OSHA’s Sorry Justification for Repealing Recordkeeping Requirements

Neither government shutdowns, requests for meetings by stakeholders, nor lack of evidence can stop an agency like OSHA from rolling back a regulation that would have increased our understanding of workplace injuries and illnesses, and made work safer for millions of American workers. As Robert Weissman of Public Citizen observed, “Although shut down, the Office of Information and Regulatory Affairs is apparently still open for business – that is, for business:

Last week, the White House Office of Management and Budget cleared OSHA’s rollback of significant parts of the Obama administration’s electronic recordkeeping regulation and today the text of that regulation was released. The rollback, nothing less than a cover-up of worker injury and illness data at the behest of the Chamber of Commerce and friends, was rushed through OSHA and OMB review in record time. The rush is apparent from reading the regulation — numerous paragraphs and argument are repeated, evidence is incorrectly cited, evidence is lacking and the arguments are laughable.

Furthermore, because of the rush or the government shutdown, the AFL-CIO’s request for a meeting with the Office of Information and Regulatory Affairs was ignored — an unprecedented action that resulted in the labor federation’s request to recall the regulation until a meeting takes place.

Furthermore, the Federal Register is scheduled to officially publish the regulation this morning, despite the Register’s government shutdown guidance document that states that

"Under DOJ guidance received January 11, 2019, we are also allowed to publish documents from funded agencies if delaying publication until the end of the appropriations lapse would prevent or significantly damage the execution of funded functions at the agency.  Agencies must submit a letter certifying that delaying publication of their documents would result in this situation. This certification provides OFR with documentation that publication in the Federal Register is a function or service excepted under the Antideficiency Act."

What “significant damage” will OSHA suffer if recordkeeping requirements aren’t rolled back today?


In order to fulfill the purpose of the Act, OSHA is authorized to require employers to record injury and illness information, as well as collect information on safety and health from employers, including injury and illness information. Consequently, OSHA has long required certain employers to keep injury and illness logs; specifically, OSHA Forms 300, 300A, and 301OSHA Form 300 is a log of all injuries and illnesses, including names of worker injured or sickened. The 300A Form is a summary of that form that contains no names or confidential information. Employers are required to post that information in the workplace between February 1 and April 30 of every year. The 301 Form is a detailed description of workplace injuries that includes information about what happened and what the employee was doing just before the incident occurred.

On May 12, 2016, OSHA amended its recordkeeping regulation to require employers to annually submit to OSHA, by electronic means, injury and illness information that employers were already required to keep. Establishments with 250 or more employees in industries that are routinely required to keep records are required to electronically submit information from their OSHA Forms 300, 300A, and 301 to OSHA or OSHA’s designee once a year. Small businesses with 20 to 249 employees in certain designated industries are only required to submit information on the summary form 300A. Employers were already required to collect this information. The only change was that they were now required to send it into OSHA through a web-based electronic system that OSHA would develop.

Until the new regulation was issued, OSHA did not — with one exception — require any of that information to be sent in to OSHA. The main purpose of requiring employers to collect the information was to help them improve their health and safety programs, and OSHA Inspectors consulted the information when conducting inspections.  The one exception was OSHA’s collection of injury and illness summary information from around 80,000 employers every year, a program that lasted from 2006 to 2013 and was used to better target OSHA inspections in the most dangerous workplaces.

The 2016 rule had two phases. The first phase was submission of the Form 300A summary data which, after several delays, took effect last year. The second part, sending in information from the 300 Form and the more detailed 301 data, was supposed to take effect this year. Submission of the 300 Log and the detailed Form 301 data are the requirements that OSHA is canceling.

Why Was The Regulation Rolled Back?

OSH’s argument for rescinding these requirements rests almost solely on a false argument that OSHA’s collection of information would put confidential information at risk of exposure:  “OSHA has determined that collecting the data from Forms 300 and 301, as was recently required under the 2016 final rule, would subject sensitive worker information to a meaningful risk of public disclosure.”

I explained last August, after OSHA issued its proposal, why this confidentialty argument is false: In 2016, when the original regulation was issued, OSHA assured the public that all confidential information would be protected. This would include the information on the left side of the 301 Form, specifically workers’ names, birthdates, names of their doctors, etc.  OSHA assured that confidential information couldn’t be accidentally released through a mistake or a data breach, or even by way of a rogue court decision by ensure that employers don’t send confidential information in to OSHA in the first place.

None of the fields on the electronic portal that employers would use to send the data to OSHA would contain fields for confidential information.  Where that couldn’t be avoided with important information like birth dates, the program would automatically convert “birth date” to “age.”

In short, if OSHA is not in possession of confidential information, there’s no way it can be released.

But Trump’s OSHA “no longer views such protections as sufficient.”

In order to undermine the facts, OSHA basically had to twist itself into a pretzel to figure out obscure and far-fetched ways that confidential information could hypothetically still be released. For example:

  • The electronic forms that employers would use to send information to OSHA contained several narrative fields where employers would describe the events leading to the injury. Hypothetically, an employer could inadvertently put confidential information into that field. This was an eventuality that OSHA anticipated in 2016, stating that “OSHA plans to review the information submitted by employers for personally identifiable information. As part of this review, the Agency will use software that will search for, and de-identify, personally identifiable information before the submitted data are posted.”OSHA nevertheless now argues that there is too much data to manually review and alleges — without evidence — that “de-identification software is not 100% effective.”  Therefore “OSHA finds that it would not be able to guarantee that all PII (personally identifiable information) inadvertently submitted to OSHA would be protected from disclosure.”
  • OSHA argues that an employee’s identity could be reverse engineered: “Several data points on Forms 300 and 301 could be combined to reveal the identity of workers who reported work-related injuries or illnesses, particularly in a small town.” According to OSHA, any commenters who disagree “fail to appreciate the real possibility of the disclosure of sensitive worker information.”A recent decision by the  the US District Court for the District of Columbia frowned on OSHA’s use of this excuse (or the excuse above) to withhold the entire database, stating that “an agency withholding records on the ground that they might be connected to a particular individual must show ‘threats to privacy interests more palpable than mere possibilities.'”  The court concludes that “Those rare instances can be addressed on a case-by-case basis.”
  • The agency continues to insist that even though confidential information is exempt from Freedom of Information Act Requests, some rogue court, somewhere in the country, could someday, nevertheless require OSHA to release confidential information. But, as we noted above, even in the extremely unlikely event that a court suddenly goes rogue and isn’t overturned by a higher court, OSHA will not be in possession of confidential information (see above.)
  • Several commenters pointed out that MSHA collects similar data and has never had a problem with inadvertent disclosure of confidential information. OSHA concedes that this is the case, but argues that the OSHA database would be bigger than MSHA’s, making cleansing of the data more burdensome for the agency.
  • OSHA concludes that the risk of violations of privacy are too large compared with the small and uncertain benefits (see below) and anyway, the agency has better things to do with its money — like focus on using the From 300A summary data to better target enforcement.

To summarize, as AFL-CIO Health and Safety Director Peg Seminario said in a statement to Bloomberg BNA after the proposal was issued last July,

OSHA’s claim that it is proposing to revoke employer requirements to submit detailed injury data in order to protect employee privacy is truly cynical.  Workers and worker representatives strongly support the collection of this information to help identify workplaces with serious injuries in order to protect workers health. Only industry groups oppose these common sense requirements. The real reason for this roll back is to protect employers who don’t want workers or the public to know about dangerous conditions and hazards at their workplaces.


OSHA alleges that the original regulation presents real risks to privacy and that “In this final rule, OSHA is balancing the issues of worker privacy and OSHA’s resource priorities against the uncertain incremental benefits of collecting the data from Forms 300 and 301.”

So what are these benefits that OSHA finds so uncertain, and what evidence does OSHA present that undermines the agency’s previous description of the benefits of collecting and publicizing the 300 and 301 data?

Here I summarize comments submitted to the public record by former OSHA director David Michaels, former Deputy Assistant Secretary of MSHA, Greg Wagner and myself, In short they are:

  • Strengthened Enforcement:  Currently, OSHA only sees the 300 and 301 data when it inspects an establishment. The detailed data gives OSHA the ability to understand the patterns of injury/illness causation even at establishments that has not been inspected, enabling the agency to effectively target high-risk industries.
  • Improved Enforcement and Compliance Assistance Targeting: OSHA currently receives valuable information about individual incidents in its Severe Injury Reports where employer must report every hospitalization, amputation or loss of an eye. But these reports miss information valuable to OSHA’s compliance assistance efforts: musculoskeletal disorders, the single most frequent type of workplace injury, and emergency room visits, for example.  Information on the Form 300A summary logs is not sufficient. Simply knowing the overall injury rate tells the agency nothing about the types of injuries that are occurring, or ways to help employers prevent future injuries.
  • Transparency Drives Positive Behavior: The original OSHA rule took advantage of the widely recognized finding that transparency can be a powerful driver of behavior.  Making injury data available to the public would likely “nudge” more dangerous employers to better protect their workers. Employees won’t want to work in more dangerous workplaces and consumers may not want to buy their products.
  • Enabling Employer Benchmarking: Employers can compare their records with their competitors and the more detailed information allows employers to benchmark by job title or department.
  • Research to improve worker safety and health: The data would enable research to be conducted on the causation and prevention of work-related injuries. Researchers could also use the data to evaluate the effectiveness of OSHA enforcement and compliance assistance programs aimed at specific types of injuries, something that cannot be done if the only data available are summary data from the Form 300A. Researchers have used MSHA’s collection of similar data to produce a series of papers that have been useful in understanding the causes of mine injuries and in preventing them from occurring.

What Benefits?

Despite the obvious benefits of the original rule described above, OSHA insists that the benefits are “uncertain” at best. OSHA complains that the agency “has no prior experience using the case-specific data collected on Forms 300 and 301 for enforcement targeting or compliance assistance and is unsure how much benefit such data would have for these purposes or the level of resources needed to attain any benefit.” And anyway, “collecting, processing, analyzing, distributing, and programmatically applying the data would be costly.”

OSHA acknowledges that its sister research agency, the National Institute for Occupational Safety and Health (NIOSH) has kindly offered

to play a leading role in helping OSHA use data collected in Forms 300 and 301 to prevent occupational injuries and illnesses.  NIOSH explained that it has the experience and capacity to analyze the data, as well as interest in using the data to provide guidance to employers for the prevention of occupational injury and illness, and to provide data analysis results and analytical tools that should enhance OSHA’s targeting….NIOSH noted that it has already developed auto-coding methods for categorizing occupation and industry based on free text data and has successfully utilized similar free text data collected from workers’ compensation claims. While NIOSH acknowledged that the data collected from Forms 300 and 301 would pose a greater analysis challenge because of the amount of data, NIOSH stated that the large data set would be useful to identify patterns and prevent workplace injuries.

Sounds good to me. But not to OSHA which responded that

OSHA appreciates the value of inter-agency efforts to achieve shared goals of preventing occupational injuries and illnesses and looks forward to continued coordination with NIOSH and other agencies where appropriate. However, OSHA has determined that NIOSH’s ability to analyze data collected from Forms 300 and 301 does not reduce the burden on OSHA to collect the data. Even if NIOSH could make the data useful for OSHA’s enforcement targeting and outreach efforts….OSHA and employers would be left covering the expense of collection,not to mention additional expense associated with the need to process and otherwise manually review data from the forms – costs that would detract from OSHA’s priorities of enforcement and compliance assistance to reduce workforce hazards.”

In other words, even if benefits exist and even if NIOSH can use the data effectively, it’s still not worth the cost to OSHA.

See No Evil, Hear No Evil

Another source confirming the benefits of collecting the data was the multi year 211 page National Academy of Sciences multi-year study, issued in early 2018 entitled A Smarter National Surveillance System for Occupational Safety and Health in the 21st Century. The purpose of the report, commissioned by OSHA, along with NIOSH and the BLS, was to examine our current systems for surveillance of occupational injury and illness and to make recommendations for improvement. Comments by Michaels, Wagner and myself summarized the report’s recommendations and endorsement of OSHA’s recordkeeping rule:

The new rule provides a much-enhanced source of injury and illnesses data that can be used for effective targeting of interventions and prevention efforts as well as compliance activity focused on hazardous industries, workplaces, exposures, and high-risk groups. Furthermore, these data are not currently available to agencies or the public from other surveys. This employer-based system also provides new opportunities to conduct outreach and build tools and provide assistance to employers to identify and address hazards at individual worksites….

The new rule will provide an extensive new data source regarding injury and illness that can be used by OSHA, NIOSH, state agencies, employers, workers, and researchers for a range of surveillance and prevention purposes… The information collected and available under the electronic reporting rule holds potential value for employers, workers, public health agencies, researchers, and others. Employers will be able to use the information to compare their experience with others in the industry. Workers will be able to have ready access to an employer’s injury reports prior to seeking employment and while employed to assess the safety record of the employer. Public health agencies will be able to determine if there are types of injuries or illnesses occurring in the workplaces of particular industries. Public health departments will be able to initiate intervention efforts, including educational efforts and adjustments to public health standards in industries such as health care facilities, food establishments, or schools, which are regulated by the states. And researchers will have ready access to a large database of injury information to assist them with better characterizing high risks as well as assessing the effectiveness of interventions.

This is a pretty strong endorsement of the original regulation, which should have been especially significant considering that Scott Mugno, President Trump’s nominee to head OSHA sat on the committee that developed the report and signed off on its conclusions.

But despite the fact that the report was released six months before OSHA issued it proposed rollback for comment, the agency failed to even mention the report in the July proposal.

OK, it’s easy to make mistakes. Anyone could overlook a well publicized major report covering the subject of your rulemaking while you’re developing your proposal.

But then we get to the final standard. The NAS study is indeed mentioned several times in the preamble to the new regulation. But OSHA states in a footnote that the agency is responding to the report only when commenters and others “have specifically referenced findings, recommendations, or other statements contained in the report in their comments.”

Otherwise, OSHA would ignore the NAS report “because the report is not, and was not intended to be, commentary on this rulemaking, the agency does not find it is appropriate or necessary to respond to statements contained therein where those statements were not referenced by commenters in their submissions to the record.”

So we have a major report focused exactly the subject of this rulemaking, compiled by the country’s top experts on workplace injury and illness recordkeeping (including the possible next head of OSHA), which strongly endorses the benefits of the original regulation that OSHA is trying to dismantle, but OSHA completely ignored it when developing the proposal and would have continued to ignore it if some commenters hadn’t cited it in their comments.

This is a scientific agency?

Hazards? What Hazards?

Another reason that OSHA argues that the Obama administration was wrong and the original regulation actually had no benefits is that injuries and illnesses allegedly don’t necessarily mean that the law was violated or that unsafe conditions existed in the workplace. OSHA quotes extensively from commenters who argue that injuries are no indication of unsafe conditions and that “According to some commenters, maintaining Form 300 and 301 data electronically would not aid OSHA in identifying, and engaging in enforcement, at high-risk workplaces, or otherwise provide any real value to the agency’s enforcement targeting strategies or decisions.”

And they quote the so-called E Recordkeeping Coalition (an employer organization) which stated that, “[b]ased on a qualitative analysis of [its] members’ 300 and 301 data, only a small percentage of that data would indicate any regulatory compliance insufficiency.”

So worker injuries don’t indicate any violation of OSHA standards (or “regulatory compliance insufficiency.”)  OSHA even uses this specious argument in an attempt to undermine evidence that publication of injury and illness information will “nudge” an employer to make its workplace safer: “because employers are required to report workplace injuries and illnesses regardless of fault, OSHA no longer considers collection of employers’ injury and illness records likely to “nudge” them to make their workplaces safer.”

Or in other words: “All those workers getting hurt in my workplace have nothing to do with unsafe working conditions or violations of OSHA standards. Workers are getting injured because they’re careless or the accidents are just an Act of God. So no matter how much OSHA embarrasses me by publicizing my injury data, there’s nothing I can possibly do to make things better.”

Or something like that.

A Self-Fulfilling Prophesy

OSHA contends that the original regulation had no certain, knowable benefits (despite the evidence presented in the NAS study and numerous other commenters), but even if collection and publication of the information did potentially have the benefits that many commenters described, OSHA would make sure those benefits would never “materialize” for any outside researchers:

OSHA begins by noting that many of the benefits discussed by commenters would not materialize. Because OSHA has determined publishing the data would do more harm than good for reasons described more fully below and in the privacy discussion above, OSHA would not make the data public even if collected. In addition, as noted above, OSHA has already taken the position that data from Form 300A is exempt from disclosure under FOIA and that OSHA will not make such data public for at least the approximately four years after its receipt that OSHA intends to use the data for enforcement purposes. Therefore, the benefits some commenters ascribed to publication of the data would not be realized. Without publication, the research benefits claimed by many commenters also fall away.

Making Shit Up: Data Accuracy

Moving into pure fantasy-land, OSHA states that the real benefits lie not in the original 2016 regulation, but in OSHA’s current repeal of those requirements.

Who may benefit by the repeal, you may ask? Not just employers, but employees as well!

The main benefit from the repeal is allegedly improved data accuracy which will  benefit workers and employers.

How is that?

OSHA suggests first, that employers would be more likely to cheat and hide injuries from the log if they know that their injury and illness data will be made public. So not making the data public will improve accuracy. They cite no evidence to this effect, except employer testimony. (This might be characterized as “The Devil Made Me Do It” defense. Don’t try it in court.)

Then we get even more ridiculous when OSHA argues that workers will also not report accurately if they feel that some of their confidential information will be revealed on OSHA’s website.

OSHA has determined that publishing the data could also cause more harm than good. Workers would know in advance that some details of their injuries would be public and on the internet. Deterring worker reporting through fear of publication could make the records less accurate.

What “evidence” did OSHA have that workers’ fear publication of injury and illness data and will therefore fail to report?   A 2013 comment from an “anonymous” commenter and another 2013 comment that can’t be found or was mis-labeled in the docket.

Nevertheless, OSHA finds that the final rule may ensure more accurate records on Forms 300 and 301 by alleviating employers’ and workers’ fears about the consequences of the records becoming public, and will allow employers to devote more of their resources towards compliance with safety and health standards.

Actually, the main fear that workers expressed during the hearings for the original regulation was the very real fear of retaliation by employers for reporting injury or illness information. That’s why OSHA inserted language into the original regulation that prohibited employers from retaliating against workers for reporting. Thankfully, the current administration left that language untouched, although it has been weakened through interpretations.

Conclusion: Is OSHA’s New Regulation Arbitrary and Capricious

There is much, much more I could go into to show the weakness of OSHA’s arguments, but you get the idea.

It’s sad and sometimes even painfully amusing. But I’m not doing this for my entertainment or your entertainment. The point I’m making is that even I (as a non-lawyer) can recognize sloppy reasoning and lack of evidence. If we had tried to get something like this cleared when I was at OSHA, we would have been laughed out of the Secretary’s office and it would never have gotten close to White House review.

The bottom line is that there are laws that prohibit agencies from trying to get away with weak justifications for rulemaking. Rulemaking in all Executive Branch agencies — including OSHA — is covered by the Administrative Procedure Act, which governs federal rulemaking and ensures that agencies follow some basic steps when they issue new regulations or change existing regulations. As I mentioned yesterday, the New York Times reported that “An analysis by the Institute for Policy Integrity at New York University School of Law shows that more than 90 percent of court challenges to major Trump deregulatory actions have been successful so far. By the institute’s count, 30 big rules have been challenged, and the courts have found for the litigants 28 times.”

But “In a typical administration, the government wins on such challenges around 70 percent of the time.” This administration, as we know, is hardly typical.

Agencies that want to issue or change regulations must do so based on evidence presented during a notice and comment period. Otherwise, if the court finds that the agency’s decision is “arbitrary and capricious,” the regulation can be thrown out.  For example, a court recently rejected the Trump administration’s attempt to add a question on American citizenship to the 2020 census, not because it’s a bad idea (which it is), but because

Judge Jesse M. Furman of the United States District Court in Manhattan said that Wilbur L. Ross Jr., the commerce secretary, broke “a veritable smorgasbord” of federal rules when he ordered the citizenship question added to the census nearly a year ago. Judge Furman said Mr. Ross cherry-picked facts to support his views, ignored or twisted contrary evidence and hid deliberations from Census Bureau experts.

This rollback will be challenged in court. And if previous court rulings are any indication, and we get good judges, it will be tossed out. Stay tuned.


Reposted from Confined Space