Markov transition models for binary repeated measures with ignorable and nonignorable missing values

Motivated by problems encountered in studying treatments for drug dependence, where repeated binary outcomes arise from monitoring biomarkers for recent drug use, this article discusses a statistical strategy using Markov transition model for analyzing incomplete binary longitudinal data. When the mechanism giving rise to missing data can be assumed to be ;ignorable', standard Markov transition models can be applied to observed data to draw likelihood-based inference on transition probabilities between outcome events. Illustration of this approach is provided using binary results from urine drug screening in a clinical trial of baclofen for cocaine dependence. When longitudinal data have ;nonignorable' missingness mechanisms, random-effects Markov transition models can be used to model the joint distribution of the binary data matrix and the matrix of missingness indicators. Categorizing missingness patterns into those for occasional or ;intermittent' missingness and those for monotonic missingness or ;missingness due to dropout', the random-effects Markov transition model was applied to a data set containing repeated breath samples analyzed for expired carbon monoxide levels among opioid-dependent, methadone-maintained cigarette smokers in a smoking cessation trial. Markov transition models provide a novel reconceptualization of treatment outcomes, offering both intuitive statistical values and relevant clinical insights.

The importance of social context: neighborhood stressors, stress-buffering mechanisms, and alcohol, drug, and mental health disorders

This study examines the relationship among neighborhood stressors, stress-buffering mechanisms, and likelihood of alcohol, drug, and mental health (ADM) disorders in adults from 60 US communities (n=12,716). Research shows that larger support structures may interact with individual support factors to affect mental health, but few studies have explored buffering effects of these neighborhood characteristics. We test a conceptual model that explores effects of neighborhood stressors and stress-buffering mechanisms on ADM disorders. Using Health Care for Communities with census and other data, we found a lower likelihood of disorders in neighborhoods with a greater presence of stress-buffering mechanisms. Higher neighborhood average household occupancy and churches per capita were associated with a lower likelihood of disorders. Cross-level interactions revealed that violence-exposed individuals in high crime neighborhoods are vulnerable to depressive/anxiety disorders. Likewise, individuals with low social support in neighborhoods with high social isolation (i.e., low-average household occupancy) had a higher likelihood of disorders. If replicated by future studies using longitudinal data, our results have implications for policies and programs targeting neighborhoods to reduce ADM disorders.

The effects of health sector market factors and vulnerable group membership on access to alcohol, drug, and mental health care

OBJECTIVE

This study adapts Andersen's Behavioral Model to determine if health sector market conditions affect vulnerable subgroups' use of alcohol, drug, and mental health services (ADM) differently than the general population, focusing specifically on community-level predisposing and enabling characteristics.

DATA SOURCES

Wave 2 data (2000-2001) from the Health Care for Communities study, supplemented with cases from wave 1 (1997-1998), were merged with area characteristics taken from Census, Area Resource File (ARF), and other data sources.

STUDY DESIGN

The study used four-level hierarchical logistic regression to examine access to ADM care from any provider and specialty ADM access. Interactions between community-level predisposing and enabling vulnerability characteristics with individual race/ethnicity, age, income category, and insurance type were explored.

PRINCIPAL FINDINGS

Nonwhites, the poor, uninsured, and elderly had lower likelihoods of service use, but interactions between race/ethnicity, income, age and insurance status with community-level vulnerability factors were not statistically significant for any service use. For ADM specialty care, those with Medicare, Medicaid, private fully managed, and private partially managed insurance, the likelihood of utilization was higher in areas with higher HMO penetration. However, for those with other insurance or no insurance plan, the likelihood of utilization was lower in areas with higher HMO penetration.

CONCLUSIONS

Community-level enabling factors explain part of the effect of disadvantaged status but, with the exception of the effect of HMO penetration on the relationship between insurance and specialty care use, do not modify any of the residual individual-level effects of disadvantage. Interventions targeting both structural and individual levels may be necessary to address the problem of health disparities. More research with longitudinal data is necessary to sort out the causal direction of social context and ADM access outcomes, and whether policy interventions to change health sector market conditions can shift ADM treatment utilization.

The mandible injury severity score: development and validity

PURPOSE

To develop and validate a clinical method for characterizing and scoring mandible injury severity.

MATERIALS AND METHODS

Constituent fracture variables (fracture type, location, occlusion, soft tissue involvement, infection, and interfragmentary displacement [FLOSID]) were used to develop the FLOSID taxonomy for characterizing injury. Each component was assigned an empirical weight to help derive a summary measure of injury severity called the UCLA Mandible Injury Severity Score (MISS). Subsequently, MISS values were calculated for a group of 336 patients treated for mandible fractures. The validity of the summary score was evaluated by relating the MISS measure to the treatment modality used and to various variables, related as well as unrelated to injury outcomes.

RESULTS

Each of the FLOSID components correlated significantly with the MISS (P < .001). Unrelated variables, including ethnicity, education, and gender, had no correlation to the MISS. On average, patients treated with rigid internal fixation had a higher MISS than patients treated with maxillomandibular fixation (P < .001). The MISS had a statistically significant association with surrogate markers of injury severity such as sensory nerve deficit, need for hospitalization, and pain at 1-month follow-up (P < .001). However, there was no significant association between MISS and indicators of postoperative complications (infection, nonunion, malunion, malocclusion).

CONCLUSIONS

The FLOSID taxonomy offers a useful alternative to narrative summarization of mandible injury. The MISS is readily derived from clinical parameters obtained at the initial patient encounter and appears to be a valid index of mandible injury severity across important clinical domains.

Robustness of a multivariate normal approximation for imputation of incomplete binary data

Multiple imputation has become easier to perform with the advent of several software packages that provide imputations under a multivariate normal model, but imputation of missing binary data remains an important practical problem. Here, we explore three alternative methods for converting a multivariate normal imputed value into a binary imputed value: (1) simple rounding of the imputed value to the nearer of 0 or 1, (2) a Bernoulli draw based on a 'coin flip' where an imputed value between 0 and 1 is treated as the probability of drawing a 1, and (3) an adaptive rounding scheme where the cut-off value for determining whether to round to 0 or 1 is based on a normal approximation to the binomial distribution, making use of the marginal proportions of 0's and 1's on the variable. We perform simulation studies on a data set of 206,802 respondents to the California Healthy Kids Survey, where the fully observed data on 198,262 individuals defines the population, from which we repeatedly draw samples with missing data, impute, calculate statistics and confidence intervals, and compare bias and coverage against the true values. Frequently, we found satisfactory bias and coverage properties, suggesting that approaches such as these that are based on statistical approximations are preferable in applied research to either avoiding settings where missing data occur or relying on complete-case analyses. Considering both the occurrence and extent of deficits in coverage, we found that adaptive rounding provided the best performance.

Chronic Posttraumatic Stress Disorder After Facial Injury: A 1-year Prospective Cohort Study

BACKGROUND

This study examined the prevalence, severity, and predictors of persistent traumatic stress symptoms in socioeconomically disadvantaged adults after orofacial injury.

METHODS

A 1-year prospective study of 336 socioeconomically disadvantaged adults treated for orofacial injury at a Level I trauma center was conducted. Univariate analyses were performed on early measures of injury characteristics, prior trauma exposure, coping resources, and psychosocial functioning to select potential predictors of 1-year posttraumatic stress disorder (PTSD) scores; independence of variable contribution was then evaluated in multiple regression analyses.

RESULTS

A substantial number of patients (23%) continued to experience significant PTSD symptomatology at 12 months. Predictors of PTSD symptoms at 12 months included current and lifetime mental health and social service needs, lifetime social service use, prior trauma exposure, sum of stressful life events in the year preceding injury, patient report of pain severity and inadequate social support at 10 days postdischarge, and PTSD scores at 1 month. One-month PTSD symptoms, unmet social service need, and need for more instrumental and emotional support were independent predictors of 12-month PTSD outcomes. Limitations include loss to follow up, use of self-report measures, and the possibility of additional traumatization in the follow-up year influencing symptom levels.

CONCLUSIONS

Many socioeconomically disadvantaged adults manifest negative psychological outcomes even 1 year after an orofacial injury. Poor social support and unmet social service needs immediately after the injury, as well as high PTSD symptoms at 1 month postinjury, are strongly associated with the risk of developing chronic PTSD. The surgical management of orofacial injuries in disadvantaged individuals should integrate case management that addresses psychosocial sequelae and patient service needs.

Multiple imputation using an iterative hot-deck with distance-based donor selection

Hot-deck imputation offers advantages in reflecting salient features of data distributions in missing-data problems, but previous implementations have lacked the appeal associated with modern Bayesian statistical-computing techniques. We outline a strategy of iterative hot-deck multiple imputation with distance-based donor selection. With distance defined as a monotonic function of the difference in predictive means between cases, donors are chosen with probability inversely proportional to their distance from the donee. This method retains the implementation ease of ad hoc techniques, while incorporating the desirable features of Bayesian approaches. Special cases of our method include nearest-neighbor imputation and a simple random hot-deck. Iterating the procedure provides an analogy to Markov Chain Monte Carlo methods and is intended to mitigate dependence on starting values. Results from imputing missing values in a longitudinal depression treatment trial as well as a simulation study are presented. We evaluate how different definitions of distance, choices of starting values, the order in which variables are chosen for imputation, and the number of iterations impact inferences. We show that our measure of distance controls the tradeoff between bias and variance of our estimates. We find that inferences from the depression treatment trial are not sensitive to most definitions of distance. In addition, while differences exist between 1 iteration and 10 iterations, there are no meaningful differences between inferences based on 10 iterations and those based on 500 iterations. The choice of starting value did not have an impact on inferences but the order in which the variables were chosen for imputation was significant even after iteration.

Geographic variation in alcohol, drug, and mental health services utilization: what are the sources of the variation?

BACKGROUND

Studies have documented geographic variation in health services utilization over a range of medical, surgical, and psychiatric conditions. These geographic differences are of concern to policy makers, as they may represent either excessive levels of unnecessary care or inappropriately low utilization of necessary services. However, the sources of geographic variation are not well understood, and variation may not represent a quality problem, to the extent that geographic variation is due to sampling variability or variation in case-mix across sites.

AIMS OF THE STUDY

Our aim was to determine the extent to which geographic variation in assessment and treatment rates for alcohol, drug, and mental disorders (ADM) was due to variation in case-mix across sites and to quantify the amount of geographic variation after case-mix adjustment.

METHODS

We analyzed data from Healthcare for Communities, a nationally representative telephone survey of ADM disorders and treatment. We utilized fixed effects and random intercept models to analyze whether individuals received a brief primary care ADM assessment, any primary care ADM treatment, any specialty ADM treatment, or any ADM treatment. Using the coefficient of variation and intra-class correlation (ICC) as summaries, we simulated reference distributions for the amount of variability in ADM assessment and treatment rates expected due to variation in case mix across 60 geographic areas. We compared this with the actual variation among our 60 sites, and the variation that remained after adjusting for ADM disorders, physical health, and socioeconomic characteristics.

RESULTS

The amount of the variation in assessment and treatment rates explained by geographic area in unadjusted fixed effects and random intercepts models was statistically significant with R2 statistics ranging from 1% to 2% in fixed effects models and ICC's ranging from 0.009 to 0.043. Considerably more variation was explained in analyses that included individual level characteristics such as ADM disorders, physical health, and socioeconomic status, with R2 statistics from 10% to 19%. In random intercept models the ICC's were decreased 20 to 100% in models that adjusted for ADM disorders, physical health, and socioeconomic status.

DISCUSSION

We found significant variation in ADM assessment and treatment rates across geographic sites. However, the magnitude of geographic variation was relatively modest, with 0.9 to 4.3% of the total variation in ADM assessment and treatment occurring at the geographic level. Further, it appears that a moderate amount of this geographic variation may be due to differences in case mix across sites.

IMPLICATIONS FOR HEALTH POLICY

Although geographic variation in assessment and treatment rates can signal variable quality, there may be only modest potential for improving quality by reducing geographic variation. Further, the success of efforts to decrease geographic variation by focusing on provider behavior may be limited by the extent to which systematic variation is explained by individual characteristics.

IMPLICATIONS FOR FURTHER RESEARCH

Future work on geographic differences in mental health care as well as in health services more generally should pay particular attention to adjusting for individual differences in morbidity, which strongly predict treatment utilization.

Using patient self-report data to evaluate orofacial surgical outcomes

OBJECTIVES

This study analyzes results of 336 patients treated for mandible fractures at King/Drew Medical Center in South Central Los Angeles, California from August 1996 to December 2001. Subjects were enrolled in a prospective study to evaluate the association between patient's subjective evaluation and objective clinical evaluations on three surgical outcome measures following orofacial surgery.

METHODS

Subjects were assessed at four time periods--hospital discharge, 10 days post-discharge, 1 month post-discharge and 6 months post-discharge. Three outcome measures were utilized to represent perceived health and oral health-related quality of life--General Oral Health Assessment Index (GOHAI); Mental Health Inventory (MHI-5); and a single-item self-reported health status measure.

RESULTS

GOHAI scores at 1 month (mean=31.5, SD=9.5) were not substantially higher than at 10 days (mean=28.6, SD=8.8), but scores did improve substantially by 6 months (mean=42.6, SD=10.6). Mean mental health scores ranged from 17.7 at 10 days to 18.0 at 1 month and 18.6 at 6 months. Mean self-reported health status score were approximately 2.2 at all recalls, describing health as 'good.' A longitudinal growth curve analysis of GOHAI scores over four time periods indicated a significantly higher average intercept for the maxillomandibular fixation (MMF) treatment group (29.67) than in the rigid internal fixation (RIF) treatment group (25.38). Meanwhile, the increase in GOHAI scores over time was significantly greater in the RIF group than in the MMF group, resulting in scores being comparable between groups after 6 months.

CONCLUSIONS

By implication, patients with MMF self-report fewer problems in the early days after placement of the intra-arch wire compared with patients with RIF.

Psychosocial impacts of a camping experience for children with cancer and their siblings

BACKGROUND

We conducted a prospective two-group evaluation of pediatric cancer patients and their siblings regarding experiences and affective changes resulting from a 1-week summer camp experience.

METHODS

The patients and siblings were assessed prior to camp (Baseline), at the end of camp (Follow-up 1), and again 4-6 months later (Follow-up 2). Assessments included standardized tests for depressive affects, social competency, and a measure of pleasure and participation in camp activities.

RESULTS

Sixty-six children were assessed, including 31 (47%) patients and 35 (53%) siblings. Ages ranged from 7 to 17 years. Of the patient campers 19 (61%) had leukemia or lymphoma and 12 (39%) had solid tumors. Results showed marked changes in affective symptoms for patient campers over time (improvements), not shown by sibling campers. For patient campers these affective changes were not present immediately after camp, but were quite significant when measured 4-6 months later. Both patient and sibling campers reflected the same positive memories and pleasure in camp activities over time. For neither group did memories or pleasure fade over time. The camping experience did not have differential impacts on first time versus returning campers. Twelve campers (18% of sample) indicated suicidal ideation on the measure of depressive affects. They did well at camp and presented no special management issues.

CONCLUSION

Expectations appear substantially different for patient versus sibling campers. The camping experience appeared to impact these groups differently, with patient campers impacted in ways not experienced by sibling campers.

Fatigue in long-term breast carcinoma survivors: a longitudinal investigation

BACKGROUND

A longitudinal study was designed to evaluate the prevalence, persistence, and predictors of posttreatment fatigue in breast carcinoma survivors.

METHODS

A sample of 763 breast carcinoma survivors completed questionnaires at 1-5 and 5-10 years after diagnosis, including the RAND 36-item Health Survey, Center for Epidemiological Studies - Depression scale (CES-D), Breast Cancer Prevention Trial Symptom Checklist, and demographic and treatment-related measures.

RESULTS

Approximately 34% of study participants reported significant fatigue at 5-10 years after diagnosis, which is consistent with prevalence estimates obtained at 1-5 years after diagnosis. Approximately 21% reported fatigue at both assessment points, indicating a more persistent symptom profile. Longitudinal predictors of fatigue included depression, cardiovascular problems, and type of treatment received. Women treated with either radiation or chemotherapy alone showed a small improvement in fatigue compared with those treated with both radiation and chemotherapy.

CONCLUSIONS

Fatigue continues to be a problem for breast carcinoma survivors many years after cancer diagnosis, with 21% reporting persistent problems with fatigue. Several factors that may contribute to long-term fatigue are amenable to intervention, including depression and comorbid medical conditions.

Do specialist self-referral insurance policies improve access to HIV-experienced physicians as a regular source of care?

Health insurance policies that require prior authorization for specialty care may be detrimental to persons with HIV, according to evidence that having a regular physician with HIV expertise leads to improved patient outcomes. The objective of this study is to determine whether HIV patients who can self-refer to specialists are more likely to have physicians who mainly treat HIV. The authors analyze cross-sectional survey data from the HIV Costs and Services Utilization Study. At baseline, 67 percent of patients had insurance that permitted self-referral. In multivariate analyses, being able to self-refer was associated with an 8-12 percent increased likelihood of having a physician at a regular source of care that mainly treats patients with HIV. Patients who can self-refer are more likely to have HIV-experienced physicians than are patients who need prior authorization. Insurance policies allowing self-referral to specialists may result in HIV patients seeing physicians with clinical expertise relevant to HIV care.

Longitudinal patterns of alcohol, drug, and mental health need and care in a national sample of U.S. adults

OBJECTIVE

Use of longitudinal data can help clarify the extent of persistent need for services or persistent problems in gaining access to services. This study examined the level of transient and persistent need and unmet need over time among respondents to a national survey and whether need was met by provision of mental health services or resolved without treatment.

METHODS

Data from the longitudinal Health Care for Communities (HCC) household telephone survey were used to produce joint distributions of need status and care for two periods (wave 1 data collected in 1997 to 1998 and wave 2 data collected in 2000 to 2001; N=6,659). Perceived need was measured as self-report of need for help with a mental or substance use problem. Probable clinical need was assessed with the Composite International Diagnostic Interview, the Alcohol Use Disorders Identification Test, and the 12-item Short Form Health Survey.

RESULTS

High levels of persistent unmet need for care (44 to 52 percent) were found among respondents who had probable clinical need in wave 1. Although a majority of those with need received some care, an equal proportion (about 30 percent) of those with perceived need only or probable clinical need in wave 1 did not receive any care. A substantial portion of need (22 to 26 percent) appears to have resolved without treatment, which may suggest high levels of transient need.

CONCLUSIONS

Persistent patterns of unmet need represent important targets for policy and programs that can improve utilization, including outreach, education, and improved insurance coverage.

Outcomes From the Moving Beyond Cancer Psychoeducational, Randomized, Controlled Trial With Breast Cancer Patients

Purpose

Evidence suggests that the re-entry phase (ie, early period after medical treatment completion) presents distinct challenges for cancer patients. To facilitate the transition to recovery, we conducted the Moving Beyond Cancer (MBC) trial, a multisite, randomized, controlled trial of psychoeducational interventions for breast cancer patients.

Methods

Breast cancer patients were registered within 6 weeks after surgery. After medical treatment, they completed baseline measures and were randomly assigned to standard National Cancer Institute print material (CTL); standard print material and peer-modeling videotape (VID); or standard print material, videotape, two sessions with a trained cancer educator, and informational workbook (EDU). Two primary end points were examined: energy/fatigue and cancer-specific distress. Secondary end points were depressive symptoms and post-traumatic growth. Perceived preparedness for re-entry was analyzed as a moderator of effects.

Results

Of 558 women randomly assigned to treatment, 418 completed the 6-month assessment and 399 completed the 12-month assessment. In analyses controlling for study site and baseline depressive symptoms, VID produced significant improvement in energy/fatigue at 6 months relative to CTL, particularly among women who felt less prepared for re-entry at baseline. No significant main effect of the interventions emerged on cancer-specific distress, but EDU prompted greater reduction in this outcome relative to CTL at 6 months for patients who felt more prepared for re-entry. Between-group differences in the primary outcomes were not significant at 12 months, and no significant effects emerged on the secondary end points.

Conclusion

A peer-modeling videotape can accelerate the recovery of energy during the re-entry phase in women treated for breast cancer, particularly among those who feel less prepared for re-entry.

Depression among Latina immigrant mothers separated from their children

OBJECTIVE

Latinas who immigrate to the United States and leave their children in their homelands may experience psychological consequences of this separation. This study examined whether immigrant Latinas who were separated from their children differed in rates of probable major depression from those who lived with their children and from those who did not have any children.

METHODS

Data were obtained between March 1997 and May 2002 from women in Women, Infants, and Children programs that target low-income pregnant and postpartum women and their children (up to five years of age), women in county-run Title X family planning clinics, women in pediatric clinics for low-income families, and women who were living in a subsidized-housing project or attending programs for county welfare recipients. Latinas in this study were all immigrants to the United States. The women were screened for major depressive disorders with the Primary Care Evaluation of Mental Disorders.

RESULTS

A total of 5,122 Latina immigrants were screened. Overall, 11.7 percent of the sample screened positive for major depression. The rates of depression were 11.4 percent for women who lived with their children, 10.9 percent for those who did not have children, and 18.1 for those who were not living with their children. When the analyses controlled for demographic differences, the odds of depression for immigrant Latinas who were separated from their children were 1.52 times as great as the odds for those whose children were currently living with them (p=.02). Odds of depression were similar among women who lived with their children and those who did not have children.

CONCLUSIONS

Separation from children during immigration may lead to increased risk of depression for immigrant Latinas. Health care clinicians who treat young immigrant women should pay close attention to signs of depression among women who have left children with relatives in their homelands.

Imputation and Variable Selection in Linear Regression Models with Missing Covariates

Across multiply imputed data sets, variable selection methods such as stepwise regression and other criterion-based strategies that include or exclude particular variables typically result in models with different selected predictors, thus presenting a problem for combining the results from separate complete-data analyses. Here, drawing on a Bayesian framework, we propose two alternative strategies to address the problem of choosing among linear regression models when there are missing covariates. One approach, which we call "impute, then select" (ITS) involves initially performing multiple imputation and then applying Bayesian variable selection to the multiply imputed data sets. A second strategy is to conduct Bayesian variable selection and missing data imputation simultaneously within one Gibbs sampling process, which we call "simultaneously impute and select" (SIAS). The methods are implemented and evaluated using the Bayesian procedure known as stochastic search variable selection for multivariate normal data sets, but both strategies offer general frameworks within which different Bayesian variable selection algorithms could be used for other types of data sets. A study of mental health services utilization among children in foster care programs is used to illustrate the techniques. Simulation studies show that both ITS and SIAS outperform complete-case analysis with stepwise variable selection and that SIAS slightly outperforms ITS.

A qualitative report of patient problems and postoperative instructions

PURPOSE

While surgery related stress may interfere with the patient's ability to concentrate on instructions, language difficulty or low health literacy may also impede appropriate doctor/patient communication. The purpose of this study is to understand from a sample of minority patients the types of problems encountered during healing and the level of information regarding elements of postoperative instructions they recalled receiving at an inner-city safety net hospital. We initiated a qualitative study to understand the care sequence process and provision of informed consent and postoperative instruction.

METHODS

African American or Latino patients, 18 years of age or older, who had third molars removed under general anesthesia or received treatment for a mandibular fracture were recruited to participate in a focus group to discuss their treatment. Patients described their problem and any informed consent given about treatment risks and benefits and postoperative information they recalled.

RESULTS

A total of 137 former patients were approached, 57 agreed to participate (42%) and 34 of those (60%) completed the interview. Subjects included 14 females and 20 males. Five categories of patient problems were reported: physical, eating, treatment-related, psychosocial, and other problems. People reported 5 categories of coping strategies: medication use, physical treatments, dietary solutions, rest, and clinical assistance. Twenty people recalled being given informed consent, and 5 participants recalled no elements of informed consent. Overall, 14 participants recalled elements of postoperative instruction.

CONCLUSION

Gaps in patient understanding of postoperative care suggest room for improvement in postoperative instructions. Additional research is necessary to design and test high-quality postoperative instructions for surgical treatment and recovery in populations with limited health related literacy.

Depression prevalence in disadvantaged young black women--African and Caribbean immigrants compared to US-born African Americans

BACKGROUND

Research with Mexican Americans suggests that immigrants have lower rates of mental disorders than U. S.-born Mexican Americans. We examine the prevalence of depression, somatization, alcohol use and drug use among black American women, comparing rates of disorders among U. S.-born, Caribbean-born, and African-born subsamples.

METHODS

Women in Women, Infants and Children (WIC) programs, county-run Title X family planning clinics, and low-income pediatric clinics were interviewed using the PRIME-MD. In total, 9,151 black women were interviewed; 7,965 were born in the U. S., 913 were born in Africa, and 273 were born in the Caribbean.

RESULTS

Controlling for other predictors, U.S.-born black women had odds of probable depression that were 2.94 times greater than the African-born women (p<0.0001, 95% CI: 2.07, 4.18) and 2.49 times greater than Caribbean-born women (p<0.0016, 95% CI: 1.41, 4.39). Likelihood of somatization did not differ among women who were U. S. born, African born, or Caribbean born. Rates of alcohol and drug problems were exceedingly low among all three groups, with less than 1% of the women reporting either alcohol or drug problems.

CONCLUSIONS

These results mirror similar findings for Mexican immigrant as compared with American-born Mexican Americans. The findings suggest that living in the U. S. might increase depression among poor black women receiving services in county entitlement clinics. Further research with ethnically validated instruments is needed to identify protective and risk factors associated with depression in immigrant and U. S.-born poor black women.

Comparison of psychiatrists' and other physicians' assessments of their ability to deliver high-quality care

OBJECTIVE

This study compared psychiatrists' assessments of their ability to deliver high-quality care with those of other physicians.

METHODS

Data were used from the Community Tracking Study Physician Survey, a national survey of 12,528 physicians. Linear regression models were used to investigate the effects of type of physician (psychiatrist or other physician), managed care involvement, and ability to obtain inpatient care on four measures of physicians' assessments of quality.

RESULTS

In models that did not control for difficulty in obtaining inpatient services, assessments of quality were significantly lower among psychiatrists than among other physicians. Furthermore, although managed care involvement was associated with lower assessments of quality for all physicians in these models, the effects were stronger for psychiatrists than for other physicians. However, after difficulty in obtaining inpatient services was controlled for, psychiatrists' and other physicians' assessments of quality were similar.

CONCLUSIONS

Compared with other physicians, psychiatrists' assessments of the quality of care that they provide were lower and their assessments were more influenced by involvement with managed care. These differences may be mediated by difficulty in obtaining inpatient services. When designing and administering behavioral health benefits packages, clinical policy makers should consider the possible effects of reduction in inpatient services on quality of care.

A comparison of imputation methods in a longitudinal randomized clinical trial

It is common for longitudinal clinical trials to face problems of item non-response, unit non-response, and drop-out. In this paper, we compare two alternative methods of handling multivariate incomplete data across a baseline assessment and three follow-up time points in a multi-centre randomized controlled trial of a disease management programme for late-life depression. One approach combines hot-deck (HD) multiple imputation using a predictive mean matching method for item non-response and the approximate Bayesian bootstrap for unit non-response. A second method is based on a multivariate normal (MVN) model using PROC MI in SAS software V8.2. These two methods are contrasted with a last observation carried forward (LOCF) technique and available-case (AC) analysis in a simulation study where replicate analyses are performed on subsets of the originally complete cases. Missing-data patterns were simulated to be consistent with missing-data patterns found in the originally incomplete cases, and observed complete data means were taken to be the targets of estimation. Not surprisingly, the LOCF and AC methods had poor coverage properties for many of the variables evaluated. Multiple imputation under the MVN model performed well for most variables but produced less than nominal coverage for variables with highly skewed distributions. The HD method consistently produced close to nominal coverage, with interval widths that were roughly 7 per cent larger on average than those produced from the MVN model.

Imputation for incomplete high-dimensional multivariate normal data using a common factor model

It is common in applied research to have large numbers of variables measured on a modest number of cases. Even with low rates of missingness on individual variables, such data sets can have a large number of incomplete cases. Here we present a new method for handling missing continuously scaled items in multivariate data, based on extracting common factors to reduce the number of covariance parameters to be estimated in a multivariate normal model. The technique is compared in several simulation settings to available-case analysis and to a multivariate normal model with a ridge prior. The method is also illustrated on a study with over 100 variables evaluating an emergency room intervention for adolescents who attempted suicide.

Psychologic functioning and needs of indigent patients with facial injury: a prospective controlled study

PURPOSE

This study sought to examine 1) temporal changes in psychologic functioning over 12 months and 2) baseline differences in mental health and social service needs between orofacial injury patients and sociodemographically comparable controls undergoing elective oral surgery.

MATERIALS AND METHODS

Prospective, case-control study of patients treated at a public hospital in Los Angeles, CA. A total of 336 subjects with mandible fractures and 119 subjects undergoing elective removal of their third molars participated in structured, repeated follow-up assessments (10 days, 6 months, and 12 months after their surgical procedures). Multiple imputation was used to manage incomplete data, and propensity score analysis was used to correct for covariate imbalances between the injury and the control cohort. A series of ANOVAs, chi(2) analyses, and odds ratios was conducted.

RESULTS

The injury patients continue to experience significant psychologic distress for up to 12 months following the traumatic event. Orofacial injury patients also tend to report more lifetime and current mental health and social service needs than the sociodemographically similar elective-surgery cohort.

CONCLUSION

The management of facial injuries in disadvantaged individuals should integrate case management that addresses psychosocial sequelae and service needs of patients.

Quality of life at the end of primary treatment of breast cancer: first results from the moving beyond cancer randomized trial

BACKGROUND

During the last decade, survival rates for breast cancer have increased as a result of earlier detection and increased use of adjuvant therapy. Limited data exist on the psychosocial aspects of the transitional period between the end of primary treatment and survivorship. We investigated the baseline psychosocial status of women enrolled in a randomized trial testing two psychosocial interventions for women at the end of primary treatment.

METHODS

Participants, identified within 1 month after surgery (registration), provided demographic information and limited measures of quality of life. They were followed until they finished primary treatment (enrollment), at which time they completed a mailed baseline survey that included standardized measures of quality of life (including standardized scales of physical and emotional functioning), mood, symptoms, and sexual functioning. A total of 558 patients (mean age = 56.9 years) were enrolled in the study between July 1, 1999, and June 30, 2002. Health outcomes were examined according to treatment received: mastectomy with and without chemotherapy, and lumpectomy with and without chemotherapy. All statistical tests were two-sided.

RESULTS

Among all treatment groups, patients who had a mastectomy had the poorest physical functioning at registration (P<.001) and at enrollment (P=.05). At enrollment, mood and emotional functioning were similar among all patients, with no differences by type of treatment received. At enrollment, symptoms, including muscle stiffness, breast sensitivity, aches and pains, tendency to take naps, and difficulty concentrating, were common among patients in all groups and were statistically significantly associated with poor physical functioning and emotional well-being. Sexual functioning was worse for women who received chemotherapy than for those who did not, regardless of type of surgery (P<.001).

CONCLUSIONS

At the end of primary treatment for breast cancer, women in all treatment groups report good emotional functioning but report decreased physical functioning, particularly among women who have a mastectomy or receive chemotherapy. Clinical interventions to address common symptoms associated with treatment should be considered to improve physical and emotional functioning at the end of primary treatment for breast cancer.

Prediction of Neurocognitive Outcome in Adult Brain Tumor Patients

PURPOSE

To determine the relative contributions of patient, disease and therapy specific factors on neurocognitive outcome of brain tumor patients.

PATIENTS AND METHODS

Seventy-nine patients (mean age = 41.1 years; range: 17-75 years; 54% male, 46% female) with glioblastoma multiforme (37%), anaplastic astrocytoma (17%), low grade astrocytoma (13%), and oligodendroglioma (10%) predominantly in the frontal regions (45%) were evaluated in an outpatient neuro-oncology clinic. A neuropsychological test battery emphasized elements of attention/concentration. Multiple regression analyses determined relationships between functional outcomes and demographic and clinical predictors.

RESULTS

Key predictors of neurocognitive functioning included age of the patient (36-59 years, p < 0.01; >/= 60 years, p < 0.05) and frontal region tumor location (p < 0.01). As expected, older patients did not perform as well as younger patients in absolute terms on neuropsychological tests; decrements persisted when comparisons were based on age-standardized versions of neurocognitive outcomes. Major depressive disorder was marginally associated with outcomes, while surgical interventions and radiotherapy did not show strong associations with test performances.

CONCLUSIONS

Primary malignant brain disease was found to be less negative on neurocognitive outcomes for younger than for either middle-aged or older patients. Treatments were not as predictive of neurocognitive outcomes as age. No single test outcome measure was as sensitive to neurocognitive status as the empirically derived index of attention and concentration.

Do malpractice concerns, payment mechanisms, and attitudes influence test-ordering decisions?

Greater understanding is needed of nonclinical factors that determine neurologists' decisions to order tests. The authors surveyed 595 US neurologists and utilized demographic information, attitude scales, and clinical scenarios to evaluate the influence of nonclinical factors on test-ordering decisions. Greater test reliance, higher malpractice concerns, and receiving reimbursement for testing were all associated with a higher likelihood of test ordering. These findings have implications for training needs and suggest malpractice worries may inflate health care costs.

Physician performance assessment: nonequivalence of primary care measures

BACKGROUND

Assessment of the performance of primary care physicians requires multiple, reliable measures. This article explores the appropriateness of selected Health Plan Employer Data and Information Set (HEDIS) measures, developed to assess health plans, to assess individual physician performance.

OBJECTIVES

To determine the consistency and reliability of 4 measures of primary care physician performance measures: cancer screening, diabetic management, patient satisfaction, and ambulatory costs.

METHODS

The study population consisted of all 194 family practitioners and general internists providing ambulatory services in 1998 to a defined patient panel of 320,000 adult health maintenance organization members. Administrative data on physician practice and performance were assessed with multiple regression and analysis of variance.

RESULTS

Each performance measure was significantly related to 1 or 2 of the other measures: high cancer screening rates with good diabetic management and high patient satisfaction, good diabetic management with high cancer screening rates, high patient satisfaction with high cancer screening rates and high ambulatory costs, or high ambulatory costs with higher patient satisfaction. Although 76% of the physicians ranked in the highest third for at least 1 measure, 81% of these high performers ranked in the lower third for at least 1 other measure. Three percent of physicians ranked exclusively in the top or bottom third on all measures.

CONCLUSIONS

Care should be taken in assessing physicians based on narrow performance measures. Assessments of individual physicians with current performance measures might identify areas in which improvement is needed and to provide feedback to improve performance quality and efficiency. However, assumptions should not be made from one measure of performance to another.

Determinants of surgical decisions about mandible fractures

PURPOSE

The study goal was to explore contextual patient- and surgeon-related characteristics that influence the perception of injury severity and treatment strategy for mandible fractures.

METHODS

After reviewing plain radiographs of 22 patients with mandible fractures, 18 oral and maxillofacial surgeons were queried on summary severity ratings and treatment decisions for each injury. Subsequently, they were asked to indicate how various hypothetical fracture and patient-specific factors would alter their perception of injury severity and original treatment recommendations. The effect of the level of clinician trauma expertise on perception of injury severity and treatment choice was also assessed.

RESULTS

Each of the fracture-specific characteristics-number of constituent fractures, fracture complexity, degree of displacement, and summary injury severity-influenced the choice of treatment modality. Surgeon-specific characteristics were related to both perception of injury severity and treatment choice. Although clinicians with greater trauma loads tended to provide higher summary severity ratings for the same range of injuries (P <.001), they appeared to recommend maxillomandibular fixation for a much broader spectrum of injury severity (R = -0.42). Surgeons' perception of injury severity appeared to escalate with increasing damage to the soft tissue envelope; the influence of patient-related risk factors was less distinct. More than half of the surgeons suggesting maxillomandibular fixation for a particular case changed their treatment recommendation to rigid internal fixation on learning that the patient was noncompliant.

CONCLUSIONS

Clinical decision making for mandible fractures is not a precise and fully reliable activity. Contextual factors (fracture, patient, and surgeon related) appear to influence the clinical decision and may be responsible for the existing variations in practice patterns.

The development of acute post-traumatic stress disorder after orofacial injury: a prospective study in a large urban hospital

PURPOSE

Psychologic distress is a common outcome among trauma survivors. This report examines both the development and predictors of acute post-traumatic stress disorder (PTSD) symptoms in a sample of US inner-city orofacial trauma survivors seeking treatment in a publicly funded hospital.

PATIENTS AND METHODS

Baseline data were collected from 336 patients seeking urgent care for an oral injury (mandibular or midfacial fracture). Participants were predominantly unemployed, unmarried, African American or Hispanic men in their 30s. One-month follow-up assessments of PTSD symptoms were conducted on the available 84% of the sample.

RESULTS

Absolute levels of PTSD symptoms were high at 1 month; 25% of the sample appeared to meet diagnostic criteria for acute PTSD, based on a self-report of symptoms. Variables associated with self-reports of higher rates of PTSD symptoms included older age, being female, prior psychologic disturbance as reflected in lifetime and current mental health and social service need and use, exposure to and distress at a prior trauma as well as overall high rates of stressful life events in the past year, injury pain, psychologic distress at hospital discharge, and unmet social support needs during the recovery phase.

CONCLUSIONS

A substantial subsample of these traumatized medical patients had negative psychologic outcomes at 1 month. Results underscore the potential use of screening survivors of orofacial injury at urban trauma centers for PTSD and developing systems of care that facilitate referral to appropriate psychologic treatment.

Collaborative care management of late-life depression in the primary care setting: a randomized controlled trial

CONTEXT

Few depressed older adults receive effective treatment in primary care settings.

OBJECTIVE

To determine the effectiveness of the Improving Mood-Promoting Access to Collaborative Treatment (IMPACT) collaborative care management program for late-life depression.

DESIGN

Randomized controlled trial with recruitment from July 1999 to August 2001.

SETTING

Eighteen primary care clinics from 8 health care organizations in 5 states.

PARTICIPANTS

A total of 1801 patients aged 60 years or older with major depression (17%), dysthymic disorder (30%), or both (53%).

INTERVENTION

Patients were randomly assigned to the IMPACT intervention (n = 906) or to usual care (n = 895). Intervention patients had access for up to 12 months to a depression care manager who was supervised by a psychiatrist and a primary care expert and who offered education, care management, and support of antidepressant management by the patient's primary care physician or a brief psychotherapy for depression, Problem Solving Treatment in Primary Care.

MAIN OUTCOME MEASURES

Assessments at baseline and at 3, 6, and 12 months for depression, depression treatments, satisfaction with care, functional impairment, and quality of life.

RESULTS

At 12 months, 45% of intervention patients had a 50% or greater reduction in depressive symptoms from baseline compared with 19% of usual care participants (odds ratio [OR], 3.45; 95% confidence interval [CI], 2.71-4.38; P<.001). Intervention patients also experienced greater rates of depression treatment (OR, 2.98; 95% CI, 2.34-3.79; P<.001), more satisfaction with depression care (OR, 3.38; 95% CI, 2.66-4.30; P<.001), lower depression severity (range, 0-4; between-group difference, -0.4; 95% CI, -0.46 to -0.33; P<.001), less functional impairment (range, 0-10; between-group difference, -0.91; 95% CI, -1.19 to -0.64; P<.001), and greater quality of life (range, 0-10; between-group difference, 0.56; 95% CI, 0.32-0.79; P<.001) than participants assigned to the usual care group.

CONCLUSION

The IMPACT collaborative care model appears to be feasible and significantly more effective than usual care for depression in a wide range of primary care practices.

A randomized trial of medical care with and without physical therapy and chiropractic care with and without physical modalities for patients with low back pain: 6-month follow-up outcomes from the UCLA low back pain study

STUDY DESIGN

A randomized clinical trial.

OBJECTIVES

To compare the effectiveness of medical and chiropractic care for low back pain patients in managed care; to assess the effectiveness of physical therapy among medical patients; and to assess the effectiveness of physical modalities among chiropractic patients.

SUMMARY OF BACKGROUND DATA

Despite the burden that low back pain places on patients, providers, and society, the relative effectiveness of common treatment strategies offered in managed care is unknown.

METHODS

Low back pain patients presenting to a large managed care facility from October 30, 1995, through November 9, 1998, were randomly assigned in a balanced design to medical care with and without physical therapy and to chiropractic care with and without physical modalities. The primary outcome variables are average and most severe low back pain intensity in the past week, assessed with 0 to 10 numerical rating scales, and low back-related disability, assessed with the 24-item Roland-Morris Disability Questionnaire.

RESULTS

Of 1,469 eligible patients, 681 were enrolled; 95.7% were followed through 6 months. The mean changes in low back pain intensity and disability of participants in the medical and chiropractic care-only groups were similar at each follow-up assessment (adjusted mean differences at 6 months for most severe pain, 0.27, 95% confidence interval, -0.32-0.86; average pain, 0.22, -0.25-0.69; and disability, 0.75, -0.29-1.79). Physical therapy yielded somewhat better 6-month disability outcomes than did medical care alone (1.26, 0.20-2.32).

CONCLUSIONS

After 6 months of follow-up, chiropractic care and medical care for low back pain were comparable in their effectiveness. Physical therapy may be marginally more effective than medical care alone for reducing disability in some patients, but the possible benefit is small.

Increasing cervical cancer screening among underserved women in a large urban county health system: can it be done? What does it take?

BACKGROUND

Further reduction in avoidable cervical cancer morbidity and mortality may require system-wide, integrated approaches implemented in the public health facilities serving the nation's indigent and minority women.

OBJECTIVES

Report on the evaluation of a 5-year demonstration project testing a multicomponent (provider, system, and patient) intervention to increase cervical cancer screening among women who receive their health care through the Los Angeles County Department of Health Services, the second largest County Health Department in the nation.

MATERIALS AND METHODS

A longitudinal nonequivalent control group design was utilized. Data were collected during a baseline (no intervention) year and 2.5 years of intervention. A large hospital, one feeder Comprehensive Health Centers (CHC), and three of the health center's feeder Public Health Centers (PHC) received the intervention. Another hospital, CHC and its three feeder PHCs (matched on size, patient characteristics, and range of services provided) served as comparison sites. Independent random samples of patients 18 years and older were drawn annually at each site (n = 18,642). The outcome measure was a receipt of a Papanicolaou smear during a 9-month period.

RESULTS

At the Hospital and CHC levels a statistically significant intervention effect was observed after controlling for baseline screening rates and case mix. No intervention effect was observed at the PHCs.

CONCLUSION

An intensive multicomponent intervention can increase cervical cancer screening in a large, urban, County health system serving a low-income minority population of under screened women. Retention of program elements in the postresearch phase, and the difficulties and importance of conducting this type of research, is described.

Quality of life in long-term, disease-free survivors of breast cancer: a follow-up study

BACKGROUND

Women with breast cancer are the largest group of female survivors of cancer. There is limited information about the long-term quality of life (QOL) in disease-free breast cancer survivors.

METHODS

Letters of invitation were mailed to 1336 breast cancer survivors who had participated in an earlier survey and now were between 5 and 10 years after their initial diagnosis. The 914 respondents interested in participating were then sent a survey booklet that assessed a broad range of QOL and survivorship concerns. All P values were two-sided.

RESULTS

A total of 817 women completed the follow-up survey (61% response rate), and the 763 disease-free survivors in that group, who had been diagnosed an average of 6.3 years earlier, are the focus of this article. Physical well-being and emotional well-being were excellent; the minimal changes between the baseline and follow-up assessments reflected expected age-related changes. Energy level and social functioning were unchanged. Hot flashes, night sweats, vaginal discharge, and breast sensitivity were less frequent. Symptoms of vaginal dryness and urinary incontinence were increased. Sexual activity with a partner declined statistically significantly between the two assessments (from 65% to 55%, P =.001). Survivors with no past systemic adjuvant therapy had a better QOL than those who had received systemic adjuvant therapy (chemotherapy, tamoxifen, or both together) (physical functioning, P =.003; physical role function, P =.02; bodily pain, P =.01; social functioning, P =.02; and general health, P =.03). In a multivariate analysis, past chemotherapy was a statistically significant predictor of a poorer current QOL (P =.003).

CONCLUSIONS

Long-term, disease-free breast cancer survivors reported high levels of functioning and QOL many years after primary treatment. However, past systemic adjuvant treatment was associated with poorer functioning on several dimensions of QOL. This information may be useful to patients and physicians who are engaging in discussion of the risks and benefits of systemic adjuvant therapy.

Second Prize: The effectiveness of physical modalities among patients with low back pain randomized to chiropractic care: findings from the UCLA low back pain study

BACKGROUND

Although chiropractors often use physical modalities with spinal manipulation, evidence that modalities yield additional benefits over spinal manipulation alone is lacking.

OBJECTIVE

The purpose of the study was to estimate the net effect of physical modalities on low back pain (LBP) outcomes among chiropractic patients in a managed-care setting.

METHODS

Fifty percent of the 681 patients participating in a clinical trial of LBP treatment strategies were randomized to chiropractic care with physical modalities (n = 172) or without physical modalities (n = 169). Subjects were followed for 6 months with assessments at 2, 4, and 6 weeks and at 6 months. The primary outcome variables were average and most severe LBP intensity in the past week, assessed with numerical rating scales (0-10), and low back-related disability, assessed with the 24-item Roland-Morris Disability Questionnaire.

RESULTS

Almost 60% of the subjects had baseline LBP episodes of more than 3 months' duration. The 6-month follow-up was 96%. The adjusted mean differences between groups in improvements in average and most severe pain and disability were clinically insignificant at all follow-up assessments. Clinically relevant improvements in average pain and disability were more likely in the modalities group at 2 and 6 weeks, but this apparent advantage disappeared at 6 months. Perceived treatment effectiveness was greater in the modalities group.

CONCLUSIONS

Physical modalities used by chiropractors in this managed-care organization did not appear to be effective in the treatment of patients with LBP, although a small short-term benefit for some patients cannot be ruled out.

Clinician variability in characterizing mandible fractures

PURPOSE

This study evaluated variability in the clinical parameters commonly used to characterize mandible fractures.

PATIENTS AND METHODS

Inter-rater reliability of 18 oral and maxillofacial surgeons was assessed using radiographs of 22 cases of mandible fractures. Raters were asked to evaluate each case based on several parameters including number, location, and displacement of the individual fractures and severity of the composite injury. To evaluate intra-rater reliability, selected cases were reviewed at a second session by a subgroup of these surgeons. Tests of concordance used to quantify measurement reliability included the interclass correlation coefficient and multiple-rater kappa statistics.

RESULTS

Inter-rater agreement on the number of constituent fractures ranged from excellent for simpler fractures to poor for complex gunshot injuries. Even within raters, the range of interclass correlation for complex injuries was only 0.33 to 0.42 between the 2 assessments. Clinicians appeared to be better at delineating coronoid, condyle, ramus, and angle fractures; symphyseal and canine region fractures had lower inter- and intrarater agreement. Tests of concordance showed moderate to excellent reliability when fracture displacements were expressed in millimeters, but only fair reliability when displacements were expressed as categories. Even when the clinicians concurred on displacement measurements, a large overlap was observed in their categorization of these displacements. Despite the differences in the assessment of individual parameters, the high intrarater reliability coefficient (0.78) indicated that the individual clinicians had a high internal consistency in their assignment of summary severity scores. Multiple regression analysis revealed the number of constituent fractures, the type of fracture, and amount of fracture displacement (millimeters) to be significant predictors of clinician ratings of injury severity.

CONCLUSIONS

The clinician variability underscores the difficulties involved in trauma description and scoring. The study identifies some sources of clinician variability and emphasizes the need to standardize the characterization of mandible fractures by using explicit guidelines.

The 18-month impact of an emergency room intervention for adolescent female suicide attempters

Following a suicide attempt by female adolescents, the impact of a specialized emergency room (ER) care intervention was evaluated over the subsequent 18 months. Using a quasi-experimental design, this study assigned 140 female adolescent suicide attempters (SA), ages 12-18 years, and their mothers (88% Hispanic) to receive during their ER visit either: (a) specialized ER care aimed at enhancing adherence to outpatient therapy by providing a soap opera video regarding suicidality, a family therapy session, and staff training; or (b) standard ER care. The adjustment of the SA and their mothers was evaluated over 18 months (follow-up, 92%) using linear mixed model regression analyses. SA's adjustment improved over time on most mental health indices. Rates of suicide reattempts (12.4%) and suicidal reideation (29.8%) were lower than anticipated and similar across ER conditions. The specialized ER care condition was associated with significantly lower depression scores by the SA and lower maternal ratings on family cohesion. Significant interactions of intervention condition with the SA's initial level of psychiatric symptomatology indicated that the intervention's impact was greatest on maternal emotional distress and family cohesion among SA who were highly symptomatic. SA's attendance at therapy sessions following the ER visit was significantly associated with only one outcome--family adaptability. Specialized ER interventions may have substantial and sustained impact over time, particularly for the parents of youth with high psychiatric symptomatology.

The 18-month impact of an emergency room intervention for adolescent female suicide attempters

Following a suicide attempt by female adolescents, the impact of a specialized emergency room (ER) care intervention was evaluated over the subsequent 18 months. Using a quasi-experimental design, this study assigned 140 female adolescent suicide attempters (SA), ages 12-18 years, and their mothers (88% Hispanic) to receive during their ER visit either: (a) specialized ER care aimed at enhancing adherence to outpatient therapy by providing a soap opera video regarding suicidality, a family therapy session, and staff training; or (b) standard ER care. The adjustment of the SA and their mothers was evaluated over 18 months (follow-up, 92%) using linear mixed model regression analyses. SA's adjustment improved over time on most mental health indices. Rates of suicide reattempts (12.4%) and suicidal reideation (29.8%) were lower than anticipated and similar across ER conditions. The specialized ER care condition was associated with significantly lower depression scores by the SA and lower maternal ratings on family cohesion. Significant interactions of intervention condition with the SA's initial level of psychiatric symptomatology indicated that the intervention's impact was greatest on maternal emotional distress and family cohesion among SA who were highly symptomatic. SA's attendance at therapy sessions following the ER visit was significantly associated with only one outcome--family adaptability. Specialized ER interventions may have substantial and sustained impact over time, particularly for the parents of youth with high psychiatric symptomatology.

Multiple imputation and posterior simulation for multivariate missing data in longitudinal studies

This paper outlines a multiple imputation method for handling missing data in designed longitudinal studies. A random coefficients model is developed to accommodate incomplete multivariate continuous longitudinal data. Multivariate repeated measures are jointly modeled; specifically, an i.i.d. normal model is assumed for time-independent variables and a hierarchical random coefficients model is assumed for time-dependent variables in a regression model conditional on the time-independent variables and time, with heterogeneous error variances across variables and time points. Gibbs sampling is used to draw model parameters and for imputations of missing observations. An application to data from a study of startle reactions illustrates the model. A simulation study compares the multiple imputation procedure to the weighting approach of Robins, Rotnitzky, and Zhao (1995, Journal of the American Statistical Association 90, 106-121) that can be used to address similar data structures.

Management of multiple sclerosis across managed care and fee-for-service systems

OBJECTIVE

To measure and compare care for adults with MS across managed care and fee-for-service (FFS) health systems.

METHODS

The authors sampled adults with MS having physician visits over a 2-year period from a group model health maintenance organization (HMO) in southern California, from a midwestern independent practice association (IPA) model managed care plan, and from the FFS portion of the practices of a random sample of southern California neurologists. The authors mailed surveys to subjects in mid-1996; 930 of 1,164 (80%) of those eligible responded. The authors measured sociodemographic and clinical characteristics, management of recent changes in mobility, bladder control, and fatigue, use of a disease-modifying agent, assessment of general health symptoms and issues, and unmet information needs. The authors adjusted comparisons between systems for comorbidity, disease severity, and disease type.

RESULTS

The groups differed on most sociodemographic and clinical characteristics. There were few differences in symptom management; differences that did exist tended toward more referrals or treatment for the FFS group. Access to the disease-modifying agent available at the time of the survey did not differ across systems, although patients' perceptions of the rationale for not using the drug did vary. General health issues and symptoms were more often assessed in the FFS and IPA systems than in the HMO, but improvement was needed across all three systems of care. There were substantial unmet information needs in all groups and especially high ones in the FFS and HMO samples.

CONCLUSIONS

Strategies to improve care for people with MS should be developed and evaluated, particularly in areas like symptom assessment and meeting patient information needs. Where variations in service delivery exist, longitudinal studies are also needed to evaluate the potential impact on outcomes and to evaluate reasons for variation.

Role of breast reconstructive surgery in physical and emotional outcomes among breast cancer survivors

BACKGROUND

Tissue-sparing approaches to primary treatment and reconstructive options provide improved cosmetic outcomes for women with breast cancer. Earlier research has suggested that conservation or restitution of the breast might mitigate the negative effects of breast cancer on women's sexual well-being. Few studies, however, have compared psychosocial outcomes of women who underwent lumpectomy, mastectomy alone, or mastectomy with reconstruction. To address some of these issues, we examined women's adaptation to surgery in two large cohorts of breast cancer survivors.

METHODS

A total of 1957 breast cancer survivors (1-5 years after diagnosis) from two major metropolitan areas were assessed in two waves with the use of a self-report questionnaire that included a number of standardized measures of health-related quality of life, body image, and physical and sexual functioning. All P: values are two-sided.

RESULTS

More than one half (57%) of the women underwent lumpectomy, 26% had mastectomy alone, and 17% had mastectomy with reconstruction. As in earlier studies, women in the mastectomy with reconstruction group were younger than those in the lumpectomy or mastectomy-alone groups (mean ages = 50.3, 55.9, and 58.9, respectively; P: =.0001); they were also more likely to have a partner and to be college educated, affluent, and white. Women in both mastectomy groups complained of more physical symptoms related to their surgeries than women in the lumpectomy group. However, the groups did not differ in emotional, social, or role function. Of interest, women in the mastectomy with reconstruction group were most likely to report that breast cancer had had a negative impact on their sex lives (45.4% versus 29.8% for lumpectomy and 41.3% for mastectomy alone; P: =. 0001).

CONCLUSIONS

The psychosocial impact of type of primary surgery for breast cancer occurs largely in areas of body image and feelings of attractiveness, with women receiving lumpectomy experiencing the most positive outcome. Beyond the first year after diagnosis, a woman's quality of life is more likely influenced by her age or exposure to adjuvant therapy than by her breast surgery.

Improving follow-up after an abnormal pap smear: results from a quasi-experimental intervention study

The success of cervical cancer control programs depends on regular screening with the Pap smear test and prompt and appropriate treatment of early neoplastic lesions. Recognizing the potentially grave consequences of lack of follow-up for abnormal Pap smears, numerous intervention studies have tested the impact of a variety of strategies to increase return for follow-up. The majority of these studies were evaluated under controlled experimental conditions. Despite the encouraging findings of these trials, the next step in the research continuum requires that the effectiveness of these interventions be demonstrated in real world settings before full implementation is initiated. We report the results of an evaluation study assessing the combined effectiveness of three intervention modalities found effective in prior randomized studies: a tracking follow-up protocol, transportation incentives, and financial incentives. This study used a before-after, nonequivalent control group design to assess the impact of a multifaceted intervention that included a computerized tracking protocol with transportation and financial incentives. The study was implemented at two major hospitals, two comprehensive health centers (CHC), and nine public health centers (PHC) under the jurisdiction of the Los Angeles County Department of Health Services. One hospital, one CHC, and the four PHC located in the catchment area of the CHC were selected as experimental sites. The control sites - one hospital, one CHC, and five PHC - provided usual care. All women with an abnormal Pap smear at the intervention and control sites were included in the study. The study consisted of a 1-year period of baseline data collection (September 1989-August 1990), followed by a 2(1/2)-year intervention period (September 1990-February 1993). During the intervention period, the intervention protocol was implemented at the experimental sites, and the control sites provided usual care. Overall, we found that the rates of receipt of follow-up care were consistent with those found in similar studies. In contrast to results obtained in these prior randomized trials, we did not find strong and consistent evidence for intervention effects. Significant findings emerged only at the CHC and hospital levels and only for selected years. Results underscore the importance of testing interventions in real world conditions before large-scale implementation is initiated. In addition, this study highlights the challenge of detecting intervention effects in large-scale studies because of the greater measurement difficulties in field studies as compared with controlled experiments.

Care for depression in HMO patients aged 65 and older

OBJECTIVE

To examine treatment for depression among older adults in a large staff model health maintenance organization (HMO).

DESIGN

A 4-year prospective cohort study (1989-1993).

SETTING

Four primary care clinics of a large staff model HMO in Seattle, Washington.

PATIENTS

A total of 2558 Medicare enrollees aged 65 and older.

MAIN OUTCOME MEASURES

Treatment of depression was defined as primary care visits resulting in depression diagnoses, use of antidepressant medications, or specialty mental health services.

MAIN RESULTS

The older adults in our sample had low rates of treatment for depression, ranging from 4 to 7% in the entire sample and from 12 to 25% among those with probable depressive disorders. Predictors of treatment included female gender, severity, and persistence of depressive symptoms, and severity of comorbid medical illness. Even when patients were treated for depression, the intensity of treatment was very low. Overall likelihood of treatment for depression increased somewhat from 1989 to 1993, but among those treated, the rate of adequate antidepressant use remained below 30%.

CONCLUSIONS

There is still considerable need to improve care for older adults with depression in primary care.

Help-seeking steps and service use for children in foster care

This study describes help-seeking steps and service-use patterns for school-age children in foster care. It also examines how these access indices are moderated by sociodemographic, enabling, and child disorder factors. Two home interviews and a telephone teacher interview were conducted using a sample of 302 randomly selected children (age 6-12 years) in foster care. The majority of children (80%) were given a psychiatric diagnosis, and 43% of the foster parents perceived a need for mental health services for the child. In the past year, about one-half of the children had received mental health (51%) and special education services (52%). Age and ethnicity, foster parent education, placement history, level of monthly benefits, number of caseworker visits, and disorder characteristics were related to help-seeking steps and mental health service use. Strategies to improve access to mental health services for children in foster care should include interventions at the caregiver and system levels.

Managing menopausal symptoms in breast cancer survivors: results of a randomized controlled trial

BACKGROUND

Menopausal symptoms (e.g., hot flashes, vaginal dryness, and stress urinary incontinence) are very common in breast cancer survivors and cannot be managed with standard estrogen replacement therapy (ERT) in these patients. The purpose of this study was to test the efficacy of a comprehensive menopausal assessment (CMA) intervention program in achieving relief of symptoms, the improvement in quality of life (QOL), and sexual functioning in breast cancer survivors.

METHODS

Using a two-group, randomized controlled design, we assigned 76 postmenopausal breast cancer survivors with at least one severe target symptom either to the intervention group or to a usual-care group. Seventy-two women were evaluable at the end of the study period. The CMA intervention, delivered by a nurse practitioner, focused on symptom assessment, education, counseling and, as appropriate, specific pharmacologic and behavioral interventions for each of the three target symptoms. Psychosocial symptoms were assessed with the use of a self-report screening instrument, and distressed women were referred for counseling if needed. The intervention took place over a 4-month period. Outcomes measured were scores on a composite menopausal symptom scale, the RAND Short Form Health Survey Vitality Scale, and the Cancer Rehabilitation Evaluation System (CARES) Sexual Functioning Scale at baseline and at 4-month follow-up. All statistical tests were two-sided and were performed at the alpha =. 05 significance level.

RESULTS

Patients receiving the intervention demonstrated statistically significant improvement (P =.0004) in menopausal symptoms but no significant change in vitality (P =.77). Sexual functioning was statistically significantly improved (P =.04) in the treatment group compared with the usual-care group.

CONCLUSIONS

A clinical assessment and intervention program for menopausal symptom management in breast cancer survivors is feasible and acceptable to patients, leading to reduction in symptoms and improvement in sexual functioning. Measurable improvement in a general QOL measure was not demonstrated.

Fatigue in breast cancer survivors: occurrence, correlates, and impact on quality of life

PURPOSE

To describe the occurrence of fatigue in a large sample of breast cancer survivors relative to general population norms and to identify demographic, medical, and psychosocial characteristics of fatigued survivors.

PATIENTS AND METHODS

Breast cancer survivors in two large metropolitan areas completed standardized questionnaires as part of a survey study, including the RAND 36-item Health Survey, Center for Epidemiological Studies-Depression Scale, Breast Cancer Prevention Trial Symptom Checklist, Medical Outcomes Study Sleep Scale, and demographic and treatment-related measures.

RESULTS

On average, the level of fatigue reported by the breast cancer survivors surveyed (N = 1,957) was comparable to that of age-matched women in the general population, although the breast cancer survivors were somewhat more fatigued than a more demographically similar reference group. Approximately one third of the breast cancer survivors assessed reported more severe fatigue, which was associated with significantly higher levels of depression, pain, and sleep disturbance. In addition, fatigued women were more bothered by menopausal symptoms and were somewhat more likely to have received chemotherapy (with or without radiation therapy) than nonfatigued women. In multivariate analyses, depression and pain emerged as the strongest predictors of fatigue.

CONCLUSION

Although the majority of breast cancer survivors in this large and diverse sample did not experience heightened levels of fatigue relative to women in the general population, there was a subgroup of survivors who did report more severe and persistent fatigue. We identified characteristics of these women that may be helpful in elucidating the mechanisms underlying fatigue in this population, as well as directing intervention efforts.

Do We Have to Count One by One?

Who Counts? The Politics of Census-Taking in Contemporary America. Margo J. Anderson and Stephen E. Fienberg. Russell Sage Foundation, New York, 1999. 329 pp. $32.50. ISBN 0-87154-256-0.

Anderson and Fieberg provide a richly informative account of the historical and scientific context of the current controversy surrounding census-taking in the United States.

Routine outcome monitoring in a public mental health system: the impact of patients who leave care

OBJECTIVE

An interest exists in using patient outcome data to evaluate the performance of publicly financed mental health organizations. Because patients leave these organizations at a high rate, the impact of patient attrition on routinely collected outcome data was examined.

METHODS

In one county mental health system, routinely collected data on a wide range of outcomes were examined, and a random sample of patients who left treatment was interviewed.

RESULTS

Of the 1,769 patients in ongoing treatment during a one-year period, 554 (31 percent) were lost to follow-up. Among a random sample of 102 patients who left treatment, two had died and 47 were interviewed. Compared with patients who left treatment, patients who stayed were older, more likely to have schizophrenia, less likely to be married, more likely to be living in an institution, more satisfied with their relationships with friends and family, and less likely to have legal problems. Average outcomes improved both for patients who stayed and for patients who left. Patients who left and could be located for follow-up were less severely ill and showed the greatest improvement and the best outcomes. Patients who left and could not be located may have been more severely ill at baseline.

CONCLUSIONS

Outcomes appear to vary substantially by whether patients stay in care and whether they can be located after leaving care. Public mental health systems that wish to evaluate treatment quality using outcome data should attend carefully to which patients are being assessed. Biases can result from convenience sampling and from patients leaving care.