The Gerontologist © The Author 2009. Published by Oxford University Press on behalf of The Gerontological Society of America.

Vol. 49, No. 6, 736–745 All rights reserved. For permissions, please e-mail: [email protected].
doi:10.1093/geront/gnp052 Advance Access Publication on June 21, 2009

Development and Psychometric Evaluation of

the Reasons for Living—Older Adults Scale: A
Suicide Risk Assessment Inventory

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

Barry A. Edelstein, PhD,1,2 Marnin J. Heisel, PhD,3,4 Deborah R. McKee, PhD,2
Ronald R. Martin, PhD,2 Lesley P. Koven, PhD,2 Paul R. Duberstein, PhD,4
and Peter C. Britton, PhD4,5
Purpose: The purposes of these studies were to with respect to lifetime history of suicidal behav-
develop and initially evaluate the psychometric prop- ior. Implications: These findings provide pre-
erties of the Reasons for Living Scale—Older Adult liminary support for the validity and reliability of the
version (RFL-OA), an older adults version of a mea- RFL-OA. The findings also support the potential value
sure designed to assess reasons for living among of attending to reasons for living during clinical treat-
individuals at risk for suicide. Design and Meth- ment with depressed older adults and others at risk
ods: Two studies are reported. Study 1 involved for suicide.
instrument development with 106 community-dwelling Key Words: Suicide, Reasons for living, Suicide risk,
older adults, and initial psychometric evaluation with Resilience
a second sample of 119 community-dwelling older
adults. Study 2 evaluated the psychometric proper-
ties of the RFL-OA in a clinical sample. One hundred Older adults, 65 years of age and older, are at
eighty-one mental health patients 50 years or older greater risk for suicide than any other age group in
completed the RFL-OA and measures of depression, the United States, with White men, aged 85 years and
suicide ideation at the current time and at the worst older, having the highest rate of suicide (National
point in one’s life, and current mental status and Center for Injury Prevention and Control [NCIPC],
physical functioning. Results: Strong psychomet- 2007). Older adults less frequently engage in self-
ric properties were demonstrated for the RFL-OA, harm behavior than do younger individuals but are
with high internal consistency (Cronbach’s alpha co- considerably more likely to die as a result of self-
efficient). Convergent validity was evidenced by neg- harm (Draper, 1996). Older adults account for 20%
ative associations among RFL-OA scores and of deaths by suicide but represent only 13% of the
measures of depression and suicide ideation. RFL-OA U.S. population (NCIPC).
scores predicted current and worst-episode suicide Although these suicide rates are astonishingly
ideation above and beyond current depression. Dis- high, little research has addressed suicidal ideation
criminant validity was evidenced with measures of and behavior among older adults (Pearson &
current mental status and physical functioning. Brown, 2000). Equally astonishing is the fact that
Criterion-related validity was also demonstrated 70% of older adults who died by suicide had seen
their primary care provider within 30 days of their
1 deaths (Conwell, Olsen, Caine, & Falnnery, 1991;
Address correspondence to Barry A. Edelstein, PhD, Department of
Psychology, West Virginia University, Morgantown, WV 26506-6040. Diekstra & van Egmond, 1989; Luoma, Martin, &
E-mail: [email protected] Pearson, 2002). These data suggest that many indi-
2
Department of Psychology, West Virginia University, Morgantown,
West Virginia. viduals who are at risk for suicide are, in principle,
3
Departments of Psychiatry and of Epidemiology & Biostatistics, identifiable, and their suicides are potentially
The University of Western Ontario, London, Ontario, Canada.
4
Department of Psychiatry, University of Rochester School of Medicine preventable. Although some predictors of older
and Dentistry, Rochester, New York.
5
Center of Excellence, Canandaigua VA Medical Center, Canandaigua,
adult suicide are known (e.g., psychiatric illness,
New York. physical illness, functional impairment; Heisel &

736 The Gerontologist

Duberstein, 2005), the lack of research focus on and older adults (Edelstein, McKee, & Martin,
older adults to date has hindered the advancement 2000) and found age-related differences in reasons
of our knowledge regarding the assessment of sui- for living for participants ranging in age from 19
cide risk and prevention of suicide in this population. to 88 years. Miller, Segal, and Coolidge (2001)
Age-related changes in the phenomenology and compared older and younger adults’ reasons for
presentation of mental disorders (e.g., Edelstein, living using the reasons for living inventory (Line-
Kalish, Drozdick, & McKee, 1999; Kogan, & han et al.) and found both overlap and differences
Edelstein, 2004; Edelstein et al., 2008) suggest the in reasons for living between these two age groups.
need for assessment instruments tailored to older These foregoing studies suggest that scales intend-
adults. Unfortunately, to date there is only one ed to measure reasons for living must be appropri-
published self-report suicide risk assessment in- ate to the age group being assessed (i.e., content

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

strument created explicitly for older adults (Heisel valid), consistent with geropsychology practice
& Flett, 2006). Much of the research on suicide guidelines (American Psychological Association,
risk among younger adults, and most of the risk 2004). Although reasons for living inventories
assessment instruments, have focused on demo- have been developed for adolescents, college stu-
graphic risk factors (e.g., marital status, age, sex), dents, young adults, and adults, no such inventory
clinical variables (e.g., depression), and behaviors had been created for older adults. The purpose of
that place individuals at risk for suicide (see the present article was to describe the development
Brown, 1999, for instrument reviews). Another and psychometric evaluation of an older adult rea-
approach to suicide risk assessment focuses on as- sons for living inventory, termed the Reasons for
sessing resiliency factors potentially preventive of Living Scale—Older Adult version (RFL-OA; Edel-
suicide risk (Heisel & Flett, 2008). One example stein et al., 2000). The first study involved the ini-
is an instrument initially developed by Linehan, tial development of the RFL-OA and was divided
Goodstein, Nielsen, and Chiles (1983) that mea- into three parts. In the first part, the items of the
sures reasons for not taking one’s life despite RFL-OA were developed. In the second part, the
suicidal thoughts or considerations. A major as- items were administered to a group of older adults
sumption of these reasons for living instruments is to examine the preliminary psychometric proper-
that suicidal individuals are lacking in adaptive ties of the instrument. Study 2 examined the psy-
beliefs present among nonsuicidal individuals chometric properties of the RFL-OA with a group
that deter suicidal behavior. The reasons for liv- of depressed older mental health patients. Specific
ing examined through these instruments can be aims included examination of the internal consis-
considered buffers or personal and environmen- tency of the RFL-OA. Construct validity was as-
tal contingencies operating against suicide. Rea- sessed by correlations among the RFL-OA and
sons for living instruments have been developed established clinical research measures of depres-
for a variety of different age groups. sion and suicide ideation (convergent validity), and
In their original research, Linehan and col- exploration of potential incremental validity of
leagues (1983) found that individuals with prior RFL-OA scores in explaining additional variability
suicidal behavior reported fewer reasons for living in suicide ideation scores above and beyond that
than individuals with no suicidal history. More- contributed by depression severity. Correlations
over, those with suicidal histories valued reasons between the RFL-OA and current mental and func-
for living to a smaller degree. That is, they rated tional status explored the measure’s discriminant
reasons for living as less important than individu- validity. Criterion-related validity was assessed by
als with no suicidal history. More recent research comparing RFL-OA scores for participants with or
(Cole, 1989; Gutierrez et al., 2002; Osman et al., without a lifetime history of suicidal behavior.
1993, 1998) has offered further support for the as-
sessment of reasons for living in diverse popula-
tions (e.g., psychiatric inpatients, college students, Study 1—Part 1: Initial Development of RFL-OA
delinquent adolescents). As one might expect, rea- Items
sons for living are different for different age groups The initial development of the RFL-OA fol-
(Koven, Edelstein, & Charlton, 2001). In a pre- lowed procedures similar to those used by Linehan
liminary study, Koven and colleagues combined and colleagues (1983). Reasons for living (not
reasons for living from scales developed for adoles- taking one’s life), and other related information,
cents (Osman et al., 1998), adults (Linehan et al.) were first obtained from community-dwelling older

Vol. 49, No. 6, 2009 737

adults. The questions most relevant to the con- Results
struction of the RFL-OA asked participants what Sixty-nine unique reasons for living were identi-
might keep them or other older adults from taking fied by the researchers in the surveys after redun-
their lives. Responses to these questions formed dancies were eliminated. The participants’ original
the basis for the second phase of the study in which wording was preserved whenever possible. When
these responses were converted into scaled items. the content of reasons was similar to the content of
an item from the original reasons for living inven-
Methods tory, the wording of the original inventory was used.
Twenty-eight of the 69 items were the same or simi-
Participants
lar to those of the original reasons for living inven-
Participants were 106 community-dwelling old- tory. In contrast to the items of the original reasons

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

er adults, 60 years of age and older, in West Vir- for living inventory, the older adult reasons included
ginia. Ages ranged from 62 to 91 years, with a substantially more reasons for living pertaining
mean of 74.3 years (SD = 6.08). The sample was to family and friends, religious beliefs, and moral
predominantly male (67.3%). All but two of the objections to suicide. This is consistent with the
participants were Caucasian. Religious affiliations findings of Miller, Segal, and Coolidge (2001).
of participants included 83.2% Protestant, 8.4%
Catholic, 0.9% Jewish, and 7.5% other religious
Study 1—Part 2: Scale Construction and Initial
faiths. Sixty-two percent of the participants were
Administration
married, 7.5% single, 4.7% divorced, and 25.2%
widowed. The second part of Study 1 involved the con-
struction and administration of the RFL-OA. Six-
ty-nine unique items comprised the RFL-OA.
Materials
These items were administered to older adults to
Materials comprised a demographic question- gather initial normative data and estimate internal
naire and a survey. The questionnaire requested consistency.
information on age, race, sex, marital status, and
religious preference. The survey informed poten- Methods
tial participants that the authors were studying
reasons that older adults want to stay alive and Participants
that this was important because older adults are at One hundred nineteen community-dwelling
the greatest risk for taking their lives. The ques- older adults, 65 years of age and older (M age =
tions posed, as per Linehan and colleagues (1983), 75.4, range = 66–90 years), were recruited through
were as follows: (1) Have you ever considered sui- a mailing to a nonoverlapping sample of 500 older
cide and, if so, what were your reasons for not tak- adults whose names were randomly drawn from a
ing your life? (2) If you were to consider taking list of all homeowners and individuals with driver
your life, what would stop you from doing so? (3) licenses in the State of West Virginia purchased
What reasons do you think other older adults have from Survey Sampling International. This repre-
for not taking their lives? sented a 24% response rate.
Forty-five participants were 65–74 years of age
Procedure (old), 33 were 75–84 years of age (old-old), and 11
were 85 years of age or older (oldest-old). Ninety-
Materials were mailed, along with instructions nine percent of the participants were Caucasian
and stamped return envelopes, to 500 older adults and 53% were men.
whose names were purchased from Survey Sam-
pling International. The names were randomly
drawn from a list of all homeowners and individu- Materials
als with driver licenses in the State of West Vir- The materials comprised a demographic ques-
ginia. The 109 returned materials represented a tionnaire and the RFL-OA, which was constructed
21% return rate, which is relatively common for from the reasons for living produced in Study 1.
mail surveys (Dillman, 2000). Respondents and The questionnaire requested information on age,
nonrespondents could not be compared, as the race, occupation, religious preference, and number
mailing was anonymous. of years since retirement. The RFL-OA included

738 The Gerontologist

69 items and employed a 6-point Likert-type scale Methods
assessing the importance of each item in deterring
Participants
suicidal behavior (1 = quite unimportant, 2 = unim-
portant, 3 = somewhat unimportant, 4 = somewhat Participants included depressed psychiatric
important, 5 = quite important, 6 = extremely patients 50 years of age and older recruited from
important). Sample items are listed in the Appen- inpatient and outpatient psychiatric services asso-
dix. Each item is a potential reason for living. ciated with three teaching hospitals in Rochester,
NY, including a community hospital, a tertiary care
Procedure facility, and an academic medical center. Research
coordinators approached 633 psychiatric inpa-
Demographic questionnaires and the RFL-OA tients older than 50 years admitted to the hospital
were mailed to 500 individuals. Participants were

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

with apparent symptoms of depression. Thirty-nine
asked, “rate how important each reason would be patients were also recruited from an older adult
for you, if you were ever to consider taking your outpatient clinic. Two hundred fifty patients con-
life, no matter how unlikely that might be.” This sented to participate in this study. For the present
instruction is similar to that used by Linehan and analyses, we excluded patients who did not com-
colleagues (1983) in the original reasons for living plete the RFL-OA (n = 46), who did not complete
inventory with two exceptions. In the present study (n = 4) or scored below 20 on the Mini-Mental
“taking your own life” was used rather than “kill State Examination (MMSE; n = 2; Folstein, Folstein,
yourself,” which was used in the original. Second, & McHugh, 1975), who did not complete the
the statement, “no matter how unlikely that might Structured Clinical Interview for the Diagnostic
be,” used in the present study, was not used in the and Statistical Manual of Mental Disorders-IV
original. (SCID; n = 5; First, Spitzer, Gibbon, & Williams,
1997) or the Scale for Suicide Ideation for current
Results episode (SSI-C; n = 3; e.g., Beck, Brown, Steer,
The following mean importance ratings were Dahlsgaard, & Grisham, 1999; Beck, Kovacs, &
calculated for each of the measure’s five-item cat- Weissman, 1979; Beck, Steer, & Brown, 1996;
egories, as aforementioned: survival = 4.57 (quite Beck, Weissman, Lester & Trexler, 1974), who
important), moral/religious objections = 4.56 were missing total scores on the revised Beck
(quite important), family/others = 4.23 (somewhat Depression Inventory-II (BDI-II; n = 6; Beck et al.,
important), fear of social disapproval = 3.26 1996), or who were missing scores on the instru-
(somewhat unimportant), and fear of suicide = mental activities of daily living (IADL) scale and/or
3.18 (somewhat unimportant). Twelve items re- the physical self-maintenance scale (PSMS; n = 3;
ceived mean importance ratings below 4.0 (some- Lawton & Brody, 1969, 1988a, 1988b). Partici-
what important), although the standard deviations pants included 181 patients (75 men and 106
for these items ranged from 1.6 to 2.0. Thus, even women) who were currently in treatment for de-
the items receiving the lowest mean ratings re- pression (154 inpatients, 4 partial hospitalization
ceived the majority of ratings at 4.6 or above. In- patients, and 23 outpatients) ranging in age from
ternal consistency was examined through 50 to 88 years (M = 60.1, SD = 10.0). The majority
calculation of Cronbach’s coefficient alpha, which of participants were White (n = 160), 10 African
was .96 for the total scale. American, 2 American Indian or Alaskan Native,
and 8 of other racial background. Nine participants
were Hispanic or Latino. One hundred forty-one
Study 2: Initial Psychometric Examination of the participants had a major depressive disorder, 23 bi-
RFL-OA in a Clinical Sample polar I disorder, 4 bipolar II disorder, 1 dysthymic
In the present study, we examined the psycho- disorder, 6 depressive disorder not otherwise speci-
metric properties of the RFL-OA in a sample of fied, 2 schizophrenia, and 4 substance-induced
mental health patients 50 years of age and older. We mood disorder. Fifty-four percent were divorced
specifically investigated the internal consistency, and (n = 63), separated (n = 19), or widowed (n = 15).
convergent and discriminant validity of the RFL- Forty percent lived alone (n = 72). Fifty-three per-
OA, and the criterion-related validity of the mea- cent were unemployed (n = 35) or receiving disabil-
sure in differentiating patients with versus without a ity benefits (n = 60). Participants completed an
lifetime history of self-injurious behavior. average of 13.3 years of education (SD = 2.6).

Vol. 49, No. 6, 2009 739

Measures patients with a possible mood disorder. Following
Reasons for living were assessed using the approval from an attending physician or primary
RFL-OA. clinician, a member of the research team ap-
Suicide ideation was assessed with the SSI (Beck proached patients seeking their written informed
et al., 1979) for respondents’ current state (SSI-C) consent to participate in an interview and complete
and for the worst point in their lives (SSI-W; e.g., measures. Trained interviewers then administered
Beck et al., 1999). The SSI is a 19-item clinician- the SCID Axis I Disorders (First et al., 1997) and
administered scale designed to assess the presence the SSI (Beck et al., 1979). Participants also com-
and severity of considerations and plans for sui- pleted the RFL-OA and measures of depression
cide. The SSI has strong reliability with older and of cognitive and physical functioning. Follow-
adults (Heisel, Flett, & Besser, 2002) and the ing the acquisition of data and reviews of medical

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

SSI-C has a reported interrater reliability coeffi- records, consensus diagnostic conferences were
cient of .87 in a clinical sample (Beck et al., 1979). held attended by at least one psychiatrist, one psy-
Scores on this measure potentially range from 0 to chologist, study investigators, and members of our
38. SSI scores were adjusted for missing data for research laboratory. The research coordinator who
up to four missing items (e.g., Beck et al., 1974, had interviewed the patient delivered a case pre-
1979, 1996, 1999). sentation incorporating information from the re-
Depression was measured with the BDI-II (Beck cord review and diagnostic interview, and the
et al., 1996), a 21-item self-report questionnaire research team reached diagnostic consensus.
assessing depressive symptomatology. The BDI-II
has demonstrated strong internal consistency, test– Statistical Analysis
retest reliability, and convergent validity, signifi-
cantly predicting hopelessness and suicide ideation Descriptive statistics are presented in Table 1,
(Beck et al.), and research supports its use with de- consisting of means, standard deviations, mea-
pressed older inpatients (Steer, Rissmiller, & Beck, sures of normality and internal consistency, and
2000). Scores on this measure potentially range intercorrelations among study measures. A pair of
from 0 to 63. BDI-II scores were prorated for up to hierarchical multiple regression analyses are pre-
for missing items. sented next, predicting current (SSI-C; Table 2)
Cognitive functioning was assessed with the and worst-episode suicide ideation (SSI-W; Table 3)
MMSE (Folstein et al., 1975), a brief global mea- with RFL-OA scores, controlling for age, sex, and
sure of an individual’s cognitive state. The MMSE depression. A logistic regression analysis is pre-
has demonstrated strong psychometric properties, sented in Table 4, predicting history of suicidal be-
with a reported test–retest reliability coefficient of havior (ever vs. never) with RFL-OA scores,
.89 over a 24-hr period, and acceptable construct controlling for age and sex. The logistic regression
and concurrent validity (Folstein et al.). Scoring on was tested using Wald’s chi-square statistic. All re-
this measure ranges from 0 to 30; this data set was ported p values are two tailed, with a set at .05.
restricted to individuals scoring 20 and above. Unadjusted between-group differences are report-
Physical functioning was assessed with the IADL ed in text comparing RFL-OA scores by sex (male
scale (IADL) and the PSMS (Lawton & Brody, 1969, vs. female), age (50–64 years vs. >65 years), educa-
1988a, 1988b), brief interviewer-rated measures of tion (<13 years vs. >13 years), and history of sui-
competence in basic (e.g., toileting, feeding, dress- cidal behavior (ever vs. never). Independent
ing, and bathing; PSMS) and instrumental daily ac- samples t tests were employed assuming equal vari-
tivities (e.g., using the telephone, shopping, preparing ances, unless Levene’s test for equality of variances
food, and housekeeping; IADL). Higher scores rep- was significant, and then, t tests were employed
resent greater functional impairment. that do not assume equal sample variances.

Procedures Results
Research coordinators screened the records of Descriptive statistics for study measures are pre-
all patients 50 years of age and older admitted to sented in Table 1. Scores on measures of depres-
one of three hospital’s inpatient units or seen for sion (BDI-II: M = 25.2, SD = 14.2, range = 0–59)
an intake session in one hospital’s ambulatory and current (SSI-C: M = 6.1, SD = 9.4, range = 0–
mental health clinic for older adults, to identify 32) and worst-episode suicide ideation (SSI-W:

740 The Gerontologist

 Table 1. Correlational Matrix for the Measures Examined in Study 2

1 2 3 4 5
1. RFL-OA 1.00 −.40*** −.42*** −.43*** −.04
2. SSI-C 1.00 .24** .49*** −.06
3. SSI-W 1.00 .29*** .05
4. BDI-II 1.00 .04
5. MMSE 1.00
M 265.3 6.1 16.4 25.2 27.5
SD 74.5 9.4 12.2 14.2 2.4
Skewness −0.45 1.43 −0.08 0.24 1.11
Kurtosis −0.55 0.62 −1.50 −0.81 0.61
a .98 .83 .87 .94

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

Notes: Correlations were computed using pairwise deletion. RFL-OA = Reasons for Living Scale—Older Adult version;
SSI-C = Scale for Suicide Ideation for current episode; SSI-W = Scale for Suicide Ideation-worst; BDI-II = Beck Depression Inven-
tory-II; MMSE = Mini-Mental State Examination; a = Cronbach’s coefficient alpha.
†p < .01. *p < .001.

M = 16.4, SD = 12.2, range = 0–36) were some- Potential associations among RFL-OA scores
what higher than published norms for clinical and demographic variables (sex, age, and educa-
samples (e.g., Beck et al., 1974, 1979, 1996, 1999). tion) were examined next. RFL-OA scores were
MMSE scores (M = 27.5, SD = 2.4, range = 20–30) not significantly different for men, M = 257.8,
were consistent with published norms (Crum, SD = 77.7, and women, M = 270.6, SD = 72.1,
Anthony, Bassett, & Folstein, 1993). All measures F(1,179) = 1.31, p = .25, h2 = .01. Participants
evidenced acceptable internal consistency. Nor- were divided into younger (50–64 years: n = 136)
mality of the distributions of scores was generally and older cohorts (65 years and older: n = 45), and
supported by skewness and kurtosis statistics. compared with respect to RFL-OA total scores.

Table 3. Summary of a Hierarchical Multiple Regression

Table 2. Summary of a Hierarchical Multiple Regression Analysis Predicting Worst-Episode Suicide Ideation
Analysis Predicting Current Suicide Ideation (SSI-C) With (SSI-W) With Reasons for Living (RFL-OA), Controlling
Reasons for Living (RFL-OA), Controlling for Demographics for Demographics (Age, Sex) and Depression (BDI-II) for
(Age, Sex) and Depression (BDI-II) in Study 2 Participants Study 2 Participants

Variable B b t Variable B b t
Step 1 Step 1
Intercept 21.836 5.15*** Intercept 34.611 6.18***
Age −0.252 −.27 −3.71*** Age −.311 −.39 −3.89***
Sex −1.511 −.08 −1.10 Sex 1.012 −.02 −.22
Step 2 Step 2
Intercept 6.112 1.36 Intercept 23.803 3.41***
Age −0.117 −.12 −1.81† Age −0.216 −.18 −2.16*
Sex −1.294 −.07 −1.05 Sex 1.376 .06 .72
BDI-II 0.297 .45 6.61*** BDI-II 0.202 .24 2.89**
Step 3 Step 3
Intercept 15.112 2.94** Intercept 42.936 5.42***
Age −0.109 −.12 −1.73† Age −0.217 −.18 −2.29*
Sex −1.638 −.09 −1.36 Sex 0.862 .04 .47
BDI-II 0.234 .36 4.89*** BDI-II 0.059 .07 .80
RFL-OA −0.029 −.23 −3.31*** RFL-OA −0.059 −.36 −4.35***
Notes: R2 = .07 (adjusted R2 = .06), F(2, 178) = 7.14, p < Notes: R2 = .07, adjusted R2 = .06, F(2, 150) = 5.54, p <
.001 for Step 1; R2 = .26 (adjusted R2 = .25, DR2 = .18), DF(1, .01 for Step 1; R2 = .12 (adjusted R2 = .10, DR2 = .05), DF(1,
177) = 43.74, p < .001 for Step 2; R2 = .30 (adjusted R2 = .29, 149) = 8.37, p < .05 for Step 2; R2 = .22 (adjusted R2 = .20,
DR2 = .04), DF(1, 176) = 10.96, p < .001 for Step 3. SSI-C = DR2 = .10), DF(1, 148) = 18.91, p < .001 for Step 3. SSI-W =
Scale for Suicide Ideation for current episode; RFL-OA = Rea- Scale for Suicide Ideation for worst episode; RFL-OA = Rea-
sons for Living Scale—Older Adult version; BDI-II = Beck De- sons for Living Scale—Older Adult version; BDI-II = Beck De-
pression Inventory-II. pression Inventory-II.
† †
p < .10. *p < .01. **p < .001. p < .10. *p < .05. **p < .01. ***p < .001.

Vol. 49, No. 6, 2009 741

 Table 4. Logistic Regression Analysis of Suicide Attempter Status as a Function of Age, Sex, and RFL-OA Scores
for Study 2 Participants

Variable B SE Wald statistic df Exp (B) 95% CI

Age −0.07 0.02 11.69*** 1 0.94 .902–.972
Sex 0.19 0.34 0.31 1 1.21 .625–2.322
RFL-OA −0.01 0.00 7.24** 1 0.99 .990–.998
Constant 5.24 1.25 17.44*** 1 187.95
Notes: RFL-OA = Reasons for Living Scale—Older Adult version; exp (B) = estimated odds ratio; CI = confidence interval.
Nagelkerke R2 = .18.
†
p < .10. *p < .05. **p < .01. ***p < .001.

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

The older cohort, M = 296.6, SD = 68.3, scored entered as covariates on Step 1, depression scores
significantly higher on the RFL-OA than the were entered as a block on Step 2, and RFL-OA
younger cohort, M = 254.9, SD = 73.9, F(1,179) = scores were entered on Step 3. RFL-OA scores ex-
11.15, p < .001, h2 = .06. Participant self-reported plained significant added variability in current sui-
education was median split at 13 years, dividing cide ideation scores, above and beyond current
participants into those who had not continued depression scores, R2 = .30, DR2 = .04, DF(1, 176)
their education beyond high school (0–12 years: = 10.96, p < .001, after controlling for the demo-
n = 76) and those who had done so (13+ years: n = graphic variables (see Table 2). Similar findings
105). Findings supported an association between emerged when worst-episode suicide ideation was
reasons for living and education, as the group with treated as the dependent variable, R2 = .22, DR2 =
less formal education, M = 297.0, SD = 63.3, .10, DF(1, 148) = 18.91, p < .001 (see Table 3).
scored significantly higher on the RFL-OA than Criterion-related validity for the RFL-OA was
those with more education, M = 242.3, SD = 73.9, examined with lifetime history of suicidal behavior
F(1,179) = 27.21, p < .001, h2 = .13. treated as the criterion. Participants were divided
into two groups based on self-reported past sui-
cidal behavior; those having engaged in suicidal
Reliability behavior at any point in their life (n = 80) and
The results indicated robust internal consistency those never having done so (n = 92). Nine partici-
for the RFL-OA (Cronbach’s a = .98). pants did not provide sufficient information to as-
sess their history of suicidal behavior. Those with
histories of suicidal behavior, M = 244.4, SD =
Validity 74.9, reported significantly lower RFL-OA scores
Construct validity was assessed with zero-order than those without, M = 281.8, SD = 72.0, F(1,170) =
correlations between RFL-OA total scores and the 11.08, p < .001, h2 = .06. A logistic regression analysis
measures examined in the present study (see Table 1). was conducted next, examining whether RFL-OA
Significant associations among RFL-OA scores and scores predicted suicidal behavior status after control-
measures of current suicide ideation (SSI-C: ling for participant age and sex (see Table 4). An om-
r = −.40, p < .001), suicide ideation at the worst nibus test of the logistic regression model was
point in one’s life (SSI-W: r = −.42, p < .001), and significant, c2(df = 3, n = 172) = 24.4, p < .001; age
depression (BDI-II: r = −.43, p < .001) attested to (Wald statistic = 11.7, p < .001) and RFL-OA
the measure’s convergent validity. Discriminant va- scores (Wald statistic = 7.2, p < .01) both signifi-
lidity was indicated by nonsignificant correlations cantly differentiated the two groups. Odds ratios
between the RFL-OA and both mental status indicated that age, exp (B) = .94, and RFL-OA
(MMSE: r = −.04, p = .64) and physical functioning scores, exp (B) = .99, were associated with a slight-
(IADL: r = −.05, p = .55; PSMS: r = .02, p = .84). ly, although significantly, lower likelihood of hav-
A pair of hierarchical multiple regression analy- ing engaged in suicidal behavior.
ses was computed next exploring the incremental
validity of the RFL-OA in predicting current and
worst-episode suicide ideation scores above and Discussion
beyond demographic variables (age and sex) The foregoing two studies described the devel-
and depression. Participant age and sex were opment and initial psychometric evaluation of the

742 The Gerontologist

RFL-OA. The initial development of the scale fol- Clinicians are encouraged to discuss both sides
lowed the procedures of those used by Linehan of the suicide equation; the reasons for wanting to
and colleagues (1983) in developing their original take one’s life, and one’s psychological strengths
reasons for living Inventory. The age-related change and life-affirming reasons for not doing so (Heisel
in content validity is revealed by the current find- & Flett, 2004, 2008). Part of a good suicide risk
ing that only 28 of the 69 RFL-OA items, devel- assessment includes assessment of the presence of
oped with older adults, were the same or similar to protective or resiliency factors, which comple-
those of the original reasons for living, which were ments the conventional assessment of pathology
developed with younger adults. This finding is con- and risk factors. This is consistent with the work
sistent with Koven and colleagues’ (2001) finding of suicide researchers and guidelines that have been
of age-related changes in reasons for living from developed for the assessment of suicide risk and

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

adolescence through older adulthood. Internal the prevention of suicide (see http://www.ccsmh.
consistency was excellent. ca/en/projects/suicide.cfm).
The results of Study 2, in which the psychomet- There are a few limitations to our findings that
ric characteristics of the RFL-OA were examined warrant consideration. The original item develop-
with mental health patients, revealed strong reli- ment sample could have been larger and more di-
ability and validity. Internal consistency statistics verse in terms of ethnicity, race, religion, and
supported the measure’s internal reliability. Find- geographic region. Thus, the generalizability of
ings indicate strong construct validity for the our results should be considered when interpreting
RFL-OA, including significant convergence with our findings. Ultimately, the utility of the RFL-OA
measures of depression and of both current and will have to be reaffirmed with additional research
worst-episode suicide ideation. RFL-OA scores with diverse populations across a wide range of
further explained significant added variance in cur- geographic regions. Another potential limitation
rent and worst-episode suicide ideation scores was the possibility that the instructions to the par-
above and beyond demographic variables (age and ticipants, which referred to suicide, may have in-
sex) and depression, providing potential evidence fluenced their ratings. Ideally, participants should
of the measure’s incremental validity with respect have been blinded to the purpose of the scale.
to recent and remote suicidal thoughts. Nonsignifi- Future researchers may consider exploring the
cant associations between reasons for living and dimensional structure of the instrument and wheth-
current mental and physical functioning attested to er specific dimensions or items better differentiate
the measure’s discriminant validity. RFL-OA scores between suicidal and nonsuicidal older adults.
distinguished those with a lifetime history of sui- Findings that older adults scored significantly
cidal behavior, providing evidence for the mea-
higher on the original reasons for living Invento-
sure’s criterion-related validity. Future research is
ry’s moral objections subscale, and higher on the
needed to demonstrate criterion validity in dis-
religion items of the RFL-OA, for example, sug-
criminating between clinical and nonclinical sam-
gest the relevance of attending to moral and spiri-
ples of older adults, and predictive validity with
tual reasons for living in older clientele (Miller et al.,
respect to the advent of suicidal thoughts and be-
2001); however, it is unclear whether these differ-
havior among older adults with no such history.
ences can be ascribed to aging or to cohort effects.
The RFL-OA scores were negatively associat-
Future research is needed to explore the stability
ed with recent and remote suicidal thoughts and
of the measure over time and its sensitivity as a
distinguished between participants with respect
potential measure of clinical change. Clinical re-
to history of suicidal behavior. The findings sup-
searchers are encouraged to attend to reasons for
port the potential value of exploring reasons for
living when assessing and treating older adults living and related adjustment and resiliency factors
who are at risk for suicide. This could be done as in promoting psychological well-being and in po-
one element of an exploration of an individual’s tentially protecting against suicide among older
overall adaptive behaviors, psychological adjust- adults (Canadian Coalition for Seniors’ Mental
ment, and coping skills. The collective consider- Health, 2006; Heisel & Flett, 2004, 2007, 2008).
ation of these more positive factors is supported Finally, The RFL-OA may also have promise
by findings of salient associations between rea- for the field of positive psychology. Seligman and
sons for living and measures of psychological Csikszentmihalyi (2000) have aptly noted that we
adjustment and coping (Range & Stringer, 1996). know little about how normal individuals thrive

Vol. 49, No. 6, 2009 743

when not faced with adversity. The RFL-OA could Current instrument status and related considerations. Clinical Geron-
tologist, 31, 1–35.
be used to begin exploring valued subjective experi- First, M. B., Spitzer, R. L., Gibbon, M., & Williams, J. B. W. (1997). Struc-
ences (e.g., happiness, hope, optimism, well-being) tured clinical interview for DSM-IV Axis I disorders-patient edition
(SCID-I/P, version 2.0, 4/97 revision). New York: Biometrics
among older adults who are not at risk for suicide. Research Department, New York State Psychiatric Institute.
Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). ‘Mini-mental
state’: A practical method for grading the cognitive state of patients for
Funding the clinician. Journal of Psychiatric Research, 12, 189–198.
Work on Study 2 was funded in part by U.S. Public Health Service Gutierrez, P. M., Osman, A., Barrios, F. X., Kopper, B. A., Baker, M. T., &
Grants R01-MH-064579 and K24MH072712 (Paul R. Duberstein); by a Haraburda, C. M. (2002). Development of the reasons for living inven-
Leonard F. Salzman Research Award of the Department of Psychiatry, tory for young adults. Journal of Clinical Psychology, 58, 339–357.
University of Rochester Medical Center (Marnin J. Heisel); by an Ameri- Heisel, M. J., & Duberstein, P. R. (2005). Suicide prevention in older
can Foundation for Suicide Prevention Young Investigator Award (Marnin adults. Clinical Psychology: Science and Practice, 12, 242–259.
J. Heisel); and by a Canadian Institutes of Health Research New Investiga- Heisel, M. J., & Flett, G. L. (2004). Purpose in life, satisfaction with life
tor Award (Marnin J. Heisel). and suicide ideation in a clinical sample. Journal of Psychopathology

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

and Behavioral Assessment, 26, 127–135.
Heisel, M. J., & Flett, G. L. (2006). The development and initial validation
Acknowledgments of the Geriatric Suicide Ideation Scale. American Journal of Geriatric
We appreciate the assistance of Anthony Beckman, Kenneth Conner, Psychiatry, 14, 742–751.
Yeates Conwell, Nathan Franus, Cindy Hutton, Kendra Marinucci, Silvia Heisel, M. J., & Flett, G. L. (2007). Meaning in life and resilience to sui-
Sörensen, J. David Useda, Holly Wadkins, and Patrick Walsh. We thank cidal thoughts among older adults. In P. T. P. Wong, L. C. J. Wong, M.
you the study participants, as well. Earlier versions of this article were McDonald, & D. Klaassen (Eds.), The positive psychology of meaning
presented at annual meetings of the American Psychological Association in and spirituality (pp. 183–196). Abbotsford, British Columbia, Canada:
Toronto, Canada (2003); the American Association of Suicidology in INPM Press.
Seattle, Washington (2006); and the Canadian Association for Suicide Heisel, M. J., & Flett, G. L. (2008). Psychological resilience to suicide
Prevention in Toronto, Canada (2006). ideation among older adults. Clinical Gerontologist, 31, 51–70.
Heisel, M. J., Flett, G. L., & Besser, A. (2002). Cognitive functioning and
geriatric suicide ideation. American Journal of Geriatric Psychiatry,
References 10, 428–436.
American Psychological Association. (2004). Guidelines for psychological Koven, L. P., Edelstein, B., & Charlton, A. S. (2001, November). Reasons
practice with older adults. American Psychologist, 59, 236–260. for living across the lifespan. Poster presented at meeting of the Geron-
Beck, A. T., Brown, G. K., Steer, R. A., Dahlsgaard, K. K., & Grisham, J. R. tological Society of America, Chicago.
(1999). Suicide ideation at its worst point: A predictor of eventual sui- Lawton, M. P., & Brody, E. M. (1969). Assessment of older people: Self-
cide in psychiatric outpatients. Suicide and Life-Threatening Behavior, maintaining and instrumental activities of daily living. Gerontologist,
29(1), 1–9. 9, 179–186.
Beck, A. T., Kovacs, M., & Weissman, A. (1979). Assessment of suicidal Lawton, M. P., & Brody, E. M. (1988a). Instrumental activities of daily
intention: The scale for suicide ideation. Journal of Consulting and living (IADL): Original observer-rated version. Psychopharmacology
Clinical Psychology, 47(2), 343–352. Bulletin, 24, 785–791.
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). BDI-II manual. San Lawton, M. P., & Brody, E. M. (1988b). Physical self-maintenance scale
Antonio, TX: Psychological Corporation. (PSMS): Original observer-related version. Psychopharmacology Bul-
Beck, A. T., Weissman, A., Lester, D., & Trexler, L. (1974). The measure- letin, 24, 793–797.
ment of pessimism: The hopelessness scale. Journal of Consulting and Linehan, M. M., Goodstein, L. J., Nielsen, S. L., & Chiles, J. A. (1983).
Clinical Psychology, 42, 861–865. Reasons for staying alive when you are thinking of killing yourself:
Brown, G. K. (1999). A review of suicide assessment measures for inter- The Reasons for Living Inventory. Journal of Consulting and Clinical
vention research with adults and older adults. Retrieved June 14, 2005, Psychology, 51, 276–286.
from http://www.nimh.nih.gov/suicideresearch/adultsuicide.pdf Luoma, J. B., Martin, C. E., & Pearson, J. L. (2002). Contact with mental
Canadian Coalition for Seniors’ Mental Health. (2006). National guide- health and primary care providers before suicide: A review of the evi-
lines for seniors’ mental health: The assessment of suicide risk and dence. American Journal of Psychiatry, 159, 909–916.
prevention of suicide. Toronto, Ontario, Canada: Author. Miller, J. S., Segal, D. S., & Coolidge, F. L. (2001). A comparison of sui-
Cole, D. A. (1989). Validation of the reasons for living inventory in gen- cidal thinking and reasons for living among younger and older adults.
eral and delinquent adolescent samples. Journal of Abnormal Child Death Studies, 25, 357–365.
Psychology, 17, 13–27. National Center for Injury Prevention and Control. (2007). 1999–2002,
Conwell, Y., Olsen, K., Caine, E. E., & Falnnery, D. (1991). Suicide in late United States suicide injury deaths and rates per 100,000. Retrieved
life: Psychological autopsy findings. International Psychogeriatrics, 3, May 14, 2008, from http://www.cdc.gov/ncipc/wisqars/
59–66. Osman, A., Downs, W. R., Kopper, B. A., Barrios, F. X., Baker, M. T., Os-
Crum, R. M., Anthony, J. C., Bassett, S. S., & Folstein, M. F. (1993). man, J. R., et al. (1998). The reasons for living inventory for adoles-
Population-based norms for the Mini-Mental State Examination by cents (RFL-A): Development and psychometric properties. Journal of
age and educational level. Journal of the American Medical Associa- Clinical Psychology, 54, 1063–1078.
tion, 269, 2386–2391. Osman, A., Gifford, J., Jones, T., Lickiss, L., Osman, J., & Wenzel, R.
Diekstra, R. F. W., & van Egmond, M. (1989). Suicide and attempted sui- (1993). Psychometric evaluation of the reasons for living inventory.
cide in general practice, 1979–1985. Acta Psychiatrica Scandinavica, Psychological Assessment, 5, 154–158.
79, 268–275. Pearson, J., & Brown, G. (2000). Suicide prevention in late-life: Directions
Dillman, D. A. (2000). Mail and internet surveys: The tailored design for science and practice. Clinical Psychology Review, 20, 685–705.
method. New York: Wiley. Range, L. M., & Stringer, T. A. (1996). Reasons for living and coping
Draper, B. (1996). Attempted suicide in old age. International Journal of abilities among older adults. International Journal of Aging and Hu-
Geriatric Psychiatry, 11, 577–587. man Development, 43, 1–5.
Edelstein, B. A., Kalish, K., Drozdick, L., & McKee, D. (1999). Assessment Seligman, M. E. P., & Csikszentmihalyi, M. (2000). Positive psychology:
of depression and bereavement in older adults. In P. A. Lichtenberg (Ed.), An introduction. American Psychologist, 55, 5–14.
Handbook of geriatric assessment (pp. 11–58). New York: Wiley. Steer, R. A., Rissmiller, D. J., & Beck, A. T. (2000). Use of the Beck De-
Edelstein, B. A., McKee, D. R., & Martin, R. R. (2000, September). Devel- pression Inventory-II with depressed geriatric inpatients. Behaviour
opment of the reasons for living scale for older adults: A suicide assess- Research and Therapy, 38, 311–318.
ment instrument. Poster presented at the Harvard Symposium on
Future Research Trends and Opportunities in Aging, Boston. Received July 23, 2008
Edelstein, B. A., Woodhead, E. L., Segal, D. L., Heisel, M. J., Bower, E. H., Accepted January 8, 2009
Lowery, A. J., et al. (2008). Older adult psychological assessment: Decision Editor: William J. McAuley, PhD

744 The Gerontologist

 Appendix
Sample Reasons for Living—Older Adults Scale Items

1 2 3 4 5 6
Extremely Quite Somewhat Somewhat Quite Extremely
unimportant unimportant unimportant important important important
It would hurt my family too much, I would not want them to suffer.
My religious beliefs forbid it.
I believe only God has the right to end life.
I am afraid of going to hell.
Tomorrow I may feel better.
I want to see my grandchildren grow up.
I love and enjoy my family too much and could not leave them.
I have the hope that things will improve and the future will be happier.

Downloaded from https://academic.oup.com/gerontologist/article-abstract/49/6/736/694744 by guest on 21 May 2020

I still have many things left to do.
My family depends on me and needs me.
Life is too beautiful and precious to end it.
I can always think of someone else who is worse off than I am.
I am concerned about what others would think of me.
I do not want to die.
I consider it morally wrong.

Vol. 49, No. 6, 2009 745