Can artificial intelligence separate the wheat from the chaff in systematic reviews of health economic articles?

OBJECTIVES: Artificial intelligence-powered tools, such as ASReview, could reduce the burden of title and abstract screening. This study aimed to assess the accuracy and efficiency of using ASReview in a health economic context. METHODS: A sample from a previous systematic literature review containing 4,994 articles was used. Previous manual screening resulted in 134 articles included for full-text screening (FT) and 50 for data extraction (DE). Here, accuracy and efficiency was evaluated by comparing the number of identified relevant articles with ASReview versus manual screening. Pre-defined stopping rules using sampling criteria and heuristic criteria were tested. Robustness of the AI-tool's performance was determined using 1,000 simulations. RESULTS: Considering included stopping rules, median accuracy for FT articles remained below 85%, but reached 100% for DE articles. To identify all relevant articles, a median of 89.9% of FT articles needed to be screened, compared to 7.7% for DE articles. Potential time savings between 49 and 59 hours could be achieved, depending on the stopping rule. CONCLUSIONS: In our case study, all DE articles were identified after screening 7.7% of the sample, allowing for substantial time savings. ASReview likely has the potential to substantially reduce screening time in systematic reviews of health economic articles.

All circulating EpCAM+CK+CD45- objects predict overall survival in castration-resistant prostate cancer.

BACKGROUND: Presence of five or more circulating tumor cells (CTC) in patients with metastatic carcinomas is associated with poor survival. Although many objects positive for epithelial cell adhesion molecules and cytokeratin (EpCAM+CK+) are not counted as CTC, they may be an important predictor for survival. We evaluated the association between these objects and survival in patients with prostate cancer. PATIENTS AND METHODS: Included in this follow-up study were 179 patients with castration-resistant prostate cancer. CellSearch was used to isolate EpCAM+ objects and to stain DNA, cytokeratin and CD45. All EpCAM+CK+ objects were subdivided into seven classes on the basis of predefined morphological appearance in 63 independent samples. Association of each class with survival was studied using Kaplan-Meier and Cox regression analyses. RESULTS: Each EpCAM+CK+CD45- class showed a strong association with overall survival (P < 0.001). This included small tumor microparticles (S-TMP), which did not require a nucleus and thus are unable to metastasize. A higher number of objects in any class was associated with decreased survival. A good prediction model included large tumor cell fragments (L-TCF), age, hemoglobin and lactate dehydrogenase. Models with S-TMP or CTC instead of L-TCF performed similarly. CONCLUSION: EpCAM+CK+CD45- that do not meet strict definitions for CTC are strong prognostic markers for survival.

Experience with multiple control groups in a large population-based case-control study on genetic and environmental risk factors.

We discuss the analytic and practical considerations in a large case-control study that had two control groups; the first control group consisting of partners of patients and the second obtained by random digit dialling (RDD). As an example of the evaluation of a general lifestyle factor, we present body mass index (BMI). Both control groups had lower BMIs than the patients. The distribution in the partner controls was closer to that of the patients, likely due to similar lifestyles. A statistical approach was used to pool the results of both analyses, wherein partners were analyzed with a matched analysis, while RDDs were analyzed without matching. Even with a matched analysis, the odds ratio with partner controls remained closer to unity than with RDD controls, which is probably due to unmeasured confounders in the comparison with the random controls as well as intermediary factors. However, when studying injuries as a risk factor, the odds ratio remained higher with partner control subjects than with RRD control subjects, even after taking the matching into account. Finally we used factor V Leiden as an example of a genetic risk factor. The frequencies of factor V Leiden were identical in both control groups, indicating that for the analyses of this genetic risk factor the two control groups could be combined in a single unmatched analysis. In conclusion, the effect measures with the two control groups were in the same direction, and of the same order of magnitude. Moreover, it was not always the same control group that produced the higher or lower estimates, and a matched analysis did not remedy the differences. Our experience with the intricacies of dealing with two control groups may be useful to others when thinking about an optimal research design or the best statistical approach.

Recurrent thrombosis and survival after a first venous thrombosis of the upper extremity.

BACKGROUND: Little is known about the consequences of a first venous thrombosis in the upper extremity. We studied the incidence of, survival, and risk factors for recurrence in a follow-up study. METHODS AND RESULTS: We followed up 224 patients 18 to 70 years of age after a first venous thrombosis of the arm. Information was collected through anticoagulation clinics, the national death registry, discharge letters, and questionnaires. The median follow-up was 3 years, during which time 30 patients experienced a recurrent event, yielding an incidence rate of 43.2 per 1000 person-years. Survival was reduced: 55 of 224 patients died, which was 5.4-fold higher than age- and sex-adjusted population rates (standardized mortality ratio, 5.4; 95% CI, 4.2 to 7.0). The risk of recurrence was 2-fold higher in women than in men (hazard ratio, 1.8; 95% CI, 0.9 to 3.9). A central venous catheter at the time of first thrombosis was associated with a reduced risk of recurrence. A body mass index > or =25 kg/m(2) and a first nonsubclavian thrombosis appeared to increase the risk of a recurrent event. Prothrombotic mutation carriers did not appear to have an increased recurrence risk. CONCLUSIONS: The risk of recurrence was high, with women, patients with body mass index > or =25 kg/m(2), and patients with a first nonsubclavian vein thrombosis having a higher risk of recurrence. Patients with a first venous thrombosis of the arm have a poor vital prognosis.

Donor profiles: demographic factors and their influence on the donor career.

BACKGROUND AND OBJECTIVES: Studying the contribution of demographic factors to the donor career provides important knowledge to be used for donor management. The aim of this study is to gain insight into donor characteristics, more specifically into the demographic profile of active vs. resigned donors, and multi-gallon vs. occasional donors. MATERIALS AND METHODS: The study population consisted of all registered Dutch whole-blood donors between 1 January 2004 and 1 January 2005 (N = 370 470). The effect of several blood donor characteristics and demographic variables on (i) resigning donating and (ii) being a multi-gallon donor were assessed. Blood donor characteristics were extracted from the blood bank information system and included age, sex, blood group, number of donations and invitations. Demographic characteristics were constituted by population data on urbanization level, socio-economic status (income, housing value), and ethnicity. RESULTS: Men clearly resigned less often than women (odds ratio (OR) 0.73, 95% confidence interval (CI) 0.72-0.75). Being older than 24 years, having a high income, a high-priced house, living in less urbanized areas or areas with relatively few ethnically diverse people also reduced the stopping risk. With respect to multi-gallon donorship, men were five times more often multi-gallon donor than women (OR 5.27, 95% CI 5.15-5.39) irrespective of the number of donation invitations. Furthermore, multi-gallon donors appeared to live in urbanized areas and have a higher income than occasional donors. CONCLUSION: Our results show that different donor profiles can be distinguished. Differences between active and resigned donors include age, the number of donations, sex, socio-economic-status, ethnicity, and urbanization level. The factors highly associated with being a multi-gallon donor are sex, age, socio-economic status, and to a lesser extent urbanization level. Donor profiles do provide the blood bank with knowledge on their donor population, which may be used as valuable information for donor recruitment and retention policies.

Mechanisms of the factor V Leiden paradox.

OBJECTIVE: Carriers of the factor V Leiden mutation (FVL-carriers) have a substantially increased risk of deep venous thrombosis (DVT), whereas the risk of pulmonary embolism (PE) is only mildly increased compared with noncarriers. So far few studies have investigated possible mechanisms for this so-called FVL paradox. METHODS AND RESULTS: Consecutive patients with a first DVT or PE were included in a large population-based case-control study (MEGA study). Patients, aged 18 to 70 years, provided a questionnaire, DNA (n=3313), or plasma (n=1474). Surgery, injury, and travel were considered thrombosis-provocative. Of 2063 patients with isolated DVT, 20% were FVL-carrier, as were 8% of the 885 patients with isolated PE. Among DVT patients, FVL-carriers had their thrombi more often proximal and a higher number of affected veins than noncarriers. No differences were observed between FVL-carriers and noncarriers in time between provocation and diagnosis, in vitro coagulation time, and thrombus density. Compared with patients with both DVT and PE, isolated DVT patients more often had thrombi located distally and had a similar number of affected veins. Compared with isolated PE patients, isolated DVT patients had a similar time between provocation and diagnosis, and similar in vitro coagulation time and thrombus density. CONCLUSIONS: Although some effects were differential for FVL-carriers and noncarriers, and some were differential for PE and DVT patients, none of the potential mechanisms offered a clear explanation.

Venous thrombosis in the elderly.

While the overall incidence of venous thrombosis is 1-2 per 1000 per year, it is close to 1% per year in the very old. The case-fatality rate of thrombosis is high in the elderly, particularly among those with cancer. The risk of major hemorrhage during anticoagulant treatment is also strongly age-dependent, contributing to the vulnerability of the old patient with thrombosis. From this perspective it is surprising that far fewer studies into the etiology and treatment of venous thrombosis have focused on the elderly than on young and middle-aged patients. In this review we discuss that, while environmental risk factors, such as immobilization and cancer, are important causes of thrombosis in the elderly, abnormalities of the coagulation system are equally, or even more, important than in young individuals. In addition to a review of the literature, new data are presented from the MEGA-study. Thrombosis in the elderly should be a focus of future studies.

Regular sports activities decrease the risk of venous thrombosis.

BACKGROUND: Stasis of the blood has been postulated as a major cause of venous thrombosis. However, little is known about the effect of stimulating the blood flow in order to prevent venous thrombosis through, for example, sports activities. OBJECTIVES: In a large population-based case-control study (MEGA study), we studied whether participating in sports activities on a regular basis was associated with venous thrombosis risk. PATIENTS/METHODS: Consecutive patients with a first venous thrombosis of the leg or a pulmonary embolism, and control subjects, consisting of partners of the patients and randomly selected control subjects from the general population, were asked to participate. Sports activities and other risk factors for venous thrombosis were reported in a standardized mailed questionnaire. Participants with malignancy were excluded. RESULTS: Out of 3608 patients, 1136 (31.5%), and of our 4252 control subjects 1686 (39.7%), participated in sports activities. Participating in sports activities reduced the risk of venous thrombosis compared with not participating in sports activities [odds ratio (OR) 0.64; 95% CI 0.58-0.71]. Risk reductions were similar after adjustment for sex, age and body mass index (OR(adj) 0.71; 95% CI 0.64-0.78) and when the analysis was restricted to healthy individuals (OR(adj) 0.67; 95% CI 0.58-0.78). No differences in risk were found for various frequencies, intensities and types of sport. CONCLUSION: Regular sports activities reduce the risk of venous thrombosis.

The association of prothrombin A19911G polymorphism with plasma prothrombin activity and venous thrombosis: results of the MEGA study, a large population-based case-control study.

BACKGROUND: Prothrombin (FII) G20210A mutation and elevated plasma prothrombin activity are known risk factors for venous thrombosis. The risk of venous thrombosis among 19911G carriers of the prothrombin A19911G polymorphism has not been extensively investigated. OBJECTIVES AND METHODS: We assessed prothrombin activity, FIIG20210A, and FIIA19911G polymorphisms in a large population-based case-control study, the Multiple Environmental and Genetic Assessment (MEGA) study of risk factors for venous thrombosis. Four thousand three hundred and sixty-five consecutive patients with a first episode of deep vein thrombosis of the leg or pulmonary embolism were included. The control group (n = 4779) consisted of partners of patients or persons gathered using a random-digit dialing method. We studied the effect of FIIA19911G polymorphism on prothrombin activity and thrombosis risk, also in combination with factor V Leiden. RESULTS: Among FII20210-GG control subjects, FII19911-GG carriers had 7.1% [95% confidence interval (CI): 5.7-8.5] higher mean prothrombin activity than FII19911-AA carriers and the risk for GG carriers was 1.43-fold increased compared to AA carriers [odds ratio (OR) 1.43; 95% CI: 1.27-1.61]. Among FII20210-GA control carriers, the mean prothrombin activity in both FII19911-AA and -AG carriers was nearly equivalent [131.7% and 133.4%; mean difference (95% CI) = 1.7% (-7.2-10.7)]. Because of genetic linkage, FII19911-GG carriers were very rare on a FII20210-GA background, as only one FII20210A carrier had FII19911-GG. In FII20210-GA carriers, the OR increased from 3.05 (95% CI: 2.17-4.27) in subjects with FII19911-AA to 3.33 (2.28-4.85) in subjects with FII19911-AG, compared to those with FII20210-GG and FII19911-AA. CONCLUSIONS: The FIIA19911G polymorphism is associated with mildly elevated prothrombin activity and is a risk factor for venous thrombosis.

Incidence and prognostic factors of chronic pain after isolated musculoskeletal extremity injury.

BACKGROUND: Chronic pain in patients is usually related to an episode of pain following acute injury, emphasizing the need to prevent progression from acute to chronic pain. Multiple factors in the acute phase might be responsible for perpetuating the pain. The presentation of patients at the emergency department (ED) presents a prime opportunity to identify patients at high risk for chronic pain and to start appropriate treatment. METHODS: The PROTACT study is a prospective follow-up study aiming to estimate the incidence and prognostic factors responsible for the development of chronic pain after musculoskeletal injury. Data including sociodemographic, pain, clinical, injury- or treatment-related and psychological factors of 435 patients were collected from registries and questionnaires at ED visit, 6-week, 3- and 6-month follow-up. RESULTS: At 6 months post-injury, 43.9% of the patients had some degree of pain (Numeric Rating Scale (NRS) ≥1) and 10.1% had chronic pain (NRS ≥4). Patients aged over 40 years, in poor physical health, with pre-injury chronic pain, pain catastrophizing, high urgency level and severe pain at discharge were found to be at high risk for chronic pain. CONCLUSIONS: Two prognostic factors, severe pain at discharge and pain catastrophizing, are potentially modifiable. The implementation of a pain protocol in the ED and the use of cognitive-behavioural techniques involving reducing catastrophizing might be useful.

Old and new risk factors for upper extremity deep venous thrombosis.

BACKGROUND: Well known risk factors for upper extremity deep venous thrombosis are the presence of a central venous catheter (CVC) and malignancy, but other potential risk factors, such as surgery, injury and hormone replacement therapy (HRT), have not yet been explored. METHODS: We performed a population-based case-control study including 179 consecutive patients, aged 18-70 years with upper extremity deep venous thrombosis and 2399 control subjects. Participants reported on acquired risk factors in a questionnaire and factor V Leiden and prothrombin 20210A mutation were ascertained. Information on CVC was obtained from discharge letters. RESULTS: Forty-two patients (23%) and one control subject (0.04%) had a CVC (ORadj: 1136, 95% CI: 153-8448, adjusted for age and sex). Cancer patients without a CVC had an eightfold increased risk of venous thrombosis of the arm (ORcrude: 7.7, 95% CI: 4.6-13.0). Other evident risk factors were prothrombotic mutations, surgery, immobilization of the arm (plaster cast), oral contraceptive use and family history, with odds ratios varying from 2.0 up to 13.1. The risk in the presence of injury and during puerperium was twofold or more increased, although not significantly. In contrast HRT, unusual exercise, travel and obesity did not increase the risk. Hormone users had an increased risk in the presence of prothrombotic mutations or surgery. Obese persons (BMI > 30 kg m(-2)) undergoing surgery had a 23-fold increased risk of arm thrombosis compared with non-obese persons not undergoing surgery. CONCLUSION: A CVC is a very strong risk factor for arm thrombosis. Most risk factors for thrombosis in the leg are also risk factors for arm thrombosis.

Pain management in trauma patients in (pre)hospital based emergency care: current practice versus new guideline.

INTRODUCTION: Acute pain in trauma patients in emergency care is still undertreated. Early pain treatment is assumed to effectively reduce pain in patients and improve long-term outcomes. In order to improve pain management in the chain of emergency care, a national evidence-based guideline was developed. The aim of this study was to assess whether current practice is in compliance with the guideline 'Pain management for trauma patients in the chain of emergency care' from the Netherlands Association for Emergency Nurses (in Dutch NVSHV), and to evaluate early and initial pain management for adult trauma patients in emergency care. METHODS: Chart reviews were conducted in three regions of the Netherlands using electronic patient files of trauma patients from the chain of emergency care. We included one after-hours General Practitioner Co-operation (GPC), one ambulance Emergency Medical Services (EMS), two Helicopter Emergency Medical Services (HEMS), and three Emergency Departments (EDs). Organisation of pain management, pain assessment, and pain treatment was examined and compared with national guideline recommendations, including quality indicators. RESULTS: We assessed a random sample of 1066 electronic patient files. The use of standardised tools to assess pain was registered in zero to 52% of the electronic patient files per organisation. Registration of (non-)pharmacological pain treatment was found in less than half of the files. According to the files, pharmacological pain treatment deviated from the guideline in 73-99% of the files. Time of administration of medication was missing in 73-100%. Reassessment of pain following pain medication was recorded in half of the files by the HEMS, but not in files of the other organisations. CONCLUSIONS: The (registration of) current pain management in trauma patients in the chain of emergency care varies widely between healthcare organisation, and deviates from national guideline recommendations. Although guideline compliance differs across groups of healthcare professionals, maximum compliance rate with indicators registered is 52%. In order to improve pain management and evaluate its effectiveness, we recommend to improve pain registration in patient files. Furthermore, we advise to identify barriers and facilitators related to the implementation of the national guideline in all emergency care organisations.

High levels of procoagulant factors mediate the association between free thyroxine and the risk of venous thrombosis: the MEGA study.

BACKGROUND: Thyroid hormone affects the coagulation system, but its effect on clinical disease is not clear. We determined the associations of levels of free thyroxine (FT4), thyroid-stimulating hormone (TSH) and anti-thyroid peroxidase antibodies (antiTPO) with levels of coagulation factors and the risk of venous thrombosis. METHODS: In a large population based case-control study (Multiple Environmental and Genetic Assessment of risk factors for venous thrombosis study) on the etiology of venous thrombosis, we determined the levels of FT4, TSH, antiTPO, factor FII, FVII, FVIII, FIX, FX, von Willebrand factor (VWF), antithrombin, protein C, protein S and fibrinogen in 2177 cases and 2826 controls. RESULTS: High levels of FT4 were associated with increased concentrations of procoagulant factors, and not with levels of anticoagulant factors. High levels of FT4 were also associated with the risk of venous thrombosis, up to an odds ratio (OR) of 2.2 (95% confidence interval [CI] 1.0-4.6) for levels above 24.4 pm relative to FT4 levels between 15.5 and 18.9 pm. In 11 cases and one control, clinical hyperthyroidism had been diagnosed within a year of the thrombotic event, leading to an OR of 17.0 (95% CI 2.2-133.0) for thrombosis. The ORs approached unity after adjustment for FVIII and VWF, which suggests that the effect was mediated by these factors. Low TSH levels were also, but less evidently, associated with thrombosis, whereas there was no association between antiTPO and venous thrombosis risk. CONCLUSIONS: High levels of FT4 increase the concentrations of the procoagulant proteins FVIII, FIX, fibrinogen, and VWF, and by this mechanism increase the risk of venous thrombosis.

Predictors of the post-thrombotic syndrome with non-invasive venous examinations in patients 6 weeks after a first episode of deep vein thrombosis.

BACKGROUND: Post-thrombotic syndrome (PTS) is a chronic complication of deep vein thrombosis (DVT) affecting a large number of patients. Because of its potential debilitating effects, identification of patients at high risk for the development of this syndrome is relevant, and only a few predictors are known. OBJECTIVES: To assess the incidence and potential predictors of PTS. METHODS: We prospectively followed 111 consecutive patients for 2 years after a first episode of objectively documented DVT of the leg. With non-invasive venous examinations, residual thrombosis, valvular reflux, calf muscle pump function and venous outflow resistance were assessed at 6 weeks, 3 months, 6 months, 1 year, and 2 years. The Clinical, Etiologic, Anatomic, and Pathophysiologi classification was used to record the occurrence and severity of PTS. Regression analysis with area under the receiver operating characteristic (ROC) curve was performed to identify potential predictors. RESULTS: The cumulative incidence of PTS was 46% after 3 months, and the incidence and severity did not increase further. Men appeared to be at increased risk as compared with women (risk ratio [RR] 1.4, 95% confidence interval [CI] 0.9-2.2), as were patients over 50 years as compared with younger patients (RR 1.4%, 95% CI 0.9-2.1). Patients with thrombosis localized in the proximal veins at diagnosis had an increased risk of PTS as compared with patients with distal thrombosis (RR 2.3%, 95% CI 1.0-5.6). PTS developed in 32 of 52 patients (62%) with residual thrombosis in the proximal veins 6 weeks after diagnosis, as compared with 17 of 45 patients (38%) without residual proximal thrombosis, leading to a 1.6-fold increased risk (95% CI 1.0-2.5). The presence of valvular reflux in the superficial veins was also a predictor at 6 weeks, with a 1.6-fold increased risk as compared with patients without superficial reflux (95% CI 1.1-2.3). A multivariate analysis of these predictors yielded an area under the ROC curve of 0.72 (95% CI 0.62-0.82). CONCLUSIONS: PTS develops in half of all patients within 3 months, with no further increase being seen up to 2 years of follow-up. Male sex, age over 50 years, proximal localization of the thrombus at entry, residual proximal thrombosis and superficial valvular reflux at 6 weeks seem to be the most important predictors of PTS in patients with a first episode of DVT. Duplex scanning 6 weeks after diagnosis appears to be clinically useful for the identification of patients at risk of PTS.