
It is well documented that mental illness influences not only the person diagnosed but also the family. The following study
explored distress in 169 caregivers of a family member with serious mental illness (SMI). Factors such as time spent caregiving,
type of familial relationship, mental health treatment, and income were explored as they associated to caregiver depression and
family discord. Regression models suggested that caregiving for a son or daughter or devoting 20 hr a week or less to caregiving
was associated with lower distress levels. Higher numbers of difficult behaviors exhibited by the person diagnosed were asso-
ciated with higher family discord. The Mental Illness Distress Scale demonstrated some capacity to measure distress related to
having a family member with SMI.

Keywords
serious mental illness, caregivers, Mental Illness Distress Scale

Despite developments in the mental health care of persons with a mental illness family functioning, family closeness, cop-
diagnosed with serious mental illness (SMI; including major ing, distress, time spent caregiving, amount of and satisfaction
depression, schizophrenia, bipolar disorder, obsessive compul- with treatment received, and communication were examined.
sive disorder [OCD], panic disorder, posttraumatic stress disor- Related specifically to caregiving, hours a week spent caregiv-
der [PTSD], and borderline personality disorder, National ing negatively predicted family flexibility, communication in
Alliance on Mental Illness [NAMI], 2012), family members are the family, and family satisfaction. There was a positive rela-
still doing much of the caregiving for their relatives with SMI, tionship between hours a week caregiving and distress. A key
including both practical assistance such as attending doctor’s finding was that family caregivers who reported higher levels
appointments, assisting with daily living skills, and providing of distress were less likely to report better family functioning,
direct care and emotional support (Abraham & Stein, 2013; family communication, and satisfaction. Mental health treat-
Crowe & Lyness, 2014). In fact, some have posited that the role ment satisfaction was related to higher levels of family support
of the family in the care of persons with SMI has increased over and coping as well as lower levels of distress, which were then
time (Ohaeri, 2003). Caregiving for a person with SMI is a linked to more balanced family flexibility and higher family
demanding and oftentimes distressing role, possibly harmful satisfaction. Readers interested in looking more deeply at the
to the caregiver’s overall quality of life (Struening et al., results related to functioning, closeness, coping, satisfaction,
2001). Effects of caregiving can range from influencing the and communication are encouraged to consult the article, as
overall functioning of the family, decreasing family finances, reporting all results is beyond the scope of this article.
and feeling stigmatized as a result of others judging or stereo- In a study (Struening et al., 2001) of 461 family caregivers,
typing mental illness (Ohaeri, 2003). Marsh (2001) has used the the extent to which participants believed that people devalued
term family burden to describe the experience of caregiving for persons with SMI and their caregivers was explored. Approxi-
a family member with SMI as ‘‘the overall level of distress mately 70% of caregivers indicated a belief that most people
resulting from the illness’’ (p. 22), including both subjective devalue persons with SMI, and 43% expressed a belief that
and objective distress. In addition to the more negative effects,
authors have also cited positive outcomes including increased
resilience, family adaptation, and coping, as well as greater 1
Department of Interdisciplinary Professions, East Carolina University,
sensitivity to others with disabilities, and a sense of family Greenville, NC, USA
2
inner strength (Pickett-Schenk et al., 2006). Department of Biostatistics, East Carolina University, Greenville, NC, USA
Given the notion that caregiving can be a distressing task,
Corresponding Author:
caregivers of a person with SMI have been studied in recent Allison Crowe, Department of Interdisciplinary Professions, East Carolina
years in order to understand their specific needs. In a study University, 225 Ragsdale Hall, Greenville, NC 27858, USA.
of 165 family caregivers (Crowe & Lyness, 2014) of someone Email: [email protected]
Crowe and Brinkley 287

most people also devalue the families of those with SMI. The findings offer some important insights into how family care-
authors (Struening et al., 2001) concluded that interventions givers are functioning, so a study within the United States
that support these families are in demand. would shed light on whether or not the same distress might
Much of the most recent literature particularly related to dis- be found in a U.S. sample.
tress in caregiving for family members with SMI has been con-
ducted outside the United States. In Greece, researchers The Current Study
(Mitsonis et al., 2012) examined a variety of factors associated
with family caregivers of individuals with schizophrenia Much of the abovementioned literature on family caregivers
including type of familial relationship (parent, spouse/partner, took place outside the United States (Mackay & Pakenham,
or sibling), gender, age, education, and living apart or together 2012; Mitsonis et al., 2012), so a similar investigation within the
with the family member with SMI, and characteristics of the United States is warranted. Similar to the Mitsonis et al. study,
person with SMI were also assessed. In order to assess severity the current investigation explored a variety of familial relation-
of the illness, the researchers recorded clinical features, for ships of caregivers of a family member with SMI, as well as the
example, duration, positive, and negative symptoms related symptoms of the person diagnosed, in order to see how they
to the mental illness. These clinical features significantly pre- might impact one another. The current study investigated the
dicted distress in caregivers. Sociodemographic characteristics amount of and satisfaction with mental health treatment
such as age, education level, type of familial relationship, and received as variables that might impact distress as well various
whether or not the caregiver cohabitated with the family mem- types of mental illnesses. Similar to Mackay and Pakenham, the
ber were not associated with caregivers’ distress with the excep- current research looked at sociodemographic factors across a
tion of gender. Female caregivers experienced significantly variety of caregivers, as well as time spent caregiving and symp-
higher levels of distress than did male caregivers. A study that toms of the family member with SMI, but within the United
replicates this type of investigation of caregiver demographics States. This study will extend the literature base with an inves-
and how this relates to caregiver distress is needed within the tigation that will include a variety of familial relationships, time
United States to see if results might be similar. spent caregiving, income, amount of and satisfaction with treat-
A similar study in Australia (Mackay & Pakenham, 2012) ment received, illness symptoms in a sample of caregivers for a
investigated a model for stress and coping for caregivers family member with SMI within the United States.
(parent, spouse, or other family/friend) of an adult with SMI. The question that framed this research was what is the
Predictors such as stress, coping, distress, and sociodemo- effect of familial relationship type, time spent caregiving,
graphic factors of caregivers and the person with SMI, relation- income, amount of and satisfaction with treatment received,
ship quality, as well as positive caregiver outcomes (e.g., negative symptoms, and difficult behaviors (e.g., unpredict-
positive affect, satisfaction) were examined. The model was able, suspicious) on family distress in families with SMI
derived from Lazarus and Folkman’s (1984) theory on stress member? Due to the small body of literature on these topics
and coping commonly used for identifying factors associated in the U.S. samples, this study was exploratory and included
with adjustment to caregiving. Relationships between stress a nondirectional hypothesis. Of particular concern was how
and coping and negative and positive caregiver adjustment out- caregiver distress might relate to mental illness symptoms
comes were examined. Sociodemographics of both caregivers of the person being cared for? By assessing mental illness
and the family member with SMI were examined including symptoms, relationships between these and the caregiver’s
gender, age, income, employment, marital status, and type of level of distress could be explored.
familial relationship. Time spent caregiving and whether or not
the caregiver was living with the family member with SMI Method
were also assessed. Amount of and satisfaction with treatment
were not assessed. Distress (as measured using the Depression
Participants
Anxiety and Stress Scale [DASS]-21; Lovibond & Lovibond, The total sample included 169 participants who self-identified
1995) was operationalized as anxiety and depression, and care- as a caregiver for at least one family member with SMI. The
givers indicated the amount of distress using a Likert-type majority (83.4%) of respondents were female, Caucasian
scale. Unlike the Mitsonis et al. (2012) research, none of the (88.2%), highly educated (76% with at least a four year college
sociodemographic characteristics of the caregivers were related degree), and approximately 70% of the sample had an income
to caregiver adjustment, and only one of the characteristics of level of US$51,000 or more. One hundred and twenty-two indi-
the family member with SMI (symptom unpredictability) was viduals (72%) were Caucasian females. Ten participants (6%)
related to higher distress and poorer health (as measured by a identified as African American, four (2.4%) as Hispanic, and six
4-point scale ranging from poor to excellent; Pakenham, (3.4%) as multiracial or other. Approximately 65% of partici-
Dadds, & Terry 1994). Daily caregiving was associated with pants were parents of a family member with SMI (25% caring
higher distress and poorer health. Higher objective burden for daughters and 40% caring for sons). Many of the family
(as measured by the objective burden subscale of the Burden members with SMI were fairly young with median age falling
Assessment Scale [BAS]; Reinhard, Gubman, Horwitz, & between 33 and 42. Over half of the family members with SMI
Minsky, 1994) was associated with higher distress. These had been diagnosed with a mental illness for more than 10 years
288 The Family Journal: Counseling and Therapy for Couples and Families 23(3)

Table 1. MIDS Frequencies and Intensity Descriptives.

Frequency Intensity

n M SD n M SD

Irrit 169 3.12 0.93 169 2.69 1.02

Lone 169 2.83 1.12 169 2.14 0.99
Anger 168 2.64 1.06 169 2.48 1.21
Insecurity 168 2.54 1.13 168 2.12 1.08
Neglect 167 2.29 1.14 167 1.97 0.98
Worry 168 2.78 1.36 169 2.33 1.26
Phys Ab 168 1.58 1.03 168 1.55 0.98
Embarr 169 2.2 1.11 166 2.01 1.01
Money Short 168 3.04 1.42 169 3.3 1.37
Lack of Stim Ad 169 2.41 1.25 168 2.07 1.09
Verb Abuse 169 2.25 1.32 169 2.04 1.27
UN Sat Rom 168 2.32 1.37 167 1.95 1.16

Note: irri ¼ irritability; Phys Ab ¼ physical abuse; Embarr ¼ embarrassment; Money short ¼ money shortages; Stim Ad ¼ stimulating adult; UN Sat Rom ¼
unsatisfactory romantic life.

(n ¼ 106, 62.7%). As well, the sample was caregiving for family the MIDS (Crowe & Lyness, 2014). This measure is a 24-item
members with serious concerns—the majority of disorders modification of the Drinker’s Partner Distress Scale (DPDS, Crisp
included psychotic disorders or mood, anxiety, and/or bipolar & Barber, 1995) used recently to measure distress related to care-
disorder (n ¼ 157, 71.9%). Seventy-five percent of respondents giving for someone with a mental illness (Crowe & Lyness, 2014).
indicated that they felt emotionally close to the family member. Participants respond on a 5-point Likert-type scale ranging from
When asked about perceptions of time and energy spent car- Never to Always (frequency) and No Distress to Extreme Distress
egiving, 85% of respondents identified that they spend at least (intensity). Subdomain (irritability, anger, insecurity, neglect,
some significant time and 90% identified with at least some worry, physical abuse, embarrassment, money shortages, lack
significant energy toward caregiving. Actual hours per week of stimulating adult company, verbal abuse, and unsatisfactory
caregiving was assessed, and participants indicated that they romantic life) scores are found by multiplying frequency of
spent less than or equal to 20 hr/week (n ¼ 99, 59.3%), 20– occurrence by intensity of event. Subdomains combine into
30 hr/week (n ¼ 37, 22.2%), 30–40 hr/week (n ¼ 13, 7.8%), two overall domains—family discord and depression. Thus,
and more than 40 hr/week (n ¼ 18, 10.8%). Caregivers indi- the MIDS offers researchers a measure of distress in the family
cated that over 98% of their family members with SMI were context since it includes both psychological distress (depres-
currently in some sort of mental health treatment, with 63% sion) and distress in the family context (family discord). Range
seeking ‘‘a lot’’ of treatment. Eighty-five percent of survey of possible scores is 1–25 for each subdomain. Directions read
respondents were satisfied with current treatment efforts. Unfortunately, an individual’s struggles with mental illness
often create problems for other people as well—especially the
person’s family and close friends. The scale has demonstrated
Instrumentation strong internal consistency (a ¼ .86 depression; a ¼ .83 marital
discord) in the original research (Crisp & Barber, 1995) as well
Participants answered a series of demographic and descriptive as more recent studies (overall a ¼ .94; Crowe & Lyness,
questions about their experiences with their family member with 2014). In the current sample, the scale’s internal consistency
SMI. These asked about the relationship (family relationship type, was .84). For this article, it was decided that a full psycho-
and how close the caregiver felt to the family member) and metric analysis was not essential to the current exploratory
descriptive questions about the family member with SMI (age, study. While it has not yet been verified whether the domain
diagnosis, perception of severity of mental illness, treatment scores in the MIDS work in the exact same manner as the orig-
received and satisfaction with treatment, time spent, and energy inal DPDS, a clinical assumption is being made in order to
toward caregiving). Income was assessed by asking participants examine the clinical utility of a repurposed instrument.
to indicate their approximate household income. Table 1 also
includes categories for demographic information used in data
analyses. Consistent with previous research (Crowe & Lyness, Experience of Caring Inventory. The Experience of Caring Inven-
2014), caregiving was defined as going to doctor’s appointments, tory (ECI; Szmukler et al., 1996) is a 66-item inventory that
assisting with daily living skills, and providing direct care. measures the experiences of caregivers of someone with SMI.
Used recently to assess caregivers’ experiences (de Araújo
Mental Illness Distress Scale (MIDS). Participants indicated their Jorge & Chaves, 2012; Grover et al., 2012), 10 subscales mea-
overall level of distress related to mental illness by completing sure factors related to the mental illness such as the difficult
Crowe and Brinkley 289

behaviors subscale (e.g., unpredictable, suspicious) and negative replication where the data were resampled 100 times and the
symptoms subscale (e.g., withdrawn, unreliable). Other sub- same final models were fit to the resampled data. The interested
scales measure factors related to the actual caregiver such as reader can see Harrell (2001) for a full discussion of the mod-
stigma (e.g., covering up the person’s mental illness), problems eling strategies employed here. R-square values are reported
with services (e.g., dealing with psychiatrists), and effects on the for each model and effect sizes were calculated for the predic-
family (e.g., the illness causing a family break up). All subscales tors in the final model and reported here in the form of partial
have demonstrated strong internal consistency (a ranging from eta-square values, where each variable is considered as being
.74–.91), and a full validation study is available for those added last to the model. This way we have a fair comparison
wishing more detail (Szmukler et al., 1996). for relative contribution to the statistical model that is irre-
For the purposes of this research, two subscales from the spective of the type of variable under consideration. A level
ECI—difficult behaviors (a ¼ .91) and negative symptoms of significance of .05 was used for all tests; analyses were per-
(a ¼ .89) were used in order to see how these might relate to formed using a combination of JMP (Version Pro 10.0) and
the caregiver’s level of distress. SAS (Version 9.3) software.

Procedures Mental Illness Distress Results

After institutional review board approval, a survey was sent Two types of questions, frequency and intensity, comprise the
using e-mail to all members of NAMI in a state branch in the MIDS. In the current study, evidence for internal consistency of
Southeast region of the United States. NAMI is the largest the MIDS is clear for both types of subscales, that is, frequency
grassroots mental health organization in the United States that (a ¼ .87) and intensity (a ¼ .75), and the total scale a level was
aims to build better lives for those with mental illness. Advo- .84, thus demonstrating the scale’s capacity for conducting
cacy is a large part of the organization’s efforts for access to research. Across the overall MIDS instrument, many of the
treatment, support, and research. The survey was housed at a individual domain median scores were smaller than the means,
commercial online site for electronic survey research. Since suggesting that many of caregivers scored low on each individ-
participants were invited to participate through online listservs, ual domain. This phenomenon occurred for all domains, with
an actual response rate could be calculated. There were approx- the exception of money shortage and overall depression
imately 1,478 members on the listserv at the time the survey domain. In addition, the authors chose to use the ECI difficult
was sent, but it is unclear how many of the surveys made it behaviors domain as a surrogate index for the severity level of
to the intended targets. Our estimate of response rate, given the the individual with SMI. For the two overall domains of the
sample size of 169, is 11%. In order to qualify, participants MIDS, depression and family discord, the caregivers who pro-
self-identified as caregiving for a family member with SMI. vided 20 hours of care or more each week had significantly
more overall depression (t test p value ¼ .0003) as well as fam-
Results ily discord (t test p value ¼ .0006) than those who provided less
than 20 hr/week. Relationship, amount of and satisfaction with
Planned Analysis treatment received, and income level were not associated with
While a variety of information was gathered from the partici- significant differences in either family discord or overall
pants for this study, here we will focus on exploring associa- depression. Table 2 presents these results. Chi-square analysis
tions between mental illness distress (as measured by the revealed associations between the amount of caregiving time
MIDS scores) and the following seven independent variables: and the type of familial relationship. Among caregivers who
relationship between caregiver and patient, time spent caregiv- care for their children, 49% of such individuals devoted more
ing per week, income level, caregiver perceived treatment than 20 hr of care per week. By contrast, among individuals
amount, caregiver perceived treatment satisfaction, and nega- who gave care to someone who was not their son or daughter,
tive symptoms and difficult behaviors (two subscales of the only 28% of such individuals devoted more than 20 hr of care
ECI). In addition to simple data summaries and descriptive per week (w2 p value ¼ .0104).
analyses, bivariate analyses consisted of a combination of In addition to looking at the actual time in hours, percep-
chi-square tests, t tests, or one-way ANOVA (as appropriate). tions of time and energy were also assessed by asking about
While some of the results of those analyses are presented time (no time, little, some, and lots) and energy (no energy,
here, the major focus of inferences come from multiple linear little, some, and lots). Most caregivers who gave more than
regression models. The authors started with a seven variable 20 hr a week perceived that they spent at least some time
‘‘full model’’ and reduced to a final three-variable model (in most cases a lot) and energy in caregiving, however, more
manually using a backward selection manner with a combina- than one in five respondents who devoted less than 20 hr a
tion of nonsignificant p values and/or large variance inflation week endorsed that they gave ‘‘lots of time’’ and almost half
factors. It was decided that if a potential predictor had value perceived that they give ‘‘lots of energy.’’ This finding sug-
in either outcome model, then it would be a predictor for both gests that there was an imbalance between quantified time
regression analyses. Because inflated type I error was a poten- devoted to caregiving and perceptions of time and energy
tial issue, each final model was validated using bootstrap caregiving (see Figure 1).
290 The Family Journal: Counseling and Therapy for Couples and Families 23(3)

Table 2. Relationship, Time Spent Caregiving, Amount of Treatment, and Income and Levels of Depression/Family Discord.

Depression Domain Score Family Discord Score

n M SD p Value n M SD p Value

Parent/child relationshipa
No 57 6.36 4.23 .7205 59 5.7 3.51 .1639
Yes 103 6.57 3.12 104 6.59 4.14
20 hr Carea
Less than or equal to 20 95 5.67 3.11 .0003* 97 5.41 3.47 .0006*
More than 20 65 7.7 3.8 66 7.53 4.25
Treatment: A lota
No 61 6.74 3.77 0.4878 63 6.57 3.76 0.4324
Yes 99 6.34 3.41 100 6.08 4.05
Incomeb
25 K or less 12 7.05 5.51 .181 10 6.62 5.35 .2975
26–50 K 35 7.44 3.18 35 7.03 4.09
51–75 K 36 6.58 3.9 39 5.26 3.39
76–100 K 29 5.3 2.29 31 5.95 4.02
Over 100 K 32 6.29 3.36 31 6.83 3.48
a
Comparisons are t test. bComparisons are an ANOVA.
*Significant at p ¼ .05.

between the MIDS and the difficult behaviors domain. Table

1 shows the mean and standard deviations of the individual
frequency and intensity scores and Table 3 summarizes the
individual domain scores and the two composite scores as
well as provides broad comparisons between individuals with
high and low scores on the ECI. Note that the money
shortages domain seems to have the highest mean frequency
and intensity levels but also the most variation, implying that
money issues were not uniform across all respondents. By
contrast, the highest mean domain scores came from the irrit-
ability domain while some of the lowest variation levels were
in the money shortages domain. Taken together this suggests that
money shortage frequency and intensity seemed to follow one
another more closely because a divergence in frequency and
intensity would have corresponded to higher standard deviation
Figure 1. Perceived time and energy separated by amount of scores (more consistent with the rest of Table 3). Testing was not
caregiving. done at this point since the ECI runs on a continuum from 1 to 32
and will be studied later as a linear predictor in regression
analyses.
Experience of Caregiving Results Given broad potential associations between level of care,
In addition to measuring factors related to distress in care- caregiving relationship, amount of treatment, household
givers, two specific domains from the ECI were used in order income, and degree of mental illness, multiple regression mod-
to measure how the family member with SMI was exhibiting els were employed to consider the multivariate effect of many
symptoms and behaviors related to the SMI. The two of these patient and household level factors on caregiver
domains, negative symptoms domain and difficult behaviors depression and family distress. Several different models were
domain, asked participants exactly what types of symptoms explored and the results presented here were decided on in
of mental illness the family member with SMI was showing order to strike the best balance of significant predictors and
and the particular behaviors the family member exhibited. model parsimony.
There were significant differences in the difficult behaviors The results for regression models on both outcome vari-
domain and negative symptoms domain between caregivers ables are listed in Table 4. Only one of the ECI behavior
who gave less than or equal to 20 hr of caregiving (t test p val- scales showed a significant effect on outcomes, with the
ues .0026 and <.0001, respectively). Significant correlations models suggesting that higher numbers of difficult beha-
were found between all of the domains of the MIDS scores viors associate with both higher family discord scores and
and the two ECI domains. Strong associations were found depression scores. Likewise, in both models the level of
Crowe and Brinkley 291

Table 3. MIDS Descriptives and Comparisons Across High- and Low-Frequency ECI Difficult Behaviors Domain.

ECI Difficult Behaviors Domain ECI Difficult Behaviors

Score Less Than Domain Score

n M SD Median M SD M SD

Irrit domain 169 9.08 5.16 9 7 4.6 10.6 5

Lone domain 169 6.86 5.03 6 4.8 4.1 8.3 5.1
Anger domain 168 7.51 5.83 6 5.6 5.6 8.9 5.7
Insecurity domain 167 6.18 5.13 6 4.5 4.2 7.5 5.6
Neglect domain 165 5.25 4.65 4 3.4 3.7 6.5 5
Worry domain 168 7.75 6.8 6 4.6 5.4 10 6.8
Phys Ab domain 167 3.23 4.87 1 2 3.3 4.2 5.8
Embarr domain 166 5.3 5.02 4 3 3 7 5.6
Money short domain 168 8.37 3.3 8 7.5 3 8.8 3.2
Lack of Stim Ad domain 168 6.07 5.63 4 3.9 5 7.7 5.6
Verb abuse domain 169 6.01 6.7 2 3.4 5.2 7.8 7
UN Sat Rom domain 167 5.74 6.07 3 3.7 4.7 7.3 6.6
Depression domain score 160 6.49 3.54 6.14 4.4 2.7 8 3.4
Family discord score 163 6.27 3.93 5.8 4.3 3 7.7 4
Note: MIDS ¼ Mental Illness Distress Scale; ECI ¼ Experience of Caring Inventory; irri ¼ irritability; Phys Ab ¼ physical abuse; Embarr ¼ embarrassment; Money
short ¼ money shortages; Stim Ad ¼ stimulating adult; UN Sat Rom ¼ unsatisfactory romantic life.

Table 4. Linear Regression Models Predictors of MIDS and Family Discord Scores.

Outcome: Depression Domain Scorea Outcome: Family Discord Domain Scoreb

Standard p Effect Bootstrap Standard p Effect Bootstrap

Term Estimate Error Value Sizec Validation Rate Estimate Error Value Sizec Validation Rate

Intercept 3.43 0.74 <.0001 N/A 98% 3.01 0.88 .0008 N/A 73%
Provide less than 20 hr of care 1.25 0.49 .0123 0.0416 55% 1.29 0.58 .0266 0.0317 37%
ECI—difficult behaviors domain 0.29 0.04 <.0001 0.3111 100% 0.27 0.04 <.0001 0.2057 73%
Parent/child relationship 1.21 0.51 .0192 0.0365 65% 0.45 0.61 .45 0.0037 21%
Note. MIDS ¼ Mental Illness Distress Scale; ECI ¼ Experience of Caring Inventory.
a 2
R ¼ .37.
b 2
R ¼ .26.
c
Effect size estimates from partial eta-square of each variable when added to the model last.

care makes a significant effect on both metrics. After adjust-

ing for the amount of time spent and the number of difficult Discussion
behaviors caregiver and family member with SMI, relation- Almost all participants indicated that their relative with SMI
ship does have a significant effect on caregiver depression. was receiving some sort of mental health services. These
Unlike other analyses, however, the relationship worked in results are encouraging, since authors had indicated that care-
reverse, the model indicates that after adjusting for time givers are doing more for the care of their relative with SMI
spent caregiving, family caregivers had significantly lower than before including direct care (Abraham & Stein, 2013;
depression scores when caring for their sons or daughters. Crowe & Lyness, 2014). Additionally, caregivers in our sam-
Bootstrapping the data for validation purposes showed that ple were generally satisfied with the mental health treatment
the significant predictors in the depression model were also their family member was receiving. Actual time spent care-
significant in many of the resamplings, only the ECI diffi- giving in hours (less than or equal to 20 hr vs. more than 20
cult behaviors domain validated at over 50% of resamplings hr/week) as well as perceived time and energy that the care-
in the family discord model. When combined with the effect giver felt that he or she dedicated to the family member was
test output, it may seem that the true impact of time spent explored. Perceived time was measured with a Likert-type
and caregiver relationship may have only weak influences scale (no time and lots of time) and energy was measured
on family discord scores. While these regression models on a similar scale (no energy and lots of energy). Largely,
offer new insights, it should be noted that the R-square val- caregivers’ perceived time and energy matched the actual
ues indicate that there is still a high level of unexplained hours spent caregiving—most caregivers who spent more than
variation in the data, which may be the result of unmeasured 20 hours each week perceived that they spent at least some
confounding or simply natural variation in the data. time (and in most cases a lot of time) and energy caregiving.
292 The Family Journal: Counseling and Therapy for Couples and Families 23(3)

More than one in five of caregivers, however, who devoted Perhaps caring for a son or daughter somehow decreases the
less than 20 hr/week endorsed that they gave ‘‘lots of time’’ feelings of depression that might be associated with caregiving.
and almost half perceived that they gave ‘‘lots of energy.’’ Caregiving as a parent might be less distressing, or demanding,
This finding suggests that there was a difference between when compared to caregiving for other family members. While
quantified time in hours and perceived time and energy of this was the case with depression, familial relationship and
caregivers. Thus, caregivers, even those who are spending hours spent caregiving did not result in a significant difference
less than 20 hr/week perceive that they are spending a lot of related to family discord. Previous literature had suggested that
time and energy. This finding perhaps may be indicative of caregiving is a distressing role (Ohaeri, 2003; Struening et al.,
the role of caregiver and how it is one that demands a lot of 2001). These findings suggest that perhaps it is less distressing
the family member. Regardless of actual time spent, caregiv- for caregivers who are parents when compared to other familial
ing feels demanding and takes considerable energy for care- relationships (related to depression).
givers. Researchers (Abraham & Stein, 2013) had posited Also in question was whether or not income level of the
that the role of the family is important in the care of people family might influence caregiver distress. Since previous stud-
with SMI, despite the developments in mental health treat- ies (Mackay & Pakenham, 2012; Mitsonis et al., 2012) had
ment. This finding suggests that caregivers might be feeling included sociodemographic factors as they applied to distress
overwhelmed with the time and energy they are devoting to in family caregivers, we wanted to include this in the current
their family member, suggesting the need for support for care- study as well. Income did not appear to affect depression or
givers of someone with SMI. For family counselors and other family discord in caregivers for a family member with SMI.
mental health professionals working with caregivers, or fam- It is noteworthy, however, that money shortages seemed to
ilies with mental illness, it seems as though it will be impor- be an important variable across the MIDS scores. While the
tant to discuss time and energy spent caregiving, in order to overall income might not impact depression or discord,
assist with coping with this substantial role. researchers and clinicians alike might want to explore the
Caregivers who provided more than or equal to 20 hr of care impact of money shortages on family caregivers for SMI.
each week had significantly more depression and family dis- Two domains from the ECI were used in order to explore par-
cord than those who indicated that they provided less than ticular behaviors of the family member with SMI to see how
20 hr of care per week. This suggests that those family mem- these might relate to family distress. Difficult behaviors and neg-
bers who are caring for their relative with SMI for 20 hr or ative symptoms were looked at together in order to determine
more a week are experiencing more of the negative symptoms whether behaviors and symptoms of the person with mental ill-
such as depression or discord than those who are providing less ness effected the caregiver’s level of distress. Significant differ-
time caregiving. Perhaps those who are caregiving at this ences were found for both negative symptoms and difficult
capacity are experiencing both of these outcomes due to the time behaviors domains for caregivers who provided more than or
and energy that caregiving demands, since researchers had dis- equal to 20 hr of caregiving. Significant correlations were also
cussed that caregiving is a demanding or distressing role (Struen- found across all domains of the MIDS and the two ECI domains,
ing et al., 2001). These results appear to support this notion. indicating that there is a relationship between the distress expe-
Family discord and depression were in fact present for caregivers, rienced by the caregiver and the behaviors and symptoms that
especially those who spent considerable time in this role. Family the family member is experiencing. Interestingly, family rela-
clinicians might remember that caregiving could be associated tionship and hours of caregiving did not impact family discord
with depression and discord in family caregivers, and assess for in the same way that it did with depression.
this when working with families. Exploring ways to lessen the
time caregiving, perhaps by looking for outside supports for both
the family member and the caregiver might prove beneficial.
Implications for Family Practice
Our study was similar to recent studies (Mackay & Pakenham, Clinical implications of the study findings are of particular
2012; Mitsonis et al., 2012) outside the United States and importance. Family counselors might consider that family
examined the type of familial relationship and distress. Among caregivers, despite actual time caregiving, are going to need
family members who cared for their children, 49% of such indi- considerable support themselves, since the study results are
viduals devoted more than 20 hr of care per week. By contrast, consistent with other research that supports that caregiving
among individuals who were caregiving for someone other than is a difficult and distressing task. Parents who are caregiving
a son or daughter, only 28% of such individuals devoted more for a son or daughter might have less depression than other
than 20 hr of care per week. Thus, parent caregivers seem to be types of family members who are caregiving, thus counselors
spending more time caregiving than caregivers who are not caring might consider this when seeing family members. When
for a son or daughter. While the increased time does have an meeting with the family, counselors might also pay close
increasing depression effect, the results indicate that caring for attention to what behaviors the person with SMI is exhibiting.
a son or daughter is associated with lower depression. The point Since difficult behaviors impacted the level of family discord,
estimates in Table 4 show that the effect of more time per week counselors might ask family members about what type of
is about the same in magnitude as whether you are the parent or behaviors the person with SMI is exhibiting at home, since
not. They almost appear to cancel each other out. these may effect discord in the family. Decreasing difficult
Crowe and Brinkley 293

behaviors might be a goal of treatment so that family discord health treatment, and future researchers might wish to investi-
can decrease. Screening for depression is another consider- gate those who have not participated in mental health treatment
ation for counselors working with families, since many hours in order to see if these families are different in any way.
caregiving, relationships (other than a son or daughter), and A lack of thorough psychometric analysis may seem to some
difficult behaviors were related to increased depression in the as a major limitation of the study. Indeed, it is unclear as to
caregivers we sampled. whether this instrument applied to this population has provided
This study used the MIDS to measure distress in caregivers the most reliable and consistent set of scores possible. Further
both in psychological terms (depression) and within the family work needs to be done to examine the full set of psychometric
context (family discord). Until now, there has not been a mea- properties of the MIDS instrument. The focus here was on clin-
sure that explores the level of distress (in a family context) that ical utility and not instrumentation, and the results in Table 4
caregivers experience even though there is evidence (Struening illustrate that the proposed set of scores do provide some clin-
et al., 2001) that they are in need of support. Although other mea- ical merit. For the purposes of this article, the authors chose to
sures had looked at the distress related to caregiving for a family rely on both the psychometric foundations of the original
member with mental illness, distress was operationalized as only DPDS and the preliminary properties found in this as well as
psychological distress rather distress as it might manifest in a previous (Crowe & Lyness, 2014) research with the MIDS.
family context. For this reason, we chose to use the MIDS, an Although significant relationships were found in this study, it
instrument that measured distress both internally and externally. is worth noting that there is still much unexplained variation in
In this sample, the vast majority of scores on the MIDS were the data. Also, further study is needed to formally examine the
skewed toward the low end of possible scores—none of the psychometric properties of the MIDS. Demonstrating that the
domains of the MIDS were extreme, so we can assume that this MIDS has the capacity to elucidate differences between groups
sample demonstrated either low frequency and/or low intensity. that are well-defined clinically is a not the same as suggesting
Indeed, many of the means and medians were at or around 6 that the MIDS is a valid and reliable instrument for this popula-
(maximum score was 17). The skewed nature of the results in tion. Additionally, due to the small sample size, this study could
this study suggests the utility of the scale for researchers inter- not explore differences in subgroups (e.g., gender, familial rela-
ested in this topic; however, for researchers interested in study- tionships) so future research with larger sample sizes will offer
ing groups who are not experiencing distress, a metric other than this to researchers. A limitation of the MIDS related to the scor-
the MIDS might be more suitable. Overall, the MIDS appears to ing is also worth noting. On each domain, total scores are calcu-
have demonstrated some capacity for comparing groups. lated by multiplying a frequency score by an intensity score (for
full instructions, see Crisp & Barber, 1995; Crowe & Lyness,
2014), so researchers interested in looking at the two dimen-
Limitations and Future Directions sions separately cannot do so with the MIDS.
As with any research, this study is not without limitations. The
sample was predominantly homogeneous, and future studies
would benefit from more diverse samples for generalizability.
Conclusion
The research was cross sectional in design and self-reported This study explored distress in caregivers of a family member
by a convenience sample. Scholars interested in families and with SMI. In addition to the overall distress related to caregiv-
SMI might longitudinally examine families over time to better ing, particular symptoms and behaviors of the person with SMI
understand the developmental nature of SMI and the family. were explored as they related to caregiver distress. Findings
Data were obtained from a single individual in the family, and suggested that caregivers who provided a lot of care did in fact
future research might focus on gathering data from multiple fam- feel depressed and experience overall family discord. Many
ily members, possibly comparing results of each member. Gath- family members who were caring for their son or daughter
ering data from the family member with the mental illness and devoted more than 20 hr/week, while a lesser percentage of
comparing this data to the other family members could yield individuals caring for those other than a son or daughter devoted
interesting findings. Authors interested in doing this could more than 20 hr/week. Regression models revealed that higher
consult Guttman and Laporte (2012) who investigated various numbers of difficult behaviors were associated with higher fam-
family members’ perceptions on intrafamilial relationships. ily discord scores. Level of care influenced both depression and
The families in this study were members of a southeastern discord, and after adjusting for the amount of time spent and the
state branch of NAMI and were responsive to a listserv request number of difficult behaviors, the type of familial relationship
from the researcher. This raises the question of how families had a significant effect on caregiver depression. Caregivers had
who are not members of NAMI might respond. Future research significantly lower depression scores when caring for their sons
might include sampling designs that capture experiences from or daughters, thus suggesting that the type of familial relation-
families who are not involved in the organization and who may ship influences depression of caregivers.
not have access to the Internet. NAMI members are typically
active in advocacy efforts for mental illness, so those in the Declaration of Conflicting Interests
general population might differ. These family members also The author(s) declared no potential conflicts of interest with respect to
indicated that their relative with SMI had already had mental the research, authorship, and/or publication of this article.
294 The Family Journal: Counseling and Therapy for Couples and Families 23(3)

Funding Mental Health Journal, 48, 450–462. doi:10.1007/s10597-011-

The author(s) received no financial support for the research, author- 9435-4
ship, and/or publication of this article. Marsh, D. T. (2001). A family-focused approach to serious mental
illness: Empirically supported interventions. Sarasota, FL:
Professional Resource Press.
References Mitsonis, C., Voussoura, E., Dimopoulos, N., Psarra, V., Kararizou, E.,
Abraham, K., & Stein, C. H. (2013). Emerging adults’ perspectives on Latzouraki, E., . . . Katsanou, M. N. (2012). Factors associated with
their relationships with mothers with mental illness: Implications caregiver psychological distress in chronic schizophrenia. Social
for caregiving. Journal of Orthopsychiatry, 82, 542–549. doi:10. Psychiatry and Psychiatric Epidemiology, 47, 331–337. doi:10.
1111/j.1939-0025.2012.01175 1007/s00127-010-0325-9
Crisp, B., & Barber, J. (1995). The drinker’s partner distress scale: An National Alliance on Mental Illness. (2012). Mental illnesses.
instrument for measuring the distress caused by drinkers to their Retrieved September 5, 2013, from http://www.nami.org/Tem-
partners. International Journal of Addictions, 30, 1009–1017. plate.cfm?Section¼By_Illness
Crowe, A., & Lyness, K. P. (2014). Family functioning, coping, and Ohaeri, J. U. (2003). The burden of caregiving in families with a men-
distress in families with serious mental illness. The Family Jour- tal illness: A review of 2002. Current Opinions in Psychiatry, 16,
nal, 22, 186–197. doi:10.1177/1066480713513552 457–465. doi: 10.1097/01.yco.0000079212.36371.c0
de Araújo Jorge, R., & Chaves, A. (2012). The experience of caregiv- Pakenham, K. I., Dadds, R. M., & Terry, D. J. (1994). The relation-
ing inventory for first-episode psychosis caregivers: Validation of ships between adjustment to HIV and both social support and
the Brazilian version. Schizophrenia Research, 138, 274–279. doi: coping strategies. Journal of Consulting and Clinical Psychology,
10.1016/j.schres.2012.03.014 62, 1203–1994.
Grover, S., Chakrabarti, S., Aggarwal, M., Avasthi, A., Kulhara, P., Pickett-Schenk, S. A., Bennett, C., Cook, J. A., Steigman, P., Lippin-
Sharma, S., & Khehra, N. (2012). Comparative study of the cott, R., Villagracia, I., & Grey, D. (2006). Changes in caregiving
experience of caregiving in bipolar affective disorder and schi- satisfaction and information needs among relatives of adults with
zophrenia. International Journal of Social Psychiatry, 58, mental illness: Results of a randomized evaluation of a family-
614–622. doi:10.1177/0020764011419054 led education intervention. American Journal of Orthopsychiatry,
Guttman, H. A., & Laporte, L. (2012). Family members’ retrospective 76, 545–553
perceptions of intrafamilial relationships. Contemporary Family Reinhard, S. C., Gubman, G. D., Horwitz, A. V., & Minsky, S. (1994).
Therapy, 24, 505–521. doi:10.1023/A:1019871417951 Burden assessment scale for families of the seriously mentally ill.
Harrell, F. E. (2001). Regression modeling strategies with applica- Evaluation and Program Planning, 17, 261–269.
tions to linear models, logistic regression and survival analysis. Szmukler, G. I., Burgess, P., Herrman, H., Benson, A., Colusa, S., &
New York, NY: Springer. Bloch, S. (1996). Caring for relatives with serious mental illness:
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal and coping. The development of the Experience of Caregiving Inventory.
New York, NY: Springer. Social Psychiatry and Psychiatric Epidemiology, 31, 137–148.
Lovibond, S. H., & Lovibond, P. F. (1995). Manual for the depression Struening, E. L., Perlick, D. A., Link, B. G., Hellman, F., Herman, D.,
anxiety stress scale. Sydney, Australia: The Psychological Founda- & Sirey, J. (2001). Stigma as a barrier to recovery: The extent to
tion of Australia. which caregivers believe most people devalue consumers and their
Mackay, C., & Pakenham, K. I. (2012). A stress and coping model of families. Psychiatric Services, 52, 1633–1638. doi:10.1176/appi.
adjustment to caring for an adult with mental illness. Community ps.52.12.1633