Elder Abuse and Chronic Pain: Cross-Sectional and
Longitudinal Results from the Preventing Elder Abuse and Neglect Initiative
Raudah M Yunus, MPH,*
†Noran N Hairi, PhD,* Wan Y Choo, PhD,* Maw P Tan, PhD,
‡Farizah Hairi, PhD,* Rajini Sooryanarayana, DrPH,* Norliana Ismail, DrPH,*
Shatanapriya Kandiben, BSc,* Devi Peramalah, BSc,* Zainudin M Ali, MPH,
§Sharifah N Ahmad, MD,
§Inayah A Razak, MD,
§Sajaratulnisah Othman, PhD,
¶Fadzilah HM Mydin, MD,
¶Karuthan Chinna, PhD,* and Awang Bulgiba, PhD*
OBJECTIVES:To examine the cross-sectional and longitu- dinal relationships between elder abuse and neglect (EAN) and chronic pain in rural older Malaysians.
DESIGN:Two-year prospective cohort study.
SETTING: Kuala Pilah, a district in Negeri Sembilan approximately 100 km from the capital city, Kuala Lumpur.
PARTICIPANTS: Community-dwelling older adults aged 60 and older. Using a multistage cluster sampling strategy, 1,927 respondents were recruited and assessed at baseline, of whom 1,189 were re-assessed 2 years later.
MEASURES: EAN was determined using the modified Conflict Tactic Scale, and chronic pain was assessed through self-report using validated questions.
RESULTS: The prevalence of chronic pain was 20.4%.
Cross-sectional results revealed 8 variables significantly associated with chronic pain—age, education, income, comorbidities, self-rated health, depression, gait speed, and EAN. Abused elderly adults were 1.52 times as likely to have chronic pain (odds ratio51.52, 95%
confidence interval (CI)51.03–2.27), although longitudi- nal analyses showed no relationship between EAN and risk of chronic pain (risk ratio51.14, 95% CI50.81–
1.60). This lack of causal link was consistent when
comparing analysis with complete cases with that of imputed data.
CONCLUSION: Our findings indicate no temporal rela- tionship between EAN and chronic pain but indicated cross-sectional associations between the two. This might indicate that, although EAN does not lead to chronic pain, individuals with greater physical limitations are more vulnerable to abuse. Our study also shows the importance of cohort design in determining causal rela- tionships between EAN and potentially linked health out- comes.J Am Geriatr Soc 66:1165–1171, 2018.
Key words: elder abuse and neglect; elder mistreatment;
elderly abuse; chronic pain; longitudinal study
E
lder abuse and neglect (EAN) is considered to be a major public health problem. With rapid demographic transition worldwide, especially in developing countries, EAN is expected to escalate. The World Health Organiza- tion defines EAN as “a single or repeated act, or lack of appropriate action, occurring within any relationship where there is an expectation of trust which causes harm or distress to an older person”.1EAN is generally catego- rized into 5 subtypes: physical, psychological, financial, sexual, and neglect.2Like the other 2 domains of family violence—child abuse and intimate partner violence (IPV)—various adverse health effects of abuse in late life have been docu- mented. A systematic review of the health consequences of EAN ranked premature mortality and depression as the 2 most credible outcomes.3 Other consequences listed were greater healthcare use, anxiety, suicidal ideation, diabetes mellitus, metabolic syndrome, digestive problems, poor
From the *Julius Centre, Department of Social and Preventive Medicine, University of Malaya, Kuala Lumpur;†Department of Population Health and Preventive Medicine, Faculty of Medicine, Universiti Teknologi MARA, Selangor;‡Geriatric Division, Faculty of Medicine, University Malaya Medical Centre, Kuala Lumpur;§Negeri Sembilan State Health Department, Seremban; and the¶Department of Primary Care Medicine, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia.
Address correspondence to Raudah Mohd Yunus, Department of Social and Preventive Medicine, Faculty of Medicine, University of Malaya, Kuala Lumpur, Malaysia. E-mail: [email protected] DOI: 10.1111/jgs.15370
JAGS 66:1165–1171, 2018 VC 2018, Copyright the Authors
Journal compilationVC2018, The American Geriatrics Society 0002-8614/18/$15.00
sleep, incontinence, allergy, and weight problems. Chronic pain was categorized as an outcome with limited evidence, needing more rigorous studies,3although it has long been associated with family violence and history of abuse.4,5
A conceptual framework was proposed on the effects of abuse in late life that encompassed physical, psychologi- cal, social, and behavioral health. The model was partly derived from child abuse literature, which highlighted dif- ferent types of pain as among the outcomes of abuse.6 Based on the existing literature on family violence and the proposed framework, the relationship between elder abuse and chronic pain is theoretically plausible, but it has not undergone sufficient empirical tests.
Chronic pain is a complex health condition believed to be an interplay between the nervous, immune, and endocrine systems.7 Despite its relevance to aging and its effect on quality of life and healthcare cost, evidence of the relationship between EAN and chronic pain is scarce.
To the best of our knowledge, only 1 study has addressed EAN and chronic pain, in which an association was reported between the 2 (odds ratio (OR)51.65, 95% con- fidence interval (CI)51.28–2.15),8 but the study used a cross-sectional design and was restricted to older U.S.
women. This study therefore aimed at investigating the cross-sectional association between EAN and chronic pain and the longitudinal relationship between EAN and new incidence of chronic pain in a cohort of rural older Malaysians.
METHODS Study design
This was a 2-year prospective cohort study. Baseline assessments were conducted from November 2013 to July 2014 and follow-up assessments from December 2015 to March 2016.
Setting
The study was conducted in Kuala Pilah, a rural district in Negeri Sembilan located approximately 100 km from the capital city, Kuala Lumpur.
Study subjects
The sampling frame comprised all community-dwelling older adults aged 60 and older. Inclusion criteria were Malaysian citizenship, ability to communicate independ- ently, and residence in Kuala Pilah for at least the previous 12 months. Individuals who lived in institutions (e.g., nursing home) or had severe hearing or cognitive impair- ment were excluded.
Sampling strategy
A multistage cluster sampling approach was used. Kuala Pilah was first divided into 254 enumeration blocks, from which 156 were randomly selected. Then 16 to 20 house- holds were randomly selected from each enumeration block. Potential participants were identified through
telephone calls and door-to-door visits and invited to join the study. If there was more than 1 eligible respondent in a household, 1 person was randomly selected. Further details on study methodology have been previously published.9
Data collection
Respondents were interviewed face to face and clinically assessed (for physical function) at baseline. A second inter- view to reassess chronic pain was conducted over the tele- phone after 2 years.
Variables and tools
EAN was the exposure variable, and chronic pain was the outcome of interest. EAN was ascertained through face-to- face interviews at baseline using the modified Conflict Tactic Scale, as revised.10Whereas some studies operation- alize EAN as abuse episodes “taking place within the last 12 months,” we included all occurrences since the age of 60, defining EAN as any events that occurred since the respondent turned 60.
Chronic pain was defined as “pain that is persistently felt, every day or almost every day, in any part of the body for longer than 3 months, and the pain experience must have occurred within the last 6 months.” The cut-off period of 3 months was said to be the most commonly used by studies on chronic pain in Asian adults, as the International Association for the Study of Pain proposes.11 Each respondent was asked 2 questions to establish chronic pain and its severity: “Within the last six months, did you experience pain in any part of your body that was continuous every day, or almost every day, for at least three months? (Yes or No). “Does the pain interrupt your daily work or activities? (Likert scale: 15not at all, 25to a little extent, 35to some extent, 45to a great extent, 55extremely disturbing).
These 2 questions had been tested and validated in older Malaysians12and used in a number of epidemiologi- cal studies.13 In analysis, the 5 Likert scale options were merged into 3 categories; 1 and 2 were grouped as mild, 3 was defined as moderate, 4 and 5 were grouped as severe.
Other covariates included age, sex, ethnicity, income group, education level, health-related variables (self-rated health, number of chronic diseases), psychosocial variables (depression, social support), and physical function (gait speed). Ethnicity was categorized as Malay, Chinese, Indian, and others, but during analysis it was dichotomized into Malay and non-Malay because of the small number of non-Malay respondents. Income group was categorized as low (monthly household income <RM1,000), middle (RM1,000–2,499), and high (RM2,500). Education level was classified as low (no formal education), medium (ele- mentary or high school), and high (college or university).
Self-rated health was grouped into poor and good, and comorbidities were established through self-report using a list of common diseases—diabetes mellitus, hypertension, cancer, high cholesterol, coronary heart disease, arthritis, stroke, and congestive heart failure. Depression was assessed using the Geriatric Depression Scale (GDS)14 and
social support using the DUKE Social Support Index (DSSI).15Gait speed was determined using the 2.4-m walk- ing test.
Ethics
Written consent was obtained from all participants. The Medical Ethics Committee of University Malaya Medical Centre approved this study (MEC ref.no: 902.2).
Owing to the sensitive nature of the subject, the ques- tionnaire was administered in a setting that ensured the individual’s privacy. Every respondent was interviewed alone without the presence of family members. If a respondent admitted to having been abused, the inter- viewer would seek his or her permission to refer him or her to the social worker. If he or she refused, the inter- viewer would provide him or her with relevant informa- tion and contact details for seeking assistance. If the respondent agreed, referral to the social worker was made.
Analytical approach
Continuous data were reported as means and standard deviations. Categorical data were reported as frequencies
and percentages. Associations between 2 continuous varia- bles were tested using the Pearson correlation coefficient if normally distributed or Spearman correlation if not nor- mally distributed, and associations between categorical variables were examined using chi-square tests. Compari- son of continuous variables between the 2 groups (abused vs not abused) was performed using the independent- sample t-test.
At follow-up, response rate and dropouts were ana- lyzed. Comparison of sociodemographic characteristics of responders and nonresponders was tabulated. Missing data and potential bias were addressed using multiple imputation using the Markov Chain Monte Carlo method and comparison of complete case analysis and imputed data. To optimize the multiple imputation model, auxil- iary variables were included. Ten datasets were generated, from which pooled estimates were taken. Variables involved in multiple imputation were income, education, self-rated health, comorbidities, gait average, depression, social support (missing values at baseline), and chronic pain (missing values at follow-up).
Cross-sectional associations between EAN (and other covariates) and chronic pain were quantified using gener- alized linear models (GLMs) and longitudinal relationships Table 1. Baseline Characteristics of Study Respondents According to Abuse and Neglect Status (N 51,927)
Variable All Abused Not Abused P-Value
Age, n (%) 0.856
60–69 980 (50.9) 80 (51.3) 897 (50.8)
70–79 769 (39.9) 60 (38.4) 709 (40.1)
80 177 (9.2) 16 (10.3) 161 (9.1)
Sex, n (%) 0.063
Male 755 (39.2) 72 (46.2) 682 (38.6)
Female 1,172 (60.8) 84 (53.8) 1086 (61.4)
Ethnicity, n (%) 0.290
Malay 1,868 (96.9) 149 (95.5) 1,716 (97.1)
Chinese 21 (1.1) 2 (1.3) 19 (1.1)
Indian 30 (1.6) 5 (3.2) 25 (1.4)
Other 8 (0.4) 0 (0) 8 (0.4)
Education level, n (%) 0.262
Low 238 (12.4) 13 (8.4) 225 (12.8)
Medium 1638 (85.3) 139 (89.7) 1,496 (84.9)
High 44 (2.3) 3 (1.9) 41 (2.3)
Household income, n (%) <0.005
Low 1,222 (64.1) 115 (74.7) 1,106 (63.3)
Middle 612 (32.1) 32 (20.8) 578 (33.1)
High 71 (3.7) 7 (4.5) 64 (3.7)
Chronic diseases, n (%)
Diabetes mellitus 531 (27.9) 50 (32.3) 481 (27.5) 0.209
Hypertension 1,034 (53.8) 81 (51.9) 952 (54.0) 0.618
Congestive heart disease 127 (6.6) 13 (8.4) 114 (6.5) 0.354
Stroke 46 (2.4) 6 (3.9) 40 (2.3) 0.214
Arthritis 387 (20.2) 38 (24.5) 349 (19.8) 0.165
Cancer 17 (0.9) 0 (0) 17 (1.0) 0.220
Self-rated health, n (%) 0.159
Poor 535 (27.8) 31 (32.7) 484 (27.4)
Good 1,389 (72.2) 105 (67.3) 1,281 (72.6)
Number of comorbidities, mean6SD 1.4861.34 1.5461.40 1.4761.33 0.15
Geriatric Depression Scale score, mean6SD 3.9663.67 4.8063.79 3.8963.65 <0.008
Duke Social Support Index score, mean6SD 27.463.3 26.864.1 27.463.1 0.023
Gait speed, m/s, mean6SD 6.6462.7 6.4761.9 6.6562.7 0.443
SD5standard deviation.
were assessed using generalized estimating equations (GEEs). Distribution of outcome variable was set as bino- mial and the link function as logit. The working correla- tion matrix in GEE was unstructured to allow for all possible correlations, but other options were tested one by one to see which gave the lowest quasi-likelihood under the independence model criterion value. Parameters of interest—OR for GLM and risk ratio for GEE—were reported with their 95% CIs and p-values, with statistical significance set at 0.05. Analyses were performed using SPSS version 20.0 (IBM Corp., Armonk, NY).
RESULTS
Two thousand four hundred ninety-six older adults were invited to participate in the study, 2,118 (84.9%) of whom agreed and were successfully interviewed. There were no significant differences between responders and nonrespond- ers at baseline. Cognitive screening using the Elderly Cogni- tive Assessment Questionnaire excluded 191 individuals, leaving a sample size of 1,927 (mean age 69.866.9, 60.8%
female), of whom 156 (8.1%) reported having experienced abuse since the age of 60. Table 1 illustrates the basic char- acteristics of the study subjects at baseline.
Of the 1,927 respondents assessed at baseline, 1,189 participated in the 2-year follow-up telephone interview (response rate 61.7%). Of the baseline cohort, 393 were found to have chronic pain and were excluded from longi- tudinal analysis. Cross-sectional analyses included all 1,927 participants at baseline, and longitudinal analysis using imputed data included 1,534 participants.
Participant Characteristics
The prevalence of chronic pain was 20.4%. Of those abused, 30.1% reported having chronic pain, versus 19.6% of those not abused. In terms of severity, 36.2% of those who were abused had mild chronic pain, 19.1% had moderate pain, and 44.7% had severe pain, whereas
44.7% of those not abused had mild chronic pain, 18.7%
had moderate pain, and 36.3% had severe pain.
Cross-sectional analysis
GLMs were used to test the association between EAN (along with other potential confounders) and chronic pain.
Table 2 shows the relationships between variables of inter- est and chronic pain.
Eight variables were significantly associated with chronic pain: age, education, income, comorbidities, self- rated health, depression, gait speed, and EAN. Those who were older, had lower income, reported more chronic dis- eases, and rated their health as poor were more likely to have chronic pain than their counterparts who were younger, wealthier, had fewer comorbidities, and rated their health positively. Similarly, older adults with higher GDS scores (depressed), slower walking speed, and experi- ence of abuse in late life had greater odds of having chronic pain than those who were not depressed, had higher walking speed, and were never abused. Older adults from the lower and medium education background appeared to have lower odds of having chronic pain than those with higher levels of education.
Longitudinal analysis
One thousand one hundred eighty-nine respondents were successfully reached for the telephone interview conducted after 2 years (response rate 61.7%). Nonresponders were older and had lower income and education. Further analysis suggested that the missing data mechanism was missing at random. The diagram below illustrates the flow of the tele- phone interview and reasons for nonresponse. Comparison of the characteristics of responders and nonresponders at follow-up is available as Supplementary Appendix S1.
Analyses of complete cases only were then performed using GEEs. Following that, GEEs were rerun on imputed data to test the robustness of results from complete case Table 2. Cross-Sectional Analyses Between Elder Abuse and Chronic Pain Using Generalized Linear Models (n5 1,927)
Variable
Odds Ratio
(95% Confidence Interval) Standard Error P-Value
Age 1.03 (1.01–1.05) 0.01 <0.001
Male 0.96 (0.75–1.25) 0.13 0.790
Ethnicity Malay 0.73 (0.39–1.40) 0.33 0.363
Education level (reference high)
Low 0.19 (0.08–0.47) 0.44 <0.001
Medium 0.26 (0.12–0.57) 0.40 <0.001
Income (reference high)
Low 4.81 (1.80–12.68) 0.49 <0.001
Medium 2.83 (1.07–7.46) 0.49 0.039
Comorbidity 1.46 (1.33–1.59) 0.05 <0.002
Self-rated health poor 1.40 (1.08–1.80) 0.13 0.006
Geriatric Depression Scale score 1.06 (1.03–1.09) 0.02 <0.002
Duke Social Support Index score 1.01 (0.97–1.05) 0.02 0.554
Gait average 1.04 (1.00–1.09) 0.02 0.050
Abuse 1.52 (1.03–2.27) 0.20 0.031
Deviance: 1,727.51, Pearson chi-square: 1,806.35, Akaike Information Criterion: 1,755.5.
analyses. As mentioned before, before GEEs were run, respondents who had chronic pain at baseline were excluded to ensure that all individuals analyzed were free of the outcome of interest. Collinearity diagnostic tests revealed no multicollinearity between any of the predictor variables. GEE results using imputed data are shown in Table 3, and results from complete case analysis are pro- vided in Supplementary Appendix S2.
Analyses in Table 4 (complete case) and Table 5 (imputed data) demonstrated fairly consistent results. On the average, multiple imputation increased the precision of estimates, which smaller standard errors across variables indicate. Female sex, having more comorbidities, and poor self-rated health contributed to greater risk of chronic pain. In complete-case analysis, social support was shown to have a protective effect against the risk of developing chronic pain, but its p-value crossed the significance cut- off in the second GEE model, with a small margin. Simi- larly, there was a slight variation in p-values of age when comparing the first and second models. EAN did not lead to higher risk of chronic pain, and this was consistent in both GEE models.
A similar GEE model was then constructed, but this time, each subtype replaced the exposure variable, EAN, one by one (except for sexual abuse because of its small number). No relationship was found between EAN sub- types and chronic pain. The results are available as Sup- plementary Appendix S3.
DISCUSSION
We examined cross-sectional and longitudinal associations between EAN and chronic pain in community-dwelling, rural, older Malaysians. We found no longitudinal rela- tionship between EAN and new incidence of chronic pain, even though individuals who were abused were more likely to have chronic pain in cross-sectional assessment.
The prevalence of EAN in this study was 8.1%. This is lower than the prevalence reported in urban Malaysia
(9.6%),16Brazil (14.4%),17and Mexico (32.1%).18A sys- tematic review found the pooled prevalence estimate of EAN to be 15.7%,19 and another reported the global prevalence of EAN as ranging between 3.2% to 27.5%.20 Ultimate comparison between studies should be done with caution because of heterogeneity in methods and opera- tional definitions used.
Initial unadjusted analyses showed associations between EAN and depression, low income, and poor social support—all of which previous findings have corro- borated as credible risk factors for EAN.21 Investigation of a nationally representative Malaysian sample reported that 15.2% of older adults had chronic pain, but it was more common in rural areas.22 Our higher finding—
20.4%—thus may be because of the study (rural) setting.
Different prevalence rates had been cited across countries, such as 31.2% of elderly Norwegians, 42.0% of Taiwan- ese, and 69.2% of Spanish,23–25 although heterogeneity in study methods may explain these variations.
Findings from cross-sectional analyses using GLMs were fairly consistent with prior studies. Older age, low income, depression, poor self-rated health, and comorbid- ities were all associated with chronic pain.26,27 Similarly, the association between EAN and chronic pain echoed that found in a previous study.8A visible contrast was the relationship between education level and chronic pain;
many studies reported greater frequency of chronic pain in individuals with less education,24,28,29 whereas our study revealed the opposite. How formal education affects pain perception, pain tolerance, and pain behavior across cul- tures perhaps has a role in this and needs further investigation.
Longitudinal findings from both datasets (complete case, imputed data) showed consistent results across all variables, including EAN, with the exception of age and social support. Female sex, chronic diseases, and poor self- rated health were found to cause higher risk of developing chronic pain, in line with previous findings,27,30 although the apparent small inconsistencies in age and social Table 3. Generalized Estimating Equation Showing Longitudinal Relationships Between Variables of Interest and Chronic Pain Using Imputed Data (n 51,534)
Variable
Risk Ratio
(95% Confidence Interval) Standard Error P-Value
Age 1.02 (0.99–1.03) 0.008 0.069
Male 0.73 (0.57–0.93) 0.121 0.010
Ethnicity Malay 2.11 (0.84–5.32) 0.463 0.107
Education level (reference high)
Low 1.23 (0.44–3.42) 0.510 0.692
Medium 1.23 (0.48–3.10) 0.464 0.653
Income (reference high)
Low 0.85 (0.48–1.51) 0.288 0.582
Medium 1.21 (0.69–2.14) 0.284 0.477
Comorbidity 1.16 (1.06–1.26) 0.043 <0.005
Self-rated health poor 1.38 (1.08–1.77) 0.124 0.010
Geriatric Depression Scale score 1.02 (0.99–1.06) 0.016 0.131
Duke Social Support Index score 0.96 (0.93–1.00) 0.016 0.069
Gait average 1.04 (0.99–1.10) 0.024 0.087
Abuse 1.14 (0.81–1.60) 0.173 0.459
Quasi-likelihood under the independence model criterion52,632.28, quasi-likelihood under the independence model information criterion (QICC)52,638.67.
support should not indefinitely exclude these 2 variables from the list of potential causes of chronic pain. Existing literature shows that age predicts chronic pain30 and that social support influences pain, pain perception, and coping ability.31,32Our short follow-up period (2 years) may have caused these irregularities; more time is perhaps needed for the effects of age and social support to manifest fully.
Abuse, despite having a significant association with chronic pain in cross-sectional analyses, did not lead to greater risk of developing chronic pain in longitudinal assessments. Cross-sectional associations do not imply cause and effect. Therefore, it could be that abuse in late life plays no role in causing chronic pain; rather it is chronic pain that makes older adults more vulnerable to abuse. This can be because of the physical limitation it causes and greater demand for care associated with it.
Studies have shown that poor physical health, functional dependency, and caregiver burden are risk factors for EAN.21,33 In addition, the cause of chronic pain is unclear.34 Although some studies suggest that stress or adverse life events are antecedents to chronic pain,35,36 others have demonstrated that psychological distress is not a cause but an outcome of chronic pain.7,37,38Our results thus appear to refute the “stress as antecedent” theory and support the latter.
The main strength of this study is that it involves a large, representative sample of rural community-dwelling older adults in Malaysia with longitudinal follow-up. Lon- gitudinal studies are important in investigating changes in health outcomes over time, although as in any longitudinal study involving older populations, high attrition rates are common. Our loss to follow-up (38.3%) comprised mainly those with lower socioeconomic background. Even though we attempted to compensate for potential bias through multiple imputation and sensitivity analysis, such procedures are not free from drawbacks and therefore may not yield full precision. We used 2 methods of data collection, face-to-face interview at baseline and telephone interview at follow-up, which may raise concerns about comparability of results and quality of data.39,40Neverthe- less, studies have shown that telephone interview is
equally reliable as physical administration of question- naires.41,42 In addition, the short follow-up period may have not allowed the effect of some variables on chronic pain to be fully established. Our findings need to be inter- preted in light of these constraints.
CONCLUSION
Our findings indicate no causal relationship between EAN and chronic pain, although given the significant associa- tion between the 2 from cross-sectional assessments, it could be that individuals with greater physical limitations, functional dependency, and higher demand for care—
which may result from chronic pain—are more vulnerable to EAN.43,44
ACKNOWLEDGMENTS
We would like to thank all members of the Julius Centre University of Malaya and the Health State Department of Negeri Sembilan for their administrative support throughout the research period.
Financial Disclosure: This study was funded by Ministry of Higher Education High Impact Research Grant E000010–
20001, the University of Malaya Grand Challenge on Pre- venting Elder Abuse and Neglect Initiative (GC001B-14 HTM) and the Population Studies Unit.
Conflict of Interest: The authors declare no conflict of interest.
Author Contributions: CWY, NNH, RMY: study design, preparation and drafting the manuscript. TMP, AB, RS, SK, NI, DP, SNA, IAR, ZMA, SO, KC, FH, FHM: con- tributions in areas of expertise. RMY, RS, SK, NI, DP, SNA, IAR, FH, ZMA: acquisition of data. RMY, CWY, KC, TMP, AB: data analyses. CWY and NNH were the principal inves- tigators and led the grant application. All authors approved the final manuscript.
Sponsor’s Role: The sponsors had no role in in the design, methods, subject recruitment, data collection, analy- sis, of preparation of the manuscript.
REFERENCES
1. World Health Organization. Elder Abuse (online). Available at http://www.
who.int/ageing/projects/elder_abuse/en/ Accessed 26 October 2017 2. Laumann EO, Leitsch SA, Waite LJ. Elder mistreatment in the United
States: Prevalence estimates from a nationally representative study.
J Gerontol B Psychol Sci Soc Sci 2008;63A:S248–S254.
3. Yunus RM, Hairi NN, Choo WY. Consequences of elder abuse and neglect: A systematic review of observational studies. Trauma Violence Abuse 2017:1524838017692798.
4. Wuest J, Merritt-Gray M, Ford-Gilboe M, Lent B, Varcoe C, Campbell JC.
Chronic pain in women survivors of intimate partner violence. J Pain 2008;9:1049–1057.
5. Springer KW, Sheridan J, Kuo D, Carnes M. Long-term physical and men- tal health consequences of childhood physical abuse: Results from a large population-based sample of men and women. Child Abuse Neglect 2007;
31:517–530.
6. Anetzberger GJ. Elderly adult survivors of family violence: Implications for clinical practice. Violence Against Women 1997;3:499–514.
7. Breen J. Transitions in the concept of chronic pain. Adv Nurs Sci 2002;24:
48–59.
8. Fisher BS, Regan SL. The extent and frequency of abuse in the lives of older women and their relationship with health outcomes. Gerontologist 2006;46:200–209.
Figure 1.Flow chart of telephone-based interview.
9. Choo WY, Hairi NN, Sooryanarayana R, et al. Elder mistreatment in a community dwelling population: The Malaysian Elder Mistreatment Project (MAESTRO) cohort study protocol. BMJ Open 2016;6:e011057.
10. Naughton C, Drennan J, Lyons I, et al. Elder abuse and neglect in Ireland:
Results from a national prevalence survey. Age Ageing 2011;41:98–103.
11. Zaki LRM, Hairi NN. A systematic review of the prevalence and measure- ment of chronic pain in Asian adults. Pain Manag Nurs 2015;16:440–452.
12. Rafidah L, Zaki M. Chronic Pain Among Older People in Malaysia: Preva- lence, Associated Factors and Healthcare Utilization. Kuala Lumpur, Malaysia: University of Malaya; 2016.
13. Ismail N. Pattern and Risk Factors of Functional Limitation and Physical Disability Among Community-Dwelling Elderly in Kuala Pilah, Malaysia:
A 12-Month Follow-Up Study. Kuala Lumpur, Malaysia: University of Malaya; 2016.
14. Nyunt MSZ, Fones C, Niti M, Ng T-P. Criterion-based validity and reli- ability of the Geriatric Depression Screening Scale (GDS-15) in a large vali- dation sample of community-living Asian older adults. Aging Ment Health 2009;13:376–382.
15. Koenig HG, Westlund RE, George LK, Hughes DC, Blazer DG, Hybels C.
Abbreviating the Duke Social Support Index for use in chronically ill elderly individuals. Psychosomatics 1993;34:61–69.
16. Sooryanarayana R, Choo WY, Hairi NN, Chinna K, Bulgiba A. Insight into elder abuse among urban poor of Kuala Lumpur, Malaysia—a middle- income developing country. J Am Geriatr Soc 2015;63:180–182.
17. Blay SL, Laks J, Marinho V, Figueira I, Maia D, Coutinho ESF, et al. Prev- alence and correlates of elder abuse in Sao Paulo and Rio de Janeiro. J Am Geriatr Soc 2017;65:2634–2638.
18. Giraldo-Rodriguez L, Rosas-Carrasco O, Mino-Leon D. Abuse in Mexican older adults with long-term disability: National prevalence and associated factors. J Am Geriatr Soc1 2015;63:1594–600.
19. Yon Y, Mikton CR, Gassoumis ZD, Wilber KH. Elder abuse prevalence in community settings: A systematic review and meta-analysis. Lancet Glob Health 2017;5:e147–e156.
20. Cooper C, Selwood A, Livingston G. The prevalence of elder abuse and neglect: a systematic review. Age Ageing 2008;37:151–160.
21. Johannesen M, LoGiudice D. Elder abuse: A systematic review of risk fac- tors in community-dwelling elders. Age Ageing 2013;42:292–298.
22. Zaki LRM, Hairi NN. Chronic pain and pattern of health care utilization among Malaysian elderly population: National Health and Morbidity Sur- vey III (NHMS III, 2006). Maturitas 2014;79:435–441.
23. Rustøen T, Wahl AK, Hanestad BR, Lerdal A, Paul S, Miaskowski C. Age and the experience of chronic pain: Differences in health and quality of life among younger, middle-aged, and older adults. Clin J Pain 2005;21:513–523.
24. Yu H-Y, Tang F-I, Kuo BI-T, Yu S. Prevalence, interference, and risk fac- tors for chronic pain among Taiwanese community older people. Pain Manag Nurs 2006;7:2–11.
25. Miro J, Paredes S, Rull M, et al. Pain in older adults: A prevalence study in the Mediterranean region of Catalonia. Eur J Pain 2007;11:83–92.
26. Abu-Saad Huijer H. Chronic pain: A review. J Med Liban 2010;58:21–27.
27. Arnow BA, Hunkeler EM, Blasey CM, et al. Comorbid depression, chronic pain, and disability in primary care. Psychosom Med 2006;68:262–268.
28. Eriksen J, Jensen MK, Sjøgren P, Ekholm O, Rasmussen NK. Epidemiology of chronic non-malignant pain in Denmark. Pain 2003;106:221–228.
29. Blyth FM, March LM, Brnabic AJ, Jorm LR, Williamson M, Cousins MJ.
Chronic pain in Australia: A prevalence study. Pain 2001;89:127–134.
30. Elliott AM, Smith BH, Penny KI, Smith WC, Chambers WA. The epidemi- ology of chronic pain in the community. Lancet 1999;354:1248–1252.
31. Zhou X, Gao D-G. Social support and money as pain management mecha- nisms. Psychol Inq 2008;19:127–144.
32. Evers AW, Bijlsma JW. Pain coping and social support as predictors of long-term functional disability and pain in early rheumatoid arthritis.
Behav Res Ther 2003;41:1295–1310.
33. Lachs MS, Williams C, O’Brien S, Hurst L, Horwitz R. Risk factors for reported elder abuse and neglect: A nine-year observational cohort study.
Gerontologist 1997;37:469–474.
34. Peplinskie K. Chronic Pain as the Embodiment of Hardship: A Narrative Study. Toronto, Canada: University of Toronto; 2016.
35. Van Houdenhove B, Egle UT. Fibromyalgia: A stress disorder? Psychother Psychosom 2004;73:267–275.
36. Otis JD, Keane TM, Kerns RD. An examination of the relationship between chronic pain and post-traumatic stress disorder. J Rehab Res Dev 2003;40:397.
37. Gamsa A. Is emotional disturbance a precipitator or a consequence of chronic pain? Pain 1990;42:183–195.
38. Fishbain DA, Cutler R, Rosomoff HL, Rosomoff RS. Chronic pain- associated depression: antecedent or consequence of chronic pain? A review. Clin J Pain 1997;13:116–137.
39. Holbrook AL GM, Krosnick JA. Telephone versus face-to-face interviewing of national probability samples with long questionnaires: Comparisons of respondent satisficing and social desirability response bias. Public Opin Q 2003;67:79–125.
40. Carr EC, Worth A. The use of the telephone interview for research. NT Res 2001;6:511–524.
41. Rohde P, Lewinsohn PM, Seeley JR. Comparability of telephone and face- to-face interviews in assessing axis I and II disorders. Am J Psychiatry 1997;154:1593–1598.
42. Aziz MA, Kenford S. Comparability of telephone and face-to-face inter- views in assessing patients with posttraumatic stress disorder. J Psychiatr Pract 2004;10:307–313.
43. Yan E, Chan KL, Tiwari A. A systematic review of prevalence and risk factors for elder abuse in Asia. Trauma Violence Abuse 2015;16:
199–219.
44. Pillemer K, Burnes D, Riffin C, Lachs MS. Elder abuse: Global situation, risk factors, and prevention strategies. Gerontologist 2016;56 Suppl 2:
S194–S205.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article.
Appendix S1: Comparison between the characteris- tics of responders and non-responders at T1 (baseline)
Appendix S2: GEE showing relationships between variables of interest and chronic pain using complete cases only
Appendix S3: GEE showing longitudinal relation- ships between EAN subtypes and chronic pain
Please note: Wiley-Blackwell is not responsible for the content, accuracy, errors, or functionality of any supporting materials supplied by the authors. Any queries (other than missing material) should be directed to the corresponding author for the article.
