Health-Related Quality of Life Among Adults Who Experienced Maltreatment During Childhood

[Headnote]
Objectives. We sought to assess the difference in a preference-based measure of health among adults reporting maltreatment as a child versus those reporting no maltreatment.
Methods. Using data from a study of adults who reported adverse childhood experiences and current health status, we matched adults who reported childhood maltreatment (n = 2812) to those who reported no childhood maltreatment (n=3356). Propensity score methods were used to compare the 2 groups. Health-related quality-of-life data (or "utilities") were imputed from the Medical Outcomes Study 36-Item Short Form Health Survey using the Short Form-6D preference-based scoring algorithm.
Results. The combined strata-level effects of maltreatment on Short Form-6D utility was a reduction of 0.028 per year (95% confidence interval=0.022, 0.034; P<.001). All utility losses for the childhood-maltreatment versus no-childhoodmaltreatment groups by age group were significantly different: 18-39 years, 0.042; 40-49 years, 0.038; 50-59 years, 0.023; 60-69 years, 0.016; 70 or more years, 0.025.
Conclusions. Persons who experienced childhood maltreatment had significant and sustained losses in health-related quality of life in adulthood relative to persons who did not experience maltreatment. These data are useful for asessing the cost-effectiveness of interventions designed to prevent child maltreatment in terms of cost per quality-adjusted life years saved. (Am J Public Health. 2008;98: 1094-1100. doi:10.2105/AJPH.2007.119826)

There is increasing evidence that exposure to childhood maltreatment can lead to greater susceptibility to lifelong physical and mental heath problems, including cardiovascular disease, hypertension, diabetes, anxiety disorders, depression, substance abuse, and perpetration of future violence.1-7 Childhood maltreatment can be defined as any act or series of acts of commission or omission by a parent or other caregiver, in the context of a relationship of responsibility, trust, or power, that results in harm, potential for harm, or threat of harm to a child's health, survival, development, or dignity.8,9

Childhood maltreatment poses a substantial risk for long-term health for many reasons. First, recurrent exposure to the stress associated with maltreatment can lead to potentially irreversible changes in the interrelated brain circuits and hormonal systems that regulate stress.10-12 Changes in these brain systems can lead to a premature physiological aging of the body that increases vulnerability to disease over the life course.11,12 Second, childhood maltreatment increases the risk of behavioral problems such as smoking, substance abuse, obesity, and sexual promiscuity. 1,13 Third, a related body of evidence indicates that early adverse childhood experiences have a profound effect on a range of cognitive, social, and emotional competencies that lay the foundation for successful learning, coping, and subsequent economic productivity. 13-16

This broad range of childhood maltreatment's impact on health suggests that it may also have an impact on victims' life expectancy and long-term health-related quality of life (HRQoL). When assessed together, these outcomes provide information on the effect that childhood maltreatment has on victims' remaining quality-adjusted life years (QALYs), which is a composite measure of health typically used in economic evaluations of health interventions such as costeffectiveness analyses.17-21

Assessment of the impact of childhood maltreatment on the first of the 2 components of the QALY-life expectancy-is relatively straightforward. It requires good epidemiological data on mortality outcomes associated with the acute and chronic phases of childhood maltreatment. Assessment of the impact of childhood maltreatment on the second component, HRQoL, is more complicated. When following national guidelines for conducting cost-effectiveness analyses,17,22,23 measures of HRQoL should reflect relative desirability of different health outcomes under consideration for the population of interest. Preference-based measures provide a summary value for a respondent's valuation of the quality of life of a particular health state, incorporating all positive and negative aspects of a health state into a single number.

A commonly used approach for valuing preferences in health is "utility." A utility weight is typically scaled between 1, representing perfect health, and 0, representing a health state judged equivalent to being dead. Decrements in HRQoL, as measured by utility weights on this scale, are then multiplied by length of life to estimate the QALYs associated with and without the intervention under consideration. These preferences, or utilities, can be directly elicited from the affected population or can be indirectly derived through the use of well-developed, generally accepted, and widely used generic HRQoL indexes whose valuation is based on general population samples.24-28

For health outcomes resulting from physical abuse, sexual abuse, psychological abuse, neglect, or any combination thereof, few if any studies have either directly or indirectly elicited utilities. The paucity of data, particularly for health states associated with childhood maltreatment, is most likely because of a variety of practical and methodological challenges.29 These include the difficulty in defining an average health state for acute or ongoing violent episodes, the cognitive challenges in eliciting preferences for health outcomes from children, proxy issues concerning parents or caregivers who are often the perpetrators of maltreatment, and other reasons associated with development of the field of childhood maltreatment prevention and priorities for research.30,31

Only a few studies have assessed the longterm impact of childhood maltreatment on HRQoL,32-35 but these have included summary measures of health that are not preference based. One summary measure of health, the Medical Outcomes Study 36-Item Short Form Health Survey (SF-36),36 is a commonly used health-state classification instrument. Edwards et al. compared self-reports of health on the SF-36 in an adult population to an index measure of the number of adverse exposures, including childhood maltreatment, experienced during childhood (the adverse childhood experiences [ACE] score).32 The authors found an inverse relationship between ACE score (on which the more adverse experiences, the higher the score) and the SF-36 overall summary measure. However, the summary measure derived from the SF-36 measures health on a scale from 0 (worst health) to 100 (best health) but does not explicitly incorporate preferences into its scoring algorithm and, therefore, cannot be used to obtain preference weights for constructing the QALY. Alternatively, preferencebased measures of HRQoL reflect relative desirability of a score (or index on a scale) based on tradeoffs that one would make on life expectancy to achieve better HRQoL.23

Fortunately, new methods have been developed that enable one to translate summary measures of HRQoL into preferencebased measures of HRQoL for use in costeffectiveness analyses. This represents an exciting advance in methodology, particularly as it is applied to health outcomes associated with violence that have received such little attention in terms of eliciting preference-based measures of HRQoL. We sought to derive preference-based values for childhood maltreatment outcomes derived from summary measures of health defined by adults self-reporting maltreatment outcomes during childhood. These results, when incorporated with epidemiological data on life expectancy, will provide a means for assessing lifetime losses in QALYs and for conducting cost-effectiveness analyses of interventions designed to prevent childhood maltreatment.

METHODS

Study Population
Data were originally collected as part of the second survey wave of the Adverse Childhood Experiences Study at Kaiser Permanente's Health Appraisal Clinic in San Diego, California, between June and October 1997. Complete descriptions of the study population and several analyses of this large database are available elsewhere.1,32 Basic demographic information was collected from participants, as well as data on adverse events experienced during childhood, current health status as measured by the SF-36, health risk behaviors, and diseases past and present. Table 1 lists the questions used to measure adverse childhood experiences. Five categories of childhood maltreatment were included, with questions adapted from previously developed scales: physical abuse,37 sexual abuse,38 emotional abuse,37 physical neglect,39 and emotional neglect.39 An additional 5 categories of questions were asked regarding other adverse experiences during childhood, including household substance abuse, household mental illness, violent treatment of mother, household member in prison, and parental separation or divorce.
Data Analysis

Our main outcome measure of interest was a preference-based HRQoL measure, or utility, for 2 populations-adults who self-reported childhood maltreatment during the first 18 years of life (cases) and those who did not report maltreatment during childhood (controls).

Health utility measures were calculated using the Brazier algorithm (provided by Brazier) that transforms a summary measure of health into a preference-based measure of health. Brazier et al.40 first reduced a summary measure of health, the SF-36, into a 6-dimensional health state classification system, the Short Form-6D (SF-6D). The SF-6D includes physical functioning, role imitations, social functioning, pain, mental health, and vitality. Then they directly elicited preference- based measures of HRQoL, or utilities, for a variety of health states defined by the SF-6D from 165 health professionals and patients in the United Kingdom. Following positive outcomes from this pilot work, Brazier et al.41 refined the original models by using a representative sample of the general public (n=836). Several models were tested, with the fixed effects and random effects models being the most appropriate, with utility values as the dependent variable and personal characteristics and dummies for each level of the SF-6D as independent variables. Parameters were estimated from these models and then used for the population to estimate utility indices from the SF-6D. Subsequent studies have tested the validity and reliability of the transformation formula, and it is now seen as a promising method for deriving utilities or preference-based measures of health states from summary health data.41,42

Because our study relied on a large observational study with cases (the childhoodmaltreatment group) being assigned to experimental units without the benefits of randomization, systematic differences were likely to exist between individuals in the childhood-maltreatment and no-childhoodmaltreatment groups with respect to confounding covariates such as other adverse childhood experiences and socioeconomic status. Simple comparisons of HRQoL measures between childhood maltreatment and no childhood maltreatment are potentially misleading or biased in that the differences of health utility between the 2 groups could be explained by systematic between-group differences rather than as the effect of maltreatment per se.

Therefore, we use the method of stratification based on the propensity score, a scalar function of the covariates, to approximate a randomized controlled setting and to reduce bias in estimating marginal impacts of childhood maltreatment on predicted utility in an observational study.43,44 The method involved dividing units into 5 age groups and then dividing them into quintiles based on the propensity score within each age group (for a total of 25 strata). Health utility measures of childhood maltreatment and no childhood maltreatment were compared for those who fell into the same strata. An overall effect of childhood maltreatment on utility was estimated by using a weighted average of the within-strata estimates with the weights equal to the proportions of the population within the strata.

To assess the marginal impact of each type of childhood maltreatment on utility, logistic regression models were estimated with imputed health utility as the outcome variable and 5 types of maltreatment as predictors for all 25 strata. Similar to estimating the overall effect of childhood maltreatment on utility, the overall impact of each type of maltreatment on utility were weighted and combined across all 25 strata to determine the overall impact of that type of childhood maltreatment on utility.

To create the propensity score, which was defined as the predicted probability of being maltreated during childhood, we estimated a multiple logistic regression predicting childhood maltreatment by using a number of covariates as explanatory variables. These covariates included basic demographics (gender, age, age squared, race), family economic variables found to be related to childhood maltreatment in previous research (mother's years of education, log of number of residential moves in childhood, whether parent owned own home),45,46 and the other 5 categories of adverse childhood experiences described previously and in Table 1. The rationale for using the other adverse childhood experiences as covariates was to determine the marginal impact of childhood maltreatment on utility. The model, therefore, adjusted for exposure to other adverse childhood experiences as potential confounders.

Significance tests for all key variables were conducted between the childhood-maltreatment and no-childhood-maltreatment groups within each of the 25 strata for both before and after subclassification. We used an analysis of variance (ANOVA) to evaluate differences in prevalence of key variables that were continuous and a 2-sided Pearson ?2 test for variables that were categorical. A P value of less than .05 was considered significant in this analysis.

RESULTS

Of the 8667 respondents in the second survey wave of the Adverse Childhood Experiences Study, 7641 (88%) agreed to complete the SF-36, and 6815 (78.6%) completed all questions. An additional 647 respondents were excluded because they were missing information on childhood maltreatment (n=25) or on covariates used to develop the propensity score (n=622). Of the 6168 respondents who remained, the average age of participants was 55.4 years (SD=14.9), 53% were women, 76% were White, and 45.6% (n=2812) self-reported some form of maltreatment during childhood. Respondents that remained did not differ substantially on demographic characteristics from the original sample. For example, those respondents who remained in the analyses were similar in age (55.4 years vs 55.9 years) and were more likely to be men (by 1.1%) and White (by 2.1%) compared with the original sample. Therefore we feel that the respondents included in this analysis were representative of Kaiser Permanente's population.

Table 2 contains the prevalence of each individual form of childhood maltreatment, as well as the correlation between maltreatment types. Physical abuse had the highest prevalence of any of the abuse types (26%), whereas physical neglect was reported by the fewest participants (9%). Each maltreatment type was modestly to moderately correlated (P<.05), with the highest correlations between emotional abuse and emotional neglect (0.43), although physical abuse and emotional abuse were nearly as highly correlated (0.42).

A number of key variables were significantly different between the maltreated and nonmaltreated populations, as previously analyzed and reported by the Adverse Childhood Experience Study investigators.47,48 In particular, persons in all age groups who reported childhood maltreatment also reported significantly higher percentages of the other 5 measured adverse childhood experiences, compared with those who reported no childhood maltreatment. The measured economic variables were also significantly associated with childhood maltreatment. After we applied the stratified propensity score method, only 1 of the 25 strata had a significantly different mean propensity score, but the magnitude of the difference within this strata was slight (a score of 0.76 in the maltreated group vs 0.73 in the nonmaltreated group). Therefore, we concluded that the overall matching process was successful in reducing bias between the childhood-maltreatment and no-childhoodmaltreatment groups.43,44,49

Table 3 shows overall mean utility differences comparing the childhood-maltreatment group with the no-childhood-maltreatment group by age group and type of maltreatment. Overall, respondents who reported childhood maltreatment had a marginal utility difference (or disutility) of 0.028 (95% confidence interval [CI]=0.022, 0.034) compared with respondents who reported no childhood maltreatment. This result is in the range of what Walters and Brazier50 estimated as a minimally important difference (0.011 to 0.097) in utility for the SF-6D as measured in 11 studies. For every age group, the overall marginal difference in utilities for those reporting childhood maltreatment compared with those reporting no maltreatment were statistically significant at P<.05, with the largest difference occurring in the group aged 20 to 39 years and the smallest difference occurring in the group aged 60 to 69 years. Imputed utility scores by age group are provided for childhood-maltreatment and no-childhoodmaltreatment groups in Table 4.

Table 3 shows that, across all ages, emotional neglect had the strongest influence on the marginal disutility, followed by sexual abuse and physical abuse. Neither emotional abuse nor physical neglect significantly impacted the disutility across all age groups. However, type of maltreatment impacted the disutility differentially within each age group. For example, among those aged 19 to 49 years, physical abuse, sexual abuse, and emotional neglect significantly impacted disutility. Among those aged 50 to 59 years, however, only physical abuse significantly impacted disutility, and among those aged 60 to 69 years, only sexual abuse and emotional neglect significantly impacted disutility. Among those 70 years and older, only emotional abuse significantly impacted disutility. In fact, the influence of emotional abuse on disutility was only significant among those 70 years and older.

DISCUSSION

We found that persons who experienced maltreatment during childhood had significant and sustained losses in preference-based HRQoL in adulthood, as measured by health utilities, compared with persons who did not experience maltreatment during childhood. Overall, adults who self-reported any form of childhood maltreatment had a yearly loss of 0.03 QALYs, or 11 days per year. Physical abuse, sexual abuse, and emotional neglect alone significantly reduced HRQoL per year by 0.015, 0.016, and 0.026 QALYs, respectively; emotional abuse or physical neglect alone did not. Preference-based HRQoL, or utility, losses among the childhood-maltreatment group compared with the no-childhood-maltreatment group significantly differed for all age groups, with higher differential losses in utilities found among the youngest age group (0.04 QALYs, or 15 days per year). These differential losses diminished with increasing age up until age 70 years and older, at which time the marginal difference in utility losses between the childhood-maltreatment and no-childhoodmaltreatment groups increased.
Limitations and Potential Biases

The retrospective nature of the self-report data may be one explanation for the declining differences in utility as age increased, with the slight exception of the group 70 years and older. One might question the reliability of older age groups in self-reporting events that may have occurred, in some cases, more then a half century ago. However, there is accumulating evidence that suggests that the unreliability of retrospective reports of trauma is overstated.51,52 For example, in another analysis that used the Adverse Childhood Experiences Study data, researchers found that Cohen's ? was in the good-to-excellent range when a test-retest reliability of the ACE measure was conducted.53 In addition, other analyses from the Adverse Childhood Experiences Study have not found that the association between adverse childhood events and HRQoL decreases with age.32

The recollection of personally experienced events such as childhood maltreatment may have more to do with when the maltreatment occurred and other factors occurring during childhood than with the age of the respondent. Memories of events that occurred before age 3.5 years are very unlikely to be recalled and memories from the 3.5- to 6-year age range are also less likely to be recalled than those that occurred during a later age.54 Older age when the maltreatment ended, maternal support following the disclosure of maltreatment, and more-severe maltreatment have all been found to be associated with an increased likelihood of disclosure.55,56

Another probable source of bias in our study relating to retrospective self reports of childhood maltreatment was that some cases of maltreatment may not have been selfidentified. In a prospective study of women's memory of childhood sexual abuse, Williams57 found, for example, that about 38% of abused women did not recall abuse that had been confirmed 17 years earlier. This type of misclassification would bias our results toward the null. It could be that the effect of childhood maltreatment on HRQoL was mediated by the biological or psychological developmental stage of the individual, with certain types of maltreatment resulting in differential effects over time. Although these data suggested that this phenomenon might exist, more research in this area is warranted, particularly surrounding the effects on HRQoL of different combinations of abuse and other adverse outcomes experienced during childhood.

There were a number of other limitations with this study that should be considered. First, type of childhood maltreatment and other adverse exposures were defined by a limited number of survey questions. As such, there could exist wide exposure variance within each category that is not accounted for in the model. Second, the sample was not representative of the US population and included a group who had good health care coverage and access to health care. Thus, we cannot easily draw the conclusion that these utility losses would be higher or lower in other populations. However, we suspect that in populations with limited access to health care, and mental health services in particular, the marginal difference in utilities between cases and controls might be even greater. Third, we excluded respondents for whom complete SF-36 data (and therefore SF-6D data) were not available, and if these data were not missing at random, our results could be biased. To the best of our knowledge, there are no methods to impute missing values for the transformed SF-6D. Fourth, others have noted that traumatic events tend to be more memorable.58,59 Therefore, adult self-reports of the neglect subtypes from the Adverse Childhood Experiences Study data may be less reliable than reports of the other maltreatment subtypes that are more traumatic.
Public Health Implications

Despite these limitations, translated over a typical lifespan of an individual (aged 75 years, for example), these data suggest that persons who experienced childhood maltreatment have a marginal decrease in at least 2 years of undiscounted quality-adjusted life expectancy, compared with persons who did not experience childhood maltreatment. A cost-effectiveness analysis of an intervention designed to prevent childhood maltreatment, therefore, would include 2 QALYs saved for every case of childhood maltreatment prevented. These results represent a floor effect of the true impact of childhood maltreatment on QALYs for 3 reasons. First, these estimates did not include losses in life years that may be associated with childhood maltreatment because of its influence on key risk factors for suicide and drugor alcohol-related fatalities.60,61 Our estimates of QALYs lost in a maltreated population also did not account for differential mortality rates associated with chronic diseases found to be correlated with childhood maltreatment. And, of potential greater impact, our estimates did not include HRQoL losses incurred during the acute stage of the maltreatment.

These utility loss estimates were also conservative in that other adverse childhood exposures were controlled for in the estimation of the propensity score, thus making the utility losses estimated in this analysis marginal to any utility losses that could occur with co-existing adverse childhood exposures. Dong et al.48 found that the presence of 1 adverse childhood exposure resulted in significantly higher odds (between 2 and 17.7 times) of reporting additional adverse childhood exposures. As a reduction in SF-36 score by increasing number of self-reported adverse childhood exposures was shown in Edwards et al.,32 we would expect utility losses to also increase with an increasing number of adverse childhood exposures. The marginal effect of the other adverse childhood exposures may be less influential then the effect of childhood maltreatment on utility, however. To test this, we estimated utility losses by ACE score and found that individuals with 5 or more adverse childhood exposures had a marginal utility difference of 0.067. Compared with individuals with zero adverse childhood exposures, an individual with 5 or more exposures would have a marginal decrease of at least 5 years (over his or her lifespan) of undiscounted quality-adjusted life expectancy.

The results presented here are an important first step for developing the benefits measure for use in economic evaluations. Economic evaluations are critical for policymakers charged with making allocation decisions with scarce public health resources. Use of a composite measure, such as the QALY, allows the decisionmaker to consider effects of the intervention on length of life and quality of life simultaneously. Applications of cost-effectiveness analyses to interventions that prevent childhood maltreatment are ideal because of the impact on life expectancy previously suggested by the literature and on quality of life as indicated by these results. If cost-effectiveness analyses of interventions to prevent childhood maltreatment are to be successful, further research to estimate the impact of childhood maltreatment severity and duration on quality of life and differential mortality losses associated with victims of childhood maltreatment are essential. This would require a serious commitment to collecting and analyzing longitudinal data on these victimized children. Improvements in HRQoL assessment of children, both in defining the dimensions of health appropriate for this age group and in improving elicitation methods, are also needed. When short-term losses in HRQoL are coupled with the long-term losses in HRQoL presented here, analysts will have a complete accounting of QALYs that could be saved per case of childhood maltreatment prevented.

[Referensi]
References
1. Felitti VJ, Anda RF, Nordenberg D, et al. Relationship of childhood abuse and household dysfunction to many of the leading causes of death in adults. The Adverse Childhood Experiences (ACE) Study. Am J Prev Med. 1998;14:245-258.
2. Kendall-Tackett K. Treating the Lifetime Health Effects of Childhood Victimization. Kingston, NJ: Civic Research Institute Inc; 2003.
3. Kolko DJ. Child physical abuse. In: Myers JEB, Berliner L, Briere J, Hendrix CT, Reid TA, Jenny CA, eds. The APSAC Handbook on Child Maltreatment. 2nd ed. Thousand Oaks, CA: Sage Publications Inc; 2002.
4. Malinosky-Rummell R, Hansen DJ. Long-term consequences of childhood physical abuse. Psychol Bull. 1993;114:68-79.
5. Putnam FW. Ten-year research update review: child sexual abuse. J Am Acad Child Adolesc Psychiatry. 2003;42:269-278.
6. Paolucci EO, Genuis ML, Violato C. A meta-analysis of the published research on the effects of child sexual abuse. J Psychol. 2001;135:17-36.
7. Fang X, Corso P. Child maltreatment, youth violence, and intimate partner violence: developmental relationships. Am J Prev Med. 2007;33:281-290.
8. Leeb RT, Paulozzi L, Melanson C, Simon T, Arias I. Child Maltreatment Surveillance: Uniform Definitions for Public Health and Recommended Data Elements, Version 1.0. Atlanta, GA: National Center for Injury Prevention and Control, Centers for Disease Control and Prevention; 2007.
9. Report of the Consultation on Child Abuse Prevention, 29-31 March 1999. Geneva, Switzerland: World Health Organization; 1999. Document WHO/HSC/ PVI/99.1.
10. Shonkoff JP, Phillips DA, eds. From Neurons to Neighborhoods: The Science of Early Childhood Development. Committee on Integrating the Science of Early Childhood Development. Washington, DC: National Academy Press; 2000.
11. McEwen BS, Seeman T. Protective and damaging effects of mediators or stress. Elaborating and testing concepts of allostasis and allostatic load. Ann N Y Acad Sci. 1999;896:30-47.
12. Seeman TE, Singer BH, Rowe JW, Horwitz RI, McEwen BS. Price of adaptation-allostatic load and its health consequences. MacArthur studies of successful aging. Arch Intern Med. 1997;157:2259-2268.
13. Repetti RL, Taylor SE, Seeman TE. Risky families: family social environments and the mental and physical health of offspring. Psychol Bull. 2002;128:330-366.
14. Heckman JJ. Skill formation and the economics of investing in disadvantaged children. Science. 2006;312: 1900-1902.
15. Taylor SE, Lerner JS, Sage RM, Lehman BJ, Seeman TE. Early environment, emotions, responses to stress, and health. J Pers. 2004;72:1365-1393.
16. Williams DR, Collins C. US socioeconomic and racial differences in health: patterns and explanations. Ann Rev Sociol. 1995;21:348-386.
17. Haddix A, Teutsch S, Corso P, eds. Prevention Effectiveness: A Guide to Economic Evaluation and Decision Analysis. New York, NY: Oxford University Press; 2003.
18. Owens DK. Analytic tools for public health decision making. Med Decis Making. 2002;22:S3-S10.
19. Aos S, Phipps P, Barnoski R, Lieb R. The Comparative Costs and Benefits of Programs to Reduce Crime. Seattle: Washington State Institute for Public Policy; 2001.
20. Hornick JP, Paetsch JJ, Bertrand LD. A Manual on Conducting Economic Analysis of Crime Prevention Programs. Ottawa, Ontario: National Crime Prevention Centre; 2002.
21. Miller TR, Levy DT. Cost outcome analysis in injury prevention and control: a primer on methods. Inj Prev. 1997;4:288-293.
22. Miller W, Robinson L, Lawrence R, eds. Valuing Health for Regulatory Cost-Effectiveness Analysis. Washington, DC: National Academies Press; 2006.
23. Gold M, Siegel J, Russell L, Weinstein M. Cost- Effectiveness in Health and Medicine. New York, NY: Oxford University Press; 1996.
24. Kaplan RM, Anderson JP. A general health policy model: update and applications. Health Serv Res. 1988; 23:203-235.
25. Patrick D, Erickson P. Health Status and Health Policy: Allocating Resources to Health Care. New York, NY: Oxford University Press; 1993.
26. Shaw JW, Johnson JA, Coons SJ. US valuation of the EQ-5D health states: development and testing of the D1 valuation model. Med Care. 2005;43:203-220.
27. Brazier JE, Roberts J. The estimation of a preference-based measure of health from the SF-12. Med Care. 2004;42:851-859.
28. Torrance G, Feeny DH, Furlong WJ, Barr RD, Zhang Y, Wang Q. Multiattribute utility function for a comprehensive health status classification system: Health Utilities Index Mark 2. Med Care. 1996;34:702-722.
29. Prosser L, Corso P. Measuring health-related quality of life for child maltreatment: a systematic literature review. Health Qual Life Outcomes. 2007;5:42.
30. Matza LS, Swensen AR, Flood EM, Secnik K, Leidy NK. Assessment of health-related quality of life in children: a review of conceptual, methodological, and regulatory issues. Value Health. 2004;7:79-92.
31. Corso PS, Lutzker JR. The need for economic analysis in research on child maltreatment. Child Abuse Negl. 2006;30:727-738.
32. Edwards VJ, Anda RF, Felitti VJ, Dube SR. Adverse childhood experiences and health-related quality of life as an adult. In: K. Kendall-Tackett, ed. Victimization and Health. Washington, DC: American Psychological Association; 2003.
33. Holbrook TL, Hoyt DB, Coimbra R, Potenza B, Sise M, Anderson JP. High rates of acute stress disorder impact on quality of life outcomes in injured adolescents: mechanism and gender predict acute stress disorder risk. J Trauma. 2005;59:1126-1130.
34. Walker EA, Gelfand A, Katon WJ, et al. Adult health status of women with histories of childhood abuse and neglect. Am J Med. 1999;107:332-339.
35. Dickinson LM, de Gruy FV III, Dickinson WP, Lark M, Candib LM. Health-related quality of life and symptom profiles of female survivors of sexual abuse. Arch Fam Med. 1999;8:35-43.
36. Ware JE Jr, Sherbourne CD. The MOS 36-item short-form health survey (SF-36). I. Conceptual framework and item selection. Med Care. 1992;30:473-483.
37. Straus MA. Measuring intrafamily conflict and violence: the Conflict Tactics (CT) scales. J Marriage Fam. 1979;41:75-88.
38. Wyatt GE. The sexual abuse of Afro-American and White-American women in childhood. Child Abuse Negl. 1985;9:507-519.
39. Bernstein DP, Fink L, Handelsman L, et al. Initial reliability and validity of a new retrospective measure of child abuse and neglect. Am J Psychiatry. 1994;151:1132-1136.
40. Brazier J, Usherwood T, Harper R, Thomas K. Deriving a preference-based single index from the UK SF-36 Health Survey. J Clin Epidemiol. 1998;51:1115-1128.
41. Brazier J, Roberts J, Deverill M. The estimation of a preference-based measure of health from the SF-36. J Health Econ. 2002;21:271-292.
42. Kaplan RM, Groessl EJ, Sengupta N, Sieber WJ, Ganiats TG. Comparison of measured utility scores and imputed scores from the SF-36 in patients with rheumatoid arthritis. Med Care. 2005;43:79-87.
43. Rosenbaum PR, Rubin DB. The central role of the propensity score in observational studies for causal effects. Biometrika. 1983;70:41-55.
44. Rosenbaum PR, Rubin DB. Reducing bias in observational studies using subclassification on the propensity score. J Am Stat Assoc. 1984;79:516-524.
45. Brown SE. Social class, child maltreatment, and delinquent behavior. Criminology. 1984;22:259-278.
46. Coleman PK, Maxey CD, Rue VM, Coyle CT. Associations between voluntary and involuntary forms of perinatal loss and child maltreatment among low-income mothers. Acta Paediatr. 2005;94:1476-1483.
47. Dube SR, Anda RF, Felitti VJ, Chapman DP, Williamson DF, Giles WH. Childhood abuse, household dysfunction and the risk of attempted suicide throughout the life span: findings from Adverse Childhood Experiences Study. JAMA. 2001;286:3089-3096.
48. Dong M, Anda RF, Felitti VJ, et al. The interrelatedness of multiple forms of childhood abuse, neglect, and household dysfunction. Child Abuse Negl. 2004; 28:771-784.
49. Cochran WG. The effectiveness of adjustment by subclassification in removing bias in observational studies. Biometrics. 1968;24:295-313.
50. Walters SJ, Brazier JE. Comparison of the minimally important difference for two health state measures: EQ-5D and SF-6D. Qual Life Res. 2005 14:1523-1532.
51. Brewin CR, Andrews B, Gotlib IH. Psychopathology and early experience: a reappraisal of retrospective reports. Psychol Bull. 1993;113:82-98.
52. Hardt J, Rutter M. Validity of adult retrospective reports of adverse childhood experiences: review of the evidence. J Child Psychol Psychiatry. 2004;45: 260-273.
53. Dube SR, Williamson DF, Thompson T, Felitti VJ, Anda RF. Assessing the reliability of retrospective reports of adverse childhood experiences among adult HMO members attending a primary care clinic. Child Abuse Negl. 2004;28:729-737.
54. Rogers ML. Factors influencing recall of childhood sexual abuse. J Trauma Stress. 1995;8:691-716.
55. Goodman GS, Ghetti S, Quas JA, et al. A prospective study of memory for child sexual abuse: new findings relevant to the repressed-memory controversy. Psychol Sci. 2003;14:113-118.
56. Ghetti S, Edelstein RS, Goodman GS, et al. What can subjective forgetting tell us about memory for childhood trauma. Mem Cognit. 2006;34:1011-1025.
57. Williams LM. Recall of childhood trauma: a prospective study of women's memories of child sexual abuse. J Consult Clin Psychol. 1994;62:1167-1176.
58. Porter S, Birt AR. Is traumatic memory special? A comparison of traumatic memory characteristics with memory for other emotional life experiences. Appl Cognitive Psychol. 2001;15:5101-5117.
59. Terr L. What happens to early memories of trauma? A study of twenty children under age five at the time of documented traumatic events. J Am Acad Child Adolesc Psychiatry. 1988;27:96-104.
60. Santa Mina EE, Gallop RM. Childhood sexual and physical abuse and adult self-harm and suicidal behavior: a literature review. Can J Psychiatry. 1998;43: 793-800.
61. Lo CC, Cheng TC. The impact of childhood maltreatment on young adults' substance abuse. Am J Drug Alcohol Abuse. 2007;33:139-146.
[Afiliasi Pengarang]
Phaedra S. Corso, PhD, Valerie J. Edwards, PhD, Xiangming Fang, PhD, and James A. Mercy, PhD

[Afiliasi Pengarang]
About the Authors
At the time of the study, Phaedra S. Corso was with the Department of Health Policy and Management, College of Public Health, University of Georgia, Athens, and the Division of Violence Prevention, Centers for Disease Control and Prevention, Atlanta, GA. Xiangming Fang and James A. Mercy are with the Division of Violence Prevention, Centers for Disease Control and Prevention, Atlanta. Valerie J. Edwards is with the Division of Adult and Community Health, Centers for Disease Control and Prevention, Atlanta.
Requests for reprints should be sent to Phaedra Corso, Department of Health Policy and Management, College of Public Health, University of Georgia, N125 Paul Coverdell Center, Athens, GA 30602-7397 (e-mail: pcorso@uga.edu).
This article was accepted October 24, 2007.
Note. The findings and conclusions in this article are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention.
Contributors

P.S. Corso originated the study and supervised all aspects of its implementation, synthesized analyses, and led the writing. V.J. Edwards and X. Fang assisted with the study and completed the analyses. J.A. Mercy assisted with the synthesis of the analyses and the writing of the article. All authors helped to conceptualize ideas, interpret findings, and review and edit drafts of the article.

Human Participant Protection
No human participants were involved in this study.

imsak-nur-subechi.blogspot.com

Estimation of LDL-Associated Apolipoprotein B from Measurements of Triglycerides and Total Apolipoprotein B

VLDL and chylomicrons may interfere with measurements of apolipoprotein B (apo B) on LDL particles. Ultracentrifugation of samples enriched in chylomicrons and VLDL and subsequent measurement of apo B in the infranate fraction [density (d) = 1.006] removes this interference. This apo B fraction is called "LDL-apo B."

METHODS: We retrospectively analyzed 64 895 measurements of triglycerides, total apo B, and LDL-apo B. Samples were ultracentrifuged, and 3 commercially available immunoassays that use different antibodies were used to measure LDL-apo B in the 1.006 infranate fraction.

RESULTS: After adjusting for triglyceride concentration, we found total apo B and LDL-apo B measurements to be strongly correlated. We derived a simple linear equation for calculating LDL-apo B concentration (in milligrams per deciliter) from measurements of total apo B and triglycerides: LDL-apo B = apo B - 10 mg/ dL - triglycerides/32. This equation accurately predicts LDL-apo B values within ± 12% of the measured value in 75% of cases.

CONCLUSIONS: Our equation provides a convenient means of estimating LDL-apo B from commonly available measurements of total apo B and triglycerides without the need for ultracentrifugation. LDL-apo B measurements were also independent of the different apo B antibodies in the 3 assays used in this study. An equation that predicts LDL-apo B particle number may be useful, regardless of the apo B assay used.


Measurements of plasma lipids (cholesterol, triglycerides) and HDL and LDL cholesterol subfractions (HDL-C, LDL-C) have been used to assess cardiovascular risk for several decades. Cholesterol, triglyceride, and HDL-C measurements have been readily available for quite some time, whereas LDL-C has been routinely estimated with the Friedewald equation. Although the reliability of this calculation decreases as the triglyceride concentration increases, the Friedewald equation is a useful clinical tool for estimating LDL-C in most cases, even with the current availability of assays that measure LDL-C directly.

Measurements of apolipoprotein B (apo B) are better for distinguishing and predicting cardiovascular disease risk than either measuring LDL-C directly (especially when triglycerides are high (1-6)) or calculating the LDL-C concentration via the Friedewald formula (7-11). Apo B measurement has been standardized, and current methods are highly precise and accurate. Apo B immunoassays use antibodies that recognize epitopes on LDL-C particles containing apo B-IOO, the main structural protein of these particles. Turbidity is a source of interference in apo B immunoassays of lipemic samples, and cross-reactivity of apo B-IOO antibodies with apo B-48 on chylomicrons can also produce interference. We propose that apo B values obtained for the 1.006 infranate fraction [density (if) = 1.006 kg/L)] of a sample after ultracentrifugation (hereby termed LDL-apo B) accurately estimate LDL particle numbers. Furthermore, we provide a simple equation for estimating LDL-apo B from measurements of triglycérides and apo B that is independent of the antibody used in the immunoassay.

In the beta-quantification reference method by which LDL-C is measured, a sample is ultracentrifuged through a solution of d = 1.006 kg/L. The supranate contains chylomicrons, VLDL1 and any beta-VLDL, whereas the infranate contains LDL, HDL, intermediate-density lipoprotein (IDL), and lipoprotein (a). Thus, ultracentrifugation can be used as a means to remove chylomicrons and VLDL from LDL and IDL. Apo B measured in the 1.006 infranate fraction is associated with LDL and IDL, which are considered atherogenic compared with VLDL particles. In addition, LDL-apo B has been associated with arteriographie changes in the FATS study and with decreased numbers of events and regression of coronary lesions in the SCRIP study (6, 12). LDL-apo B is likely to be a better marker of cardiovascular disease risk than total apo B. LDL-apo B has been measured >60 000 times at Berkeley HeartLab using 3 distinct commercially available assays with different specificities for apo B. In each measurement, 2 parts of an aqueous NaCl solution ( 11.5 g NaCl plus O. l g EDTA per 1 L water; d = 1.006) are added to 1 part patient serum, and the procedure is followed by ultracentrifugation in a Beckman 50.4 Ti fixed-angle rotor with a Beckman Coulter L8-80MR or L80-XP ultracentrifuge (Beckman Coulter) at 45 000 rpm for 11 h at 10 °C, resulting in 218 068& The top third of the centrifuged sample contains chylomicrons and VLDL and is removed. The apo B content in the bottom fraction (d > 1.006} is then measured with commercially available reagents in automated chemistry analyzers. We have used the following 3 methods to measure apo B in the 1.006 fraction: (a) Abbott reagents (cat. no. 9D93) with the Abbott Aeroset, (b) Roche reagents (cat. no. 03032639) with the Roche/Hitachi Module P-800, and (c) Kamiya K-Assay reagents (cat. no. KAI-024) with the Roche/ Hitachi Module P-800.

Perbesar 200%
Perbesar 400%

Fig. 1. Correlation between LDL-apo B and total apo B values at 210 mg/dL triglycerides (n = 1 000) (A) and at 700-950 mg/dL triglycerides (n = 1 000) (B).
The mean difference between apo B and LDL-apo B values is correlated with the triglyceride (TG) concentration (n = 8940) (C). Closed circles represent the mean difference at each triglyceride concentration; the line is a plot of the equation (in figure). Calculated and measured LDL-apo B values are strongly correlated (n = 1200, method a) (D).


We used deidentified data from 64 895 samples in our laboratory information system in a retrospective analysis in which we retrieved triglyceride, total apo B, and LDL-apo B data for each sample. LDL-apo B was measured with one of the 3 methods mentioned above. The data set was limited to samples with triglyceride concentrations >200 mg/dL,1 the threshold above which LDL-apo B was measured. This threshold was chosen because it is near the 95th percentile of triglyceride values for our laboratory's patient population and because our laboratory can accommodate routine measurement of LDL-apo B for this quantity of samples. Total apo B and LDL-apo B were strongly correlated, especially after adjusting for triglyceride concentration (Fig. 1, A and B). Plotting the difference between apo B and LDL-apo B values against triglyceride concentration revealed a linear relationship until the triglyceride concentration exceeded approximately 800 mg/dL (Fig. 1C). The following linear equation fits the data obtained with any of the 3 methods:
LDL-apo B = apo B - 10 mg/dL - triglycerides/32,1
where LDL-apo B, apo B, and tnglycerides are in milligrams per deciliter and apo B is the concentration of total apo B.


To demonstrate that this equation accurately predicts measured LDL-apo B values, we compared LDLapo B values calculated for subsets of data (n = 1200) with measured values obtained with each of the 3 methods (Table 1). The equation predicted measured LDL-apo B values to within ± 15% for all 3 methods in >85% of the cases (n = 3600). A comparison of all the LDL-apo B measurements (n = 64 895) with calculated LDL-apo B values demonstrated that the equation predicted the measured value to within ±15% in 84% of the cases. The correlations between measured and calculated LDL-apo B values were high (R^sup 2^ ≥ 0.88; Fig. 1D) and were independent of the method used. Thus, our equation appears to correctly predict measured LDL-apo B values. Although the performance criteria for apo B measurements have not been formally established, the total error for LDL-C has been established by the National Cholesterol Education Program to be ≤12% (13). Our equation was able to meet this 12% performance criterion with respect to measured LDL-apo B values in 75% of the cases (n = 64 895, all data for the 3 methods).

Other, more complex equations fit the data slightly better; however, the simplicity of our equation does not appear to compromise its predictive power compared with the more complex equations. Furthermore, its simplicity facilitates its use because it is easier for a clinician to remember. The triglycerides/32 term takes into consideration the VLDL contribution to apo B and is reminiscent of the triglycerides/5 term in the Friedewald equation. Because ultracentrifugation is not readily available in most clinical laboratories, our equation conveniently estimates LDL-apo B from commonly available measurements for total apo B and triglycerides. Furthermore, because our data suggest that the LDL-apo B values obtained with the 3 apo B assays are similar, this equation is likely to be broadly applicable and independent of the particular apo B assay used. Clinically, this equation can be used to estimate atherogenic apo B particle number without requiring specialized laboratory measurements. Further investigation is needed to determine whether this or a similar equation can be used universally for calculating LDL-apo B.

Grant/Funding Support: None declared.
Financial Disclosures: None declared.

[Catatan Kaki]
1 Editor's footnote: Although this journal has a policy of using SI nomenclature and expressing triglyceride and apo B concentrations in milligrams per liter, consideration of the likely clinical application of the equation presented herein requires these concentrations to be left expressed in milligrams per deciliter, the units in common clinical usage.


[Referensi]
References
1. Walldius G, Jungner I, Holme I, Aastveit AH, Kolar W, Steiner E. High apolipoprotein B, low apolipoprotein A-I. and improvement in the prediction of fatal myocardial infarction (AMORIS study): a prospective study. Lancet 2001;358:2026-33.
2. Yusuf S, Hawken S, Ounpuu S, Dans T. Avezum A, Lanas F, et al. Effect of potentially modifiable risk factors associated with myocardial infarction in 52 countries (the INTERHEART study): case-control study, lancet 2004;364:937-52.
3. Gotto AM Jr, Whitney E, Stein EA, Shapiro DR, Clearlield M, Weis S, et al. Relation between baseline and on-treatment lipid parameters and first accurate major coronary events in the Air Force/Texas Coronary Atherosclerosis Prevention Study (AFCAPS/TexCAPS). Circulation 2000;101: 477-84.
4. Contois JH, McNamara JR, Lammi-Keefe CJ, Wilson PW, Massow T, Schaefer EJ. Reference interval for plasma apolipoprotein B determined with a standardized commercial immunoturbidimetric assay: results from the Framingham Offspring Study. Clin Giern 1996;42:515-23.
5. Sniderman AD. Apolipoprotein B and apolipoprotein A1 as predictors of coronary artery disease. Can J Cardiol 1988;4(Suppl A):24A-30A.
6. Brown G, Albers JJ, Fisher LD. Regression of coronary artery disease as a result of intensive lipid-lowering therapy in men with high levels of apolipoprotein B. N Engl J Med 1990;323:128998.
7. Marcovina SM, Albers JJ, Dati F, Ledue TB, Ritchie RF. International Federation of Clinical Chemistry standardization project for measurements of apolipoproteins A-I and B. Clin Chem 1991;37:167682.
8. Albers JJ. Marcovina SM, Kennedy H. International Federation of Clinical Chemistry standardization project for measurements of apolipoproteins A-I and B, II. Evaluation and selection of candidate reference materials. Clin Chem 1992; 38:658-62.
9. Marcovina SM, Albers JJ, Henderson LO, Hannon WH. International Federation of Clinical Chemistry standardization project for measurements of apolipoproteins A-I and B. III. Comparability of apolipoprotein A-I values by use of international reference materiai. Clin Chem 1993;39:773-81.
10. Marcovina SM, Albers JJ, Kennedy H, Mei JV, Henderson LO. Hannon WH. International Federation of Clinical Chemistry standardization project for measurements of apolipoproteins A-I and B. IV. Comparability of apolipoprotein B values by use of international reference material. Clin Chem 1994;40:586-92.
11. Jungner I, Marcovina SM, Walldius G, Holme I, Kolar W, Steiner E. Apolipoprotein B and A-I values in 147576 Swedish males and females, standardized according to the World Health Organization-International Federation of Clinical Chemistry First International Reference Materials. Clin Chem 1998;44:1641-9.
12. Haskell WL. Alderman EL, Fair JM, Superko HR, Maron DJ, Champagne MA, et al. Beneficial angiographie and clinical response to multifactor modification in the Stanford Coronary Risk Intervention Project (SCRIP). Circulation 1994;89: 975-90.
13. Myers GL, Cooper GR, Greinberg N, Kimberly MM, Waymack PP, Hassemer DJ. Standardization of iipid and lipoprotein measurements. In: Rifai N, Warnick G, Dominiczak MH, eds. Handbook of lipoprotein testing, 2nd ed. Washington, DC: AACC Press, 2000:726.
DOI: 10.1373/clinchem.2007.100941

imsak-nur-subechi.blogspot.com