Accounting for comorbidity in etiological research

Accounting for comorbidity in etiological research | medRxiv /* */ /* */ <!-- <!-- /*! * yepnope1.5.4 * (c) WTFPL, GPLv2 */ (function(a,b,c){function d(a){return"[object Function]"==o.call(a)}function e(a){return"string"==typeof a}function f(){}function g(a){return!a||"loaded"==a||"complete"==a||"uninitialized"==a}function h(){var a=p.shift();q=1,a?a.t?m(function(){("c"==a.t?B.injectCss:B.injectJs)(a.s,0,a.a,a.x,a.e,1)},0):(a(),h()):q=0}function i(a,c,d,e,f,i,j){function k(b){if(!o&&g(l.readyState)&&(u.r=o=1,!q&&h(),l.onload=l.onreadystatechange=null,b)){"img"!=a&&m(function(){t.removeChild(l)},50);for(var d in y[c])y[c].hasOwnProperty(d)&&y[c][d].onload()}}var j=j||B.errorTimeout,l=b.createElement(a),o=0,r=0,u={t:d,s:c,e:f,a:i,x:j};1===y[c]&&(r=1,y[c]=[]),"object"==a?l.data=c:(l.src=c,l.type=a),l.width=l.height="0",l.onerror=l.onload=l.onreadystatechange=function(){k.call(this,r)},p.splice(e,0,u),"img"!=a&&(r||2===y[c]?(t.insertBefore(l,s?null:n),m(k,j)):y[c].push(l))}function j(a,b,c,d,f){return q=0,b=b||"j",e(a)?i("c"==b?v:u,a,b,this.i++,c,d,f):(p.splice(this.i++,0,a),1==p.length&&h()),this}function k(){var a=B;return a.loader={load:j,i:0},a}var l=b.documentElement,m=a.setTimeout,n=b.getElementsByTagName("script")[0],o={}.toString,p=[],q=0,r="MozAppearance"in l.style,s=r&&!!b.createRange().compareNode,t=s?l:n.parentNode,l=a.opera&&"[object Opera]"==o.call(a.opera),l=!!b.attachEvent&&!l,u=r?"object":l?"script":"img",v=l?"script":u,w=Array.isArray||function(a){return"[object Array]"==o.call(a)},x=[],y={},z={timeout:function(a,b){return b.length&&(a.timeout=b[0]),a}},A,B;B=function(a){function b(a){var a=a.split("!"),b=x.length,c=a.pop(),d=a.length,c={url:c,origUrl:c,prefixes:a},e,f,g;for(f=0;f<d;f++)g=a[f].split("="),(e=z[g.shift()])&&(c=e(c,g));for(f=0;f<b;f++)c=x[f](c);return c}function g(a,e,f,g,h){var i=b(a),j=i.autoCallback;i.url.split(".").pop().split("?").shift(),i.bypass||(e&&(e=d(e)?e:e[a]||e[g]||e[a.split("/").pop().split("?")[0]]),i.instead?i.instead(a,e,f,g,h):(y[i.url]?i.noexec=!0:y[i.url]=1,f.load(i.url,i.forceCSS||!i.forceJS&&"css"==i.url.split(".").pop().split("?").shift()?"c":c,i.noexec,i.attrs,i.timeout),(d(e)||d(j))&&f.load(function(){k(),e&&e(i.origUrl,h,g),j&&j(i.origUrl,h,g),y[i.url]=2})))}function h(a,b){function c(a,c){if(a){if(e(a))c||(j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}),g(a,j,b,0,h);else if(Object(a)===a)for(n in m=function(){var b=0,c;for(c in a)a.hasOwnProperty(c)&&b++;return b}(),a)a.hasOwnProperty(n)&&(!c&&!--m&&(d(j)?j=function(){var a=[].slice.call(arguments);k.apply(this,a),l()}:j[n]=function(a){return function(){var b=[].slice.call(arguments);a&&a.apply(this,b),l()}}(k[n])),g(a[n],j,b,n,h))}else!c&&l()}var h=!!a.test,i=a.load||a.both,j=a.callback||f,k=j,l=a.complete||f,m,n;c(h?a.yep:a.nope,!!i),i&&c(i)}var i,j,l=this.yepnope.loader;if(e(a))g(a,0,l,0);else if(w(a))for(i=0;i (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0];var j=d.createElement(s);var dl=l!='dataLayer'?'&l='+l:'';j.src='//www.googletagmanager.com/gtm.js?id='+i+dl;j.type='text/javascript';j.async=true;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-P4HH5NV'); Skip to main content Home About Submit ALERTS / RSS Search for this keyword Advanced Search Accounting for comorbidity in etiological research View ORCID Profile Vahe Khachadourian , View ORCID Profile Magdalena Janecka doi: https://doi.org/10.1101/2025.01.19.25320775 Vahe Khachadourian 1 Department of Child and Adolescent Psychiatry, NYU Grossman School of Medicine , New York, NY, US Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Vahe Khachadourian For correspondence: vahe.khachadourian{at}gmail.com Magdalena Janecka 1 Department of Child and Adolescent Psychiatry, NYU Grossman School of Medicine , New York, NY, US 2 Department of Population Health, NYU Grossman School of Medicine , New York, NY, US Find this author on Google Scholar Find this author on PubMed Search for this author on this site ORCID record for Magdalena Janecka Abstract Full Text Info/History Metrics Data/Code Preview PDF ABSTRACT Introduction Despite the theoretical advancements and recommendations regarding covariate adjustment in causal inference, clinical studies often fail to explicitly state the underlying assumptions related to causal structure among the study variables. Specifically, despite the pervasive nature of comorbidity, explicit causal assumptions about the role of comorbidity in exposure-outcome relationships are often lacking, potentially leading to inappropriate accounting for comorbid conditions and resulting in biased effect estimates. This study aims to explore common causal structures involving comorbidity and provide guidance for handling it in etiologic research. Methods We use Directed Acyclic Graphs (DAGs) to depict six causal scenarios involving comorbidity as a confounder, mediator, collider, or consequence of the exposure or outcome. Simulations were conducted across 5,000 iterations for each scenario, assessing the impact of conditioning on comorbidity under three effect measures (mean difference, odds ratio, risk ratio). Bias was evaluated by comparing adjusted and unadjusted effect estimates to the true values. Results The impact of conditioning on comorbidity varied by its causal role. Adjusting for comorbidity mitigated bias when it acted as a confounder, but introduced bias when it was a mediator or collider. In instances where comorbidity was a consequence of either the exposure or outcome, the decision to adjust depended on the research objectives. Nonlinear models revealed differences in marginal and conditional effects due to non-collapsibility. Discussion Explicit causal assumptions are essential for selecting appropriate analytical strategies in etiologic research. This study provides practical guidance on handling comorbidity-related challenges, highlighting the need for study design and analysis to align with research objectives. Future work should address more complex causal structures and other methodological challenges. INTRODUCTION Over the past decades, etiologic research on human disease has advanced significantly due to improvements in causal inference methodology, availability of health records – including both registry-based studies and real-world data – and establishment of international consortia allowing to gauge the generalizability of results. Throughout, a considerable focus in observational studies has been placed on reducing confounding and other biases 1 . Although numerous publications 2 – 5 have now addressed various aspects of “adjustment” in design or analyses, many of these works are either theory-based or focused on specific causal structures 3 , 4 , limiting their wider implementation in applied research. Specifically, the issue of comorbidity, its various causal relationships with the outcome of interest, and downstream analytical recommendations have not been widely discussed. Therefore, addressing the issue of comorbidity in etiological studies, integrating clinical perspective and focusing on clinical concepts, could help bridge the gap between methodological knowledge and practice. Comorbidity, first defined by Feinstein (1970) 6 , is the presence (existence or occurrence) of another distinct health condition in the individual with the index disease. Although there is variation in the definition and interpretation of comorbidity 7 , 8 , studies consistently show that comorbidity is a common phenomenon 9 . Given the pervasive nature of comorbidity in many disorders 10 , etiologic studies often consider comorbidity in their design or analyses 11 – 16 . The motivations for such considerations are different and could include understanding of the risk factors for the disease of interest “independent” of its comorbidity 17 , control of confounding 18 , minimizing misclassification 19 , or study effect measure modification by comorbidity 20 . Despite these various contexts for considering comorbidity in etiologic research, the underlying motivation is most often to reduce bias or obtain a valid estimate for the parameters of interest. Given that there is no universally assumed (or universally true) causal structure between comorbidity 8 and other variables in a study, a one-size-fits-all solution for accounting for comorbidity in etiologic research does not exist. For instance, comorbidity could be a risk factor for the index disease, its consequence, or share a common cause with it. Depending on the causal structure at hand, adjusting for comorbidity can change the effect estimate of the exposure on the outcome, sometimes resulting in attenuation, amplification, or even reversal of the estimated association relative to the comorbidity unadjusted associations 2 – 4 , 21 . Understanding the scenarios under which comorbidity adjustment is made and its impact on the main effect of exposure on the index diagnosis is therefore critical. As the definition of comorbidity does not imply any explicit or implicit causal relation or even temporal order between the index disease and the co-occurring comorbid conditions, it is important to be clear about the assumed structure(s) – which often is critical in informing the correct design or analytical strategy. However, a review of the literature reveals that the handling of comorbidity in the design and analysis of etiological studies often lacks explicit assumptions about their relationship with other variables, or may in fact not align with the study objective 11 , 16 . In this paper, we therefore explore several common causal structures involving comorbidity and other study variables. For each scenario, we discuss the impact of adjusting for comorbidity on the effect estimates and potential considerations. We hope this piece can provide guidance for accounting for comorbidity (and potentially other covariates) in epidemiological research and improve causal inference. Controlling for comorbidity extends beyond adjustment in the analytical model and can occur by restricting the sample based on comorbidity, stratified sampling or analysis based on comorbidity, or handling differential loss to follow-up or non-random missingness patterns. Here forth, for ease of read we use conditioning and adjusting interchangeably, and unless otherwise specified, the arguments and statements about adjustment are applicable to all other conditioning approaches. METHODS Possible Structure of Associations Explicitly stating the causal assumptions is critical for the choice of the study design or analytical methods, and clear communication and justification of these decisions. Causal diagrams, specifically Directed Acyclic Graphs (DAGs) 22 are a common tool in causal inference for depicting assumptions and assessing bias. Concise and accessible description of the causal diagrams are provided by Greenland et. al (1999) 23 . Here we use DAGs to depict several possible common scenarios underpinning the associations between comorbid conditions and the exposure and outcome in study. We use X to denote the key exposure, Y for the index condition, and C for the comorbidity. Comorbidity as a common cause of X and Y Comorbidity as a mediator of the effects of X on Y Comorbidity as a collider in the association between X and Y Comorbidity as a consequence of X, but not a risk factor for Y Comorbidity as a risk factor of Y, but independent of X Comorbidity as a consequence of Y, but not directly associated with X Effect measures The impact of adjusting for comorbidity on the estimate of the total effect depends on the effect measures. To extend the applications of our work, for each of the above scenarios presented, we will assess and discuss the effect of conditioning on comorbidity for three different effect measures; 1) mean difference (least squares regression), 2) odds ratio (logistic regression), and 3) risk ratio (log binomial regression). Simulation setting We conducted a simulation study to assess the effect of conditioning on covariate C in each of the causal structures presented in Figure 1 . The parameters for the simulation analyses, along with the analytical code, are presented in the supplemental material. Briefly, for each of the causal structures presented in Figure 1 , we made 5,000 iterations of simulated studies with sample size 1000. For simplicity, we assumed X and C are both binary variables. For each causal diagram we simulated 3 sets of outcomes (Y): two binary endpoints (e.g., hip fracture) corresponding to the total effect of X on Y on the RR and OR scales to be 2; and one normally distributed, continuous endpoint (e.g., serum cholesterol level) for which the total effect of X on Y measured by the mean difference was equal to 2. Download figure Open in new tab Figure 1. Depicting causal relation between exposure (X), index disease (Y), and comorbidity (C) in selected common scenarios using Directed Acyclic Graphs The simulation for the causal structure depicted in Figure 1 panel 1 included the following steps: Simulate binary covariate, C, with a probability of pc. Simulate a binary exposure variable, X (0, 1), where the probability of X, px, was assumed to be a function of C, P(X = 1|C). Simulate a response Y, where Y was assumed to be a function of X and C. ∘ The response for the log binomial model estimating RR was simulated by: P(Y = 1|X, C) = exp(log(0.2) + log(2)*x + log(2)*c) ∘ The response for the logistic regression model estimating OR was simulated by: P(Y = 1|X, C) = 1 / (1 + exp(-(log(0.4) + log(2)*x + log(2)*c))) ∘ The response for the linear regression mode estimating mean difference was simulated by E(Y|X, C) = N~(2, 2) + N~(2*x, 2) + N~(2*c, 2) Run the following models: ∘ logistic regression model estimating effect of X on Y, with and without adjusting for C. Exponentiate coefficient of X to obtain OR. ∘ log binomial regression model estimating effect of X on Y, with and without adjusting for C. Exponentiate coefficient of X to obtain RR. ∘ least square regression model estimating effect of X on Y, with and without adjusting for C. Repeat steps 1-4, 5,000 times. The point estimate for each effect measure was the median of the estimates from the 5,000 bootstraps and the lower and upper limits of the 95%CI corresponded to the 2.5 th and 97.5 th percentile of these 5,000 estimates. Given the true effect estimate for OR, RR, and least square mean difference was 2, the coefficient of x from each model was evaluated based on the relative bias (B1 −2)/2. RESULTS Anticipated effect of conditioning on comorbidity and simulation results Table 1 presents simulation results. For each of these causal diagrams the crude and C adjusted estimated association between X and Y along with their bootstrap based 95%CI intervals were summarized on for OR, RR, and mean difference. The relative bias of the point estimate was reported as percentage. View this table: View inline View popup Download powerpoint Table 1. Association between exposure (X) and index disease (Y) in crude and comorbidity (C) adjusted models for different causal scenarios DAG 1: Comorbidity as a common cause of X and Y Real-world example: Obesity (C) in a study of the effects of type 2 diabetes mellitus (X) on ischemic heart disease (Y). In this scenario, comorbidity is the common cause of exposure and outcome, fulfilling the definition of a confounder. Using DAGs, we observe the open back door path from X to C to Y, resulting in bias. Conditioning on comorbidity is warranted and failure to condition on comorbidity in this case will result in confounding of the effect estimate of X on Y. The crude estimate was biased for all three models, and the adjusted estimates remained unbiased 21 Recommended strategy: adjusting for comorbidity. DAG 2: Comorbidity as a mediator of the effects of X on Y Real-world example: hyperlipidemia (C) in a study of the effects of sedentary lifestyle (X) on ischemic heart disease (Y). In this case, the path from X to C to Y is causal. C is a mediator on the path reflecting the indirect effect of X to Y. Therefore, conditioning on C resulted in bias in the estimate of the total effect of X on Y whereas the unconditional effect remained unbiased. Recommended strategy: not adjusting for comorbidity. DAG 3: Comorbidity as a collider Real-world example: injury (C) in a study of the effects of alcohol use (X) on depression (Y) When the comorbidity is caused by both exposure and index disease, the path connecting exposure and index disease through comorbidity is closed and the comorbidity serves as a collider on that path. Here, the crude estimate provided unbiased effect estimates, and adjusting for C resulted in bias (collider stratification bias) in the estimates 2 , 4 . This bias is present regardless of the analytical model used, and its magnitude depends on the strength of the association between X and C; and Y and C. Recommended strategy: not adjusting for comorbidity. DAG 4: Comorbidity as a consequence of X, but not a risk factor for Y Real world example: liver failure (C) in a study of the effects of alcohol use (X) on dementia (Y). In this case either conditioning on C or not will not bias the estimate of the effect of E on Y. However, even though both approaches will result in unbiased estimates, the estimated effects have a different causal interpretation and might be different. Scenario 4, both crude and C- adjusted models provided unbiased estimates of OR, RR and LSMD 3 . Recommended strategy: both adjusting or not adjusting will not result in bias. DAG 5: Comorbidity as a risk factor of Y, but independent of X Real-world example: alcohol abuse (C) in a study of the effects of air pollution (X) on hepatic cancer (Y). When comorbidity is a risk factor for the disease but is not associated with the exposure, adjusting for comorbidity does not introduce bias. RR and LSMD are unbiased in both crude and adjusted models. For the OR, the results for the adjusted and crude models are slightly different, nevertheless none are biased (assuming model specifications are correct) 3 . However, because of the non-collapsibility of nonlinear models (e.g. logistic regression), the conditional and marginal effects of X on Y are different, and the difference in the estimates between the crude and adjusted models reflects the difference in the underlying effects being estimated, and not necessarily possible bias 24 – 26 . The choice of the model should be guided by clear definition of the underlying effect of interest. Recommended strategy: both adjusting or not adjusting for comorbidity will not result in bias. DAG 6: Comorbidity as a consequence of Y, but not directly associated with X Real world example: opportunistic infection (C) in a study of the effects of syringe service program (X) on risk of HIV infection (Y). When the disease is a risk factor for comorbidity, the crude estimates are unbiased. However, under the assumption that X has an effect on Y, conditioning on comorbidity in the linear models (e.g., RR and LSMD) leads to biased estimates while the OR from the adjusted model remained unbiased 3 . None of the models will be biased by adjusting for C if the effect of X on Y is null. Recommended strategy: both adjusting or not adjusting for comorbidity will not result in bias in OR. Not adjusting in log binomial and least squares regression models. DISCUSSION This study provides a detailed exploration of the potential roles of comorbidity in the association between exposure and index disease, highlighting the diverse causal relationships that can exist between these variables. Accounting for the valid causal structure has extensive implications for causal inference in etiologic research. By systematically examining common causal structures and their effects on bias, this work addresses a critical but underexplored issue: the appropriate handling of comorbidity in causal research. Our findings confirm that the role of comorbidity in the exposure-outcome relationship determines the impact of conditioning. When comorbidity act as a confounder, conditioning effectively mitigate confounding bias across all the effect measures (e.g., odds ratio [OR], relative risk [RR], and least-squares mean difference [LSMD]). Conversely, when comorbidity serves as a mediator or collider, conditioning introduces bias into the total effect estimate of the exposure on the outcome. This highlights the critical need to avoid inappropriate adjustments. When comorbidity is either a cause or consequence of the exposure but has no direct connection to the outcome (except through the exposure), adjusting for comorbidity does not affect the estimated effect. However, if comorbidity is a consequence of the outcome, adjustment introduces bias in log-binomial and linear regression models, whereas such adjustment does not bias estimates in logistic regression models. Furthermore, adjusting for a risk factor of the outcome does not result in bias. However, in nonlinear models like logistic regression, the distinction between marginal and conditional effects due to the non- collapsibility of effect measures emphasizes the importance of aligning analytical strategies with study objectives, which extends beyond merely minimizing bias. In such cases marginal effects provide better insight about the effect of the exposure on the outcome on the population level, whereas, the conditional effect could be more relevant for inference on an individual level. Although our results demonstrate that inappropriate handling of comorbidity in the analyses could lead to bias, it is important to note that the direction and magnitude of such bias depend on the distribution and prevalence of the exposure, outcome, comorbidity, and the nature of the associations among these variables. Furthermore, while our examples focused on a single comorbidity at a time, in practice, multiple comorbid conditions may be associated with the index disease. The relation of these comorbidities with the study exposure and the index disease could be different, and each should be handled according to underlying assumptions and their causal relations. Importantly, in our study, we assumed no measurement error, no selection bias, and no model misspecification. In real-life setting, and in presence of one or more of these issues, the impact of adjusting for comorbidity on the overall bias might be different than the impact of such adjustment in isolation and absence of any other source of bias. The simulation results corroborate these theoretical insights, illustrating how the causal role of comorbidity influences bias under various analytical strategies, emphasizing the importance of explicit causal assumptions and appropriate model selection to ensure valid inferences. By examining several common causal structures between exposure, index disease and its comorbidity, this study provides practical guidance for refining study design and analytical methods. While no universal approach exists for handling comorbidity, integrating methodological rigor with clinical insight – essential to decide which of the causal structures is most applicable in the study setting - enhances the validity and applicability of etiologic research. Data Availability All data produced in the present study are available upon reasonable request to the authors. References 1. ↵ Grimes DA , Schulz KF . Bias and causal associations in observational research . The Lancet . 2002 ; 359 ( 9302 ): 248 – 252 . doi: 10.1016/S0140-6736(02)07451-2 OpenUrl CrossRef PubMed Web of Science 2. ↵ Digitale JC , Martin JN , Glidden D V. , Glymour MM . Key concepts in clinical epidemiology: collider-conditioning bias . J Clin Epidemiol . 2023 ; 161 : 152 – 156 . doi: 10.1016/j.jclinepi.2023.07.004 OpenUrl CrossRef PubMed 3. ↵ Schisterman EF , Cole SR , Platt RW . Overadjustment bias and unnecessary adjustment in epidemiologic studies . Epidemiology . 2009 ; 20 ( 4 ): 488 – 495 . doi: 10.1097/EDE.0b013e3181a819a1 OpenUrl CrossRef PubMed Web of Science 4. ↵ Holmberg MJ , Andersen LW . Collider Bias . JAMA . 2022 ; 327 ( 13 ): 1282 . doi: 10.1001/jama.2022.1820 OpenUrl CrossRef PubMed 5. ↵ Etminan M , Brophy JM , Collins G , Nazemipour M , Mansournia MA . To Adjust or Not to Adjust: The Role of Different Covariates in Cardiovascular Observational Studies . Am Heart J . 2021 ; 237 : 62 – 67 . doi: 10.1016/j.ahj.2021.03.008 OpenUrl CrossRef PubMed 6. ↵ Feinstein AR . The pre-therapeutic classification of co-morbidity in chronic disease . J Chronic Dis . 1970 ; 23 ( 7 ): 455 – 468 . doi: 10.1016/0021-9681(70)90054-8 OpenUrl CrossRef PubMed Web of Science 7. ↵ van den Akker M , Buntinx F , Knottnerus JA . Comorbidity or multimorbidity . European Journal of General Practice . 1996 ; 2 ( 2 ): 65 – 70 . doi: 10.3109/13814789609162146 OpenUrl CrossRef 8. ↵ Valderas JM , Starfield B , Sibbald B , Salisbury C , Roland M. Defining Comorbidity: Implications for Understanding Health and Health Services . The Annals of Family Medicine . 2009 ; 7 ( 4 ): 357 – 363 . doi: 10.1370/afm.983 OpenUrl Abstract / FREE Full Text 9. ↵ Starfield B. Threads and Yarns: Weaving the Tapestry of Comorbidity . The Annals of Family Medicine . 2006 ; 4 ( 2 ): 101 – 103 . doi: 10.1370/afm.524 OpenUrl FREE Full Text 10. ↵ Khachadourian V , Mahjani B , Sandin S , et al. Comorbidities in autism spectrum disorder and their etiologies . Transl Psychiatry . 2023 ; 13 ( 1 ): 71 . doi: 10.1038/s41398-023-02374-w OpenUrl CrossRef 11. ↵ Kanina A , Larsson H , Sjölander A , et al. Association between cumulative psychosocial adversity in the family and ADHD and autism: a family-based cohort study . Transl Psychiatry . 2023 ; 13 ( 1 ): 282 . doi: 10.1038/s41398-023-02571-7 OpenUrl CrossRef PubMed 12. Zhuang S , Na M , Winkelman JW , et al. Association of Restless Legs Syndrome With Risk of Suicide and Self-harm . JAMA Netw Open . 2019 ; 2 ( 8 ): e199966 . doi: 10.1001/jamanetworkopen.2019.9966 OpenUrl CrossRef 13. Alonso J , de Jonge P , Lim CCW , et al. Association between mental disorders and subsequent adult onset asthma . J Psychiatr Res . 2014 ; 59 : 179 – 188 . doi: 10.1016/j.jpsychires.2014.09.007 OpenUrl CrossRef PubMed 14. Borrelli A , Spencer K , Birch R. Adjusting for comorbidity in observational cancer studies: A systematic review to assess alignment between index and study . Int J Oncol . 2022 ; 62 ( 1 ): 9 . doi: 10.3892/ijo.2022.5457 OpenUrl CrossRef PubMed 15. Federico A , Zgodic A , Flory K , et al. Predictors of Autism Spectrum Disorder and ADHD: Results from the National Survey of Children’s Health . Disabil Health J . 2024 ; 17 ( 1 ): 101512 . doi: 10.1016/j.dhjo.2023.101512 OpenUrl CrossRef PubMed 16. ↵ Song IG , Kim HS , Cho YM , et al. Association between birth weight and neurodevelopmental disorders assessed using the Korean National Health Insurance Service claims data . Sci Rep . 2022 ; 12 ( 1 ): 2080 . doi: 10.1038/s41598-022-06094-x OpenUrl CrossRef PubMed 17. ↵ Navarro-Cano G , del Rincón I , Pogosian S , Roldán JF , Escalante A. Association of mortality with disease severity in rheumatoid arthritis, independent of comorbidity . Arthritis Rheum . 2003 ; 48 ( 9 ): 2425 – 2433 . doi: 10.1002/art.11127 OpenUrl CrossRef PubMed Web of Science 18. ↵ Schneeweiss S , Maclure M. Use of comorbidity scores for control of confounding in studies using administrative databases . Int J Epidemiol . 2000 ; 29 ( 5 ): 891 – 898 . doi: 10.1093/ije/29.5.891 OpenUrl CrossRef PubMed Web of Science 19. ↵ Bossini L , Casolaro I , Caterini C , et al. Diagnostic mistakes in post-traumatic stress disorder . The problem of symptom overlap with depression. Psychiatr Pol . 2013 ; 47 ( 6 ): 1051 – 1063 . OpenUrl PubMed 20. ↵ Liu YC , Liao YT , Wen MH , Chen VCH , Chen YL . The Association between Autism Spectrum Disorder and Precocious Puberty: Considering Effect Modification by Sex and Neuropsychiatric Comorbidities . J Pers Med . 2024 ; 14 ( 6 ): 632 . doi: 10.3390/jpm14060632 OpenUrl CrossRef PubMed 21. ↵ Arah OA . Bias Analysis for Uncontrolled Confounding in the Health Sciences . Annu Rev Public Health . 2017 ; 38 ( 1 ): 23 – 38 . doi: 10.1146/annurev-publhealth-032315-021644 OpenUrl CrossRef PubMed 22. ↵ Pearl J. Causal diagrams for empirical research . Biometrika . 1995 ; 82 ( 4 ): 669 – 688 . doi: 10.1093/biomet/82.4.669 OpenUrl CrossRef Web of Science 23. ↵ Greenland S , Pearl J , Robins JM . Causal diagrams for epidemiologic research . Epidemiology . 1999 ; 10 ( 1 ): 37 – 48 . OpenUrl CrossRef PubMed Web of Science 24. ↵ Steingrimsson JA , Hanley DF , Rosenblum M. Improving precision by adjusting for prognostic baseline variables in randomized trials with binary outcomes, without regression model assumptions . Contemp Clin Trials . 2017 ; 54 : 18 – 24 . doi: 10.1016/j.cct.2016.12.026 OpenUrl CrossRef PubMed 25. Greenland S. Noncollapsibility, confounding, and sparse-data bias . Part 1: The oddities of odds. J Clin Epidemiol . 2021 ; 138 : 178 – 181 . doi: 10.1016/j.jclinepi.2021.06.007 OpenUrl CrossRef PubMed 26. ↵ Greenland S. Noncollapsibility, confounding, and sparse-data bias. Part 2: What should researchers make of persistent controversies about the odds ratio? J Clin Epidemiol . 2021 ; 139 : 264 – 268 . doi: 10.1016/j.jclinepi.2021.06.004 OpenUrl CrossRef PubMed View the discussion thread. Back to top Previous Next Posted January 21, 2025. Download PDF Data/Code Email Thank you for your interest in spreading the word about medRxiv. NOTE: Your email address is requested solely to identify you as the sender of this article. Your Email * Your Name * Send To * Enter multiple addresses on separate lines or separate them with commas. You are going to email the following Accounting for comorbidity in etiological research Message Subject (Your Name) has forwarded a page to you from medRxiv Message Body (Your Name) thought you would like to see this page from the medRxiv website. Your Personal Message CAPTCHA This question is for testing whether or not you are a human visitor and to prevent automated spam submissions. Share Accounting for comorbidity in etiological research Vahe Khachadourian , Magdalena Janecka medRxiv 2025.01.19.25320775; doi: https://doi.org/10.1101/2025.01.19.25320775 Share This Article: Copy Citation Tools Accounting for comorbidity in etiological research Vahe Khachadourian , Magdalena Janecka medRxiv 2025.01.19.25320775; doi: https://doi.org/10.1101/2025.01.19.25320775 Citation Manager Formats BibTeX Bookends EasyBib EndNote (tagged) EndNote 8 (xml) Medlars Mendeley Papers RefWorks Tagged Ref Manager RIS Zotero Tweet Widget Facebook Like Google Plus One Subject Area Epidemiology Subject Areas All Articles Addiction Medicine (568) Allergy and Immunology (863) Anesthesia (297) Cardiovascular Medicine (4421) Dentistry and Oral Medicine (443) Dermatology (382) Emergency Medicine (606) Endocrinology (including Diabetes Mellitus and Metabolic Disease) (1507) Epidemiology (15212) Forensic Medicine (30) Gastroenterology (1121) Genetic and Genomic Medicine (6581) Geriatric Medicine (667) Health Economics (996) Health Informatics (4520) Health Policy (1366) Health Systems and Quality Improvement (1611) Hematology (539) HIV/AIDS (1264) Infectious Diseases (except HIV/AIDS) (15906) Intensive Care and Critical Care Medicine (1103) Medical Education (620) Medical Ethics (144) Nephrology (667) Neurology (6580) Nursing (345) Nutrition (998) Obstetrics and Gynecology (1141) Occupational and Environmental Health (956) Oncology (3324) Ophthalmology (970) Orthopedics (369) Otolaryngology (420) Pain Medicine (435) Palliative Medicine (129) Pathology (663) Pediatrics (1689) Pharmacology and Therapeutics (691) Primary Care Research (710) Psychiatry and Clinical Psychology (5432) Public and Global Health (9212) Radiology and Imaging (2193) Rehabilitation Medicine and Physical Therapy (1368) Respiratory Medicine (1194) Rheumatology (593) Sexual and Reproductive Health (709) Sports Medicine (529) Surgery (709) Toxicology (99) Transplantation (288) Urology (265) (function(){function c(){var b=a.contentDocument||a.contentWindow.document;if(b){var d=b.createElement('script');d.innerHTML="window.__CF$cv$params={r:'9ff4c88bfdfbdfa9',t:'MTc3OTM3OTExNA=='};var a=document.createElement('script');a.src='/cdn-cgi/challenge-platform/scripts/jsd/main.js';document.getElementsByTagName('head')[0].appendChild(a);";b.getElementsByTagName('head')[0].appendChild(d)}}if(document.body){var a=document.createElement('iframe');a.height=1;a.width=1;a.style.position='absolute';a.style.top=0;a.style.left=0;a.style.border='none';a.style.visibility='hidden';document.body.appendChild(a);if('loading'!==document.readyState)c();else if(window.addEventListener)document.addEventListener('DOMContentLoaded',c);else{var e=document.onreadystatechange||function(){};document.onreadystatechange=function(b){e(b);'loading'!==document.readyState&&(document.onreadystatechange=e,c())}}}})();