SEARCH · PubMed Health

Results for “Logistic Models”

Explore indexed PubMed citations for clinical trials, systematic reviews and public health research. Read source abstracts and follow each citation to its original PubMed record.

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

At least 955 records · Page 53Linked to original sources

PubMed1993

Detection of bulimia in a primary care setting.

OBJECTIVE: To develop a screening tool for the identification of bulimia in ambulatory practice. DESIGN: Administration of a 112-item questionnaire about eating and weight-control practices to women with known bulimia and to healthy control patients. Questions were compared with DSM-III-R criteria of bulimia as a "gold standard." SETTING: Self-help group for eating disorders and hospital-based primary care practice. SUBJECTS: Thirty of 42 women with known bulimia met DSM-III-R criteria for current bulimia, and 124 of 130 control patients met the criterion of no history of an eating disorder. MAIN RESULTS: Thirteen individual questions discriminated between bulimic subjects and control subjects with a sensitivity and specificity of > 75%. When these questions were entered into a stepwise logistic model, two questions were independently significant. A "no" response to the question "Are you satisfied with your eating patterns?" or a "yes" response to "Do you ever eat in secret?" had a sensitivity of 1.00 and a specificity of 0.90 for bulimia. The positive predictive value, based on a 5% prevalence, was 0.36. CONCLUSIONS: A set of two questions may be as effective as a more extensive questionnaire in identifying women with eating disorders, and could be easily incorporated into the routine medical history obtained from all women.

Adolescent↗

PubMed2006

Histopathology of the right ventricular outflow tract and its relationship to clinical outcomes and arrhythmias in patients with tetralogy of Fallot.

OBJECTIVES: The purposes of this study were to evaluate the myocardial histopathology and ultrastructure in patients with tetralogy of Fallot and to identify the histopathologic characteristics that may predispose patients to postoperative myocardial dysfunction and arrhythmias. PATIENTS AND METHODS: Operatively resected crista supraventricularis muscle from 183 patients undergoing intracardiac repair of tetralogy of Fallot aged 12 months to 42 years (mean, 106.84 +/- 79.35 months) were studied by light and electron microscopy. Biventricular function and cardiac rhythm were assessed by 2-dimensional echocardiography and electrocardiography. RESULTS: The incidence of moderate or severe cellular hypertrophy, endocardial thickening, and interstitial fibrosis was 36%, 68.3%, and 65%, respectively. Logistic regression analysis demonstrated age greater than 4 years, systemic arterial desaturation, higher hematocrit values, and elevated ventricular end-diastolic pressures as the major predisposing risk factors for pathologic changes. Twenty-seven (81.8%) patients more than 15 years of age and 29 (29.3%) patients between 4 and 15 years of age had predominant right ventricular dysfunction and low cardiac output (chi(2) [1 degree of freedom (df)] = 27.95; P < .001; odds ratio [OR] = 10.86 [3.75-33.10]). Ventricular arrhythmia was detected in 11 patients in whom repair was performed between 4 and 15 years of age and in 13 patients whose age at operation was 15 years or older. According to an additive logistic model, the effect of age at repair on the influence of ventricular arrhythmia was significant (chi(2) [1 df] = 24.4; P < .001; OR = 8.21 (2.96-23.11]). CONCLUSIONS: The great majority of myocardial tissues in cyanotic tetralogy of Fallot indicates pre-existing ultrastructural hypertrophic and degenerative changes. The changes are more pronounced in older patients subjected to long-standing cyanosis and pressure overload and may account for or may coexist with the higher incidence of myocardial dysfunction and ventricular arrhythmia.

Adolescent↗

PubMed1994

The impact of membership in a health maintenance organization on hospital admission rates for acute chest pain.

OBJECTIVE: We evaluate the impact of membership in a staff-model health maintenance organization (HMO) on hospital admission rates for patients presenting to an emergency department with acute chest pain. DATA SOURCES AND STUDY SETTING: Primary prospective data were gathered from all 3,006 patients presenting with a chief complaint of chest pain to the emergency department (ED) of a university teaching hospital from October 1987 to November 1989. STUDY DESIGN: Prospective cohort analysis used clinical data to stratify patients into groups at high (> or = 25%), medium (8-24%), and low risk (< or = 7%) of acute myocardial infarction (AMI). Insurance status was determined as either HMO, Medicare, commercial, Medicaid, or self-pay. Triage decisions were recorded, and patient outcomes of AMI and other final diagnoses were determined for all patients. DATA COLLECTION METHODS: Clinical data were recorded by the physicians in the ED as part of a detailed protocol. Insurance data were recorded separately by the ED staff as part of the hospital administrative database. Patient outcomes were recorded daily by research nurses for hospitalized patients; for patients who were discharged from the ED, telephone or physician follow-up was accomplished within seven days after discharge. PRINCIPAL FINDINGS: HMO patients were more likely to be admitted to the hospital than patients in other insurance groups in both the medium- and low-risk patient categories. Within the low-risk category, after controlling for clinical differences in a multiple logistic model, HMO membership retained an independent positive association with hospital admission compared to all other insurance groups except Medicaid. CONCLUSIONS: For patients with acute chest pain who were at medium and low risk of acute myocardial infarction, HMO membership was associated with higher rates of hospital admission. These findings suggest that organizational factors beyond financial incentives may exercise an important influence on hospitalization rates for HMO patients.

Acute Disease↗

PubMed2005

The relationship between selected measures of periodontal status and demographic and behavioural risk factors.

OBJECTIVE: To assess differences between selected periodontal measures by demographic and behavioural factors in a nationally representative sample of the United States. METHODS: Data for 11,347 person's ages 20-79 years from the third National Health and Nutrition Examination Survey (NHANES III) were used. Indices and measures constructed from NHANES III data used for this study were: derived community periodontal index (dCPI), attachment loss extent index (ALEI), attachment loss (AL) scores, and a Periodontal Status Measure (PSM) developed for this study. RESULTS: The influence of demographic and behavioural factors varied across the four indices examined in multivariate cumulative logistic models. Moreover, there was significant effect modification by cigarette smoking with age in the ALEI and AL models. The odds ratio (OR) of increasing periodontal disease status among 20-39 year olds as measured by AL or ALEI for current smokers compared with non-smokers were OR=6.2 (95% confidence interval (CI)=4.1, 8.7) and OR=5.6 (95% CI=3.7, 8.7), respectively. In a similar comparison, the OR for dCPI was 2.6 (95% CI=1.7, 3.8). Furthermore, Mexican American ethnicity was generally not significant in any models using dCPI, PSM, AL, or ALEI and prior dental visit was more likely to be significant only in the dCPI and PSM models. DISCUSSION: Among the well-known demographic and behavioural influences on periodontal health status, some, such as race/ethnicity and prior dental visit status have different relationships with differing periodontal measures employed to assess periodontal status. Moreover, potential interactions among cofactors also are dependent upon the measure selected. Periodontal research findings may be influenced significantly by periodontal measure selection and its affect on measurement validity. This may have particular relevance to issues concerning disease surveillance and assessing reduction of disparities in oral health. Consequently, a renewed approach to developing appropriate measures for periodontal epidemiology is needed.

Adult↗

PubMed2002

Improvement in spine bone density and reduction in risk of vertebral fractures during treatment with antiresorptive drugs.

PURPOSE: To estimate how much the improvement in bone mass accounts for the reduction in risk of vertebral fracture that has been observed in randomized trials of antiresorptive treatments for osteoporosis. METHODS: After a systematic search, we conducted a meta-analysis of 12 trials to describe the relation between improvement in spine bone mineral density and reduction in risk of vertebral fracture in postmenopausal women. We also used logistic models to estimate the proportion of the reduction in risk of vertebral fracture observed with alendronate in the Fracture Intervention Trial that was due to improvement in bone mineral density. RESULTS: Across the 12 trials, a 1% improvement in spine bone mineral density was associated with a 0.03 decrease (95% confidence interval [CI]: 0.02 to 0.05) in the relative risk (RR) of vertebral fracture. The reductions in risk were greater than predicted from improvement in bone mineral density; for example, the model estimated that treatments predicted to reduce fracture risk by 20% (RR = 0.80), based on improvement in bone mineral density, actually reduce the risk of fracture by about 45% (RR = 0.55). In the Fracture Intervention Trial, improvement in spine bone mineral density explained 16% (95% CI: 11% to 27%) of the reduction in the risk of vertebral fracture with alendronate. CONCLUSION: Improvement in spine bone mineral density during treatment with antiresorptive drugs accounts for a predictable but small part of the observed reduction in the risk of vertebral fracture.

Alendronate↗

PubMed2006

Population distribution and zoonotic potential of gastrointestinal helminths of wild rats Rattus rattus and R. norvegicus from Jamaica.

The population distribution and zoonotic potential of gastrointestinal helminths in a naturally infected population of wild rats (Rattus rattus and Rattus norvegicus) in Jamaica are described. One hundred and thirty (29.7%) of 437 rats captured in the study were infected: 104 (35%) of 297 R. rattus compared with 26 (18.6%) of 140 R. norvegicus. Nine species of gastrointestinal helminths were recovered: Raillietina sp. (0.2%), Trichuris sp. (0.2%), Rictularia sp. (0.7%), Syphacia obvelata (1.1%), Strongyloides ratti (1.4%), Hymenolepis diminuta (3.8%), Protospirura muricola (4.3%), Moniliformis moniliformis (11.2%), and Nippostrongylus brasiliensis (14.2%). In a logistic model, the single risk factor identified for both M. moniliformis and P. muricola was R. rattus, compared with R. norvegicus (OR = 8.369 and 9.714, respectively). In comparison, the risk factor predicted for infection with N. brasiliensis was the northeastern section of Jamaica (OR = 11.000) compared with western Jamaica. Rictularia sp. represents a new geographic distribution record for the Caribbean region. Hymenolepis diminuta, M. moniliformis, Raillietina sp., and Rictularia sp. are potentially zoonotic, but only human infection with H. diminuta has been previously reported in the Caribbean.

Animals↗

PubMed2004

Further evidence of the association between periodontal conditions and coronary artery disease.

BACKGROUND: There is increasing evidence that chronic infections, such as periodontal diseases, could play a role in the initiation and development of coronary artery disease (CAD). The present study was intended to test for a possible association between presence and severity of periodontitis and coronary artery disease in a Belgian population. METHODS: A total of 108 CAD patients (mean age 59.2 +/- 11 years) and 62 presumably healthy controls (mean age 57.7 +/- 9 years) were enrolled in the study. Probing depth, periodontal pocket bleeding index (PPBI), plaque index, furcation involvements, and tooth mobility were evaluated to compare periodontal health in both groups. The subjects were also ranked according to a novel index of periodontitis severity, the periodontal index for risk of infectiousness (PIRI), aimed at quantifying the risk of release of proinflammatory mediators from the periodontal sites. RESULTS: Periodontitis was significantly more frequent in CAD patients than in controls (CAD patients: 91%; controls: 66%). The mean number of pockets was 18 +/- 17.1 in cardiac patients versus 7.6 +/- 12.7 in controls (P < 0.0001), despite the fact that the mean number of missing teeth was significantly greater in cases than in controls (14 +/- 7.1 versus 9 +/- 5.2; P < 0.0001). Furthermore, proportions of mobile teeth, bleeding sites, periodontal pockets, and involved furcations were significantly higher in CAD patients than in controls. In addition, the extent of the periodontal disease present was also greater in cases than in controls. A logistic model, adjusted for known cardiovascular risk factors, showed a strong association between CAD and periodontitis (odds ratio [OR] = 6.5). Moreover, there was a significant dose-response relationship between increasing scores of the periodontal risk of infectiousness and the presence of CAD (adjusted OR = 1.3 per PIRI unit). CONCLUSION: In the present study, periodontitis was revealed to be a significant risk factor for CAD after adjusting for other confounding factors, with the level of association increasing with the individual extent of the periodontal lesions.

Belgium↗

PubMed2006

[The study on the prevalence and associated risk factors of female urinary incontinence in Beijing women].

OBJECTIVES: To evaluate the prevalence, the clinical characteristics and associated risk factors of urinary incontinence in Beijing women. METHODS: In the cross-sectional study, 5300 Beijing women aged over 20 years were interviewed through a questionnaire including International Consultation on Incontinence Questionnaire Female Lower Urinary Tract Symptoms Module (ICIQ-FLUTS). Aged from 20 to 99 years old (mean 46 +/- 17). Divided into 8 groups every 10 years. 4812 cases are married. RESULTS: The overall prevalence of UI was 38.5% (2008/5221), among theses women, 22.9% (1197/5221), had stress incontinence, 2.8% (147/5221) urge incontinence and 12.4% (649/5221) mixed incontinence. In multiple logistic models, age, delivery mode, smoking, hypertension, constipation, body mass index, chronic pelvic pain (CPP) were relation factors for stress incontinence. CONCLUSIONS: The prevalence of urinary incontinence is high (38.5%). Urinary incontinence is major disease that affects female quality of life. 22.9% (1197/5221) of the incontinence women had stress urinary incontinence, 2.8% (147/5221) urge urinary incontinence and 12.4% (649/5221) mixed urinary incontinence. Old, vaginal delivery, smoking, hypertension, constipation, CPP and higher BMI are potential risk factors for stress urinary incontinence. Cesarean section are protective factors for stress urinary incontinence.

Adult↗

PubMed2006

Relation between severity of dysmenorrhea and endometrioma.

BACKGROUND: To evaluate the relationship between the severity of dysmenorrhea and endometrioma. METHODS: Descriptive study with prospective design. Two hundred and thirty-nine women with histologically proved endometriomas. The severity of dysmenorrhea was assessed prospectively with a 10-cm visual analog scale. Various indicators concerning the endometrioma and the extent of deep infiltrating endometriosis were recorded during surgery in 239 patients. Correlations were sought with a multiple regression logistic model. RESULTS: According to univariate analysis, the following variables were related to more severe dysmenorrhea: subperitoneal infiltration (uterosacral ligament and rectal infiltration) and R-AFS score of implants. None of the specific characteristics of endometriomas were associated with severe dysmenorrhea. After multiple regression analysis, rectal infiltration and R-AFS score of implants were the only factors that remained related to dysmenorrhea severity. CONCLUSIONS: When there is an endometrioma, severe dysmenorrhea is not directly related with the characteristics specific to these ovarian cysts. The associated deep infiltrating endometriotic lesions and in particular rectal infiltration could explain these symptoms.

Adult↗

PubMed2000

[Attitude of primary care physicians to early detection of prostatic cancer through prostate-specific antigen].

OBJECTIVES: To study the attitude of primary care doctors when a high (> or = 4 ng/ml) prostate-specific antigen (PSA) is found and to examine the variables linked to a prostate biopsy and the diagnosis of prostate cancer (PC). DESIGN: Descriptive, observational study. SETTING: Urban health district. PATIENTS: Ninety-four men not previously diagnosed with PC who in 1998 had a PSA figure > or = 4 ng/ml. The list was obtained from the pertinent laboratory. MEASUREMENTS: The following variables were gathered from review of clinical records: family background of PC, age, PSA figure, reason for request for PSA (if not given, it was considered a screening), referral to the urologist, rectal touch, transrectal echography, prostate biopsy and final diagnosis. RESULTS: Average age was 70 (SD, 9.31). The reason for requesting PSA was: urine symptoms in 25 (26.6%), other signs or symptoms in 25 (26.6%), request of patient in 2 cases (2.1%) and screening in 42 (44.7%). Rectal touch took place in 16 cases. Twenty-nine people were referred for examination to the urologist. 36 patients had an echography and biopsy. Variables linked to the prostate biopsy in the logistic model were: higher value of the PSA (OR 1.1; 95% CI, 1.03-1.18), being older (OR 0.92; CI, 0.87-0.98) and rectal touch performed (OR 3.58; CI, 1.02-12.51). Ten cases of PC were diagnosed. CONCLUSIONS: The most common reason for a PSA request was screening. Prostate biopsy was not requested for 58 men. A primary care guide to action concerning PC diagnosis in cases of PSA > or = 4 ng/ml would be useful.

Age Factors↗

PubMed2005

Tourette Syndrome and learning disabilities.

BACKGROUND: Tourette Syndrome (TS) is a neurodevelopmental disorder of childhood. Learning disabilities are frequently comorbid with TS. Using the largest sample of TS patients ever reported, we sought to identify differences between subjects with TS only and subjects with TS and a comorbid learning disability. METHODS: We used the Tourette Syndrome International Consortium database (TIC) to compare subjects with comorbid Tourette Syndrome and learning disabilities (TS + LD) to subjects who did not have a comorbid learning disability (TS-LD). The TIC database contained 5,500 subjects. We had usable data on 5,450 subjects. RESULTS: We found 1,235 subjects with TS + LD. Significant differences between the TS + LD group and the TS-LD group were found for gender (.001), age onset (.030), age first seen (.001), age at diagnosis (.001), prenatal problems (.001), sibling or other family member with tics (.024), two or more affected family members (.009), and severe tics (.046). We used logistic modeling to identify the optimal prediction model of group membership. This resulted in a five variable model with the epidemiologic performance characteristics of accuracy 65.2% (model correctly classified 4,406 of 5,450 subjects), sensitivity 66.1%, and specificity 62.2%. CONCLUSION: Subjects with TS have high prevalence rates of comorbid learning disabilities. We identified phenotype differences between the TS-LD group compared to TS + LD group. In the evaluation of subjects with TS, the presence of a learning disability should always be a consideration. ADHD may be an important comorbid condition in the diagnosis of LD or may also be a potential confounder. Further research on etiology, course and response to intervention for subjects with TS only and TS with learning disabilities is needed.

Attention Deficit Disorder with Hyperactivity↗

PubMed2000

[Risk factors of anastomotic leak following operations for gastric cancer in the elderly].

The risk factors of anastomotic leak in the elderly following operations for gastric cancer were evaluated by multiple logistic model analysis. Data were taken from 705 operations over a 14-year period. The mean age of patients was 75.8 +/- 7.6 years. The significant risk factors for anastomotic leak were amounts of intra-operative bleeding and male gender. No other factors were significant, including age, preoperative associated diseases, preoperative nutritional states and postoperative complications, some of which were, however, significant factors by univariate analyses. We conclude that we should make every endeavor to lower the amount of intra-operative bleeding in order to prevent postoperative anastomotic leaks in the elderly, especially in male patients.

Adult↗

PubMed2004

Did Finland's Tobacco Control Act of 1976 have an impact on ever smoking? An examination based on male and female cohort trends.

OBJECTIVE: To examine patterns of ever smoking among Finnish adults by gender and birth cohort from 1978 to 2001, with special emphasis on the possible effects of the 1976 Tobacco Control Act (TCA). METHODS: The data were derived from independent, annual cross sectional postal surveys among 15-64 year olds (n = 91,342), average response rate 75%. For the analyses 13 five year birth cohorts from 1916 to 1980 were constructed. Birth cohort variations in ever regular smoking were first examined graphically, and then logistic models were used to test the impact of the TCA. RESULTS: Among men there was a decrease in smoking from older to younger cohorts. For women an increase in smoking was observed between successive cohorts. A clear decline in the prevalence of ever smokers concurrent with the TCA was found among both men and women. CONCLUSIONS: The smoking behaviour trends across successive birth cohorts suggest the impact of tobacco policy in decreasing smoking initiation in youth. These findings thus support the acceptability and effectiveness of antismoking and smoke free policy measures in society.

Adolescent↗

PubMed2005

Risk factors for falling among community-dwelling veterans and their caregivers.

PURPOSE: To assess the prevalence and strength of association of risk factors for falling in Canadian veterans of World War II and Korea and their caregivers. METHODS: Questionnaires were sent to addresses of 3,000 Canadian veterans (response rate(1/4)70%). Risk factors for falls and the frequency of falls and injurious falls in the past 12 months were collected. RESULTS: Veterans had more risk factors than did caregivers, and more had fallen in the past year (39.8% vs. 29.7%). Risk factors in the logistic model for veterans included lower extremity disability (odds ratio 1.98, 95% confidence interval 1.5-2.6); lower extremity weakness (OR 1.75, 95% CI 1.3-2.3); worse memory than peers (OR 1.67, 95% CI 1.1-2.5); one or more visits to the family doctor in the past month (OR 1.53, 95% CI 1.2-2.0); and worse memory than 5 years ago (OR 1.36, 95% CI 1.0-1.8). CONCLUSIONS: Veterans appear more frail and prone to falling than their caregivers.

Accidental Falls↗

PubMed1999

Ocular predictors of the onset of juvenile myopia.

PURPOSE: The purpose of this study was to identify reliable predictors of the onset of juvenile myopia. METHODS: The data from 554 children enrolled in the Orinda Longitudinal Study of Myopia (OLSM) as nonmyopes with baseline data from the third grade were evaluated to develop a predictive profile for later onset of juvenile myopia. Myopia was defined as at least -0.75 D of myopia in the vertical and horizontal meridians of the right eye as measured by cycloplegic autorefraction (n = 45 children). Chosen predictors were refractive error and the ocular components: corneal power, Gullstrand crystalline lens power, and axial length. Sensitivity and specificity were calculated. Receiver operating characteristic (ROC) curves were generated to evaluate and compare these predictors singly and combined. RESULTS: Refractive error, axial length, Gullstrand lens and pod corneal power were all significant predictive factors for the onset of juvenile myopia. The best single predictor of future myopia onset in the right eye was the right eye's cycloplegic autorefraction spherical refractive error value (mean sphere across 10 readings) at baseline. For a cut point of less than +0.75 D hyperopia in the third grade, sensitivity was 86.7% and specificity was 73.3%. The area under the ROC curve for this mean sphere was 0.880. Producing a logistic model combining mean sphere, corneal power, Gullstrand lens power, and axial length results in a slight improvement in predictive ability (area under the ROC curve = 0.893). CONCLUSIONS: Onset of juvenile myopia can be predicted with moderate accuracy using the mean cycloplegic, spherical refractive error in the third grade. Measurement of other ocular components at this age improves predictive ability, albeit incrementally. Further improvements in the prediction of myopia onset will require the use of longitudinal data in addition to one-time measurement of refractive error and the ocular components.

Adolescent↗

PubMed1997

Comparing risk-adjustment methods for provider profiling.

Risk-adjustment and provider profiling have become common terms as the medical profession attempts to measure quality and assess value in health care. One of the areas of care most thoroughly developed in this regard is quality assessment for coronary artery bypass grafting (CABG). Because in-hospital mortality following CABG has been studied extensively, risk-adjustment mechanisms are already being used in this area for provider profiling. This study compares eight different risk-adjustment methods as applied to a CABG surgery population of 28 providers. Five of the methods use an external risk-adjustment algorithm developed in an independent population, while the other three rely on an internally developed logistic model. The purposes of this study are to: (i) create a common metric by which to display the results of these various risk-adjustment methodologies with regard to dichotomous outcomes such as in-hospital mortality, and (ii) to compare how these risk-adjustment methods quantify the 'outlier' standing of providers. Section 2 describes the data, the external and internal risk-adjustment algorithms, and eight approaches to provider profiling. Section 3 then demonstrates the results of applying these methods on a data set specifically collected for quality improvement.

Aged↗

PubMed2005

The effects of geography and spatial behavior on health care utilization among the residents of a rural region.

OBJECTIVE: This analysis determines the importance of geography and spatial behavior as predisposing and enabling factors in rural health care utilization, controlling for demographic, social, cultural, and health status factors. DATA SOURCES: A survey of 1,059 adults in 12 rural Appalachian North Carolina counties. STUDY DESIGN: This cross-sectional study used a three-stage sampling design stratified by county and ethnicity. Preliminary analysis of health services utilization compared weighted proportions of number of health care visits in the previous 12 months for regular check-up care, chronic care, and acute care across geographic, sociodemographic, cultural, and health variables. Multivariable logistic models identified independent correlates of health services utilization. DATA COLLECTION METHODS: Respondents answered standard survey questions. They located places in which they engaged health related and normal day-to-day activities; these data were entered into a geographic information system for analysis. PRINCIPAL FINDINGS: Several geographic and spatial behavior factors, including having a driver's license, use of provided rides, and distance for regular care, were significantly related to health care utilization for regular check-up and chronic care in the bivariate analysis. In the multivariate model, having a driver's license and distance for regular care remained significant, as did several predisposing (age, gender, ethnicity), enabling (household income), and need (physical and mental health measures, number of conditions). Geographic measures, as predisposing and enabling factors, were related to regular check-up and chronic care, but not to acute care visits. CONCLUSIONS: These results show the importance of geographic and spatial behavior factors in rural health care utilization. They also indicate continuing inequity in rural health care utilization that must be addressed in public policy.

Adult↗

PubMed1998

Study of the variations of the cesarean sections rate in the Rhône-Alpes region (France): effect of women and maternity service characteristics.

OBJECTIVE: To analyze the variations of the rate of prophylactic cesarean sections and cesarean sections during labor according to the women's and maternity services' characteristics. STUDY DESIGN: The study population was a sample of 8470 women having delivered in one of the 86 maternity services of the Rhône-Alpes region (France) in 1990. The women's characteristics were risk factors for pregnancy or delivery. The maternity services' characteristics included those describing the service structure and the mode of recruitment. For the statistical analysis we used a two-level logistic model. RESULTS: After taking into account the women's characteristics, there remained a variation between the maternity services. This variation was in part explained by the type of recruitment of the maternity service. When the proportion of women with a previous cesarean section or a dystocia increased in the maternity service the cesarean sections rate decreased in women presenting similar characteristics. CONCLUSION: Thus, the maternity services which receive many "at risk" women tend to have higher cesarean section rates than the others. However, in comparison with women presenting similar characteristics, these same maternity services tend to perform fewer cesarean sections.

Cesarean Section↗