Dec 29, 2025
Woods, Roberson, Booker, Wood, Booker (2024) Longitudinal Associations of Family Relationship Quality with Chronic Pain Incidence & Persistence Among Aging African Americans
- Sarah.Woods 1
- 1UT Southwestern Medical Center
Protocol Citation: Sarah.Woods 2025. Woods, Roberson, Booker, Wood, Booker (2024) Longitudinal Associations of Family Relationship Quality with Chronic Pain Incidence & Persistence Among Aging African Americans. protocols.io https://dx.doi.org/10.17504/protocols.io.ewov1kydkgr2/v1
Manuscript citation:
Woods, S.B., Roberson, P.N.E., Booker, Q., Wood, B., & Booker, S. (2024). Longitudinal associations of family relationship quality with chronic pain incidence & persistence among aging African Americans. Journals of Gerontology, Series B: Psychological and Social Sciences, 79(7), gbae064. doi: 10.1093/geronb/gbae064
License: This is an open access protocol distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited
Protocol status: Other
Protocol functional and completed.
Created: December 25, 2025
Last Modified: December 29, 2025
Protocol Integer ID: 235793
Keywords: aging, Black or African American, chronic pain, family, minority and diverse populations, longitudinal modeling, personal relationships, pain incidence, pain persistence, relationship quality, resilience factors for chronic pain outcome, persistence among aging african american, chronic pain outcome, better pain outcomes over time, aging african american, chronic pain, longitudinal associations of family relationship quality, risk factor for worse pain outcome, better pain outcome, positive family emotional climate, negative family emotional climate, family relationship quality, worse pain outcome, aging health, specific family risk, positive family, african american, negative family, resilience factor, longitudinal association, midlife in the united state, health, emotional climate, supported research, greater strain, less support, greater support, midlife
Funders Acknowledgements:
National Institute on Aging
Grant ID: R21AG082344
Abstract
This study protocol outlines the steps needed to execute, in part, Aim 1 of NIH HEAL Initiative-supported research to identify specific family risk and resilience factors for chronic pain outcomes for aging African Americans (R21AG082344). We hypothesized that a positive family emotional climate (greater support, less strain) promotes better pain outcomes over time. Conversely, we hypothesized a negative family emotional climate (greater strain, less support) is a risk factor for worse pain outcomes over time. These analyses utilize publicly available secondary data from two nationally representative studies of aging health: Midlife in the United States and the Health and Retirement Study.
Attachments
Troubleshooting
Data Cleaning and Management
Retrieve and clean data from Midlife in the United States (MIDUS) including: MIDUS 2 core, MIDUS 2 Milwaukee, MIDUS 3 core, and MIDUS 3 Milwaukee datasets via https://www.icpsr.umich.edu/sites/icpsr/home with more information at https://midus.wisc.edu/data-access/
Select subsamples identifying as Black and/or African American from MIDUS 2 and MIDUS 2 Milwaukee (baseline)
Recode chronic pain incidence items: B1SA15/BACAS15, C1SA15/CACAS15 to 1 (yes) and 0 (no)
Remove participants whose pain is malignant in nature, resulting in N = 757 for MIDUS 2 and MIDUS 2 Milwaukee participants who identify as Black/African American and complete pain incidence items at baseline
Calculate scale scores for four relationship quality scales to align with HRS scoring such that the family support scale consists of three items (MIDUS includes four items), the intimate partner support measure includes three items (MIDUS includes six items), and the intimate partner strain measure includes four items (MIDUS includes six items). Family strain items are unchanged as MIDUS includes the same four items as HRS. Scale scores are calculated via averaging reverse-coded item responses, as indicated by the original scale scoring instructions.
Calculate family relationship quality indices via averaging scale scores across relationship quality measures of support and, separately, strain, completed by each participant. (Scales not completed by a participant and/or relationships participants do not report are not included for a participant's relationship quality index.) Average scores should range from 1 to 4 for average support and 1 to 4 for average strain.
Recode sex to 1 (female) and 0 (male).
Code for occurrence of comorbidities at baseline to include diabetes or high blood sugar, multiple sclerosis, or other neurological disorders. E.g., in MIDUS 2 core, reports of "yes" to either B1SA11X or B1SA11Y should be coded as 1, versus 0 (denies diagnosis/presence of any of these conditions).
Retrieve and clean data from Health and Retirement Study (HRS) including biennial, public survey, core datasets for: HRS 2006 and HRS 2016 (to align with 10-year timespan of MIDUS) via https://hrs.isr.umich.edu/data-products
Select subsample identifying as Black and/or African American from HRS 2006 (baseline)
Recode chronic pain incidence items: KC104 and PC104 to 1 (yes) and 0 (no)
Remove participants whose pain is malignant in nature, resulting in N = 2,585 for HRS 2006 participants who identify as Black/African American and complete pain incidence item at baseline
Calculate scale scores for six relationship quality scales via averaging reverse-coded item responses, as indicated by the original scale scoring instructions and to align with MIDUS scoring procedures. (e.g., for family strain, items include KLB012D-G at baseline)
Calculate family relationship quality indices via averaging scale scores across relationship quality measures of support and, separately, strain, completed by each participant. (Scales not completed by a participant and/or relationships participants do not report are not included for a participant's relationship quality index.) Average scores should range from 1 to 4 for average support and 1 to 4 for average strain.
Recode sex to 1 (female) and 0 (male).
Code for occurrence of comorbidities at baseline to include diabetes (KC010), multiple sclerosis, or other neurological conditions (KC1081M1M and KC1081M2M responses coded as 163). Reports of "yes" to either item should be coded as 1, versus 0 (denies diagnosis/presence of any of these conditions).
Assess missingness in each dataset as tests of the likelihood of attrition tied to sample characteristics.
All models account for sample characteristics including baseline age, sex, prescription pain medication use, and incidence of pain-related comorbidities to account for possible long-term impacts on neuropathic pain development.
For both datasets, pain incidence is operationalized as individuals who deny chronic pain at baseline and new reports of chronic pain at follow-up, while pain persistence is operationalized as individuals who report chronic pain at baseline and continue to report chronic pain at follow-up.
Baseline Analyses
Calculate relationship quality scale score averages, standard deviations, and internal reliabilities (Cronbach's alphas).
Calculate correlations among independent variables and pain outcomes for each dataset.
All models tested in MIDUS and then replicated and extended in HRS. All of the following models tested using logistic regression via MPlus (version 8.10, Muthén & Muthén, 2017) with Monte Carlo integration.
Independent Effects Analyses
Test independent effect of each individual, observed measure of relationship quality (e.g., family strain, parent-child support, etc.) on later risk of pain incidence and, separately, pain persistence via logistic regressions controlling for sample characteristics. Logistic regression models tested for each dataset, and for pain incidence (or, pain development) and pain persistence, separately. Total number of models equals 20 (equaling the number of observed relationship quality scale scores from each dataset).
Relationship Type Analyses
Test effects of relationship quality within relationship types, for each dataset, on pain incidence and pain persistence, separately via logistic regression models controlling for sample characteristics. Total number of models equals 10 (family and intimate partner models for MIDUS and HRS, and parent-child models for HRS). E.g., logistic regression model with MIDUS participants who are pain-free at baseline including family strain and family support at baseline as independent variables and reports of chronic pain at follow-up (MIDUS 3/MIDUS 3 Milwaukee), controlling for sex, age, prescription pain medication use, and incidence of pain-related comorbidities to test the effects of family strain and family support on pain incidence while accounting for both dimensions of relationship quality. Purpose: to estimate the unique variance in pain incidence or persistence risk explained by positive versus negative relationship quality (while accounting for valence measured in the opposite direction) within each type of relationship.
Average Relationship Strain or Average Relationship Support Analyses
Test effects average relationship strain and, separately, average relationship support on pain incidence and, separately, pain persistence, for each dataset, via logistic regression models controlling for sample characteristics. Total number of models equals eight (pain incidence or pain persistence regressed on either average strain or average support, for MIDUS or HRS samples). Average strain and average support are calculated via the family relationship quality indices procedures described above, within each dataset.
Average Relationship Strain and Average Relationship Support Analyses
Model average strain and average support simultaneously in order to estimate the association of average relationship strain with pain incidence or persistence, while accounting for average relationship support, and controlling for sample characteristics. Total number of models equals four (pain incidence or pain persistence regressed on both average strain and average support, for MIDUS or HRS samples). Average strain and average support are calculated via the family relationship quality indices procedures described above, within each dataset.
Protocol references
Crosswell, A. D., Suresh, M., Puterman, E., Gruenewald, T. L., Lee, J., & Epel, E. S. (2020). Advancing Research on Psychosocial Stress and Aging with the Health and Retirement Study: Looking Back to Launch the Field Forward. Journals of Gerontology. Series B, Psychological Sciences and Social Sciences, 75(5), 970-980.
https://doi.org/10.1093/geronb/gby106
Health and Retirement Study, (2006 HRS Core) public use dataset. (2021). Produced and distributed by the University of Michigan with funding from the National Institute on Aging (NIA U01 AG009740). Ann Arbor, MI. https://hrsdata.isr.umich.edu/data-products/2006-hrs-core
Health and Retirement Study, (2016 HRS Core) public use dataset. (2019). Produced and distributed by the University of Michigan with funding from the National Institute on Aging (NIA U01 AG009740). Ann Arbor, MI.
https://hrsdata.isr.umich.edu/data-products/2016-hrs-core
Ryff, C., Almeida, D., Ayanian, J., Binkley, N., Carr, D. S., Coe, C., Davidson, R., Grzywacz, J., Karlamangla, A., Krueger, R., Lachman, M., Love, G., Mailick, M., Mroczek, D., Radler, B., Seeman, T., Sloan, R., Thomas, D., Weinstein, M., & Williams, D. (2017). Midlife in the United States (MIDUS 3), 2013-2014. Distributed by the Inter-University Consortium for Political and Social Research at University of Michigan with funding from the National Institute on Aging (P01 AG020166; U19 AG051426). Ann Arbor, MI. https://doi.org/10.3886/ICPSR36346.v6
Ryff, C., Almeida, D. M., Ayanian, J., Carr, D. S., Cleary, P. D., Coe, C., Davidson, R., Krueger, R. F., Lachman, M. E., Marks, N. F., Mroczek, D. K., Seeman, T., Seltzer, M. M., Singer, B. H., Sloan, R. P., Tun, P. A., Weinstein, M., & Williams, D. (2017). Midlife in the United States (MIDUS 2), 2004-2006. Distributed by the Inter-University Consortium for Political and Social Research at University of Michigan with funding from the National Institute on Aging (P01 AG020166;
U19 AG051426). Ann Arbor, MI. https://doi.org/10.3886/ICPSR04652.v7
Ryff, C. D., Almeida, D., Ayanian, J., Binkley, N., Carr, D. S., Coe, C., Davidson, R., Grzywacz, J. G., Karlamangla, A., Krueger, R., Lachman, M. E., Love, G., Seltzer, M., Mroczek, D. K., Radler, B., Seeman, T. E., Sloan, R., Thomas, D., Weinstein, M., & Williams, D. (2018). Midlife in the United States (MIDUS 3): Milwaukee African American Sample, 2016-2017. Distributed by the Inter-University Consortium for Political and Social Research at University of Michigan with funding from the National Institute on Aging (P01 AG020166; U19AG051426). Ann Arbor, MI. https://doi.org/10.3886/ICPSR37120.v2
Ryff, C. D., Almeida, D., Ayanian, J., Carr, D., Cleary, P. D., Coe, C., Davidson, R., Kruger, R., Lachman, M., Marks, N., Mroczek, D., Seeman, T., Seltzer, M., Singer, B., Sloan, R., Tun, P., Winstein, M., & Williams, D. (2018). Midlife in the
United States (MIDUS 2): Milwaukee African American Sample, 2005-2006. Distributed by the Inter-University Consortium for Political and Social Research at University of Michigan with funding from the National Institute on Aging (P01AG020166; U19 AG051426). Ann Arbor, MI. https://doi.org/10.3886/ICPSR22840.v5
Acknowledgements
This work was supported by the National Institute on Aging of the National Institutes of Health (R21AG082344). The
content is solely the responsibility of the authors and does not necessarily represent the official views of the National
Institutes of Health.
Data used for this study are publicly available and distributed via the University of Michigan, including the Interuniversity Consortium for Political and Social Research. Metadata for this study are available via the NIH HEAL Data Platform. There is no additional public preregistration affiliated with this study.