Permutation Tests for Assessing Potential Non-Linear Associations between Treatment Use and Multivariate Clinical Outcomes.

In many psychometric applications, the relationship between the mean of an outcome and a quantitative covariate is too complex to be described by simple parametric functions; instead, flexible nonlinear relationships can be incorporated using penalized splines. Penalized splines can be conveniently represented as a linear mixed effects model (LMM), where the coefficients of the spline basis functions are random effects. The LMM representation of penalized splines makes the extension to multivariate outcomes relatively straightforward. In the LMM, no effect of the quantitative covariate on the outcome corresponds to the null hypothesis that a fixed effect and a variance component are both zero. Under the null, the usual asymptotic chi-square distribution of the likelihood ratio test for the variance component does not hold. Therefore, we propose three permutation tests for the likelihood ratio test statistic: one based on permuting the quantitative covariate, the other two based on permuting residuals. We compare via simulation the Type I error rate and power of the three permutation tests obtained from joint models for multiple outcomes, as well as a commonly used parametric test. The tests are illustrated using data from a stimulant use disorder psychosocial clinical trial.

A note on median regression for complex surveys.

There is a great need for statistical methods for analyzing skewed responses in complex sample surveys. Quantile regression is a logical option in addressing this problem but is often accompanied by incorrect variance estimation. We show how the variance can be estimated correctly by including the survey design in the variance estimation process. In a simulation study, we illustrate that the variance of the median regression estimator has a very small relative bias with appropriate coverage probability. The motivation for our work stems from the National Health and Nutrition Examination Survey where we demonstrate the impact of our results on iodine deficiency in females compared with males adjusting for other covariates.

Meta-analysis of Total Effect Decomposition in the Presence of Multiple Mediators: The Example of Schizophrenia Treatment.

BACKGROUND: Causal mediation analysis addresses mechanistic questions by decomposing and quantifying effects operating through different pathways. Because most individual studies are underpowered to detect mediating effects, we outlined a parametric approach to meta-analyzing causal mediation and interaction analyses with multiple mediators, compared it with a bootstrap-based alternative, and discussed its limitations. METHODS: We employed fixed- and random-effects multivariate meta-analyses to integrate evidence on treatment-mediators and mediators-outcome associations across trials. We estimated path-specific effects as functions of meta-analyzed regression coefficients; we obtained standard errors using the delta method. We evaluated the performance of this approach in simulations and applied it to assess the mediating roles of positive symptoms of schizophrenia and weight gain in the treatment effect of paliperidone ER on negative symptoms across four efficacy trials. RESULTS: Both simulations and the application showed that the meta-analytic approaches increased statistical power. In the application, we observed substantial mediating effects of positive symptoms (proportions mediated from fixed-effects meta-analysis: (Equation is included in full-text article.)). Weight gain may have beneficial mediating effects; however, such benefit may disappear at high doses when metabolic side effects were excessive. CONCLUSIONS: Meta-analyzing causal mediation analysis combines evidence from multiple sources and improves power. Targeting positive symptoms may be an effective way to reduce negative symptoms that are challenging to treat. Future work should focus on extending the existing methods to allow for more flexible modeling of mediation.

The Treatment Ambassador Program: A Highly Acceptable and Feasible Community-Based Peer Intervention for South Africans Living with HIV Who Delay or Discontinue Antiretroviral Therapy.

We conducted a novel pilot randomized controlled trial of the Treatment Ambassador Program (TAP), an 8-session, peer-based, behavioral intervention for people with HIV (PWH) in South Africa not on antiretroviral therapy (ART). PWH (43 intervention, 41 controls) completed baseline, 3- and 6-month assessments. TAP was highly feasible (90% completion), with peer counselors demonstrating good intervention fidelity. Post-intervention interviews showed high acceptability of TAP and counselors, who supported autonomy, assisted with clinical navigation, and provided psychosocial support. Intention-to-treat analyses indicated increased ART initiation by 3 months in the intervention vs. control arm (12.2% [5/41] vs. 2.3% [1/43], Fisher exact p-value = 0.105; Cohen's h = 0.41). Among those previously on ART (off for > 6 months), 33.3% initiated ART by 3 months in the intervention vs. 14.3% in the control arm (Cohen's h = 0.45). Results suggest that TAP was highly acceptable and feasible among PWH not on ART.

Temporal patterns of suicide and circulatory system disease-related mortality are inversely correlated in several countries.

BACKGROUND: Nearly 800,000 suicides occur worldwide annually and suicide rates are increasing faster than population growth. Unfortunately, the pathophysiology of suicide remains poorly understood, which has hindered suicide prevention efforts. However, mechanistic clues may be found by studying effects of seasonality on suicide and other mortality causes. Suicides tend to peak in spring-summer periods and nadir in fall-winter periods while circulatory system disease-related mortality tends to exhibit the opposite temporal trends. This study aimed to determine for the first time whether monthly temporal cross-correlations exist between suicide and circulatory system disease-related mortality at the population level. If so and if common biological factors moderate risks for both mortality types, such factors may be discoverable and utilized to improve suicide prevention. METHODS: We conducted time series analyses of monthly mortality data from northern (England and Wales, South Korea, United States) and southern (Australia, Brazil) hemisphere countries during the period 2009-2018 (N = 41.8 million all-cause mortality cases). We used a Poisson regression variant of the standard cosinor model to determine peak months of mortality. We also estimated cross-correlations between monthly mortality counts from suicide and from circulatory system diseases. RESULTS: Suicide and circulatory disease-related mortality temporal patterns were negatively correlated in Australia (- 0.32), Brazil (- 0.57), South Korea (- 0.32), and in the United States (- 0.66), but no temporal correlation was discernable in England and Wales. CONCLUSIONS: The negative temporal cross-correlations between these mortality types we found in 4 of 5 countries studied suggest that seasonal factors broadly and inversely moderate risks for circulatory disease-related mortality and suicide, but not in all regions, indicating that the effect is not uniform. Since the seasonal factors of temperature and light exert opposite effects on suicide and circulatory disease-related mortality in several countries, we propose that physiologically-adaptive circulatory system responses to heat and light may increase risk for suicide and should be studied to determine whether they affect suicide risk. For example, heat and light increase production and release of the bioactive gas nitric oxide and reduce circulatory system disease by relaxing blood vessel tone, while elevated nitric oxide levels are associated with suicidal behavior, inverse effects that parallel the inverse temporal mortality patterns we detected.

Levels of Depersonalization and Derealization Reported by Recovered and Non-recovered Borderline Patients Over 20 Years of Prospective Follow-up.

Borderline personality disorder (BPD) is a serious psychiatric illness, and it is often associated with dissociative symptoms. The purpose of this study was to assess the course of depersonalization and derealization symptoms in recovered and non-recovered borderline patients over 20 years of prospective follow-up. The Dysphoric Affect Scale (DAS) - a 50-item self-report measure was administered to 290 borderline inpatients at baseline, and the remaining participants (85%) at 10 follow-up interviews conducted over 20 years. The level of depersonalization and derealization experienced by borderline patients was assessed using three items (feeling unreal, feeling completely numb, and feeling like people and things aren't real) from the DAS. The patients who recovered from BPD reported significantly lower scores in all three inner states (62 - 63%) at baseline compared to those patients who did not recover. Furthermore, scores of recovered and non-recovered groups decreased significantly in all three inner states studied over 20 years of prospective follow-up. Overall, these results suggest that the severity of depersonalization and derealization symptoms decreased significantly over 20 years of prospective follow-up and had a strong association with BPD recovery status.

Impact of transversus abdominis plane blocks versus non-steroidal anti-inflammatory on post-operative opioid use in ERAS ovarian cancer surgery.

BACKGROUND: Treatment of ovarian cancer often requires extensive surgical resection. The transversus abdominis plane (TAP) block has been utilized in benign gynecologic surgery to decrease post-operative pain and opioid use. We hypothesized that TAP blocks would decrease total opioid use in the first 24 hours and decrease length of stay following staging and cytoreductive surgery for ovarian cancer compared with either no local anesthetic or local wound infiltration alone. METHODS: All patients undergoing surgery for ovarian cancer from November 2016 to June 2017 were included in this retrospective cohort study. Median opioid use at 24, 48, and 72 hours post-operatively, as well as length of stay, were assessed. Multivariate median regression analysis was performed to adjust for potential confounders: post-operative non-steroidal anti-inflammatory drug (NSAID) usage, pre-operative opioid consumption, and extent of cytoreductive surgery. Length of stay was compared using Cox regression analysis. RESULTS: One-hundred-and-twenty patients were included in the analysis. Eighty-two patients had a TAP block, while 38 did not. After adjusting for potential confounders there was no difference in median 24 hours opioid consumption (p=0.29) or length of stay (HR 0.95, p=0.78) between patients receiving TAP block compared with those without. After adjusting for potential confounders, patients receiving scheduled NSAIDs had a 2.6-fold greater likelihood of being discharged (p<0.0005) and a significant reduction in median opioid use at 24 hours (23.5 vs 14.5 mg) (p=0.017) compared with those patients without NSAIDs. DISCUSSION: Post-operative administration of NSAIDs, but not TAP block, was associated with a decrease in post-operative opioid use and length of stay following surgery for ovarian cancer. Further investigation into alternative regional anesthetics for Enhanced Recovery after Surgery (ERAS) protocols is warranted.

Longitudinal Associations Between Bullying and Intimate Partner Violence Among Adolescents and Young Adults.

This study assessed longitudinal associations between bullying and intimate partner violence (IPV) among adolescents and young adults in a U.S.-based cohort study. Participants (N = 5,279) reported past-year bullying when they were 14-20 years old and reported lifetime experiences of IPV when they were 20-27 years old. The results indicate that participants reporting being bullied more than twice were at elevated risk of IPV victimization compared to participants reporting no bullying victimization, adjusting for bullying perpetration and covariates. Participants reporting bullying others once or more were at elevated risk of IPV perpetration compared to participants reporting no bullying perpetration, adjusting for bullying victimization and covariates. There was no evidence that the associations differed by gender. Results suggest that adolescents carry forward behaviors from their peer relationships to their dating relationships. Findings may have implications for school-based programs, which should explicitly integrate IPV prevention into bullying prevention efforts.

Parental history of moderate to severe infantile malnutrition is associated with cognitive deficits in their adult offspring.

OBJECTIVES: We compared the IQ and academic achievement of the young adult offspring of parents malnourished in infancy and those of a healthy control group in order to test the hypothesis that the offspring of previously malnourished individuals would show IQ and academic deficits that could be related to reduced parental socioeconomic status. METHODS: We conducted a group comparison study based on a community sample in Barbados (Barbados Nutrition Study). Participants were adult children ≥16 years of age whose parents had been malnourished during the first year of life (n = 64; Mean age 19.3 years; 42% male) or whose parents were healthy community controls (n = 50; Mean age 19.7 years; 48% male). The primary outcome was estimated IQ (Wechsler Abbreviated Scale of Intelligence); a secondary outcome was academic achievement (Wide Range Achievement Test - Third Edition). Data were analyzed using PROC MIXED with and without adjusting for parental socioeconomic status (Hollingshead Index of Social Position). RESULTS: IQ was reduced in the offspring of previously malnourished parents relative to the offspring of controls (9.8 point deficit; P < 0.01), but this difference was not explained by parental socioeconomic status or parental IQ. The magnitude of the group difference was smaller for basic academic skills and did not meet criteria for statistical significance. DISCUSSION: The deleterious impact of infant malnutrition on cognitive function may be transmitted to the next generation; however, this intergenerational effect does not appear to be explained by the reduced socioeconomic status or IQ of the parent generation.

Nabilone pharmacotherapy for cannabis dependence: A randomized, controlled pilot study.

BACKGROUND AND OBJECTIVES: We assessed the safety, tolerability, and preliminary efficacy of nabilone, a cannabinoid agonist, to treat cannabis dependence. METHODS: Eighteen adults with DSM-IV cannabis dependence were randomized to receive either 2 mg/day of nabilone (n = 10) or placebo (n = 8) for 10 weeks in addition to medication management. Twelve participants, six in each group, completed treatment. The safety and tolerability of nabilone was assessed at each visit. Any side effects from nabilone or the placebo were documented. Cannabis use outcomes were assessed via self-report of days of use and twice-weekly urine cannabinoid tests; secondary outcomes included cannabis craving and anxiety. RESULTS: We assessed safety and tolerability at each study visit. A total of eight adverse events, all mild or moderate, were reported in two participants in the nabilone group, and six events were reported in four participants in the placebo group during study treatment. A total of eight adverse events were reported in two participants in the nabilone group and six events were reported in four participants in the placebo group during study treatment. All reported adverse events were rated mild-to-moderate. There were no side effects deemed serious enough to be classified as an FDA-defined serious adverse event. In general, participants in both groups reported reduced cannabis use according to self-report over the course of the study, although these reductions were not statistically discernible. Moreover, there was no difference in cannabis use between the nabilone group and the placebo group as measured by self-report. DISCUSSION AND CONCLUSIONS: Nabilone pharmacotherapy was safe and well-tolerated in participants with cannabis dependence. Future studies might evaluate a higher dose of nabilone to determine its effects on cannabis use outcomes in participants with cannabis dependence. SCIENTIFIC SIGNIFICANCE: There remains a clear need for additional pharmacotherapy trials for cannabis dependence, and nabilone remains a candidate for such trials. (Am J Addict 2017;26:795-801).

Approximate median regression for complex survey data with skewed response.

The ready availability of public-use data from various large national complex surveys has immense potential for the assessment of population characteristics using regression models. Complex surveys can be used to identify risk factors for important diseases such as cancer. Existing statistical methods based on estimating equations and/or utilizing resampling methods are often not valid with survey data due to complex survey design features. That is, stratification, multistage sampling, and weighting. In this article, we accommodate these design features in the analysis of highly skewed response variables arising from large complex surveys. Specifically, we propose a double-transform-both-sides (DTBS)'based estimating equations approach to estimate the median regression parameters of the highly skewed response; the DTBS approach applies the same Box-Cox type transformation twice to both the outcome and regression function. The usual sandwich variance estimate can be used in our approach, whereas a resampling approach would be needed for a pseudo-likelihood based on minimizing absolute deviations (MAD). Furthermore, the approach is relatively robust to the true underlying distribution, and has much smaller mean square error than a MAD approach. The method is motivated by an analysis of laboratory data on urinary iodine (UI) concentration from the National Health and Nutrition Examination Survey.

Grit in patients with substance use disorders.

BACKGROUND AND OBJECTIVES: Grit is an emerging concept in positive psychology, defined as the ability to be persistent and focused in pursuit of long-term goals. This concept has received a great deal of interest recently because of its robust ability to predict success and well-being across a wide variety of domains. The study aim was to examine the clinical relevance of the construct of grit among patients with substance use disorders. METHODS: Inpatients on a detoxification unit were enrolled from September 2013 to August 2015 (N = 673). Psychometric properties of the Short Grit Scale (Grit-S) were reported. We then examined sociodemographic and clinical variables that might be associated with grit in this population. RESULTS: In this sample of patients with substance use disorders, the total Grit-S demonstrated strong psychometric properties. Grit-S scores were higher among older patients and those who were employed; scores were lower among those never married, diagnosed with a co-occurring psychiatric disorder, or who had used heroin during the past month, according to bivariate analyses. Grit-S scores remained associated with age, employment, and presence of a co-occurring psychiatric disorder in adjusted analysis. CONCLUSIONS AND SCIENTIFIC SIGNIFICANCE: This study provides initial support for the utility of the Grit-S among those with substance use disorders; this novel measure has not been previously reported in clinical populations. Research examining grit prospectively is needed to determine whether the links between grit and outcomes observed in other populations apply to patients with substance use disorders. (Am J Addict 2016;25:652-658).

Measuring affiliation in group therapy for substance use disorders in the Women's Recovery Group study: Does it matter whether the group is all-women or mixed-gender?

BACKGROUND AND OBJECTIVES: A Stage II, two-site randomized clinical trial compared the manualized, single-gender Women's Recovery Group (WRG) to mixed-gender group therapy (Group Drug Counseling; GDC) and demonstrated efficacy. Enhanced affiliation and support in the WRG is a hypothesized mechanism of efficacy. This study sought to extend results of the previous small Stage I trial that showed the rate of supportive affiliative statements occurred more frequently in WRG than GDC. METHODS: Participants (N = 158; 100 women, 58 men) were 18 years or older, substance dependent, and had used substances within the past 60 days. Women were randomized to WRG (n = 52) or GDC (n = 48). Group therapy videos were coded by two independent raters; Rater 1 coded 20% of videos (n = 74); Rater 2 coded 25% of videos coded by Rater 1 (n = 19). RESULTS: The number of affiliative statements made in WRG was 66% higher than in GDC. Three of eight affiliative statement categories occurred more frequently in WRG than GDC: supportive, shared experience, and strategy statements. DISCUSSION AND CONCLUSIONS: This larger Stage II trial provided a greater number of group therapy tapes available for analysis. Results extended our previous findings, demonstrating both greater frequency of all affiliative statements, as well as specific categories of statements, made in single-gender WRG than mixed-gender GDC. SCIENTIFIC SIGNIFICANCE: Greater frequency of affiliative statements among group members may be one mechanism of enhanced support and efficacy in women-only WRG compared with standard mixed-gender group therapy for substance use disorders. (Am J Addict 2016;25:573-580).

Almost efficient estimation of relative risk regression.

Relative risks (RRs) are often considered the preferred measures of association in prospective studies, especially when the binary outcome of interest is common. In particular, many researchers regard RRs to be more intuitively interpretable than odds ratios. Although RR regression is a special case of generalized linear models, specifically with a log link function for the binomial (or Bernoulli) outcome, the resulting log-binomial regression does not respect the natural parameter constraints. Because log-binomial regression does not ensure that predicted probabilities are mapped to the [0,1] range, maximum likelihood (ML) estimation is often subject to numerical instability that leads to convergence problems. To circumvent these problems, a number of alternative approaches for estimating RR regression parameters have been proposed. One approach that has been widely studied is the use of Poisson regression estimating equations. The estimating equations for Poisson regression yield consistent, albeit inefficient, estimators of the RR regression parameters. We consider the relative efficiency of the Poisson regression estimator and develop an alternative, almost efficient estimator for the RR regression parameters. The proposed method uses near-optimal weights based on a Maclaurin series (Taylor series expanded around zero) approximation to the true Bernoulli or binomial weight function. This yields an almost efficient estimator while avoiding convergence problems. We examine the asymptotic relative efficiency of the proposed estimator for an increase in the number of terms in the series. Using simulations, we demonstrate the potential for convergence problems with standard ML estimation of the log-binomial regression model and illustrate how this is overcome using the proposed estimator. We apply the proposed estimator to a study of predictors of pre-operative use of beta blockers among patients undergoing colorectal surgery after diagnosis of colon cancer.

Rates of psychotropic medication use reported by borderline patients and axis II comparison subjects over 16 years of prospective follow-up.

The purpose of this study was to assess the classes and types of psychotropic medication reported by borderline patients and axis II comparison subjects over 16 years of prospective follow-up. Medication use was assessed at baseline using a semistructured interview of proven reliability and validity as well as its follow-up analog at 8 contiguous 2-year follow-up periods. A significantly higher percentage of borderline patients than axis II comparison subjects reported taking an antidepressant, an anxiolytic, an antipsychotic, and a mood stabilizer over time. They also reported more commonly taking 7 of the 10 more specific types of medication studied (ie, all but tricyclic antidepressants, monoamine oxidase inhibitor antidepressants, and atypical antipsychotics). The rates over time of taking antipsychotics and mood stabilizers were stable, whereas there was a significant decline in the rates of antidepressants and anxiolytics from baseline to 8-year follow-up (but not from 8- to 16-year follow-up) reported by those in both study groups. In terms of specific medications, rates of atypical antidepressants and anticonvulsants were the most stable. In contrast, nonbenzodiazepine anxiolytics declined the most steadily over time, whereas rates of atypical antipsychotics increased significantly over the 16 years of prospective follow-up. Taken together, the results of this study suggest that a substantial percentage of borderline patients continue to use the major classes of medication over time. They also suggest that the declining rates of use tend to stabilize less than a decade after index admission.

Testing for independence in J×K contingency tables with complex sample survey data.

The test of independence of row and column variables in a (J×K) contingency table is a widely used statistical test in many areas of application. For complex survey samples, use of the standard Pearson chi-squared test is inappropriate due to correlation among units within the same cluster. Rao and Scott (1981, Journal of the American Statistical Association 76, 221-230) proposed an approach in which the standard Pearson chi-squared statistic is multiplied by a design effect to adjust for the complex survey design. Unfortunately, this test fails to exist when one of the observed cell counts equals zero. Even with the large samples typical of many complex surveys, zero cell counts can occur for rare events, small domains, or contingency tables with a large number of cells. Here, we propose Wald and score test statistics for independence based on weighted least squares estimating equations. In contrast to the Rao-Scott test statistic, the proposed Wald and score test statistics always exist. In simulations, the score test is found to perform best with respect to type I error. The proposed method is motivated by, and applied to, post surgical complications data from the United States' Nationwide Inpatient Sample (NIS) complex survey of hospitals in 2008.

Using the jackknife for estimation in log link Bernoulli regression models.

Bernoulli (or binomial) regression using a generalized linear model with a log link function, where the exponentiated regression parameters have interpretation as relative risks, is often more appropriate than logistic regression for prospective studies with common outcomes. In particular, many researchers regard relative risks to be more intuitively interpretable than odds ratios. However, for the log link, when the outcome is very prevalent, the likelihood may not have a unique maximum. To circumvent this problem, a 'COPY method' has been proposed, which is equivalent to creating for each subject an additional observation with the same covariates except the response variable has the outcome values interchanged (1's changed to 0's and 0's changed to 1's). The original response is given weight close to 1, while the new observation is given a positive weight close to 0; this approach always leads to convergence of the maximum likelihood algorithm, except for problems with convergence due to multicollinearity among covariates. Even though this method produces a unique maximum, when the outcome is very prevalent, and/or the sample size is relatively small, the COPY method can yield biased estimates. Here, we propose using the jackknife as a bias-reduction approach for the COPY method. The proposed method is motivated by a study of patients undergoing colorectal cancer surgery.

Health-related quality of life among prescription opioid-dependent patients: Results from a multi-site study.

BACKGROUND: Although prescription opioid use disorder has recently increased sharply in the United States, relatively little is known about the general well-being of this population. Assessment of quality of life in patients with substance use disorders has been recommended to improve clinical care. OBJECTIVES: Health-related quality of life was examined in prescription opioid-dependent patients at entry to a national multi-site clinical trial, to compare quality of life scores in the study sample to other populations; further, background variables associated with quality of life in the literature were examined. METHODS: Prescription opioid-dependent patients (N = 653) were compared to general populations on the Medical Outcome Study Short Form-36 (SF-36) quality of life measure; and the association between patient background variables and quality of life was examined. RESULTS: Compared to a general population, the current sample of prescription opioid-dependent patients had worse physical (-1.7 points, p < .001) and mental quality of life (-12.3 points, p < .001) as measured by the SF-36, similar to other opioid-use disorder populations. Within our sample, women showed more impairment than men in mental quality of life (-4.3 points, p < .001); older patients scored worse on physical (-5.2 points, p < .001), but not mental, quality of life. Chronic pain was associated with poorer physical quality of life (-9.0 points, p < .001). CONCLUSIONS AND SCIENTIFIC SIGNIFICANCE: The growing focus on wellness underscores the importance of measuring quality of life in addition to substance use outcomes. Routine assessment of health-related quality of life can add an important dimension to overall evaluation of patients' treatment response.

Improving the efficiency of drug use disorder screening in psychiatric settings: validation of a single-item screen.

BACKGROUND: Co-occurring drug use disorders are under-detected in psychiatrically ill populations highlighting the need for more efficient screening tools. OBJECTIVES: This study compares a single-item screening tool, previously validated in a primary care setting, to the 10-item Drug Abuse Screening Test (DAST-10) for identifying co-occurring drug use disorders among patients with severe psychiatric illness. METHODS: A total of 395 patients attending a psychiatric partial hospital program completed both the single-item screen and DAST-10. A subsample of consecutive patients (n = 67) was also administered the Structure Clinical Interview for DSM-IV (SCID-IV) as a diagnostic reference standard. RESULTS: Concordance between screening measures was moderate (κ = 0.4, p < 0.01). Sensitivity and specificity of the single-item screen and DAST-10 as compared to the SCID-IV were comparable, while area under the receiver operating curve showed better discriminatory power for the identification of drug use disorders with the single-item screen. CONCLUSIONS: In comparison to the DAST-10, the single-item screen appears to be a more efficient tool to identify co-occurring drug use disorders in a psychiatric treatment setting among patients with a range of psychiatric diagnoses.

Pro re nata (as needed) psychotropic medication use in patients with borderline personality disorder and subjects with other personality disorders over 14 years of prospective follow-up.

The use of pro re nata (PRN; as needed) psychotropic medication in patients with borderline personality disorder (BPD) has not been well characterized. This study had 3 purposes, which are as follows: (1) to describe the prevalence of PRN psychotropic medication use among patients with BPD and comparison subjects with other personality disorders (OPD) over 14 years of prospective follow-up, (2) to examine the rates reported by patients with BPD who ever recovered and never recovered, and (3) to examine the reasons for taking PRN medication reported by these patients. Overall, the prevalence of PRN psychotropic medication use was initially approximately 3 times higher among patients with BPD than comparison subjects with OPD, with a significant one-third decline in the use of PRN medication reported by patients with BPD over time. In analyses restricted to patients with BPD, patients with BPD who never recovered were approximately twice as likely to use PRN medication than patients with BPD who ever recovered over time. In reasons for use, the rates of PRN medication use to decrease agitation for both diagnostic groups declined significantly over time, whereas they remained significantly higher among patients with BPD. Likewise, patients with BPD who never recovered reported higher use of PRN medication to decrease agitation than patients with BPD who ever recovered over time. The results of this study indicate that PRN psychotropic medication is widely used for the treatment of patients with BPD, particularly those who have not achieved a recovery in both the symptomatic and psychosocial realms. They also suggest that patients with BPD use proportionally more PRN medication to decrease agitation than comparison subjects with OPD, with lower proportional use to reduce agitation found among recovered patients with BPD.