Background: Traumatic injury remains a major cause of mortality in the US. Older Americans experience lower rates of injury and higher rates of death at lower injury severity than their younger counterparts. The objectives of this study were to explore pre-injury factors and injury patterns that are associated with post-discharge mortality among injured elderly surviving index hospitalization. Methods: We queried a 5% random sample of Medicare beneficiaries (n=2,002,420) for any hospitalization with a primary ICD-9 diagnosis code for injury. Patients admitted without urgent/emergent admission were excluded, as well as patients presenting from inpatient hospitalization or rehabilitation. The primary endpoint was all-cause mortality. Patients were categorized into three mortality groups: death within 0-30 days, 31-90 days, or 91- 365 days post-discharge from the index hospitalization. These groups were compared with those who survived greater than one year post-discharge. Univariate tests of association and multivariable logistic regression models were utilized to identify factors associated with mortality during the 3 examined periods. Results: 83,439 elderly patients (4.2%) were admitted with new injuries. 63,628 met inclusion criteria. 1,936 patients (3.0%) died during their index hospitalization, 2,410 (3.8%) died within 0-30 days, 3,084 (4.8%) died within 31-90 days, and 5,718 (9.0%) died within 91- 365 days after discharge. In multivariable adjusted models, advanced age, male sex, and higher Elixhauser score were associated with post-discharge mortality. The presence of critical injury had the greatest effect on mortality early after injury (0-30 days, OR 1.81, CI 1.64-2.00). Discharge to anywhere other than home without services was associated with an increased odds of dying. Conclusions: Socio-demographic characteristics, disposition, and co-morbid factors were the strongest predictors of post-discharge mortality. Efforts to reduce injury-related mortality should focus on injury prevention and modification of co-morbidities.

Background: Guideline-concordant therapies for survivors of an acute coronary syndrome (ACS) hospitalization require healthcare access, something that millions of Americans lack. Methods and Results: Using data from a prospective cohort study of over 2,000 survivors of a hospitalization for an ACS in central Massachusetts and Georgia from 2011 to 2013, the first two aims of this thesis sought to identify the post-discharge consequences for survival and health status of having: 1) financial barriers to healthcare, 2) no usual source of care, and 3) transportation barriers. We found that patients lacking a usual source of care and having a transportation barrier were more likely to have died within two years following hospital discharge compared to those without such barriers. Also, patients with financial barriers to healthcare were more likely to experience clinically meaningful declines in physical and mental health-related quality of life over the six months after hospital discharge. The third aim sought to better understand factors influencing the success of care transitions home after an unplanned hospitalization through a qualitative study of 22 patients. Participants described how adequate healthcare access, particularly having insurance and transportation to clinical appointments, facilitated the receipt of follow-up care and adherence to treatments. Conclusions: Limitations in healthcare access may contribute to poorer survival, health-related quality of life, and survival. Additional research is needed to identify interventions to improve healthcare access and test whether improved access leads to better patient outcomes.

Depression is a common comorbidity in rheumatoid arthritis (RA), yet it may not be adequately recognized during routine clinical care. RA symptoms may confer a risk for depression, and vice versa; depression may affect RA disease activity and response to treatment. The study aims were to compare patient- and physician-reported depression measures, evaluate the temporal bi-directional association between RA disease activity and depressive symptomology, and assess depression as a moderator of RA treatment. Patients were identified using a national RA registry sample (Consortium of Rheumatology Researchers of North America; CORRONA). Depression prevalence and incidence rates were estimated, and concordance and disagreement using measures reported separately by patients and physicians, as well as baseline cross-sectional associations between RA disease and a history of depression. A survival analysis was conducted to temporally predict the incident onset of self-reported depressive symptoms using the different metrics of RA disease activity. Also, mixed effects models were used to assess prospective changes in RA disease activity by prevalent and incident depressive symptom status. Lastly, logistic regression models compared the likelihood of clinical response to RA treatment during follow-up in those with and without depression when starting biologic disease modifying anti-rheumatic drug (DMARD) therapy. Patient-reported depression rates were much higher and significantly different from physician based comorbidity estimates. Patient and physician RA disease activity measures were associated with an increased risk for depression onset, but not laboratory-reported serum biomarkers. Similarly, depression was temporally associated with significantly slower rates of decline regarding every patient-reported disease activity measure, some physician-reported metrics, but not acute phase reactants. Moreover, there was a significantly lower probability of achieving clinical remission among those with depression on a biologic DMARD after 6 months and an analogous effect at 12-months that was slightly lower in magnitude, which did not reach statistical significance. Rheumatologists under-reported the occurrence of prevalent and incident depressive symptoms, and thus are likely unaware of its presence in their RA patients. Further, the results suggest the bi-directional effects between these conditions are related to the cognitive and behavioral aspects of depression and their interactions with disease activity, rather than shared immunological mechanisms in the context of cell-mediated immunity. When also considering the impact on clinical response to biologic DMARDS, the findings collectively imply that rheumatologists must address any challenges due to depression to provide the best care to their patients.

Gestational diabetes mellitus (GDM), a pregnancy-induced metabolic disorder that affects 2-10% of pregnancies poses future risk for diabetes mellitus (DM) and cardiovascular disease in mother and child. However, few prospective studies have examined the effect of GDM on altered maternal and cord blood lipids, specifically HDL, LDL, triglycerides, and total cholesterol, both during and after pregnancy. We have evaluated the association between GDM and lipid metabolism in pregnant mothers and their infants using data from a prospective cohort study conducted at Baystate Medical Center’s Wesson Women and Infant’s Unit. GDM was assessed prenatally by 3-hr GTT blood samples and was confirmed by obstetrician review. Lipids were assessed via fasting and non-fasting blood samples obtained during 3-hr GTTs performed at 24-28 weeks of gestation and 6-8 weeks post-partum. Data for covariates were collected via an interview form administered at the time of recruitment. We used multivariable linear regression to evaluate the association between GDM status and maternal lipids during and after pregnancy as well as cord lipids. These study results inform future research on GDM as a risk factor for future metabolic disorders in mother and child.

The condition of multimorbidity — the presence of two or more medical conditions in an individual — is a growing phenomenon worldwide. In the United States, multimorbid patients represent more than a third of the population and the trend is steadily increasing in an already aging population. There is thus a pressing need to understand the patterns in which multimorbidity occurs, and to better understand the nature of the care that is required to be provided to such patients. In this thesis, we use data from the Medical Expenditure Panel Survey (MEPS) from the years 2011 to 2015 to identify combinations of multiple chronic conditions (MCCs). We first quantify the significant heterogeneity observed in these combinations and how often they are observed across the five years. Next, using two criteria associated with each combination -- (a) the annual prevalence and (b) the annual median expenditure -- along with the concept of non-dominated Pareto fronts, we determine the degree of impact each combination has on the healthcare system. Our analysis reveals that combinations of four or more conditions are often mixtures of diseases that belong to different clinically meaningful groupings such as the metabolic disorders (diabetes, hypertension, hyperlipidemia); musculoskeletal conditions (osteoarthritis, spondylosis, back problems etc.); respiratory disorders (asthma, COPD etc.); heart conditions (atherosclerosis, myocardial infarction); and mental health conditions (anxiety disorders, depression etc.). Next, we use unsupervised learning techniques such as association rule mining and hierarchical clustering to visually explore the strength of the relationships/associations between different conditions and condition groupings. This interactive framework allows epidemiologists and clinicians (in particular primary care physicians) to have a systematic approach to understand the relationships between conditions and build a strategy with regards to screening, diagnosis and treatment over a longer term, especially for individuals at risk for more complications. The findings from this study aim to create a foundation for future work where a more holistic view of multimorbidity is possible.

<p><strong>Background and objectives</strong></p> <p>We investigate three issues related to the adjustment for baseline covariates in late phase clinical trials: (1) the analysis of correlated outcomes in multicentre RCTs, (2) the assessment of the probability and implication of prognostic imbalance in RCTs, and (3) the adjustment for baseline confounding in cohort studies.</p> <p><strong>Methods</strong></p> <p>Project 1: We investigated the properties of six statistical methods for analyzing continuous outcomes in multicentre randomized controlled trials (RCTs) where within-centre clustering was possible. We simulated studies over various intraclass correlation (ICC) values with several centre combinations.</p> <p>Project 2: We simulated data from RCTs evaluating a binary outcome by varying risk of the outcome, effect of the treatment, power and prevalence of a binary prognostic factor (PF), and sample size. We compared logistic regression models with and without adjustment for the PF, in terms of bias, standard error, coverage of confidence interval, and statistical power. A tool to assess sample size requirement to control for chance imbalance was proposed.</p> <p>Project 3: We conducted a prospective cohort study to evaluate the effect of tuberculosis (TB) at the initiation of antiretroviral therapy (ART) on all cause mortality using Cox proportional hazard model on propensity score (PS) matched patients to control for potential confounding. We assessed the robustness of results using sensitivity analyses.</p> <p><strong>Results and conclusions</strong></p> <p>Project 1: All six methods produce unbiased estimates of treatment effect in multicentre trials. Adjusting for centre as a random intercept leads to the most efficient treatment effect estimation, and hence should be used in the presence of clustering.</p> <p>Project 2: The probability of prognostic imbalance in small trials can be substantial. Covariate adjustment improves estimation accuracy and statistical power, and hence should be performed when strong PFs are observed.</p> <p>Project 3: After controlling for the important confounding variables, HIV patients who had TB at the initiation of ART have a moderate increase in the risk of overall mortality.</p> / Doctor of Philosophy (PhD)

Background: Sperm contain highly compact nuclei, inhibiting DNA extraction using traditional techniques. Current methods extracting sperm DNA involve lengthy lysis and no means of stabilizing DNA, hindering clinical research. Objective: We sought to optimize an efficient method of extracting high quality human sperm DNA. Methods: Sperm from three volunteers were isolated using PureCeption. We tested 1) proteinase K with DNA/RNA Shield, 2) DTT and TCEP as reducing agents, 3) QIAshredder homogenization, and 4) stability of sperm DNA fresh (baseline) or after 4 weeks of storage at 4OC in DNA/RNA Shield using modified Quick-gDNA MiniPrep. DNA was PCR amplified using ALU primers and digested with Hinf1 restriction enzymes. DNA methylation was measured by MassARRAY. Results: DNA concentrations were similar with (30.1+0.28ng/μL, 33.4+0.21ng/μL) and without (28.9+0.00ng/μL, 30.9+0.85ng/μL) proteinase K. Sperm cells were lysed after 1 and 20 minutes with 25mM TCEP and 100mM DTT respectively. TCEP (50mM) produced greater DNA concentrations (17.2+0.50ng/μL, 21.3+0.71ng/μL) than 50mM DTT (12.6+0.28ng/μL, 12.3+0.35ng/μL). Adding QIAshredder to 50mM TCEP increased DNA concentrations (25.9+0.35ng/μL, 21.7+0.49ng/μL versus 18.6+0.99ng/μL, 12.3+0.35ng/μL). At baseline and 4 weeks: 1) DNA concentrations were similar (36.2+2.75 ng/μL, 32.2+1.38ng/μL, 44.3+3.93ng/μL versus 40.0+2.98ng/μL, 37.6+1.38ng/μL, 38.7+3.93ng/μL respectively) and 2) DNA was equally amplified by PCR and digested with restriction enzymes. DNA methylation was similar at baseline and 4 weeks for SNURF (1.43+1.02%, 1.55+0.95%), PEG10 (3.69+0.66%, 4.28+1.52%), and H19 (88.93+3.24%, 91.78+2.00%). Conclusions: We stabilized and isolated high quality DNA from human sperm using 5 minute versus > 2 hour lysis in other methods. Our methods may facilitate efficient clinical research.

Evidenzbasierte Medizin (EbM) versteht sich als Disziplin, die zwischen der klinischen Forschung und der Gesundheitsversorgung eine Brücke schlägt. Die vorliegende Habilitationsschrift behandelt in 4 Einzelprojekten das Thema "Die klinische Epidemiologie in der ärztlichen Entscheidungsfindung". In dem ersten Projekt "Beobachtung oder Experiment" haben wir nachgewiesen, dass die Randomisierung für den Wirksamkeitsnachweis klinischer Interventionen unverzichtbar ist, um eine möglichst unverzerrte Effektmessung sicherzustellen. Die Studienpopulation waren systematische Übersichtsarbeiten, die randomisierte und nicht-randomisierte Studien zu einem breiten Spektrum an Interventionen aus dem Gesundheitsbereich einschlossen und mehr als 3000 Primärstudien umfassten. In der Auswertung konnten wir zeigen, dass - verglichen mit einer randomisierten Patientenzuordnung - bei einer nicht-randomisierten Patientenzuordnung in Studien der Effekt der Intervention häufig überschätzt wird, aber auch unterschätzt oder sogar invers geschätzt werden kann. Allerdings ist es auch möglich, dass vergleichbare Effekte beobachtet werden. Aufgrund der grossen Gefahr für Effektverzerrung in nicht vorhersagbarer Richtung ist die Randomisierung bei Interventionsstudien absolut erforderlich, um Gruppen mit vergleichbaren Ausgangskriterien zu generieren und damit eine möglichst biasfreie Effektschätzung sicherzustellen. In dem zweiten Projekt haben wir in einer Simulationsstudie demonstriert, wie es durch systematische Fehler in der Durchführung klinischer Studien zu klinischen Fehlentscheidungen kommen kann. Mit Hilfe klinischer Daten von Intensivpatienten, unterschiedlichen Annahmen über das Ausgangsrisiko für gastrointestinale Blutungen und der relativen Risikoreduktion für gastrointestinale Blutung durch H2-Blocker wurden typische Risikokonstellationen identifiziert: Klinische Situationen mit einem moderaten bis niedrigen Patienten-Grundrisiko und moderater bis geringer Wirksamkeit der medizinischen Maßnahme waren für Fehlentscheidungen infolge verzerrter Studienergebnisse besonders anfällig. Diese Konstellation kommt in der Patientenversorgung häufig vor, wodurch die Erkenntnisse unserer Studie einen ganz konkreten Praxisbezug erhalten. In dem dritten Projekt, einer klinischen Studie über die Wirksamkeit von ärztlichen Fortbildungen in evidenzbasierter Medizin ("Berliner EbM-Studie"), haben wir auf der Grundlage des Berliner Gegenstandskatalogs EbM ein Instrument (2 Fragebögen à 15 Fragen) entwickelt und validiert, mit dem man zuverlässig und reproduzierbar zwischen unterschiedlichen Kenntnissen und Fertigkeiten von EbM differenzieren kann. In einer dreijährigen Studie konnten wir nachweisen, dass durch kurze intensive Kurse in evidenzbasierter Medizin (wie z.B. den Berliner EbM-Kursen) bei den 203 Teilnehmern ein statistisch signifikanter und klinisch relevanter Wissenszuwachs erzielt werden kann (vor dem Kurs 6,3 + 2,9, nach dem Kurs 9,9 + 2,4 richtige Antworten; p< 0.001). Im vierten Projekt ging es um die Implementierung der evidenzbasierten Medizin in die tägliche Praxis von Hausärzten. In einer clusterrandomisierten Studie unter Hausärzten hatten wir untersucht, ob man durch kurze, evidenzbasierte Erläuterungen zu im Krankenhaus neu angesetzten Behandlungen, die im Entlassungsbrief beigefügt werden, Hausärzte motivieren kann, diese Behandlung fortzusetzen. 178 Praxen nahmen an der Studie teil. Dabei wurden 417 Entlassungsbriefe mit insgesamt 59 unterschiedlichen evidenzbasierten Medikamentenempfehlungen versandt und nach 3-4 Monaten 268 Interviews erfolgreich durchgeführt. Ärzte in der Interventionsgruppe hatten eine statistisch signifikant geringere Wahrscheinlichkeit, von den Krankenhausempfehlungen abzuweichen als Ärzte in der Kontrollgruppe, die nur den üblichen Entlassungsbrief enthielten (absolute Risikoreduktion 12,5%; p=0.039). Die Ärzte waren über die zusätzliche Information sehr zufrieden, auch wenn diese Information i.a. keine neuen Erkenntnisse lieferte, vielmehr den gegenwärtigen Kenntnisstand der Ärzte bestätigte. Kurze evidenzbasierte Medikamenteninformationen können das rationale Verschreibungsverhalten von Hausärzten positiv beeinflussen. / Evidence-based medicine (ebm) can be described as the discipline bridging research and health care. This thesis covers 4 individual projects on the role of clinical epidemiology / evidence-based medicine in rational clinical decision-making. The first study "Observation or Experiment" addressed the methodological issue of the impact of observational studies versus randomised allocation to any intervention on the estimated effect of the intervention. The study population were systematic reviews including randomised and non-randomised studies on a broad spectrum of interventions and comprising more than 3000 primary studies. In the empirical assessment, we could demonstrate that lack of randomisation tended to exaggerate the estimated effect of the intervention, but could also result in underestimation of the effect, in similar effect sizes or even in inverse effects. Therefore randomisation is mandatory in intervention studies to generate comparable baseline groups and thereby ensure an unbiased assessment of the underlying treatment effect. The second project was a simulation study investigating the impact of bias on clinical decision-making. Based on empirical data from ITU-patients, various assumptions on baseline risks of gastrointestinal bleeding and an estimate of the relative risk reduction for bleeding by H2-antagonists from a recent meta-analysis we investigated the potential for erroneous clinical decisions induced by systematic errors in the performance of clinical trials. We could demonstrate that certain clinical situations are particularly susceptible for errors in decision-making, in particular, if a patient’s baseline risk for an adverse event or the effectiveness of the intervention is only moderate or small. As low baseline risk and / or moderate treatment effects tend to occur frequently in physician-patient-encounters, physicians need to be aware of the increased risk for errors and pay meticulous attention on a reliable evidence base. The third project was a clinical trial on the effectiveness of teaching evidence-based medicine to physicians. The trial comprises two phases: Instrument development and performance of the trial. Starting from a comprehensive curriculum of evidence-based-medicine we identified relevant core items of ebm. Based on this curriculum we developed and validated a before-after instrument with 15 questions each that was able to distinguish varying degrees of knowledge and skills of ebm. Over a period of 3 years, the instrument was randomly administered to 203 participants in consecutive ebm-courses. Prior to the course, the participants scored a mean of 6,3 + 2,9, after the course the number of correct answers increased to 9,9 + 2,4 (p< 0.001). The scores of ebm-experts or ebm-naïve controls was significantly higher resp. lower. Thereby we could demonstrate that short intensive courses in evidence-based medicine (such as the Berlin courses) can lead to a significant and clinically meaningful increase in knowledge and skills. The forth project focussed on the implementation of evidence-based medicine in the day-to-day practice of family doctors. In a cluster-randomised study we investigated whether short evidence-based drug information in hospital discharge letters can influence the prescribing behaviour of general practitioners. 178 practices participated in the study, 417 discharge letters with 59 different evidence-based drug information were sent out and 268 interviews were successfully performed after an interval of 3 - 4 months. Physicians in the intervention group were statistically less likely to depart from the hospital recommendations than physicians in the control group who only received a regular discharge letter group (absolute risk reduction 12,5%; p=0.039). Furthermore, physicians were very satisfied with this additional piece of information, which tended to assure their own knowledge and reminded them to apply it in this individual patient (rather than providing new information). It was concluded that short evidence-based information in discharge letters could have a positive impact on a rational prescribing behaviour of physicians.

Nurses are frequently seen in public in their “scrubs,” which could mean that contaminated clothing is being brought into the community, thereby posing an infection risk. The purpose of this study is to investigate if and which contaminants are present on the fabrics and the actions nurses are taking to eliminate contamination risks. Eleven scrub tops were worn on hospital units over one twelve-hour shift. The contaminated scrubs and three control tops were then swabbed and used to inoculate agar plates. After incubation, colonies were counted, streaked onto nutrient and Mannitol-salt agar for isolation, and incubated. Using API Staph strips and Gram staining, the bacteria were identified. The nurses also completed a short survey on laundering and scrub care. All scrub tops, except the controls, were contaminated with multiple species of bacteria including Staphylococcus species. Responses to the survey showed that no two nurses washed their scrubs in the same manner and many wear them in public. The results determined that bacteria can survive on clothing and pose the possibility of transmission throughout the hospital and public venues. The survey results indicate a need for employer laundering policies, public awareness of the risk for transmission of disease from contaminated clothing, and stricter regulations about employees wearing scrubs outside of health care facilities.

Objectives To examine the relationship between Concussion Symptom Clusters (CSCs) and return-to-play time using a representative sample of U.S. college athletes with sports-related concussions. Background Recent evidence regarding concussion symptoms have been observed to be an important element of concussion severity, and potentially a predictor of return-to-play time. However, there is a paucity of data examining the associations between Concussion Symptom Clusters (CSCs) and return-to-play time in the U.S. college athlete population. Methods Data from the 2009-2010 to 2013-2014 academic years (n=1670) were obtained from the Datalys Center for Sports Injury and Prevention Inc. database. Exploratory factor analytic methods were applied, and the resulting factors were used in multinomial regression modeling to identify associations between CSCs and return-to-play time. ResultsA 4-factor solution accounted for 48.8% of the variance and included: audio-vestibular, somatic, amnesic, and affective factor structure. Audio-vestibular symptoms were associated with increased odds of prevented participation at 7-13 days, 14-29 days, greater than 30 days, and out for remainder of season, respectively (p Conclusion Specific CSCs were significantly associated with return-to-play time in college athletes, (p<0.05).