Choice of baseline hazards in joint modeling of longitudinal and time-to-event cancer survival data.

Longitudinal time-to-event analysis is a statistical method to analyze data where covariates are measured repeatedly. In survival studies, the risk for an event is estimated using Cox-proportional hazard model or extended Cox-model for exogenous time-dependent covariates. However, these models are inappropriate for endogenous time-dependent covariates like longitudinally measured biomarkers, Carcinoembryonic Antigen (CEA). Joint models that can simultaneously model the longitudinal covariates and time-to-event data have been proposed as an alternative. The present study highlights the importance of choosing the baseline hazards to get more accurate risk estimation. The study used colon cancer patient data to illustrate and compare four different joint models which differs based on the choice of baseline hazards [piecewise-constant Gauss-Hermite (GH), piecewise-constant pseudo-adaptive GH, Weibull Accelerated Failure time model with GH & B-spline GH]. We conducted simulation study to assess the model consistency with varying sample size (N = 100, 250, 500) and censoring (20 %, 50 %, 70 %) proportions. In colon cancer patient data, based on Akaike information criteria (AIC) and Bayesian information criteria (BIC), piecewise-constant pseudo-adaptive GH was found to be the best fitted model. Despite differences in model fit, the hazards obtained from the four models were similar. The study identified composite stage as a prognostic factor for time-to-event and the longitudinal outcome, CEA as a dynamic predictor for overall survival in colon cancer patients. Based on the simulation study Piecewise-PH-aGH was found to be the best model with least AIC and BIC values, and highest coverage probability(CP). While the Bias, and RMSE for all the models showed a competitive performance. However, Piecewise-PH-aGH has shown least bias and RMSE in most of the combinations and has taken the shortest computation time, which shows its computational efficiency. This study is the first of its kind to discuss on the choice of baseline hazards.

Reproductive factors and thyroid cancer risk: Meta-analysis.

BACKGROUND: Assessed pooled risk on reproductive factors and oral contraceptives (OC) on thyroid cancer (TC) using published studies (1996-2017). METHODS: Summary odds ratio (OR) for case-control studies (n = 10) and risk ratio (RR) for cohort studies (n = 9) was done. RESULTS: OR was 1.43 (95% CI: 1.16-1.77) for age at menarche >14 years, 1.49 (95% CI: 1.19-1.86) for parity >2, 1.38 (95% CI: 1.18-1.61) for miscarriage/abortion, and 2.05 (95% CI: 1.39-3.01) for artificial menopause. A protective effect (ORs: 0.85; 95% CI: 0.72-0.99) on TC was observed for prolonged use of OCs. RR was 1.17 (95% CI: 0.90-1.57) for age at menarche >14 years, 1.10 (95% CI: 0.94-1.27) for parity >2, 1.20 (95% CI: 1.03-1.40) for miscarriage/abortion, and 2.16 (95% CI: 1.41-3.31) for artificial menopause and protective effect (RR: 0.78; 95% CI: 0.65-0.92) for prolonged use of OCs. CONCLUSIONS: This meta-analysis supports an association due to changes in female hormones during menstrual cycle and pregnancy with the risk of TC and explains female preponderance.

Development of a field-friendly automated dietary assessment tool and nutrient database for India.

Studies of diet and disease risk in India and among other Asian-Indian populations are hindered by the need for a comprehensive dietary assessment tool to capture data on the wide variety of food and nutrient intakes across different regions and ethnic groups. The nutritional component of the India Health Study, a multicentre pilot cohort study, included 3908 men and women, aged 35-69 years, residing in three regions of India (New Delhi in the north, Mumbai in the west and Trivandrum in the south). We developed a computer-based, interviewer-administered dietary assessment software known as the 'NINA-DISH (New Interactive Nutrition Assistant - Diet in India Study of Health)', which consisted of four sections: (1) a diet history questionnaire with defined questions on frequency and portion size; (2) an open-ended section for each mealtime; (3) a food-preparer questionnaire; (4) a 24 h dietary recall. Using the preferred meal-based approach, frequency of intake and portion size were recorded and linked to a nutrient database that we developed and modified from a set of existing international databases containing data on Indian foods and recipes. The NINA-DISH software was designed to be easily adaptable and was well accepted by the interviewers and participants in the field. A predominant three-meal eating pattern emerged; however, patterns in the number of foods reported and the primary contributors to macro- and micronutrient intakes differed by region and demographic factors. The newly developed NINA-DISH software provides a much-needed tool for measuring diet and nutrient profiles across the diverse populations of India with the potential for application in other South Asian populations living throughout the world.

Assessment of follow-up, and the completeness and accuracy of cancer case ascertainment in three areas of India.

BACKGROUND: A prospective study of diet and cancer has not been conducted in India; consequently, little is known regarding follow-up rates or the completeness and accuracy of cancer case ascertainment. METHODS: We assessed follow-up in the India Health Study (IHS; 4671 participants aged 35-69 residing in New Delhi, Mumbai, or Trivandrum). We evaluated the impact of medical care access and relocation, re-contacted the IHS participants to estimate follow-up rates, and conducted separate studies of cancer cases to evaluate registry coverage (604 cases in Trivandrum) and the accuracy of self- and proxy-reporting (1600 cases in New Delhi and Trivandrum). RESULTS: Over 97% of people reported seeing a doctor and 85% had lived in their current residence for over six years. The 2-year follow-up rate was 91% for Trivandrum and 53% for New Delhi. No cancer cases were missed among public institutions participating in the surveillance program in Trivandrum during 2003-2004; but there are likely to be unmatched cases (ranging from 5 to 13% of total cases) from private hospitals in the Trivandrum registry, as there are no mandatory reporting requirements. Vital status was obtained for 36% of cancer cases in New Delhi as compared to 78% in Trivandrum after a period of 4 years. CONCLUSIONS: A prospective cohort study of cancer may be feasible in some centers in India with active follow-up to supplement registry data. Inclusion of cancers diagnosed at private institutions, unique identifiers for individuals, and computerized medical information would likely improve cancer registries.

Multi-center feasibility study evaluating recruitment, variability in risk factors and biomarkers for a diet and cancer cohort in India.

BACKGROUND: India's population exhibits diverse dietary habits and chronic disease patterns. Nutritional epidemiologic studies in India are primarily of cross-sectional or case-control design and subject to biases, including differential recall of past diet. The aim of this feasibility study was to evaluate whether a diet-focused cohort study of cancer could be established in India, providing insight into potentially unique diet and lifestyle exposures. METHODS: Field staff contacted 7,064 households within three regions of India (New Delhi, Mumbai, and Trivandrum) and found 4,671 eligible adults aged 35-69 years. Participants completed interviewer-administered questionnaires (demographic, diet history, physical activity, medical/reproductive history, tobacco/alcohol use, and occupational history), and staff collected biological samples (blood, urine, and toenail clippings), anthropometric measurements (weight, standing and sitting height; waist, hip, and thigh circumference; triceps, sub-scapula and supra-patella skin fold), and blood pressure measurements. RESULTS: Eighty-eight percent of eligible subjects completed all questionnaires and 67% provided biological samples. Unique protein sources by region were fish in Trivandrum, dairy in New Delhi, and pulses (legumes) in Mumbai. Consumption of meat, alcohol, fast food, and soft drinks was scarce in all three regions. A large percentage of the participants were centrally obese and had elevated blood glucose levels. New Delhi participants were also the least physically active and had elevated lipids levels, suggesting a high prevalence of metabolic syndrome. CONCLUSIONS: A high percentage of participants complied with study procedures including biological sample collection. Epidemiologic expertise and sufficient infrastructure exists at these three sites in India to successfully carry out a modest sized population-based study; however, we identified some potential problems in conducting a cohort study, such as limited number of facilities to handle biological samples.

HbA(1c) values for defining diabetes and impaired fasting glucose in Asian Indians.

AIM: To determine the glycosylated haemoglobin (HbA(1c)) cut-points for diabetes and impaired fasting glucose (IFG) among Asian Indians. METHODS: Participants (n=525) were a random sample selected from the India Health Study. Based on history and fasting plasma glucose (FPG), participants were classified into known diabetes, newly diagnosed diabetes (NDD), impaired fasting glucose (IFG) [ADA and WHO criteria] or normal fasting glucose (NFG). Receiver Operating Characteristic curves were used to identify the optimum sensitivity and specificity for defining HbA(1c) cut-points for NDD and IFG against the FPG criteria. RESULTS: There were 64 participants with a known history of diabetes. Of the remaining 461, IFG was present in 44.7% (ADA) and 18.2% (WHO), and 10.4% were NDD. Mean HbA(1c) were 5.4 (±0.04)% for NFG; 5.7 (±0.06)% among IFG-ADA, 5.8 (±0.09)% among IFG-WHO; 7.5 (±0.33)% for NDD and 8.4 (±0.32)% for known diabetes. Optimal HbA(1c) cut-point for NDD was 5.8% (sensitivity=75%, specificity=75.5%, AUC=0.819). Cut-point for IFG (ADA) was 5.5% (sensitivity=59.7%, specificity=59.9%, AUC=0.628) and for IFG (WHO) was 5.6% (sensitivity=60.7%, specificity=65.1%, AUC=0.671). CONCLUSION: In this study population from north and south regions of India, the HbA(1c) cut-point that defines NDD (≥5.8%) was much lower than that proposed by an international expert committee and the American Diabetes Association (≥6.5%). A cut-point of ≥5.5% or ≥5.6% defined IFG, and was slightly lower than the ≥5.7% for high risk proposed, but accuracy was less than 70%.

A cross-sectional investigation of regional patterns of diet and cardio-metabolic risk in India.

BACKGROUND: The role of diet in India's rapidly progressing chronic disease epidemic is unclear; moreover, diet may vary considerably across North-South regions. METHODS: The India Health Study was a multicenter study of men and women aged 35-69, who provided diet, lifestyle, and medical histories, as well as blood pressure, fasting blood, urine, and anthropometric measurements. In each region (Delhi, n=824; Mumbai, n=743; Trivandrum, n=2,247), we identified two dietary patterns with factor analysis. In multiple logistic regression models adjusted for age, gender, education, income, marital status, religion, physical activity, tobacco, alcohol, and total energy intake, we investigated associations between regional dietary patterns and abdominal adiposity, hypertension, diabetes, and dyslipidemia. RESULTS: Across the regions, more than 80% of the participants met the criteria for abdominal adiposity and 10 to 28% of participants were considered diabetic. In Delhi, the "fruit and dairy" dietary pattern was positively associated with abdominal adiposity [highest versus lowest tertile, multivariate-adjusted OR and 95% CI: 2.32 (1.03-5.23); Ptrend=0.008] and hypertension [2.20 (1.47-3.31); Ptrend<0.0001]. In Trivandrum, the "pulses and rice" pattern was inversely related to diabetes [0.70 (0.51-0.95); Ptrend=0.03] and the "snacks and sweets" pattern was positively associated with abdominal adiposity [2.05 (1.34-3.14); Ptrend=0.03]. In Mumbai, the "fruit and vegetable" pattern was inversely associated with hypertension [0.63 (0.40-0.99); Ptrend=0.05] and the "snack and meat" pattern appeared to be positively associated with abdominal adiposity. CONCLUSIONS: Cardio-metabolic risk factors were highly prevalent in this population. Across all regions, we found little evidence of a Westernized diet; however, dietary patterns characterized by animal products, fried snacks, or sweets appeared to be positively associated with abdominal adiposity. Conversely, more traditional diets in the Southern regions were inversely related to diabetes and hypertension. Continued investigation of diet, as well as other environmental and biological factors, will be needed to better understand the risk profile in this population and potential means of prevention.

Measurement of spices and seasonings in India: opportunities for cancer epidemiology and prevention.

Bioactive components of many foods added during cooking have potential antioxidant, anti-inflammatory, antimicrobial, antibacterial and chemopreventive properties. However, epidemiologic studies generally do not collect detailed information on these items, which include spices, chilies, coconuts, garlic, onions, and oils. Since India has some of the highest spice consumption in the world, we developed a computer-based food preparer questionnaire to estimate per capita consumption of 19 spices, chilies, coconuts, garlic, onions, and 13 cooking oils among 3,625 participants in the India Health Study, a multicenter pilot study in three regions of India. We observed notable regional differences in consumption of spices, chilies, coconut, garlic, and onions. In Trivandrum, over 95 percent of the participants consumed 12 different spices, while in New Delhi and Mumbai, 95 percent of participants consumed only four and five spices, respectively. Cooking oil use also varied, as ghee was most common in New Delhi (96.8%) followed by mustard seed oil (78.0%), while in Trivandrum the primary oil was coconut (88.5%) and in Mumbai it was peanut (68.5%). There was some variation in consumption by education, income, and religion. Using a novel method for assessing food items primarly added during cooking, we successfully estimated per capita consumption within an epidemiologic study. Based on basic science research and suggestive ecologic level data on cancer incidence and spice consumption, improving epidemiologic assessment of these potentially chemopreventive food items may enhance our understanding of diet and cancer risk.