The document discusses the importance of using statistics to make accurate exposure risk decisions, as individual exposure measurements can vary significantly. It emphasizes targeting at least 95% confidence that the true 95th percentile exposure is less than the occupational exposure limit. The document also outlines the AIHA exposure rating categories and the types of controls recommended based on where exposures fall within those categories.
Turning your organization into a high reliability organization just makes sense. Implementing predictable behaviors and reliable processes create a culture that strives to achieve error-free performance and safety in every procedure, every time. This increases safety and satisfaction for both patients and staff while reducing costs and improving clinical results.
Join HRO expert Tony Gorski and learn steps that you can take to turn your organization into the efficient and safe environment you know it can be.
This presentation explores the challenges, opportunities and available tools in developing a safety case regime for operators of major hazardous installations.
This presentation describes the key performance indicators to assess the quality of work in microbiology department. The KPIs in common use are mentioned and other indicators are summarized.
Turning your organization into a high reliability organization just makes sense. Implementing predictable behaviors and reliable processes create a culture that strives to achieve error-free performance and safety in every procedure, every time. This increases safety and satisfaction for both patients and staff while reducing costs and improving clinical results.
Join HRO expert Tony Gorski and learn steps that you can take to turn your organization into the efficient and safe environment you know it can be.
This presentation explores the challenges, opportunities and available tools in developing a safety case regime for operators of major hazardous installations.
This presentation describes the key performance indicators to assess the quality of work in microbiology department. The KPIs in common use are mentioned and other indicators are summarized.
5 essential steps for sample size determination in clinical trials slidesharenQuery
In this free webinar hosted by nQuery Researcher & Statistician Eimear Keyes, we map out the 5 essential steps for sample size determination in clinical trials. At each step, Eimear will highlight the important function it plays and how to avoid the errors that will negatively impact your sample size determination and therefore your study.
Watch the Video: https://www.statsols.com/webinar/the-5-essential-steps-for-sample-size-determination
Laboratory Management With Constrains Iamm 2010PathKind Labs
Clinical laboratory services are a critical yet much neglected component of health systems in resource poor countries. They are crucial for public health, disease control and surveillance, and guide patient diagnosis and care, but their key role is often not recognized by governments or donors. Laboratory tests should be used to improve the outcome for individual patients or to provide public health information. However, if the quality of laboratory tests is poor, resources will be wasted on repeat tests or inappropriate management and the laboratory service will be inefficient.
The primary goal of Laboratory Medicine is to provide information that is useful to assist medical decision-making, allowing optimal health care. This can only be obtained by generating reliable analytical results on patient samples. Meaningful measurements are indeed essential for the diagnosis, monitoring, treatment, and risk assessment of patients. Inadequate laboratory performance may have extensive consequences for practical medicine, healthcare system, and, in conclusion, for the patient. Poor quality results may actually lead to incorrect interpretation by the clinician, impairing the patient’s
situation.
Accreditation authorities have identified twelve quality system essentials that need to be in place for a laboratory to perform clinical tests adequately and in a quality assured manner. Along with each laboratory performing tests that are in its scope, it is essential that duplication and excess capacity is addressed by forging and operating a network of laboratories leading to consolidation and integration of clinical testing. A network would have collection centers at places convenient to the patients, supported by frequent transfer of samples in appropriate conditions to the laboratory. In the laboratory there is a need for increased automation and relevant training of personnel and the setting up of centralized accessioning, pneumontic chutes for transport of samples to the work bench and for bidirectional interphased equipment to transfer results to desk top of laboratory physicians and after validation of results for the results to be electronically transferred by SMS and/or PDF files via email and/or becoming available online for clients, supplemented by delivery of hard copies of the results.
The challenge in the next decade for laboratory medicine is to accomplish these major changes in organization to meet fiscal restraint and shortage of adequately trained laboratory personnel. Collaborative networks, constructive use of point of care devices, and the development of rapport between laboratories and their clients leading to cost effective utilization of limited resources, are some of the strategies that will maximize patient benefit
In September, the IAASB held two webinars on its recently proposed International Standard on Sustainability Assurance 5000. The webinars featured presentations and discussions from IAASB members, including those who helped draft the standard.
Find more information on ISSA 5000 on the IAASB website: iaasb.org/ISSA5000
Purpose of the Call:
•Recap of aggregated MedRec audit month data that identifies potential opportunities for improvement
•Review quality improvement concepts as it relates to measuring for quality improvement
•Hear how Horizon Health team (NB) is using their data to improve MedRec processes
•Receive a tutorial on how to access your MedRec Quality Score run charts in Patient Safety Metrics.
WATCH: http://bit.ly/1EVcREL
RUNNING HEADER: Potential Risk Factors
Potential Risk Factors
Potential Risk Factors
BUS475
Understanding the risks listed below is regular will be indispensable to assessing an association's necessary arrangement. Besides, seeing how to quantify and screen these risks can assist organizations with recognizing and relieve barricades in the essential provision.
1. Economic Struggles
Changing large scale and microeconomic conditions can cause increasingly significant expenses underway; for instance, required materials can turn out to be scant or have lower edges causing lower benefit. Checking the changing monetary conditions can assist with envisioning the impacts on the business and change techniques varying.
2. Political vulnerability.
The administration assumes an indispensable job in the maintainability and strength, all things considered, legislative unsteadiness, such as visit changes in arrangements, can prompt vulnerabilities and lower benefits. Observing the world of politics can help in the capacity to make inside approach changes to relieve outside risks.
3. Demographic changes.
Changes in populace demographics of the objective market can be gainful because, as it may, gone unchecked can prompt misfortunes. Checking deals information, client profiles, and dissecting buyer conduct can quantify the demographic changes that can compromise the organization.
4. Increasing competition.
With a profitable business comes increasingly extensive measures of competition, and the risk for impersonation increments. The degree of competition can be persistently checked and estimated through statistical surveying and examination, enabling a business to keep its upper hand.
5. Quality Control.
The test of meeting and surpassing the degrees of quality wanted by purchasers frequently represent a risk because of the capacity for new organizations to improve and enter the market. Checking clients' assessment and revamping items to line up with showcase needs can help decrease losses because of quality issues.
Contingency planning
A business contingency plan is a game-plan that your association would take if a surprising occasion or circumstance happens. In some cases, a contingency can be sure, for example, an unexpected flood of cash—however, regularly, the term alludes to an adverse occasion that influences an association's notoriety, money-related well-being, or capacity to remain in business. These incorporate a fire, flood, information penetrates, significant system disappointment, and only the tip of the iceberg.
Contingency plans are a significant part of your general business coherence methodology since they help you guarantee your association is prepared for anything. Numerous huge organizations and government associations make different arrangements of contingency designs with the goal that an assortment of potential dangers is very much looked into, and their proper reactions are thoroughly drilled before.
IAASB Webinar on Assurance on Sustainability Reporting
In September, the IAASB held two webinars on its recently proposed International Standard on Sustainability Assurance 5000. The webinars featured presentations and discussions from IAASB members, including those who helped draft the standard.
Find more information on ISSA 5000 on the IAASB website: iaasb.org/ISSA5000
Measurement System Analysis is the first step of the Measure Phase of an improvement project. Before you can pass judgment on the process, you need to ensure that your measurement system is accurate, precise, capable and in control.
Industrial Training at Shahjalal Fertilizer Company Limited (SFCL)MdTanvirMahtab2
This presentation is about the working procedure of Shahjalal Fertilizer Company Limited (SFCL). A Govt. owned Company of Bangladesh Chemical Industries Corporation under Ministry of Industries.
5 essential steps for sample size determination in clinical trials slidesharenQuery
In this free webinar hosted by nQuery Researcher & Statistician Eimear Keyes, we map out the 5 essential steps for sample size determination in clinical trials. At each step, Eimear will highlight the important function it plays and how to avoid the errors that will negatively impact your sample size determination and therefore your study.
Watch the Video: https://www.statsols.com/webinar/the-5-essential-steps-for-sample-size-determination
Laboratory Management With Constrains Iamm 2010PathKind Labs
Clinical laboratory services are a critical yet much neglected component of health systems in resource poor countries. They are crucial for public health, disease control and surveillance, and guide patient diagnosis and care, but their key role is often not recognized by governments or donors. Laboratory tests should be used to improve the outcome for individual patients or to provide public health information. However, if the quality of laboratory tests is poor, resources will be wasted on repeat tests or inappropriate management and the laboratory service will be inefficient.
The primary goal of Laboratory Medicine is to provide information that is useful to assist medical decision-making, allowing optimal health care. This can only be obtained by generating reliable analytical results on patient samples. Meaningful measurements are indeed essential for the diagnosis, monitoring, treatment, and risk assessment of patients. Inadequate laboratory performance may have extensive consequences for practical medicine, healthcare system, and, in conclusion, for the patient. Poor quality results may actually lead to incorrect interpretation by the clinician, impairing the patient’s
situation.
Accreditation authorities have identified twelve quality system essentials that need to be in place for a laboratory to perform clinical tests adequately and in a quality assured manner. Along with each laboratory performing tests that are in its scope, it is essential that duplication and excess capacity is addressed by forging and operating a network of laboratories leading to consolidation and integration of clinical testing. A network would have collection centers at places convenient to the patients, supported by frequent transfer of samples in appropriate conditions to the laboratory. In the laboratory there is a need for increased automation and relevant training of personnel and the setting up of centralized accessioning, pneumontic chutes for transport of samples to the work bench and for bidirectional interphased equipment to transfer results to desk top of laboratory physicians and after validation of results for the results to be electronically transferred by SMS and/or PDF files via email and/or becoming available online for clients, supplemented by delivery of hard copies of the results.
The challenge in the next decade for laboratory medicine is to accomplish these major changes in organization to meet fiscal restraint and shortage of adequately trained laboratory personnel. Collaborative networks, constructive use of point of care devices, and the development of rapport between laboratories and their clients leading to cost effective utilization of limited resources, are some of the strategies that will maximize patient benefit
In September, the IAASB held two webinars on its recently proposed International Standard on Sustainability Assurance 5000. The webinars featured presentations and discussions from IAASB members, including those who helped draft the standard.
Find more information on ISSA 5000 on the IAASB website: iaasb.org/ISSA5000
Purpose of the Call:
•Recap of aggregated MedRec audit month data that identifies potential opportunities for improvement
•Review quality improvement concepts as it relates to measuring for quality improvement
•Hear how Horizon Health team (NB) is using their data to improve MedRec processes
•Receive a tutorial on how to access your MedRec Quality Score run charts in Patient Safety Metrics.
WATCH: http://bit.ly/1EVcREL
RUNNING HEADER: Potential Risk Factors
Potential Risk Factors
Potential Risk Factors
BUS475
Understanding the risks listed below is regular will be indispensable to assessing an association's necessary arrangement. Besides, seeing how to quantify and screen these risks can assist organizations with recognizing and relieve barricades in the essential provision.
1. Economic Struggles
Changing large scale and microeconomic conditions can cause increasingly significant expenses underway; for instance, required materials can turn out to be scant or have lower edges causing lower benefit. Checking the changing monetary conditions can assist with envisioning the impacts on the business and change techniques varying.
2. Political vulnerability.
The administration assumes an indispensable job in the maintainability and strength, all things considered, legislative unsteadiness, such as visit changes in arrangements, can prompt vulnerabilities and lower benefits. Observing the world of politics can help in the capacity to make inside approach changes to relieve outside risks.
3. Demographic changes.
Changes in populace demographics of the objective market can be gainful because, as it may, gone unchecked can prompt misfortunes. Checking deals information, client profiles, and dissecting buyer conduct can quantify the demographic changes that can compromise the organization.
4. Increasing competition.
With a profitable business comes increasingly extensive measures of competition, and the risk for impersonation increments. The degree of competition can be persistently checked and estimated through statistical surveying and examination, enabling a business to keep its upper hand.
5. Quality Control.
The test of meeting and surpassing the degrees of quality wanted by purchasers frequently represent a risk because of the capacity for new organizations to improve and enter the market. Checking clients' assessment and revamping items to line up with showcase needs can help decrease losses because of quality issues.
Contingency planning
A business contingency plan is a game-plan that your association would take if a surprising occasion or circumstance happens. In some cases, a contingency can be sure, for example, an unexpected flood of cash—however, regularly, the term alludes to an adverse occasion that influences an association's notoriety, money-related well-being, or capacity to remain in business. These incorporate a fire, flood, information penetrates, significant system disappointment, and only the tip of the iceberg.
Contingency plans are a significant part of your general business coherence methodology since they help you guarantee your association is prepared for anything. Numerous huge organizations and government associations make different arrangements of contingency designs with the goal that an assortment of potential dangers is very much looked into, and their proper reactions are thoroughly drilled before.
IAASB Webinar on Assurance on Sustainability Reporting
In September, the IAASB held two webinars on its recently proposed International Standard on Sustainability Assurance 5000. The webinars featured presentations and discussions from IAASB members, including those who helped draft the standard.
Find more information on ISSA 5000 on the IAASB website: iaasb.org/ISSA5000
Measurement System Analysis is the first step of the Measure Phase of an improvement project. Before you can pass judgment on the process, you need to ensure that your measurement system is accurate, precise, capable and in control.
Industrial Training at Shahjalal Fertilizer Company Limited (SFCL)MdTanvirMahtab2
This presentation is about the working procedure of Shahjalal Fertilizer Company Limited (SFCL). A Govt. owned Company of Bangladesh Chemical Industries Corporation under Ministry of Industries.
NO1 Uk best vashikaran specialist in delhi vashikaran baba near me online vas...Amil Baba Dawood bangali
Contact with Dawood Bhai Just call on +92322-6382012 and we'll help you. We'll solve all your problems within 12 to 24 hours and with 101% guarantee and with astrology systematic. If you want to take any personal or professional advice then also you can call us on +92322-6382012 , ONLINE LOVE PROBLEM & Other all types of Daily Life Problem's.Then CALL or WHATSAPP us on +92322-6382012 and Get all these problems solutions here by Amil Baba DAWOOD BANGALI
#vashikaranspecialist #astrologer #palmistry #amliyaat #taweez #manpasandshadi #horoscope #spiritual #lovelife #lovespell #marriagespell#aamilbabainpakistan #amilbabainkarachi #powerfullblackmagicspell #kalajadumantarspecialist #realamilbaba #AmilbabainPakistan #astrologerincanada #astrologerindubai #lovespellsmaster #kalajaduspecialist #lovespellsthatwork #aamilbabainlahore#blackmagicformarriage #aamilbaba #kalajadu #kalailam #taweez #wazifaexpert #jadumantar #vashikaranspecialist #astrologer #palmistry #amliyaat #taweez #manpasandshadi #horoscope #spiritual #lovelife #lovespell #marriagespell#aamilbabainpakistan #amilbabainkarachi #powerfullblackmagicspell #kalajadumantarspecialist #realamilbaba #AmilbabainPakistan #astrologerincanada #astrologerindubai #lovespellsmaster #kalajaduspecialist #lovespellsthatwork #aamilbabainlahore #blackmagicforlove #blackmagicformarriage #aamilbaba #kalajadu #kalailam #taweez #wazifaexpert #jadumantar #vashikaranspecialist #astrologer #palmistry #amliyaat #taweez #manpasandshadi #horoscope #spiritual #lovelife #lovespell #marriagespell#aamilbabainpakistan #amilbabainkarachi #powerfullblackmagicspell #kalajadumantarspecialist #realamilbaba #AmilbabainPakistan #astrologerincanada #astrologerindubai #lovespellsmaster #kalajaduspecialist #lovespellsthatwork #aamilbabainlahore #Amilbabainuk #amilbabainspain #amilbabaindubai #Amilbabainnorway #amilbabainkrachi #amilbabainlahore #amilbabaingujranwalan #amilbabainislamabad
Final project report on grocery store management system..pdfKamal Acharya
In today’s fast-changing business environment, it’s extremely important to be able to respond to client needs in the most effective and timely manner. If your customers wish to see your business online and have instant access to your products or services.
Online Grocery Store is an e-commerce website, which retails various grocery products. This project allows viewing various products available enables registered users to purchase desired products instantly using Paytm, UPI payment processor (Instant Pay) and also can place order by using Cash on Delivery (Pay Later) option. This project provides an easy access to Administrators and Managers to view orders placed using Pay Later and Instant Pay options.
In order to develop an e-commerce website, a number of Technologies must be studied and understood. These include multi-tiered architecture, server and client-side scripting techniques, implementation technologies, programming language (such as PHP, HTML, CSS, JavaScript) and MySQL relational databases. This is a project with the objective to develop a basic website where a consumer is provided with a shopping cart website and also to know about the technologies used to develop such a website.
This document will discuss each of the underlying technologies to create and implement an e- commerce website.
Water scarcity is the lack of fresh water resources to meet the standard water demand. There are two type of water scarcity. One is physical. The other is economic water scarcity.
Saudi Arabia stands as a titan in the global energy landscape, renowned for its abundant oil and gas resources. It's the largest exporter of petroleum and holds some of the world's most significant reserves. Let's delve into the top 10 oil and gas projects shaping Saudi Arabia's energy future in 2024.
Explore the innovative world of trenchless pipe repair with our comprehensive guide, "The Benefits and Techniques of Trenchless Pipe Repair." This document delves into the modern methods of repairing underground pipes without the need for extensive excavation, highlighting the numerous advantages and the latest techniques used in the industry.
Learn about the cost savings, reduced environmental impact, and minimal disruption associated with trenchless technology. Discover detailed explanations of popular techniques such as pipe bursting, cured-in-place pipe (CIPP) lining, and directional drilling. Understand how these methods can be applied to various types of infrastructure, from residential plumbing to large-scale municipal systems.
Ideal for homeowners, contractors, engineers, and anyone interested in modern plumbing solutions, this guide provides valuable insights into why trenchless pipe repair is becoming the preferred choice for pipe rehabilitation. Stay informed about the latest advancements and best practices in the field.
RAT: Retrieval Augmented Thoughts Elicit Context-Aware Reasoning in Long-Hori...
Video 1A Handout 2023.pptx
1. AIHA VIDEO SERIES:
MAKING ACCURATE EXPOSURE RISK
DECISIONS
Video 1A
Exposure Variability and the Importance of
Using Statistics to Improve Judgements
1
John R. Mulhausen, PhD, CIH, CSP, FAIHA
2. Disclaimer &
Copyright
Although the information contained in this session has been compiled from sources believed to be reliable, the presenter and AIHA®
make no guarantee as to, and assumes no responsibility for, the correctness, sufficiency, or completeness of such information.
Since standards and codes vary from one place to another, consult with your local Occupational or Environmental Health and Safety
professional to determine the current state of the art before applying what you learn from this webinar.
AIHA must ensure balance, independence, objectivity, and scientific rigor in its educational events. Instructors are expected to
disclose any significant financial interests or other relationships. The intent of this disclosure is not to prevent an instructor from
presenting, but to provide participants with information to base their own judgments. It remains up to the participant to determine
whether an instructor’s interests or relationships may influence the presentation.
Session presentation material belongs to the presenter with usage rights given to AIHA. This session and associated materials can
be reproduced, rebroadcast, or made into derivative works without express written permission.
3. Disclaimer &
Copyright
Handout Information
All AIHA University session handouts are produced by AIHA as submitted by the instructors, and in the instructor determined order.
AIHA does not change or modify handout content; we may adjust images to improve layout and clarity.
4. AGENDA
• Why Important?
• Decision Statistic
• AIHA Categories
• Exposure Judgment Accuracy
• Exposure Variability
• Inferential IH Statistics
• Data Interpretation Using IHSTAT™
• Examples
• Review: Data Interpretation
• Key Resources
6. Effective and Efficient
Exposure Risk Management
Effective:
Ensure that no worker has
unacceptable exposures
Efficient:
Do it for minimum cost
6
7. 7
7
What if our exposure assessment is wrong?
If we underestimate the exposure?
• Increased risk to employees
If we overestimate the exposure?
• Unnecessary expenditures for controls
• Unnecessary constraints for employees
and production
7
Well-Designed Exposure Risk
Management Strategy
We Want: We Don’t Want:
Good Data Bad Data
To Be Effective To Not Be Protective
To Be Efficient To Waste Resources
No Biases Biases (High or Low)
Low Uncertainty High Uncertainty
Correct Decisions Wrong Decisions
8. AIHA Exposure Risk Management Process
Comprehensive Exposure Assessment
and Management
• Goal is to understand and manage
all exposures
• Document both qualitative and
quantitative exposure assessments
• Document low exposures
• Emphasize Accurate and Efficient
Exposure Decisions
• Drive Effective and Efficient
Exposure Risk Management
• Devise OELs when needed
• Continuous Prioritization
• Continuous Improvement
• Identify the critical SEGs
• Anticipate and manage change
8
9. CLICK TO EDIT MASTER TITLE
STYLE
9
DECISION
STATISTIC
10. Decision Statistic:
•Relative to Appropriate OEL
•Defined BEFORE conducting exposure risk assessment
Defines Objective for Acceptable Exposure
(i.e. No Further Intervention Needed)
10
11. Decision Statistic:
1st Framing Question
An employee performs a job 100 days per year. If
you collected personal samples on the employee all
100 days, how many days is it acceptable for
exposures to exceed the Occupational Exposure
Limit (OEL) without a respirator?
1) 0 days?
2) 1 days?
3) 5 days?
4) 10 days?
5) 25 days?
6) 50 days?
11
12. Decision Statistic:
1st Framing Question
An employee performs a job 100 days per year. If
you collected personal samples on the employee all
100 days, how many days is it acceptable for
exposures to exceed the Occupational Exposure
Limit (OEL) without a respirator?
1) 0 days?
2) 1 days?
3) 5 days?
4) 10 days?
5) 25 days?
6) 50 days?
12
• Answers emphasize the desire for
very few days above the OEL
• Professional consensus developing
around targeting for no more than
5 days out of 100 above the OEL
(i.e. 95th Percentile)
95%ile
5/100 (5%) above
95/100 (95%) below
16. Decision Statistic:
2nd Framing Question
How sure do you want to
be in your judgment?
16
?
?
1) 100% Sure?
2) 99%?
3) 95%?
4) 90%?
5) 70%?
6) 50%?
17. Decision Statistic:
2nd Framing Question
How sure do you want to
be in your judgment?
17
?
?
1) 100% Sure?
2) 99%?
3) 95%?
4) 90%?
5) 70%?
6) 50%?
• Answers express the desire for high confidence that
employees are protected.
• Implementing the AIHA Strategy with its emphasis on
driving follow-up actions and continuous improvement
enables a program to strive for high confidence.
• Common to strive for 95% confidence.
18. Decision Statistic:
“Strive for at least 95% confidence that the
true 95th percentile is less than the OEL”
OEL
95%ile
95%ile UCL
0 1 2 3 4
-2
0
2
4
6
8
10
0 0.2 0.4 0.6 0.8 1 1.2
Best Estimate
95% Upper Confidence
Best Estimate
Exposure Profile
95% Upper Confidence
Exposure Profile
Pulling Together 1st and 2nd Framing Questions:
18
20. Exposure Rating Category** Recommended Control
0 (<1% of OEL) No action
1 (<10% of OEL) Procedures and Training; General Hazard Communication
2 (10-50% of OEL)
+ Chemical Specific Hazard Communication; Periodic Exposure
Monitoring,
3 (50-100% of OEL)
+ Required Exposure Monitoring, Workplace Inspections to Verify
Work Practice Controls; Medical Surveillance, Biological Monitoring
4 (>100% of OEL)
+ Implement Hierarchy of Controls; Monitoring to Validate
Respirator Protection Factor Selection.
Multiples of OEL (>500% of OEL
or others based on respirator APF)
+Immediate Engineering Controls or Process Shut Down, Validate
Acceptable Respirators
** Decision statistic = 95th percentile
AIHA Exposure Rating and Control Categories
20
21. AIHA Exposure Rating and Control Categories
Increase Effectiveness and Efficiency
• Avoid diminishing returns from
“over-refining” exposure estimates
• Streamline Documentation
• Facilitate Qualitative Exposure
Judgements
• Drive consistent follow-up
management and control activities
which lead to consistent risk
management.
21
22. CLICK TO EDIT MASTER TITLE
STYLE
22
EXPOSURE JUDGMENT
ACCURACY
23. ?
Exposure Risk Decisions:
How Accurate Are We?
** Decision statistic = 95th percentile
Sample Results
(ppm)
18
15
5
8
12
When We Have Monitoring Data . . .
23
24. Video Tasks: Quantitative Judgment Accuracy
Pre- and Post- Statistical Training
24
P. Logan, G. Ramachandran, J. Mulhausen and P. Hewett “Occupational Exposure Decisions: Can Limited Data
Interpretation Training Help Improve Accuracy?”. Annals of Occupational Hygiene - 2009
Biased Low
Pre-Training
Statistical Training
Increased Accuracy
and Eliminated Bias
26. Task results, Pre and Post training
7%
18%
28%
46%
1% 0% 0%
2%
11%
16%
69%
1% 0% 0%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
-3 -2 -1 0 1 2 3
%
correct
judgments
Pre Training Post training
Number of AIHA Categories Away From the Correct Category
Vadali, Ramachandran, Mulhausen & Banerjee (2012): Effect of Training on Exposure Judgment
Accuracy of Industrial Hygienists, Journal of Occupational and Environmental Hygiene, 9:4, 242-256
Statistical Training
Increased Accuracy
NIOSH Funded U of MN Study
Actual Workplace Assessments
26
27. Monitoring-Based
Exposure Judgments
• Bad News
• Often incorrect
• Good News
• Simple statistical training
improves judgments
• GREAT NEWS!!!
• Using statistical tools when we
make monitoring-based
exposure judgments will greatly
improve accuracy
27
Industrial Hygiene Statistics Beta 0.9 - For trial and testing only - Please do not distribute
Data Description: John Mulhausen
OEL DESCRIPTIVE STATISTICS
5 Number of Samples (n) 15
Maximum (max) 5.5
Sample Data Minimum (min) 1.2
(max n=50) Range 4.3
No less-than (<) Percent above OEL (%>OEL) 6.667
or greater-than (>) Mean 2.680
1.3 Median 2.500
1.8 Standard Deviation (s) 1.138
1.2 Mean of Log (LN) Transformed Data 0.908
4.5 Std Deviation of Log (LN) Transformed Data 0.407
2 Geometric Mean (GM) 2.479
2.1 Geometric Standard Deviation (GSD) 1.502
5.5
2.2 TEST FOR DISTRIBUTION FIT
3 W Test of Log (LN) Transformed Data 0.974
2.4 Lognormal (a=0.05)? Yes
2.5
2.5 W Test of Data 0.904
3.5 Normal (a=0.05)? Yes
2.8
2.9 LOGNORMAL PARAMETRIC STATISTICS
Estimated Arithmetic Mean - MVUE 2.677
1,95%LCL - Land's "Exact" 2.257
1,95%UCL - Land's "Exact" 3.327
95th Percentile 4.843
Upper Tolerance Limit (95%, 95%) 7.046
Percent Exceeding OEL (% > OEL) 4.241
1,95% LCL % > OEL 0.855
1,95% UCL % > OEL 15.271
NORMAL PARAMETRIC STATISTICS
Mean 2.680
1,95%LCL - t stats 2.162
1,95%UCL- t stats 3.198
95th Percentile - Z 4.553
Upper Tolerance Limit (95%, 95%) 5.60
Percent Exceeding OEL (% > OEL) 2.078
Linear Probability Plot and Least Squares
Best Fit Line
1%
2%
5%
10%
16%
25%
50%
75%
84%
90%
95%
98%
99%
-5 0 5 10
Concentration
Log-Probability Plot and Least Squares Best Fit Line
1%
2%
5%
10%
16%
25%
50%
75%
84%
90%
95%
98%
99%
0 1 10
Concentration
Idealized Lognormal Distribution
AM and CI's 95%ile
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0 1 2 3 4 5 6 7
Concentration
Sequential Data Plot
0
1
2
3
4
5
6
0 2 4 6 8 10 12 14 16
Sample Number
Concentration
AIHA IHSTAT
28. Use Statistical Tools!!
28
95%ile = 1.2
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0 0.5 1.0 1.5 2.0
Concentration (mg/M3)
UTL95%,95% =
16 mg/M3
AIHA
IHSTAT
Industrial Hygiene Statistics Beta 0.9 - For trial and testing only - Please do not distribute
Data Description: John Mulhausen
OEL DESCRIPTIVE STATISTICS
5 Number of Samples (n) 15
Maximum (max) 5.5
Sample Data Minimum (min) 1.2
(max n=50) Range 4.3
No less-than (<) Percent above OEL (%>OEL) 6.667
or greater-than (>) Mean 2.680
1.3 Median 2.500
1.8 Standard Deviation (s) 1.138
1.2 Mean of Log (LN) Transformed Data 0.908
4.5 Std Deviation of Log (LN) Transformed Data 0.407
2 Geometric Mean (GM) 2.479
2.1 Geometric Standard Deviation (GSD) 1.502
5.5
2.2 TEST FOR DISTRIBUTION FIT
3 W Test of Log (LN) Transformed Data 0.974
2.4 Lognormal (a=0.05)? Yes
2.5
2.5 W Test of Data 0.904
3.5 Normal (a=0.05)? Yes
2.8
2.9 LOGNORMAL PARAMETRIC STATISTICS
Estimated Arithmetic Mean - MVUE 2.677
1,95%LCL - Land's "Exact" 2.257
1,95%UCL - Land's "Exact" 3.327
95th Percentile 4.843
Upper Tolerance Limit (95%, 95%) 7.046
Percent Exceeding OEL (% > OEL) 4.241
1,95% LCL % > OEL 0.855
1,95% UCL % > OEL 15.271
NORMAL PARAMETRIC STATISTICS
Mean 2.680
1,95%LCL - t stats 2.162
1,95%UCL- t stats 3.198
95th Percentile - Z 4.553
Upper Tolerance Limit (95%, 95%) 5.60
Percent Exceeding OEL (% > OEL) 2.078
Linear Probability Plot and Least Squares
Best Fit Line
1%
2%
5%
10%
16%
25%
50%
75%
84%
90%
95%
98%
99%
-5 0 5 10
Concentration
Log-Probability Plot and Least Squares Best Fit Line
1%
2%
5%
10%
16%
25%
50%
75%
84%
90%
95%
98%
99%
0 1 10
Concentration
Idealized Lognormal Distribution
AM and CI's 95%ile
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0 1 2 3 4 5 6 7
Concentration
Sequential Data Plot
0
1
2
3
4
5
6
0 2 4 6 8 10 12 14 16
Sample Number
Concentration
IH
Data
Analyst
Exposure Rating Category
<1%OEL <10% OEL 10 – 50% 50 – 100% >100% OEL
Probability
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0
0.087
0.4
0.513
OEL
Likelihood that
95%ile falls into
indicated
Exposure Rating
Category
Initial
Qualitative
Assessment
or Validated
Model
Prior
Exposure Rating
0 1 2 3 4
Decision
Probability
1
0.8
0.6
0.4
0.2
0
0.05
0.2
0.5
0.2
0.05
Monitoring
Results
Likelihood
Exposure Rating
0 1 2 3 4
Decision
Probability
1
0.8
0.6
0.4
0.2
0
0 0 0.06
0.376
0.564
Integrated
Exposure
Assessment
Posterior
Exposure Rating
0 1 2 3 4
Decision
Probability
1
0.8
0.6
0.4
0.2
0
0 0
0.225
0.564
0.211
Expostats
Traditional Statistics Bayesian Statistics
28
29. CLICK TO EDIT MASTER TITLE
STYLE
29
EXPOSURE
VARIABILITY
31. Annual population of exposures for one worker: 250 Worker-days per Year
Measurement
250
200
150
100
50
0
Concentration
6.5
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
Trying to understand this . . . .
31
32. Annual population of exposures for one worker: 250 Worker-days per Year
Measurement
250
200
150
100
50
0
Concentration
6.5
6
5.5
5
4.5
4
3.5
3
2.5
2
1.5
1
0.5
0
Based on this (n=5 samples) . . . .
“See” Only 2%
of Exposures
32
34. CLICK TO EDIT MASTER TITLE
STYLE
34
INFERENTIAL IH
STATISTICS
35. Solution: Inferential Statistics . . . .
Estimate From What We Looked At
(Our Five Samples) . . .
The Actual Population Exposure
Profile (SEG of 10 Workers)
Using Knowledge of Underlying
Shape (Lognormal Distribution) . . .
35
36. Things can go wrong at several stages when
extrapolating sample data to make inferences
about the underlying population
Data
(measurements)
Sample
(Actual exposure
levels during
measurements)
Distribution of
which the sample
is representative
• Intrument
quality
• Interferences
• Exposure
periods missed
• Worst case
• Night shift
• Summer/winter
• Process changes
• Selection of workers
Inductive Inference
Target population
(exposure
distribution)
• Sample size
Confidence
intervals
36
37. Things can go wrong at several stages when
extrapolating sample data to make inferences
about the underlying population
Data
(measurements)
Sample
(Actual exposure
levels during
measurements)
Distribution of
which the sample
is representative
• Intrument
quality
• Interferences
• Exposure
periods missed
• Worst case
• Night shift
• Summer/winter
• Process changes
• Selection of workers
Inductive Inference
Target population
(exposure
distribution)
• Sample size
Confidence
intervals
Critical Data Quality Considerations
• Defined decision statistic
• Well defined SEG
• Appropriate OEL
• Well described exposure question
• Appropriate sampling strategy
• Valid and appropriate monitoring method
• Validated analytical method
37
38. Lognormal Model Most Appropriate?
38
• Many papers dating back to the 60s, in Europe and the US, have shown the lognormal
distribution to fit occupational exposure data reasonably well.
• Noise exposure data also follow a lognormal distribution when expressed as dose.
• Formal statistical tests exist but they have low power for small sample sizes, and reject
lognormality very (too) quickly for large sample sizes.
39. Lognormal Model Most Appropriate?
39
• Many papers dating back to the 60s, in Europe and the US, have shown the lognormal
distribution to fit occupational exposure data reasonably well.
• Noise exposure data also follow a lognormal distribution when expressed as dose.
• Formal statistical tests exist but they have low power for small sample sizes, and reject
lognormality very (too) quickly for large sample sizes.
A Pragmatic Approach:
• Assume lognormality based on historical weight of evidence
• Make a graphical check (Quantile - Quantile or log – probit
plot) to detect obvious departures from the model
“All models are wrong, some are useful”
- George E. P. Box
40. Always Check the
Lognormal Assumption
• Check your monitoring data for
lognormal distribution fit before
detailed analysis.
• If data is not lognormal go back
and verify SEG is constructed well.
• Are jobs/tasks truly similar?
• Does the data have errors?
• Should SEG be broken down to
smaller levels?
• Challenge your SEG assumptions.
• Many other factors . . .
40
Probability
Probit
3
2
1
0
-1
-2
-3
99
98
95
90
84
75
50
25
16
10
5
2
1 99
98
95
90
84
75
50
25
16
10
5
2
1
Concentration
0.01
0.1
1
42. probability
density
Less frequent
More frequent
Area under
the curve sums to
100% of the population
Values of the quantity of interest
Less frequent
𝑋𝑃%ile = 𝐺𝑀 ⋅ 𝐺𝑆𝐷𝑍𝑝
Interpreting a probability density curve is just like
interpreting a histogram . . .
42
43. 95%ile = 𝐺𝑀 ⋅ 𝐺𝑆𝐷1.645
95%ile
95% 5%
probability
density
44. Lognormal Distribution
Defined by GM and GSD
Geometric mean (GM)
Measure location. Central parameter. Cuts
the distribution into 2 equal parts (median).
Geometric standard deviation (GSD)
Measure variability (~ distance between
the lowest and highest values)
Concentration Concentration
Median = GM
Mean
Mode
44
45. Calculate GM and GSD
Samples
xi (mg/m3)
0.84
0.98
0.42
1.16
1.36
2.66
46. Calculate GM and GSD
Samples
xi (mg/m3)
0.84
0.98
0.42
1.16
1.36
2.66
yi=ln(xi)
-0.1744
-0.0202
-0.8675
0.1484
0.3075
0.9783
1. Log-Transform Data:
e.g. yi=ln(xi)
47. Calculate GM and GSD
Samples
xi (mg/m3)
0.84
0.98
0.42
1.16
1.36
2.66
yi=ln(xi)
-0.1744
-0.0202
-0.8675
0.1484
0.3075
0.9783
2. Calculate Parameter of
Interest:
e.g. mean (y) and
standard deviation (sy)
1. Log-Transform Data:
e.g. yi=ln(xi)
y = 0.062
sy = 0.606
48. Calculate GM and GSD
Samples
xi (mg/m3)
0.84
0.98
0.42
1.16
1.36
2.66
yi=ln(xi)
-0.1744
-0.0202
-0.8675
0.1484
0.3075
0.9783
2. Calculate Parameter of
Interest:
e.g. mean (y) and
standard deviation (sy)
1. Log-Transform Data:
e.g. yi=ln(xi)
3. Calculate Anti-Log of the
Parameter of Interest:
e.g. GM=exp(y) and
GSD=exp(sy)
GM = 1.06
GSD = 1.83
y = 0.062
sy = 0.606
50. Calculate 95%ile
Worked Example for Reference
Welding Fume Sample Data
𝑋𝑃%ile = 𝐺𝑀 ⋅ 𝐺𝑆𝐷𝑍𝑝
95%ile = 𝐺𝑀 ⋅ 𝐺𝑆𝐷1.645
• Six full-shift TWA welding
fume measurements resulted
in the following statistics:
GM = 1.06 mg/m3
GSD = 1.83
• What is the point estimate
(i.e., best estimate) of the true
95th percentile?
95%ile = 1.06 ⋅ 1.831.645
95%ile = 2.86 mg/m3
95%ile = 2.86 mg/m3
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0 2 4 6 8 10
50
51. Upper Tolerance Limit (UTL) for the 95th Percentile
[Same as 95%ile Upper Confidence Limit (UCL)]
• Concept
• Calculate the 95% upper tolerance limit
(same as upper confidence limit) for the
95th percentile statistic to characterize
uncertainty in the point estimate
• Interpretation
• If the UTL95%,95% is less than the OEL,
then we can say that we are at least 95%
confident that the true 95th percentile is
less than the OEL
95%ile
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
Distribution of
SEG Exposures
(Exposure Profile)
51
52. Upper Tolerance Limit (UTL) for the 95th Percentile
[Same as 95%ile Upper Confidence Limit (UCL)]
• Concept
• Calculate the 95% upper tolerance limit
(same as upper confidence limit) for the
95th percentile statistic to characterize
uncertainty in the point estimate
• Interpretation
• If the UTL95%,95% is less than the OEL,
then we can say that we are at least 95%
confident that the true 95th percentile is
less than the OEL
95%ile
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
95%ile point
estimate
95%ile point
estimate
uncertainty
distribution
Distribution of
SEG Exposures
(Exposure Profile)
52
53. Upper Tolerance Limit (UTL) for the 95th Percentile
[Same as 95%ile Upper Confidence Limit (UCL)]
• Concept
• Calculate the 95% upper tolerance limit
(same as upper confidence limit) for the
95th percentile statistic to characterize
uncertainty in the point estimate
• Interpretation
• If the UTL95%,95% is less than the OEL,
then we can say that we are at least 95%
confident that the true 95th percentile is
less than the OEL
95%ile
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
95%ile point
estimate
95%ile point
estimate
uncertainty
distribution
95% upper confidence limit
for the 95%ile estimate
UTL95%,95%
Distribution of
SEG Exposures
(Exposure Profile)
53
54. Upper Tolerance Limit (UTL) for the 95th Percentile
[Same as 95%ile Upper Confidence Limit (UCL)]
• Concept
• Calculate the 95% upper tolerance limit
(same as upper confidence limit) for the
95th percentile statistic to characterize
uncertainty in the point estimate
• Interpretation
• If the UTL95%,95% is less than the OEL,
then we can say that we are at least 95%
confident that the true 95th percentile is
less than the OEL
95%ile
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
95%ile point
estimate
95%ile point
estimate
uncertainty
distribution
95% upper confidence limit
for the 95%ile estimate
UTL95%,95%
UTL95%,95% = UCL95,95
Distribution of
SEG Exposures
(Exposure Profile)
54
55. Upper Tolerance Limit (UTL) for the 95th Percentile
[Same as 95%ile Upper Confidence Limit (UCL)]
• Concept
• Calculate the 95% upper tolerance limit
(same as upper confidence limit) for the
95th percentile statistic to characterize
uncertainty in the point estimate
• Interpretation
• If the UTL95%,95% is less than the OEL,
then we can say that we are at least 95%
confident that the true 95th percentile is
less than the OEL
95%ile
0
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
UTL95%,95% = UCL95,95
Distribution of
SEG Exposures
(Exposure Profile)
55
56. Calculate UTL95%,95% for the 95th Percentile
(UTL95%,95% = UCL95%,95%)
Procedure:
1.Calculate the GM and GSD
2.Using n, read the UTL K-value from
the appropriate table
g = confidence level, e.g., 0.95
p = proportion, e.g., 0.95
n = sample size
3.Using GM, GSD, and K, calculate the
UTL95%,95%:
56
𝑦 = ln 𝐺𝑀
𝑠𝑦 = ln 𝐺𝑆𝐷
𝑈𝑇𝐿95%,95% = exp 𝑦 + 𝐾 ⋅ 𝑠𝑦
Factors for one-sided tolerance limits
for normal distributions*
*Partial Table IV.11 - A Strategy for Assessing
and Managing Occupational Exposures, Fourth
Edition. AIHA Press 2015.
57. Worked Example for Reference:
Calculate 95%ile UTL95%,95
Welding Fume Sample Data
• Six full-shift TWA welding fume
measurements:
GM = 1.06 mg/m3
GSD = 1.83
• What is the 95th percentile UTL95%,95%?
𝑈𝑇𝐿95%,95% = exp 𝑦 + 𝐾 ⋅ 𝑠𝑦
𝑈𝑇𝐿95%,95% = exp 0.058 + 3.707 ⋅ 0.604
𝑠𝑦 = ln 𝐺𝑆𝐷 = ln( 1.83) = 0.604
𝑦 = ln 𝐺𝑀 = ln 1.06 = 0.058
𝑈𝑇𝐿95%,95% = 10mg/m3
95%ile =
2.86 mg/m3
UTL95%,95% =
10 mg/m3
0
0.2
0.4
0.6
0.8
0 2 4 6 8 10
57
Factors for one-sided tolerance limits
for normal distributions
58. CLICK TO EDIT MASTER TITLE
STYLE
58
aiha.org | 58
DATA INTERPRETATION
USING IHSTAT
59. Use Statistical Tools!!
59
95%ile = 1.2
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0 0.5 1.0 1.5 2.0
Concentration (mg/M3)
UTL95%,95% =
16 mg/M3
AIHA
IHSTAT
Industrial Hygiene Statistics Beta 0.9 - For trial and testing only - Please do not distribute
Data Description: John Mulhausen
OEL DESCRIPTIVE STATISTICS
5 Number of Samples (n) 15
Maximum (max) 5.5
Sample Data Minimum (min) 1.2
(max n=50) Range 4.3
No less-than (<) Percent above OEL (%>OEL) 6.667
or greater-than (>) Mean 2.680
1.3 Median 2.500
1.8 Standard Deviation (s) 1.138
1.2 Mean of Log (LN) Transformed Data 0.908
4.5 Std Deviation of Log (LN) Transformed Data 0.407
2 Geometric Mean (GM) 2.479
2.1 Geometric Standard Deviation (GSD) 1.502
5.5
2.2 TEST FOR DISTRIBUTION FIT
3 W Test of Log (LN) Transformed Data 0.974
2.4 Lognormal (a=0.05)? Yes
2.5
2.5 W Test of Data 0.904
3.5 Normal (a=0.05)? Yes
2.8
2.9 LOGNORMAL PARAMETRIC STATISTICS
Estimated Arithmetic Mean - MVUE 2.677
1,95%LCL - Land's "Exact" 2.257
1,95%UCL - Land's "Exact" 3.327
95th Percentile 4.843
Upper Tolerance Limit (95%, 95%) 7.046
Percent Exceeding OEL (% > OEL) 4.241
1,95% LCL % > OEL 0.855
1,95% UCL % > OEL 15.271
NORMAL PARAMETRIC STATISTICS
Mean 2.680
1,95%LCL - t stats 2.162
1,95%UCL- t stats 3.198
95th Percentile - Z 4.553
Upper Tolerance Limit (95%, 95%) 5.60
Percent Exceeding OEL (% > OEL) 2.078
Linear Probability Plot and Least Squares
Best Fit Line
1%
2%
5%
10%
16%
25%
50%
75%
84%
90%
95%
98%
99%
-5 0 5 10
Concentration
Log-Probability Plot and Least Squares Best Fit Line
1%
2%
5%
10%
16%
25%
50%
75%
84%
90%
95%
98%
99%
0 1 10
Concentration
Idealized Lognormal Distribution
AM and CI's 95%ile
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
0.45
0 1 2 3 4 5 6 7
Concentration
Sequential Data Plot
0
1
2
3
4
5
6
0 2 4 6 8 10 12 14 16
Sample Number
Concentration
IH
Data
Analyst
Exposure Rating Category
<1%OEL <10% OEL 10 – 50% 50 – 100% >100% OEL
Probability
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0 0
0.087
0.4
0.513
OEL
Likelihood that
95%ile falls into
indicated
Exposure Rating
Category
Initial
Qualitative
Assessment
or Validated
Model
Prior
Exposure Rating
0 1 2 3 4
Decision
Probability
1
0.8
0.6
0.4
0.2
0
0.05
0.2
0.5
0.2
0.05
Monitoring
Results
Likelihood
Exposure Rating
0 1 2 3 4
Decision
Probability
1
0.8
0.6
0.4
0.2
0
0 0 0.06
0.376
0.564
Integrated
Exposure
Assessment
Posterior
Exposure Rating
0 1 2 3 4
Decision
Probability
1
0.8
0.6
0.4
0.2
0
0 0
0.225
0.564
0.211
Expostats
Traditional Statistics Bayesian Statistics
Focus on Traditional
IH Statistics
59
67. Steps in Data Analysis and Interpretation*
1. Enter Data Into Appropriate Statistical Tool
2. Evaluate the Goodness-of-fit
3. Review Descriptive and Inferential Statistics
Compare…
• the “decision statistic” (e.g., sample 95th percentile) to the OEL.
• the 95%UCL to the OEL.
4. Assign a Final Rating and Certainty Level
• Final Rating: Compare the sample 95th percentile to the Exposure
Control Categories (ECCs) and select a category.
• Certainty Level: Compare the 95%UCL to the ECCs:
• Low certainty if > 2 categories above the chosen ECC
• Medium certainty if only 1 category above
• High certainty if within chosen category
5. Document the Analysis and Recommendations
Recommend controls and/or PPE; work practice evaluation; additional
sampling; surveillance sampling, etc.
67
*After Executing a Carefully
Defined Monitoring Plan:
• Defined decision statistic
• Well defined SEG
• Appropriate OEL
• Well described exposure question
• Appropriate sampling strategy
• Valid and appropriate monitoring
method
• Validated analytical method
Hewett’s
ROT
67
82. Change in UCL95%,95% and LCL95%,95%
With Increasing Number of Samples
Idealized
GSD = 2 Random measurements generated
from a distribution where the true
GSD=3.0 and the true 95%ile=1.
Simulated
Number of Samples Number of Samples
82
83. * Decision statistic = 95th percentile
?
Into which AIHA Exposure Category will the 95th percentile
MOST LIKELY fall?
OEL = 100 ppm
83
Example 5
Sample Results
(ppm)
8
55
5
37
12
84. Exposure Rating
Category*
0
(<1% of
OEL)
1
(<10% of
OEL)
2
(10-50% of
OEL)
3
(50-100%
of OEL)
4
(>100% of
OEL)
Multiples of OEL
(>500% of OEL
or others based
on respirator
APF)
Recommended Control No action Procedures and
Training;
General Hazard
Communication
+ Chemical
Specific Hazard
Communication;
Periodic
Exposure
Monitoring,
+ Required
Exposure
Monitoring,
Workplace
Inspections to
Verify Work
Practice
Controls; Medical
Surveillance,
Biological
Monitoring
+ Implement
Hierarchy of
Controls;
Monitoring to
Validate
Respirator
Protection Factor
Selection.
+Immediate
Engineering
Controls or
Process Shut
Down, Validate
Acceptable
Respirators
Sample Results
(ppm)
8
55
5
37
12
OEL = 100 ppm GM = 15.8 ppm
GSD = 2.8
95%ile =84.1 ppm
95%ile UCL95,95 = 1135.3 ppm OEL
95%ile
0
0.005
0.01
0.015
0.02
0.025
0.03
0.035
0.04
0.045
0 20 40 60 80 100 120
95% UCL
95%ile Likely in
Category 3 ? 4 ???
(Low Certainty)
Less Than 95%
Confident That True
95%ile Exposure <OEL
Follow-Up Actions:
+ Required Exposure
Monitoring, Workplace
Inspections to Verify Work
Practice Controls; Medical
Surveillance, Biological
Monitoring
+ Consider Implementing
Hierarchy of Controls;
84
85. A Few Words About Handling Censored Data . . .
Sample
Results
(ppm)
8
25
<5
10
<3
7
11
86. A Few Words About Handling Censored Data . . .
Do:
• Minimize the Likelihood and Impact of Censored Data
with Good Sample Planning
• Strive for a detection limit that is less than 10% of the OEL.
• Ask the laboratory performing sample analysis if they would
calculate results down to their limit of detection (LOD) in
addition to their limit of quantification (LOQ) as the LOD is
often significantly lower than the LOQ.
Sample
Results
(ppm)
8
25
<5
10
<3
7
11
87. A Few Words About Handling Censored Data . . .
Do:
• Minimize the Likelihood and Impact of Censored Data
with Good Sample Planning
• Strive for a detection limit that is less than 10% of the OEL.
• Ask the laboratory performing sample analysis if they would
calculate results down to their limit of detection (LOD) in
addition to their limit of quantification (LOQ) as the LOD is
often significantly lower than the LOQ.
Don’t:
• Remove the non-detects from the statistical analysis.
• Perform data analysis with the detection limit
substituted for the less-than values.
Sample
Results
(ppm)
8
25
<5
10
<3
7
11
88. Parametric Censored Data Analysis Methods
(Assumes Lognormal Distribution)
• Simple Substitution - DL/2 or DL/sqrt(2)
• Very easy to implement
• Reasonable performance [particularly DL/sqrt(2) for 95%ile estimation] for low n (<20)
and low (<25%) to moderate (25-50%) censoring.
• Maximum Likelihood Estimates (MLE)
• Complex calculations
• Closest to best universal method
• Beta Substitution
• Straight forward to program in a spreadsheet
• Performance similar to MLE
• Log-Probit Regression (LPR) - also called Regression on Order Statistics (ROS)
• Straight forward to program in a spreadsheet
• Good choice for 25% to 50% censored data if n greater than 10 or 15.
Sample
Results
(ppm)
8
25
<5
10
<3
7
11
89. Parametric Censored Data Analysis Methods
(Assumes Lognormal Distribution)
• Simple Substitution - DL/2 or DL/sqrt(2)
• Very easy to implement
• Reasonable performance [particularly DL/sqrt(2) for 95%ile estimation] for low n (<20)
and low (<25%) to moderate (25-50%) censoring.
• Maximum Likelihood Estimates (MLE)
• Complex calculations
• Closest to best universal method
• Beta Substitution
• Straight forward to program in a spreadsheet
• Performance similar to MLE
• Log-Probit Regression (LPR) - also called Regression on Order Statistics (ROS)
• Straight forward to program in a spreadsheet
• Good choice for 25% to 50% censored data if n greater than 10 or 15.
Simple
Option for
IHSTAT
Sample
Results
(ppm)
8
25
<5
10
<3
7
11
90. Parametric Censored Data Analysis Methods
(Assumes Lognormal Distribution)
• Simple Substitution - DL/2 or DL/sqrt(2)
• Very easy to implement
• Reasonable performance [particularly DL/sqrt(2) for 95%ile estimation] for low n (<20)
and low (<25%) to moderate (25-50%) censoring.
• Maximum Likelihood Estimates (MLE)
• Complex calculations
• Closest to best universal method
• Beta Substitution
• Straight forward to program in a spreadsheet
• Performance similar to MLE
• Log-Probit Regression (LPR) - also called Regression on Order Statistics (ROS)
• Straight forward to program in a spreadsheet
• Good choice for 25% to 50% censored data if n greater than 10 or 15.
• Bayesian Decision Analysis
• BDA uses same equations as MLE
• Superior performance for characterizing parameter uncertainty
• Can readily analyze censored data, including fully censored datasets
Simple
Option for
IHSTAT
Sample
Results
(ppm)
8
25
<5
10
<3
7
11
91. Parametric Censored Data Analysis Methods
(Assumes Lognormal Distribution)
• Simple Substitution - DL/2 or DL/sqrt(2)
• Very easy to implement
• Reasonable performance [particularly DL/sqrt(2) for 95%ile estimation] for low n (<20)
and low (<25%) to moderate (25-50%) censoring.
• Maximum Likelihood Estimates (MLE)
• Complex calculations
• Closest to best universal method
• Beta Substitution
• Straight forward to program in a spreadsheet
• Performance similar to MLE
• Log-Probit Regression (LPR) - also called Regression on Order Statistics (ROS)
• Straight forward to program in a spreadsheet
• Good choice for 25% to 50% censored data if n greater than 10 or 15.
• Bayesian Decision Analysis
• BDA uses same equations as MLE
• Superior performance for characterizing parameter uncertainty
• Can readily analyze censored data, including fully censored datasets
Simple
Option for
IHSTAT
Sample
Results
(ppm)
8
25
<5
10
<3
7
11
92. Sample
Results
(ppm)
8
25
<5
10
<3
7
11
OEL = 100 ppm
Example:
IHSTAT Analysis of Censored Data Using Simple Substitution:
Detection Limit Divided by Square Root of Two [DL / sqrt(2)]
29% censored
93. Sample
Results
(ppm)
8
25
<5
10
<3
7
11
OEL = 100 ppm
Sample Results
With Substitutions
for Non-Detects
(ppm)
8
25
3.54
10
2.12
7
11
Substitute
𝐷𝐿
2
=
𝐷𝐿
1.4142
Example:
IHSTAT Analysis of Censored Data Using Simple Substitution:
Detection Limit Divided by Square Root of Two [DL / sqrt(2)]
29% censored
94. Sample
Results
(ppm)
8
25
<5
10
<3
7
11
OEL = 100 ppm
Sample Results
With Substitutions
for Non-Detects
(ppm)
8
25
3.54
10
2.12
7
11
Substitute
𝐷𝐿
2
=
𝐷𝐿
1.4142
Example:
IHSTAT Analysis of Censored Data Using Simple Substitution:
Detection Limit Divided by Square Root of Two [DL / sqrt(2)]
29% censored
95. Exposure Rating
Category*
0
(<1% of
OEL)
1
(<10% of
OEL)
2
(10-50%
of OEL)
3
(50-100%
of OEL)
4
(>100% of
OEL)
Multiples of OEL
(>500% of OEL
or others based
on respirator
APF)
Recommended Control No action Procedures and
Training; General
Hazard
Communication
+ Chemical
Specific
Hazard
Communicatio
n; Periodic
Exposure
Monitoring,
+ Required
Exposure
Monitoring,
Workplace
Inspections to
Verify Work
Practice Controls;
Medical
Surveillance,
Biological
Monitoring
+ Implement
Hierarchy of
Controls;
Monitoring to
Validate
Respirator
Protection Factor
Selection.
+Immediate
Engineering
Controls or
Process Shut
Down, Validate
Acceptable
Respirators
OEL
95%ile 95%ile
UCL
0
0.005
0.01
0.015
0.02
0.025
0.03
0 20 40 60 80 100 120 140 160 180 200
OEL = 100 ppm
Sample Results
DL/sqrt(2) Substitution
for Non-Detects
(ppm)
8
25
<5 3.54
10
<3 2.12
7
11
GM = 7.35 ppm
GSD = 2.3
95%ile =27.4 ppm
UTL95%,95% = 112 ppm
95%ile Most Likely
in Category 2
(Low Certainty)
Less Than 95% Confident That
True 95%ile Exposure <OEL
29% censored
97. Example:
BDA Analysis of Censored Data
No Substitution
Needed
Enter Directly Into Bayesian
Statistical Analysis Tool
(IHDA or Expostats)
Sample
Results
(ppm)
8
25
<5
10
<3
7
11
OEL = 100 ppm
29% censored
Expostats
98. Learn More: Censored Data References
• IHDA Help File
• Hewett, P. Appendix VIII: Analysis of Censored Data. A Strategy for Assessing and Managing
Occupational Exposures. 4th Ed. AIHA Press. 2015.
• Hewett, P., and G. Ganser. “A Comparison of Several Methods for Analyzing Censored
Data”. Ann. Occup. Hyg., Vol. 51, No. 7, pp. 611–632, 2007
• Ganser, G. and P. Hewett. “An Accurate Substitution Method for Analyzing Censored Data”.
Journal of Occupational and Environmental Hygiene, 7:4, 233-244, 2010.
• Huynh, Tran, Harrison Quick, Gurumurthy Ramachandran, Sudipto Banerjee, Mark Stenzel,
Dale P Sandler, Lawrence S Engel, Richard K Kwok, Aaron Blair, and Patricia A Stewart. “A
Comparison of the β-Substitution Method and a Bayesian Method for Analyzing Left-
Censored Data.” The Annals of Occupational Hygiene 60, no. 1 (January 2016): 56–73.
• Helsel, D. Non Detects and Data Analysis - Statistics for Censored Environmental Data.
Hoboken, NJ: John Wiley & Sons, Inc., 2005.
• Helsel, Dennis R. Statistics for Censored Environmental Data Using Minitab and R
(CourseSmart). Wiley, 2012.
99. CLICK TO EDIT MASTER TITLE
STYLE
99
aiha.org | 99
REVIEW:
DATA
INTERPRETATION
100. REVIEW: Steps in Data Analysis and Interpretation*
100
*After Executing a Carefully
Defined Monitoring Plan:
• Defined decision statistic
• Well defined SEG
• Appropriate OEL
• Well described exposure question
• Appropriate sampling strategy
• Valid and appropriate monitoring
method
• Validated analytical method
100
Preparation:
Collect High Quality Data
101. *After Executing a Carefully
Defined Monitoring Plan:
• Defined decision statistic
• Well defined SEG
• Appropriate OEL
• Well described exposure question
• Appropriate sampling strategy
• Valid and appropriate monitoring
method
• Validated analytical method
REVIEW: Steps in Data Analysis and Interpretation*
1. Enter Data Into Appropriate Statistical Tool
2. Evaluate the Goodness-of-fit
101
101
Evaluate Lognormal Assumption
102. *After Executing a Carefully
Defined Monitoring Plan:
• Defined decision statistic
• Well defined SEG
• Appropriate OEL
• Well described exposure question
• Appropriate sampling strategy
• Valid and appropriate monitoring
method
• Validated analytical method
REVIEW: Steps in Data Analysis and Interpretation*
1. Enter Data Into Appropriate Statistical Tool
2. Evaluate the Goodness-of-fit
3. Review Descriptive and Inferential Statistics
Compare…
• the “decision statistic” (e.g., sample 95th percentile) to the OEL.
• the 95%UCL to the OEL.
102
102
Compare 95%ile Estimate
and 95%UCL to OEL
OEL
95%ile
95%ile UCL
0 1 2 3 4
-0.02
0
0.02
0.04
0.06
0.08
0.1
0 20 40 60 80 100 120
103. REVIEW: Steps in Data Analysis and Interpretation*
1. Enter Data Into Appropriate Statistical Tool
2. Evaluate the Goodness-of-fit
3. Review Descriptive and Inferential Statistics
Compare…
• the “decision statistic” (e.g., sample 95th percentile) to the OEL.
• the 95%UCL to the OEL.
4. Assign a Final Rating and Certainty Level
• Final Rating: Compare the sample 95th percentile to the Exposure
Control Categories (ECCs) and select a category.
• Certainty Level: Compare the 95%UCL to the ECCs:
• Low certainty if > 2 categories above the chosen ECC
• Medium certainty if only 1 category above
• High certainty if within chosen category
103
*After Executing a Carefully
Defined Monitoring Plan:
• Defined decision statistic
• Well defined SEG
• Appropriate OEL
• Well described exposure question
• Appropriate sampling strategy
• Valid and appropriate monitoring
method
• Validated analytical method
Hewett’s
ROT
103
Assign Exposure Rating and Certainty
OEL
95%ile 95%il
e
UCL
01 2 3 4
-0.02
0
0.02
0.04
0.06
0.08
0.1
0 20 40 60 80 100 120
Category 2
Medium Certainty
104. REVIEW: Steps in Data Analysis and Interpretation*
1. Enter Data Into Appropriate Statistical Tool
2. Evaluate the Goodness-of-fit
3. Review Descriptive and Inferential Statistics
Compare…
• the “decision statistic” (e.g., sample 95th percentile) to the OEL.
• the 95%UCL to the OEL.
4. Assign a Final Rating and Certainty Level
• Final Rating: Compare the sample 95th percentile to the Exposure
Control Categories (ECCs) and select a category.
• Certainty Level: Compare the 95%UCL to the ECCs:
• Low certainty if > 2 categories above the chosen ECC
• Medium certainty if only 1 category above
• High certainty if within chosen category
5. Document the Analysis and Recommendations
Recommend controls and/or PPE; work practice evaluation; additional
sampling; surveillance sampling, etc.
104
*After Executing a Carefully
Defined Monitoring Plan:
• Defined decision statistic
• Well defined SEG
• Appropriate OEL
• Well described exposure question
• Appropriate sampling strategy
• Valid and appropriate monitoring
method
• Validated analytical method
Hewett’s
ROT
104
Document Results and Implement Follow-Up Actions
106. Key References
• Papers:
• Hewett, P., Logan, P., Mulhausen, J., Ramachandran, G., and Banerjee, S.: “Rating Exposure Control
using Bayesian Decision Analysis”, Journal of Occupational and Environmental Hygiene, 3: 568–581,
2006
• Logan P., G. Ramachandran, J. Mulhausen, S. Banerjee, and P. Hewett “Desktop Study of Occupational
Exposure Judgments: Do Education and Experience Influence Accuracy?” Journal of Occupational and
Environmental Hygiene, 8:12, 746-758, 2011.
• Logan P., G. Ramachandran, J. Mulhausen, and P. Hewett:” Occupational Exposure Decisions: Can
Limited Data Interpretation Training Help Improve Accuracy?” Annals of Occupational Hygiene, Vol. 53,
No. 4, pp. 311–324, 2009.
• Vadali, M. G. Ramachandran, J. Mulhausen, S. Banerjee, "Effect of Training on Exposure Judgment
Accuracy of Industrial Hygienists”. Journal of Occupational & Environmental Hygiene. 9: 242–256, 2012.
• Vadali, M., G. Ramachandran, and J. Mulhausen, “Exposure Modeling in Occupational Hygiene Decision
Making”, Journal of Occupational and Environmental Hygiene: 6,353 — 362, 2009.
106
107. Key References - Continued
• Papers:
• Hewett, P., and G. Ganser. “A Comparison of Several Methods for Analyzing Censored Data”. Ann.
Occup. Hyg., Vol. 51, No. 7, pp. 611–632, 2007
• Ganser, G. and P.Hewett. “An Accurate Substitution Method for Analyzing Censored Data”. Journal of
Occupational and Environmental Hygiene, 7:4, 233-244, 2010.
• Arnold S., M. Stenzel, D. Drolet, G. Ramachandran; Journal of Occupational and Environmental Hygiene,
13, 159-168, 2016
• Jérôme Lavoué, Lawrence Joseph, Peter Knott, Hugh Davies, France Labrèche, Frédéric Clerc, Gautier
Mater, Tracy Kirkham, “Expostats: A Bayesian Toolkit to Aid the Interpretation of Occupational Exposure
Measurements”, Annals of Work Exposures and Health, Volume 63, Issue 3, April 2019, Pages 267–279,
• Books:
• A Strategy for Assessing and Managing Occupational Exposures. 4th Ed. AIHA Press. 2015.
107
108. Key Resources
• IH Stats and Other Exposure Assessment Tools:
AIHA Exposure Assessment Strategies Committee Website “Tools and Links for
Exposure Assessment Strategies”:
https://www.aiha.org/public-resources/consumer-resources/topics-of-interest/ih-apps-tools
• BDA Computer Tool:
Paul Hewett’s Website - Exposure Assessment Solutions:
http://www.EASInc.co
• ExpoStats Bayesian IH Data Analysis Tools
http://expostats.ca/site/en/index.html
• Competency Demonstration:
AIHA Exposure Decision Analysis Registry
108
109. Exposure Decision Analysis:
Competency Assessment
Exposure Decision Criteria
• Allowable Exceedance
• Needed Confidence
• Use of Exposure Categories
Traditional Industrial Hygiene Stats
• Properties of a lognormal distribution
• Upper percentile estimate calculation & interpretation
• Tolerance Limit calculation & interpretation
Bayesian Decision Analysis (BDA)
• Properties of a lognormal distribution
• Upper percentile estimate calculation & interpretation
• Tolerance Limit calculation & interpretation
Data and Similar Exposure Groups (SEGs)
• Rules for combining data
• Indications that a SEG may need refining
Decision Heuristics and Human Biases
• Common sources of bias in data interpretation and
exposure assessment
• How to avoid bias in data interpretation
Exposure Data Interpretation
• Most likely exposure category given data
• Meet the certainty requirement given data
Techniques for Improving Professional Judgments
• Feedback loops (quantitative judgment > monitoring >
qualitative judgment)
• Group judgment sessions
• Documentation of rationale
• Break decisions into aggregate parts (Modeling)
109
110. AIHA VIDEO SERIES:
MAKING ACCURATE EXPOSURE RISK
DECISIONS
Video 1A
Exposure Variability and the Importance of
Using Statistics to Improve Judgements
110
John R. Mulhausen, PhD, CIH, CSP, FAIHA
111. AIHA VIDEO SERIES:
MAKING ACCURATE EXPOSURE RISK DECISIONS
Video 1A: Exposure Variability and the Importance of Using Statistics to
Improve Judgements
Video 1B: Rules of Thumb for Interpreting Exposure Monitoring Data
Video 2: Introduction to Bayesian Statistical Approaches and Their
Advantages
Video 3A: Free Bayesian Statistical Tools: IHDA Student Edition
Video 3B: Free Bayesian Statistical Tools: Expostats
Video 4: Implementing AIHA Strategy Using Statistical Tools: Examples
111
Join us for the next video in the series . . .