Search

To estimate HIV incidence among female sex workers (FSW) in Zimbabwe: using HIV prevalence by age and number of years since started selling sex (YSSS).We pooled data from FSW aged 18-39 participating in respondent-driven sampling surveys conducted in Zimbabwe between 2011 and 2017.For each year of age, we estimated: HIV prevalence ( Pt ) and the change in HIV prevalence from the previous age ( Pt - Pt -1 ). We then estimated the rate of new HIV infections during that year of age: It = Pt - Pt -1 /(1 - Pt -1 ), and calculated HIV incidence for 18-24 and 25-39 year-olds separately as the weighted average of It . We estimated HIV incidence for FSW 1-5 years and 6-15 years since first selling sex using the same approach, and compared HIV prevalence among FSW first selling sex at their current age with the general population.Among 9906 women, 50.2% were HIV positive. Based on HIV prevalence increases by age, we estimated an HIV incidence of 6.3/100 person-years at risk (pyar) (95% confidence interval [CI] 5.3, 7.6) among 18-24 year-olds, and 3.3/100 pyar (95% CI 1.3, 4.2) among 25-39 year-olds. Based on prevalence increases by YSSS, HIV incidence was 5.3/100 pyar (95% CI 4.3, 8.5) between 1 and 5 years since first selling sex, and 2.1/100 pyar (95% CI -1.3, 7.2) between 6 and 15 years.Our analysis is consistent with very high HIV incidence among FSW in Zimbabwe, especially among those who are young and recently started selling sex. There is a critical need to engage young entrants into sex work in interventions that reduce their HIV risk.

The creation of The 100 Million Brazilian Cohort was motivated by the availability of high quality but dispersed social and health databases in Brazil and the need to integrate data and evaluate the impact of policies aiming to improve the social determinants of health (e.g. social protection policies) on health outcomes, overall and in subgroups of interest in a dynamic cohort. • The baseline of The 100 Million Brazilian Cohort comprises 131 697 800 low-income individuals in 35 358 415 families from 2011 to 2018. The Cohort population is mostly composed of children and young adults, with a higher proportion of females than the general Brazilian population, who identify themselves as Brown and live in the urban area of the country. • Exposure to social protection and the follow-up of individuals are obtained through: (i) deterministic linkage using the Social Identification Number (NIS) to link the Cohort baseline to social protection programmes and to periodically renewed socioeconomic information in Cadatro U ́ nico datasets; and/or (ii) non-deterministic linkage using the CIDACS-RL non-deterministic linkage tool, to link the Cohort baseline to administrative health care datasets such as mortality (Mortality Information System, SIM), disease notification (Information System for Notifiable Diseases, SINAN), birth information (Live Birth Information System, SINASC) and nutrition status (Food and Nutrition Surveillance System, SISVAN). • So far, studies have used The 100 Million Brazilian Cohort to investigate the socioeconomic and demographic determinants of leprosy, leprosy treatment outcomes and low birthweight and to evaluate the impact of the Bolsa Familia Programme (BFP) on leprosy and child mortality. Other studies are now being conducted that are of utmost relevance to the health inequalities of Brazil and many low- and middle-income countries, and many research opportunities are being opened up with the linkage of a range of health outcomes.

Background Although routine NHS data potentially include all patients, confounding limits their use for causal inference. Methods to minimise confounding in observational studies of implantable devices are required to enable the evaluation of patients with severe systemic morbidity who are excluded from many randomised controlled trials. Objectives Stage 1 – replicate the Total or Partial Knee Arthroplasty Trial (TOPKAT), a surgical randomised controlled trial comparing unicompartmental knee replacement with total knee replacement using propensity score and instrumental variable methods. Stage 2 – compare the risk benefits and cost-effectiveness of unicompartmental knee replacement with total knee replacement surgery in patients with severe systemic morbidity who would have been ineligible for TOPKAT using the validated methods from stage 1. Design This was a cohort study. Setting Data were obtained from the National Joint Registry database and linked to hospital inpatient (Hospital Episode Statistics) and patient-reported outcome data. Participants Stage 1 – people undergoing unicompartmental knee replacement surgery or total knee replacement surgery who met the TOPKAT eligibility criteria. Stage 2 – participants with an American Society of Anesthesiologists grade of ≥ 3. Intervention The patients were exposed to either unicompartmental knee replacement surgery or total knee replacement surgery. Main outcome measures The primary outcome measure was the postoperative Oxford Knee Score. The secondary outcome measures were 90-day postoperative complications (venous thromboembolism, myocardial infarction and prosthetic joint infection) and 5-year revision risk and mortality. The main outcome measures for the health economic analysis were health-related quality of life (EuroQol-5 Dimensions) and NHS hospital costs. Results In stage 1, propensity score stratification and inverse probability weighting replicated the results of TOPKAT. Propensity score adjustment, propensity score matching and instrumental variables did not. Stage 2 included 2256 unicompartmental knee replacement patients and 57,682 total knee replacement patients who had severe comorbidities, of whom 145 and 23,344 had linked Oxford Knee Scores, respectively. A statistically significant but clinically irrelevant difference favouring unicompartmental knee replacement was observed, with a mean postoperative Oxford Knee Score difference of Limitations Although some propensity score methods successfully replicated TOPKAT, unresolved confounding may have affected stage 2. Missing Oxford Knee Scores may have led to information bias. Conclusions Propensity score stratification and inverse probability weighting successfully replicated TOPKAT, implying that some (but not all) propensity score methods can be used to evaluate surgical innovations and implantable medical devices using routine NHS data. Unicompartmental knee replacement was safer and more cost-effective than total knee replacement for patients with severe comorbidity and should be considered the first option for suitable patients. Future work Further research is required to understand the performance of propensity score methods for evaluating surgical innovations and implantable devices. Trial registration This trial is registered as EUPAS17435. Funding This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full in Health Technology Assessment; Vol. 25, No. 66. See the NIHR Journals Library website for further project information.

Interrupted time series (ITS) analysis is being increasingly used in epidemiology. Despite its growing popularity, there is a scarcity of guidance on power and sample size considerations within the ITS framework. Our aim of this study was to assess the statistical power to detect an intervention effect under various real-life ITS scenarios. ITS datasets were created using Monte Carlo simulations to generate cumulative incidence (outcome) values over time. We generated 1,000 datasets per scenario, varying the number of time points, average sample size per time point, average relative reduction post intervention, location of intervention in the time series, and reduction mediated via a 1) slope change and 2) step change. Performance measures included power and percentage bias. We found that sample size per time point had a large impact on power. Even in scenarios with 12 pre-intervention and 12 post-intervention time points with moderate intervention effect sizes, most analyses were underpowered if the sample size per time point was low. We conclude that various factors need to be collectively considered to ensure adequate power for an ITS study. We demonstrate a means of providing insight into underlying sample size requirements in ordinary least squares (OLS) ITS analysis of cumulative incidence measures, based on prespecified parameters and have developed Stata code to estimate this.

Bisphosphonates are contraindicated in patients with stage 4+ chronic kidney disease. However, they are widely used to prevent fragility fractures in stage 3 chronic kidney disease, despite a lack of good-quality data on their effects.The aims of each work package were as follows. Work package 1: to study the relationship between bisphosphonate use and chronic kidney disease progression. Work package 2: to study the association between using bisphosphonates and fracture risk. Work package 3: to determine the risks of hypocalcaemia, hypophosphataemia, acute kidney injury and upper gastrointestinal events associated with using bisphosphonates. Work package 4: to investigate the association between using bisphosphonates and changes in bone mineral density over time.This was a new-user cohort study design with propensity score matching.Data were obtained from UK NHS primary care (Clinical Practice Research Datalink GOLD database) and linked hospital inpatient records (Hospital Episode Statistics) for work packages 1-3 and from the Danish Odense University Hospital Databases for work package 4.Patients registered in the data sources who had at least one measurement of estimated glomerular filtration rate of 45 ml/minute/1.73 mBisphosphonate use, identified from primary care prescriptions (for work packages 1-3) or pharmacy dispensations (for work package 4), was the main exposure.Work package 1: chronic kidney disease progression, defined as stage worsening or starting renal replacement. Work package 2: hip fracture. Work package 3: acute kidney injury, hypocalcaemia and hypophosphataemia identified from Hospital Episode Statistics, and gastrointestinal events identified from Clinical Practice Research Datalink or Hospital Episode Statistics. Work package 4: annualised femoral neck bone mineral density percentage change.Bisphosphonate use was associated with an excess risk of chronic kidney disease progression (subdistribution hazard ratio 1.12, 95% confidence interval 1.02 to 1.24) in work package 1, but did not increase the probability of other safety outcomes in work package 3. The results from work package 2 suggested that bisphosphonate use increased fracture risk (hazard ratio 1.25, 95% confidence interval 1.13 to 1.39) for hip fractures, but sensitivity analyses suggested that this was related to unresolved confounding. Conversely, work package 4 suggested that bisphosphonates improved bone mineral density, with an average 2.65% (95% confidence interval 1.32% to 3.99%) greater gain in femoral neck bone mineral density per year in bisphosphonate users than in matched non-users.Confounding by indication was a concern for the clinical effectiveness (i.e. work package 2) data. Bias analyses suggested that these findings were due to inappropriate adjustment for pre-treatment risk. work packages 3 and 4 were based on small numbers of events and participants, respectively.Bisphosphonates were associated with a 12% excess risk of chronic kidney disease progression in participants with stage 3B+ chronic kidney disease. No other safety concerns were identified. Bisphosphonate therapy increased bone mineral density, but the research team failed to demonstrate antifracture effectiveness.Randomised controlled trial data are needed to demonstrate antifracture efficacy in patients with stage 3B+ chronic kidney disease. More safety analyses are needed to characterise the renal toxicity of bisphosphonates in stage 3A chronic kidney disease, possibly using observational data.This study is registered as EUPAS10029.This project was funded by the National Institute for Health Research (NIHR) Health Technology Assessment programme and will be published in full inBisphosphonates are used to prevent fractures in people with fragile bones. People with chronic kidney disease have a high risk of fracturing, but the safety and effectiveness of bisphosphonates in severe chronic kidney disease is unclear. The aim of this study was to assess the benefits (e.g. bone strength improvement and fracture prevention) and the risks of unwanted effects associated with bisphosphonates for people with moderate to severe chronic kidney disease.Anonymised primary and secondary care electronic medical records data from the UK NHS were used, as well as a Danish equivalent that included bone density scans. Anyone in these databases with a measure of reduced kidney function that suggested moderate to severe chronic kidney disease was eligible, which was 220,000 people from the UK. Over 20,000 of them used bisphosphonates. Bisphosphonate users were matched to non-users with similar age, sex and other characteristics.Bisphosphonate users had a 12% higher risk of their chronic kidney disease getting worse than non-users. Their risks of other side effects, such as acute kidney injuries and gastrointestinal problems, did not change. Bisphosphonate users had a 25% higher risk of fractures than non-users in the UK database, probably because the matching methods did not create similar-enough groups of users and non-users. However, it was found that bisphosphonate improved bone density in the Danish database. Bone density is a proxy for bone strength, so better bone density should mean fewer fractures.These results suggest that bisphosphonate therapy may make moderate to severe chronic kidney disease worse. More studies are needed on how bisphosphonates affect milder chronic kidney disease. Bisphosphonates were associated with better bone strength, but it could not be demonstrated that they reduced fracture risk. More data are required, probably from a placebo-controlled trial, to determine whether or not bisphosphonates prevent fractures in people with moderate to severe chronic kidney disease and whether or not this is worth the risk of their chronic kidney disease worsening.

Bisphosphonates are the first-line treatment for preventing fractures in osteoporosis patients. However, their use is contraindicated or to be used with caution in chronic kidney disease (CKD) patients, primarily because of a lack of information about their safety and effectiveness. We aimed to investigate the safety of oral bisphosphonates in patients with moderate to severe CKD, using primary-care electronic records from two cohorts, CPRD GOLD (1997-2016) and SIDIAP (2007-2015) in the UK and Catalonia, respectively. Both databases were linked to hospital records. SIDIAP was also linked to end-stage renal disease registry data. Patients with CKD stages 3b to 5, based on two or more estimated glomerular filtration rate measurements less than 45 mL/min/1.73 m2 , aged 40 years or older were identified. New bisphosphonate users were propensity score-matched with up to five non-users to minimize confounding within this population. Our primary outcome was CKD stage worsening (estimated glomerular filtration rate [eGFR] decline or renal replacement therapy). Secondary outcomes were acute kidney injury, gastrointestinal bleeding/ulcers, and severe hypocalcemia. Hazard ratios (HRs) were estimated using Cox regression and Fine and Gray sub-HRs were calculated for competing risks. We matched 2447 bisphosphonate users with 8931 non-users from CPRD and 1399 users with 6547 non-users from SIDIAP. Bisphosphonate use was associated with greater risk of CKD progression in CPRD (sub-HR [95% CI]: 1.14 [1.04, 1.26]) and SIDIAP (sub-HR: 1.15 [1.04, 1.27]). No risk differences were found for acute kidney injury, gastrointestinal bleeding/ulcers, or hypocalcemia. Hence, we can conclude a modest (15%) increased risk of CKD progression was identified in association with bisphosphonate use. No other safety concerns were identified. Our findings should be considered before prescribing bisphosphonates to patients with moderate to severe CKD. © 2020 American Society for Bone and Mineral Research (ASBMR).

Background Record linkage is the process of identifying and combining records about the same individual from two or more different datasets. While there are many open source and commercial data linkage tools, the volume and complexity of currently available datasets for linkage pose a huge challenge; hence, designing an efficient linkage tool with reasonable accuracy and scalability is required. Methods We developed CIDACS-RL (Centre for Data and Knowledge Integration for Health – Record Linkage), a novel iterative deterministic record linkage algorithm based on a combination of indexing search and scoring algorithms (provided by Apache Lucene). We described how the algorithm works and compared its performance with four open source linkage tools (AtyImo, Febrl, FRIL and RecLink) in terms of sensitivity and positive predictive value using gold standard dataset. We also evaluated its accuracy and scalability using a case-study and its scalability and execution time using a simulated cohort in serial (single core) and multi-core (eight core) computation settings. Results Overall, CIDACS-RL algorithm had a superior performance: positive predictive value (99.93% versus AtyImo 99.30%, RecLink 99.5%, Febrl 98.86%, and FRIL 96.17%) and sensitivity (99.87% versus AtyImo 98.91%, RecLink 73.75%, Febrl 90.58%, and FRIL 74.66%). In the case study, using a ROC curve to choose the most appropriate cut-off value (0.896), the obtained metrics were: sensitivity = 92.5% (95% CI 92.07–92.99), specificity = 93.5% (95% CI 93.08–93.8) and area under the curve (AUC) = 97% (95% CI 96.97–97.35). The multi-core computation was about four times faster (150 seconds) than the serial setting (550 seconds) when using a dataset of 20 million records. Conclusion CIDACS-RL algorithm is an innovative linkage tool for huge datasets, with higher accuracy, improved scalability, and substantially shorter execution time compared to other existing linkage tools. In addition, CIDACS-RL can be deployed on standard computers without the need for high-speed processors and distributed infrastructures.

Summary Bisphosphonates are contraindicated in moderate-to-severe chronic kidney disease patients. However, they are used to prevent fragility fractures in patients with impaired kidney function, despite a lack of evidence on their effects on bone density in these patients. We demonstrated that Alendronate had a positive effect on bone in these patients. Purpose This study aimed to assess the association between alendronate use and bone mineral density (BMD) change in subjects with moderate-severe chronic kidney disease (CKD). Methods We created a cohort of CKD stage 3B–5 patients by linking all DXA-based measurements in the Funen area, Denmark, to biochemistry, national health registries and filled prescriptions. Exposure was dispensation of alendronate and the outcome was annualized percentage change in BMD at the femoral neck, total hip and lumbar spine. Individuals were followed from first BMD to the latest of subsequent DXA measurements. Alendronate non-users were identified using incidence density sampling and matched groups were created using propensity scores. Linear regression was used to estimate average differences in the annualized BMD. Results Use of alendronate was rare in this group of patients: propensity score matching (PSM) resulted in 71 alendronate users and 142 non-users with stage 3B–5 CKD (as in the 1 year before DXA). Whilst alendronate users gained an average 1.07% femoral neck BMD per year, non-users lost an average of 1.59% per annum. The PSM mean differences in annualized BMD were + 2.65% (1.32%, 3.99%), + 3.01% (1.74%, 4.28%) and + 2.12% (0.98%, 3.25%) at the femoral neck, total hip and spine BMD, respectively, all in favour of alendronate users. Conclusion In a real-world cohort of women with stage 3B–5 CKD, use of alendronate appears associated with a significant improvement of 2–3% per year in the femoral neck, total hip and spine BMD. More data are needed on the anti-fracture effectiveness and safety of bisphosphonate therapy in moderate-severe CKD.

Additional file 1: Description of: (1) the different linkage tools; (2) the datasets and the creation of the gold standard dataset; (3) the comparison metrics; (4) the descriptive analysis of the linkage between CadUnico and SINAN-TB; (5) the determination of Threshold in Weight Vector Classification.

Pharmacoepidemiology is used extensively in osteoporosis research and involves the study of the use and effects of drugs in large numbers of people. Randomized controlled trials are considered the gold standard in assessing treatment efficacy and safety. However, their results can have limited external validity when applied to day‐to‐day patients. Pharmacoepidemiological studies aim to assess the effect/s of treatments in actual practice conditions, but they are limited by the quality, completeness, and inherent bias due to confounding. Sources of information include prospectively collected (primary) as well as readily available routinely collected (secondary) (eg, electronic medical records, administrative/claims databases) data. Although the former enable the collection of ad hoc measurements, the latter provide a unique opportunity for the study of large representative populations and for the assessment of rare events at relatively low cost. Observational cohort and case‐control studies, the most commonly implemented study designs in pharmacoepidemiology, each have their strengths and limitations. However, the choice of the study design depends on the research question that needs to be answered. Despite the many advantages of observational studies, they also have limitations. First, missing data is a common issue in routine data, frequently dealt with using multiple imputation. Second, confounding by indication arises because of the lack of randomization; multivariable regression and more specific techniques such as propensity scores (adjustment, matching, stratification, trimming, or weighting) are used to minimize such biases. In addition, immortal time bias (time period during which a subject is artefactually event‐free by study design) and time‐varying confounding (patient characteristics changing over time) are other types of biases usually accounted for using time‐dependent modeling. Finally, residual “uncontrolled” confounding is difficult to assess, and hence to account for it, sensitivity analyses and specific methods (eg, instrumental variables) should be considered. © 2018 The Authors JBMR Plus published by Wiley Periodicals, Inc. on behalf of American Society for Bone and Mineral Research.

ObjectivesPrevious ecological data suggest a decline in the need for joint replacements in RA patients following the introduction of TNF inhibitor (TNFi) therapy, although patient-level data are lacking. Our primary aim was to estimate the association between TNFi use and subsequent incidence of total hip replacement (THR) and total knee replacement.MethodsA propensity score matched cohort was analysed using the British Society for Rheumatology Biologics Registry (2001–2016) for RA data. Propensity score estimates were used to match TNFi users to similar conventional synthetic DMARD users (with replacement) using a 1:1 ratio. Weighted multivariable Cox regression was used to estimate the impact of TNFi on study outcomes. Effect modification by baseline age and disease severity were investigated. Joint replacement at other sites was also analysed. An instrumental variable sensitivity analysis was also performed.ResultsThe matched analysis contained a total of 19 116 patient records. Overall, there was no significant association between TNFi use vs conventional synthetic DMARD on rates of THR (hazard ratios = 0.86 [95% CI: 0.60, 1.22]) although there was significant effect modification by age (P < 0.001). TNFi was associated with a reduction in THR among those >60 years old (hazard ratio = 0.60 [CI: 0.41, 0.87]) but not in younger patients. No significant associations were found for total knee replacement or other joint replacement.ConclusionOverall, no association was found between the use of TNFi and subsequent incidence of joint replacement. However, TNFi was associated with a 40% relative reduction in THR rates among older patients.

Oral bisphosphonates (oBPs) have been associated with reduced fractures and mortality. However, their risks and benefits are unclear in patients with moderate–severe CKD. This study examined the association between oBPs and all-cause mortality in G3B-5D CKD. This is a population-based cohort study including all subjects with an estimated glomerular filtration rate (eGFR) 2 (G3B: eGFR 2 G4: eGFR 15–29/mL/min/1.73 m 2 G5: eGFR 2 G5D: hemodialysis) aged 40+ years from the UK Clinical Practice Research Datalink (CPRD) and the Catalan Information System for Research in Primary Care (SIDIAP). Previous and current users of other anti-osteoporosis drugs were excluded. oBP use was modeled as a time-varying exposure to avoid immortal time bias. Treatment episodes in oBP users were created by concatenating prescriptions until patients switched or stopped therapy or were censored or died. A washout period of 180 days was added to (date of last prescription +180 days). Propensity scores (PSs) were calculated using prespecified predictors of mortality including age, gender, baseline eGFR, socioeconomic status, comorbidities, previous fracture, co-medications, and number of hospital admissions in the previous year. Cox models were used for PS adjustment before and after PS trimming (the first and last quintiles). In the CPRD, of 19,351 oBP users and 210,954 non-oBP users, 5234 (27%) and 85,105 (40%) deaths were recorded over 45,690 and 915,867 person-years of follow-up, respectively. oBP users had 8% lower mortality risk compared to non-oBP users (hazard ratio [HR] 0.92; 95% CI, 0.89 to 0.95). Following PS trimming, this became nonsignificant (HR 0.98; 95% CI, 0.94 to 1.04). In the SIDIAP, of 4146 oBP users and 86,127 non-oBP users, 1330 (32%) and 36,513 (42%) died, respectively. oBPs were not associated with mortality in PS adjustment and trimming (HR 1.04; 95% CI, 0.99 to 1.1 and HR 0.95; 95% CI, 0.89 to 1.01). In this observational, patient-based cohort study, oBPs were not associated with increased mortality among patients with moderate–severe CKD. However, further studies are needed on other effects of oBPs in CKD patients.

Background: Although contraindicated in patients with severe chronic kidney disease (CKD) due to a lack of safety data, bisphosphonates are commonly used to prevent or treat bone fragility in these patients. We studied the safety of bisphosphonates in patients with moderate-to-severe CKD. Methods: Patients with stage 3b+ CKD based on 2+ estimated glomerular filtration rate (eGFR) measures were identified in two primary-care records databases, the UK CPRD GOLD during 1997-2016 and the Catalan (Spain) SIDIAP during 2007-2015. New bisphosphonate users were propensity-scorematched with up to five non-users to minimise confounding, and followed until treatment discontinuation, switching, dropout, death, or 10 years. The primary outcome was CKD stage worsening (eGFR decline or renal replacement therapy). Secondary outcomes included hospitalised acute kidney injury (AKI), gastrointestinal bleeding/ulcers (GIE), and hypocalcaemia. Hazard ratios (HRs) were estimated using Cox regression for SIDIAP participants, and sub-HRs (sHRs) calculated using Fine-Gray models for CPRD. Findings: We matched 2,447 bisphosphonate users with 8,931 non-users from CPRD, and 1,399 users with 6,547 non-users from SIDIAP. Bisphosphonate use was associated with greater risk of CKD progression in both CPRD (HR [95% CI]: 1·14 [1·04, 1·26]) and SIDIAP (HR: 1·16 [1·04, 1·29]). No statistically significant risk differences were found for AKI, GIE, or hypocalcaemia. Interpretation: A modest (10%-15%) increased risk of CKD progression was identified in association with bisphosphonate use amongst patients with moderate-severe disease. No other safety concerns were identified. Our findings should be considered in combination with effectiveness data before prescribing bisphosphonates to patients with moderate-severe CKD. Funding Statement: Funding was provided by the National Institute of Health Research Health Technology Assessment project number 14/36/02. The project was partially funded by CIBERFES, Instituto Carlos III, Spain. JP and MJP-S are supported by PI16/00619 (Spanish Ministry of Health ISCIII FIS-FEDER) and RD16/0009/0013 (ISCIII FEDER RedinRen). Declaration of Interests: B Abrahamsen reports research grants from UCB and Novartis. NK Arden reports personal fees from Flexion, Freshfields, Janssen, Merck, and Regeneron. C Cooper reports lecture fees and honoraria from the Alliance for Better Bone Health Amgen, Eli Lilly, GSK, Medtronic, Merck, Novartis, Pfizer, Roche, Servier, Takeda, and UCB. A Judge is a subpanel member of the NIHR PGfAR, has received consultancy fees from Freshfields Bruckhaus Deringer, and has held advisory board positions (which involved receipt of fees) at Anthera Pharmaceuticals, INC. MK Javaid reports research support and speaker fees from Amgen. D Prieto-Alhambra reports research grants from UCB, Amgen, and Les Laboratoires Servier, and departmental (not personal) speaker and consultancy fees from UCB and Amgen. D Prieto-Alhambra is a member of the NIHR HTA CET panel since November 2017. DE Robinson, MS Ali, VY Strauss, L Elhussein, Y Ben-Shlomo, F Caskey, D Dedman, and A Delmestri have nothing to disclose. Ethics Approval Statement: This study was approved by the CPRD Independent Scientific Advisory Committee (protocol number 15_ 53R2A).

BACKGROUND: Observational studies including time-varying treatments are prone to confounding. We compared time-varying Cox regression analysis, propensity score (PS) methods, and marginal structural models (MSMs) in a study of antidepressant [selective serotonin reuptake inhibitors (SSRIs)] use and the risk of hip fracture. METHODS: A cohort of patients with a first prescription for antidepressants (SSRI or tricyclic antidepressants) was extracted from the Dutch Mondriaan and Spanish Base de datos para la Investigacion Farmacoepidemiologica en Atencion Primaria (BIFAP) general practice databases for the period 2001-2009. The net (total) effect of SSRI versus no SSRI on the risk of hip fracture was estimated using time-varying Cox regression, stratification and covariate adjustment using the PS, and MSM. In MSM, censoring was accounted for by inverse probability of censoring weights. RESULTS: The crude hazard ratio (HR) of SSRI use versus no SSRI use on hip fracture was 1.75 (95%CI: 1.12, 2.72) in Mondriaan and 2.09 (1.89, 2.32) in BIFAP. After confounding adjustment using time-varying Cox regression, stratification, and covariate adjustment using the PS, HRs increased in Mondriaan [2.59 (1.63, 4.12), 2.64 (1.63, 4.25), and 2.82 (1.63, 4.25), respectively] and decreased in BIFAP [1.56 (1.40, 1.73), 1.54 (1.39, 1.71), and 1.61 (1.45, 1.78), respectively]. MSMs with stabilized weights yielded HR 2.15 (1.30, 3.55) in Mondriaan and 1.63 (1.28, 2.07) in BIFAP when accounting for censoring and 2.13 (1.32, 3.45) in Mondriaan and 1.66 (1.30, 2.12) in BIFAP without accounting for censoring. CONCLUSIONS: In this empirical study, differences between the different methods to control for time-dependent confounding were small. The observed differences in treatment effect estimates between the databases are likely attributable to different confounding information in the datasets, illustrating that adequate information on (time-varying) confounding is crucial to prevent bias. Copyright © 2016 John Wiley & Sons, Ltd.

The purpose of this study is to evaluate the performance and validity of the case-crossover (CCO) and self-controlled case-series (SCCS) designs when studying the association between hip/femur fracture (HF) and antidepressant (AD) use in general practitioner databases. In addition, comparability with cohort and case-control designs is discussed. Adult patients with HF and who received an AD prescription during 2001-2009 were identified from UK's The Health Improvement Network (THIN) and the Dutch Mondriaan databases. AD exposure was classified into current, recent and past/non-use (reference). In the CCO, for each patient, a case moment (date of HF) and four prior control moments at -91, -182, -273 and -365 days were defined. In SCCS, incidence of HF was compared between exposure states. Conditional logistic regression was used in the CCO and Poisson regression in the SCCS to compute odds ratios and incidence rate ratios, respectively. In CCO, we adjusted for time-varying co-medication and in SCCS for age. Adjusted estimates for the effect of current AD exposure on HF were higher in the CCO (co-medication-adjusted odds ratio, THIN: 2.24, 95% confidence interval [CI]: 2.04-2.47; Mondriaan: 2.57, 95%CI [1.50, 4.43]) than in the SCCS (age-adjusted incidence rate ratio, THIN: 1.41, 95%CI [1.32, 1.49]; Mondriaan: 2.14, 95%CI [1.51, 3.03]). The latter were comparable with the traditional designs. Case-only designs confirmed the association between AD and HF. The CCO design violated assumptions in this study with regard to exchangeability and length of exposure, and transient effects on outcome. The SCCS seems to be an appropriate design for assessing AD-HF association. Copyright © 2016 John Wiley & Sons, Ltd.

Sara Khalid,1,2 Sara Calderon-Larrañaga,3 Samuel Hawley,2 M Sanni Ali1,2,4 Andrew Judge,1,2,5 Nigel Arden,6 Tjeerd van Staa,7,8 Cyrus Cooper,2,9,10 Muhammad Kassim Javaid,2 Daniel Prieto-Alhambra2,11 1Centre for Statistics in Medicine, Nuffield Department of Orthopaedics, Rheumatology, and Musculoskeletal Sciences, 2Oxford NIHR Biomedical Research Centre, University of Oxford, Oxford, United Kingdom; 3Family and Community Medicine Teaching Unit of Granada. Cartuja University Health Centre. Andalusian Health Service (SAS), Granada, Spain; 4London School of Hygiene and Tropical Medicine, London, United Kingdom; 5Bristol NIHR Biomedical Research Centre, Musculoskeletal Research Unit, Southmead Hospital, University of Bristol, Bristol, United Kingdom; 6Arthritis Research UK Centre for Sport, Exercise, and Osteoarthritis, University of Oxford, Oxford, United Kingdom; 7Farr Institute, University of Manchester, Manchester, United Kingdom; 8Utrecht Institute for Pharmaceutical Sciences, Utrecht University, Utrecht, the Netherlands; 9Medical Research Council Lifecourse Epidemiology Unit, University of Southampton, Southampton, United Kingdom; 10NIHR Southampton Biomedical Research Centre, University of Southampton and University Hospital Southampton NHS Foundation Trust, Southampton, United Kingdom; 11GREMPAL (Grup de Recerca en Malalties Prevalents de l’Aparell Locomotor) Research Group, Idiap Jordi Gol Primary Care Research Institute, Universitat Autònoma de Barcelona, Barcelona, Spain Purpose: This paper aims to compare the clinical effectiveness of oral anti-osteoporosis drugs based on the observed risk of fracture while on treatment in primary care actual practice.Materials and methods: We investigated two primary care records databases covering UK National Health Service (Clinical Practice Research Datalink, CPRD) and Catalan healthcare (Information System for Research in Primary Care, SIDIAP) patients during 1995–2014 and 2006–2014, respectivey. Treatment-naive incident users of anti-osteoporosis drugs were included and followed until treatment cessation, switching, death, transfer out, or study completion. We considered hip fracture while on treatment as main outcome and major osteoporotic fractures (hip, clinical spine, wrist, and proximal humerus) as secondary outcome. Users of alendronate (reference group) were compared to those of (1) OBP, (2) strontium ranelate (SR), and (3) selective estrogen receptor modulators (SERMs), after matching on baseline characteristics using propensity scores. Multiple imputation was used to handle missing data on confounders and competing risk modelling for the calculation of relative risk according to therapy. Country-specific data were analyzed separately and meta-analyzed.Results: A total of 163,950 UK and 145,236 Catalan patients were identified. Hip (sub-hazard ratio [SHR] [95% CI] 1.04 [0.77–1.40]) and major osteoporotic (SHR [95% CI] 1 [0.78–1.27]) fracture risks were similar among OBP compared to alendronate users. Both hip (SHR [95% CI] 1.26 [1.14–1.39]) and major osteoporotic (SHR [95% CI] 1.06 [1.02–1.12]) fracture risk were higher in SR compared to alendronate users. SERM users had a reduced hip (SHR [95% CI] 0.75 [0.60–0.94]) and major osteoporotic (SHR [95% CI] 0.77 [0.72–0.83]) fracture risk compared to alendronate users.Conclusion: We found a 26% excess hip fracture risk among SR compared to matched alendronate users, in line with placebo-controlled RCT findings. Conversely, in a lower risk population, SERM users had a 25% reduced hip fracture risk compared to alendronate users. Head-to-head RCTs are needed to confirm these findings. Keywords: pharmaco-epidemiology, anti-osteoporosis medication, osteoporosis, fracture risk, electronic health records

Appropriate drug prescription for an increasingly ageing and multi-morbid population can be a challenge for general practitioners. This study uses unsupervised learning methods to identify different types of patient profiles which could inform policymakers and regulators about patterns of drug use, and identify specific clusters of users with unknown drug effects (risk and benefit). Hard and soft clustering methods are proposed to detect patterns of medication use by patients and to estimate the probability of belonging to a certain patient profile. Results showed the presence of expected as well as a surprising patient profile based on fracture risk factors. Challenges associated with unsupervised learning using electronic medical record data are described and an approach for evaluating models in the presence of unlabeled data using internal and external cluster evaluation methods is presented, such that it can be extended to other unsupervised learning applications within healthcare and beyond. To our knowledge, this is the first study proposing cluster analysis for detecting drug utilisation patterns from electronic healthcare records in the routinely-collected SIDIAP database.

On December 8, 2016, the New England Journal of Medicine published a sounding board on Real World Evidence (RWE)1 by the US Food and Drug Administration (FDA) leadership. While the value of RWE based on nonrandomized observational studies was appreciated, such as for hypothesis generating, safety, and measuring quality in healthcare delivery, the authors expressed concerns on the quality of data sources and the ability of methodologies to control for confounding. In response, we offer a few considerations regarding these concerns.

Stratification and conditioning on time-varying cofounders which are also intermediates can induce collider-stratification bias and adjust-away the (indirect) effect of exposure. Similar bias could be expected when one conditions on time-dependent PS. We explored collider-stratification and confounding bias due to conditioning or stratifying on time-dependent PS using a clinical example on the effect of inhaled short- and long-acting beta2-agonist use (SABA and LABA, respectively) on coronary heart disease (CHD). In an electronic general practice database we selected a cohort of patients with an indication for SABA and/or LABA use and ascertained potential confounders and SABA/LABA use per three month intervals. Hazard ratios (HR) were estimated using PS stratification as well as covariate adjustment and compared with those of Marginal Structural Models (MSMs) in both SABA and LABA use separately. In MSMs, censoring was accounted for by including inverse probability of censoring weights.The crude HR of CHD was 0.90 [95 % CI: 0.63, 1.28] and 1.55 [95 % CI: 1.06, 2.62] in SABA and LABA users respectively. When PS stratification, covariate adjustment using PS, and MSMs were used, the HRs were 1.09 [95 % CI: 0.74, 1.61], 1.07 [95 % CI: 0.72, 1.60], and 0.86 [95 % CI: 0.55, 1.34] for SABA, and 1.09 [95 % CI: 0.74, 1.62], 1.13 [95 % CI: 0.76, 1.67], 0.77 [95 % CI: 0.45, 1.33] for LABA, respectively. Results were similar for different PS methods, but higher than those of MSMs. When treatment and confounders vary during follow-up, conditioning or stratification on time-dependent PS could induce substantial collider-stratification or confounding bias; hence, other methods such as MSMs are recommended.

Instrumental variable analysis has been used to control for unmeasured confounding in nonrandomized studies. An instrumental variable 1) is associated with exposure, 2) affects outcome only through exposure, and 3) is the independent of confounders. If these key assumptions are satisfied (together with additional assumptions such as homogeneity) instrumental variable analysis could consistently estimate the average causal effect of exposure. However if one of the assumptions is violated, the estimate can be severely biased. Several methods are available for checking the first assumption but there is no well-established method of checking the second and third assumptions. Some authors have argued that these assumptions are untestable, as they involve unmeasured confounding. We proposed a standardized difference a robust balance measure used in propensity score analysis to falsify the third assumption by checking independence between an instrumental variable and measured confounders Language: en

Ankarfeldt MZ, Thorsted BL, Groenwold RHH, et al. Clin Epidemiol. 2017;9:19–30.On page 23, Figure 1 Notes section was marked incorrectly. The correct Notes should read as follows:Figure 1 Propensity score over time for GLP-1 versus basal insulin initiators.Notes: Blue: insulin, red: GLP-1.Abbreviation: GLP-1, glucagon-like peptide-1 analogs.On page 24, Figure 2 Notes section was marked incorrectly. The correct Notes should read as follows:Figure 2 Propensity score over time for DPP-4i versus sulfonylurea initiators.Notes: Blue: sulfonylurea, red: DPP-4i.Abbreviation: DPP-4i, dipeptidyl peptidase-4 inhibitors.On Page 3 of Supplementary materials, Figure S2 caption was shown incorrectly. The correct caption should read as follows:Figure S2 Histograms of propensity score over time. Intention-to-treat and perprotocol cohorts of all identified initiators of GLP-1 and insulin, and DPP-4i and sulfonylurea, respectively. Blue: insulin and sulfonylurea initiators, respectively. Red: GLP-1 and DPP-4i initiators, respectively.Despite the above corrections, the interpretation of these figures in the published proof and the Supplementary materials was correct.Read the original article

Observational studies including time-varying treatments are prone to confounding. We compared time-varying Cox regression analysis, propensity score (PS) methods, and marginal structural models (MSMs) in a study of antidepressant [selective serotonin reuptake inhibitors (SSRIs)] use and the risk of hip fracture.A cohort of patients with a first prescription for antidepressants (SSRI or tricyclic antidepressants) was extracted from the Dutch Mondriaan and Spanish Base de datos para la Investigación Farmacoepidemiológica en Atención Primaria (BIFAP) general practice databases for the period 2001-2009. The net (total) effect of SSRI versus no SSRI on the risk of hip fracture was estimated using time-varying Cox regression, stratification and covariate adjustment using the PS, and MSM. In MSM, censoring was accounted for by inverse probability of censoring weights.The crude hazard ratio (HR) of SSRI use versus no SSRI use on hip fracture was 1.75 (95%CI: 1.12, 2.72) in Mondriaan and 2.09 (1.89, 2.32) in BIFAP. After confounding adjustment using time-varying Cox regression, stratification, and covariate adjustment using the PS, HRs increased in Mondriaan [2.59 (1.63, 4.12), 2.64 (1.63, 4.25), and 2.82 (1.63, 4.25), respectively] and decreased in BIFAP [1.56 (1.40, 1.73), 1.54 (1.39, 1.71), and 1.61 (1.45, 1.78), respectively]. MSMs with stabilized weights yielded HR 2.15 (1.30, 3.55) in Mondriaan and 1.63 (1.28, 2.07) in BIFAP when accounting for censoring and 2.13 (1.32, 3.45) in Mondriaan and 1.66 (1.30, 2.12) in BIFAP without accounting for censoring.In this empirical study, differences between the different methods to control for time-dependent confounding were small. The observed differences in treatment effect estimates between the databases are likely attributable to different confounding information in the datasets, illustrating that adequate information on (time-varying) confounding is crucial to prevent bias.

The purpose of this study is to evaluate the performance and validity of the case-crossover (CCO) and self-controlled case-series (SCCS) designs when studying the association between hip/femur fracture (HF) and antidepressant (AD) use in general practitioner databases. In addition, comparability with cohort and case-control designs is discussed.Adult patients with HF and who received an AD prescription during 2001-2009 were identified from UK's The Health Improvement Network (THIN) and the Dutch Mondriaan databases. AD exposure was classified into current, recent and past/non-use (reference). In the CCO, for each patient, a case moment (date of HF) and four prior control moments at -91, -182, -273 and -365 days were defined. In SCCS, incidence of HF was compared between exposure states. Conditional logistic regression was used in the CCO and Poisson regression in the SCCS to compute odds ratios and incidence rate ratios, respectively. In CCO, we adjusted for time-varying co-medication and in SCCS for age.Adjusted estimates for the effect of current AD exposure on HF were higher in the CCO (co-medication-adjusted odds ratio, THIN: 2.24, 95% confidence interval [CI]: 2.04-2.47; Mondriaan: 2.57, 95%CI [1.50, 4.43]) than in the SCCS (age-adjusted incidence rate ratio, THIN: 1.41, 95%CI [1.32, 1.49]; Mondriaan: 2.14, 95%CI [1.51, 3.03]). The latter were comparable with the traditional designs.Case-only designs confirmed the association between AD and HF. The CCO design violated assumptions in this study with regard to exchangeability and length of exposure, and transient effects on outcome. The SCCS seems to be an appropriate design for assessing AD-HF association.

BACKGROUND: Conditional on the propensity score (PS), treated and untreated subjects have similar distribution of observed baseline characteristics when the PS model is appropriately specified. The performance of several PS balance measures in assessing the balance of covariates achieved by a specific PS model and selecting the optimal PS model was evaluated in simulation studies. However, these studies involved only normally distributed covariates. Comparisons in binary or mixed covariate distributions with rare outcomes, typical of pharmacoepidemiologic settings, are scarce. METHODS: Monte Carlo simulations were performed to examine the performance of different balance measures in terms of selecting an optimal PS model, thus reduction in bias. The balance of covariates between treatment groups was assessed using the absolute standardized difference, the Kolmogorov-Smirnov distance, the Levy distance, and the overlapping coefficient. Spearman's correlation coefficient (r) between each of these balance measures and bias were calculated. RESULTS: In large sample sizes (n ≥ 1000), all balance measures were similarly correlated with bias (r ranging between 0.50 and 0.68) irrespective of the treatment effect's strength and frequency of the outcome. In smaller sample sizes with mixed binary and continuous covariate distributions, these correlations were low for all balance measures (r ranging between 0.11 and 0.43), except for the absolute standardized difference (r = 0.51). CONCLUSIONS: The absolute standardized difference, which is an easy-to-calculate balance measure, displayed consistently better performance across different simulation scenarios. Therefore, it should be the balance measure of choice for measuring and reporting the amount of balance reached, and for selecting the final PS model. Copyright © 2014 John Wiley & Sons, Ltd.

Conditional on the propensity score (PS), treated and untreated subjects have similar distribution of observed baseline characteristics when the PS model is appropriately specified. The performance of several PS balance measures in assessing the balance of covariates achieved by a specific PS model and selecting the optimal PS model was evaluated in simulation studies. However, these studies involved only normally distributed covariates. Comparisons in binary or mixed covariate distributions with rare outcomes, typical of pharmacoepidemiologic settings, are scarce.Monte Carlo simulations were performed to examine the performance of different balance measures in terms of selecting an optimal PS model, thus reduction in bias. The balance of covariates between treatment groups was assessed using the absolute standardized difference, the Kolmogorov-Smirnov distance, the Lévy distance, and the overlapping coefficient. Spearman's correlation coefficient (r) between each of these balance measures and bias were calculated.In large sample sizes (n ≥ 1000), all balance measures were similarly correlated with bias (r ranging between 0.50 and 0.68) irrespective of the treatment effect's strength and frequency of the outcome. In smaller sample sizes with mixed binary and continuous covariate distributions, these correlations were low for all balance measures (r ranging between 0.11 and 0.43), except for the absolute standardized difference (r = 0.51).The absolute standardized difference, which is an easy-to-calculate balance measure, displayed consistently better performance across different simulation scenarios. Therefore, it should be the balance measure of choice for measuring and reporting the amount of balance reached, and for selecting the final PS model.

Objectives To assess the current practice of propensity score (PS) analysis in the medical literature, particularly the assessment and reporting of balance on confounders. Study Design and Setting A PubMed search identified studies using PS methods from December 2011 through May 2012. For each article included in the review, information was extracted on important aspects of the PS such as the type of PS method used, variable selection for PS model, and assessment of balance. Results Among 296 articles that were included in the review, variable selection for PS model was explicitly reported in 102 studies (34.4%). Covariate balance was checked and reported in 177 studies (59.8%). P-values were the most commonly used statistical tools to report balance (125 of 177, 70.6%). The standardized difference and graphical displays were reported in 45 (25.4%) and 11 (6.2%) articles, respectively. Matching on the PS was the most commonly used approach to control for confounding (68.9%), followed by PS adjustment (20.9%), PS stratification (13.9%), and inverse probability of treatment weighting (IPTW, 7.1%). Balance was more often checked in articles using PS matching and IPTW, 70.6% and 71.4%, respectively. Conclusion The execution and reporting of covariate selection and assessment of balance is far from optimal. Recommendations on reporting of PS analysis are provided to allow better appraisal of the validity of PS-based studies.

Background Cardiovascular disease (CVD) has a disproportionate effect on mortality among the poorest people. We assessed the impact on CVD and all-cause mortality of the world's largest conditional cash transfer, Brazil’s Bolsa Família Programme (BFP). Methods We linked administrative data from the 100 Million Brazilian Cohort with BFP receipt and national mortality data. We followed individuals who applied for BFP between 1 January 2011 and 31 December 2015, until 31 December 2015. We used marginal structural models to estimate the effect of BFP on all-age and premature (30–69 years) CVD and all-cause mortality. We conducted stratified analyses by levels of material deprivation and access to healthcare. We checked the robustness of our findings by restricting the analysis to municipalities with better mortality data and by using alternative statistical methods. Results We studied 17 981 582 individuals, of whom 4 855 324 were aged 30–69 years. Three-quarters (76.2%) received BFP, with a mean follow-up post-award of 2.6 years. We detected 106 807 deaths by all causes, of which 60 893 were premature; and 23 389 CVD deaths, of which 15 292 were premature. BFP was associated with reductions in premature all-cause mortality [hazard ratio (HR) = 0.96, 95% CI = 0.94–0.98], premature CVD (HR = 0.96, 95% CI = 0.92–1.00) and all-age CVD (HR = 0.96, 95% CI = 0.93–1.00) but not all-age all-cause mortality (HR = 1.00, 95% CI = 0.98–1.02). In stratified and robustness analyses, BFP was consistently associated with mortality reductions for individuals living in the two most deprived quintiles. Conclusions BFP appears to have a small to null effect on premature CVD and all-cause mortality in the short term; the long-term impact remains unknown.