Advancing depression assessment in older adults with cancer: Development and validation of the Older Adults with Cancer‐Depression Scale (OAC‐D), a novel, patient‐reported outcome

Abstract

Background

Depression in older adults with cancer (OACs) is poorly captured by patient‐reported outcomes (PROs) because traditional criteria from the Diagnostic and Statistical Manual of Mental Disorders (DSM) may overlook unique symptoms in OACs. By using US Food and Drug Administration Guidance for Industry on PRO development, the authors conceptualized depression in OACs and created the Older Adults with Cancer‐Depression Scale (OAC‐D). This study evaluates the psychometric properties of this novel measure.

Methods

Based on a refined conceptual model informed by literature reviews and qualitative work with patients and experts in geriatric oncology, a 35‐item draft PRO was developed. The draft was administered, alongside legacy measures, to OACs aged >70 years and older at a comprehensive cancer center and across the United States. Exploratory graph analysis was used to identify items and factors.

Results

The mean ± standard deviation age of the 155 participants was 76 ± 5 years, and 56% were women. Exploratory graph analysis yielded an 18‐item measure with five domains: (1) interest and enjoyment, (2) purpose and meaning, (3) loneliness, (4) social withdrawal, and (5) regret and guilt. Internal consistency (Cronbach alpha, .85–.95) and test‐retest reliability (intraclass correlation coefficient = 0.61–0.80) were strong. Convergent validity was supported by correlations with the Patient Health Questionnaire‐9 (r = 0.75), the Patient‐Reported Outcomes Measurement Information System anxiety score (r = 0.69), and the Patient‐Reported Outcomes Measurement Information System physical score (r = −0.39). Known groups analysis demonstrated higher OAC‐D scores for those with a history of depression (p < .001).

Conclusions

The OAC‐D identified five unique domains of depression, only one of which overlapped with DSM criteria. It demonstrated robust psychometric properties, providing a nuanced alternative to DSM‐based measures and addressing the distinctive psychological challenges of cancer and aging.

Short abstract

The Older Adults with Cancer‐Depression Scale is a novel patient‐reported outcome designed to address the unique expression of depression in older adults with cancer. Grounded in a comprehensive conceptual model informed by systematic literature reviews, patient perspectives, and expert input, the Older Adults with Cancer‐Depression Scale offers a meaningful alternative to traditional Diagnostic and Statistical Manual of Mental Disorders‐based measures by incorporating domains that reflect the nuanced experiences of this population, such as loneliness and loss of purpose.

INTRODUCTION

Addressing depression among older adults with cancer (OACs) is a US National Cancer Institute (NCI) priority. ¹ , ² , ³ , ⁴ , ⁵ , ⁶ Nearly 60% of new cancer diagnoses occur in those aged 65 years and older. ⁴ , ⁷ According to the 2020 census, there are greater than 55.8 million older adults living in the United States, and this number is expected to climb to 80 million by 2040. Thus the number of OACs will continue to significantly increase in the next several decades. Cancer in older age is often associated with significant mental health concerns, such as depression, with a prevalence in OACs as high as 25%. ⁸ , ⁹ , ¹⁰ , ¹¹ Comorbid cancer and depression can lead to declines in functioning and health‐related quality of life, increased pain and fatigue, treatment interference and discontinuation, higher health care costs, and, in the extreme, premature mortality and suicide. ¹² , ¹³ , ¹⁴

Despite the high prevalence and deleterious effects of depression, older adults are far less likely to be appropriately diagnosed with depression (major or minor) than any other age group. ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹ , ²⁰ For patients with cancer of any age, a primary source of difficulty lies in the overlap between the diagnostic criteria for depression and many cancer symptoms and treatment side effects (e.g., weight loss, abnormal sleep, fatigue, diminished concentration). Consequently, current research supports using the two gateway symptoms of depressed mood and loss of interest as the most clinically relevant symptoms to differentiate depression in patients with cancer. ²¹ However, in patients without cancer, depressed older adults, compared with depressed younger adults, may be less likely to endorse these two gateway symptoms with depression. ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ Depression in older adults is often characterized as depression without sadness because older adults tend not to report depressed mood as a salient symptom of depression. ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ Thus for OACs, when combining the challenges of diagnosing depression in individuals who have cancer with the complexities of detecting depression in older adults, there may be few, if any, Diagnostic and Statistical Manual of Mental Disorders (DSM)‐based symptoms that consistently identify depression.

Although patient‐reported outcomes (PROs) are one way to systematically identify patients who are suffering from depression or depressive symptoms, PROs are generally based on DSM diagnostic criteria or classic symptoms of depression. This, as stated above, may not be ideal for identifying depressive symptoms in OACs. Indeed, there is evidence that self‐report scales significantly underrepresent the severity of depressive symptoms compared with expert clinician diagnosis in older adults. ²⁴ Researchers reported that, in 810 older patients who had mood disorders, the severity of depressive symptoms decreased with age when depression was assessed with self‐report instruments, whereas an increase in the severity of depression with age was observed when analyzing expert clinician assessments. ²⁴ These data indicate that older patients do not endorse items on standard PRO depression instruments. A study of 201 OACs supports this conclusion as OACs rarely endorsed items on commonly used PROs to assess depression and the PROs had the potential to miss up to 83% of OACs who were indeed experiencing clinically significant depressive symptoms. ²⁸ When items were endorsed, the overlap between classic depressive symptoms and the symptoms of cancer and side effects becomes problematic.

Given the importance of identifying and treating depression, the difficulty in diagnosing depression, and the potential shortcomings of existing PROs for assessing depression in OACs, we sought to investigate the conceptualization of depression and improve the accuracy of depression screening in OACs. This includes the development of a new PRO using the US Food and Drug Administration's (FDA's) Guidance for Industry for PRO Use in Medical Product Development to Support Labeling Claims (i.e., the gold standard for PRO development; hereafter referred to as FDA PRO Guidance), which highlights the importance of establishing the validity of item content with substantial patient qualitative input. This report describes the initial psychometric validation of a newly developed PRO, the Older Adults with Cancer‐Depression Scale (OAC‐D).

MATERIALS AND METHODS

Scale development

The FDA PRO Guidance provided the framework for developing the OAC‐D with the completion of phases 1 and 2. For a detailed description of the previous steps completed to date, please see Saracino et al., 2025. ²⁹ A brief description of these steps is listed below. This article is primarily focused on phase 2.

FDA PRO Guidance phase 1: Hypothesize conceptual framework

By using the following steps, the research team identified gaps in conceptualizing and screening for depression in OACs indicating a clear opportunity to improve the assessment of depression for OACs: (1) we conducted a systematic review of the development and psychometric properties of the eight most used depression PROs in geriatric, cancer, and geriatric cancer populations ³⁰ ; (2) we conducted a study grounded in classical test theory to examine the performance of common depression PROs among OACs ²⁸ ; (3) we conducted an extensive narrative review of 90 articles that focused on the phenomenology (i.e., unique symptom presentation) of depression in older adults, individuals with cancer, and their overlap with one another and with DSM diagnostic criteria ³¹ (Table 1); and (4) we implemented semistructured interviews of internationally recognized experts in geriatric psychiatry and oncology (N = 8 ³² ; Table 1).

TABLE 1.

Possible signs, symptoms, and themes of depression in older adults with cancer: Divergence from Diagnostic and Statistical Manual of Mental Disorders criteria.

Criteria	Narrative literature review (Saracino 2016 31 )	Clinical experts (Saracino 2024 32 )	Qualitative study with patients (Saracino 2023 33 )	Final OAC‐D domains
DSM criteria	‐Anhedonia ‐Concentration disturbances	‐Anhedonia	‐Interest and enjoyment (anhedonia) a ‐Depressed mood	‐Loss of interest and enjoyment (anhedonia)
Non‐DSM criteria	‐Irritability ‐Social withdrawal ‐Loneliness ‐Loss of purpose ‐Having “nothing to do” ‐Memory changes ‐Psychic Tension (e.g., worry, apprehension)	‐Loneliness and social withdrawal ‐Malaise ‐Tearfulness and crying ‐Loss of meaning and purpose ‐Negative ruminations ‐Hopelessness	‐Loneliness/social withdrawal a ‐Lack of meaning/purpose a ‐Lack of usefulness/feeling like a burden a ‐Regret and guilt ‐Attitude toward treatment ‐Attitude toward physical symptoms/limitations	‐Loss of purpose and meaning ‐Loneliness ‐Social withdrawal ‐Regret and guilt

Abbreviations: DSM, Diagnostic and Statistical Manual of Mental Disorders; OAC‐D, Older Adults with Cancer‐Depression Scale.

^a In the column headed Qualitative study with patients, this is a major theme; others in the column are minor themes.

FDA PRO Guidance phase 2: Adjust conceptual framework, elicit content from patients, and develop and establish content validity of draft instrument items

The research team took the following steps to adjust the conceptual framework of depression in OACs and develop the draft instrument: (1) We used the refined conceptual framework of depression (e.g., based on existing literature and expert interviews used in phase 1) through semistructured qualitative interviews with 26 older adults who had a history of cancer (depressed, n = 13; nondepressed; n = 13). ³³ Qualitative analyses revealed four major and four minor themes (Table 1). Six of these themes departed from DSM symptoms. (2) With these constructs defined, the team generated a list of 38 candidate items (i.e., four or five items for each of the eight domains) for a draft OAC‐D scale. (3) We obtained feedback through cognitive interviews with 10 OACs (i.e., aged 70 years and older, diagnosed with cancer, English‐speaking) who had a history of depression. We conducted qualitative analyses to refine items and language, eliminate redundancies, and determine the patient‐preferred format and response options before piloting it, resulting in a content‐valid 35‐item draft measure with a recall period of 1 week. Each question was asked with a beginning stem stating, “In the last week I noticed that…”, and the response format was a five‐point Likert‐type scale scored from 0 to 4 with response options of never, rarely, sometimes, often, or always. ²⁹ For a detailed description of the cognitive interview process, see Saracino et al., 2025. ²⁹

The next step in phase 2 of the FDA PRO Guidance is pilot testing and establishing the initial psychometric evaluation of a measure. In the current study, we conducted a study of the psychometric properties of the OAC‐D.

Participants

OACs were recruited from the Memorial Sloan Kettering Cancer Center (MSK) and across the United States using the National Institutes of Health–funded platform ResearchMatch.org. Inclusion criteria were: (1) current or previous cancer diagnosis and treatment (any site and any stage); (2) aged 70 years or older; and (3) able to communicate, comprehend, and complete questionnaires in English. The age cutoff of 70 years and older was chosen because, in previous qualitative work with patients and collaborating researchers, 65 years was considered too young to be classified as older, whereas 70 years was viewed as a more appropriate threshold. In addition, to test for known group differences, approximately one quarter of the sample who had a history of depressive symptoms was recruited from the MSK Counseling Center to help define a depressive symptoms group. This additional criterion was assessed according to medical record or self‐report as a current diagnosis or history of one of the following to have a wide range of depressive symptom presentation: adjustment disorder with depressed mood, adjustment disorder with mixed depressed mood and anxiety, mood disorder (i.e., because of general medical condition, not otherwise specified), and depressive disorder (i.e., major depressive disorder single episode or recurrent, depressive disorder not otherwise specified, dysthymia). Exclusion criteria were severe psychopathology or cognitive impairment likely to interfere with participation, completion of the protocol, or ability to provide meaningful information. All participants provided informed consent under MSK Institutional Review Board‐approved protocol 14‐101.

Procedures

Participants completed the novel OAC‐D and several existing questionnaires using REDCap (Research Electronic Data Capture; Vanderbilt University) or by using pen and paper. To assess test‐retest reliability, a second timepoint consisting only of the draft OAC‐D was administered approximately 1–2 weeks after the first assessment.

Measures

To assess convergent and discriminant validity, participants completed the following questionnaires: a sociodemographic questionnaire, the Center for Epidemiological Studies Depression Scale, the Patient Health Questionnaire‐9, the Hospital Anxiety and Depression Scales, the Patient‐Reported Outcomes Measurement Information System (PROMIS)‐Anxiety short form 8a (PROMIS‐Anxiety), and the PROMIS‐Physical Function short form 10a. ³⁴

Statistical methods

Exploratory graph analysis (EGA) ³⁵ , ³⁶ was used to estimate the number of dimensions and identify the most suitable structure among the items of the OAC‐D. Generally, this technique involves determining and eliminating high redundancies among the items (step 1), performing the EGA to identify structures (step 2), and checking stability using the bootstrap EGA procedure (step 3). ³⁷ The sample of respondents who completed the OAC‐D and distributions of each item are described. Redundant items were eliminated if a pair had a weighted topological overlap (WTO) of 0.3. ³⁸ For the EGA, we used the Glasso technique with the walktrap algorithm and generated 1000 bootstrap samples with resampling. The replication estimate indicates the proportion of bootstrap samples with consistent membership; according to Christensen and Golino (2021), replication estimates ≥75% indicate good stability. ³⁷ A structural equation model was used to test a total score of these factor scores with equal loadings; the comparative fit index, the Tucker–Lewis index, and the root mean square error of approximation were used to assess fit. The model supported by the EGA was then assessed for reliability using the Cronbach alpha to measure the internal consistency of each factor, the intraclass correlation coefficient (ICC) for test‐retest reliability, ³⁹ and a series of correlations with other known constructs (i.e., psychosocial measures, disease characteristics, and sociodemographics) for construct validity. Analyses were conducted in R version 4.3.3using the EGAnet package (The R Project for Statistical Computing), SAS version 9.4 (SAS Institute, Inc.), and MPlus (version 8.11; Muthen & Muthen) for the structural equation modeling.

RESULTS

Sample descriptives

Although 156 participants enrolled in the study, one did not complete any of the questionnaire data, resulting in the inclusion of 155 participants in the analysis. Two thirds (n = 104; 67%) of participants were recruited from MSK, and the remainder were recruited from ResearchMatch. Over one third (n = 61; 39%) were either recruited from the MSK Counseling Center (n = 40) or reported having been diagnosed with depression or bipolar mood disorder (12 from ResearchMatch and nine from MSK). This formed a depressive symptoms group to test known group differences. The sample was primarily White (n = 145; 94%), non‐Hispanic (n = 149; 97%), married (n = 99; 64%), retired (n = 115; 75%), and had at least a college degree (n = 115; 75%). Participants ranged in age from 70 to 94 years (median, 76 years). There were slightly more women (n = 86; 56%) than men, and about one half (n = 79; 51%) of the group indicated they had previously received treatment or counseling for an emotional problem. One half (n = 69; 52%) of the group had a disease stage of III or IV, and about one quarter (n = 41; 27%) had received chemotherapy in the previous 6 months. Only 7% (n = 11) of the sample rated their health status as excellent, whereas 22% (n = 34) rated their health status as fair or poor. Descriptive characteristics are listed in Table 2.

TABLE 2.

Participant characteristics, N = 155.

Characteristic	No. (%)	Characteristic	No. (%)
Site: MSK	104 (67.0)	Previous counseling/treatment: Yes	79 (51.0)
From counseling center	40 (26.0)	Current counseling/treatment: Yes	46 (30.0)
Age: Mean ± SD, years	76.4 ± 4.7	Disease stage, n = 132
Sex, women	86 (56.0)	0	14 (11.0)
Race		I	35 (27.0)
White	145 (94.0)	II	14 (11.0)
Black or African American	5 (3.0)	III	30 (23.0)
Other	4 (3.0)	IV	39 (30.0)
Hispanic: Yes	5 (3.0)	Chemotherapy in past 6 months: Yes	41 (27.0)
Marital status		Hormonal therapy in past 6 months: Yes	31 (20.0)
Married/live with partner	99 (64.0)	Surgery in past 6 months: Yes	26 (17.0)
Single	17 (11.0)	Radiotherapy in past 6 months: Yes	30 (19.0)
Divorced/separated	16 (10.0)	Self‐rated health status, n = 153
Widowed	22 (14.0)	Excellent	11 (7.0)
Educational attainment		Very good	41 (27.0)
High school or less	13 (8.0)	Good	67 (44.0)
Some college	26 (17.0)	Fair	28 (18.0)
College graduate	36 (23.0)	Poor	6 (4.0)
Graduate degree	79 (51.0)	Psychosocial measures: Mean ± SD score
Employment: Retired	115 (75.0)	CES‐D	16.4 ± 7.3
Annual household income		PHQ‐9, n = 153	4.8 ± 4.9
<$40,000	23 (15.0)	HADS Anxiety, n = 152	4.8 ± 3.8
$40,000–$75,000	29 (19.0)	HADS Depression, n = 152	4.3 ± 3.8
$75,000–$150,000	50 (32.0)	PROMIS Physical Function, n = 151	40.4 ± 7.3
>$150,000	40 (26.0)	PROMIS Anxiety, n = 151	13.8 ± 6.4
Missing/refused	12 (8.0)	Depressive symptoms: Yes	61 (39.0)

Note: One participant was missing all baseline data and thus is included only in the site and group characteristics.

Abbreviations: CES‐D, Center for Epidemiological Studies Depression Scale; HADS, Hospital Anxiety and Depression Scale; MSK, Memorial Sloan Kettering Cancer Center; PHQ‐9, Patient Health Questionnaire‐9; PROMIS, Patient‐Reported Outcomes Measurement Information System; SD, standard deviation.

Item descriptives

The 35 original items with their response distributions are depicted in Figure 1. Items that were most heavily endorsed included: “It was hard for me to push myself to do new things” (19% responded often or always, n = 30), “I spent time thinking about my physical limitations” (17% responded often or always, n = 26), and “My physical problems kept me from living my life” (20% responded often or always, n = 31). Items that were the least endorsed included: “I felt like a burden” (5% responded often or always, n = 8; 63% responded never, n = 97), “I did not feel needed” (2% responded often or always, n = 3; 65% responded never, n = 101), “I no longer felt useful to others” (3% responded often or always, n = 5; 62% responded never, n = 96), and “I could not be cheered up” (4% responded often or always, n = 6; 61% responded never, n = 94).

FIGURE 1.

Item endorsement distributions. Distribution of responses for each of the 35 original Older Adults with Cancer‐Depression Scale items.

EGA results

In the redundancy analysis (step 1), six pairs of items were identified that had large redundancies. This resulted in removal of the following six items: “I lacked companionship” (redundant with “I am lonely”; WTO = 0.35), “I no longer felt useful to others” (“I did not feel needed”; WTO = 0.33), “I worried about being a burden to my family” (“I felt like a burden”; WTO = 0.32), “My treatment was a hassle” (“I felt resentful about my medical treatments”; WTO = 0.43), “I looked back on my life and had regrets” (“I spent time thinking about my regrets”; WTO = 0.42), and “My physical problems kept me from living my life” (“I spent time thinking about my physical limitations”; WTO = 0.42). Thus 29 items were eligible for step 2. Although the initial EGA supported a six‐dimensional solution, the bootstrap stability analysis (step 3) supported a five‐dimensional solution (median, five dimensions; 36% of simulations).

The bootstrap EGA item replication analysis identified from three to five items for each of five factors, for a total of 18 items. These five factors appeared to delineate subdomains of (1) interest and enjoyment, (2) purpose and meaning, (3) loneliness, (4) social withdrawal, and (5) regret and guilt. All items for three of these factors (interest and enjoyment, social withdrawal, and regret and guilt) had replication estimates indicating robust stability. Replication for the items on the other two factors (purpose and meaning and loneliness) ranged between 58% and 66%, indicating suboptimal stability. The final 18 items with replication results are provided in Table 3 (for the final OAC‐D PRO, see Supporting Information). Finally, a structural equation model with the mean as a summary score of the five factors indicated good Tucker–Lewis index and comparative fit index values (0.97 for each) and an acceptable root mean square error of approximation (0.055; 95% confidence interval, 0.04–0.07).

TABLE 3.

Domain and item reliability statistics.

Domain/item	Replication percent	Cronbach alpha	ICC (test‐retest reliability)
Interest and enjoyment		.93	0.80
I did not enjoy things.	0.84
I was just going through the motions.	0.84
It was harder for me to try new things.	0.87
I was not as interested in things.	0.86
It was hard for me to push myself to do new things.	0.87
Purpose and meaning		.93	0.76
My life lacked a sense of purpose.	0.62
I felt like I was just taking up space.	0.58
The things I did were not worthwhile.	0.61
Life was not as meaningful to me.	0.61
Loneliness		.86	0.74
I was lonely.	0.66
I had no one to talk to when I was feeling down.	0.63
I wished I had more contact with other people.	0.65
Social withdrawal		.84	0.61
I avoided spending time with others.	0.84
I struggled to maintain my relationships.	0.84
I did not look forward to spending time with my family or friends.	0.76
Regret and guilt		.85	0.79
I felt guilty about things (from my past).	0.88
I spent time thinking about my regrets.	0.87
I blamed myself for the bad things that happened in my life.	0.88
Total		.95	0.81

Abbreviation: ICC, intraclass correlation coefficient.

Factor distributions, reliability, and validity

Figure 2 depicts the distribution for each factor at baseline. Approximately one fifth of participants had a mean score of at least 2 (indicating sometimes) on the factors related to interest/enjoyment (21%) and loneliness (18%); and slightly lower proportions had a score of at least 2 on purpose/meaning (13%), social withdrawal (14%), and regret and guilt (17%).

FIGURE 2.

Domain score distributions (baseline). Distributions of scores for each of the five identified Older Adults with Cancer‐Depression Scale domains at baseline: (1) interest and enjoyment, (2) purpose and meaning, (3) loneliness, (4) social withdrawal, and (5) regret and guilt.

Cronbach alpha was excellent for all factors and for the total score (i.e., Cronbach alpha, .84–.93; Table 3). Test‐retest reliability (ICC) was excellent for four factors (interest/enjoyment, ICC = 0.80; purpose/meaning, ICC = 0.76; loneliness, ICC = 0.74; guilt/regret, ICC = 0.79) and the total score (ICC = 0.81), and the remaining factor had acceptable reliability (social withdrawal, ICC = 0.61; Table 3). Construct validity (Table 4) indicated that all five factor scores were significantly, positively associated with self‐rated health status (with higher scores indicating poorer health): Patient Health Questionnaire‐9, the Center for Epidemiological Studies Depression Scale, the Hospital Anxiety and Depression Scales, and the PROMIS‐Anxiety measure. All of these factor scores also were negatively correlated with PROMIS physical functioning. The strongest correlations for all factor scores were with the legacy depression (r = 0.45–0.71) and anxiety (r = 0.52–0.63) measures. Correlations to health status (r = 0.29–0.44) and PROMIS physical functioning (r = −0.27 to −0.39) were in the medium range. There were no significant correlations with age. Furthermore, all factor scores were significantly associated with known groups (i.e., the outpatient oncology group vs. the depressive symptoms group). The mean scores were higher in the depressive symptoms group versus the outpatient oncology group (Figure 3), with the smallest group difference in the guilt and regret domain (mean difference ± standard deviation, 0.68 ± 0.80; p < .001) and the largest group difference in the interest and enjoyment domain (mean difference ± SD, 0.89 ± 0.84; p < .001). The total score followed the patterns of the domain scores.

TABLE 4.

Construct validity and distributions.

Measure	Domain: Correlation (p)
	Interest and enjoyment	Purpose and meaning	Loneliness	Social withdrawal	Regret and guilt	Total score
Age, years	0.11 (.195)	0.07 (.378)	0.04 (.593)	0.05 (.518)	−0.13 (.114)	0.04 (.645)
Self‐rated health status	0.44 (< .001)	0.37 (< .001)	0.28 (< .001)	0.33 (< .001)	0.29 (< .001)	0.42 (< .001)
CES‐D	0.58 (< .001)	0.52 (< .001)	0.51 (< .001)	0.56 (< .001)	0.52 (< .001)	0.65 (< .001)
PHQ‐9	0.72 (< .001)	0.65 (< .001)	0.55 (< .001)	0.58 (< .001)	0.58 (< .001)	0.75 (< .001)
HADS Anxiety	0.58 (< .001)	0.55 (< .001)	0.59 (< .001)	0.63 (< .001)	0.60 (< .001)	0.71 (< .001)
HADS Depression	0.75 (< .001)	0.70 (< .001)	0.45 (< .001)	0.64 (< .001)	0.49 (< .001)	0.73 (< .001)
PROMIS Physical Function	−0.39 (< .001)	−0.32 (< .001)	−0.34 (< .001)	−0.29 (< .001)	−0.27 (.001)	−0.39 (< .001)
PROMIS Anxiety	0.56 (< .001)	0.58 (< .001)	0.59 (< .001)	0.61 (< .001)	0.52 (< .001)	0.69 (< .001)
Depressive symptoms, t (p value)	−6.00 (< .001)	−5.43 (< .001)	−4.80 (< .001)	−5.91 (< .001)	−4.84 (< .001)	−6.89 (< .001)
Distribution, mean ± SD	1.15 ± 0.93	0.79 ± 0.94	0.89 ± 0.94	0.81 ± 0.88	0.85 ± 0.86	0.90 ± 0.76

Abbreviations: CES‐D, Center for Epidemiological Studies Depression Scale; HADS, Hospital Anxiety and Depression Scale; PROMIS, Patient‐Reported Outcomes Measurement Information System; PHQ‐9, Patient Health Questionnaire‐9; SD, standard deviation; t, t value.

FIGURE 3.

Known group differences across Older Adults with Cancer‐Depression Scale domains. Comparison of mean domain scores between the outpatient oncology group and the depressive symptoms group. All five domains showed significantly higher scores among the depressive symptoms group.

DISCUSSION

This study outlines a program of research that sought to reconsider how depression is evaluated in OACs and to improve the accuracy of depression screening in this population. ²⁸ , ³⁰ , ³¹ , ³² , ³³ In doing so, the OAC‐D, a new PRO, was developed with close adherence to phases 1 and 2 of the FDA PRO Guidance. ⁴⁰ , ⁴¹ The OAC‐D is grounded in an enhanced conceptual model of depression based on systematic literature reviews and extensive, semistructured interviews with patients and experts in geriatric oncology and psychiatry. ³² , ³³ The psychometric analysis in the current study produced an 18‐item scale with good validity and reliability and with meaningful and significant differences among known groups consisting of five domains: (1) interest and enjoyment, (2) purpose and meaning, (3) loneliness, (4) social withdrawal, and (5) regret and guilt.

In developing the OAC‐D, care was taken to use a phenomenological approach to conceptualize the experiences of OACs with depression. Consistent with our prior literature reviews and qualitative analyses, symptoms of depression emerged that are unique compared with those outlined in DSM criteria for depression. This is seen in the final domains of the OAC‐D because only one domain, interest and enjoyment, overlaps with current DSM criteria (i.e., anhedonia; Table 1). It is noteworthy that DSM criteria for depression remain essentially unchanged since their original form in the 1980s, ⁴² and reliance on these criteria may limit the field's ability to consider the potential differential expression of depressive symptoms in individuals with cancer and in older adults. Thus the innovation of this approach is a program of research that proposes a theoretical reconceptualization of depression in OACs based on important patient feedback, rather than relying solely on DSM criteria. While beginning to consider alternative depression criteria, other studies have continued to rely on existing self‐report measures of depression that are largely grounded in DSM criteria and have not included patient input in their development (i.e., to ensure that information obtained from patients accurately describes their experiences and symptoms as recommended by both the FDA and the National Institutes of Health). This retrofitting of existing measures developed in other settings with other patient populations limits the ability to explore alternative conceptualizations of depression in the context of cancer and aging.

Theoretically, the conceptual foundation of the OAC‐D represents a shift in the field's understanding of depression in OACs. It is important to note that this conceptual shift may have application outside of cancer and, in the future, may be explored in older adults with chronic health conditions or in healthy older adults. The results may also have future implications for defining DSM depression criteria for older adults, potentially arguing for new criteria to be included in the DSM, such as the inclusion of loneliness or loss of meaning to sit alongside depressed mood as a gateway diagnostic criterion, similar to how irritability is used in pediatrics as a DSM gateway symptom for depression. ⁴² Clinically, in providing a rigorously developed instrument, we hope to improve the detection of depression to facilitate a timely, evidence‐based intervention for depression in this medically vulnerable population. This PRO can be integrated into existing comprehensive geriatric assessments and depression screening programs nationwide without significantly increasing patient or clinician burden because of its targeted and relatively brief structure. For research purposes, the use of the rigorous FDA PRO Guidance may allow the scale to be used as a primary outcome in FDA‐approved clinical trials.

There are important strengths of this work, including extensive literature reviews, substantial patient and expert input, an approach open to non‐DSM criteria, and a resulting PRO with sound psychometric properties. However, there are also weaknesses. This psychometric analysis included a homogenous patient population with respect to sociodemographic (i.e., race, ethnicity, education) and linguistic diversity. In addition, the development of PROs should be a dynamic process that continues to advance aspects of the measure. Therefore, the next steps in this program of research should include expanding the psychometric validation to a larger, more representative sample, determining the optimal cutoff score to maximize sensitivity and specificity, and defining the scale's minimal clinically important difference. There is also the need to increase the potential reach and impact of the proposed measure by translating the OAC‐D into other languages. These future directions would assist in integrating the OAC‐D into clinical and research settings. In addition, because the purpose and meaning domain as well as the loneliness domain demonstrated lower replication stability, further work should focus on clarifying this stability in other samples and modifying these domains as needed.

CONCLUSION

This report presents the development and preliminary validation of the OAC‐D, a novel PRO designed to address the unique expression of depression in OACs. Grounded in a comprehensive conceptual model informed by systematic literature reviews, patient perspectives, and expert input, the OAC‐D offers a meaningful alternative to traditional DSM‐based measures by incorporating domains that reflect the nuanced experiences of this population, such as loneliness and loss of purpose. Our findings highlight the limitations of relying solely on DSM criteria in this context and underscore the importance of a targeted approach to depression screening that captures the distinctive psychological impact of cancer and aging.

AUTHOR CONTRIBUTIONS

Christian J. Nelson: Conceptualization, methodology, data curation, investigation, validation, supervision, funding acquisition, project administration, resources, writing–original draft, and writing–review and editing. Kathleen Flaherty: Data curation, writing–original draft, and writing–review and editing. Elizabeth Schofield: Methodology, data curation, formal analysis, and writing–review and editing. Thomas M. Atkinson: Methodology, data curation, writing–original draft, writing–review and editing, and supervision. Hayley Pessin: Methodology, data curation, writing–original draft, and writing–review and editing. Addison Kitrel: Writing–original draft and writing–review and editing. Barry Rosenfeld: Conceptualization, methodology, funding acquisition, and writing–review and editing. Rebecca M. Saracino: Conceptualization, methodology, data curation, supervision, project administration, writing–original draft, and writing–review and editing.

CONFLICT OF INTEREST STATEMENT

Hayley Pessin reports an ownership interest in PsyOnc Partners outside the submitted work. The remaining authors disclosed no conflicts of interest.

Supporting information

Supplementary Material

Nelson CJ, Flaherty K, Schofield E, et al. Advancing depression assessment in older adults with cancer: development and validation of the Older Adults with Cancer‐Depression Scale (OAC‐D), a novel, patient‐reported outcome. Cancer. 2026;e70255. doi: 10.1002/cncr.70255

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.