Use of Effect Size Measures along with p-Value in Scientific Publications

Background: To describe various measures for estimation of effect size, how it can be calculated and the scenarios in which each measures of effect size can be applied. Methods: The researchers can display the effect size measures in research articles which evaluate the difference between the means of continuous variables in different groups or the difference in proportions of outcomes in different groups of individuals. When p-value alone is displayed in a research article, without mentioning the effect size, reader may not get the correct pictures regarding the effect or role of independent variable on the outcome variable. Results: Effect size is a statistical concept that measures the actual difference between the groups or the strength of the relationship between two variables on a numeric scale. Conclusion: Effect size measures in scientific publications can communicate the actual difference between groups or the estimate of association between the variables, not just if the association or difference is statistically significant. The researchers can make their findings more interpretable, by displaying a suitable measure of effect size. Effect size measure can help the researchers to do meta-analysis by combining the data from multiple research articles.

36 Effect of Blending and Tuning Relationship Matrices in Single-step Genomic BLUP

Single-step GBLUP (ssGBLUP) relies on the combination of genomic (G) and pedigree relationships for all (A) and genotyped animals (A22). The procedure implemented in the BLUPF90 software suite first involves combining a small percentage of A22 into G (blending) to avoid singularity problems, then an adjustment to account for the fact the genetic base in G and A22 is different (tuning). However, blending before tuning may not reflect the actual difference between pedigree and genomic base because the blended matrix already contains a portion of A22. The objective of this study was to evaluate the impact of tuning before blending on predictivity, bias, and inflation of GEBV, indirect predictions (IP), and SNP effects from ssGBLUP using American Angus and US Holstein data. We used four different scenarios to obtain genomic predictions: BlendFirst_TunedG2, TuneFirst_TunedG2, BlendFirst_TunedG4, and TuneFirst_TunedG4. TunedG2 adjusts mean diagonals and off-diagonals of G to be similar to the ones in A22, whereas TunedG4 adjusts based on the fixation index. Over 6 million growth records were available for Angus and 5.9 million udder depth records for Holsteins. Genomic information was available on 51,478 Angus and 105,116 Holstein animals. Predictivity and reliability were obtained for 19,056 and 1,711 validation Angus and Holsteins, respectively. We observed the same predictivity and reliability for GEBV or IP in all four scenarios, ranging from 0.47 to 0.60 for Angus and was 0.67 for Holsteins. Slightly less bias was observed when tuning was done before blending. Correlation of SNP effects between scenarios was > 0.99. Refined tuning before blending had no impact on GEBV and marginally reduced the bias. This option will be implemented in the BLUPF90 software suite.

A Simple, Rapid, and Inexpensive Color-Based Hemoglobin Assay As a Robust Screening Test for Severe Anemia in Limited Resource Settings

Severe anemia is common in limited-resource settings (LRS) of sub-Saharan Africa, affecting up to 30% of hospitalized patients and associated with high mortality rates. The World Health Organization (WHO) recommends blood transfusion for patients with hemoglobin (Hb) concentration <4 g/dL in all cases, and for patients with Hb <6 g/dL if there are other clinical complications. In reality, a Hb ≤5 g/dL is roughly used as a threshold for transfusion in many limited-resource settings. The accurate diagnosis of anemia is critical to appropriately utilize the very limited blood supply, and to triage and treat patients in a timely manner to provide life-saving care. Lack of timely transfusions for children with severe anemia has been associated with significant increase in mortality in several studies within Africa. However, accurate diagnosis of anemia is often difficult or impossible in LRS due to lack of equipment, inadequate replenishment of reagents, poor maintenance or calibration of existing equipment, inadequately trained laboratory personnel, or unreliable and inconsistent access to electricity. A recent publication (Opoka RO et al. BMC Health Serv Res2018) examining transfusion practices in Uganda demonstrated that over half of blood transfusions were administered inappropriately, either without a Hb measurement or with Hb values for which transfusion would not be recommended. The WHO Haemoglobin Colour Scale (HCS), a simple, rapid, point-of-care (POC) test requiring a single drop of blood on filter paper, was developed in an effort to improve the accurate diagnosis of anemia in settings where more definitive laboratory testing is not available. There have been many published reports of the HCS with heterogeneous results, including a recent meta-analysis demonstrating significant variation in sensitivity for the detection of anemia and severe anemia, defined as Hb ≤11 g/dL and Hb ≤7 g/dL respectively (36-76%, Marn H. Lancet Glob Health 2016). The lower limit of the HCS is 4 g/dL, and there is very limited data on its accuracy to diagnose severe anemia requiring transfusion (Hb ≤5 g/dL). The AnemoCheckTMis a rapid, inexpensive, color-based POC test initially designed to diagnose anemia in higher-resource settings, and subsequently modified for use in limited-resource settings (Tyburski et al. J Clin Invest 2014). Based on mixed results both in controlled settings within the US and in the field in Tanzania, AnemoCheck was adapted to focus on its ability to detect severe anemia warranting transfusion. The newly modified AnemoCheck-LRS test allows identification of Hb ranging from 2-8 g/dL, with sharp color changes especially differentiable at the clinically relevant 5 g/dL threshold (Figure 1) to allow accurate diagnosis of clinically meaningful severe anemia in settings where this is common. We prepared and tested 200 samples of varying Hb concentration, each of which was evaluated independently by three separate readers, for a total of 1200 evaluations. By standard CBC measurement, the 200 samples had a mean (±SD) Hbof 5.2±1.9 g/dL (range 1.7 to 10.1 g/dL). Of the 570 samples with Hb ≤8 g/dL (upper limit of AnemoCheck), the absolute difference between AnemoCheck and the gold standard CBC Hb was 0.5±0.4 g/dL (actual difference 0.2±0.7 g/dL). The WHO HCS was inaccurate, with Hb readings consistently higher than the CBC Hb, and an absolute and actual difference of 4.9±1.3 g/dL for samples within the detection range of the HCS (4-14 g/dL, n=405). Table 1 summarizes the ability of both tests to diagnose severe anemia (Hb ≤5 g/dL), demonstrating the accuracy of AnemoCheck-LRS. Furthermore, given that most readings that missed the diagnosis of anemia were within 1 g/dL of actual Hb, if 6 g/dL was used as the screening threshold to identify patients who need a more definitive diagnosis, the sensitivity of AnemoCheck increased to 99%. This study demonstrates that the new AnemoCheck-LRS assay is an accurate POC test to diagnosis very severe anemia in a controlled laboratory setting. The widely cited WHO HCS performed poorly, with a stark inability to diagnose very severe anemia in most cases. This study provides important preliminary data to support follow-up field studies that will use AnemoCheck-LRS to diagnose severe anemia, with the goal of the assay serving as a primary screening test for rapid and accurate diagnosis in settings where laboratory capacity and resources are limited. Disclosures Tyburski: Sanguina, LLC: Employment, Equity Ownership. Ware:Bristol Myers Squibb: Research Funding; Nova Laboratories: Consultancy; Biomedomics: Research Funding; Global Blood Therapeutics: Other: advisory board; Addmedica: Research Funding; Agios: Other: advisory board; Novartis: Membership on an entity's Board of Directors or advisory committees.

Do weak stationary magnetic fields affect the perceived astringency of red wine?

To investigate claims that products containing weak stationary magnets can reduce the astringency of tannic red wine, a double- blind randomized trial was carried out, in which 96 paired tastings were conducted of magnetized and non-magnetized samples of a young Nebbiolo. The data showed no association between reported differences in astringency and actual difference in the magnetic treatments given to the wine (χ2=0.135, degrees of freedom=1, P=0.71). This study confirms an earlier work that magnets have no observable effect on red wine.

506 PB 248 DETERMINING THE EFFECT OF CLOUD CONDITIONS ON THE VARIATION IN DAYLENGTH PERCEIVED BY PLANTS

Quantum sensors were placed at plant canopy height inside and outside a glass greenhouse. Photosynthetic photon flux (PPF) was measured during September for a 3-hour period near sunrise and sunset, which were determined from US Naval Observatory Circular #171. Under clear skies, the PPF at the canopy exceeded 0.25 μmol·m-2·s-1 for nearly 20 minutes before sunrise through 20 minutes after sunset. Under heavy overcast, the duration was only 5 minutes before sunrise through 5 minutes after sunset. The PPF at the canopy reached 0.25 μmol·m-2·s-1 approximately 12 minutes later in the morning and 12 minutes earlier in the evening than it did outside the greenhouse. The length of the dark period perceived by plants in a greenhouse on September 21st (assuming plants perceive light at 0.25 μmol·m-2·s-1) can range from 11:37 (hr:min) during cloudy conditions to 11:15 during clear ones, a difference of 22 minutes. At 43°N latitude, the maximum difference in date of flower initiation because of an extended period of heavily overcast versus clear weather on a crop such as poinsettias would be one week since the night length during September increases by 3 minutes per day. The actual difference from year to year is probably less because a seven-day duration of heavily overcast weather is unlikely.

Pilot Study of Conditioned Relaxation during Simulated Meditation

For 6 male and 6 female college students (22.5 yr.) EMGs were recorded from the frontalis for 5 days during 4 10-min. periods of EMG-feedback on Days 1 to 4 while repeating a mantra or relaxing. Significantly lower frontalis EMG on Day 5 occurred while subjects repeated the mantra although the actual difference was too small to be of clinical value.

The conceptual impact of linguistic input

ABSTRACTThis study was concerned with the question of how linguistic input of the social environment might affect children's acquisition of word meaning. It was argued that the conceptual impact of linguistic input could be demonstrated where the conceptual components underlying word meaning are not primarily based on perceptual categories, as in the case of kinship terms. In two experiments, 84 German children living in families, and 84 orphans, aged from 6 to 10, were compared in their ability to use natural or artificial kinship names as terms for kinship relations. The results suggested that a mere familiarity interpretation for the actual difference in the ability to use kinship terms could be ruled out. Orphans showed a specific conceptual deficit in handling relational components even when the familiarity of the terms was equated for the two groups. Two issues these results raise are whether the orphans' deficit will disappear with increasing age, and to what extent the correct use of kinship terms is due to differences in general intelligence.

Grapheme-Phoneme and Phoneme-Grapheme Correspondences

The ability to make grapheme-phoneme correspondences (phonemic recall) comprises a necessary part of decoding unknown written words to their oral equivalents, that is, working out the pronunciation of words while reading. However, group-decoding tests and skills-management systems, both of which measure phoneme-grapheme correspondences (graphemic recall), are frequently used to infer students' facility with phonemic recall. The notion that graphemic recall has properties that yield higher scores than phonemic recall and the notion both types of correspondences require identical skills were tested by having 26 second graders make both types of correspondences. Although a design with counterbalanced sequence by correspondences showed a significantly higher mean score for graphemic recall, the actual difference between the two scores was small. A significant rank-order correlation indicates a relationship between the two types of correspondences, but the skills required to make the two types of correspondences are not identical. The results suggest caution should be exercised when inferring students' facility with phonemic recall from performance on instruments which measure graphemic recall.