omit the page number.
APA Citation Generator
Get started
APA references generally include information about the author , publication date , title , and source . Depending on the type of source, you may have to include extra information that helps your reader locate the source.
It is not uncommon for certain information to be unknown or missing, especially with sources found online. In these cases, the reference is slightly adjusted.
Missing element | What to do | Reference format |
---|---|---|
Author | Start the reference entry with the source title. | Title. (Date). Source. |
Date | Write “n.d.” for “no date”. | Author. (n.d.). Title. Source. |
Title | Describe the work in square brackets. | Author. (Date). [Description]. Source. |
On the first line of the page, write the section label “References” (in bold and centered). On the second line, start listing your references in alphabetical order .
Apply these formatting guidelines to the APA reference page:
On the reference page, you only include sources that you have cited in the text (with an in-text citation ). You should not include references to personal communications that your reader can’t access (e.g. emails, phone conversations or private online material).
In addition to the APA Citation Generator, Scribbr provides many more tools and resources that help millions of students and academics every month.
Know the Differences & Comparisons
While writing an article, one must quote or refer to the original source of the information, fact or idea, from which it has been taken. It not just to supports your own points, but also to prevent plagiarism and denote that a variety of sources are used to write the piece.
And so, the author provides citations, which are correspondingly mentioned under the head reference, indicating the complete details of the resource. In this article, we are going to talk about the differences between citation and reference.
Comparison chart.
Basis for Comparison | Citation | Reference |
---|---|---|
Meaning | Citation is a way of disclosing within the main body, that the quote, image, chart, statistics, etc. are taken from an outside source. | Reference is a list which contains all the sources which have been sought or cited while writing the article or assignment. |
Use | It informs the readers, the basic source of information. | It informs the reader, the complete source of information. |
Purpose | To indicate the source of the material taken. | To support or criticize an argument or point. |
Placement | Presented in the bracket. | Presented as endnote or end of the document. |
Information | It contains information like publication year and last name of the author. | It contains information like publication date, title of book/journal, author's name, page number. |
In the citation, the author cites or refer to the source in the text to represent that the information is derived from an external source and to mention that source, in brief. Basically, it is an abbreviated reference, which you can find in the main body of the article or assignment, addressing the source of a quote, image, video, paraphrase, chart, table, etc. Due to this very reason, it is also called as “in-text citation”, which includes a set of parenthesis.
Simply put, it is a short notation, within the intellectual work, which points out to a complete notation, at the end of the page, providing full details of the source i.e. all the authors or publications which have been cited, are to be mentioned in the list of references.
Citation is like a credit to the author, editor or publisher, for their work and to help the readers in consulting the same source when they want more information in this regard. While citing the source of information in the document, you indicate the last name of author and year of publication.
Reference can be understood as the list of items which you have read and considered in your piece of work. While providing references, the author actually tells its readers about what kind of source he/she has used in the document.
Along with that it also helps the readers to identify the difference between the author’s words, theories and ideas and those of other authors. Further, it helps the reader to refer to the source for further information in that area, whenever required.
You can find references at the end of the document or article (before bibliography), in alphabetical order, by the first or main author’s last name. One should always use a genuine, reliable and authentic source of information, to ensure support, credence and authority, to the information, ideas and arguments stated.
Reference can be given to books, articles from journals, legal documents, webpage, blogs, official report of government departments and agencies, interview transcripts, conference papers, newspaper articles, films, television, video, etc.
The difference between citation and reference can be drawn clearly on the following grounds:
Basically, reference and citation are provided to give the user or reader, the source of facts, images, statistics, charts, tables and diagrams, which are a part of the article or assignment. It may also be used when you are talking about a theory, method or model, discovered by or linked to a specific person or writer.
With the help of these two, the author shows acknowledgement to those scholars, whose work or ideas has been used in the article or assignment.
Your email address will not be published. Required fields are marked *
Save my name, email, and website in this browser for the next time I comment.
This page addresses when to include digital object identifiers (DOIs) and uniform resource locators (URLs) in APA Style references. Also check out the related topic of when to include database information in references .
The DOI or URL is the final component of a reference list entry. Because so much scholarship is available and/or retrieved online, most reference list entries end with either a DOI or a URL.
Follow these guidelines for including DOIs and URLs in references:
DOIs and URLs are covered in the seventh edition APA Style manuals in the Publication Manual Sections 9.34 to 9.36 and the Concise Guide Sections 9.34 to 9.36
Follow these guidelines to format DOIs and URLs:
https://doi.org/xxxxx
When a DOI or URL is long or complex, you may use shortDOIs or shortened URLs if desired.
The mysterious Altar Stone at Stonehenge might have been transported more than 700km from north east Scotland. Scientists have revealed the fascinating study about the stone circle but the revelation brings as many questions as it does answers.
FILE - The world heritage site of Stonehenge is seen in Wiltshire, England on Dec. 17, 2013. (AP Photo/Alastair Grant, File)
In this photo provided by researchers in August 2024, Stonehenge’s Altar Stone lies underneath two Sarsen stones in Wiltshire, England. (Nick Pearce/Aberystwyth University via AP)
WASHINGTON (AP) — The ancient ritual meaning of Stonehenge is still a mystery, but researchers are one step closer to understanding how the famous stone circle was created.
The unique stone lying flat at the center of the monument was brought to the site in southern England from near the tip of northeast Scotland, researchers reported Wednesday in the journal Nature. It’s not clear whether the 16-foot (5-meter) stone was carried by boat or across land — a journey of more than 460 miles (740 kilometers).
“It’s a surprise that it’s come from so far away,” said University of Exeter archaeologist Susan Greaney, who was not involved in the study.
For more than a hundred years, scientists believed that Stonehenge’s central sandstone slab — long called the “altar stone” — came from much closer Wales. But a study last year by some of the same researchers showed that the stone didn’t match the geology of Wales’ sandstone formations. The actual source of the stone remained unknown until now.
AP correspondent Karen Chammas reports on a new study into the origins of the Stone Henge altar stone.
For the study, the team was not permitted to chip away rocks at the site, but instead analyzed minerals in bits of rock that had been collected in previous digs, some dating back to the 1840s. They found a match in the sandstone formations of Orcadian Basin in northeast Scotland, a region that includes parts of the tip of the Scottish peninsula as well as the Orkney Islands.
“That geological ‘fingerprint’ isn’t repeated in any other area of sediment in the U.K.,” said Aberystwyth University geologist Nick Pearce, a study co-author.
Greaney said the difficult logistics of moving the stone such a long distance show a high level of coordination and cultural connection between these two regions of ancient Britain.
Stonehenge was constructed around 5,000 years ago, with stones forming different circles brought to the site at different times. The placement of stones allows for the sun to rise through a stone “window” during summer solstice. The ancient purpose of the altar stone — which lies flat at the heart of Stonehenge, now beneath other rocks — remains a mystery.
“Stonehenge isn’t a settlement site, but a place of ceremony or ritual,” said Heather Sebire, senior curator at English Heritage, who was not involved in the study. She said that past archaeological excavations had not uncovered evidence of feasting or daily living at the site.
Previous research has shown cultural connections — such as similarities in pottery styles — between the area around Stonehenge and Scotland’s Orkney Islands. Other stones at Stonehenge came from western Wales.
While Britain is dotted with other Neolithic stone circles, “the thing that’s unique about Stonehenge is the distance from which the stones have been sourced,” said Aberystwyth University’s Richard Bevins, a study co-author.
The Associated Press Health and Science Department receives support from the Howard Hughes Medical Institute’s Science and Educational Media Group. The AP is solely responsible for all content.
Advertisement
Supplemental digital content, hypotension prediction index is equally effective in predicting intraoperative hypotension during noncardiac surgery compared to a mean arterial pressure threshold: a prospective observational study.
Submitted for publication September 26, 2023. Accepted for publication March 22, 2024. Published online first on April 1, 2024.
This article is featured in “This Month in A nesthesiology ,” page A1.
This article is accompanied by an editorial on p. 421.
This article has a related Infographic on p. A16.
Supplemental Digital Content is available for this article. Direct URL citations appear in the printed text and are available in both the HTML and PDF versions of this article. Links to the digital files are provided in the HTML text of this article on the Journal’s Web site ( www.anesthesiology.org ).
Marijn P. Mulder , Mirjam Harmannij-Markusse , Libera Fresiello , Dirk W. Donker , Jan-Willem Potters; Hypotension Prediction Index Is Equally Effective in Predicting Intraoperative Hypotension during Noncardiac Surgery Compared to a Mean Arterial Pressure Threshold: A Prospective Observational Study. Anesthesiology 2024; 141:453–462 doi: https://doi.org/10.1097/ALN.0000000000004990
Download citation file:
The Hypotension Prediction Index is designed to predict intraoperative hypotension in a timely manner and is based on arterial waveform analysis using machine learning. It has recently been suggested that this algorithm is highly correlated with the mean arterial pressure itself. Therefore, the aim of this study was to compare the index with mean arterial pressure–based prediction methods, and it is hypothesized that their ability to predict hypotension is comparable.
In this observational study, the Hypotension Prediction Index was used in addition to routine intraoperative monitoring during moderate- to high-risk elective noncardiac surgery. The agreement in time between the default Hypotension Prediction Index alarm (greater than 85) and different concurrent mean arterial pressure thresholds was evaluated. Additionally, the predictive performance of the index and different mean arterial pressure–based methods were assessed within 5, 10, and 15 min before hypotension occurred.
A total of 100 patients were included. A mean arterial pressure threshold of 73 mmHg agreed 97% of the time with the default index alarm, whereas a mean arterial pressure threshold of 72 mmHg had the most comparable predictive performance. The areas under the receiver operating characteristic curve of the Hypotension Prediction Index (0.89 [0.88 to 0.89]) and concurrent mean arterial pressure (0.88 [0.88 to 0.89]) were almost identical for predicting hypotension within 5 min, outperforming both linearly extrapolated mean arterial pressure (0.85 [0.84 to 0.85]) and delta mean arterial pressure (0.66 [0.65 to 0.67]). The positive predictive value was 31.9 (31.3 to 32.6)% for the default index alarm and 32.9 (32.2 to 33.6)% for a mean arterial pressure threshold of 72 mmHg.
In clinical practice, the Hypotension Prediction Index alarms are highly similar to those derived from mean arterial pressure, which implies that the machine learning algorithm could be substituted by an alarm based on a mean arterial pressure threshold set at 72 or 73 mmHg. Further research on intraoperative hypotension prediction should therefore include comparison with mean arterial pressure–based alarms and related effects on patient outcome.
The Hypotension Prediction Index is an alarm system approved by the Food and Drug Administration used to predict a mean arterial pressure less than 65 mmHg for at least 1 min in the operating room and critical care environments
It is based on a proprietary algorithm derived using machine learning from components of the arterial waveform processed using pulse contour analysis methods
Preliminary reports have suggested that the Hypotension Prediction Index is highly correlated with mean arterial pressure and that a mean arterial pressure threshold alone might predict imminent hypotension
Using prospectively obtained observational single-center data from 100 patients undergoing elective noncardiac surgery with invasive arterial monitoring, the authors used correlation, receiver operating characteristic curves, and precision-recall curves analyses to characterize the relationship between the index and simultaneous mean arterial pressure
A mean arterial pressure threshold of 73 mmHg correlated 97% of the time with the default Hypotension Prediction Index alarm (greater than 85)
A mean arterial pressure threshold of 72 mmHg had the most comparable predictive performance
Predictive performance metrics were essentially identical for the index and the mean arterial pressure across all analyses, and both surpassed either linearly extrapolated mean arterial pressure or delta mean arterial pressure
The positive predictive value for either the index or a mean arterial pressure threshold of 72 mmHg were both low, suggesting a high rate of false alarms
Intraoperative hypotension is common in surgical patients and is associated with an increased risk for postoperative mortality, acute kidney injury and myocardial injury. 1 , 2 A method to predict hypotension could help anesthesiologists to treat hypotension in a proactive instead of a reactive way and thus ultimately prevent it. The Hypotension Prediction Index has recently been introduced as an innovative method, consisting of a validated machine learning model based on the characteristics of the arterial waveform. 3 It is specifically developed to predict hypotension, defined as a mean arterial pressure (MAP) less than 65 mmHg for at least 1 min, up to 15 min in advance. 3
Recently, a modeling study hypothesized that the performance of the Hypotension Prediction Index may be overestimated due to the data selection process in the development phase of the machine learning algorithm, which can introduce systematic bias. 4 This elaborate theoretical underpinning further substantiates the notion that MAP may indirectly be over-represented in the index. This is also supported by our recent observational pilot study revealing a high cross-correlation and no time difference between the Hypotension Prediction Index and MAP signals. 5 It is conceivable that with MAP available, the added predictive ability of the Hypotension Prediction Index algorithm is relatively limited, because it largely represents an inverse reflection of the concurrent MAP.
The predictive performance of the Hypotension Prediction Index was compared to several standard clinical hemodynamic variables, including MAP. 6 For the index, an area under the curve (AUC) of the receiver operating characteristic of 0.926 was reported for the prediction of hypotension 5 min in advance. The concurrent MAP was found to be the best predictor among the standard hemodynamic variables tested (AUC of 0.807). 6 However, in a recent erratum to this publication, it was clarified that different data analyses were applied for the Hypotension Prediction Index and other tested variables, and therefore it could not be concluded that the index outperforms concurrent MAP. 6 This erratum was not supported by numeric data or new results, so a formal direct comparison between the Hypotension Prediction Index and MAP is still lacking. Therefore, we set out to compare the Hypotension Prediction Index with MAP-based predictions in a large group of moderate- to high-risk noncardiac surgery patients and hypothesized that respective alarms are equally effective in predicting hypotension.
This single-center prospective, observational cohort study was conducted within the Department of Anesthesiology of the Medisch Spectrum Twente (Enschede, The Netherlands), a tertiary referral hospital. This article is prepared in accordance with the STROBE and STARD reporting guidelines. 7 , 8 Ethical approval for this study was waived by the local ethical committee (#K22-42). All consecutive adult patients aged 18 yr or older and undergoing moderate- to high-risk, noncardiac, elective surgery were included. All patients were scheduled for and received general anesthesia with the need for invasive blood pressure monitoring by means of an arterial line. The inclusion period ranged from March 2022 until July 2023 while aiming for a convenience sample size of 100 patients. There was no follow-up of this cohort. A subanalysis of the first 33 patients from this cohort was published in a research letter before. 5
In addition to standard care in line with current practice guidelines, the patients were monitored with the HemoSphere advanced monitor including the Acumen Hypotension Prediction Index software (version 2.1, Edwards Lifesciences, Inc., USA). The HemoSphere advanced monitor, including all available additional parameters and secondary screen, was not blinded. The attending anesthesiologists could manage blood pressure according to their preferences; no specific treatment protocol was used. Demographic data, medical history, and details on the surgery performed were obtained from the electronic health records of the patients enrolled. The measured and averaged MAP (in mmHg) and computed Hypotension Prediction Index values (unitless, ranging from 0 to 100) were downloaded from the HemoSphere monitor, with 20-s interval samples.
Data analysis was performed in Python (version 3.9, Python Software Foundation, USA). The Python script, featuring a documented, step-by-step methodology along with example data is available at https://github.com/crph-utwente/HPIvalidation . All data from the start of incision till the end of surgery were used for the analysis; no data were removed. Missing data and artifacts, where a single data point dropped 30 mmHg 3 or greater below its surrounding values, were replaced by a linear interpolation. The analysis included the incidence and severity of hypotension, defined as a MAP less than 65 mmHg for at least 1 min. 3 All other 20-s data points were defined as nonhypotensive. The “area under the threshold” was calculated by multiplying the depth of hypotension (in mmHg) below a MAP of 65 mmHg by the duration of time spent in hypotension (in min). This calculation yields an area value that takes both depth and duration of hypotensive events into account. Additionally, the area under the threshold was normalized by dividing it by the total duration of surgery, resulting in the “time-weighted average” in hypotension.
All statistical analysis were performed in RStudio (version 2023.09.1, RStudio, PBC, USA). The descriptive statistics are presented as means ± SD for normally distributed data, or median [25th, 75th percentiles] for data not normally distributed. The normality of distribution was assessed visually with histograms and Q-Q plots, and categorical data are presented as n (%).
A cross-correlation analysis between the time series of the Hypotension Prediction Index and the MAP was conducted per patient. This is a technique commonly used in the fields of signal processing and statistical time series analysis to quantify the similarity between the two signals as a function of the time shift relative to each other. 9 , 10 Both the index and MAP signals were normalized for this analysis by subtracting their means and dividing by their SDs. Because this method incorporates the temporal relation between data points, it is also used to estimate the time delay between two signals. The cross-correlation at zero time shift is essentially Pearson’s correlation. Because a nonlinear, but monotonic, relation was found between the Hypotension Prediction Index and MAP samples (see fig. 2 ) and the residuals generated during the correlation estimation process were not normally distributed, performing only Pearson’s correlation may formally not be completely sufficient. Therefore, a Spearman’s rank correlation was also calculated to better capture the relationship between the Hypotension Prediction Index values and the concurrent MAP values at zero time shift. The latter two correlation measures were calculated on a per-patient basis and for the entire pooled data set.
Calculation of mean arterial pressure (MAP)–based predictors of hypotension, including concurrent MAP, delta MAP, and linearly extrapolated MAP, illustrated based on a representative time frame of a patient.
Additionally, the timewise relationship between the alarms derived from the Hypotension Prediction Index and different MAP thresholds was evaluated. The Hypotension Prediction Index alarm was defined at the default value greater than 85, whereas different MAP thresholds were defined at concurrent values ranging from 65 to 80 mmHg retrospectively. The proportionate agreement between different MAP thresholds and the index alarm, defined as the percentage of time the thresholds were surpassed or not and the alarms were simultaneously on or off, was calculated. In addition, the Cohen’s kappa coefficient was analyzed, which takes into account the possibility of this agreement occurring by chance.
The performance of predicting hypotension within 5, 10, and 15 min in advance was evaluated for the Hypotension Prediction Index and concurrent MAP. The prediction windows were defined as follows: 1 to 5 min, 1 to 10 min, and 1 to 15 min. The first 1 min was excluded from the prediction windows, because this short prediction time is not deemed clinically relevant when aiming to prevent intraoperative hypotension by starting an intervention. The performance was assessed by means of a receiver operating characteristic curve and the AUC including 95% CI using the bootstrap method with 2,000 repetitions. These were constructed using a generalized estimating equations model that allowed to correct for repeated measures within a single patient. 11 Comparably, precision recall curves were also constructed, in which sensitivity (recall) is plotted against positive predictive value (precision) for each threshold value. The latter curves are more commonly used to assess the performance of tests in an imbalanced data set, because they hold more clinical relevance. For different Hypotension Prediction Index and MAP threshold sensitivity, specificity and positive and negative predictive values were assessed. Additionally, the threshold values at a minimal difference between sensitivity and both specificity and positive predictive values were defined.
For the 5 min prediction window, the performance of ∆MAP and linearly extrapolated MAP was also assessed. 12 ∆MAP was defined as the difference between the concurrent MAP and the MAP 5 min before. The linearly extrapolated MAP served as prediction of the MAP 5 min ahead and was determined on a trendline between concurrent MAP and MAP 5 min before, assuming linearity ( fig. 1 ).
A sensitivity analysis was performed to evaluate the potential effect of different data selection methods, (non)hypotension definitions, and prediction windows.
To achieve a more clinically relevant positive predictive value estimate, a forward analysis was conducted. 13 In this forward analysis, alarms were defined as a Hypotension Prediction Index greater than 85 or a MAP less than 72 mmHg for at least 1 min. From the onset of the alarm, i.e. , the first sample, the presence of hypotension was searched in the upcoming 1- to 5-min, 1- to 10-min, and 1- to 15-min windows. The positive predictive value was defined as the percentage of alarms truly followed by hypotension. Additionally, this forward analysis was repeated while excluding alarms that were followed by a sudden increase in MAP to correct for possible antihypotensive treatments that might conceal the potential predictive performance by preventing hypotension. Sudden increases in MAP were defined as MAP increase of more than 5 mmHg in 20 s or more than 8 mmHg in 2 min, when MAP was initially less than 70 mmHg. 6
A total of 100 patients were included in the study; no patients were excluded from the analysis. Table 1 shows the baseline characteristics of the study population: 49 of 100 (49%) American Society of Anesthesiologists (ASA) Physical Status II, 42 of 100 (42%) ASA III, and 9 of 100 (9%) ASA IV patients, with hypertension as the most common chronic disease (49 of 100, 49%), followed by diabetes mellitus (24 of 100, 24%). The mean age was 68 ± 11 yr, and about half of the patients (51 of 100, 51%) underwent gastrointestinal surgery. The total number of data points analyzed was 68,583, with 2,061 (3.0%) defined as hypotensive, grouped into 231 hypotensive events. Table 2 shows the hypotension metrics for this study population. Hypotension was present in 62 of 100 (62%) patients, with an area under the threshold of 5 [0, 35] mmHg · min and time-weighted average of 0.03 [0.0, 0.2] mmHg.
Baseline Characteristics of the Study Population
Hypotension Metrics
The maximal cross-correlation between the Hypotension Prediction Index signal and MAP signal was −0.91 ± 0.05 with 0.0 ± 0.0 min time shift. This indicates a strong inverse relation between the two signals and no time delay. For the pooled data set of the index and concurrent MAP data points, the Pearson’s correlation was −0.88. A sigmoidal monotonic relation between Hypotension Prediction Index and concurrent MAP values is visualized in the scatter plot in figure 2 . The Spearman’s rank correlation coefficient between the index and concurrent MAP values per patient was −0.99 ± 0.01, with a −0.97 correlation for the pooled data set.
In figure 2 , the default index alarm and the most corresponding MAP thresholds (72 and 73 mmHg) resulting from the analyses detailed below are also indicated. At least one Hypotension Prediction Index alarm was detected in 89 of 100 (89%) patients, with a concurrent MAP of 71 ± 4 mmHg at the alarm onset (histogram displayed in supplemental digital content, fig. S1, https://links.lww.com/ALN/D530 ). The proportionate agreement between the default index alarm and the different MAP thresholds (65 to 80 mmHg) ranges between 0.79 and 0.97. The specific results of the agreement and Cohen’s kappa analysis can be found in table 3 and more details in supplemental digital content 1, table S1 ( https://links.lww.com/ALN/D530 ). The highest proportionate agreement (0.97) and Cohen’s kappa coefficient (0.85) were found with a MAP threshold of 73 mmHg, indicating that this value has the highest agreement with the default Hypotension Prediction Index alarm. In figure 3 , the agreement between these alarms is displayed in the time frames of two representative hypotensive patients.
Agreement between Default Hypotension Prediction Index (Greater than 85) and Concurrent MAP-based Alarms
Scatter plot of Hypotension Prediction Index and concurrent mean arterial pressure (MAP) values of all 100 patients. Red points indicate presence of hypotension, and blue points indicate its absence within 1 to 5 min. Values for the default index and proposed MAP thresholds of 72 and 73 mmHg ( dashed lines ) were used.
Agreement of the Hypotension Prediction Index default alarm (greater than 85, blue ) and the proposed alarm based on a mean arterial pressure (MAP) threshold (greater than 73 mmHg, red ) during two representative time frames of patients experiencing hypotension, defined as MAP less than 65 mmHg ( dotted line ).
In figure 4A , the receiver operating characteristic curves of the Hypotension Prediction Index, concurrent MAP, ∆MAP, and linearly extrapolated MAP to predict hypotension up to 5 min are shown. The Hypotension Prediction Index and concurrent MAP receiver operating characteristic curves are almost identical, with an AUCs of 0.89 (0.88 to 0.89) and 0.88 (0.88 to 0.89), respectively, yielding a higher AUC than the other two methods: AUC of ∆MAP is 0.66 (0.65 to 0.67), and AUC of linearly extrapolated MAP is 0.85 (0.84 to 0.85). Figure 4B shows the precision recall curves for the four prediction methods within 5 min. Again, Hypotension Prediction Index and concurrent MAP are similar with overlapping CI of the AUC: 0.56 (0.55 to 0.57) and 0.55 (0.54 to 0.56). Linearly extrapolated MAP has a substantially lower precision recall AUC of 0.32 (0.32 to 0.33). The coinciding performance curves for the index and concurrent MAP are also present at the other prediction windows. In figures S2 and S3 (supplemental digital content 1, https://links.lww.com/ALN/D530 ), receiver operating characteristic and precision recall curves for the 10- and 15-min prediction windows can be found.
Performance curves for hypotension prediction with Hypotension Prediction Index ( blue line ), concurrent mean arterial pressure (MAP; red dashed line ), linearly extrapolated MAP ( green line ), and ∆MAP ( orange line ) within a time window of 1 to 5 min. ( A ) Receiver operating characteristic curve. ( B ) Precision recall curve. AUC, area under the curve.
The performance metrics of a selection of the Hypotension Prediction Index alarms with corresponding MAP thresholds and the other MAP-based methods are reported in table 4 . When predicting hypotension within 5 min, the default Hypotension Prediction Index alarm has a sensitivity of 72.4 (71.1 to 73.8)%, specificity of 89.9 (89.7 to 90.1)% and positive predictive value of 31.9 (31.3 to 32.9)%. The MAP threshold set at 72 mmHg was the most comparable to the default Hypotension Prediction Index alarm, gaining a 72.6 (71.3 to 73.9)% sensitivity, 90.3 (90.1 to 90.5)% specificity, and 32.9 (32.2 to 33.6)% positive predictive value. When balancing sensitivity and specificity, the thresholds are 63 for the Hypotension Prediction Index, 75 mmHg for concurrent MAP, −3 mmHg for ∆MAP, and 73 mmHg for linearly extrapolated MAP. Those thresholds are 97 for the Hypotension Prediction Index, 67 mmHg for concurrent MAP, −16 mmHg for ∆MAP, and 57 mmHg for linearly extrapolated MAP when aiming for a minimal difference between sensitivity and positive predictive value. A more complete overview of the performance metrics for other Hypotension Prediction Index and MAP thresholds at different prediction windows can be found in supplemental digital content 1, tables S2 to S4 ( https://links.lww.com/ALN/D530 ).
Performance of Different Hypotension Prediction Index and MAP-based Thresholds in Predicting Hypotension
The sensitivity analysis revealed that absolute performance of both Hypotension Prediction Index and MAP-based prediction methods are affected by data processing, whereas they still perform equally, both being affected in the same way. The performance curves for different data selection and prediction times are displayed in figures S4 and S5 (supplemental digital content 1, https://links.lww.com/ALN/D530 ). At times, we observed an oscillatory behavior in the precision recall curves at low—but not clinically relevant—sensitivity, which is caused by few data points with extremely low Hypotension Prediction Index or MAP values.
The results of the forward analysis are shown in table 5 . The time to hypotension within 15 min from start of a Hypotension Prediction Index alarm was 2.3 [1.0, 7.3] min. The longer the prediction window, the higher the positive predictive value; all the values increase with correction for antihypotensive interventions. The positive predictive values for the Hypotension Prediction Index range between 21.5 and 42.1%; for a MAP alarm of 72 mmHg, they are systematically slightly higher, ranging between 24.3 and 52.0%.
Positive Predictive Values from Forward Analysis for Default Hypotension Prediction Index (Greater than 85) and MAP (Less than 72) Alarm
This observational clinical investigation in 100 surgical patients demonstrates a high correlation between the Hypotension Prediction Index and MAP values in real-world intraoperative practice. Specifically, the MAP thresholds of 72 and 73 mmHg were found to correspond closely to the default Hypotension Prediction Index alarm. When comparing the predictive performance of those methods using AUCs, those for Hypotension Prediction Index and concurrent MAP are virtually identical, while outperforming ∆MAP and linearly extrapolated MAP.
The cross-correlation analysis revealed that the MAP signal is highly similar to the index signal, being largely its mirrored version, as suggested before. 5 , 14–16 When accounting for the nonlinear relation between the two values, an even higher correlation was found (Spearman’s ρ = −0.99). This may well be due to a presumed selection bias in the training of the machine learning model 4 and the fact that MAP is related to the arterial pressure waveform features from which the index is derived. 3 Because no time delay between the two signals was found at maximum correlation, the Hypotension Prediction Index does not seem to be ahead of concurrent MAP in time. The cross-correlation analysis performed on a subgroup published earlier is confirmed by the results found in this larger study population. 5 These results indicate that the Hypotension Prediction Index value, although based on multiple waveform features, does essentially not comprise more or different predictive information than the MAP itself.
The Hypotension Prediction Index was originally not compared to concurrent MAP, but only to ∆MAP, which showed AUCs of 0.50 to 0.62 for predicting hypotension. 3 This low predictive power of ∆MAP is confirmed in the current study, ∆MAP being a parameter that does not take the actual absolute blood pressure into account. Instead, a linear extrapolation of the MAP, as suggested recently, includes both the change and absolute value of MAP 12 and shows a substantially higher performance. Therefore, linearly extrapolated MAP seems a better comparator than ∆MAP when evaluating machine learning models or other innovative methods to predict hypotension, but it still performs substantially less than concurrent MAP.
In our current study, we directly compared the Hypotension Prediction Index and concurrent MAP using identical data and methodology and found that the receiver operating characteristic curves are coinciding. This indicates that not only the overall performance of the two prediction methods is the same, as reflected by the AUCs, but that for every specific alarm threshold the same sensitivity and specificity are achieved. There is no added predictive value of the index over concurrent MAP for any chosen threshold. This is in agreement with the receiver operating characteristic curves in the supplementary material of a recent study on the Hypotension Prediction Index performance in liver transplant surgery. 17 In addition, the precision recall curves of the index and concurrent MAP are also almost identical. The sensitivity analyses revealed that despite the spectrum of studied definitions and the data selection and prediction windows, the index and MAP perform similar. This clinical study highlights the importance of a proper and accurate comparison between the Hypotension Prediction Index and other clinical measures to prevent ambiguous assumptions about its added value.
During this clinical study, no specific protocol was used for treating hypotension, allowing assessment of the net effect of the presence of the Hypotension Prediction Index. This resulted in real-world data of a large group of patients showing the potential of the Hypotension Prediction Index in the intended clinical context of use and without strict boundaries imposed by a rigid research protocol. We realize that our study represents an observational, nonblinded analysis, that might have led the attending anesthesiologists to act upon Hypotension Prediction Index alarms. The resulting performance measures, especially positive predictive value, may therefore be underestimated because hypotension might have been prevented by timely intervention. We aimed to retrospectively correct for this in the forward analysis. This limits general conclusions about the absolute performance of the index and comparison with blinded studies. However, MAP is never blinded in a routine clinical setting; for a proper comparison of both performances, it is therefore beneficial that the index is also visible to the clinician.
Compared to the original Hypotension Prediction Index article, the AUCs of the Index’s performance found in this study for 5-, 10-, and 15-min prediction are somewhat lower than reported before, when all AUCs were greater than 0.95. 3 In this study, all data points were included to prevent a selection bias, as recently underscored in this context, 4 and consequently reducing the performance of the index. In addition, we used prediction windows instead of a single prediction time point in the future. The current approach may thus result in a more accurate reflection of index performance in everyday clinical practice.
Although our study is confined to a single-center analysis, its adequate representation of real-world clinical data is underlined by a comparison with the recently published European multicenter prospective observational registry: EU HYPROTECT. 18 This registry contains data of 749 surgeries in which the Hypotension Prediction Index was also used in a nonblinded fashion and without a treatment protocol, showing comparable hypotension metrics: 59% incidence with 1 [0, 3] event per patient, 2 [0, 9] min duration, and 0.03 [0.00, 0.20] mmHg time-weighted average 18 (see also supplemental digital content 1, fig. S1, https://links.lww.com/ALN/D530 ).
Probably, the most clinically relevant measure is the positive predictive value resulting from the forward analysis. This indicates the likelihood of hypotension occurring upon the onset of an alarm, which would be the trigger to consider a therapeutic intervention. Our results reveal that only around a quarter to half of the Hypotension Prediction Index alarms are actually followed by hypotension. Despite performing a retrospective correction of possible preventive interventions, the positive predictive value still remains rather limited. The MAP alarm of 72 mmHg seems to perform even slightly better than the default Hypotension Prediction Index alarm. The positive predictive values for the index in this study are lower than in previous validation studies using a prediction window of 0 to 20 min. 13 , 19
Overall, the results of this study imply that the index’s performance in predicting hypotension can equally be achieved by adjusting MAP thresholds to higher set points than usually applied in conventional clinical routine. Notably, the default Hypotension Prediction Index alarm is fixed at 85, whereas theoretically a better balance between sensitivity and specificity could be achieved by setting a much lower alarm value (46 to 63; see table 4 ). This might not be adequate in a clinical setting, potentially inducing alarm fatigue and overtreatment from false positive alarms. Nevertheless, the Hypotension Predicting Index alarm could be set lower to improve sensitivity, for example, at 75 as it is currently proposed for the planned HYPE-2 trail 20 ; or alternatively, concurrent MAP thresholds could be set higher (75 to 79 mmHg; see table 4 ), as supported by our data. Depending on the clinical need, a higher index threshold (90 to 97) or lower MAP (67 to 71 mmHg) could also be chosen to achieve a better precision.
Additionally, an important limitation of the Hypotension Prediction Index is that it can only be used to predict hypotension defined as a MAP of less than 65 mmHg, whereas MAP thresholds can easily be adjusted to predict any desired blood pressure in a timely manner. Therefore, intraoperative alarms based on MAP thresholds seem more suitable than the index when aiming for patient-specific blood pressure targets in relation to the patient’s baseline blood pressure, co-morbidities, and type of surgery to balance the risks of organ ischemia and bleeding. 21 In addition, the predictive value of the MAP thresholds found in this study are limited to our specific population; optimal thresholds could be different for patient groups with different age, sex, and ASA score. 22
In general, innovative methodologies like the Hypotension Prediction Index are deemed beneficial to prevent intraoperative hypotension, as reflected by a recent systematic review. 23 The data synthesis of several randomized controlled clinical trials 24–28 showed a decrease in the occurrence, duration, and severity of hypotension during noncardiac surgery while using the index, yet these results were confined to limited and low- to very-low-quality evidence. 23 It still needs to be verified whether a MAP-based alarm would yield the same hypotension reduction in clinical practice as the Hypotension Prediction Index in every single patient at every single moment. The current study was not designed to investigate hypotension reduction, treatment decisions, or the use of any other parameters available the HemoSphere platform. However, hypotension was found to be less prevalent than in both the control and intervention groups of the randomized clinical trials as reflected by different hypotension metrics. 23 However, this might also be based on a psychologic effect that arises from the presence of the Hypotension Prediction Index technique, its novelty, and potentially induced increased awareness, leading to more proactive hemodynamic management. The same phenomenon could be present in those randomized controlled trails. Moreover, because most studies used a treatment protocol in addition to the Hypotension Prediction Index, it remains unresolved which of the two would lead to improved outcomes.
Hypotension Prediction Index and concurrent MAP values are highly correlated, their respective alarms and thresholds show strong agreement and interchangeable performance in predicting intraoperative hypotension. A MAP threshold set at 72 or 73 mmHg seems a straightforward clinical alternative to the Hypotension Prediction Index. It is important to compare new hypotension prediction methods against parameters derived from routine hemodynamic monitoring, notably concurrent MAP, to study their added value including patient outcome.
Support was provided solely from institutional and/or departmental sources.
Dr. Mulder, Dr. Donker, and Dr. Fresiello provide research consultancy to Maquet Critical Care AB (Solna, Sweden). Dr. Donker and Dr. Fresiello have a research corporation with Sonion Nederland B.V. (Hoopddorf, The Netherlands), and Dr. Donker provides research consultancy to HBOX Therapies GmbH (Aachen, Germany). None of these authors receive personal fees. The other authors declare no competing interests.
Additional results: Figures and tables, https://links.lww.com/ALN/D530
Most viewed, email alerts, related articles, social media, affiliations.
Sign In or Create an Account
Objectives Disease activity control in patients with systemic lupus erythematosus (SLE) with corticosteroid and immunosuppressant withdrawal is a treatment goal. We evaluated whether this could be attained with sequential subcutaneous belimumab (BEL) and one cycle of rituximab (RTX).
Methods In this phase 3, double-blind BLISS-BELIEVE trial (GSK Study 205646), patients with active SLE initiating subcutaneous BEL 200 mg/week for 52 weeks were randomised to intravenous placebo (BEL/PBO) or intravenous RTX 1000 mg (BEL/RTX) at weeks 4 and 6 while stopping concomitant immunosuppressants/tapering corticosteroids; standard therapy for 104 weeks (BEL/ST; reference arm) was included. Primary endpoint: proportion of patients achieving disease control (SLE Disease Activity Index-2000 (SLEDAI-2K) ≤2; without immunosuppressants; prednisone equivalent ≤5 mg/day) at week 52 with BEL/RTX versus BEL/PBO. Major (alpha-controlled) secondary endpoints: proportion of patients with clinical remission (week 64; clinical SLEDAI-2K=0, without immunosuppressants/corticosteroids); proportion of patients with disease control (week 104). Other assessments: disease control duration, anti-dsDNA antibody, C3/C4 and B cells/B-cell subsets.
Results The modified intention-to-treat population included 263 patients. Overall, 16.7% (12/72) of BEL/PBO and 19.4% (28/144) of BEL/RTX patients achieved disease control (OR (95% CI) 1.27 (0.60 to 2.71); p=0.5342) at week 52. For major secondary endpoints, differences between BEL/RTX and BEL/PBO were not statistically significant. Anti-dsDNA antibodies and most assessed B cells/B-cell subsets were lower with BEL/RTX versus BEL/PBO. Mean disease control duration through 52 weeks was significantly greater with BEL/RTX versus BEL/PBO.
Conclusions BEL/RTX showed no superiority over BEL/PBO for most endpoints analysed; however, it led to significant improvements in disease activity markers compared with BEL/PBO. Further investigation of combination treatment is warranted.
Trial registration number NCT03312907 .
Data are available upon reasonable request. GSK is committed to publicly disclosing the results of GSK-sponsored clinical research that evaluates GSK medicines, and as such was involved in the decision to submit the results of this study (GSK Study 205646). Anonymised individual patient data and study documents can be requested for further research from https://www.gsk-studyregister.com/en/ .
This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/ .
https://doi.org/10.1136/ard-2024-225686
Request permissions.
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.
Achieving low disease activity in the absence of corticosteroids remains an important treatment goal in patients with systemic lupus erythematosus (SLE).
Preliminary studies suggested that sequential therapy with belimumab and rituximab in patients with active SLE may provide clinical benefits with an acceptable safety profile.
In this robust phase 3 BLISS-BELIEVE study of sequential belimumab and rituximab administration, while the primary and major secondary endpoints were not met, the mean duration of longest disease control was nominally significantly greater in patients treated with belimumab and rituximab sequential therapy compared with belimumab and placebo.
Significant reductions in anti-dsDNA antibody levels, CD19+ B cells and B-cell subsets, were observed with belimumab and rituximab sequential therapy versus belimumab and placebo.
This is the first randomised study in SLE to prospectively investigate a novel treatment regimen that incorporated a rapid reduction and withdrawal of standard immunosuppressants and thereby it sets the stage for future trials in SLE to aim for the stringent, clinically meaningful endpoint of remission off-therapy.
Despite traditional standard therapy (ST), including medications such as corticosteroids, antimalarials and immunosuppressants, a significant proportion of patients with systemic lupus erythematosus (SLE) do not achieve long-term disease control. 1–3 As prolonged exposure to glucocorticoids increases the risk of organ damage accrual, 4 management guidelines recommend tapering corticosteroids to ≤5 mg/day or withdrawing entirely when possible. 3 Furthermore, the treat-to-target principle has been embraced by SLE experts where, besides achieving low disease activity, treatment goals should be remission on-therapy and, even more aspirational, remission off-therapy. 5 6 Notwithstanding these ambitious goals, achieving disease control without corticosteroids remains an unmet treatment goal, but no randomised trial has previously employed these endpoints. 2 These goals could be achieved with disease-modifying therapies targeting the underlying pathogenesis of SLE. 7
B cells play a key role in the pathogenesis of SLE. 8 B-lymphocyte stimulator (BLyS) promotes B-cell activation and differentiation, 9–11 and elevated serum BLyS is associated with higher disease activity, disease relapse and increased numbers of autoantibody-secreting plasma cells. 9 12
Belimumab, a human IgG1λ monoclonal antibody that selectively inhibits soluble BLyS, is approved in combination with ST for treating SLE and lupus nephritis (LN). 13 14 Rituximab, a B-cell-depleting anti-CD20 monoclonal antibody, is used off-label in clinical practice as clinical trials have not demonstrated clinical efficacy. 8 15–18
The scientific justification for sequential therapy with belimumab and rituximab is twofold. Elevated BLyS levels, which occur following B-cell depletion, promote the maturation of autoreactive B cells by allowing them to bypass tolerance checkpoints and enter the immune repertoire. 19–22 Conversely, B-cell reconstitution without high levels of BLyS might result in tolerised B cells without autoreactivity and an enhanced clinical response. This potentially explains the inability of rituximab alone to show superiority over ST in SLE studies. 23 A second rationale for dual therapy is that although rituximab rapidly depletes peripheral B cells, tissue-resident B cells are less affected. 24–26 Thus, since belimumab was shown to increase circulating B-cell levels by either disrupting lymphocyte trafficking and preventing B cells from transmigrating from the blood into tissue or by preventing B cells from being retained at the tissue level, greater B-cell depletion may occur when belimumab is administered before rituximab. 27–34
Three small phase 2 trials, Synbiose (in patients with severe, refractory SLE, of whom the majority had active LN), Belimumab after B cell depletion therapy in patients with SLE (BEAT-LUPUS; in refractory SLE) and The Combination of Antibodies in Lupus Nephritis: Belimumab and Rituximab Assessment of Tolerance and Efficacy (CALIBRATE; in LN), have evaluated the efficacy and safety of rituximab and belimumab sequential therapy in improving disease biomarkers and outcomes, including B-cell depletion. 31 32 35 We have therefore hypothesised that the belimumab/rituximab combination may result in greater B-cell depletion and thus yield better control of disease activity with less need for concomitant immunosuppressants and corticosteroids than belimumab alone. To test this hypothesis, we evaluated the efficacy, safety and tolerability of belimumab with or without a single cycle of rituximab while stopping concomitant immunosuppressants and tapering corticosteroids in adult patients with SLE, using novel and stringent disease control endpoints.
This phase 3, multicentre, randomised, double-blind, placebo-controlled, 104-week study (GSK Study 205646, NCT03312907 ) was conducted at 71 sites globally between March 2018 and July 2021 ( online supplemental figure S1 ). The study protocol has been published previously. 28
Eligible patients were ≥18 years of age, fulfilled ≥4 of the 11 American College of Rheumatology classification criteria for SLE, 36 37 and were required to have serum positivity for anti-nuclear antibody (titre ≥1:80) and/or anti-dsDNA antibody (≥30 IU/mL) and an SLE Disease Activity Index-2000 (SLEDAI-2K) score ≥6. Patients with severe LN or severe, active central nervous system lupus were excluded. Full eligibility criteria have been published previously. 28
Patients were centrally randomised in a 1:2:1 ratio to: (1) BEL/PBO arm: subcutaneous (SC) belimumab 200 mg/week for 52 weeks and intravenous placebo at weeks 4 and 6; (2) BEL/RTX arm: belimumab 200 mg/week SC for 52 weeks and rituximab 1000 mg/week intravenous at weeks 4 and 6; or (3) BEL/ST arm: open-label belimumab 200 mg/week SC and ST for 104 weeks. The BEL/ST arm was included to provide an exploratory reference for assessing the relative performance of BEL/RTX versus BEL/PBO. Patients were stratified by their screening SLEDAI-2K score (≤9 or ≥10), immunosuppressant use (yes/no) and corticosteroid dose (prednisone equivalent ≤10 or >10 mg/day).
BEL/PBO and BEL/RTX patients received study treatment until week 52 (primary efficacy endpoint assessment) and then entered a 52-week, treatment-free (no belimumab or rituximab) double-blind observational phase through week 104 to allow assessment of the durability of remission. During the double-blind observational phase, comprehensive clinical assessments were scheduled at week 60, week 64 and every 8 weeks thereafter up to week 104. Data were also collected at any unscheduled visits that occurred. Patients were considered to have failed treatment if they received, at the investigator’s discretion, corticosteroids (>5 mg/day), any immunosuppressants, and/or open-label belimumab ( online supplemental materials ). BEL/ST patients received belimumab and continued their ST throughout the 104 weeks.
Antimalarials, non-steroidal anti-inflammatory drugs (NSAIDs), and/or prednisone-equivalent doses of ≤5 mg/day were allowed within protocol-defined limits in weeks 53 through 104 in all treatment groups.
All BEL/PBO and BEL/RTX patients discontinued immunosuppressants at or before week 4. Reinitiation of immunosuppressants post week 4 in BEL/PBO and BEL/RTX groups (on the physician discretion) resulted in treatment failure. In the BEL/ST group, an increase of immunosuppressant dose post week 12 resulted in treatment failure. After the initial 12 weeks of study treatment, all treatment groups were required to taper prednisone-equivalent dose to ≤5 mg/day by week 26. Patients were considered treatment failures if their dose exceeded 5 mg/day after week 26 ( online supplemental materials ).
The primary endpoint was the proportion of patients achieving a state of disease control at week 52, defined as a SLEDAI-2K score ≤2 achieved without immunosuppressants and with a prednisone-equivalent dose of ≤5 mg/day.
Major secondary endpoints were the proportion of patients in clinical remission at week 64, defined as a clinical SLEDAI-2K score of 0 (without a serological activity component) achieved without immunosuppressants and with a prednisone-equivalent dose of 0 mg/day, and the proportion of patients with a state of disease control (defined as per primary endpoint) at week 104. Additional secondary efficacy endpoints included analysis of the primary efficacy endpoint by baseline characteristics (age, race, sex, SLEDAI-2K, use of immunosuppressants, corticosteroid dose, complement 3/4 (C3/C4) and anti-dsDNA antibody levels, BLyS levels, and region). As the BEL/ST arm was open label, the primary and major secondary assessments were conducted by independent blinded assessors (IBA).
Secondary efficacy endpoints that were assessed by principal investigators (PI) included duration of disease control through 52 and 104 weeks in patients achieving disease control at ≥1 time point; time to disease control (defined as per primary endpoint); time to clinical remission (defined as per secondary endpoint); change from baseline in SLEDAI-2K score and the proportion of patients achieving Lupus Low Disease Activity State (LLDAS) 38 39 at weeks 52, 64 and 104.
Time-to-first severe flare was measured through week 104 by a modified SLE Flare Index (SFI), which did not consider severe flares that were triggered only by an increase in SLEDAI-2K score to >12, but included patients classified as a treatment failure by our stringent definition (eg, receiving corticosteroid dose >5 mg/day after week 26, restarting belimumab in year 2 (BEL/PBO or BEL/RTX)), as having a severe flare. Post hoc analyses were also performed to assess disease control at week 52 by baseline proteinuria (≤0.5 g/24 hours vs >0.5 g/24 hours), time to severe flare (modified SFI) without imputing treatment failures as a flare, proportions of patients with proteinuria shift from >0.5 g/24 hours to normal (≤0.5 g/24 hours) and prednisone-equivalent reduction by ≥25% from baseline to ≤7.5 mg/day during weeks 40 through 52.
Prespecified biomarker endpoints included changes from baseline in IgG, anti-dsDNA antibody and complement C3/C4 levels. Proportions of patients with anti-dsDNA antibody shift from positive at baseline to negative, and with C3/C4 shifts from low at baseline to normal/high were also assessed. B cells and B-cell subsets were assessed by flow cytometric characterisation. Central laboratory testing was used for all biomarker measurements. Biomarker endpoints were assessed statistically at weeks 52, 64 and 104.
Safety assessments included adverse events (AEs), serious AEs, AEs of special interest (malignant neoplasms; post-injection/infusion anaphylaxis/hypersensitivity reactions; all infections of special interest (opportunistic infections, herpes zoster, tuberculosis and sepsis); depression/suicide/self-injury) and deaths. AEs are reported on treatment for week 52 (year 1) and on study for week 104 (year 1+2).
The target sample size of 280 patients was expected to achieve ≥98% power to detect a treatment difference in the primary endpoint at the 5% two-sided level of significance, assuming response rates of 10% and 35% in the BEL/PBO and BEL/RTX groups (1:2 randomisation), respectively.
All randomised patients who received ≥1 dose of study treatment were included in the intention-to-treat (ITT) population. Efficacy endpoints were assessed using a modified ITT (mITT) population, which comprised the ITT population but excluded 29 patients from the BEL/ST group due to IBAs being potentially unblinded.
Primary and the two major secondary efficacy endpoints were alpha controlled using a prespecified testing order; endpoints could be tested in sequence (two-sided alpha=0.05), provided that statistical significance was achieved in all prior tests. These endpoints were analysed using logistic regression adjusted for baseline SLEDAI-2K score (≤9 or ≥10), immunosuppressant use (use or no use), prednisone-equivalent dose (≤10 mg/day or >10 mg/day) and treatment group. As the primary endpoint was not statistically significant, the p values for the two major secondary endpoints are nominal.
Other endpoints were not adjusted for multiplicity, and all associated p values are nominal.
Patients who met treatment-failure criteria, withdrew or missed assessments (and subsequent data collection was not possible) were regarded as non-responders for primary and secondary efficacy assessments and as experiencing a severe flare for the modified SFI flare endpoint. Other analyses were not alpha controlled; preplanned analyses were performed at weeks 52, 64 and 104.
The post hoc analysis for proteinuria is described in the online supplemental materials .
Biomarker outcomes were performed on observed data for ITT patients who were ongoing after 52 weeks and received both intravenous doses of rituximab or placebo or who remained on open-label belimumab through week 52 (for BEL/ST group) and did not have more than 28 consecutive days from baseline to week 51 without a belimumab dose. Year 1 and 2 analyses used data collected in year 1 before belimumab restart (BEL/PBO and BEL/RTX groups) or discontinuation (BEL/ST group) for patients who completed week 52 on treatment. No imputation was carried out for missing data in these analyses.
Safety endpoints were summarised descriptively.
Patients and/or public were not involved in the design, conduct, reporting or dissemination of this research.
This study (GSK Study 205646) was funded by GSK. GSK was involved in designing the study, contributed to the collection, analysis and interpretation of the data, supported the authors in the manuscript development, and funded the medical writing assistance. All authors approved the content of the submitted manuscript and were involved in the decision to submit the manuscript for publication.
Of 396 patients screened, 292 were included in the ITT population and 263 in the mITT population. Overall, 226/292 (77.4%) patients completed the study at week 104, with 215 of these patients having completed the scheduled study treatment up to Week 52. Across all groups, the most common reason for not completing the study was the patient’s decision to withdraw ( figure 1 ).
Patient disposition summary. mITT population excludes 29 patients from the BEL/ST group, due to IBAs being potentially unblinded. BEL, belimumab; IBA, independent blinded assessor; ITT, intention-to-treat; mITT, modified intention-to-treat; RTX, rituximab; ST, standard therapy.
Baseline characteristics were similar across treatment groups. The mean (SD) glucocorticoid dose was 10.0 (10.49) mg/day; most patients (80.8%) were taking glucocorticoids, followed by antimalarials (79.8%; table 1 ).
Patient baseline characteristics (ITT population, N=292)
At week 52, 16.7% of BEL/PBO-treated and 19.4% of BEL/RTX-treated patients achieved disease control, which was not statistically significant ( table 2 ). Of note, 25.5% of BEL/ST-treated patients achieved disease control. The observed differences in individual components of disease control between BEL/RTX and BEL/PBO groups are summarised in online supplemental table S1 , and the first reasons for not achieving disease control at Week 52 in online supplemental table S2 .
Summary of primary and major secondary efficacy endpoints, based on IBA assessment (mITT population,* N=263)
Proportions of patients off-treatment achieving clinicalremission at week 64 or disease control at week 104 showed no statistically significant differences between BEL/RTX and BEL/PBO ( table 2 ). The most common first reason for not achieving disease control at week 104 was continuation or restart of belimumab at the investigator’s discretion ( online supplemental table S2 ). In the ITT population, including those for whom it was the first reason for not achieving disease control, 54.8% (n=34/62) of BEL/PBO and 50.0% (n=62/124) of BEL/RTX patients continued or restarted belimumab in year 2.
Few patients achieved disease control or clinical remission that was sustained for at least 24 weeks and maintained through week 104, with no significant difference between BEL/RTX and BEL/PBO ( online supplemental table S3 ).
Prespecified subgroup analyses by baseline characteristics for the primary endpoint, disease control at week 52, showed no significant differences. However, there were numerically higher ORs in favour of BEL/RTX versus BEL/PBO for achieving disease control in patients with low C3/C4 levels and anti-dsDNA antibodies ≥30 IU/mL or SLEDAI-2K score ≥10 at baseline ( figure 2 ). When analysed by baseline proteinuria (post hoc) at week 52, disease control was achieved by 18.0% (n=11/61) of BEL/PBO and 20.8% (n=25/120) of BEL/RTX patients with baseline proteinuria ≤0.5 g/24 hours (OR (95% CI) 1.21 (0.54 to 2.71), p=0.6353), and by 9.1% (n=1/11) of BEL/PBO and 12.5% (n=3/24) of BEL/RTX patients with baseline proteinuria >0.5 g/24 hours (OR (95% CI) 1.74 (0.12 to 25.15), p=0.6845).
Disease control* by baseline characteristics subgroups at week 52, based on IBA assessment (mITT population † ; N=263). Note: OR (95% CI) and p value are from a logistic regression model with covariates: baseline SLEDAI-2K, baseline immunosuppressants, baseline prednisone-equivalent dose and treatment group (however, covariates were excluded from the corresponding subgroup models, for example, baseline SLEDAI-2K was not included in analysis by baseline SLEDAI-2K). *Disease control is defined as a SLEDAI-2K score ≤2 achieved without immunosuppressants and with a prednisone-equivalent dose of ≤5 mg/day. † mITT population excludes 29 patients from the BEL/ST group, due to IBAs being potentially unblinded. BEL, belimumab; BLyS, B-lymphocyte stimulator; C3/4, complement 3/4; IBA, independent blinded assessor; mITT, modified intention-to-treat; PBO, placebo; RTX, rituximab; SLEDAI-2K, SLE Disease Activity Index-2000; SLE, systemic lupus erythematosus; ST, standard therapy.
The mean duration of the longest disease control response (prespecified analysis) was significantly greater in BEL/RTX versus BEL/PBO patients through 52 weeks (adjusted treatment difference (95% CI) 47.0 (8.0 to 86.0) days, p=0.0188). Duration of disease control over 104 weeks, including the observational phase where all immunosuppression was stopped, was not significantly different between treatments ( table 3 ). Reductions in SLEDAI-2K scores through 104 weeks were observed in all treatment groups, with a significantly larger reduction observed with BEL/RTX versus BEL/PBO (adjusted treatment difference (95% CI) −1.6 (−2.4 to –0.7), p=0.0003; online supplemental figure S2A ). The odds of achieving LLDAS were not significantly different for BEL/RTX versus BEL/PBO at week 52 or 104 ( online supplemental figure S2B ).
Duration of disease control,* based on PI assessment (mITT population, † N=263)
Reduction from baseline in SLEDAI-2K ≥4 at week 52 by anti-dsDNA antibody and C3/C4 levels is shown in online supplemental table S4 .
SLE severe flares were assessed using a modified SFI as defined. As such, through week 52, 37.5% of BEL/PBO and 36.1% of BEL/RTX patients had a severe SFI flare. Over 104 weeks, this increased to 75.0% and 64.6%, respectively (HR (95% CI) 0.81 (0.57 to 1.13), p=0.2150) ( online supplemental figure S2C ). The median number of days to the first severe SFI flare was 372 in the BEL/PBO group and 379 in the BEL/RTX group. In post hoc analysis of severe flares that did not impute treatment failure, 20.8% of BEL/PBO-treated patients and 15.3% of BEL/RTX-treated patients experienced severe SFI flares over 104 weeks (HR (95% CI) 0.69 (0.36 to 1.34), p=0.2745). Among these patients, the median number of days to first severe flare was 202.0 and 226.5 in the BEL/PBO and BEL/RTX groups, respectively.
Within the small number of patients with proteinuria >0.5 g/24 hours at baseline (BEL/PBO, n=11; BEL/RTX, n=24; BEL/ST, n=9), at week 52, a numerically greater proportion of BEL/RTX-treated patients versus BEL/PBO-treated patients shifted from high (>0.5 g/24 hours) to normal (≤0.5 g/24 hours), but the difference was not significant (p=0.1318; post hoc analysis; online supplemental figure S3A ).
In a post hoc analysis, among patients with baseline prednisone-equivalent corticosteroid dose of >7.5 mg/day, dose reduction by ≥25% from baseline to ≤7.5 mg/day during week 40 through week 52 was observed in 50.0% (n=18/36) of BEL/PBO and 54.7% (n=41/75) of BEL/RTX patients, with no statistical difference (OR (95% CI) 1.29 (0.57 to 2.90), p=0.5379).
By week 52, a decrease from baseline in IgG was observed in all groups, with the greatest decreases observed with BEL/RTX. On belimumab discontinuation from week 52, IgG levels gradually increased but remained below baseline ( online supplemental figure S3B ).
A significantly greater decrease in anti-dsDNA antibodies from baseline was observed with BEL/RTX versus BEL/PBO at weeks 52 (p=0.0495) and 64 (p=0.0230), but not at week 104 (p=0.2501; online supplemental figure S3C ). Among anti-dsDNA antibody-positive patients at baseline (BEL/PBO, 63.6%; BEL/RTX, 71.4%), a significantly greater proportion of BEL/RTX versus BEL/PBO patients shifted to negative at week 52 only (n=18/74, 24.3% vs n=2/35, 5.7%; p=0.0186).
Increases from baseline in median C3 and C4 levels were observed in all groups, particularly BEL/RTX; levels were generally maintained above baseline during the observation phase ( online supplemental figure S3D, E ). Significant differences between BEL/RTX and BEL/PBO for median change from baseline were observed only for C3 levels at week 64 (p=0.0315). Among patients with low C3 levels (BEL/PBO, 43.6%; BEL/RTX, 44.8%) and those with low C4 levels (BEL/PBO, 18.2%; BEL/RTX, 32.4%) at baseline, there was no difference in the shift to normal/high levels between BEL/RTX and BEL/PBO groups.
At week 52, decreases in B cells and B-cell subsets were observed across all three groups, with significantly greater decreases observed with BEL/RTX versus BEL/PBO (p <0.0001) for all assessed B cells and B-cell subsets ( online supplemental figure S4 ). At week 104, the only statistically significant BEL/RTX versus BEL/PBO treatment difference was for memory CD20+ CD27+ B cells (p=0.0026).
Similar proportions of patients across treatment groups experienced on-treatment AEs (year 1) and on-study AEs (years 1+2; table 4 ). Proportions of patients experiencing belimumab-related AEs on-study were 34.7%–38.2% across all treatment groups and study periods ( table 4 ). Infections and infestations were the most frequently reported serious and non-serious AEs by system organ class, experienced by a higher proportion of the BEL/RTX group than other treatment groups ( table 4 ). In year 1, serious infections and infestations occurred in eight, two and one patients in the BEL/RTX, BEL/PBO and BEL/ST groups, respectively.
Summary of AEs and AESI for year 1 and years 1 + 2 (ITT population; N=292)
The incidence of infections of special interest was low, with one patient reporting sepsis in the BEL/RTX group.
Overall, 9 (12.5%) BEL/PBO-treated patients, 16 (11.1%) BEL/RTX-treated patients, and 5 (6.6%) BEL/ST-treated patients experienced psychiatric disorders of special interest and events of depression/suicide/self-injury ( table 4 ). Suicidal behaviour was reported in one patient in each group. No completed suicides were reported. Three deaths were recorded: one in the BEL/PBO group due to cholangiocarcinoma (fatal SAE started on study, but death occurred after study withdrawal) and two on study in the BEL/RTX group (one due to pneumonia and one attributed to ‘sudden death’).
This study addressed a novel treatment approach of adding rituximab to belimumab in patients with active SLE while investigating a rigorous withdrawal of background immunosuppressants and tapering corticosteroids. As such, BLISS-BELIEVE is the first randomised clinical trial in SLE to investigate disease control and remission off-therapy as stringent study outcomes. Similar proportions of BEL/RTX-treated and BEL/PBO-treated patients achieved disease control; therefore, the study’s primary endpoint was not met. However, the study demonstrated that disease control without immunosuppressants and low-dose corticosteroids can be achieved in 16.7%–19.4% of patients with active SLE.
Belimumab and rituximab have complementary mechanisms of action, providing an immunological rationale for their combined use, 28 as supported by case reports showing improvement with belimumab either preceding rituximab in primary Sjögren’s Syndrome (pSS) or following rituximab in pSS, SLE and LN. 30 33 40–43 While the BEAT-LUPUS trial demonstrated a reduced risk of severe flares among patients receiving belimumab after rituximab compared with placebo after rituximab, this was not observed in the present study of belimumab and subsequent rituximab therapy. The two study designs, sizes and populations differed substantially. For example, BEAT-LUPUS, a small phase 2 trial of 52 patients with refractory SLE, incorporated less stringent corticosteroid tapering and permitted immunosuppression. 32 Furthermore, the risk of severe flares was a secondary endpoint, and a different definition of severe flare was used in BEAT-LUPUS; thus, caution is required in comparing findings across the two trials.
In subgroup analyses, disease control rates at week 52 tended to numerically favour BEL/RTX versus BEL/PBO in patients with low C3/C4 and high anti-dsDNA antibody levels at baseline and in patients with baseline SLEDAI-2K score ≥10. Also, a numerically greater improvement in proteinuria was observed with BEL/RTX versus BEL/PBO in a post hoc analysis; however, these data must be interpreted with caution due to small patient numbers.
In the present study, BEL/RTX led to significant reduction in anti-dsDNA antibody levels at week 52 versus BEL/PBO corroborating the primary outcome results of anti-dsDNA antibody levels from the BEAT-LUPUS trial. 32 Similarly, C3/C4 levels increased from baseline to week 104 in all treatment groups, with a trend for larger increases with BEL/RTX at both week 52 and week 104. With respect to B-cell phenotyping, B-cell subsets (such as naïve B cells) decreased within 16 weeks of belimumab treatment in our study and memory B cells rapidly increased in circulation, consistent with observations made in previous studies. 44–46 More recently, the surge in circulating memory B cells following belimumab initiation has been associated with disrupted lymphocyte trafficking. 34 47 As previously demonstrated, the surge of BLyS levels after rituximab treatment can be targeted by sequentially treating with belimumab and vice versa as the memory B-cell surge following initiation of belimumab can be well targeted by rituximab. It remains, however, to be established whether disrupting lymphocyte trafficking and depriving memory B cells of cell-to-cell interactions can adequately modulate memory B cell-mediated immunological processes driving pathology and/or whether depletion of circulating memory B cells is an absolute requirement to impact the pathological pathways driven by this B-cell subset. We favour the latter, as anti-dsDNA antibodies have been shown to be markedly reduced in all randomised studies of sequential therapy, regardless of the order in which belimumab and rituximab are administered. 31 32 35 Significant decreases in anti-dsDNA antibody levels and increases in complement levels were observed in patients with complete depletion of B cells. 48 Successful depletion of B cells was also shown to predict response to rituximab, whereas repopulation of B cells increased the risk of clinical relapse. 48 Therefore, repopulation of B cells or subsets thereof (such as memory B-cells) could potentially act as an important biomarker for personalised treatment decisions when employing B-cell targeted therapies in active SLE.
Thus, the biomarker investigations as secondary outcomes of this study should prompt further studies to address whether changes seen in serological biomarkers and degree of B-cell depletion can be translated to identify a subgroup of patients with SLE that benefit the most from belimumab/rituximab combination therapy.
Safety findings were consistent with the known safety profiles of belimumab and rituximab, although incidences of AEs and serious infections and infestations, and depression were lower than in the BEAT-LUPUS phase 2 study of rituximab followed by belimumab treatment of patients with SLE. 32 Comparable proportions of patients in all treatment groups experienced ≥1 AE during our study. Although incidence of serious infections and infestations was higher in the BEL/RTX group than in other treatment groups, there was no evidence of new or unexpected safety signals using the combination. Nonetheless, the benefit from sequential belimumab and rituximab treatment in any subgroup of patients will need to be balanced against risk of serious infections and infestations.
In the present study, 25.5% of BEL/ST patients achieved disease control at week 52, which was higher than either BEL/PBO or BEL/RTX patients. One could speculate that a less stringent tapering of immunosuppression in the primary comparator groups BEL/PBO and BEL/RTX could have resulted in better disease control. 49 50 The patients in this study had a median disease duration of 7 years and a baseline mean daily corticosteroid dose of 10 mg/day, perhaps making them better suited to a less stringent tapering regimen, further supporting one of the important learnings of this informative study design.
This study has several limitations. Occurrence of severe SFI flare during the 2-year study period was relatively high, due partly to including the stringent study-specific definition of treatment failure (eg, corticosteroids >5 mg/day rather than 7.5 mg/day, restarting belimumab in year 2 (BEL/PBO or BEL/RTX); see online supplemental materials for full definition). Also, the BEL/ST group was included for descriptive reference only and cannot be directly compared with BEL/RTX; though, in real-world practice, belimumab is administered as an add-on to ST. The study aimed to investigate the hypothesis that combined B-cell targeting therapy would be effective without other ST. As such, the present study was not designed to reflect clinical practice accurately, and the use of less stringent tapering of IS, repetitive treatment cycles of rituximab and a comparison with continued BEL/ST would have been counterproductive to the study’s goal to achieve clinical remission off-therapy. BLISS-BELIEVE used SLEDAI-2K in our stringent efficacy outcomes, which, as a binary assessment, does not measure the improvement in symptoms, nor does it capture incomplete improvement in those symptoms. Furthermore, it is difficult to make direct comparisons between the BLISS-BELIEVE study and previous interventional SLE trials, in which response was determined by the SLE Responder Index 4 or British Isles Lupus Assessment Group-based Composite Lupus Assessment.
Primary and major secondary efficacy endpoints used IBA assessments, while other efficacy endpoints and treatment decisions were based on PI assessments. Some residual bias cannot be ruled out, as PIs were not blinded to the BEL/ST reference group.
The BLISS-BELIEVE study incorporated a unique and stringent efficacy endpoint requiring immunosuppression-free disease remission in patients with active SLE, a concept consistent with treat-to-target criteria. 38 50 Although BEL/RTX treatment was not superior to BEL/PBO and the addition of rituximab to patients’ therapy did not improve disease control using the stringent outcome measures employed by this study, there were significant reductions in circulating B-cells and improvement in serological biomarkers. This warrants further investigation to determine if combination with rituximab provides better disease control for patients. Nonetheless, this large, robust study provides valuable information to clinicians to support their treatment decision-making and informs future studies into combination therapy, perhaps with less rapid immunosuppressant tapering.
Patient consent for publication.
Consent obtained directly from patient(s).
This study involves human participants and was approved by "Argentina: 1. Framingham Centro Medico 2. Comite de Etica en Investigacion 3. Comité Independiente de Ética Médica del Noroeste Argentino (CIEM-NOA)""Brazil: 1. CEP do Hospital Universitario Julio Muller / MT 2. Fundação Faculdade Regional de Medicina de são José do Rio Preto 3. Hospital Santa Izabel 4. Comitê de Ética em Pesquisa do Hospital Universitário da Universidade Federal de Juiz de For a""Canada: 1. Advarra Institutional Review Board 2. Hamilton Integrated Research Ethics Board 3. University Health Network, Research Ethics Board""France: 1. Comite de Protection des Personnes Nord-Ouest II - CHU d’Amiens Hôpital Nord ""Germany: 1. Ethik-Kommission des Fachbereichs Medizin der Johann Wolfgang Goethe-Universität 2. Ethik-Kommission bei der Ärztekammer Hamburg 3. Ethik-Kommission der Medizinischen Hochschule Hannover 4. Ethik-Kommission der Med. Fakultaet der Christian-Albrechts-Universitaet zu Kiel""Republic of Korea: 1. IRB of Hanyang University Hospital, College of Medicine 2. IRB of Kyungpook National University Hospital 3. IRB of Chonnam National University Hospital, Clinical trial center 4. IRB of Seoul National University Hospital, Biomedical Research Institute 5. IRB of Seoul Saint Mary’s hospital 6. IRB of Inha University Hospital 7. Ajou University Hospital IRB, 5th Floor Annex building""Mexico: 1. Investigación Biomédica para el Desarrollo de Fárm 2. Hospital Centro de Especialidades Médicas CEM""Netherlands: 1. MEC-U""Russian Federation: 1. Republican Clinical Hospital 2. Local Ethics Committee of The State Educational Institution of the Highest Professional Education 3. University of Roszdrav 4. Institute of Cytology and Genetics 5. BUZ Regional Clinical Hospital 6. I.M. Sechenov First Moscow State Medical University, Local Ethics Committee of Federal State Autonomous Educational Institution of Higher Education 7. Chelyabinsk Regional Clinical Hospital 8. Tver Regional clinical hospital 9. Krasnoyarsk State Medical University 10. Local Ethics Committee at “Clinical Hospital #2, City Hospital #8 11. Kemerovo Regional Clinical Hospital 12. Republican Hospital nom Baranov, Ethics comitee 13. Ulyanovsk Regional Clinical Hospital 14. Local Ethics Committee at LLC Research Medical Complex ""Your Health 15. Local Ethics Committee at State Budgetary Institution of Ryazan Region ‘Regional Clinical Cardiology Dispensary 16. City Rheumatology Center, Hospital #25""Spain1. Comité Etico Hospital U. Germans Trias i Pujol""United States: 1. Advarra Institutional Review Board 2. Biomedical Research Alliance of New York, LLC, Institutional Review Board 3. WCG Institutional Review Board 4. Columbia University Medical Center, Institutional Review Board 5. Ochsner Clinic Foundation Institutional Review Board 6. NYU School of Medicine Institutional Review Board" Participants gave informed consent to participate in the study before taking part.
The authors would like to thank the participating patients and their families, clinicians and study investigators. Medical writing support was provided by Olga Conn, PhD, of Fishawack Indicia, UK, part of Avalere Health, and was funded by GSK.
Handling editor Josef S Smolen
Contributors Conception or design: INB, RF, JG, RvV, NLF, RBH, AvM, MO, DAR, PPT and YOT. Acquisition of data: CA, CFA, ZA, PCC, WWC and MBU. Data analysis or interpretation: CA, CFA, ZA, INB, PCC, WWC, KLC, RF, JG, MBU, RvV, MD, MO, NLF, YIG, AvM, JCO-A, DAR, DS, PPT and YOT. JCO-A is guarantor for this publication (KLC, JG, MD, NLF, YIG, DS, PPT: at the time of the author’s contribution to this study).
Funding This study (GSK Study 205646) was funded by GSK.
Competing interests CA has received research support from GSK; consulting fees from GSK, AstraZeneca, BMS, Kezar Life Sciences Inc., Merck Sharp & Dohme and Alumis Inc. CFA has received research support from Janssen, AbbVie, and Eli Lilly. ZA has received research support from GSK, Roche, AstraZeneca, and Amgen; and consulting fees from GSK, AstraZeneca, Amgen, Kezar Life Sciences Inc, and Novartis. INB has received research support from Genzyme, Sanofi, and GSK; consulting fees from AstraZeneca, Eli Lilly, Aurinia Pharmaceuticals, GSK, and ILTOO; and has served as an advisory board member for AstraZeneca and Merck Serono. INB is a National Institute for Health Research (NIHR) Senior Investigator Emeritus and is funded by the NIHR Manchester Biomedical Research Centre (NIHR203308). PCC has received consulting and speaker fees from GSK, Aurinia Pharmaceuticals, and Eli Lilly. WWC has received research support from GSK, UCB, Amgen, Pfizer, and BMS; consulting fees from GSK and Aurinia Pharmaceuticals; honoraria from GSK and Aurinia Pharmaceuticals for disease awareness presentations and curricula on achieving disease control and remission in SLE. RF has received research support and consulting fees from GSK. MBU has received research support from GSK; consulting fees from GSK and UCB; and speaker fees from GSK, Eli Lilly, and AstraZeneca. RvV has received research support for educational programmes and institutional grants from AstraZeneca, Pfizer, and UCB; consulting fees from AbbVie, AstraZeneca, Biogen, Biotest, BMS, Galapagos, Gilead, Janssen, Pfizer, Sanofi, Servier, UCB, and Vielabio; speaker fees from AbbVie, Galapagos, GSK, Janssen, Pfizer, and UCB; and research support from BMS, GSK Eli Lilly, and UCB. RBH, AvM, JCO-A, MO and DAR are employees of GSK and hold stocks and shares in the company. RBH is also an inventor for the following patents: US-20220195062-A1, US-11180569-B2, and US-20180265588-A1. KLC, JG, MD, YIG, DS and PPT were employees of GSK at the time of the study and hold stocks and shares in the company. NLF is a former employee and consultant for GSK and holds stocks in the company. YOT has received unrestricted research grants from GSK, Aurinia Pharmaceuticals, and Vifor Pharma; and consulting fees from Aurinia Pharmaceuticals, Novartis, GSK, Kezar Life Sciences Inc, Vifor Pharma, and Otsuka Pharmaceuticals (paid to Leiden University Medical Center).
Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.
Provenance and peer review Not commissioned; externally peer reviewed.
Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.
Welcome to the Money blog, your place for personal finance and consumer news and tips. Today's posts include Morrisons getting rid of some self-checkouts and a Money Problem on topping up your national insurance. Leave your consumer issue below - remember to include contact details.
Monday 19 August 2024 20:13, UK
Morrisons has admitted it "went a bit too far" with self-checkouts.
Chief executive Rami Baitiéh says the supermarket is "reviewing the balance between self-checkouts and manned tills".
Some will be removed.
Mr Baitiéh told The Telegraph : "Morrisons went a bit too far with the self-checkout. This had the advantage of driving some productivity. However, some shoppers dislike it, mainly when they have a full trolley."
The executive also said self checkouts had driven more shoplifting.
What have other supermarkets said about self-checkouts?
In April, the boss of Sainsbury's said customers liked self-checkouts...
That prompted us to ask readers for their thoughts - and we carried out a poll on LinkedIn which suggested the Sainsbury's boss was right...
Asda's chief financial officer Michael Gleeson said last week the technology had reached its limit - and said his firm would be putting more staff on tills.
Northern grocer Booths ditched almost all self-checkouts last year amid customer service concerns.
Over at Marks & Spencer, chairman Archie Norman last year blamed self-checkouts for a rise in "middle-class shoplifting".
But Tesco CEO Ken Murphy is an advocate: "We genuinely believe, at the end of the day, it provides a better customer experience."
The number of drivers visited by bailiffs due to unpaid traffic fines has increased substantially, according to a report.
Four million penalty charge notices (PCNs) were referred to bailiffs in England and Wales in the 2023-24 financial year, it is claimed.
This is up from 2.4 million during the previous 12 months, 1.9 million in 2019-20 and 1.3 million in 2017-18.
Read more ...
Ted Baker is the latest in a string of high-street giants to call in administrators in recent years, with shops set to disappear this week.
But how does it affect you?
Purchases and returns
You can still buy items online and in store until they close, but you could run into trouble returning them.
If the retailer stops trading, it may not be able to get your money back to you.
If that is the case, you would have to file a claim with Teneo (Ted Baker's administrator) to join a list of creditors owed money by Ted Baker – and even then there's no guarantee you'd get your money back.
If you have a gift card, you need to use it while you still can.
Credits and debits
You can file a claim with your debit or credit card provider to recover lost funds - but how exactly does that work?
Many retailers boosted wages after living wage/minimum wage changes in spring.
Figures show German discount brands Aldi and Lidl top the list of major UK supermarkets when it comes to staff hourly pay - after Lidl introduced its third pay increase of the year in May to match its closest rival.
Meanwhile, Morrisons is at the bottom of the pack for staff pay outside London, with hourly wages starting at the National Living Wage (£11.44).
How do other companies compare when it comes to pay and benefits? We've taken a look...
Pay: £12.40 an hour outside London and £13.65 inside the M25
Aldi announced in March it was bringing in its second pay rise of the year as part of its aim to be the best-paying UK supermarket.
From 1 June, hourly pay rose from £12 an hour to £12.40 outside the M25 and £13.55 to £13.65 in London.
Aldi is one of the only supermarkets to give staff paid breaks. It also offers perks such as discounted gym membership and cinema tickets, and financial planning tools. However, there are no cheaper meals, staff discounts or bonus schemes.
Pay: £12.04 an hour outside London and £13.21 inside the M25
As of 1 July, hourly wages for Asda supermarket staff rose to £12.04 per hour from £11.11, with rates for London staff also going up to £13.21.
As part of the July changes, Asda brought in the option for free later-life care or mortgage advice. The company also offers a pension and a free remote GP service.
Pay: £12 an hour outside London and £13.15 inside the M25
Co-op boosted its minimum hourly wage for customer team members from £10.90 to £12 nationally as the national living wage rose to £11.44 in April.
For staff inside the M25, rates rose from £12.25 to £13.15.
The perks are better than some. Workers can get 30% off Co-op branded products in its food stores as well as 10% off other brands. Other benefits include a cycle to work scheme, childcare vouchers and discounts on its other services.
Pay: £11.50 an hour outside London and £12.65 inside the M25
Iceland says it pays £11.50 for staff aged 21 and over - 6p above the minimum wage. Employees in London receive £12.65 per hour.
Staff are also offered a 15% in-store discount, which was raised from 10% in 2022 to help with the cost of living.
The firm says it offers other perks such as a healthcare scheme and Christmas vouchers.
Pay: £12.40 an hour outside London and £13.65 inside the M25
From June, Lidl matched its rival Aldi by raising its hourly wage to £12.40 for workers outside the M25 and £13.55 for those inside.
Lidl also offers its staff a 10% discount card from the first working day, as well as other perks such as dental insurance and fertility leave.
Marks and Spencer's hourly rate for store assistants was hiked from £10.90 to £12 for staff outside London and from £12.05 to £13.15 for London workers from April.
The grocer also offers a 20% staff discount after the probation period as well as discretionary bonus schemes and a free virtual GP service.
Pay: £11.44 an hour outside London and £12.29 inside the M25
Along with many other retailers, Morrisons increased the hourly wage for staff outside the M25 in line with the national living wage of £11.44 in April.
Employees in London receive an 85p supplement.
While it's not the most competitive for hourly pay, Morrisons offers perks including staff discounted meals, a 15% in-store discount and life assurance scheme.
Sainsbury’s
Sainsbury's hourly rate for workers outside London rose to £12 from March, and £13.15 for staff inside the M25.
The company also offers a 10% discount card for staff to use at Sainsbury's, Argos and Habitat, as well as a range of benefits including season ticket loans and long service rewards.
Pay: £12.02 an hour outside London and £13.15 inside the M25
Since April, Tesco staff have been paid £12.02 an hour nationally - up from £11.02 - while London workers get £13.15 an hour.
The supermarket giant also provides a 10% in-store discount, discounted glasses, health checks and insurance, and free 24/7 access to a virtual GP.
Staff get their pay boosted by 10% on a Sunday if they joined the company before 24 July 2022.
Pay: £11.55 an hour outside London and £12.89 inside the M25
Waitrose store staff receive £11.55 an hour nationally, while workers inside the M25 get at least £12.89.
Staff can also get access to up to 25% off at Waitrose's partner retailer John Lewis as well as 20% in Waitrose shops.
JLP (the John Lewis Partnership) gives staff a bonus as an annual share-out of profit determined by the firm's performance. In 2021-22 the bonus was 3% of pay; however, it has not paid the bonus for the past two years.
Dozens of Ted Baker stores will shut for the last time this week amid growing doubts over a future licensing partnership with the retail tycoon Mike Ashley.
Sky News understands that talks between Mr Ashley's Frasers Group and Authentic, Ted Baker's owner, have stalled three months after it appeared that an agreement was imminent.
Administrators are overseeing the closure of its remaining 31 UK shops.
One store source said they had been told that this Tuesday would be the final day of trading.
The housing market experienced a surge in activity following the Bank of England's recent decision to cut interest rates, according to a leading property website.
Estate agents reported a 19% jump in enquiries about properties for sale after 1 August, when compared with the same period last year, research by Rightmove found.
It came after the Bank cut rates for the first time in more than four years from 5.25% to 5%.
The lead negotiator for major train union ASLEF has denied the union sees the new government as a "soft touch" after announcing fresh strikes two days after train drivers were offered a pay deal.
Drivers working for London North Eastern Railway will walk out on weekends from the end of August in a dispute over working agreements.
Lead negotiator Nigel Roebuck said it is a separate issue from the long-running row over pay, which looks likely to be resolved after a much-improved new offer from the government.
Over 40 bottles of fake vodka have been seized from a shop in Scotland after a customer reported "smelling nail varnish".
The 35cl bottles, fraudulently labelled as the popular brand Glen's, were recovered from the shop in Coatbridge, North Lanarkshire.
Officers from the council's environmental health officers and Food Standard Scotland (FSS) sent them for analysis after a customer raised the alarm by saying they smelt nail varnish from one of the bottles.
The bottles were found to be counterfeit.
Britons don't have long left to claim cost of living assistance from the Household Support Fund.
Introduced in October 2021, the scheme provides local councils with funding which can be used to support those struggling most with the rising cost of living.
The vast majority of councils operate their version of the Household Support Fund on a "first come, first serve" basis and will officially end the schemes once the funding has run out in September.
The help provided by councils has ranged from free cash payments, council tax discounts, and vouchers for supermarkets and energy providers.
Who is eligible?
Local authorities were instructed to target the funding at "vulnerable households in most need of support to help with significantly rising living costs" when it was first rolled out.
In particular, councils were guided to make priority considerations for those who:
If you do not meet these criteria, you can still contact your local council , with many having broadened their criteria for eligibility.
By Daniel Binns, business reporter
Weapons maker BAE Systems is the big loser on the FTSE 100 this morning, with its shares down almost 3% in early trading.
It comes following reports over the weekend that the German government is planning to scale back aid to Ukraine in its war with Russia – in what would be a blow to the arms industry.
German media said ministers are set to slash support for Kyiv to 6% of current levels by 2027 in their upcoming budget.
However, the government there has rejected the reports and has denied it is "stopping support" to Ukraine.
Whatever the truth, the reports appear to have spooked traders.
Other companies involved in the defence sector, including Rolls-Royce Plc and Chemring Group, are also down more than 2% and 1% respectively on Monday.
It comes amid a slight slump in early trading, with the FTSE 100 down just over 0.2%, although the FTSE 250 is up 0.07%.
Gainers this morning include housebuilders Barratt Developments, up 1.5%, and Redrow Plc, which is up almost 3%.
Barratt said today it intends to push ahead with a planned £2.5bn merger with its rival despite concerns from the competition regulator.
Meanwhile, the price of oil is down amid concerns of weaker demand in China.
Ongoing ceasefire talks in the Israel-Hamas conflict have also raised hopes of cooling tensions in the Middle East, which would help ease supply risks and worries.
A barrel of the benchmark Brent Crude is currently priced at just over $79 (£61).
On the currency markets, this morning £1 buys $1.29 US or €1.17.
Winter energy bills are projected to rise by 9%, according to a closely watched forecast.
The price cap from October to December will go up to £1,714 a year for the average user, Cornwall Insight says.
It would be a £146 rise from the current cap, which is controlled by energy regulator Ofgem and aims to prevent households on variable tariffs being ripped off.
The cap doesn't represent a maximum bill. Instead it creates an average bill by limiting how much you pay per unit of gas and electricity, as well as setting a maximum daily standing charge (which all households must pay to stay connected to the grid).
Ofgem will announce the October cap this Friday.
"This is not the news households want to hear when moving into the colder months," said the principal consultant at Cornwall, Dr Craig Lowrey.
"Following two consecutive falls in the cap, I'm sure many hoped we were on a steady path back to pre-crisis prices.
"However, the lingering impact of the energy crisis has left us with a market that's still highly volatile and quick to react to any bad news on the supply front.
"Despite this, while we don't expect a return to the extreme prices of recent years, it's unlikely that bills will return to what was once considered normal. Without significant intervention, this may well be the new normal."
Cornwall Insight warned that the highly volatile energy market and unexpected global events, such as the recent escalating tensions in the Russia-Ukraine war, could see prices rise further at the start of the new year.
To avoid this vulnerability, Cornwall Insight said domestic renewable energy production should increase and Britain should wean itself off energy imports.
Kellogg's appears to have shrunk its packets of Corn Flakes.
Two of its four different pack sizes have reduced in weight by 50g, according to The Sun.
What used to be 720g boxes are now 670g, while 500g boxes have become 450g.
The newspaper says the 670g boxes are being sold for £3.20 in Tesco - the same price customers were paying for the larger box back in May.
The 450g boxes are being sold for £2.19, only slightly less than the previous price of £2.25.
Other supermarkets have similar pricing, although in Morrisons the price has gone down in proportion to the size reduction.
The 250g and 1kg pack sizes remain unchanged.
Kellogg's has said it is up to shops to choose what they charge, but Tesco said the manufacturer should comment on pricing.
Sky News has contacted Kellogg's for comment.
A spokesperson is quoted by The Sun: "Kellogg's Corn Flakes are available in four different box sizes to suit different shopper preferences and needs.
"As the cost of ingredients and production processes increase, it costs us more to make our products than it used to.
"This can impact the recommended retail price. It's the grocer's absolute discretion and decision what price to charge shoppers."
Be the first to get Breaking News
Install the Sky News app for free
Configuration engineer
"I decided to make a career change when I was twenty-four. I chose to pursue Cisco Certifications because I knew it would put me in the best position to start a career in networking."
My dad is a systems administrator, so I grew up around computers. Just watching him tinker with them was cool. However, when I went to college, I earned a bachelor’s degree in philosophy. Next, I got a teaching credential, then taught elementary school for a few years. But when I turned 24, I decided I wanted a career change. So, I picked up a book on TCP/IP networking, started studying it, and realized that I really enjoyed it. My dad wasn’t into networking that much, so I didn’t know much about it, and got into it more organically. I considered pursuing Cisco certifications right away since the idea of learning about networking led straight to these credentials. I took a networking class at a local community college, then obtained my Cisco CCNA certification.
I enjoy system administration, I’ve gained communications skills, and I’ve learned tech skills, in Microsoft active directory systems, administration, and networking virtualization. General troubleshooting is a big part of my job, too. Most importantly, I’ve gained the flexibility to work around difficulties and learn new technologies. Just being able to keep up with the constant change in technologies is incredibly valuable.
Columbus, OH
Reading Hiking Learning new tech
"I’ve gained more knowledge and different skill sets. I’ve opened myself up for more opportunities. And it’s a validation of me and what I’ve learned."
The biggest challenge was where to start, there where a lot of options. A lot of them involved going back to school or spending a lot of money. CCNA was the most cost-effective, and it would put me in the best position to start a career. Cisco is a leader in that. I took a bootcamp course at the local community college that lasted 8 weeks in total.
I was already in my job in IT in tier 1 support, help desk and the CCNA Certification was a catalyst for my first promotion as a second tier support security.
Definitely try different things and go all in on whatever interests you the most.
I’d tell them to go for it. It will open up a lot of opportunities for them. They’ll have all the knowledge they need to go forward in their career.
Career path.
Current role
Configuration engineers work on systems and network administration.
Previous role
IT Help Desk Support Primary School Educator
Most recent certification
Cisco Certified Network Associate (CCNA) certification is the first step toward a career in IT Networking. The CCNA exam covers networking fundamentals, IP services, security fundamentals, automation and programmability.
"The guidance I gained from earning the certification helped— in near real time—to determine what was happening on the job, when I became an SOC operator."
Network security analyst CyberOps Associate
"You can do anything. You just need to decide to do it, have the will to do it, and never give up. Be confident in yourself and stop the barriers in your mind."
Network security architect CCNA, CCNP Enterprise, CCNP Security, CyberOps Associate, CCIE Enterprise Infrastructure, Cisco Certified DevNet Associate, Cisco Certified DevNet Professional
"The knowledge that I’ve got from studying for those certifications gave me the confidence that I’ll always be able to get a job if I need to."
Systems engineer CCNA, CCNP Enterprise, DevNet Associate
Has earning a Cisco certification positively changed your life or career, or both? Do you think your Cisco certification story would help encourage other people to earn their Cisco certification? If so, we want to talk to you!
IMAGES
COMMENTS
References in Research. Definition: References in research are a list of sources that a researcher has consulted or cited while conducting their study. They are an essential component of any academic work, including research papers, theses, dissertations, and other scholarly publications.
Reference in research papers: A reference is a detailed description of the source of information that you want to give credit to via a citation. The references in research papers are usually in the form of a list at the end of the paper. The essential difference between citations and references is that citations lead a reader to the source of ...
References provide the information necessary for readers to identify and retrieve each work cited in the text. Check each reference carefully against the original publication to ensure information is accurate and complete. Accurately prepared references help establish your credibility as a careful researcher and writer. Consistency in reference ...
Referencing correctly: helps you to avoid plagiarism by making it clear which ideas are your own and which are someone else's. shows your understanding of the topic. gives supporting evidence for your ideas, arguments and opinions. allows others to identify the sources you have used.
Referencing is used to tell the reader where ideas from other sources have been used in an assignment. There are many reasons why it is important to reference sources correctly: It shows the reader that you can find and use sources to create a solid argument. It properly credits the originators of ideas, theories, and research findings.
Creating a Harvard reference list. A bibliography or reference list appears at the end of your text. It lists all your sources in alphabetical order by the author's last name, giving complete information so that the reader can look them up if necessary. The reference entry starts with the author's last name followed by initial(s).
Reference sources are generally the place to begin your research, especially when you're starting out with an unfamiliar field. But they're also where you return when you need to look up formulas, facts, definitions, and other standard details; they tend to pack a lot of information into simple, easy-to-use packages.
footnotes, endnotes, reference list or bibliography. (The format and terms used depend on the citation style.) The terms reference list and bibliography are sometimes used to mean the same thing, that is, the complete list of references or bibliographic details for the sources you have cited. However, bibliography can be
Referencing is a way to provide evidence to support the assertions and claims in your own assignments. By citing experts in your field, you are showing your marker that you are aware of the field in which you are operating. Your citations map the space of your discipline and allow you to navigate your way through your chosen field of study, in ...
There are three main approaches: Parenthetical citations: You include identifying details of the source in parentheses in the text—usually the author's last name and the publication date, plus a page number if relevant ( author-date ). Sometimes the publication date is omitted ( author-page ). Numerical citations: You include a number in ...
In-text citations are quick references to your sources. In Harvard referencing, you use the author's surname and the date of publication in brackets. Up to three authors are included in a Harvard in-text citation. If the source has more than three authors, include the first author followed by ' et al. '.
At college level, you must properly cite your sources in all essays, research papers, and other academic texts (except exams and in-class exercises). Add a citation whenever you quote, paraphrase, or summarize information or ideas from a source. You should also give full source details in a bibliography or reference list at the end of your text.
Referencing. Referencing is one of the most important aspects of any academic research and poor or lack of referencing will not only diminish your marks, but such practices may also be perceived as plagiarism by your university and disciplinary actions may follow that may even result in expulsion from the course. Difference between References ...
Examples of reference sources include: Encyclopedias; Dictionaries; Almanacs; Indexes; Atlases; Bibliographies; We realize that the term "reference sources" used this way may be a bit confusing, since your professors might also talk about "references" as a way of describing anything that you might cite in a research paper. Always be sure to ask ...
Articles & Research Databases Literature on your research topic and direct access to articles online, when available at UW.; E-Journals Alphabetical list of electronic journal titles held at UW.; Encyclopedias & Dictionaries Resources for looking up quick facts and background information.; E-Newspapers, Media, Maps & More Recommendations for finding news, audio/video, images, government ...
Here's one example of writing references in research papers - 'Nature 171: 737' is a code that, if you know how to decipher it, tells you that it means an article published in Nature (a weekly journal published from the UK) that begins on page 737 of volume 171 of that journal. However, it does not tell you what the article was about ...
For example, you are citing study notes titled "Health Effects of Exposure to Forest Fires," but you do not know the author's name, your reference entry will look like this: Health effects of exposure to forest fires [Lecture notes]. (2005). Walden University Canvas. https://waldenu.instructure.com.
Reference is a citation of legal authority, cases, Acts, regulations, law books, or journals systematically and acceptably. A researcher should understand primary and secondary texts, locate cases and statutes, and use treaties, periodicals, digests, and standard practitioner texts. The citation may provide important information, support one ...
References (summarised from here ): demonstrate the foundation of the study. support the novelty and value of the study. link one study to others creating a web of knowledge that carries meaning. allows researchers to identify work as relevant in general and relevant to them. create values that are internal to science (e.g., relevance, credit).
For three or more authors, provide the first author's name surname first then followed by "et al." Books with three or more authors : Joseph, Gary, et al. Changing shirts. Generic Publishing House, 2011. When you want to cite a chapter or an essay in a book, follow this basic format.
APA Style is widely used by students, researchers, and professionals in the social and behavioral sciences. Scribbr's free citation generator automatically generates accurate references and in-text citations. This citation guide outlines the most important citation guidelines from the 7th edition APA Publication Manual (2020).
A reference page organizes all of these types of references in one place. It is a list of all sources cited and is the final page of an APA Style paper. A consistent and organized reference page ...
Meaning. Citation is a way of disclosing within the main body, that the quote, image, chart, statistics, etc. are taken from an outside source. Reference is a list which contains all the sources which have been sought or cited while writing the article or assignment. Use. It informs the readers, the basic source of information.
The reference should be the same as the reference for a print version of the work. For works from databases that publish original, proprietary material available only in that database (such as the UpToDate database) or for works of limited circulation in databases (such as monographs in the ERIC database), include the name of the database or ...
WASHINGTON (AP) — The ancient ritual meaning of Stonehenge is still a mystery, but researchers are one step closer to understanding how the famous stone circle was created.. The unique stone lying flat at the center of the monument was brought to the site in southern England from near the tip of northeast Scotland, researchers reported Wednesday in the journal Nature.
Using prospectively obtained observational single-center data from 100 patients undergoing elective noncardiac surgery with invasive arterial monitoring, the authors used correlation, receiver operating characteristic curves, and precision-recall curves analyses to characterize the relationship between the index and simultaneous mean arterial pressure. A mean arterial pressure threshold of 73 ...
Objectives Disease activity control in patients with systemic lupus erythematosus (SLE) with corticosteroid and immunosuppressant withdrawal is a treatment goal. We evaluated whether this could be attained with sequential subcutaneous belimumab (BEL) and one cycle of rituximab (RTX). Methods In this phase 3, double-blind BLISS-BELIEVE trial (GSK Study 205646), patients with active SLE ...
Welcome to the Money blog, your place for personal finance and consumer news and tips. Today's posts include a Money Problem on the benefits or otherwise of topping up your national insurance.
I enjoy system administration, I've gained communications skills, and I've learned tech skills, in Microsoft active directory systems, administration, and networking virtualization.