- Research
- Open Access
- Published:
Researchers in rheumatology should avoid categorization of continuous predictor variables
BMC Medical Research Methodology volume 23, Article number: 104 (2023)
Abstract
Background
Rheumatology researchers often categorize continuous predictor variables. We aimed to show how this practice may alter results from observational studies in rheumatology.
Methods
We conducted and compared the results of two analyses of the association between our predictor variable (percentage change in body mass index [BMI] from baseline to four years) and two outcome variable domains of structure and pain in knee and hip osteoarthritis. These two outcome variable domains covered 26 different outcomes for knee and hip combined. In the first analysis (categorical analysis), percentage change in BMI was categorized as ≥ 5% decrease in BMI, < 5% change in BMI, and ≥ 5% increase in BMI, while in the second analysis (continuous analysis), it was left as a continuous variable. In both analyses (categorical and continuous), we used generalized estimating equations with a logistic link function to investigate the association between the percentage change in BMI and the outcomes.
Results
For eight of the 26 investigated outcomes (31%), the results from the categorical analyses were different from the results from the continuous analyses. These differences were of three types: 1) for six of these eight outcomes, while the continuous analyses revealed associations in both directions (i.e., a decrease in BMI had one effect, while an increase in BMI had the opposite effect), the categorical analyses showed associations only in one direction of BMI change, not both; 2) for another one of these eight outcomes, the categorical analyses suggested an association with change in BMI, while this association was not shown in the continuous analyses (this is potentially a false positive association); 3) for the last of the eight outcomes, the continuous analyses suggested an association of change in BMI, while this association was not shown in the categorical analyses (this is potentially a false negative association).
Conclusions
Categorization of continuous predictor variables alters the results of analyses and could lead to different conclusions; therefore, researchers in rheumatology should avoid it.
Background
Epidemiological research can suggest potential risk factors and strategies to prevent, delay or reverse osteoarthritis and other rheumatic diseases. In epidemiological research in osteoarthritis and other rheumatic diseases, it is common practice to categorize a continuous variable that is a predictor of an outcome (the ‘predictor variable’), as evident in studies published in the past two years (2020 to 2022 [1,2,3,4,5,6,7]). These studies categorized continuous predictor variables such as: change in weight [8]; change in body mass index (BMI) [2]; risk score for mortality [3]; age [3]; appendicular lean mass index [4]; fat mass index [4]; disease activity score [9]; years of use of analgesics [5]; patient global visual analogue scale assessment of disease activity [10]; the Stanford Health Assessment Questionnaire-Disability Index [10]; glucocorticoid drug dosage [10]; swollen joint count [6]; tender joint count [6]; and alcohol unit consumption per week [7]. To further exemplify how common the practice of categorization of continuous predictor variables is in rheumatology research, we surveyed articles published in 2021 (the year before commencing this current study) in the journal titled Arthritis Care & Research. Arthritis Care & Research is an official journal of the American College of Rheumatology, which is a leading professional organization in rheumatology. In our survey, we only included articles reporting observational studies and randomized trials. Our survey revealed that 49% (101 [11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111] of 208 [11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131,132,133,134,135,136,137,138,139,140,141,142,143,144,145,146,147,148,149,150,151,152,153,154,155,156,157,158,159,160,161,162,163,164,165,166,167,168,169,170,171,172,173,174,175,176,177,178,179,180,181,182,183,184,185,186,187,188,189,190,191,192,193,194,195,196,197,198,199,200,201,202,203,204,205,206,207,208,209,210,211,212,213,214,215,216,217,218]) of those articles categorized the continuous predictor variables that were used in their primary analysis. Categorizing continuous variables is not specific to research in osteoarthritis and other rheumatic diseases. Indeed, a review of 58 articles published in the two months of December 2007 and January 2008 in ten journals (five epidemiological and five general medicine) found that 86% of these articles categorized the primary predictor variable [219]. A more recent review of 23 observational studies published between April and June 2015 found that 61% categorized continuous predictor variables [220].
Although it is widely used, categorization of continuous predictor variables [221,222,223,224,225,226,227] or continuous outcome variables [228,229,230,231,232,233,234] is not recommended in research because of several issues: distortion of associations [235]; loss of power and precision [236, 237]; increased probability of biased estimates [237, 238]; type I errors (false positives) [239]; type II errors (false negatives) [240]; and inflated effect sizes (odds ratio) [223, 241].
The common practice of categorizing continuous predictor variables in epidemiological research in osteoarthritis and other rheumatic diseases despite the drawbacks mentioned above may be due to the need for clarity on how this practice changes the results and conclusions. Therefore, our primary aim in this study was to investigate the extent to which the categorization of continuous predictor variables changes findings in epidemiological rheumatology research. For this study, we will use as our example the percentage change in BMI as a predictor variable, with the percentage change in BMI treated either as 1) a categorical variable or 2) a continuous variable, and the two outcome variable domains of structure and pain in knee and hip osteoarthritis.
Methods
We revisited a study by Joseph et al. [2] that investigated the association between percentage change in BMI over four years and the two outcome variable domains of structure and pain in knee and hip osteoarthritis using data from the Osteoarthritis Initiative (OAI) study. The authors treated the percentage change in BMI as a categorical variable. From their results, they suggested that while a decrease of 5% or more in BMI may protect against overall structural changes in the knee (as assessed by radiography) and may decrease pain in the knee over four years, an increase of 5% or more in BMI may exacerbate medial joint space narrowing (JSN) in the knee and the development of pain in the knee over four years [2]. There was no association of the percentage change in BMI – when treated as a categorical variable – with any outcomes of hip osteoarthritis [2].
Data
We used data from the OAI study [242]. OAI data is openly available to researchers for scientific and educational purposes. The OAI is a multi-center longitudinal study that collected data over four years from a total of 4796 adults (45 to 79 years of age) with or at risk of clinically significant knee osteoarthritis. The local institutional review boards of the OAI centers reviewed and approved the informed consent documentation and ethics approval.
Exposures
Our predictor variable was the percentage change in BMI between baseline and four years, calculated as follows [2]. We fitted a simple linear regression line for each participant to estimate their annual rate of change in BMI, based on their data for BMI at baseline and other available time points. We then multiplied the slope of this regression line by 4 to estimate the absolute change in BMI over four years. The percentage change in BMI for each individual was then calculated as the absolute change in BMI over four years divided by the baseline BMI of that individual [2]. Fitting a simple linear regression line for each participant allowed us to estimate the change in BMI in cases of missing data, by using all available data points.
For the ‘categorical analysis’, we created 3 weight change groups: ≥ 5% decrease in BMI, < 5% change in BMI (i.e., stable BMI, the reference category), and ≥ 5% increase in BMI between baseline and four years. As opposed to the study by Joseph et al. [2], we did not exclude participants who showed a modest change in BMI (3–5%); and we defined the “stable BMI” category (which was the reference category) as those individuals who exhibited a change in BMI of less than 5%, whereas it was defined by Joseph et al. [2] as a change in BMI of less than 3%. By including participants that exhibited modest change in BMI, we have increased our sample size by 26.2%, and have therefore increased statistical power in our study [243, 244]. We used a 5% weight change threshold because prior studies suggest that this degree of weight change is clinically relevant [2].
For the ‘continuous analysis’, we treated the percentage change in BMI between baseline and four years as the continuous variable that it is.
Outcomes
Our two outcome variable domains of structure and pain of knee and hip osteoarthritis covered a total of 26 outcomes (18 in the structure and 8 in the pain outcome variable domains). The definitions of these 26 outcomes are detailed in the supplementary file. These outcomes were defined based on the definitions in the study by Joseph et al. [2].
The 18 outcomes that were in the outcome variable domains of structure were as follows: eight outcomes related to the progression of knee osteoarthritis as assessed by radiography at four years’ follow up; eight outcomes related to the progression of hip osteoarthritis, also assessed by radiography at four years’ follow up; one outcome for the incidence of total knee replacement (TKR) over four years; and one outcome for the incidence of total hip replacement (THR) over four years. For our eight outcomes related to the progression of knee osteoarthritis, we separately investigated the overall structure of the knee joint, and the following seven individual structural features (ISFs) of the knee joint: 1) joint space narrowing (JSN) in the medial or lateral compartment; 2) JSN in the medial compartment; 3) JSN in the lateral compartment; 4) osteophytes on the medial tibial surface; 5) osteophytes on the lateral tibial surface; 6) osteophytes on the medial femoral surface; and 7) osteophytes on the lateral femoral surface. For our eight outcomes for the progression of hip osteoarthritis, we also separately investigated the overall structure of the hip joint, and the following seven ISFs of the hip joint: 1) JSN in the medial or lateral compartment; 2) JSN in the medial compartment; 3) JSN in the lateral compartment; 4) osteophytes on the superior acetabular surface; 5) osteophytes on the superior inferior surface; 6) osteophytes on the superior femoral surface; and 7) osteophytes on the inferior femoral surface.
In the outcome variable domain of pain, two types of pain were investigated for the knee and hip: “frequent pain” and “any pain”. For frequent pain in the knee and hip, we used the following 4 outcomes in the analyses: 1) development of frequent pain in the knee 2) development of frequent pain in the hip; 3) resolution of frequent pain in the knee; and 4) resolution of frequent pain in the hip, by four years’ follow up. For any pain in the knee and hip, we used the following 4 outcomes in the analyses: 1) development of any pain in the knee 2) development of any pain in the hip; 3) resolution of any pain in the knee; and 4) resolution of any pain in the hip, by four years’ follow up.
Participant selection
We applied exclusion criteria for participant selection as per the study by Joseph et al. [2]. Firstly, we excluded participants that had BMI data at less than three of the five available timepoints (Fig. 1). This was due to needing a minimum of three timepoints with BMI data to determine weight cycling (to be explained below) from BMI fluctuation. Secondly, we excluded participants who had end stage osteoarthritis of knees or hips at baseline (Fig. 1). End stage osteoarthritis of knees was defined as having a Kellgren Lawrence (KL) grade of 4 (the highest possible KL grade) in both knees. End stage osteoarthritis of hips was defined as having JSN that had an Osteoarthritis Research Society International (OARSI) grade of 3 (the highest possible OARSI grade) in both hips, in any of the two sides of the hip (i.e., lateral or medial). Exclusion of these participants was done to avoid any possible confounding effect of their data on the study results due to their potentially reduced mobility and / or reduced ability to exercise. Additionally, there is no way to assess further change in the structure of the knee or hip joints as assessed radiographically once a participant has reached end-stage osteoarthritis. Thirdly, we also excluded participants with rheumatoid arthritis, cancer, or cardiac failure at baseline, as these conditions may cause pathological weight change, which in turn can impact change in BMI (Fig. 1). Fourthly, using BMI fluctuation information, we excluded participants who had ‘weight cycling’ during follow up. Weight cycling refers to a repetitive pattern of weight loss and regain [245]. We excluded participants with weight cycling as they would not completely be classified in the weight loss or weight gain categories. Moreover, weight cycling is associated with increased progression of structural defects in osteoarthritis, regardless of whether there is net weight gain or net weight loss [246]. Weight cycling was defined based on BMI fluctuation. BMI fluctuation was calculated as the root mean square error (RMSE) of the regression line of BMI over time that was calculated for each individual [2]. The participants with a RMSE value in the top 10% of all RMSE values were determined as having weight cycling and were thus excluded [2] (Fig. 1).
Selection of participants for each cohort. OAI: Osteoarthritis Initiative; BMI, body mass index
With the application of these four selection criteria, the ‘main cohort’ was created, which was used for investigating the 18 outcomes in the outcome variable domain of structure for the progression of knee and hip osteoarthritis and the incidence of TKR and THR. Further, we created four additional sub-cohorts (the ‘frequent knee pain cohort’, ‘frequent hip pain cohort’, ‘any knee pain cohort’, and ‘any hip pain cohort’) which was used for investigating the 8 outcomes in the outcome variable domain of pain (Fig. 1). The 4 outcomes for frequent knee and hip pain were investigated in the ‘frequent knee pain cohort’ and ‘frequent hip pain cohort’, respectively. The 4 outcomes for any knee and hip pain were investigated in the ‘any knee pain cohort’ and ‘any hip pain cohort’, respectively.
Statistical analyses
We used STATA/BE 17.0 for our analyses. We set our threshold for statistical significance as a two-tailed P-value of less than 0.05, as in the study by Joseph et al. [2]. We have not adjusted the significance level for multiple testing (e.g., Bonferroni adjustment).
We investigated the association between the percentage change in BMI (treated categorically and continuously) and the outcomes described above using generalized estimating equations with a logistic link function [247], sometimes referred to as logistic regression with clustering within individuals. In this case, the clustering is of the left and right knee or hip. This approach takes into account the within-person correlation between the two knees or hips and allows for a more accurate estimation of any association between the exposure and outcome. All analyses were adjusted for the following variables: age, sex, and baseline BMI.
For the continuous analysis, we first determined whether the percentage change in BMI had a linear relationship with each of our outcomes using the Box-Tidwell method. In this method, an interaction between the percentage change in BMI and its natural logarithmic value is added to the model. A significant interaction indicated a nonlinearity between the percentage change in BMI and the outcome variable [248]. While our statistical analysis suggested that 25 of the 26 outcomes had an apparent linear relationship with the percentage change in BMI, there may be some degree of uncertainty regarding the existence of these relationships, as the inference of linear relationships was based on the results of statistical tests. The remaining outcome, overall structural defects in knee osteoarthritis, did not show any apparent linear relationship with change in BMI. For those 25 outcomes that did have a linear relationship, we fitted a line from the available continuous range of BMI change where the relationship with the outcome variable is linear on the log odds ratio scale, then estimated the effect sizes (odds ratios) from that line. We reported the point estimates of a 5% decrease and a 5% increase in BMI. For the one outcome that did not show any apparent linear association with the percentage change in BMI (i.e., overall structural defects in knee osteoarthritis), we used the statistical method of piecewise linear spline regression. In this method, we divided the data into three separate segments: a decrease of ≥ 5% in BMI; a change of < 5% in BMI; and an increase of ≥ 5% in BMI. In each of the three segments, the change in BMI was linear, but with each segment potentially having a different effect size. We calculated effect sizes from two of these 3 separate segments; one effect size from the segment of decrease of 5% or more in BMI; the other effect size from the segment of increase of 5% or more in BMI. We used these two segments to calculate the point estimates of the effect sizes at a 5% decrease in BMI and a 5% increase in BMI.
Sensitivity analyses
In our primary analyses (where we investigated the association between the percent change in BMI and 26 outcomes from the outcome variable domains of structure and pain in knee and hip osteoarthritis), the estimates were calculated using a 5% change in BMI in the categorical and continuous analyses. We performed sensitivity analyses to assess if our conclusions from the results that were obtained in our primary analyses would still hold for different percentage changes in BMI. For that, we performed sensitivity analyses by repeating the primary analyses but this time instead of 5%, using a 3% change in BMI categories (i.e., ≥ 3% decrease in BMI, < 3% change in BMI, and ≥ 3% increase in BMI) and a 10% change in BMI categories (i.e., ≥ 10% decrease in BMI, < 10% change in BMI, and ≥ 10% increase in BMI).
Results
Participant characteristics
There were 3378 participants with 6756 knees and 6756 hips in the main cohort (the cohort in which we investigated the 18 outcomes in the outcome variable domain of structure). There were 3108 participants with 5728 knees in the frequent knee pain cohort, 3312 participants with 6644 hips in the frequent hip pain cohort, 2065 participants with 3128 knees in the any knee pain cohort, and 3022 participants with 5364 hips in the any hip pain cohort (Fig. 1).
Table 1 shows characteristics of the participants included in each of the five cohorts in this study. The mean age of participants in each cohort was similar, ranging from 61.1 (standard deviation [SD] 9.2) to 61.9 (SD 9.3) years. The percentage of female participants was higher than that of male participants in each cohort, ranging from 55.9 to 57.4%. The mean BMI of participants in each cohort was also similar, ranging from 27.7 (SD 4.4) to 28.1 (SD 4.6) kg/m2.
Figure 2 shows the distribution of participants by the percentage change in BMI from baseline to four years’ follow up in the main cohort. Of the 3378 participants in this cohort, there were 469 (13.9%) that had a decrease in BMI of 5% or more, 2223 (65.8%) that had a stable BMI (change of less than 5%), and 686 (20.3%) that had an increase in BMI of 5% or more. The distribution of percentage change in BMI was similar in all the other four sub-cohorts (data not shown).
Distributions of participants by percentage change in body mass index (BMI) from baseline. BMI: Body Mass Index
Incidence of outcomes
The incidence count of outcomes in the five cohorts can be found in Tables S1 to S5 in the Supplementary file. In comparison to knee, the numbers for the incidence of outcomes for hip were generally lower, with the exception of development and resolution of any hip pain (12.7% versus 14.2% for the development of pain, and 13.6% versus 15.4% for the resolution of pain of the knee and hip, respectively).
Association between percentage change in BMI and outcomes
Table 2 shows the results from our two analyses (categorical and continuous) for the associations of a 5% change in BMI with the 26 outcomes. Of the 26 outcomes investigated, 18 (69%) showed the same result in both the categorical and continual analyses. Of these 18 outcomes, 17 showed no association with the percentage change in BMI when treated either categorically or continuously. These 17 outcomes were: (for the knee) progression in lateral JSN; progression in medial tibial osteophytes; progression in lateral tibial osteophytes; progression in lateral femoral osteophytes; and incidence of TKR; (and for the hip) progression in overall structural defects in hip osteoarthritis; progression in medial or lateral JSN; progression in medial JSN; progression in lateral JSN; progression in superior acetabular osteophytes; progression in superior femoral osteophytes; progression in inferior femoral osteophytes; development of frequent pain in hip; resolution of frequent pain in hip; development of any pain in hip; resolution of any pain in hip; and incidence of THR (Table 2). The remaining one of these 18 outcomes (i.e., progression in overall structural defects in knee osteoarthritis) showed an association with a decrease in BMI but not with an increase in BMI in both the continuous and the categorical analysis.
Of the 26 outcomes investigated, the remaining eight outcomes (31%) showed association with a 5% change in BMI in either the categorical or the continuous analysis, but not in both analyses (Table 2). These eight outcomes were: (for the knee) 1) progression in medial or lateral JSN; 2) progression in medial JSN; 3) progression in medial femoral osteophytes; 4) development of frequent pain in knee; 5) resolution of frequent pain in knee; 6) development of any pain in knee; 7) resolution of any pain in knee; and (for the hip) 8) inferior acetabular osteophyte progression (Table 2).
Although these eight outcomes were associated with the percentage change in BMI when treated categorically or continuously, there were three types of differences between the associations in the two analyses. These differences will be explained below in points a, b, and c.
-
a)
Outcomes showed associations with percentage change in BMI only in one direction in the categorical analyses but in both directions in the continuous analyses
Six of these eight outcomes were positively associated with the percentage change in BMI (i.e., both increase and decrease in BMI) when BMI was treated as a continuous variable. However, in the categorical analysis, these six outcomes had an association with either an increase in BMI or a decrease in BMI, but not both. In the categorical analyses, five of these six outcomes were only associated with a decrease in BMI but not an increase in BMI. These five outcomes (all for knee) were: 1) progression in medial or lateral JSN; 2) progression in medial JSN; 3) progression in medial femoral osteophytes; 4) resolution of frequent pain in knee; and 5) resolution of any pain in knee. The remaining one outcome (development of frequent pain in knee) showed an association with an increase in BMI but not with a decrease in BMI in the categorical analysis (Table 2).
-
b)
Outcomes showed associations with percentage change in BMI in the categorical analysis but not in the continuous analysis (possible false positive)
When BMI was treated as a categorical variable, one of these eight outcomes, namely progression in inferior acetabular osteophytes in the hip, showed an association with a decrease in BMI (but not an increase in BMI). This may be a false positive, because 1) the outcome showed no significant association with the percentage change in BMI when percentage change in BMI was treated as a continuous variable (Table 2); 2) there was no other significant association for the any of the 8 outcomes for progression of hip osteoarthritis assessed by radiography in the categorical analysis (Table 2); and 3) acetabular osteophytes are not a reliable measure for the progression of hip osteoarthritis [249, 250] as it is difficult to distinguish them from normal anatomy [250].
-
c)
Outcomes showed associations with percentage change in BMI in the continuous analysis but not in the categorical analysis (possible false negative)
When BMI was treated as a categorical variable, one of these eight outcomes, namely the development of any pain in the knee, showed no association with the percentage change in BMI (either a decrease in BMI or an increase in BMI). This may be a false negative, because 1) the outcome showed an association with the percentage change in BMI when percentage change in BMI was treated as a continuous variable (Table 2); 2) all the other 3 of 4 outcomes for knee pain showed an association either with a decrease or increase in BMI in the categorical analysis, suggesting likelihood of an association; and 3) other studies showed an association of change in BMI with the development of knee pain due to osteoarthritis [251, 252].
Sensitivity analyses
The results from the sensitivity analyses using 3% and 10% change in BMI (decrease or increase) showed that eight and seven of the 26 outcomes investigated, respectively, differed in the categorical compared to the continuous analysis, showing all of the three different types of differences that were shown in our primary analyses (i.e., using 5% change in BMI) (Table S6).
Discussion
This study in osteoarthritis showed that categorizing the continuous predictor variable in the analysis (in this example, the percentage change in BMI) could influence the results in three ways. The first of these three ways was that statistically significant associations were found in one direction when percentage change in BMI was treated categorically. In contrast, they were found in both directions when percentage change in BMI was left as a continuous variable. The second of these three ways was by showing statistically significant associations that are non-existent when the variable is left as a continuous variable (possible false positives) [240]. Specifically, in our categorical analysis, the outcome of progression in inferior acetabular osteophytes in the hip was associated with a decrease in BMI. In contrast, it was not associated with either a decrease or increase in BMI in the continuous analysis (Table 2). The third way was that the analyses with categorized continuous variables might mask statistically significant associations when the variable is left as a continuous variable (possible false negatives) [239]. Specifically, in our categorical analysis, the outcome of the development of any pain in knee was not associated with a decrease nor an increase in BMI. In contrast, it was associated with both a decrease and an increase in BMI in the continuous analysis (Table 2). Further, our sensitivity analyses using 3% and 10% changes in BMI delivered the same conclusions as our primary analyses (that used a 5% change in BMI), showing that these three issues with categorization of continuous variables are independent of these different thresholds of the continuous variable (percentage change in BMI).
Of these three ways that results differed depending on whether the predictor variable was treated as a continuous or a categorical variable, the first one was a major problem as the conclusions drawn from the continuous and categorical analyses results would be different. From the continuous analyses in this study, we would conclude a beneficial association between a decrease in BMI and a harmful association of an increase in BMI for structural changes and pain in knee osteoarthritis over four years, as the effect of percentage change in BMI was shown in both directions (decrease and increase). However, from the categorical analysis in this study, we would have concluded that a decrease in BMI is associated with beneficial effects for knee structure and pain in osteoarthritis but that an increase in BMI is not associated with harmful effects. The conclusion about lack of association between an increase in BMI from the categorical analyses conflicts with the conclusion from the continuous analyses, as well as from previous research showing that weight gain is associated with harmful effects of structural changes and pain in knee osteoarthritis while weight loss is associated with beneficial effects [249, 253,254,255,256]. It is difficult to reconcile that a decrease in BMI is associated with one effect, whereas an increase in BMI is not associated with the opposite effect. This difficulty in the reconciliation of the results from the analysis using categorized continuous variables can also be seen in the study by Joseph et al.[2] which we revisited. That study, which used categorized percentage change in BMI, showed the association of either a decrease in BMI or an increase in BMI with outcomes of structural changes and pain in knee osteoarthritis, but not for both a decrease and an increase in BMI for any outcome.
We acknowledge the limitations in our study. There were 276 (5.8%) of the 4,796 participants with missing data for whom we could not estimate their BMI change. Therefore, we cannot exclude the possibility that missing data could have resulted in bias in our estimates. Additionally, it is important to note that our study only investigated the impact of categorizing one predictor variable (percentage change in BMI) on several outcomes in one specific population (OAI). Therefore, our results cannot be generalized to all situations in which researchers categorize variables in rheumatology. Categorization of continuous predictor variables can be useful in rheumatology research when there is strong prior knowledge or established cut-offs for a particular variable, such as disease activity scores, categorization of antibody titers (anti-citrullinated protein antibody, ACPA positive/negative), or achieving remission or not (yes/no). In such cases, categorization can aid in simplifying the analysis and interpretation of the results.
Conclusions
In conclusion, our study demonstrated that categorizing continuous predictor variables in rheumatology may result in associations being shown in only one direction, and could also lead to possible false positive and possible false negative associations, which may lead to erroneous conclusions. We suggest that researchers in rheumatology, including clinicians and peer reviewers, consider the potential drawbacks of categorizing continuous predictor variables and prioritize the use of continuous variables.
Availability of data and materials
OAI data is openly available to researchers for scientific and educational purposes in the following website https://nda.nih.gov/oai/. The STATA code can be obtained from the authors upon request.
Abbreviations
- BMI:
-
Body Mass Index
- CI:
-
Confidence Interval
- ISF:
-
Individual Structural Feature
- JSN:
-
Joint Space Narrowing
- KL:
-
Kellgren Lawrence
- OAI:
-
Osteoarthritis Initiative
- OARSI:
-
Osteoarthritis Research Society International
- OR:
-
Odds Ratio
- RMSE:
-
Root Mean Square Error
- SD:
-
Standard Deviation
- THR:
-
Total Hip Replacement
- TKR:
-
Total Knee Replacement
References
Jin X, et al. Does weight loss reduce the incidence of total knee and hip replacement for osteoarthritis?-A prospective cohort study among middle-aged and older adults with overweight or obesity. Int J Obes (Lond). 2021;45(8):1696–704.
Joseph GB, et al. Effects of Weight Change on Knee and Hip Radiographic Measurements and Pain Over Four Years: Data From the Osteoarthritis Initiative. Arthritis Care Res (Hoboken). 2023;75(4):860–8.
Huang Y-C, et al. Site-specific mortality in native joint septic arthritis: a national population study. Rheumatology. 2020;59(12):3826–33.
Baker JF, et al. Sarcopenic obesity in rheumatoid arthritis: prevalence and impact on physical functioning. Rheumatology (Oxford). 2022;61(6):2285–94.
Swain S, et al. Comorbidities and use of analgesics in people with knee pain: a study in the Nottingham Knee Pain and Health in the Community (KPIC) cohort. Rheumatol Adv Pract. 2022;6(2):rkac049.
Glintborg B, et al. Is the risk of infection higher during treatment with secukinumab than with TNF inhibitors? An observational study from the Nordic countries. Rheumatology. 2023;62(2):647–58.
Nakafero G, et al. Incidence and pattern of mycophenolate discontinuation associated with abnormal monitoring blood-test results: cohort study using data from the Clinical Practice Research Datalink Aurum. Rheumatol Adv Pract. 2022;6(2):rkac046.
Tanamas SK, et al. Association of weight gain with incident knee pain, stiffness, and functional difficulties: a longitudinal study. Arthritis Care Res (Hoboken). 2013;65(1):34–43.
Nikiphorou E, et al. Remission vs low disease activity: function, quality of life and structural outcomes in the Early Rheumatoid Arthritis Study and Network. Rheumatology. 2019;59(6):1272–80.
Ugarte-Gil MF, et al. Characteristics associated with poor COVID-19 outcomes in individuals with systemic lupus erythematosus: data from the COVID-19 Global Rheumatology Alliance. Ann Rheum Dis. 2022;81(7):970–8.
Gilbert AL, et al. Relationship between self-reported restless sleep and objectively measured physical activity in adults with knee osteoarthritis. Arthritis Care Res (Hoboken). 2021;73(5):687–92.
Wyndow N, et al. Foot orthoses and footwear for the management of patellofemoral osteoarthritis: a pilot randomized trial. Arthritis Care Res (Hoboken). 2021;73(2):240–9.
Buckley LH, et al. Psoriasis associated with tumor necrosis factor inhibitors in children with inflammatory diseases. Arthritis Care Res (Hoboken). 2021;73(2):215–20.
Sari I, et al. Factors predictive of radiographic progression in ankylosing spondylitis. Arthritis Care Res (Hoboken). 2021;73(2):275–81.
Kiadaliri A, Englund M. Intersectional inequalities and individual heterogeneity in chronic rheumatic diseases: an intersectional multilevel analysis. Arthritis Care Res (Hoboken). 2021;73(2):296–304.
Mørk S, et al. Spondyloarthritis and outcomes in pregnancy and labor: a nationwide register-based cohort study. Arthritis Care Res (Hoboken). 2021;73(2):282–8.
Li L, et al. Impact of systemic lupus erythematosus on the risk of newly diagnosed hip fracture: a general population-based study. Arthritis Care Res (Hoboken). 2021;73(2):259–65.
Guinot M, et al. Effects of repetitive transcranial magnetic stimulation and multicomponent therapy in patients with fibromyalgia: a randomized controlled trial. Arthritis Care Res (Hoboken). 2021;73(3):449–58.
Thomsen RS, et al. Adiposity and physical activity as risk factors for developing psoriatic arthritis: longitudinal data from a population-based study in Norway. Arthritis Care Res (Hoboken). 2021;73(3):432–41.
Jetha A, Johnson SR, Gignac MAM. Unmet workplace support needs and lost productivity of workers with systemic sclerosis: a path analysis study. Arthritis Care Res (Hoboken). 2021;73(3):423–31.
Horton DB, et al. Making decisions about stopping medicines for well-controlled juvenile idiopathic arthritis: a mixed-methods study of patients and caregivers. Arthritis Care Res (Hoboken). 2021;73(3):374–85.
Rogers JL, et al. Using clinical characteristics and patient-reported outcome measures to categorize systemic lupus erythematosus subtypes. Arthritis Care Res (Hoboken). 2021;73(3):386–93.
Iijima H, et al. low back pain as a risk factor for recurrent falls in people with knee osteoarthritis. Arthritis Care Res (Hoboken). 2021;73(3):328–35.
Garg S, et al. High burden of premature arteriosclerosis on renal biopsy results in incident lupus nephritis. Arthritis Care Res (Hoboken). 2021;73(3):394–401.
Zaccardelli A, et al. Elevated anti-citrullinated protein antibodies prior to rheumatoid arthritis diagnosis and risks for chronic obstructive pulmonary disease or asthma. Arthritis Care Res (Hoboken). 2021;73(4):498–509.
Mehta B, et al. Community deprivation index and discharge destination after elective hip replacement. Arthritis Care Res (Hoboken). 2021;73(4):531–9.
Eder L, et al. Health care utilization for musculoskeletal issues during the prediagnosis period in psoriatic arthritis: a population-based study. Arthritis Care Res (Hoboken). 2021;73(5):680–6.
Salmasi S, et al. Adherence to antimalarial therapy and risk of type 2 diabetes mellitus among patients with systemic lupus erythematosus: a population-based study. Arthritis Care Res (Hoboken). 2021;73(5):702–6.
Gao Y, et al. Adherence to the dietary approaches to stop hypertension diet and hyperuricemia: a cross-sectional study. Arthritis Care Res (Hoboken). 2021;73(4):603–11.
Wang Y, et al. Knee osteoarthritis, potential mediators, and risk of all-cause mortality: data from the Osteoarthritis Initiative. Arthritis Care Res (Hoboken). 2021;73(4):566–73.
Schletzbaum M, et al. Predictors of smoking cessation in patients with rheumatoid arthritis in two cohorts: most predictive health care factors. Arthritis Care Res (Hoboken). 2021;73(5):633–9.
Shaw Y, et al. Acceptability and content validity of patient-reported outcome measures considered from the perspective of patients with rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(4):510–9.
Simard JF, et al. Maternal hypertensive disorders in pregnant women with systemic lupus erythematosus and future cardiovascular outcomes. Arthritis Care Res (Hoboken). 2021;73(4):574–9.
Fairley JL, et al. Clinical features of systemic sclerosis-mixed connective tissue disease and systemic sclerosis overlap syndromes. Arthritis Care Res (Hoboken). 2021;73(5):732–41.
Escobar Martinez A, Perera RA, Riddle DL. Development and underlying structure of a second-generation appropriateness classification system for total knee arthroplasty. Arthritis Care Res (Hoboken). 2021;73(6):801–9.
Lee YC, et al. Defining pain that does not interfere with activities among rheumatoid arthritis patients. Arthritis Care Res (Hoboken). 2021;73(5):626–32.
Berkovic D, et al. “The financial impact is depressing and anxiety inducing”: a qualitative exploration of the personal financial toll of arthritis. Arthritis Care Res (Hoboken). 2021;73(5):671–9.
Arnold JB, et al. Foot and leg muscle weakness in people with midfoot osteoarthritis. Arthritis Care Res (Hoboken). 2021;73(6):772–80.
Sauer BC, et al. Potential for major therapeutic changes to produce significant clinical response across a broad range of disease activity: an observational study of US veterans with rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(7):964–74.
Cozier YC, et al. Association of child abuse and systemic lupus erythematosus in Black women during adulthood. Arthritis Care Res (Hoboken). 2021;73(6):833–40.
Roth M, et al. Changes in medial meniscal three-dimensional position and morphology as predictors of knee replacement in rapidly progressing knee osteoarthritis: data from the Osteoarthritis Initiative. Arthritis Care Res (Hoboken). 2021;73(7):1031–7.
Carlesso LC, et al. Association of intermittent and constant knee pain patterns with knee pain severity and with radiographic knee osteoarthritis duration and severity. Arthritis Care Res (Hoboken). 2021;73(6):788–93.
Holmqvist M, et al. Findings on coronary angiographies in patients with rheumatoid arthritis and ischemic heart disease: are they different from patients without rheumatoid arthritis? Arthritis Care Res (Hoboken). 2021;73(5):658–65.
Knudsen SS, et al. Association of rheumatoid arthritis in pregnancy with school performance of offspring: a Danish nationwide register-based study. Arthritis Care Res (Hoboken). 2021;73(7):975–82.
Wiese MD, et al. Precision medicine with leflunomide: consideration of the DHODH haplotype and plasma teriflunomide concentration and modification of outcomes in patients with rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(7):983–9.
Berry KM, et al. Obesity progression between young adulthood and midlife and incident arthritis: a retrospective cohort study of US adults. Arthritis Care Res (Hoboken). 2021;73(3):318–27.
Curtis JR, et al. Real-world outcomes associated with methotrexate, sulfasalazine, and hydroxychloroquine triple therapy versus tumor necrosis factor inhibitor/methotrexate combination therapy in patients with rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(8):1114–24.
Touma Z, et al. Identifying a response for the systemic lupus erythematosus disease activity 2000 glucocorticoid index. Arthritis Care Res (Hoboken). 2021;73(9):1243–9.
Olofsson T, et al. Pain over two years after start of biologic versus conventional combination treatment in early rheumatoid arthritis: results from a Swedish randomized controlled trial. Arthritis Care Res (Hoboken). 2021;73(9):1312–21.
Wilfong JM, et al. Participation in informal caregiving among people with arthritis: findings from a Canadian longitudinal study on aging. Arthritis Care Res (Hoboken). 2021;73(10):1528–36.
Hernández-Muñoz JJ, Wei W, Sierra-Zorita R. Prevalence of rheumatoid arthritis and drug dispensing patterns among Medicaid and Medicaid-Medicare dually eligible beneficiaries in Puerto Rico. Arthritis Care Res (Hoboken). 2021;73(2):199–206.
Li CKH, et al. Safety and tolerability of subcutaneous methotrexate in routine clinical practice. Arthritis Care Res (Hoboken). 2021;73(9):1306–11.
Baker JF, Katz P, Michaud K. Defining minimum clinically important changes for the Patient Activity Scale II. Arthritis Care Res (Hoboken). 2021;73(10):1456–60.
Helget LN, et al. Incidence, prevalence, and burden of gout in the Veterans Health Administration. Arthritis Care Res (Hoboken). 2021;73(9):1363–71.
Collins JE, Neogi T, Losina E. Trajectories of structural disease progression in knee osteoarthritis. Arthritis Care Res (Hoboken). 2021;73(9):1354–62.
Prior JA, et al. Rheumatic conditions as risk factors for self-harm: a retrospective cohort study. Arthritis Care Res (Hoboken). 2021;73(1):130–7.
Williams JN, et al. Racial differences in contraception encounters and dispensing among female medicaid beneficiaries with systemic lupus erythematosus. Arthritis Care Res (Hoboken). 2021;73(10):1396–404.
Singh JA, Cleveland JD. Epidemiology, time trends, and outcomes of serious infections in patients with vasculitis: a nineteen-year national study. Arthritis Care Res (Hoboken). 2021;73(10):1544–51.
Miura D, et al. Lifetime risk of primary shoulder arthroplasty from 2008 to 2017: a population-level analysis using National Registry Data. Arthritis Care Res (Hoboken). 2021;73(10):1511–7.
Anastasiou C, et al. Mortality among hospitalized individuals with systemic lupus erythematosus in the US between 2006 and 2016. Arthritis Care Res (Hoboken). 2021;73(10):1444–50.
Katz P, et al. Impact of limited health literacy on patient-reported outcomes in systemic lupus erythematosus. Arthritis Care Res (Hoboken). 2021;73(1):110–9.
Jacobs H, et al. Use of physical therapy in patients with osteoarthritis in Germany: an analysis of a linkage of claims and survey data. Arthritis Care Res (Hoboken). 2021;73(7):1013–22.
Perruccio AV, et al. Understanding the association between osteoarthritis and social participation: the Canadian longitudinal study on aging. Arthritis Care Res (Hoboken). 2021;73(11):1638–47.
Jetha A, et al. Depressive symptoms and the arthritis-employment interface: a population-level study. Arthritis Care Res (Hoboken). 2021;73(1):65–77.
George MD, et al. Variability in glucocorticoid prescribing for rheumatoid arthritis and the influence of provider preference on long-term use of glucocorticoids. Arthritis Care Res (Hoboken). 2021;73(11):1597–605.
Hitchon CA, et al. Impact of psychiatric comorbidity on health care use in rheumatoid arthritis: a population-based study. Arthritis Care Res (Hoboken). 2021;73(1):90–9.
Stojan G, et al. Cachexia in systemic lupus erythematosus: risk factors and relation to disease activity and damage. Arthritis Care Res (Hoboken). 2021;73(11):1577–82.
Minhas D, et al. Access and cost-related nonadherence to prescription medications among lupus patients and controls: the Michigan lupus epidemiology and surveillance program. Arthritis Care Res (Hoboken). 2021;73(11):1561–7.
Dietz B, et al. Major depression and adverse patient-reported outcomes in systemic lupus erythematosus: results from a prospective longitudinal cohort. Arthritis Care Res (Hoboken). 2021;73(1):48–54.
Ra JH, et al. Patient perspective on using digital resources to address unmet needs in systemic lupus erythematosus. Arthritis Care Res (Hoboken). 2021;73(11):1568–76.
Partington R, et al. Mortality among patients with polymyalgia rheumatica: a retrospective cohort study. Arthritis Care Res (Hoboken). 2021;73(12):1853–7.
Siviero P, et al. Factors associated with functional decline in hand and hip/knee osteoarthritis after one year: data from a population-based study. Arthritis Care Res (Hoboken). 2021;73(9):1343–53.
Jun JB, et al. Significant decline in the incidence of Behçet’s disease in South Korea: a nationwide population-based study (2004–2017). Arthritis Care Res (Hoboken). 2021;73(11):1689–96.
Bruera S, et al. Cervical cancer screening in women with systemic lupus erythematosus. Arthritis Care Res (Hoboken). 2021;73(12):1796–803.
Davis AM, et al. Limiting the risk of osteoarthritis after anterior cruciate ligament injury: are health care providers missing the opportunity to intervene? Arthritis Care Res (Hoboken). 2021;73(12):1754–62.
Hernández-Hernández MV, et al. Impact of disease activity on physical activity in patients with psoriatic arthritis. Arthritis Care Res (Hoboken). 2021;73(12):1834–44.
Charen DA, et al. Examining the association of knee pain with modifiable cardiometabolic risk factors. Arthritis Care Res (Hoboken). 2021;73(12):1777–83.
Flowers P, et al. Foot osteoarthritis frequency and associated factors in a community-based cross-sectional study of White and African American adults. Arthritis Care Res (Hoboken). 2021;73(12):1784–8.
Chew E, et al. Measurement of minimal disease activity in psoriatic arthritis using the patient-reported outcomes measurement information system-physical function or the health assessment questionnaire disability index. Arthritis Care Res (Hoboken). 2022;74(1):151–60.
Carlesso LC, et al. Association of pain sensitization and conditioned pain modulation to pain patterns in knee osteoarthritis. Arthritis Care Res (Hoboken). 2022;74(1):107–12.
Sparks JA, et al. Depression and subsequent risk for incident rheumatoid arthritis among women. Arthritis Care Res (Hoboken). 2021;73(1):78–89.
Barbhaiya M, et al. Association of dietary quality with risk of incident systemic lupus erythematosus in the nurses’ health study and nurses’ health study II. Arthritis Care Res (Hoboken). 2021;73(9):1250–8.
Zhao N, et al. Sunlight exposure, sun-protective behavior, and anti-citrullinated protein antibody positivity: a general population-based study in Quebec, Canada. Arthritis Care Res (Hoboken). 2022;74(2):236–42.
Sepriano A, et al. Inflammation of the sacroiliac joints and spine and structural changes on magnetic resonance imaging in axial spondyloarthritis: five-year data from the DESIR cohort. Arthritis Care Res (Hoboken). 2022;74(2):243–50.
Ardalan K, et al. Parent perspectives on addressing emotional health for children and young adults with juvenile myositis. Arthritis Care Res (Hoboken). 2021;73(1):18–29.
Morlà RM, et al. Multidimensional health assessment questionnaire as an effective tool to screen for depression in routine rheumatology care. Arthritis Care Res (Hoboken). 2021;73(1):120–9.
Curtis JR, et al. Data-driven patient clustering and differential clinical outcomes in the Brigham and women’s rheumatoid arthritis sequential study registry. Arthritis Care Res (Hoboken). 2021;73(4):471–80.
D’Silva KM, et al. Hypouricemia and mortality risk in the US general population. Arthritis Care Res (Hoboken). 2021;73(8):1171–9.
Dell’Isola A, et al. Willingness to undergo joint surgery following a first-line intervention for osteoarthritis: data from the better management of people with osteoarthritis register. Arthritis Care Res (Hoboken). 2021;73(6):818–27.
Singh JA, et al. Management of rheumatic diseases during the COVID-19 pandemic: a national veterans affairs survey of rheumatologists. Arthritis Care Res (Hoboken). 2021;73(7):998–1003.
Barbhaiya M, et al. Development of a new international antiphospholipid syndrome classification criteria phase I/II report: generation and reduction of candidate criteria. Arthritis Care Res (Hoboken). 2021;73(10):1490–501.
Maldonado D, et al. Association of medication access difficulty and COVID-19-related distress with disease flares in rheumatology patients during the COVID-19 pandemic. Arthritis Care Res (Hoboken). 2021;73(8):1162–70.
Barbhaiya M, et al. Pregnancy and rheumatic disease: experience at a single center in New York city during the COVID-19 pandemic. Arthritis Care Res (Hoboken). 2021;73(7):1004–12.
Jamshidi A, et al. Machine learning-based individualized survival prediction model for total knee replacement in osteoarthritis: data from the Osteoarthritis Initiative. Arthritis Care Res (Hoboken). 2021;73(10):1518–27.
Michaud K, et al. Changes in disease-modifying antirheumatic drug treatment for patients with rheumatoid arthritis in the US during the COVID-19 pandemic: a three-month observational study. Arthritis Care Res (Hoboken). 2021;73(9):1322–31.
Jetha A, et al. Impact of the COVID-19 pandemic on the employment of Canadian young adults with rheumatic disease: findings from a longitudinal survey. Arthritis Care Res (Hoboken). 2021;73(8):1146–52.
George MD, et al. Disruptions in rheumatology care and the rise of telehealth in response to the COVID-19 pandemic in a community practice-based network. Arthritis Care Res (Hoboken). 2021;73(8):1153–61.
Feldman CH, Xu C, Costenbader KH. Avoidable acute care use for vaccine-preventable illnesses among Medicaid beneficiaries with lupus. Arthritis Care Res (Hoboken). 2021;73(9):1236–42.
To-Miles F, et al. Exploring the associations among occupational balance and health of adults with and without inflammatory arthritis. Arthritis Care Res (Hoboken). 2022;74(1):22–30.
Albrecht K, et al. Association between rheumatoid arthritis disease activity and periodontitis defined by tooth loss: longitudinal and cross-sectional data from two observational studies. Arthritis Care Res (Hoboken). 2021. https://doi.org/10.1002/acr.24799.
Austin K, et al. Patient perceptions of physical activity after a diagnosis of giant cell arteritis: analysis of multinational qualitative data. Arthritis Care Res (Hoboken). 2022;74(1):99–106.
Huang S, et al. Association between inflammation, incident heart failure, and heart failure subtypes in patients with rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021. https://doi.org/10.1002/acr.24804.
Chernis J, et al. Predictors of perceived functional status in early systemic sclerosis: a prospective longitudinal study of an early disease cohort. Arthritis Care Res (Hoboken). 2021. https://doi.org/10.1002/acr.24812.
Clowse MEB, et al. Pregnancy outcomes in patients with interstitial lung disease. Arthritis Care Res (Hoboken). 2021. https://doi.org/10.1002/acr.24814.
Rubinstein TB, et al. Multicenter study of utility and acceptability of depression and anxiety screening in adolescents and young adults with childhood-onset systemic lupus. Arthritis Care Res. 2023;75(4):724–33.
Perruccio AV, et al. Cardiovascular risk profile and osteoarthritis-considering sex and multisite joint involvement: a Canadian longitudinal study on aging population-based study. Arthritis Care Res. 2021;75:893–901.
Allison K, et al. Effects of an online education program on physical therapists’ confidence in weight management for people with osteoarthritis: a randomized controlled trial. Arthritis Care Res. 2021;75:835–47.
Kellahan SR, et al. Depressed symptomatology in systemic lupus erythematosus patients. Arthritis Care Res. 2021;75:749–57.
Losina E, et al. Opioid use among medicare beneficiaries with knee osteoarthritis: prevalence and correlates of chronic use. Arthritis Care Res. 2021;75:876–84.
Jorge A, et al. Hydroxychloroquine use and cardiovascular events among patients with systemic lupus erythematosus and rheumatoid arthritis. Arthritis Care Res. 2021;75:743–8.
Hunter DJ, et al. Health literacy and appropriateness of self-care and pain management in osteoarthritis: an understanding of the patient’s perspective. Arthritis Care Res. 2021;75:848–59.
Al-Mayouf SM, et al. Cumulative damage in juvenile idiopathic arthritis: a multicenter study from the Pediatric Rheumatology Arab Group. Arthritis Care Res (Hoboken). 2021;73(4):586–92.
Andersen JB, et al. Incidence trends and mortality of giant cell arteritis in southern Norway. Arthritis Care Res (Hoboken). 2021;73(3):409–14.
Avrech Bar M, et al. Fatherhood experiences of men with inflammatory arthritis: a preliminary grounded theory. Arthritis Care Res (Hoboken). 2021;73(6):885–92.
Baer AN, et al. Ultrasound-guided biopsy of suspected salivary gland lymphoma in Sjögren’s syndrome. Arthritis Care Res (Hoboken). 2021;73(6):849–55.
Baker JF, et al. Association of low muscle density with deteriorations in muscle strength and physical functioning in rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(3):355–63.
Baker JF, et al. Obesity and the risk of incident chronic opioid use in rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(10):1405–12.
Bakker MM, et al. Addressing health literacy needs in rheumatology: which patient health literacy profiles need the attention of health professionals? Arthritis Care Res (Hoboken). 2021;73(1):100–9.
Barber CEH, et al. A population-based approach to reporting system-level performance measures for rheumatoid arthritis care. Arthritis Care Res (Hoboken). 2021;73(5):640–8.
Bernatsky S, et al. Cancer risk in a large inception systemic lupus erythematosus cohort: effects of demographic characteristics, smoking, and medications. Arthritis Care Res (Hoboken). 2021;73(12):1789–95.
Bingham CO 3rd, et al. Conversion of functional assessment of chronic illness therapy-fatigue to patient-reported outcomes measurement information system fatigue scores in two phase III baricitinib rheumatoid arthritis trials. Arthritis Care Res (Hoboken). 2021;73(4):481–8.
Bitencourt N, et al. Time to first completed visit and health care utilization among young adults transferring from pediatric to adult rheumatologic care in a safety-net hospital. Arthritis Care Res (Hoboken). 2021;73(12):1730–8.
Blalock SJ, et al. Enhancing patient understanding of medication risks and benefits. Arthritis Care Res (Hoboken). 2022;74(1):142–50.
Bortoluzzi A, et al. Improved pregnancy outcome in patients with rheumatoid arthritis who followed an ideal clinical pathway. Arthritis Care Res (Hoboken). 2021;73(2):166–72.
Bove AM, et al. Race differences in postacute physical therapy utilization and patient-reported function after total knee arthroplasty. Arthritis Care Res (Hoboken). 2022;74(1):79–88.
Bowers E, et al. Quality improvement intervention to reduce thirty-day hospital readmission rates among patients with systemic lupus erythematosus. Arthritis Care Res (Hoboken). 2022;74(1):126–30.
Brunner HI, et al. Functional ability and health-related quality of life in randomized controlled trials of tocilizumab in patients with juvenile idiopathic arthritis. Arthritis Care Res (Hoboken). 2021;73(9):1264–74.
Bühler M, et al. Impact of thumb carpometacarpal joint osteoarthritis: a pragmatic qualitative study. Arthritis Care Res (Hoboken). 2021;73(3):336–46.
Buldt AK, et al. Structural characteristics associated with radiographic severity of first metatarsophalangeal joint osteoarthritis. Arthritis Care Res (Hoboken). 2021;73(7):1023–30.
Chang JC, et al. Impact of psychiatric diagnosis and treatment on medication adherence in youth with systemic lupus erythematosus. Arthritis Care Res (Hoboken). 2021;73(1):30–8.
Coates LC, et al. Performance and predictors of minimal disease activity response in patients with peripheral spondyloarthritis treated with adalimumab. Arthritis Care Res (Hoboken). 2022;74(2):259–67.
Coates LC, et al. The phenotype of axial spondyloarthritis: is it dependent on HLA-B27 status? Arthritis Care Res (Hoboken). 2021;73(6):856–60.
Coussens M, et al. Does muscle strength change over time in patients with hypermobile Ehlers-Danlos syndrome/hypermobility spectrum disorder? An eight-year follow-up study. Arthritis Care Res (Hoboken). 2021;73(7):1041–8.
Davis A, et al. Immunomodulatory medication use in newly diagnosed youth with systemic lupus erythematosus. Arthritis Care Res (Hoboken). 2021;73(11):1672–7.
de Jesús GR, et al. Soluble Flt-1, placental growth factor, and vascular endothelial growth factor serum levels to differentiate between active lupus nephritis during pregnancy and preeclampsia. Arthritis Care Res (Hoboken). 2021;73(5):717–21.
Desai SP, et al. Treat-to-target approach in rheumatoid arthritis: a quality improvement trial. Arthritis Care Res (Hoboken). 2021;73(2):207–14.
Dhingra R, et al. Prescription claims for immunomodulator and antiinflammatory drugs among persons with Ehlers-Danlos syndromes. Arthritis Care Res (Hoboken). 2021. https://doi.org/10.1002/acr.24819.
Doumen M, et al. Psychosocial burden predicts sustained remission in early rheumatoid arthritis: unraveling the complex interplay of well-being and disease activity. Arthritis Care Res. 2021;75:758–67.
Eitner A, et al. Impact of diabetes mellitus on knee osteoarthritis pain and physical and mental status: data from the Osteoarthritis Initiative. Arthritis Care Res (Hoboken). 2021;73(4):540–8.
Fadanelli G, et al. Methotrexate in linear scleroderma: long-term efficacy in fifty children from a single pediatric rheumatology center. Arthritis Care Res (Hoboken). 2021;73(9):1259–63.
Fairchild R, et al. Ultrasound detection of calcinosis and association with ulnar artery occlusion in patients with systemic sclerosis. Arthritis Care Res (Hoboken). 2021;73(9):1332–7.
Fairchild R, et al. Development and assessment of novel lung ultrasound interpretation criteria for the detection of interstitial lung disease in systemic sclerosis. Arthritis Care Res (Hoboken). 2021;73(9):1338–42.
Fan L, et al. Clinical scenario and long-term outcome of childhood Takayasu arteritis undergoing 121 endovascular interventions: a large cohort over a fifteen-year period. Arthritis Care Res (Hoboken). 2021;73(11):1678–88.
Ferreira RJO, et al. Patient global assessment of disease activity and radiographic progression in early arthritis: three-year results from the ESPOIR cohort. Arthritis Care Res (Hoboken). 2021;73(9):1300–5.
Forman CJ, et al. Association of race and risk of future scleroderma renal crisis at systemic sclerosis diagnosis. Arthritis Care Res. 2021;75:801–807.
Garrido-Cumbrera M, et al. Patient-reported impact of axial spondyloarthritis on working life: results from the European map of axial spondyloarthritis survey. Arthritis Care Res (Hoboken). 2021;73(12):1826–33.
Giancane G, et al. Determinants of discordance between criteria for inactive disease and low disease activity in juvenile idiopathic arthritis. Arthritis Care Res (Hoboken). 2021;73(12):1722–9.
Gilbert AL, et al. What is an important difference in gait speed in adults with knee osteoarthritis? Arthritis Care Res (Hoboken). 2021;73(4):559–65.
Gupta A, et al. Association of hydroxychloroquine use with decreased incident atrial fibrillation in systemic lupus erythematosus. Arthritis Care Res (Hoboken). 2021;73(6):828–32.
Hahn J, et al. Associations between smoking and systemic lupus erythematosus-related cytokines and chemokines among US female nurses. Arthritis Care Res (Hoboken). 2021;73(11):1583–9.
Hammer NM, et al. Preferences for self-management and support services in patients with inflammatory joint disease: a Danish nationwide cross-sectional study. Arthritis Care Res (Hoboken). 2021;73(10):1479–89.
Hart HF, et al. Associations between cadence and knee loading in patients with knee osteoarthritis. Arthritis Care Res (Hoboken). 2021;73(11):1667–71.
Hatfield GL, et al. Baseline gait muscle activation patterns differ for osteoarthritis patients who undergo total knee arthroplasty five to eight years later from those who do not. Arthritis Care Res (Hoboken). 2021;73(4):549–58.
Haugeberg G, et al. Physical and psychosocial burden of psoriatic arthritis: longitudinal data from a population-based study in Norway. Arthritis Care Res (Hoboken). 2021;73(1):138–45.
Häuser W, et al. Modified 2016 American College of Rheumatology Fibromyalgia Criteria, the Analgesic, Anesthetic, and Addiction Clinical Trial Translations Innovations Opportunities and Networks-American Pain Society Pain Taxonomy, and the Prevalence of Fibromyalgia. Arthritis Care Res (Hoboken). 2021;73(5):617–25.
Hegarty RSM, et al. Understanding fatigue-related disability in rheumatoid arthritis and ankylosing spondylitis: the importance of daily correlates. Arthritis Care Res (Hoboken). 2021;73(9):1282–9.
Herrera A, et al. Illness experiences of Chilean women with Sjögren’s syndrome: the patient perspective. Arthritis Care Res (Hoboken). 2021;73(8):1210–8.
Hirsch JK, et al. Positive psychological factors and impairment in rheumatic and musculoskeletal disease: do psychopathology and sleep quality explain the linkage? Arthritis Care Res (Hoboken). 2021;73(1):55–64.
Hoque A, et al. Measuring inflammatory foot disease in rheumatoid arthritis: development and validation of the rheumatoid arthritis foot disease activity index-5. Arthritis Care Res (Hoboken). 2021;73(9):1290–9.
Hughes M, et al. Factors influencing patient decision-making concerning treatment escalation in Raynaud’s phenomenon secondary to systemic sclerosis. Arthritis Care Res (Hoboken). 2021;73(12):1845–52.
Iragorri N, et al. Model to determine the cost-effectiveness of screening psoriasis patients for psoriatic arthritis. Arthritis Care Res (Hoboken). 2021;73(2):266–74.
Izadi Z, et al. Significant gains in rheumatoid arthritis quality measures among RISE registry practices. Arthritis Care Res (Hoboken). 2022;74(2):219–28.
Jiang X, et al. Precision medicine approach to develop and internally validate optimal exercise and weight-loss treatments for overweight and obese adults with knee osteoarthritis: data from a single-center randomized trial. Arthritis Care Res (Hoboken). 2021;73(5):693–701.
Johnson AJ, et al. Patterns and correlates of self-management strategies for osteoarthritis-related pain among older non-Hispanic Black and non-Hispanic White adults. Arthritis Care Res (Hoboken). 2021;73(11):1648–58.
Jones JT, et al. Down syndrome-associated arthritis cohort in the new childhood arthritis and rheumatology research alliance registry: clinical characteristics, treatment, and outcomes. Arthritis Care Res (Hoboken). 2021;73(12):1739–45.
Jorge A, et al. Hydroxychloroquine and mortality among patients with systemic lupus erythematosus in the general population. Arthritis Care Res (Hoboken). 2021;73(8):1219–23.
Karpes Matusevich AR, et al. Treatment sequences after discontinuing a tumor necrosis factor inhibitor in patients with rheumatoid arthritis: a comparison of cycling versus swapping strategies. Arthritis Care Res (Hoboken). 2021;73(10):1461–9.
Klimek M, et al. Long-term mortality of patients with osteoarthritis after joint replacement: prognostic value of preoperative and postoperative pain and function. Arthritis Care Res (Hoboken). 2021;75(4):869-875.
Kowitt SD, et al. How are neighborhood characteristics associated with mental and physical functioning among older adults with radiographic knee osteoarthritis? Arthritis Care Res (Hoboken). 2021;73(3):308–17.
Kronzer VL, et al. Family history of rheumatic, autoimmune, and nonautoimmune diseases and risk of rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(2):180–7.
Kumar D, et al. Association of quadriceps adiposity with an increase in knee cartilage, meniscus, or bone marrow lesions over three years. Arthritis Care Res (Hoboken). 2021;73(8):1134–9.
Kuntze G, et al. Consequences of juvenile idiopathic arthritis on single leg squat performance in youth. Arthritis Care Res (Hoboken). 2021;73(8):1187–93.
Kuntze G, et al. Vertical drop jump performance in youth with juvenile idiopathic arthritis. Arthritis Care Res (Hoboken). 2021;73(7):955–63.
Lanni S, et al. Comparison between clinical and ultrasound assessment of the ankle region in children with juvenile idiopathic arthritis. Arthritis Care Res (Hoboken). 2021;73(8):1180–6.
Leung YY, et al. Comparing the patient-reported physical function outcome measures in a real-life international cohort of patients with psoriatic arthritis. Arthritis Care Res (Hoboken). 2021;73(4):593–602.
Liao KP, et al. Coronary microvascular dysfunction in rheumatoid arthritis compared to diabetes mellitus and association with all-cause mortality. Arthritis Care Res (Hoboken). 2021;73(2):159–65.
Lopez-Olivo MA, et al. Randomized controlled trial of patient education tools for patients with rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(10):1470–8.
Magnusson K, et al. Relationship between magnetic resonance imaging features and knee pain over six years in knees without radiographic osteoarthritis at baseline. Arthritis Care Res (Hoboken). 2021;73(11):1659–66.
Mancuso CA, et al. Modifications in systemic rheumatic disease medications: patients’ perspectives during the height of the COVID-19 pandemic in New York city. Arthritis Care Res (Hoboken). 2021;73(6):909–17.
Marino A, et al. Early joint replacement in juvenile idiopathic arthritis: trend over time and factors influencing implant survival. Arthritis Care Res (Hoboken). 2021;73(9):1275–81.
Marshall M, et al. Identifying long-term trajectories of foot pain severity and potential prognostic factors: a population-based cohort study. Arthritis Care Res (Hoboken). 2021. https://doi.org/10.1002/acr.24823.
Mattap SM, et al. Hand examination, ultrasound, and the association with hand pain and function in community-based older adults. Arthritis Care Res (Hoboken). 2021;73(3):347–54.
Matza MA, et al. Ustekinumab for the treatment of giant cell arteritis. Arthritis Care Res (Hoboken). 2021;73(6):893–7.
McClendon J, et al. Cumulative disadvantage and disparities in depression and pain among veterans with osteoarthritis: the role of perceived discrimination. Arthritis Care Res (Hoboken). 2021;73(1):11–7.
Merkel PA, et al. Long-term safety of rituximab in granulomatosis with polyangiitis and in microscopic polyangiitis. Arthritis Care Res (Hoboken). 2021;73(9):1372–8.
Muratore F, et al. Association between specimen length and number of sections and diagnostic yield of temporal artery biopsy for giant cell arteritis. Arthritis Care Res (Hoboken). 2021;73(3):402–8.
Murphy SL, et al. Fatigue and its association with social participation, functioning, and quality of life in systemic sclerosis. Arthritis Care Res (Hoboken). 2021;73(3):415–22.
O’Leary H, et al. Relationship between pain and sedentary behavior in rheumatoid arthritis patients: a cross-sectional study. Arthritis Care Res (Hoboken). 2021;73(7):990–7.
Ogura T, et al. Finger joint cartilage evaluated by semiquantitative ultrasound score in patients with rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(2):173–9.
Pedersini P, et al. Effects of neurodynamic mobilizations on pain hypersensitivity in patients with hand osteoarthritis compared to robotic assisted mobilization: a randomized controlled trial. Arthritis Care Res (Hoboken). 2021;73(2):232–9.
Peterson RG, et al. Variation in treatment of children hospitalized with new-onset systemic juvenile idiopathic arthritis in the US. Arthritis Care Res (Hoboken). 2021;73(12):1714–21.
Quevedo-Abeledo JC, et al. Differences in capacity of high-density lipoprotein cholesterol efflux between patients with systemic lupus erythematosus and rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(11):1590–6.
Riddle DL, et al. Racial differences in pain and function following knee arthroplasty: a secondary analysis from a multicenter randomized clinical trial. Arthritis Care Res (Hoboken). 2021;73(6):810–7.
Robbins SR, et al. Effectiveness of stepped-care intervention in overweight and obese patients with medial tibiofemoral osteoarthritis: a randomized controlled trial. Arthritis Care Res (Hoboken). 2021;73(4):520–30.
Robinson PC, Dalbeth N, Donovan P. Cost-effectiveness of colchicine prophylaxis for gout flares when commencing allopurinol. Arthritis Care Res (Hoboken). 2021;73(10):1537–43.
Rumsey DG, et al. Juvenile spondyloarthritis in the Childhood Arthritis and Rheumatology Research Alliance Registry: high biologic use, low prevalence of HLA-B27, and equal sex representation in sacroiliitis. Arthritis Care Res (Hoboken). 2021;73(7):940–6.
Santo L, et al. Trends in office visits during which opioids were prescribed for adults with arthritis in the US, 2006–2015. Arthritis Care Res (Hoboken). 2021;73(10):1430–5.
Schmajuk G, et al. Quality of care for patients with systemic lupus erythematosus: data from the American College of Rheumatology RISE registry. Arthritis Care Res (Hoboken). 2022;74(2):179–86.
Seven S, et al. Anatomic distribution of sacroiliac joint lesions on magnetic resonance imaging in patients with axial spondyloarthritis and control subjects: a prospective cross-sectional study, including postpartum women, patients with disc herniation, cleaning staff, runners, and healthy individuals. Arthritis Care Res (Hoboken). 2021;73(5):742–54.
Sharma S, et al. Reliability, validity, responsiveness, and minimum important change of the stair climb test in adults with hip and knee osteoarthritis. Arthritis Care Res (Hoboken). 2021. https://doi.org/10.1002/acr.24821.
Shoop-Worrall SJW, et al. Common functional ability score for young people with juvenile idiopathic arthritis. Arthritis Care Res (Hoboken). 2021;73(7):947–54.
Singh JA, Cleveland JD. Serious infections in patients with gout in the US: a national study of incidence, time trends, and outcomes. Arthritis Care Res (Hoboken). 2021;73(6):898–908.
Slight-Webb S, et al. Unique serum immune phenotypes and stratification of Oklahoma native American rheumatic disease patients. Arthritis Care Res. 2021;75(4):936–46.
Sneed RS, et al. Using critical race theory to understand trial participation among Black individuals with systemic lupus erythematosus: a qualitative study of patients and caregivers. Arthritis Care Res (Hoboken). 2021;73(10):1387–95.
Tedeschi SK, et al. Classifying pseudogout using machine learning approaches with electronic health record data. Arthritis Care Res (Hoboken). 2021;73(3):442–8.
Tolend M, et al. Discrete choice experiment on a magnetic resonance imaging scoring system for temporomandibular joints in juvenile idiopathic arthritis. Arthritis Care Res (Hoboken). 2022;74(2):308–16.
Tollisen A, et al. Treatment satisfaction with and adherence to disease-modifying antirheumatic drugs in adult patients with juvenile idiopathic arthritis. Arthritis Care Res (Hoboken). 2021;73(2):221–31.
Touma Z, et al. Cognitive function trajectories in association with the depressive symptoms trajectories in systemic lupus erythematosus over time. Arthritis Care Res (Hoboken). 2021;73(10):1436–43.
Tu L, et al. Effects of Vitamin D supplementation on disabling foot pain in patients with symptomatic knee osteoarthritis. Arthritis Care Res (Hoboken). 2021;73(6):781–7.
Ugarte-Gil MF, et al. Applying the 2019 European Alliance of Associations for Rheumatology/American College of Rheumatology lupus criteria to patients from the LUMINA cohort: results from the multiethnic, multicenter US cohort. Arthritis Care Res (Hoboken). 2021;73(10):1451–5.
Ursin K, et al. Factors associated with time to pregnancy in women with axial spondyloarthritis: a registry-based multicenter study. Arthritis Care Res (Hoboken). 2021;73(8):1201–9.
van Beest S, et al. Two-year changes in magnetic resonance imaging features and pain in thumb base osteoarthritis. Arthritis Care Res (Hoboken). 2021;73(11):1628–37.
van Heuckelum M, et al. Are health care professionals’ implicit and explicit attitudes toward conventional disease-modifying antirheumatic drugs associated with those of their patients? Arthritis Care Res (Hoboken). 2021;73(3):364–73.
Wang EJ, et al. Targeted program in an academic rheumatology practice to improve compliance with opioid prescribing guidelines for the treatment of chronic pain. Arthritis Care Res (Hoboken). 2021;73(10):1425–9.
Weiss PF, et al. Variability in interpretation of magnetic resonance imaging of the pediatric sacroiliac joint. Arthritis Care Res (Hoboken). 2021;73(6):841–8.
Yen EY, Singh DR, Singh RR. Trends in systemic sclerosis mortality over forty-eight years, 1968–2015: a US population-based study. Arthritis Care Res (Hoboken). 2021;73(10):1502–10.
Yoshida K, et al. Canakinumab’s effect against subsequent gout flares and high-sensitivity C-reactive protein levels: a causal mediation analysis. Arthritis Care Res. 2021;75(4):817–24.
Yoshida K, et al. Roles of postdiagnosis accumulation of morbidities and lifestyle changes in excess total and cause-specific mortality risk in rheumatoid arthritis. Arthritis Care Res (Hoboken). 2021;73(2):188–98.
Turner EL, Dobson JE, Pocock SJ. Categorisation of continuous risk factors in epidemiological publications: a survey of current practice. Epidemiol Perspect Innov. 2010;7(1):9.
Mabikwa OV, et al. Assessing the reporting of categorised quantitative variables in observational epidemiological studies. BMC Health Serv Res. 2017;17(1):201.
Royston P, Altman DG, Sauerbrei W. Dichotomizing continuous predictors in multiple regression: a bad idea. Stat Med. 2006;25(1):127–41.
Altman DG. Categorizing continuous variables. Wiley StatsRef: Statistics Reference Online; 2014.
Naggara O, et al. Analysis by categorizing or dichotomizing continuous variables is inadvisable: an example from the natural history of unruptured aneurysms. Am J Neuroradiol. 2011;32(3):437–40.
Altman DG. Categorizing continuous variables. Encycl Bciostat. 1998;1:563–7.
Altman DG, Royston P. The cost of dichotomising continuous variables. BMJ. 2006;332(7549):1080.
Senn S. Disappointing dichotomies. Pharm Stat. 2003;2(4):239–40.
Owen SV, Froman RD. Why carve up your continuous data? Res Nurs Health. 2005;28(6):496–503.
Schmitz S, Adams R, Walsh C. The use of continuous data versus binary data in MTC models: a case study in rheumatoid arthritis. BMC Med Res Methodol. 2012;12:167.
Felson DT. Whither the ACR20? J Rheumatol. 2004;31(5):835–7.
Snapinn SM, Jiang Q. Responder analyses and the assessment of a clinically relevant treatment effect. Trials. 2007;8:31.
Uryniak T, et al. Responder analyses—a PhRMA position paper. Stat Biopharm Res. 2011;3(3):476–87.
Siegel JN, Zhen BG. Use of the American College of Rheumatology N (ACR-N) index of improvement in rheumatoid arthritis: argument in favor. Arthritis Rheum. 2005;52(6):1637–41.
Anderson JJ, Bolognese JA, Felson DT. Comparison of rheumatoid arthritis clinical trial outcome measures: a simulation study. Arthritis Rheum. 2003;48(11):3031–8.
Wason J, McMenamin M, Dodd S. Analysis of responder-based endpoints: improving power through utilising continuous components. Trials. 2020;21(1):427.
Nguyen TV. Common methodological issues and suggested solutions in bone research. Osteoporos Sarcopenia. 2020;6(4):161–7.
Faraggi D, Simon R. A simulation study of cross-validation for selecting an optimal cutpoint in univariate survival analysis. Stat Med. 1996;15(20):2203–13.
Taylor JMG, Yu M. Bias and efficiency loss due to categorizing an explanatory variable. J Multivar Anal. 2002;83(1):248–63.
Becher H. The concept of residual confounding in regression models and some applications. Stat Med. 1992;11(13):1747–58.
Austin PC, Brunner LJ. Inflation of the type I error rate when a continuous confounding variable is categorized in logistic regression analyses. Stat Med. 2004;23(7):1159–78.
Gupta R, et al. Understanding the effect of categorization of a continuous predictor with application to neuro-oncology. Neuro-Oncol Pract. 2021;9(2):87–90.
O’Gorman TW, Woolson RF. The effect of category choice on the odds ratio and several measures of association in case-control studies. Commun Stat Theory Methods. 1993;22(4):1157–71.
Osteoarthritis Initiative (OAI) study protocol. https://nda.nih.gov/static/docs/StudyDesignProtocolAndAppendices.pdf. Accessed 8 Feb 2023.
Serdar CC, et al. Sample size, power and effect size revisited: simplified and practical approaches in pre-clinical, clinical and laboratory studies. Biochem Med (Zagreb). 2021;31(1):010502.
Columb M, Atkinson M. Statistical analysis: sample size and power estimations. BJA Educ. 2015;16(5):159–61.
Mackie GM, Samocha-Bonet D, Tam CS. Does weight cycling promote obesity and metabolic risk factors? Obes Res Clin Pract. 2017;11(2):131–9.
Joseph GB, et al. Weight cycling and knee joint degeneration in individuals with overweight or obesity: four-year magnetic resonance imaging data from the Osteoarthritis Initiative. Obesity (Silver Spring). 2021;29(5):909–18.
Ballinger GA. Using generalized estimating equations for longitudinal data analysis. Organ Res Methods. 2004;7(2):127–50.
Zhang S, Su P, Liu S. Fusion of cognitive information: evaluation and evolution method of product image form. Comput Intell Neurosci. 2021;2021:5524093.
Salis Z, Sainsbury A. Association of change in BMI with incidence and progression of the structural defects of hip osteoarthritis: data from the OAI and CHECK studies. Arthritis Care Res. 2022. Vol. 0, No. 0, Month 2023, pp 1–11
Lane NE, et al. Progression of radiographic hip osteoarthritis over eight years in a community sample of elderly white women. Arthritis Rheum. 2004;50(5):1477–86.
Bliddal H, Leeds AR, Christensen R. Osteoarthritis, obesity and weight loss: evidence, hypotheses and horizons - a scoping review. Obes Rev. 2014;15(7):578–86.
Raud B, et al. Level of obesity is directly associated with the clinical and functional consequences of knee osteoarthritis. Sci Rep. 2020;10(1):3601.
Salis Z, et al. Weight loss is associated with reduced risk of knee and hip replacement: a survival analysis using Osteoarthritis Initiative data. Int J Obes. 2022;46(4):874–84.
Felson DT, et al. Risk factors for incident radiographic knee osteoarthritis in the elderly: the Framingham Study. Arthritis Rheum. 1997;40(4):728–33.
Riddle DL, Stratford PW. Body weight changes and corresponding changes in pain and function in persons with symptomatic knee osteoarthritis: a cohort study. Arthritis Care Res (Hoboken). 2013;65(1):15–22.
Salis Z, et al. Decrease in body mass index is associated with reduced incidence and progression of the structural defects of knee osteoarthritis: a prospective multi-cohort study. Arthritis Rheumatol. 2023;75(4):533–43.
Acknowledgements
We acknowledge the provision of datasets from the Osteoarthritis Initiative (OAI) study.
The OAI is a collaborative informatics system created by the National Institute of Mental Health and the National Institute of Arthritis, Musculoskeletal and Skin Diseases (NIAMS) to provide a worldwide resource to quicken the pace of biomarker identification, scientific investigation, and osteoarthritis drug development. The OAI is a public-private partnership comprised of five contracts (N01-AR-2-2258; N01-AR-2-2259; N01-AR-2-2260; N01-AR-2-2261; N01-AR-2-2262) funded by the National Institutes of Health, a branch of the Department of Health and Human Services, and conducted by the OAI Study Investigators. Private funding partners include Merck Research Laboratories; Novartis Pharmaceuticals Corporation; GlaxoSmithKline; and Pfizer, Inc. Private sector funding for the OAI is managed by the Foundation for the National Institutes of Health. This manuscript was prepared using OAI public-use data sets and does not necessarily reflect the opinions or views of the OAI investigators, the NIH, or the private funding partners. The OAI data repository is housed within the National Institute of Mental Health (NIMH) Data Archive (NDA).
Funding
ZS is the recipient of an Australian Government Research Training Program Scholarship. Additionally, the National Health and Medical Research Council of Australia supported this work via Senior Research Fellowship 1135897 to AS. The funder had no role in data collection, data analysis, data interpretation, or writing of the manuscript.
Author information
Authors and Affiliations
Contributions
ZS and AS conceptualized and designed the study. ZS performed the statistical analysis. ZS wrote the first draft of the manuscript. BG and AS contributed to all parts of the manuscript. All authors read and approved the final manuscript.
Corresponding author
Ethics declarations
Ethics approval and consent to participate
We used data from the Osteoarthritis Initiative (OAI) study. OAI data is openly available to researchers for scientific and educational purposes. The local institutional review boards of the OAI study centers obtained informed consent from all subjects and/or their legal guardian(s). The local institutional review boards of the OAI study centers reviewed and approved the ethics approval and any experimental protocols. The methods were carried out in accordance with relevant guidelines and regulations.
Consent for publication
Not applicable.
Competing interests
ZS and AS own 50% of each of the shares in Zuman International Pty Ltd, which receives royalties and other payments for educational resources and services in adult weight management and research methodology. AS additionally reports receiving presentation fees and travel reimbursements from Eli Lilly and Co, the Pharmacy Guild of Australia, Novo Nordisk, the Dietitians Association of Australia, Shoalhaven Family Medical Centres, the Pharmaceutical Society of Australia, and Metagenics, and serving on the Nestlé Health Science Optifast VLCD advisory board from 2016 to 2018. BG has no competing interest.
Additional information
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary Information
Additional file 1: Definitions of outcomes. Table S1.
Incidence of outcomes in main cohort (the cohort for investigating the progression of knee and hip osteoarthritis and the incidence of total joint replacement). Table S2. Incidence of outcomes in frequent knee pain cohort. Table S3. Incidence of outcomes in frequent hip pain cohort. Table S4. Incidence of outcomes in any knee pain cohort. Table S5. Incidence of outcomes in any hip pain cohort. Table S6. The associations of change in BMI change with outcomes, as treating change in BMI categorical (using 3, 5 and 10% weight change categories) and continuous variable.
Rights and permissions
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
About this article
Cite this article
Salis, Z., Gallego, B. & Sainsbury, A. Researchers in rheumatology should avoid categorization of continuous predictor variables. BMC Med Res Methodol 23, 104 (2023). https://doi.org/10.1186/s12874-023-01926-4
Received:
Accepted:
Published:
DOI: https://doi.org/10.1186/s12874-023-01926-4
Keywords
- Categorization 1
- Dichotomization 2
- Predictor Variable 3
- Covariate 4