Design and rationale of the atrial fibrillation occurring transiently with stress (AFOTS) incidence study.

BACKGROUND: Atrial fibrillation (AF) is often detected for the first time in patients hospitalized for medical illness or non-cardiovascular surgery. AF occurring transiently with stress (AFOTS) describes this manifestation of AF, which may either be the result of a non-cardiac stressor, or existing paroxysmal AF that was not previously detected. Current estimates of AFOTS incidence are imprecise: ranging from 1 to 44%, owing to the marked heterogeneity in patient populations, identification and methods used to detect AFOTS. METHODS: The prospective, two-centre epidemiological AFOTS Incidence study will enroll 250 consecutive participants without a history of AF but with at increased risk of AF (Age ≥ 65 or >50 with one risk factor for AF) admitted to intensive care units (ICUs) for medical illness or non-cardiac surgery. Upon admission, participants will wear an ECG patch monitor that will remain in place for 14 days, or until discharge from hospital. Patients' consent to participation is deferred for up to 72 h after admission. The primary endpoint is the incidence of AF lasting ≥30 s. The study is powered to detect an AF incidence of 17% ±â€¯5%. RESULTS: We conducted a vanguard feasibility study, and 55 participants have completed participation. The median duration of monitoring was seven days. AF was detected by the clinical team in 8 participants (14%; 95% Confidence Interval 7-26%). CONCLUSIONS: The AFOTS Incidence study will employ a systematic and highly sensitive protocol for detecting AFOTS in medical illness and non-cardiac surgery ICU patients. This study is feasible and will provide a reliable estimate of the true incidence of AFOTS in this population.

Development and validation of the Patient-Physician Relationship Scale among patients with irritable bowel syndrome.

BACKGROUND: An effective patient-physician relationship (PPR) is essential to the care of patients with irritable bowel syndrome (IBS). We sought to develop and validate an IBS-specific instrument to measure expectations of the PPR. METHODS: We conducted structured focus groups about PPRs with 12 patients with IBS. Qualitative analysis was used to generate a questionnaire (the Patient-Physician Relationship Scale [PPRS]), which was modified with input from content experts and usability testing. For validation, we administered it online to US adults with IBS. Participants also completed the Functional Bowel Disorder Severity Index, the Rome III Adult Functional gastrointestinal (GI) Disorder Criteria Questionnaire, and modified versions of the Communication Assessment Tool (CAT-15) and Patient-Doctor Relationship Questionnaire (PDRQ-9). We performed principal components factor analysis for the PPRS. KEY RESULTS: The PPRS contained 32 questions with responses on a 7-item Likert scale. Themes included interpersonal features, clinical care expectations, and aspects of communication. One thousand and fifty-four eligible individuals completed the survey (88% completion rate). Most participants were middle aged (mean 48 years, SD 16.3), white (90%), and female (86%). Factor analysis showed only one relevant factor, relating to quality of PPR. The final scale ranged from possible-96 to +96 (mean 62.0, SD 37.6). It correlated moderately with the CAT-15 (r=.40, P<.001) and PDRQ-9 (r=.30, P<.001), establishing concurrent validity. CONCLUSIONS & INFERENCES: We describe the development and validation of the first questionnaire for use in measuring patient expectations of the PPR, which can be used for future outcomes studies and training physicians.

Association between exposure to household cigarette smoking behavior and cigarette smoking in Hispanic adults: Findings from the Hispanic Community Health Study/Study of Latinos.

BACKGROUND: Cigarette smoking behavior is highly determined by social influences during childhood and adolescence. This phenomenon has not been fully evaluated in the Hispanic/Latino population. PURPOSE: To examine the association between exposure to household cigarette smoking behavior (HCSB) and adult cigarette smoking among a diverse Hispanic/Latino population living in four US urban centers. The effect of acculturation on cigarette smoking was also evaluated. METHODS: Data from the Hispanic Community Health Study/Study of Latinos (HCHS/SOL) (n=13,231, ages 18-74years, collected between March 2008 and June 2011) were analyzed using logistic regression. RESULTS: HCSB exposure was an independent risk factor for adult current cigarette smoking in Hispanic/Latinos (OR 1.7; 95% CI 1.4, 2.1) after controlling for relevant confounders including socio-demographic and cultural factors. Cubans and Puerto Ricans had the highest prevalence of HCSB exposure (59% and 47% respectively) and highest prevalence of current cigarette smoking (26% and 32%) compared with other Hispanic/Latino groups, (p<.01). CONCLUSIONS: Our data suggest that exposure to HCSB in Hispanics/Latinos living in the US is an independent predictor of adult cigarette smoking, and this association appears to be strongest in Cubans and Puerto Ricans.

Applicability of new supervised statistical models to assess burn injury patterns, outcomes, and their interrelationship.

The aim of the study was to investigate the possibility of using supervised statistical models to assess burn injury patterns, outcomes and their interrelationship. Using burn study data, a preliminary principal component analysis was carried out and two separate clusters were observed. Observations were split into two classes and analysed by partial least squares (PLS) regression discriminant analysis to assess possible predictors of each class. To assess predictors of total body surface area burned (TBSA), the orthogonal projections to latent structures (OPLS) model was used after PLS regression. The identified classes were later designated as high-risk burn victims and low-risk burn victims. Female gender fell into the high-risk class. Many possible predictors were found to be associated with burn injury extent, after modelling the natural logarithm of TBSA by OPLS. The fitted model explained 76% of variation in Y. It excluded up to 9% of orthogonal variation captured in two orthogonal components. This seems to be the first application of the OPLS model in public health epidemiology. The results of this study were promising regarding the use of supervised models in injury pattern analysis.

Addressing healthcare needs of people living below the poverty line: a rapid assessment of the Andhra Pradesh Health Insurance Scheme.

BACKGROUND: Families living below the poverty line in countries which do not have universal healthcare coverage are drawn into indebtedness and bankruptcy. The state of Andhra Pradesh in India established the Rajiv Aarogyasri Community Health Insurance Scheme (RACHIS) in 2007 with the aim of breaking this cycle by improving the access of below the poverty line (BPL) families to secondary and tertiary healthcare. It covered a wide range of surgical and medical treatments for serious illnesses requiring specialist healthcare resources not always available at district-level government hospitals. The impact of this scheme was evaluated by a rapid assessment, commissioned by the government of Andhra Pradesh. The aim of the assessment was to explore the contribution of the scheme to the reduction of catastrophic health expenditure among the poor and to recommend ways by which delivery of the scheme could be improved. We report the findings of this assessment. METHODS: Two types of data were used for the assessment. Patient data pertaining to 89 699 treatment requests approved by the scheme during its first 18 months were examined. Second, surveys of scheme beneficiaries and providers were undertaken in 6 randomly selected districts of Andhra Pradesh. RESULTS: This novel scheme was beginning to reach the BPL households in the state and providing access to free secondary and tertiary healthcare to seriously ill poor people. CONCLUSION: An integrated model encompassing primary, secondary and tertiary care would be of greater benefit to families below the poverty line and more cost-effective for the government. There is considerable potential for the government to build on this successful start and to strengthen equity of access and the quality of care provided by the scheme.

Secular trend in age at menarche in indigenous and nonindigenous women in Chile.

OBJECTIVES: To estimate the secular trend in age at menarche, comparing indigenous and nonindigenous women, and its relationship with socio-demographic, family and nutritional factors. METHODS: A study (historical cohorts) of 688 indigenous and nonindigenous women, divided into four birth cohorts (1960-69, 1970-79, 1980-89, and 1990-96) in an area in central southern Chile was carried out. Data and measurements were collected by health professionals using a previously validated questionnaire. Age at menarche was self-reported (recall). Adjusted differences among cohorts were estimated using a multivariate regression model. RESULTS: A secular trend (P < 0.001) in age at menarche was found in both ethnic groups, with no significant differences between them (P > 0.05). In an adjusted model, a reduction in age at menarche was estimated at 3.7 months per decade between 1960 and 1990. This trend was moderated by higher socio-economic level, smaller number of siblings, and cohabitation with a single parent during infancy. CONCLUSIONS: The trend has occurred in a steady progression over time in indigenous women, whereas in nonindigenous women, it was slow initially but has accelerated in recent years. Nonindigenous women have maintained a slightly lower age of menarche than their indigenous counterparts.

The development and validation of a new coeliac disease quality of life survey (CD-QOL).

BACKGROUND: Previous studies on coeliac disease (CD)-related quality of life (QOL) have been limited by their use of a 'generic' rather than coeliac disease-specific assessment instruments. AIM: To develop and psychometrically validate a new coeliac disease-specific instrument, the CD-QOL. METHODS: Through a series of focus groups, we elicited items from patients that related to the specific nature of their disease and its impact on their basic needs. Through expert review, cognitive debriefing with patients and pilot testing, a scale was developed, refined and administered to 387 patients on a gluten-free diet from both community-based support groups and a tertiary care referral centre. Finally, a formal validation study was conducted to assess the psychometric properties of the CD-QOL. RESULTS: The final CD-QOL has 20 items across four clinically relevant subscales (Limitations, Dysphoria, Health Concerns, and Inadequate Treatment). The CD-QOL has high internal consistency, reliability, and psychometric validation indicates both convergent and discriminate validity. CONCLUSIONS: The CD-QOL is a reliable and valid measure of coeliac disease related QOL. As a new disease-specific instrument, it is likely to be a useful tool for evaluating patients with this disorder.

Constipation does not develop following elective hysterectomy: a prospective, controlled study.

Although there have been reports that women develop constipation following hysterectomy, previous studies were either retrospective or uncontrolled. The aim of this prospective, controlled study was to assess whether constipation develops after elective hysterectomy. Women undergoing elective gynaecological surgery were compared to matched non-surgery controls at enrollment and 3 and 12 months after surgery. The subset of women who underwent elective hysterectomy was the study group for the present report. Fifty-eight of the 132 elective surgery patients underwent hysterectomy and were compared to 123 controls. There was no difference between the groups at any follow-up point in functional constipation (P = 1.0), frequency of stools (P = 0.92), stool consistency (P = 0.42), straining (P = 0.43), feeling of obstruction (P = 0.6) or need to manually evacuate stool (P = 1.0). Significantly, more hysterectomy patients without baseline pain did develop abdominal pain at 3 or 12 months than non-surgery controls (16.7% vs 3.6%, P = 0.008). We conclude that there was no significant change in bowel habit or stool characteristics in women undergoing hysterectomy even though many developed abdominal pain. This prospective, controlled study challenges existing data regarding the effect of hysterectomy on constipation.

Epidemiology of Cleistanthus collinus (oduvan) poisoning: clinical features and risk factors for mortality.

Cleistanthus collinus (local name: oduvan) poisoning is a common suicidal poisoning method used in rural southern India. There are few studies on this issue and they have small sample sizes. This study examines the epidemiology of oduvan poisoning in a large sample, highlighting socio-demographic and clinical profile, complications, and risk factors for mortality. This study is a retrospective case series of 127 oduvan intentionally poisoned patients presenting at a tertiary care teaching hospital between the years 1990-1999. Descriptive statistics, cumulative case fatality rates and time to death from ingestion of poison were calculated. Cox regression adjusting for left truncation was used to investigate the effects of covariates on death. Patients' average (sd) age was 29.1 (10.9), 62% were female, 76% were married and 49% were housewives. The cumulative case fatality rate was 30%. The median time to death after oduvan ingestion was 3 days. Common signs and symptoms included hypokalaemia, vomiting, hyponatraemia, altered sensorium, bradycardia and abnormal ECG. There was a 58% risk reduction (95% CI: 29-75) in death with each 1 mmol/l increase in plasma potassium level. Patient's age was associated with an increased risk of death and the estimated hazard ratio for an increase of 10 years in age was 1.56 (95% CI: 1.18-2.07). Use of boiled oduvan extract was associated with an increased mortality (HR: 2.71, 95% CI: 1.17-6.32) compared to ingesting fresh leaves. Risk factors for oduvan poisoning mortality were hypokalaemia, older age, mode of consumption and an elevated risk for death with presence of chronic disease. When consumed as a boiled extract, oduvan is more lethal. We recommend close monitoring, correction of plasma potassium and appropriate supportive measures.

Trends in injury-related deaths before and after dissolution of the Union of Soviet Socialist Republics.

The relationship of death rates due to injury with the dissolution of the Soviet empire deserves special focus because of the impact of injury deaths on the productive working population between the ages of 15 and 65 years. A retrospective review was performed of annual mortality rates due to intentional and unintentional injuries, using WHO data from 1980 to 2003. Using data from a comparison group in Western Europe, the nations of the former Union of Soviet Socialist Republics (USSR) were studied. Annual death rates were examined using polynomial regression after 3-year moving averages smoothed the plots. The majority of trends in injury-related death noted in the comparison group decreased constantly over time. In contrast, many nations of the former USSR experienced an initial decrease, reaching a nadir in 1985 - 1987, followed by an increase in injury-related deaths. Moreover, many of these nations experienced a subsequent decrease following the dissolution of the USSR in 1991. The monitoring of trends in injury mortality in countries experiencing social and political upheavals is warranted.

Sexual and physical abuse are not associated with rectal hypersensitivity in patients with irritable bowel syndrome.

BACKGROUND: Patients with irritable bowel syndrome (IBS) have reduced pain thresholds for rectal distension. In addition, the prevalence of sexual/physical abuse in referred IBS patients is high and is associated with greater pain reporting, poorer health status, and poorer outcome. This lead to a hypothesis that abuse history may sensitise patients to report pain at a lower threshold. AIM: To compare rectal pain thresholds in women with IBS who had a history of severe abuse to IBS women with no history of abuse. METHODS: We studied 74 IBS patients with a history of severe physical and/or sexual abuse and 85 patients with no history of abuse. Abuse history was assessed by a previously validated self-report abuse screening questionnaire. Rectal sensory thresholds were assessed using an electronic barostat and determined by the ascending method of limit (AML) and by the tracking technique. RESULTS: IBS patients with a history of severe abuse had significantly higher rectal pain thresholds, as measured by AML (F (1, 111) = 6.06; p = 0.015) and the tracking technique (F (1, 109) = 5.21; p = 0.024). Patients with a history of severe abuse also reported a significantly higher threshold for urgency to defecate (F (1, 113) = 11.23; p =.001). CONCLUSION: Severe sexual/physical abuse is associated with higher urge and pain thresholds for rectal distension in IBS patients. This suggests that the greater pain reporting and poorer health status in IBS patients with abuse history are not related to increased rectal pain sensitivity. Further studies are needed to determine the causes of these findings.

Determinants of work injuries in mines - an application of structural equation modelling.

In spite of stringent regulations and much attention towards reducing risks in the physical environment, the mining industry continues to be associated with high levels of accidents, injuries and illnesses. Only engineering solutions to accident prevention are inappropriate unless coupled with focused attention to the attitudes and behaviours of the mineworkers in coping with the inherent physical, technical and situational risks. The present study identified these various risk factors and analysed their influences on work injury in a causal framework. Data were collected from an underground coalmine of India. The pattern and strength of relationships of 16 causal factors with work injuries were assessed through structural equation modelling. The case study results showed that negatively personified individuals are of major concern for safety improvement in the mine studied. They not only fail to avoid work injuries, they are unable to extend safe work behaviours in their work. The variable safety environment is negatively affected by personality, whereas social support has a positive relationship with safety environment. The variable job hazards appeared to have a significant relationship with job involvement, which has a negative relationship with work injury. Elimination of negative behaviours must be focused and committed by the mine safety management. Long term planning through (i) identification of negative individuals, (ii) proper councelling of adverse effects of negative behaviours, and (iii) special training with psychological treatment is highly required. Identification may begin while recruiting new workers through interview. Proper allocation of jobs (right person for right job) may be a judicial solution to this end.

[The use of state of the art statistical methodology in biomedical research: the example of methodology for interim analyses in clinical trials]. / El uso de metodología estadística actualizada en la investigación biomédica: el ejemplo de metodología para análisis interino en ensayos clínicos.

BACKGROUND: Clinical research stresses the need for evidence as a basis for setting research agendas. The extent to which current clinical research involves recent advances in biostatistics is not well known. AIM: This manuscript focuses on the use of interim analyses accumulating data in clinical trials. MATERIAL AND METHODS: A PubMed search was conducted for the period 1990-2000. Keywords used included 'interim analysis', 'alpha-spending function', and 'early termination', after subsetting to 'randomised controlled trials (RCT)' that were 'multicenter' since these particular studies are more likely to have external monitoring boards conducting interim analyses. Studies were grouped by clinical discipline and also by journals of high or low impact index. RESULTS: Thirteen thousand two hundred eighty two articles with both 'RCT' and 'multicenter' were found. Of these, the 'interim analysis' keyword yielded 879 articles, of which only 163 (19%) studies reported their methodology. The percentages varied by clinical discipline and by journal type. A 25% random sample of the articles were further reviewed to investigate the appropriateness of the interim analysis methods used. DISCUSSION: Recent articles in the medical literature attempt to make the interim analysis methods more accessible to clinicians. However, further training of statisticians involved in clinical trials is needed as well.

Gender differences in the longitudinal predictors of adolescent dating violence.

BACKGROUND: Adolescent dating violence is a public health problem. The public health approach to prevention is to identify predictors of problem behaviors and develop interventions to eliminate or reduce those predictors with the intention of altering the chain of causation. Longitudinal data are preferred for identifying predictors of behavior but all dating violence studies have used cross-sectional data. We use longitudinal data to examine predictors of adolescent dating violence from several domains guided by an ecological perspective. METHODS: Eighty percent (N = 1,965) of the 8th- and 9th-graders in one county completed baseline questionnaires in school and 90% (N = 1,759) of those adolescents completed questionnaires again 1&1/2 years later. Proportional odds models were used to identify cross-sectional correlates and longitudinal predictors of dating violence perpetration that occurred between baseline and follow-up. RESULTS: Most of the study variables were correlated with dating violence in cross-sectional analyses. Having friends who are victims of dating violence, using alcohol, and being of a race other than white predicted dating violence perpetration by females. Holding attitudes that are accepting of dating violence predicted dating violence perpetration by males. CONCLUSION: The findings suggest that intervention strategies should vary for males and females and that when basing interventions on cross-sectional findings, scarce resources may be stretched to address persons who may not truly be at risk of beginning to perpetrate dating violence.

[Interim analysis in clinical trials: a methodological guide]. / Análisis interino en ensayos clínicos: una guía metodológica.

Interim analysis of data accumulated in clinical trials is one aspect of the monitoring of the study progress. It is usually done to assess whether there are significant differences in efficacy between the experimental and control treatment groups, in order to decide whether to stop or no the trial prematurely. Among many reasons for early interruption of a trial is the ethical consideration that subjects should not be exposed to an unsafe, inferior or ineffective treatment. Statistical methods suited for doing interim analysis, that allow to control the probability of incorrectly rejecting the null hypothesis of no treatment differences, are often not well understood by researchers. In this article we present an intuitive, non-mathematical explanation and review of the statistical methods for doing interim analysis in clinical trials along with an illustrative example of the application of the methods on a hypothetical dataset.

What determines severity among patients with painful functional bowel disorders?

OBJECTIVE: For patients with painful functional bowel disorders (FBD), physicians frequently make diagnostic and treatment decisions based on the severity of the pain reported; patients with severe painful complaints may receive extensive diagnostic tests and treatments. Therefore, it would be important to determine what clinical factors contribute to the judgment of severity among patients with FBD. The aim of this study was to identify the psychosocial, behavioral, and physiological (visceral sensitivity) factors that predicted severity in patients with moderate to severe FBD. METHODS: Two hundred eleven female patients with moderate or severe FBD, as determined by the Functional Bowel Disorder Severity Index, entered a multicenter treatment trial at the University of North Carolina and the University of Toronto. Patients filled out diary cards and were given questionnaires and physiological testing (rectal sensitivity using barostat). Analysis of covariance and logistic regression adjusting for demographic factors were performed to determine which factors distinguished patients at study entry with moderate from those with severe FBD. RESULTS: Patients with severe FBD were characterized by greater depression and psychological distress, poorer physical functioning and health-related quality of life, more maladaptive coping strategies, and greater health care utilization. There was a trend for patients with severe FBD to have lower rectal sensation thresholds. Regression analysis indicated that severity was best predicted by behavioral features: poorer daily physical function, difficulties related to eating, more phone calls to the physician, and more days in bed for GI symptoms. CONCLUSIONS: We conclude that patient illness behaviors are best correlated with severity in FBD. The use of psychopharmacological agents (e.g., antidepressants) and psychological treatments to treat psychiatric comorbidity and to improve behavioral coping styles is recommended. Training to help medical physicians identify and respond to psychosocial and behavioral features of these conditions is likely to improve patient satisfaction with their care and the clinical outcome.

Further validation of the IBS-QOL: a disease-specific quality-of-life questionnaire.

OBJECTIVE: There has been growing interest in the investigation of health-related quality of life (HRQOL) among patients with gastrointestinal (GI) disorders. We recently reported on the development and preliminary validation of the IBS-QOL, a specific quality-of-life measure for irritable bowel syndrome (IBS). The aim of this study was to determine the longitudinal construct validity (responsiveness) of the IBS-QOL. METHODS: Female patients enrolled in a multicenter treatment trial for functional bowel disorders were studied pre- and posttreatment with the IBS-QOL and other health status measures. Based on the response to treatment for several variables (pain/14-day score, daily function, and days in bed/3 months), patients were stratified into Responders, Partial Responders, and Nonresponders. Change scores in the IBS-QOL were then statistically compared with changes in the other variables to determine their correlation and whether Responders were significantly different from non- and Partial Responders on the IBS-QOL. RESULTS: There was a significant correlation between change scores on the IBS-QOL and the other measures of treatment effect (Pain/14 days, r = 0.25, p < 0.002; Sickness Impact Profile [SIP] Total Score, r = 0.28, p < 0.0004). In addition, the IBS-QOL scores significantly differentiated Responders from Nonresponders for most of the variables tested (regression trend test for Pain/14 days, p < 0.04; SIP Total, p < 0.0001; SIP Physical, p < 0.0001; SIP Psychosocial, p < 0.002, and SIP Eating, p < 0.04). CONCLUSION: The IBS-QOL is responsive to treatment in a referral-based clinical population of patients with functional bowel disorders.

The influence of physical activity, socioeconomic status, and ethnicity on the weight status of adolescents.

OBJECTIVE: This study examined the effects of physical activity, television viewing, video game play, socioeconomic status (SES), and ethnicity on body mass index (BMI). RESEARCH METHODS AND PROCEDURES: The sample was 2389 adolescents, 10 to 16 years of age (12.7 +/- 1.0 years); 1240 (52%) females and 1149 (48%) males; 77% white and 23% African American; from rural (77%) and urban (23%) settings. BMI and skinfolds were directly assessed. All other data were obtained from questionnaires. RESULTS: Watching television on non-school days was related to being overweight (p < 0.005). However, when BMI analyses were adjusted for ethnicity and SES, there were no significant effects of television viewing on BMI (p > 0.061). Increased hours of video game play enhanced the risk of being overweight for both genders when analyses were adjusted for ethnicity and SES (p < 0.019). In males, participation in as little as one high-intensity physical activity 3 to 5 days a week decreased the ethnic- and SES-adjusted relative risk of being overweight (RR = 0.646; CI: 0.427 to 0.977). For females, the ethnic- and SES-adjusted relative risk for being overweight was not significantly altered by physical activity. The logistic analyses further indicated the influence of low SES and African American ethnicity overshadowed any direct effect of television or videos. DISCUSSION: Because weight status of male adolescents appears to be more related to exercise habits than to television or video game habits, increased participation in high-intensity exercise appears to be important. For females, neither videos nor exercise habits appear to be related to risk of being overweight. However, ethnicity and SES may be important factors that can influence body weight status, while television viewing may be of some importance. Thus, programs to reduce obesity in female adolescent should focus their efforts in lower SES communities.

Seroprevalence and seroincidence of Norwalk-like virus infection among Brazilian infants and children.

To determine the importance of Norwalk-like viruses (NLVs) as pediatric pathogens in a developing country, the seroprevalence and seroincidence of this group of viruses in a cohort of children less than 4 years of age in an urban shantytown in northeastern Brazil was examined. Serum samples were collected approximately every 6 months from 135 children who were surveyed three times each week for diarrhea and vomiting. NLV IgG was measured by an enzyme immunosorbent assay (EIA) with recombinant Norwalk virus capsid protein. Overall NLV seroprevalence was 71%, and the overall NLV seroconversion rate was 0.7 seroconversions per child-year. The highest age-specific NLV seroconversion rate (0.8 seroconversions per child-year) was observed in the 13-24-month age group. For all study children, the incidence of diarrhea and vomiting was significantly greater (P < 0.01) during time periods spanned by serum pairs that indicated NLV seroconversion compared with time periods without NLV seroconversion. However, NLV seroconversion was not associated with gastrointestinal symptoms during the first year of life.

[Categorizing variables in the statistical analysis of data: consequences for interpreting the results]. / Categorización de variables en el análisis estadístico de datos: consecuencias sobre la interpretación de resultados.

Frequently during the process of data analysis in epidemiological studies, the scale of one or more continuous variables is changed. The objective of this paper was to assess the consequences of categorizing variables during data analysis. We studied three situations with different scenarios for statistical analysis with regression models. The results show that dichotomizing continuous variables can substantially modify the relationships between dependent and independent variables. Thus, for example, in epidemiological studies trying to evaluate the effect of an exposure on a response, the magnitude and/or the direction of this effect can be biased by dichotomizing a variable. We therefore recommend avoiding, as much as possible, the categorization of variables when doing analyses.