Search PubMed⌕ Search

SEARCH · Search PubMed

Results for “Linear Regression”

Search indexed PubMed citations on genomics, clinical trials, systematic reviews and public health. Explore titles, authors and supplied subject terms, then open the PubMed record.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

At least 775 records · Page 43Linked to original sources

Survival of Staphylococcus aureus ATCC 13565 in intermediate moisture foods is highly variable.

Staphylococcus aureus is a gram-positive, enterotoxin-producing coccus. It is a hardy organism and known to survive over a wide range of water activities, pH values, and temperatures. The objective of this study was to model the survival or gradual inactivation of S. aureus ATCC 13565 in intermediate moisture foods (IMFs). Various initial concentrations (approximately 10(1), 10(2), 10(3), and 10(4) CFU/g) were used to inoculate three different IMFs (beefsteak, bread, and chicken pockets). Viable counts were determined up to 60 days using tryptic soy agar. Inoculum size did not influence the survival or gradual inactivation of S. aureus in these foods. The rate of change (increase or decrease) in log CFU/day was calculated for every consecutive pair of data points and by linear regression for each inactivation curve. Both consecutive pair and linear regression rates of change were fit to logistic distributions (with parameters alpha and beta) for each food. Based on the distribution parameters, survival or gradual inactivation of S. aureus was predicted by computer simulation. The simulations indicated an overall decline in S. aureus population over time, although a small fraction of samples in the consecutive pair simulation showed a slight population increase even after 60 days, consistent with the observed data. Simulation results were compared to predictions from other computer models. The models of Stewart et al., were fail-safe, predicting the possibility of significant growth only after > 3,000 days. The USDA pathogen modeling program predictions were found to be fail-dangerous, predicting declines at least four times faster than observed.

Animals↗

Human gaze stabilization during active head translations.

This study investigated how binocular gaze is controlled to compensate for self-generated translational movements of the head where geometric requirements dictate that the ideal gaze signal needs to be modulated by the inverse of target distance. Binocular gaze (eye plus head) was measured for visual and remembered targets at various distances in six human subjects during active head translations at frequencies of 0.25, 0.5, 1.0, and 1.5 Hz. We found that, during head translations, gaze changes were achieved by a combination of eye and head rotations. Accordingly, stabilization performance was characterized by the gaze response parameters sensitivity and phase, where sensitivity is defined as the ratio of gaze velocity and translational eye velocity and where phase refers to the phase delay of gaze velocity relative to translational eye velocity. In the analysis, we used a binocular coordinate system yielding a version and a vergence component. We examined how frequency and target distance, estimated from the vergence angle, affected sensitivity and phase of the version component of the gaze signal and compared the results to the requirements for ideal performance. The relation between gaze sensitivity and the inverse of distance was characterized by a linear regression analysis. The ratio of the slope of the linear regression and the slope required for ideal stabilization provided a measure for the degree of "distance compensation." The results show that distance compensation was better for a visual target than for remembered targets in darkness, and behaved according to low-pass characteristics in both target conditions. It declined from 1.00 to 0.84 for visual targets and from 0.87 to 0.57 for remembered targets in the frequency range 0.25-1.5 Hz. The intercept obtained from the regression yielded the gaze response at zero vergence and specified a "default sensitivity" of gaze compensation. Default sensitivity increased with frequency from 0.02 at 0.25 Hz to 0.10 degrees/cm at 1.5 Hz for visual targets and from 0.04 to 0.16 degrees/cm in darkness. The phase delays of the gaze response increased on average from 2 to 7 degrees in the frequency range 0.25-1.5 Hz. In comparison with earlier passive studies, active translation compensation in the dark is superior at all frequencies where comparison was possible. We conclude that a nonvestibular signal with low-pass characteristics contributes to gaze during active head translations.

Adult↗

Standardized single breath normal values for carbon monoxide diffusing capacity.

Prediction equations for DLCO and diffusing capacity per unit of lung volume (DL/VA) were generated from 245 normal subjects (122 women and 123 men) using a standardized technique for measuring DLCO. Measurements were made at an altitude of 1,400 meters. Multiple linear regressions were made using standard and robust regression techniques. The resultant equations predicted values for DLCO and DL/VA that were higher than most previously reported values. The use of robust regressions did not add to the predictability of standard linear regressions.

Adolescent↗

The effect of outflow pressure upon thoracic duct lymph flow rate in fetal sheep.

Edema develops when lymph does not return to the venous circulation at a rate equal to the rate of capillary filtration. Fetal sheep develop edema as well as an increased central venous pressure while undergoing atrial pacing at 320 beats per min. We hypothesized that the increased central venous pressure augmented the appearance of fetal edema by impairing the return of thoracic duct lymph to the venous circulation. To investigate this hypothesis, we studied the effect of outflow pressure upon thoracic duct lymph flow in 10 unanesthetized fetal sheep who had low resistance lymph catheters placed in the cervical thoracic duct near its junction with the left jugular vein. After the ewe and fetus recovered for 5 d, we altered the outflow pressure of the lymph catheter by adjusting its height with respect to amniotic fluid pressure and measured the resultant change in thoracic duct lymph flow rate. We found that lymph flow rate was constant over the range of outflow pressures (central venous pressures) normally encountered but decreased in a linear fashion at pressures greater than 0.68 kPa (5.1 torr). Lymph flow stopped at an outflow pressure of 2.40 kPa (18 torr). The data points are best fit by two lines obtained by a piecewise linear regression rather than a single line obtained from a linear regression. We conclude that fetal thoracic duct lymph flow is sensitive to elevations in outflow pressure. Lymph flow begins to diminish at outflow pressures corresponding to central venous pressures commonly encountered in pathologic conditions and may augment the appearance of fetal edema.

Animals↗

The predictive power of surface profile parameters on the amount of wear measured in vitro on metal-on-polyethylene artificial hip joints.

Various studies report a weak correlation between the average surface roughness R(a) of metallic ballheads and the observed wear rate of the polyethylene cups coupled to them. The aim of this study is to verify, with controlled in vitro experiments, whether other parameters such as the total roughness R(t), and the skewness R(sk) are better predictors of the polyethylene wear when the metallic heads present a surface conditioning that varies substantially from specimen to specimen, as is usually the case with retrieved specimens. Twelve CoCrMo (cobalt-chromium-molybdenum) ballheads were intentionally damaged in order to reproduce a wide spectrum of surface conditioning and were then subjected to the standard wear test against polyethylene cups, using a hip joint wear simulator. After 2 x 10(6) cycles the weight lost by the cups was assessed with a gravimetric procedure, and the surface roughness of the metallic ballheads was qualified in terms of R(a), R(t), and R(sk). The various parameters were correlated to the weight loss using a linear regression analysis. The skewness R(sk) showed a coefficient linear regression R2 = 0.80, while the average roughness R(a), used in most previous studies, presented an R2 = 0.56. It was concluded that when specimens with substantially different surface conditioning are compared, as in retrievals analysis, it is also important to report the skewness R(sk) so that qualify the surface roughness of the specimens can be qualified.

Coated Materials, Biocompatible↗

[Secular trends in the stature of Brazilian Navy recruits born from 1940 and 1965].

The present paper reports data on secular trends in the stature of Brazilian Navy recruits born from 1940 to 1965. The final sample included 3269 individuals aged 18.00-18.99. Statistics performed were: ANOVA (one-way and two-way), Sheffe test, simple linear regression between stature and year of birth, and multiple linear regression adjusting for level of schooling (beta coefficient) and chi-square. Results indicated a progressive growth trend in stature of 0.1 cm/yr. for the country as a whole. The trend was also observed for nearly all regions and two out of three levels of schooling and can be explained by improvement in some of the country's health indicators. One important characteristic was a higher level of schooling observed among Navy recruits, suggesting that these individuals represent a highly select group, and that therefore data on the Navy cannot be applied directly to the Brazilian population as a whole.

Adolescent↗

High-frequency toneburst-evoked ABR latency-intensity functions in sensorineural hearing-impaired humans.

The latency-intensity functions (LIFs) of ABRs elicited by high-frequency (8, 10, 12, and 14 kHz) toneburst stimuli were evaluated in 20 subjects with confirmed 'moderate' high-frequency sensorineural hearing loss. Wave V results from clicks and tonebursts revealed all intra- and intersession data to be reliable (p > 0.05). Linear regression curves were highly significant (p < or = 0.0001), indicating linear relationships for all stimuli analyzed. Comparisons between the linear regression curves from a previously reported normal-hearing subject group and this sensorineural hearing-impaired group showed no significant differences. This study demonstrated that tonebursts at 8, 10, and 12 kHz evoked ABRs which decreased in latency as a function of increasing intensity and that these LIFs were consistent and orderly (14 kHz was not determinable). These results will contribute information to facilitate the establishment of change criteria used to predict change in hearing during treatment with ototoxic medications.

Acoustic Stimulation↗

Competitive interactions of drugs with their targets: a computerized improvement of Dixon's algorithm.

Many drugs are competitive and reversible enzyme inhibitors. When the target enzyme kinetics follows the Michaelis-Menten equation, the enzyme affinity of the inhibitor is characterized by a single parameter: the Ki value. This parameter is usually determined via Dixon's procedure: (i). the rate of reaction (V) is measured in the presence of a few concentrations of the inhibitor (I); (ii.) 1/V versus I gives a straight line, which allows a graphic determination of the inhibitory constants, or better via a least-square fit the linear regression. The introduction of appropriate weighting factors in the linear regression may improve the accuracy of the Ki determinations.

Algorithms↗

Regulatory motif finding by logic regression.

MOTIVATION: Multiple transcription factors coordinately control transcriptional regulation of genes in eukaryotes. Although many computational methods consider the identification of individual transcription factor binding sites (TFBSs), very few focus on the interactions between these sites. We consider finding TFBSs and their context specific interactions using microarray gene expression data. We devise a hybrid approach called LogicMotif composed of a TFBS identification method combined with the new regression methodology logic regression. LogicMotif has two steps: First, potential binding sites are identified from transcription control regions of genes of interest. Various available methods can be used in this step when the genes of interest can be divided into groups such as up-and downregulated. For this step, we also develop a simple univariate regression and extension method MFURE to extract candidate TFBSs from a large number of genes in the availability of microarray gene expression data. MFURE provides an alternative method for this step when partitioning of the genes into disjoint groups is not preferred. This first step aims to identify individual sites within gene groups of interest or sites that are correlated with the gene expression outcome. In the second step, logic regression is used to build a predictive model of outcome of interest (either gene expression or up- and down-regulation) using these potential sites. This 2-fold approach creates a rich diverse set of potential binding sites in the first step and builds regression or classification models in the second step using logic regression that is particularly good at identifying complex interactions. RESULTS: LogicMotif is applied to two publicly available datasets. A genome-wide gene expression data set of Saccharomyces cerevisiae is used for validation. The regression models obtained are interpretable and the biological implications are in agreement with the known resuts. This analysis suggests that LogicMotif provides biologically more reasonable regression models than previous analysis of this dataset with standard linear regression methods. Another dataset of S.cerevisiae illustrates the use of LogicMotif in classification questions by building a model that discriminates between up- and down-regulated genes in iron copper deficiency. LogicMotif identifies an inductive and two repressor motifs in this dataset. The inductive motif matches the binding site of the transcription factor Aft1p that has a key role in regulation of the uptake process. One of the novel repressor sites is highly present in transcription control regions of FeS genes. This site could represent a TFBS for an unknown transcription factor involved in repression of genes encoding FeS proteins in iron deficiency. We establish the robustness of the method to the type of outcome variable used by considering both continuous and binary outcome variables for this dataset. Our results indicate that logic regression used in combination with cluster/group operating binding site identification methods or with our proposed method MFURE is a powerful and flexible alternative to linear regression based motif finding methods. AVAILABILITY: Source code for logic regression is freely available as a package of the R programming language by Ruczinski et al. (2003) and can be downloaded at http://bear.fhcrc.org/~ingor/logic/download/download.html an R package for MFURE is available at http://www.stat.berkeley.edu/~sunduz/software.html

Algorithms↗

The 1.5-Mb region spanning the myotonic dystrophy locus shows uniform recombination frequency.

The myotonic dystrophy (DM) mutation has been identified as a heritable unstable CTG trinucleotide repeat sequence. The intergenerational amplification of this sequence is an example of a new class of dynamic mutations responsible for human genetic diseases. To ascertain whether recombination activity influences, or is affected by, the presence of this unique sequence, a comprehensive study of the physical and genetic mapping data for the 1.5-Mb region of human chromosome 19q13.3, which contains the DM locus, was conducted. The recombination rate for this region was examined by correlating genetic distance to physical distance for six selected marker loci. The following markers span the DM region: 19qCEN-p alpha 1.4 (D19S37)-APOC2-CKM-pE0.8 (D19S115)-pGB2.6 (DM)-p134c (D19S51)-19qTER. Initial linear regression analysis of these two parameters failed to reveal a significant linear correlation (coefficient of determination, r2 = .19), suggesting nonuniform rates of recombination. However, the presence of a recombination hot spot was believed to be unlikely, as the marker-to-marker pairs that showed the greatest deviation in recombination frequency were not restricted to a specific region of the 1.5 Mb studied and had relatively broad confidence intervals, as reflected by low LOD values. A second linear regression analysis using only marker intervals with high LOD scores (Zmax > 22) showed linear correlation (r2 = .68) for the entire 1.5-Mb region. This analysis indicated a relatively uniform recombination frequency in the 1.5-Mb region spanning the DM locus. Furthermore, the recombinations observed were neither under- nor overrepresented on DM chromosomes. Consequently, recombination activity is unlikely to influence, or be affected by, the presence of the DM mutation.

Chromosomes, Human, Pair 19↗

Prediction equations for spirometric parameters in Mexican adult population.

BACKGROUND: Our objective was to develop prediction equations for spirometric parameters that included the following: forced vital capacity (FVC); 1st-sec forced expiratory volume (FEV1); FEV1/FVC ratio; peak expiratory flow (PEF); maximal expiratory flow 50% FVC (VF50) and maximal expiratory flow 75% FVC (VF75); maximal mid-expiratory flow (FEF25-75), and forced late expiratory flow rate (FEF75-85) in a sample of adult Mexican population of both sexes. Age and height variables were utilized. METHODS: Spirometric studies were carried out, observing the international recommendations and norms in force and effected under basal conditions and post-bronchodilator. Linear regression equations were generated based on gender, age, and height. For each spirometric parameter, we selected a linear model. Studies were carried out with spirometry that was in agreement with quality criteria recommended by the American Thoracic Society (ATS). Bronchodilatator administration allowed for elimination of subjects with subclinical bronchial hyperreactivity RESULTS: We studied 436 patients with normal clinical radiographic and pulmonary function, with negative smoking and age range between 17 and 63 years; 206 were females and 230, males. Linear regression equations obtained had direct linear correlation with height and inverse linear correlation with age; in addition, equations had a determination coefficient equal to or less than those reported by authors recommended by the ATS. CONCLUSIONS: Equations obtained in this study possessed the quality required for application in adult Mexican population exposed or not to occupational and environmental contaminants.

Adolescent↗

The art and science of regression modelling; methods for building valid models to explore hormone and body composition interactions.

Multiple linear regression modelling is commonly used to investigate how hormones and body composition interact, but for valid interpretation a sound methodological approach must be used. It is particularly important that the assumptions for regression are met so that spurious associations are not generated. In this article we show how different approaches to building a multiple linear regression model can influence perceived associations, using examples from the literature and our own data related to predicting fasting insulin and leptin levels from total body fat and fat distribution in children.

Adipose Tissue↗

Reducing major cause-specific hospitalization rates and shortening hospital stays after influenza vaccination.

BACKGROUND: The impact of influenza vaccination on major cause-specific hospitalization and the duration of hospital stay is rarely reported. Our purpose was to study the effect of vaccine efficacy on major disease-specific hospitalization and the duration of hospital stays among elderly persons. SUBJECTS AND METHODS: From 1 January through 30 June 2001, we prospectively observed 35,637 vaccinated elderly persons (age, >or=65 years) and 53,094 unvaccinated elderly persons in Kaohsiung County, Taiwan, by computerized linkage to the National Health Insurance database. Of these persons, 21,347 had been assigned a high-risk status by the Department of Health, Taiwan. Univariate and multivariate logistic regression were used for determining vaccine efficacy in hospitalization. Multiple linear regression analyses were performed for determining the length of hospital stays. RESULTS: In both high-risk and low-risk groups, vaccination was associated with reducing the rates of hospitalization for all causes (20% vs. 23%), lung diseases, congestive heart failure (43% vs. 32%), renal disease, and liver disease (P<.05). It was also significant for stroke, hypertension, diabetes, neoplasm, and injury in low-risk patients (P<.05). Multivariate logistic regression showed that vaccination was significantly associated with reducing the rate of hospitalization (odds ratio [OR], 0.89; 95% confidence interval [CI], 0.86-0.92), but those with high-risk status had an increased risk of hospitalization (OR, 3.69; 95% CI, 3.56-3.82). Multiple linear regression analysis showed that vaccination decreased the duration of all-cause hospital stays (coefficient, -2.4 days; 95% CI, -2.7 to -2.1 days) and of hospitalization due to lung disease (coefficient, -4.9 days; 95% CI, -6.0 to -3.8 days). CONCLUSION: Influenza vaccination may reduce hospitalization rates and shorten hospital stays not only for lung diseases but also for other common diseases in high-risk and low-risk elderly populations.

Aged↗

Variation in the use of red blood cell transfusions. A study of four common medical and surgical conditions.

This study assessed variation in red cell transfusion practice among adult patients hospitalized with ulcer disease (ULCER), and those undergoing coronary artery bypass grafting (CABG), hip surgery (HIP), or total knee replacement (KNEE). The study design was a retrospective analysis of the 1989 MedisGroups Hospital Comparative Database, and the participants were adult patients presenting for their first admission with ULCER (N = 4,664), CABG (N = 6,812), HIP (N = 4,131) or KNEE (N = 3,042) in the MedisGroups Hospital Comparative Database. Outcome measures were whether a patient was transfused, and the number of units transfused. Logistic regression was used to analyze the decision to transfuse, and linear regression to analyze the number of units transfused. In these analyses, patient characteristics, hospital characteristics, and unique hospital identity were used as independent variables. The percentage of patients transfused was ULCER 50%, CABG 81%, HIP 69%, and KNEE 51%. The range among hospitals in the percentage of patients transfused was ULCER 11% to 76%, CABG 51% to 100%, HIP 36% to 95%, and KNEE 9% to 97%. When only patient characteristics were entered in the linear regression analyses, the R2 values were ULCER 0.33, CABG 0.11, HIP 0.11, and KNEE 0.07. When hospital was added, the R2 increased to ULCER 0.38, CABG 0.29, HIP 0.19, and KNEE 0.20 (P < 0.0001 for the change for all analyses). The results of the logistic regression analyses of the probability of transfusion were similar. There is substantial interhospital variation in the proportion of patients transfused and number of units transfused in the four conditions studied. Patient demographic and clinical characteristics explain a substantial proportion of the variation in transfusion practices for ulcer patients, but little of the variation in the three surgical conditions.

Adult↗

[Regression analysis of idiopathic endothelial cell loss after perforating normal risk keratoplasty: basic principles for long-term analysis of endothelial risk factors in a retrospective clinical study].

BACKGROUND: For the characterisation of influencing factors on chronic endothelial cell loss after penetrating keratoplasty by means of multivariate statistics, a mathematical description of the course of the individual postoperative endothelial cell density is a prerequisite. This mathematical description should result in a standardized index value describing course and amount of the postoperative endothelial cell loss over time in a canonical way. The slopes of the linear regression lines for each individual scatter plot of a) the endothelial cell density values plotted against the respective postoperative time directly (linear regression), and b) after logarithmic transformation (exponential regression) are evaluated, respectively. PATIENTS AND METHODS: 58 patients after normal-risk keratoplasty (26x keratokonus, 22x Fuchs-dystrophy and 10 cases of corneal decompensation after cataract surgery) with 5 or more postoperatively acquired endothelial density values and without any episodes of graft rejection were included in this study. Mean follow up was 2.9 +/- 1.1 years. The postoperative endothelial cell density values were plotted against the respective time for each patient individually. The coefficients of variation (R2) derived from the linear and the exponential regression models were calculated for each of these scatter plots. The pairs of R2 values (linear vs. exponential) were compared statistically. A dependence of the difference of linear and exponential R2-values on the ophthalmologic diagnosis was tested as well. RESULTS: The linear model is able to declare 83% the total variance of the course of the endothelial cell density. The exponential model even declares 86%. This small difference was statistically significant. Since both methods of regression describe the course of the cell density well, intra/and extrapolation of missing endothelial values is possible with both models. No dependence of the difference of linear and exponential R2-values on the ophthalmologic diagnosis could be demonstrated. CONCLUSIONS: Both, the intuitively understandable slope of the linear regression line and the constant of decay of the exponential regression curve, are suitable for describing the amount of the postoperative loss of endothelial cells after normal-risk keratoplasty independent of the ophthalmologic diagnosis. Both can thus be used as target variable in forthcoming statistical analyses for chronic endothelial cell loss.

Adult↗

Literacy and self-reported educational levels in relation to Mini-mental State Examination scores.

BACKGROUND AND OBJECTIVES: Because of its brevity and ease of use, the Mini-mental State Examination (MMSE) is commonly used to screen and follow patients with cognitive impairment. This pilot study attempted to determine the relationships between literacy, age, and self-reported educational level and the total MMSE score. METHODS: Cross-sectional analysis of all patients was followed by a family practice group at five local nursing homes. The associations between the patients' MMSE scores; literacy, as measured by the Rapid Estimate of Adult Literacy in Medicine (REALM); self-reported educational level; and age were determined using Pearson's correlation coefficient and stepwise multivariate linear regression. RESULTS: A total of 105 patients completed the study. Linear regression analysis showed that MMSE scores were significantly predicted by REALM score (P < .001) and the patient's age (P < .02). However, after accounting for REALM score and age, the self-reported educational level was not related to the MMSE score (P < .8). A significant relationship was seen between the REALM score and the subjects' self-reported educational levels (r = .44, P < .001) but not the subjects' ages (r = -.17, P < 0.09). A significant linear correlation was found between the MMSE and REALM scores (r = .71, P < .0001) and a significant inverse correlation was seen between MMSE scores and the patients' ages (r = -.28, P < .004). The correlation coefficient between the patients' MMSE scores and the self-reported educational levels was .33 (P < .0007). CONCLUSIONS: Practitioners who rely on the MMSE should be aware that patients may score in the demented range because they cannot read well enough to accurately complete the test. Literacy testing with REALM or other instruments may help identify such patients.

Aged↗

Weight-height relationships among eight populations of West African origin: the case against constant BMI standards.

OBJECTIVE: To ascertain whether constant body mass index (BMI) standards are appropriate in genetically similar populations. DESIGN: Data are taken from the International Collaborative Study of Hypertension in Blacks (ICSHIB), an observational study. SUBJECTS: Individuals of African descent who were included in ICSHIB. Subjects lived in eight different sites: Barbados; Cameroon (urban and rural); Jamaica; Manchester, UK; Maywood, IL; urban Nigeria; and St Lucia. MEASUREMENTS: Weight and height. RESULTS: Constant BMI standards effectively argue for the constancy of slope of the linear regression equations of In(weight) on In(height) across populations. Linear regression results indicate that the height/weight relationship implied by the use of constant BMI standards, is not found in these populations and that there is much variation across groups. CONCLUSION: The use of constant BMI standards in classifying individuals prognostically may be unwise, even in genetically similar populations.

Africa, Western↗

Potential predictive factors of osteoporosis in HIV-positive subjects.

Recent reports showed a high frequency of osteopenia/osteoporosis in HIV-infected subjects. Mechanism on the basis of this alteration is still unclear, as the direct effect of virus or of antiretroviral drugs. One hundred sixty-one consecutive HIV-infected outpatients aged 30-50 years, both naive and HAART-treated for >1 year, were included. An interview questionnaire was performed to establish prior pathological, toxic, epidemiological histories, medications intake, physical activity and eating habits. Blood and urinary tests were checked to exclude concomitant diseases, as were markers of bone metabolism and vitamin D3-metabolites. Each subject underwent to a lumbar spine and left hipbone mineral density by DEXA, using WHO criteria for diagnosis of osteopenia/osteoporosis. Radiologist was unaware if the subject was receiving HAART or not. For groups' homogeneity Chi-square, Fisher's exact and Student's t tests were used. Logistic regression analysis was used to find predictors of osteopenia/osteoporosis and linear regression model to find differences in bone mass density. The demographic characteristics of the 48 naive subjects and the 113 on HAART were comparable. Eighty subjects (49.7%) showed osteopenia/osteoporosis: 22 (45.8%) naive and 58 (51.3%) on HAART (P = 0.46). Independent predictors of osteopenia/osteoporosis were female gender (OR: 3.02, 95% CI: 1.26-7.25, P = 0.01 vs. male), older age (OR: 1.10, 95% CI: 1.01-1.20, P = 0.03, for each additional year), low body mass index (OR: 0.78, 95% CI: 0.68-0.91, P = 0.001 for each additional unit) and higher HIV-RNA levels at DEXA (OR: 1.97, 95% CI: 1.16-3.34, P = 0.01 for each additional Log(10)), whereas the use of HAART (OR: 2.61, 95% CI: 0.66-10.27, P = 0.17 vs. naive) and the alterations of markers of bone metabolism were not significantly related to osteopenia/osteoporosis. Similar findings were obtained using linear regression model analysis. HIV-infected subjects have a high frequency of osteopenia/osteoporosis. Traditional risk factors are predictive of osteopenia/osteoporosis also in HIV-subjects; the association with higher HIV-RNA levels can suggest a direct role of HIV itself in the occurrence of bone disease.

Adult↗