Underreporting of Gender and Race/Ethnicity Differences in NIH-Funded Dementia Caregiver Support Interventions

Abstract

Women and racial/ethnic minority dementia caregivers have unique caregiving experiences and support needs. To ensure the identification of potentially important differences in outcomes within these groups, the amended National Institutes of Health (NIH) Policy on Inclusion of Women and Minorities mandates reporting by gender and race/ethnicity. The objective of this study was to determine the inclusion and reporting rates among NIH-funded dementia caregiver support interventions. A focused systematic literature review of studies published from 1994 to 2015 located 48 articles meeting inclusion criteria. The majority of studies included women and racial/ethnic minorities; however, 67% did not report results by gender or racial/ethnic group. Acknowledgment of underreporting was more common for race/ethnicity than gender. Our findings suggest limited NIH guideline compliance that may reflect a lack of awareness regarding potential gender disparities in caregiving roles. Ensuring NIH guideline compliance necessitates shared investments from researchers, editors, and reviewers to ensure group differences are systematically identified and reported.

Introduction

Global concern is growing regarding the high levels of stress, burden, depression, and poor physical health outcomes experienced by informal caregivers of people with Alzheimer’s disease and related dementias, resulting in an urgent need to better support dementia caregivers. ¹ Research that addresses dementia caregiver disparities and specific gender and/or racial/ethnic needs and supports is needed. ² Both women and racial/ethnic minorities are disproportionately affected by Alzheimer’s disease and related dementias by virtue of increased disease risk, hastened rates of decline, disproportionate use of services, and poorer clinical outcomes. ^3,4

Approximately two-thirds of informal dementia caregivers are women. ^5,6 Women dementia caregivers tend to provide more intense care for longer periods of time and are likely to experience higher levels of stress, anxiety, and depression than their male counterparts. ⁷ Women are also more likely than men to be dementia caregivers across all racial/ethnic groups with the exception of Asian American dementia caregivers who are equally likely to be male as female. ² There are also well-documented differences in social expectations between genders, such as the expectation the females adopt the role of caregiver, ⁸ that may affect the experience of dementia caregiving. Some of these differences may arise from traditional cultural views that women’s work should be concentrated in the home and that it is natural for women to adopt family caregiving roles due to nurturing abilities that are thought to be inherent feminine qualities. ^9,10 Women are also reported to provide more intense care for longer periods of time and are more likely to support care for basic physical needs than male caregivers. ^8,11 Past research has found that women find caregiving to be more physically and emotionally stressful than men, reporting higher levels of depressive and anxious symptoms than male caregivers. ^11,12 These differences could be due in part to different coping processes engaged in light of the caregiver’s gender identity and perception of their caregiving role. ^8,13 Gender differences in dementia caregiving experiences may also partially reflect underlying differences in vulnerability and risk of depressive disorders or underlying risk factors such as adverse childhood events, both of which disproportionately affect women. ^14,15

Attention to racial/ethnic differences in dementia caregiver support interventions has been limited. ^{16 –18} Evidence suggests that racial/ethnic minority dementia caregivers have different support needs and poorer self-rated health and that some groups have higher levels of distress and depression than nonminority dementia caregivers. ¹¹ Although previous research has consistently identified racial/ethnic differences in caregiving experiences, these have not necessarily translated to expected findings on traditional psychological outcome measures. ^17,19 For example, racial/ethnic minority caregivers report more adverse physical health outcomes, have higher caregiving load in emotional support and instrumental activities of daily living, and have greater resource constraints; however, African American caregivers do not consistently report higher levels of common metrics such as stress, burden, and depression. ^{17,19 –23} Conversely, Asian American and Hispanic caregivers report higher levels of depression than Caucasian caregivers and engage less formal support services than Caucasians. ^{4,24 –26} Additionally, previous research suggests that self-assessment of key outcomes in dementia caregiver support interventions such as depression and burden differ across racial/ethnic minority groups. These differences may be due to unique ethnocultural explanations for these experiences and related values such as faith, different views of the aging process, and a sense of obligation toward immediate and extended family members. ^{19,26 –28}

Racial/ethnic status has been described as providing a “context” for caregiving experiences, which can produce important differences at the interpersonal, temporal, personal, situational, and sociocultural levels of caregiving—ultimately influencing coping processes engaged by caregivers. ²⁹ Previous studies suggest that some racial/ethnic groups may have lower levels of social support ³⁰ and that different racial/ethnic groups may have unique ethnocultural perspectives about dementia and family caregivers’ roles. ²⁷ Much of this literature has addressed views about familism and how it influences caregivers’ appraisal of their role, coping style, and health outcomes. ^31,32 These various cultural perspectives may influence how certain intervention components are perceived, integrated, and applied by the caregiver—thus influencing outcomes such as burden or depression.

Some studies have found that racial/ethnic minority participants may not benefit to the same extent as nonminority dementia caregivers for all intervention components. ^23,30,33 These findings provide support for the notion that there are meaningful differences and mechanisms for role appraisal and uptake of interventions across racial/ethnic groups and that further information about disparities in intervention outcomes is needed to develop culturally appropriate intervention components.

In 2001, the National Institutes of Health (NIH) updated the 1994 policy on the inclusion of women and minorities as clinical research participants. The amended policy provides additional guidance regarding the expectation that “all NIH-funded clinical research will be carried out in a manner sufficient to elicit information about individuals of both sexes/genders and diverse racial and ethnic groups.” ³⁴ The guidelines further require separate planning, conduct, and reporting of analyses for these groups in phase III clinical trials and the broader category of clinical research. The NIH defines clinical research as (1) patient-oriented research involving contact with human participants, (2) epidemiologic and behavioral studies, and (3) outcomes research and health services research. Patient-oriented research is further identified as including studies on (a) mechanisms of human diseases, (b) therapeutic interventions, (c) clinical trials, and (d) development of new technologies. ³⁴ As dissemination of research findings in peer-reviewed journals is a standard expectation following completion of NIH-funded research, the guidelines also specifically address the importance of including results of sex/gender and race/ethnicity and other subpopulations in all publications.

Despite these guidelines, there is some evidence to suggest that reporting omissions in caregiver support interventions using cultural tailoring is common. ¹⁷ Responses to caregiver support interventions by women and racial/ethnic minorities may be further influenced by widespread disparities in health, which are influenced through gender and racial/ethnic differences in environment, socioeconomic status, behaviors, and perceptions. ^{35 –37} Limited reporting of intervention results by racial/ethnic subgroups may ultimately further exacerbate disparities across groups by limiting our ability to account for any differences in treatment outcomes. Addressing potential disparities in treatment outcomes related to dementia caregiver support interventions is reliant on recognition of the potentially significant role these differences may have in interventions. Some research suggests that dementia caregiver interventions do not consistently report treatment outcomes by caregiver ethnicity. ¹⁷ To intervene appropriately, we must acknowledge and examine the potential impact that these factors have on treatment outcomes.

Advances in research on dementia caregiver interventions warrant an exploration of the extent to which gender and/or racial/ethnic differences are currently addressed within the field across NIH-funded dementia caregiver intervention research. We conducted a focused systematic review of published reports of NIH-funded dementia caregiver support interventions to determine (1) the level of reporting of study results by gender and/or racial/ethnic group and (2) the proportion of studies that conducted separate analyses by gender and/or racial/ethnic group.

Methods

Search Strategy

Search and analyses were guided by Preferred Reporting Items for Systematic and Meta-Analyses (PRISMA) guidelines for conducting systematic reviews. ³⁸ We conducted a focused systematic review of published reports of the PubMed and PsychInfo databases using a combination of the key words (“intervention” or “support”) and MeSH headings (“caregivers” and “dementia”). The PubMed database was searched for the dates of 1994 through 2015; and the PsychInfo database was searched for the dates of 2008 through 2015 only as the NIH Public Access Policy, which became effective for manuscripts accepted during or after April 2008, mandated that all final peer-reviewed manuscripts be posted in PubMed. ³⁹

Inclusion Criteria

Decisions regarding study inclusion were made using the following inclusion criteria: (1) studies evaluating dementia caregiver support interventions, (2) English language, (3) NIH-funded, and (4) published in peer-reviewed journals after 1994 (ie, after the NIH initial guidelines on Inclusion of Women and Minority in Clinical Research became effective; Figure 1).

Figure 1.

Dementia caregiver support intervention Preferred Reporting Items for Systematic and Meta-Analyses (PRISMA) flow chart of study selection.

Study Selection

Search result titles and abstracts were reviewed separately by 2 reviewers (L.W. and R.J.). Studies were excluded if they were not NIH-funded, no dementia caregiver support intervention/program was delivered, outcome measures were not specific to dementia caregivers or were not collected at the individual level, the sample included caregivers for conditions other than dementia, the study included paid/formal caregivers, or the study specifically recruited one gender or racial/ethnic group. Dementia caregiver support interventions were defined as interventions that aimed to address some element of support for dementia caregivers and did not include pharmacologic interventions. An informal dementia caregiver was defined as an unpaid individual who did not work or provide care to the person with dementia in a professional capacity; the individual is often but not always a family member of the person with dementia. Forty-six studies met these criteria and were included in the review for both the gender review and the race/ethnicity review, respectively.

Data Extraction

Studies were evaluated to identify the characteristics of intervention studies, characteristics of reporting on primary study outcomes, and approaches used to address potential differences between genders and racial/ethnic groups. Specifically, we identified (1) the rates of inclusion of participants from relevant subgroups across studies and the specific dementia caregiver outcomes measured; (2) whether studies reported gender-specific and/or racial/ethnic group–specific results for the primary outcome; (3) whether they reported statistically significant results by group membership or identified lack of subgroup reporting as a limitation; and (4) approaches described by authors for addressing or mitigating potential differences by gender, racial/ethnic group including analytic approaches such as using group membership as a covariate or testing the effect of group membership in a statistical model. Each study was reviewed separately by 2 reviewers in its entirety. Information was captured using a standardized data collection form. Reporting and analysis by racial/ethnic group membership were considered present if any minority or nonwhite group was identified separately as specific categories/groups identified varied across studies.

Results

Characteristics of Included Studies

The search identified 48 eligible studies, 44 of which were included in both the gender and race/ethnicity review. Two additional distinct articles were relevant to only the gender or race/ethnicity review, which focused on one gender or population for a total of 46 studies included in each review (Figure 1; Online Appendix Table 1). Of the 46 studies included in the gender review, 2 did not specify the proportion of participants who were male or female, and the mean percentage of female participants across the remaining studies was 74.6%. Of the 46 studies included in the race/ethnicity review, 10 did not report the proportion of participants from specific racial/ethnic groups. Of the remaining 36 studies, 7 studies only reported participants from a Caucasian background and an “other” background. The mean percentage of Caucasian participants was 73.5%, 11.3% African American, 12.3% Hispanic, 0.28% Asian American, 0.08% Native American, 0.03% Hawaiian/Pacific Islander, and 2.6% other.

Studies included addressed a range of caregiver outcomes. Common outcome measures were depression (N = 36), burden (N = 17), and stress or strain (N = 13). Additional outcome measures included dementia caregiver ratings of perceptions/beliefs (N = 8), social support (N = 8), well-being/quality of life (N = 6), and dementia caregiver reaction to behaviors in the person with dementia (N = 7).

Reporting of Intervention Results by Gender and Racial/Ethnic Group Membership

Thirty-one of the 46 studies included in the gender review (67%) did not report results by gender. Of these 31 studies, only 2 provided an explanation for this omission in reporting or cited it as a limitation. Of the remaining 15 studies that did report intervention results by gender, 8 (53.3%) found statistically significant differences in results by gender.

Of the 46 studies included in the race/ethnicity review, 31 (67%) did not report results by racial/ethnic group membership. Of these 31 studies, 16 (52%) identified this omission as a limitation. Of the 15 studies reviewed that reported results by racial/ethnic group membership, 12 (80%) reported statistically significant differences by group membership.

There was a substantial overlap in studies that did not report outcomes by gender or racial/ethnic group and moderate overlap in studies that reported outcomes by both groups. Of the 48 eligible studies, 44(96%) were included in the review for both gender and racial/ethnic group. Of these 44 studies, 26 (59%) did not report results for either subgroup, whereas 9 (20%) studies included results for both groups. Five studies (11%) reported outcomes by gender subgroup alone, whereas 4 (9%) reported differences by racial/ethnic subgroup only.

Approaches Engaged to Address Gender, Racial/Ethnic Differences

Among the 31 studies that did not report their outcome by group membership, the most common approach reported for addressing potential differences was stating that intervention and control groups were equal in specific demographic characteristics on baseline, with more studies specifying this for gender (55%) than race/ethnicity (26%; Table 1). For studies in both the gender and the race/ethnicity review, describing equivalence in general terms, that is, stating participant demographics between the control and intervention group are equivalent, without specifically referring to gender and/or race/ethnicity, was the second most frequent approach identified. It was also common for studies to not provide any description of how differences in gender or racial/ethnic groups were explored or addressed (16% of studies in the gender review and 48% of studies in the race/ethnicity review).

Table 1.

Approaches to Addressing Gender and Racial/Ethnic Group Differences in NIH-Funded Dementia Caregiver Support Intervention Studies Reporting and Not Reporting Results by Group Membership.

Analytic Approaches Described to Address Gender and/or Racial/Ethnic Group Differences among Studies Not Reporting Results by Group Membershipa	Gender Review (N = 31)	Race/Ethnicity Review (N = 31)
Reported randomized intervention and control groups to be equivalent in general demographic characteristics	7 (22.6)	6 (19.4)
Reported randomized intervention and control groups to be equivalent specifically in gender and/or race/ethnicity	17 (54.8)	8 (25.8)
Controlled for gender and/or race/ethnicity in analysisb	2 (6.5)	2 (6.5)
Did not describe any strategy or approach to address group differences	5 (16.1)	15 (48.4)
Total	31 (100)	31 (100)
Analytic Approaches Described to Address Gender and/or Racial/Ethnic Group Differences among Studies that Did Report Results by Group Membershipc	Gender Review (N = 15)	Race/Ethnicity Review (N = 15)
Included gender and/or race/ethnicity as a variable in statistical modeld	8 (53.3)	11 (73.3)
Included gender and/or race/ethnicity in correlational analysis with primary outcome(s)	1 (6.7)	1 (6.7)
Included gender and/or race/ethnicity as a covariate, that is, controlled for/adjusted for, in statistical model	6 (40.0)	3 (20.0)
Total	15 (100)	15 (100)

^aOf the 31 studies not reporting results by gender, 28 randomized participants into the intervention and control groups. Of the 31 studies not reporting results by race/ethnicity, 27 randomized participants into groups.

^bThese studies performed hierarchical linear modeling with individual characteristics as dependent variables or mediation analyses, controlling for individual characteristics.

^cAll studies reporting results by group membership performed randomization.

^dStatistical models included regression models, multivariate regression, ANOVA, and growth curve models.

Nearly all the studies that reported primary outcome by group membership either reported testing the relationship between group membership and the primary outcome or controlling for group membership as a covariate. One study that reported outcomes by group membership only reported examining correlations between group membership and the primary outcome (Table 1). Of the 8 studies that found statistically significant differences in results by gender, 5 (63%) included gender as a variable in a regression model. Of the 12 studies that found statistically significant differences in results by race/ethnicity, 9 (75%) include race/ethnicity in their regression model.

Discussion

Our results revealed that underreporting of treatment outcomes by both gender and racial/ethnic group membership among NIH-funded dementia caregiver support interventions is common and inconsistent with guidelines of the NIH Revitalization Act. Acknowledgment of underreporting as a limitation was more common for racial/ethnic group than gender. This may reflect a lack of awareness regarding gender disparities in dementia caregiving roles.

The current review focused solely on reports from studies that did not target a specific gender or racial/ethnic group. Findings from studies that have targeted ethnically diverse groups of dementia caregivers have demonstrated that tailoring of effective support approaches to caregivers’ unique sociocultural context can lead to improvements in quality of life and coping. Substantial evidence suggests that tailoring of dementia caregiver interventions to specific needs and contexts can positively affect treatment outcomes. ^{40 –42} Tailored caregiver interventions have found that, based upon intervention components, different groups may benefit in different ways. For example, the Environmental Skill-Building Program associated with the REACH Initiative found that women and spouses benefited more from the intervention than other participants, experiencing less need for help from others, reduced strain from dementia-related behaviors, and improved affect and well-being. ³³ Other studies from the REACH Initiative have found differential responses to certain intervention components by different racial/ethnic groups. For example, African Americans demonstrated lower rates of improvement across a range of outcomes including satisfaction with social support, depression, self-care, and problem behaviors. ^{23, 30}

Underreporting of treatment outcomes by both gender and racial/ethnic group membership undermines progress in advancing tailored dementia caregiver interventions to specific subgroup needs. In order to adequately address the needs of informal dementia caregivers from various genders and diverse racial/ethnic backgrounds, more consistent identification of cultural differences for different interventional approaches is needed so that these can be proactively addressed in the development and refinement of future interventions.

Our findings regarding approaches described in intervention studies for examining gender and/or race/ethnicity differences shed light on potential reasons investigators may not report findings for individual subgroups. About half of the studies reviewed reported testing control and intervention groups for baseline differences in gender and around a third did the same for race. It is possible that investigators may perceive the commonly small numbers of participants from minority groups and group equivalence at baseline to merit subgroup reporting unnecessary, redundant, or inappropriate. This could be reflective of common misconceptions that baseline statistical equivalence either demonstrates successful randomization or identifies actual or meaningful differences between groups that need to be accounted for statistically (or the absence of differences that can effectively be ignored). However, this approach has important limitations in that not all important differences are identified through statistical significance, ⁴³ and conversely, a significant difference does not always identify what contributes to the significance. In addition, it is reasonable to argue that the conceptual relevance of key variables to intervention outcomes, rather than statistical significance of those variables, should inform analytic decisions. Further, relying solely on statistical equivalence at baseline to decide which variables may be meaningful has important implications for the advancement of science—namely, that conclusions regarding the effectiveness of caregiver interventions might be different if these variables were handled differently. This notion is reflected in findings from studies that reported outcomes by group membership—most of which tested the effect of group membership on the primary outcome—wherein a statistically significant difference was found for 53% of studies in the gender review and 80% of studies in the race/ethnicity review. When these variables are dismissed due to baseline equivalence, these differences are not discovered.

Over 2 decades after the initial NIH Revitalization Act was passed, NIH-funded dementia caregiver support interventions fail to comply with NIH guidelines. Intervention outcomes for gender and/or racial/ethnic minorities are underreported, and this omission is underacknowledged. Responsibility for compliance with NIH guidelines begins with individual investigators who should familiarize themselves with NIH inclusion and reporting expectations. Additional research and discussion in the field regarding the relevance of both gender and racial/ethnic group membership to the experience of dementia caregiving may also help to increase awareness regarding the potential importance of these group differences. Additionally, even in cases where prior studies have not shown significant group differences in outcomes, inclusion and analysis of sex/gender and/or racial/ethnic subgroups are still “strongly encouraged” by NIH in all publications. ³⁴

Publication of NIH-funded studies is under the purview of journal editors and peer reviewers, the gatekeepers of the science. Editors and peer reviewers are uniquely poised to improve compliance with NIH inclusion and reporting guidelines. Potential steps forward include required compliance with the NIH Revitalization Act as part of journal submission (i.e., similar to COREQ, PRISMA, and other research reporting requirements) and journal editors highlighting the need for NIH Revitalization Act compliance in their instructions to peer reviewers. Furthermore, provided existing evidence regarding documented disparities in caregiver’s health outcomes and distinct support needs, subgroup reporting may be considered an ethical imperative regardless of funding or regulatory requirements. Required reporting by journal editors and the subsequent implementation of this compliance by peer review will serve to highlight the critical importance of disparities in dementia support intervention research and is likely to increase compliance with the Revitalization Act. It is only through such efforts that we can advance the science of dementia caregiver interventions to address potential gender and/or race/ethnic disparities.

Limitations

A limitation of the present study is that studies were identified via 2 databases. Although it is likely that a majority of NIH-funded dementia caregiver support interventions were published in MEDLINE journals, particularly following the passing of the Public Access Policy, some NIH-funded studies may have been excluded. Also, the present study only focused on NIH-funded studies, which may not be reflective of the extent of underreporting in other dementia caregiver support intervention studies.

Conclusion

To our knowledge, this is the first study to evaluate the extent of reporting of gender and racial/ethnic minority treatment outcomes among NIH-funded dementia caregiver support interventions. We found that nonadherence to NIH guidelines is common and not broadly reported as a research limitation. Greater attention to these groups’ experiences and unique service needs is paramount to addressing the disproportionate impact of dementia and dementia caregiving on both women and racial/ethnic minorities. Furthermore, examining reasons for omission in reporting may help inform efforts to improve compliance among researchers.