Objective To aid trialists, systematic reviewers and others, we evaluated the degree of standardisation of control measure reporting that has occurred in atrial fibrillation (AF) and venous thromboembolism (VTE) studies since 2000; and attempted to determine whether the prior recommendation of reporting ≥2 measures per study has been employed.
Design Systematic review.
Search strategy We searched bibliographic databases (2000 to June 2013) to identify AF and VTE studies evaluating dose-adjusted vitamin K antagonists (VKAs) and reporting ≥1 control measure. The types of measures reported, proportion of studies reporting ≥2 measures and mean (±SD) number of measures per study were determined for all studies and compared between subgroups.
Data extraction Through the use of a standardised data extraction tool, we independently extracted all data, with disagreements resolved by a separate investigator.
Results 148 studies were included, 57% of which reported ≥2 control measures (mean/study=2.13±1.36). The proportion of time spent in the target international normalised ratio range (TTR) was most commonly reported (79%), and was frequently accompanied by time above/below range (52%). AF studies more frequently reported ≥2 control measures compared with VTE studies (63% vs 37%; p=0.004), and reported a greater number of measures per study (mean=2.36 vs 1.53; p<0.001). Observational studies were more likely to provide ≥2 measures compared with randomised trials (76% vs 33%; p<0.001) and report a greater number of measures (mean=2.58 vs 1.63; p<0.001). More recent studies (2004–2013) reported ≥2 measures more often than older (2000–2003) studies (59% vs 35%; p=0.05) and reported more measures per study (mean=2.23 vs 1.48; p=0.02).
Conclusions While TTR was often utilised, studies reported ≥2 measures of VKA control only about half of the time and lacked consistency in the types of measures reported. A trend towards studies reporting greater numbers of VKA control measures over time was observed over our review time horizon, particularly, with AF and observational studies.
Statistics from Altmetric.com
Strengths and limitations of this study
This large systematic review (N=148 studies) adds to the existing literature by providing updated results and new data regarding the frequency and consistency of vitamin K antagonists (VKA) control measure reporting.
While the previous review by Fitzmaurice et al included studies of all VKA indications; ours evaluated atrial fibrillation (AF) and VTE studies only.
Unlike previous reviews, our systematic review examined VKA control measure reporting over time and differences in reporting between AF and VTE studies and randomised trials and observational studies. In addition, we explored the way in which VKA control measures are concomitantly reported in studies.
Adjusted-dose vitamin K antagonists (VKAs) are frequently used, and are the standard-of-care anticoagulants that most new oral anticoagulants for the prevention of thrombotic events in patients with atrial fibrillation (AF) and following venous thromboembolism (VTE) are compared with.1–10 VKAs have substantial evidence from clinical trials supporting their efficacy, and their use is endorsed by multiple national guidelines11 ,12; however, they are often underused due to difficulty in maintaining the international normalised ratio (INR) in the narrow therapeutic range (often 2.0–3.0).13 ,14
Fitzmaurice et al15 performed a systematic review of studies published between 1995 and 1999 in order to evaluate the manner in which VKA control was reported and to provide recommendations for reporting of VKA control measures (parameters used to summarise the level of anticoagulation). Their review found that a wide range of measures had been used in the literature, but with little consistency between studies. Since studies also suggest different VKA control measures (eg, percentage of time spent in range, proportion of tests in range, point prevalence) used in the same population can result in different conclusions regarding the quality of VKA control,16–18 researchers recommended ≥2 VKA control measures be reported per study.
In order to aid researchers (eg, clinical trialists and systematic reviewers) and other end users, we performed a systematic review to assess the degree of standardisation in VKA control measure reporting that has occurred in AF and VTE studies since the publication of the paper by Fitzmaurice et al; and to determine whether their recommendation of reporting ≥2 control measures has been widely employed.
A systematic review of MEDLINE, CENTRAL and EMBASE (from 2000 to June 2013) was conducted to identify published studies (English full-text randomised controlled trials, prospective cohort studies or retrospective analyses) including at least one dose-adjusted VKA treatment arm and reporting a minimum of one VKA control measure in adult patients being treated for AF or VTE as their primary reason for anticoagulation. Our search strategy for MEDLINE (PubMed) is provided in online supplementary appendix 1. Studies were excluded if they included <50 patients or planned to treat patients for <3 months. Manual backwards citation tracking of references from identified studies and review articles was also performed to identify additional relevant studies. All citations were screened by two independent investigators (ESM and J-SS) with discrepancies resolved by a third investigator (CIC).
Through the use of a standardised data extraction tool, we independently extracted all data (ESM, J-SS and JH), with disagreements resolved by a separate investigator (CIC). Collected study-level data included: study identifier and year of publication; indication(s) for VKA therapy; sample size; study design (prospective, retrospective or randomised study); duration of VKA treatment; mean age of participants; and the type(s) of VKA used. The types of VKA control measures reported were also extracted from each study. These included (but were not limited to): percentage of time in range (target international normalised ratio (TTR), calculated using Rosendaal's linear interpolation method19), below and/or above range, TTR in an extended range (ie, 1.8–3.2) and extreme ranges (ie, <1.5 and/or >5.0); proportion of INR measurements in range (PINRR), below and/or above extreme range; mean/median INR; mean/median VKA dose; frequency of INR monitoring (number of INR measures per patient over the course of the study); INR variability; INR monitoring interval (number of days between each INR measure); point prevalence (eg, the proportion of patients in range and/or out of range, proportion of patients in range >50% of time or proportion of patients with ≥50% of INR measures <3.0); number of VKA dosage changes; INR measure after a previously subtherapeutic or supratherapeutic INR; proportion of patients with ≥1 INR measure below range after reaching an adequate INR; number of days until the next INR measure after an extreme measure; proportion of days with treatment stability (two consecutive INR measures in range); days to reach a therapeutic INR; mean time until stable and minimum and maximum INR values per patient.
The types of measures reported were summarised and displayed using tables and figures, and the proportion of studies reporting ≥2 measures along with the mean number of measures per study (±SD) were reported for all identified studies. We also compared these same end points between select study subgroups (primary indication for anticoagulation (AF vs VTE); study design (randomised trial vs observational study); and year of publication (2000–2003 vs 2004–2013)). The year categorisations were chosen based on the year of publication of the review by Fitzmaurice et al.15 Finally, in order to assess the concomitant use of VKA control measures within studies, a diagram depicting per study measure linkages was created.
Between-group comparisons were made using χ2 tests (or Fisher's exact tests, where appropriate) for categorical data and unpaired t tests for continuous data. A p value of <0.05 was considered statistically significant in all situations. Statistical analysis was performed using SPSS V.17.0 (SPSS Inc, Chicago, Illinois, USA).
Of the 5301 citations initially identified, 1119 full-text articles were reviewed for inclusion. A total of 148 studies met all inclusion and exclusion criteria and were included in the analysis (figure 1, table 1).1–9 ,18 ,20–157 Of note, 112 VKA studies were excluded from our systematic review because they did not report a VKA control measure although study participants were receiving a VKA for AF or VTE as their primary reason for treatment for greater than 3 months.
Overall, 57% of studies reported ≥2 VKA control measures (mean/study=2.13±1.36; table 2). TTR was the most common measure reported (79%), and in a little more than half of these studies, was accompanied by the proportion of time above and/or below range. Other common metrics (used in ≥20% of studies) included mean/median INR, frequency of INR monitoring, INR testing interval and the proportion of patients in/out of range. Subgroup analysis found AF studies were 1.7-fold more likely than VTE studies (table 3), observational studies were more than twice as likely as randomised trials (p≤0.05 for all comparisons; table 4) and recently published studies were 70% more likely than older studies (2000–2003; table 5) to report ≥2 control measures. Moreover, the AF, observational and later time period study subgroups were also more likely to report a greater absolute number of measures per study (p<0.02 for all comparisons). When studies that included a new oral anticoagulant (n=30; all published after 2003) were analysed exclusively, only eight (26.7%) reported ≥2 VKA control measures (mean/study=1.37±0.72). At the same time, however, TTR was reported in all but five studies. Finally, AF and observational studies were more likely to report less common metrics, such as extended range time in the therapeutic range, INR testing interval and frequency of INR monitoring (p<0.05 for all comparisons).
Our assessment of the concomitant use of VKA control measures in identified studies suggested there was little consistency in their use (figure 2). TTR (the most frequently reported measure overall) was most often reported with mean INR, frequency of INR monitoring and INR testing interval.
We performed a systematic review to assess the degree of standardisation of VKA control measures reported in AF and VTE studies since 2000, and to determine the proportion of studies reporting ≥2 control measures. We found that while TTR was frequently reported in identified studies; other measures were more sporadically provided. Our analysis also demonstrated AF studies (compared with VTE studies), observational studies (compared with randomised trials) and more recently published studies (2004–2013) (compared with older ones) were more likely to report ≥2 VKA control measures per study and report a greater absolute number of measures per study. New oral anticoagulant studies utilised TTR quite frequently (>80% of the time), suggesting further standardisation in VKA control measure reporting. Finally, we observed little consistency in the combinations of measures used in identified studies.
We believe the results of our systematic review extend current knowledge regarding the frequency and consistency of VKA control measure reporting in anticoagulation studies, and can serve as a valuable tool for clinical trialists and systematic reviewers. The aforementioned review performed by Fitzmaurice et al15 included only 15 studies across varying indications (not just AF and VTE), making it only a fraction of the size of our own and suggesting that direct comparison between these systematic reviews should be made with caution. Fitzmaurice et al found 60% of VKA studies published between 1995 and 1999 reported ≥2 control measures (mean=1.93/study), but with a wide variation in the type of measures reported. TTR (47%), mean/median INR (33%), PINRR (40%) and mean/median warfarin dose (33%) were the most frequently reported VKA control measures identified in their review; however, none of their studies reported point prevalence despite its easy calculation and recommended use at the time.158
While our systematic review appears to confirm a number of findings of Fitzmaurice et al, our review also suggests that since 2000, additional measures of VKA control—including point prevalence—have become at least to some degree more common in the anticoagulation literature. Moreover, our findings of discrepancies in the number of control measures reported between AF and VTE studies and observational and randomised studies are novel.
There are a number of reasons why reporting multiple measures of VKA control in anticoagulation studies (as originally suggested by Fitzmaurice et al) seems wise. First, by reporting multiple measures of control, the likelihood that potentially comparable studies share at least one measure in common is increased. Furthermore, studies suggesting different VKA control measures may yield disparate findings even when utilised in the same patient population.16–18 In a retrospective cohort study of 633 patients undergoing anticoagulation with a VKA, Schmitt et al17 observed 24%, 24% and 22% absolute differences between TTR and PINRR estimates of INR control and 22%, 26% and 17% absolute differences between TTR and point prevalence (cross-sectional) estimates of INR control after 2-month, 3-month and 6-month time intervals of follow-up, respectively. Moreover, in a randomised trial of 367 patients receiving a VKA, Fitzmaurice et al16 demonstrated up to a 9% variance between TTR, point prevalence and PINRR estimates in the same population. Finally, a retrospective study of 1511 patients performed by Ansell et al18 provided yet more evidence to support population differences in VKA control estimates when different measures are used, demonstrating a 5–10% absolute difference between TTR and PINRR estimates for patients from five different countries.
A final reason for including multiple measures on VKA control in anticoagulation studies is that different VKA control measures have their own unique strengths and weaknesses.17 While TTR takes into account actual days in the target INR range (typically by assuming values vary linearly between two measures19), its calculation is more complex than other measures; it makes assumptions about INR values between actual tests and can be biased by extreme out-of-range INR values. In addition, while we were not able to assess this in our systematic review because of a lack of consistent reporting, there appears to be variability in what INR values are included in TTR calculations, with some studies excluding INR values occurring during the initiation phase (ie, first week) and/or around temporary interruptions of a VKA.6 ,8 PINRR is a simpler measure to calculate than TTR; it requires only one INR measurement per patient and is not influenced by the extent INRs are out of range; nevertheless, it fails to take into account actual days of anticoagulation like TTR and may underestimate control when more frequent INR testing occurs in unstable patients. Point prevalence is perhaps the simplest measure to calculate because it takes only one time point into consideration (a cross-sectional method), and like PINRR, it is not influenced by the extent an INR value is out of range; however, unlike the aforementioned methods, point prevalence takes individual patients into account. Finally, it is worth noting that VKA control measures may tend to stabilise over time, suggesting duration of study follow-up should be considered when interpreting a control measure.
On the basis of the results of our systematic review, we agree with the previous recommendation of Fitzmaurice et al of reporting at least two measures of VKA control. However, we would like to emphasise that while we recommend multiple measures be reported, we are by no means suggesting that the quantity of measures reported is more important than the quality of the measures. For this reason, we further suggest TTR be one of the measures because of its frequent study in the literature (use in studies and linkage to anticoagulation outcomes).
There are several limitations of our systematic review worth discussion. First, like any other systematic review, the possibility that we missed eligible studies could exist. However, we consider this risk to be minimal due to our systematic search strategy and manual backwards citation tracking. In addition, the large number of included studies in this review lessens the impact that missed studies might have on our overall conclusions. Next, it is reasonable to question the inclusion of mean/median warfarin dose as a true measure of VKA control, since unlike other measures, it does not consider INR values. However, we opted to include it as a measure in order to stay consistent with the methods of the prior review by Fitzmaurice et al.15 Finally, the possibility that journal word limits may have played some role in the under-reporting of VKA control measures should be considered.
VKA studies lack consistency in the types and combinations of control measures reported. A trend towards studies reporting greater numbers of VKA control measures over time was observed over our review time horizon, particularly, with AF and observational studies. The findings of this systematic review should be taken into consideration by researchers when performing future work in this area.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online supplement
Contributors ESM and CIC participated in study concept and design, drafting of the manuscript, administrative, technical and material support. ESM, JH, J-SS and CIC participated in acquisition of data, analysis and interpretation of data, critical revision of the manuscript for important intellectual content. CIC participated in study supervision. ESM and CIC had full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. All authors read and approved the final manuscript.
Funding This study was funded by Janssen Scientific Affairs. The authors maintained full control over the design and conduct of the study; collection, management, analysis and interpretation of the data and preparation and review of the manuscript.
Competing interests CIC has received honoraria for participation on advisory boards and speaker's bureaus and has received research funding from Janssen Scientific Affairs, LLC.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement Dataset available from the corresponding author via email.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.