Estimating Multilevel Models on Data Streams.

Social scientists are often faced with data that have a nested structure: pupils are nested within schools, employees are nested within companies, or repeated measurements are nested within individuals. Nested data are typically analyzed using multilevel models. However, when data sets are extremely large or when new data continuously augment the data set, estimating multilevel models can be challenging: the current algorithms used to fit multilevel models repeatedly revisit all data points and end up consuming much time and computer memory. This is especially troublesome when predictions are needed in real time and observations keep streaming in. We address this problem by introducing the Streaming Expectation Maximization Approximation (SEMA) algorithm for fitting multilevel models online (or "row-by-row"). In an extensive simulation study, we demonstrate the performance of SEMA compared to traditional methods of fitting multilevel models. Next, SEMA is used to analyze an empirical data stream. The accuracy of SEMA is competitive to current state-of-the-art methods while being orders of magnitude faster.

Insulin, androgen, and gonadotropin concentrations, body mass index, and waist to hip ratio in the first years after menarche in girls with regular menstrual cycles, irregular menstrual cycles, or oligomenorrhea.

Data on changes in hormone concentrations during the first years after menarche are scarce. We studied the relation between gynecological age (age minus age at menarche), hormone concentrations, and body measurements from the lst to the 6th yr after menarche in 229 observations of girls with regular menstrual cycles, 157 observations of girls with irregular menstrual cycles, and 104 observations of girls with oligomenorrhea. Body Mass Index, waist circumference, hip circumference, LH, androstenedione, testosterone, and dehydro-epiandrosterone sulphate increased significantly (linear regression, P < 0.05) by gynecological age in all menstrual cycle pattern groups. For PRL and estradiol a significant increase with gynecological age was only documented in the regular menstrual cycle group and for waist to hip ratio only in the irregular menstrual cycle group. No significant correlation could be documented between gynecological age and overnight fasting insulin concentrations or glucose to insulin ratio. We found no significant correlation between insulin concentrations or glucose to insulin ratio and androgen concentrations. Significant positive correlations were found between LH and androgens. LH and androgen levels increase during the first years after menarche, and reference values should be adjusted for gynecological age. In these years, no significant correlation between hyperinsulinemia and hyperandrogenemia could be documented.

Polycystic ovaries in adolescents and the relationship with menstrual cycle patterns, luteinizing hormone, androgens, and insulin.

STUDY OBJECTIVE: To evaluate the possible role of inappropriate LH secretion, hyperandrogenism, and hyperinsulinemia in the development of polycystic ovaries (PCO) and the polycystic ovary syndrome. DESIGN: Observational. SETTING: General population samples. PARTICIPANTS: 58 adolescents with regular menstrual cycles, 50 with irregular menstrual cycles, and 29 with oligomenorrhea (age 16.7+/-0.9 years). INTERVENTIONS: Transabdominal pelvic ultrasonography and vena puncture. MAIN OUTCOME MEASURES: PCO; LH, androstenedione, and testosterone levels; overnight fasting insulin concentrations; and oligomenorrhea. RESULTS: The prevalence of PCO increased significantly with the irregularity of the menstrual cycle pattern, as illustrated by the study, finding PCO in 9% of the girls with regular menstrual cycles, 28% of those with irregular menstrual cycles, and 45% of oligomenorrheic girls. The LH and androgen concentrations were significantly higher in girls with PCO; the insulin levels and the glucose-insulin ratio did not differ when the girls with PCO were compared with girls with normal ovaries. Oligomenorrheic girls with PCO had the highest androgen and LH concentrations; their insulin concentrations and glucose-insulin ratio were in the same range as girls with regular menstrual cycles and normal ovaries; and both their hip and waist girths were wider, although their waist-hip ratio was normal. CONCLUSIONS: PCO in adolescents is associated with irregular menstrual cycles, oligomenorrhea, and/or high androgen and LH levels; but no relationship was found with the insulin level or glucose-insulin ratio. Thus, it is doubtful that hyperinsulinemia is an important factor in the development of PCO or polycystic ovary syndrome.

[Use of the pill by adolescents for contraception or as therapy]. / Pilgebruik door adolescenten voor anticonceptie of als therapie.

OBJECTIVE: To describe frequency of oral contraceptive (OAC) use, indications for OAC use and influence of calender age, gynaecological age (age minus age at menarche) and level of education on OAC use among ninth grade secondary school girls. SETTING: Secondary schools in Amstelland-de Meerlanden, a combined urban and rural region to the south of Amsterdam. METHOD: By questionnaire and individual interview on menstrual cycle pattern and OAC use, school girls were asked about duration and reasons for OAC use and what preparation they used. RESULTS: The response to the questionnaire was 92%, that to the interview 83%. Of the 2248 responders 248 (11%; mean age 15.3 years (SD: 0.6)) used OAC: 74% used low dose 'sub 50' preparations, 3% pills with 50 micrograms oestrogen, 3% 3-phase preparations and 17% pills with antiandrogens. Of girls aged 14, 15 and 16 years 4%, 12% and 28%, respectively, used OAC. Of the 15-year-olds 36% mentioned contraception as most important reason, 20% menstrual cycle irregularity, 28% dysmenorrhoea, 12% acne and 5% other reasons. With multiple logistic regression analysis age, gynaecological age and level of education were independent variables for OAC use in general and for OAC use for contraception or dysmenorrhoea, but less so for OAC use for menstrual cycle irregularity or acne. CONCLUSION: During adolescence low dose OACs are frequently used. Of adolescent OAC users 36% mentioned contraception as most important reason. OAC use in the Netherlands among girls aged 15 and 16 years doubled in comparison with 1982. Gynaecological age and calender age (determinants of biological maturation), and level of education (a determinant of life style) were associated with OAC use.

PIP: The objective was to describe frequency of oral contraceptive (OC) use, indications for OC use and influence of calendar age, gynecological age (age minus age at menarche), and level of education on OC use among 9th grade secondary school girls. Secondary schools of 4 different types in Amstelland-de Meerlanden, a combined urban and rural region to the south of Amsterdam, were surveyed during the 1990-91 school year by questionnaire and individual interview on menstrual cycle pattern and OC use, and school girls were asked about the duration and reasons for OC use and what preparation they used. The response rate was 92% to the questionnaire and 83% to the interview. Of the 2248 respondents, 248 (11%; mean age 15.3 years) used OCs: 74% used low-dose sub-50 mcg preparations, 3% used pills with 50 mcg of estrogen, 3% used 3-phase preparations, and 17% used pills with antiandrogens. 4%, 12%, 28%, and 42%, respectively, of girls 14, 15, 16, and 17 years old used OCs. 36% of the 15-year-old girls mentioned contraception as the most important reason, 20% menstrual cycle irregularity, 28% dysmenorrhea, 12% acne, and 5% other reasons. Among 14-, 15-, and 16-year old girls, contraception was cited as the most important reason for OC use by 4%, 31%, and 36%, respectively. 39 girls used OCs for acne, of whom 34 used a preparation containing cyproterone acetate. Multiple logistic regression analysis indicated age, gynecological age, and level of education as independent variables for OC use in general and for OC use for contraception or dysmenorrhea, but less so for OC use for menstrual cycle irregularity or acne. In the model of OC use for contraception, age was the most important determinant, while in the model of OC use for dysmenorrhea it was the gynecological age. OC use for menstrual cycle control was associated significantly with age and gynecological age, but not with school type (p=0.07). In the first year after menarche not a single girl used OCs for menstrual cycle regulation.

[The atrophic maxilla. Rehabilitation by bone reconstruction, implants and permanent or removable dentures]. / De geresorbeerde bovenkaak. Rehabilitatie met behulp van botreconstructie, implantaten en vaste of uitneembare prothetische voorzieningen.

Maxillary bone reconstruction in combination with placement of endosseous dental implants is a new treatment modality for patients suffering from maxillary atrophy and difficulties with wearing removable prosthesis. Among 88 patients the experiences with and their satisfaction at parts of the treatment were investigated. Patients could express their appreciation in a 1-5 scale (1 = bad, 5 = good). The satisfaction at the total treatment was given the mean figure 4.1 +/- 0.9. Ninety percent of the patients would go through the treatment again when necessary.

Survival of cylindrical implants in composite grafted maxillary sinuses.

PURPOSE: This retrospective study investigated the survival of dental implants placed in the maxilla after composite grafting of the sinus and an average of 55 months of loading. PATIENTS AND METHODS: Maxillary sinuses of 88 patients were grafted with autogenous cancellous bone combined with dense hydroxyapatite particles. After an average healing period of 3.4 months, hydroxyapatite-coated titanium endosseous implants were placed. A total of 388 implants were placed in grafted sinus floors, and 82 were placed in onlay grafted nonsinus position in the canine region. The implants were loaded with overdentures and fixed bridges 4 months (mean) after implantation, with a follow-up for a mean of 55 months. RESULTS: The cumulative implant survival was calculated according to the Kaplan-Meier method. Implant survival from the time of loading was 89% in full reconstructed cases and 90% in partially edentulous cases. The overall cumulative implant survival rate, including the loss in the surgical stage, was 82%. CONCLUSION: Implant loss in composite grafted maxillae after 70 months of follow-up was similar to loss in nongrafted maxillae.

Peri-implant tissue health in reconstructed atrophic maxillae--report of 88 patients and 470 implants.

The aim of this study was to evaluate the soft-tissue health around 470 implants placed in the upper jaw of 88 patients with severe maxillary atrophy. All patients underwent a reconstruction procedure which included a composite bone grafting from the iliac crest to the maxillary sinus. Evaluation was performed at a mean implant lifetime of 21.6+/-10.9 months. The following parameters were used: probing depth, plaque-, gingival-, bleeding index and width of the buccal keratinized mucosa. They were calculated as means for each implant and related to the type of superstructure and the thickness of the original sinus floor. (higher or lower than 5 mm). A high percentage of bleeding on probing was observed in obviously clinically healthy peri-implant pockets. Bleeding on probing and deeper probing depth were especially observed in the group with severe maxillary atrophy (less than 5 mm initial height of the sinus floor.) There was a significant increase in mean peri-implant probing depth and gingiva index in overdentures versus fixed bridges. No differences were observed with respect to the width of the keratinized buccal mucosa. It was concluded that implants in the reconstructed maxilla and supporting overdentures had a higher risk for bone loss, based on the worse peri-implant tissue health observed.

A clinical evaluation of 76 implant-supported superstructures in the composite grafted maxilla.

The aim of the study was to evaluate the material behaviour and nature of implant-supported superstructures in function. A total of 37 overdentures, 26 fixed partial and 13 fixed full dentures with a mean lifetime of 40 months were clinically evaluated in relation to damage, occlusion and articulation, parafunction, retentive devices, hygiene and speech problems. In general 70% of all superstructures were damaged. Fixed partial dentures were significantly less damaged than overdentures and fixed full dentures (P<0.001). A significant high percentage of newly gained parafunction was found (P<0.05).

Dental implants in the atrophic maxilla: measurements of patients' satisfaction and treatment experience.

Maxillary bone reconstruction in combination with placement of dental implants is a treatment modality reported since 1980 for patients suffering from maxillary atrophy and difficulties with wearing removable prosthesis. This study reports on patient experience and satisfaction among 88 consecutive patients concerning surgical and prosthetic aspects of their treatment. A questionnaire was used to gather the data. Patient satisfaction was measured on a scale from 1 to 5 (1 = bad/few, 5 = very good/much). The satisfaction index with regard to the total treatment was 4.1 +/- 0.9. A total of 90% of the patients responded that they would go through the treatment again or recommend the treatment to a friend if necessary. The post-operative pain index scored 2.6 +/- 1.2. The acceptation of the post-operative pain at the crista iliaca was 86.1% and of the maxilla 89.9%. It is concluded that the results of measurements of patient satisfaction after maxillary bone reconstruction in combination with placement of dental implants were acceptable.

The use of oral contraceptives by adolescents for contraception, menstrual cycle problems or acne.

BACKGROUND: Oral contraceptives are prescribed as contraception but also as therapy for menstrual cycle disturbances and acne. We studied the prevalence of oral contraceptive (OC) use and the indications to start OC use among adolescents. METHODS: A cohort consisting of ninth grade secondary school girls (mean age 15.3+/-0.6 (s.d.) years) answered a questionnaire on their menstrual cycle. OC users were asked about duration and reasons for OC-use and the name of the preparation they used. The influence of calendar age, gynecological age and level of education on the prevalence of OC was studied by multiple logistic regression analysis. RESULTS: The response on the questionnaire was 92%. Of 2248 responders 248 (11%) used oral contraceptives: 74% used low dose 'sub 50' preparations, 3% pills with 50 microg estrogen, 3% tri-phase preparations and 17% pills with antiandrogens. Of girls aged 14, 15 and 16 years 4%, 12% and 28% respectively, used OC. Of the 15-year-olds 31% mentioned contraception as most important reason for OC use, 18% menstrual cycle irregularity, 26% dysmenorrhea, 10% acne and 5% other reasons. Calendar age, gynecological age and level of education were independent variables for OC use in general and for OC use for contraception or dysmenorrhea, but less so for OC use for menstrual cycle irregularity or acne. CONCLUSIONS: During adolescence low dose OC's were frequently used. In The Netherlands OC use among girls aged 15 and 16 years doubled in comparison with 1982. One third of the adolescent OC-users mentioned contraception as most important reason to start OC. Gynecological age (a determinant of biological maturation), calendar age (a determinant of biological maturation and lifestyle in peer groups), and level of education (a determinant of lifestyle in peer groups) were associated with OC use.

Relationship of the menstrual cycle pattern in 14-17 year old old adolescents with gynaecological age, body mass index and historical parameters.

In a cross-sectional population-based study the association between the menstrual pattern in ninth grade schoolgirls and calendar age, gynaecological age, body mass index (BMI) and historical parameters was investigated. The survey was held in a combined urban and rural region, south of Amsterdam. A total of 2480 adolescents, mean age 15.3 +/- 0.6 (SD) years, answered a questionnaire: response 92%. The menstrual cycle patterns were categorized to regular menstrual cycles (RMC), irregular menstrual cycles (IMC), oligomenorrhoea, polymenorrhoea, pre-menarche, <6 months after menarche, and oral contraceptive use. Gynaecological age was strongly associated with the prevalence of IMC but only weakly with the prevalence of oligomenorrhoea. In a logistic regression analysis gynaecological age, subjective acne and intellectual performance were independently associated with oligomenorrhoea. Gynaecological age, low BMI, chronic non-specific lung disease (CNSLD) or allergic disease, stress and strain, weight loss of >5 kg were independently associated with IMC. More than 8 h sports per week was associated with not having experienced menarche in the ninth grade but not with menstrual cycle disturbances. The association between CNSLD or allergic disease and IMC has not previously been described. The associations between weight loss, low body weight, stress, physical exercise or signs of hyperandrogenism and menstrual cycle patterns in adolescents are weak when studied on a population basis. The value of these parameters to explain abnormal menstrual cycle patterns is limited.

Endocrine features of polycystic ovary syndrome in a random population sample of 14-16 year old adolescents.

Hospital based studies have shown that oligomenorrhoeic adolescents have high luteinizing hormone (LH) and androgen concentrations, endocrine signs of polycystic ovary syndrome (PCOS). The prevalence of these abnormalities in an unselected population of adolescents is not known. We determined LH, follicle stimulating hormone (FSH), androstenedione, testosterone, dehydroepiandrosterone sulphate (DHEAS), oestradiol and prolactin concentrations in unselected population samples of adolescents with oligomenorrhoea, secondary amenorrhoea and regular menstrual cycles. A total of 2248 white, west European adolescents, aged 15.3 +/- 0.6 (mean +/- SD) years, participated. Blood was taken from 107 adolescents with regular menstrual cycles, 52 with oligomenorrhoea and four with secondary amenorrhoea. Oligomenorrhoeic adolescents had higher mean LH, androstenedione, testosterone, DHEAS and oestradiol concentrations compared with girls with regular menstrual cycles; 57% of the oligomenorrhoeic girls had LH or androgen concentrations above the 95th centile of adolescents with regular menstrual cycles. None of the 52 oligomenorrhoeic girls and only one of four girls with secondary amenorrhoea had a hypogonadotrophic endocrine pattern. The present study and available literature support the view that oligomenorrhoea in adolescents is not a stage in the physiological maturation of the hypothalamic pituitary-ovarian axis but an early sign of PCOS associated with subfertility. Physicians should consider endocrine evaluation before reassuring oligomenorrhoeic girls or prescribing oral contraceptives to these girls.

Predictive value of menstrual cycle pattern, body mass index, hormone levels and polycystic ovaries at age 15 years for oligo-amenorrhoea at age 18 years.

BACKGROUND: On the question of how to counsel adolescents with irregular menstrual cycles or oligomenorrhoea, no clear answer has been given. Adolescents with oligomenorrhoea especially show endocrine abnormalities and may be at risk for ovulatory dysfunction and the polycystic ovary syndrome in adulthood. METHODS: We followed a cohort of adolescents to document changes in menstrual cycle pattern between ages 15 and 18 years in the general population. RESULTS: Two per cent (2/128) of adolescents with regular menstrual cycles developed oligomenorrhoea, and 12% (17/148) of those with irregular menstrual cycles did so. Fifty-one per cent (34/67) of the oligomenorrhoeic adolescents remained oligomenorrhoeic. Increase in body mass index (BMI), concentration of LH, androstenedione or testosterone, and polycystic ovaries (PCO) were associated with persistence of oligomenorrhoea. In multivariate analysis only a normal to high BMI (>19.6 kg/m(2)) consistently contributed significantly to predict persistent oligomenorrhoea. Glucose:insulin ratio as a marker for insulin resistance was not associated with an increased risk for oligomenorrhoea. CONCLUSIONS: Oligomenorrhoea at age 18 years is better predicted by menstrual cycle pattern at age 15 years than by LH or androgen concentrations or PCO at this age. Not only obese, but also normal weight oligomenorrhoeic, adolescents have a high risk of remaining oligomenorrhoeic.