Advocacy & Safety - Bicyclists’ Injuries and the Cycling Environment: Findings

http://cyclingincities.spph.ubc.ca/files/2011/10/BICEmethods1.png

Here are the results of a 2 year study involving 690 cyclists in Vancouver and Toronto, who were injured while cycling and wound up at an emergency room because of it. The study aimed to find which infrastructure features contribute to safety, and which ones cyclists prefer. The results were published in the American Journal of Public Health (http://ajph.aphapublications.org/doi/pdf/10.2105/AJPH.2012.300762) and elsewhere (http://www.biomedcentral.com/1471-2458/12/765/abstract).


Of the 690 injured cyclists in the study, 59% were male. The injury trips were mainly on weekdays (77%), less than 5 km long (68%), and for utilitarian purposes (74%). Of the injury events, 72% were collisions (with motor vehicles, route features, people, or animals) and 28% were falls.

We found that route infrastructure does affect the risk of cycling injuries. The most commonly observed route type was major streets with parked cars and no bike infrastructure. It had the highest risk. In comparison, the following route types had lower risks (starting with the safest route type):


cycle tracks (bike lanes physically separated from motor vehicle traffic) alongside major streets (about 1/10 the risk)
residential street bike routes (about 1/2 the risk)
major streets with bike lanes and no parked cars (about 1/2 the risk)
off-street bike paths (about 6/10 the risk)

The following infrastructure features had increased risk:

streetcar or train tracks (about 3 times higher than no tracks)
downhill grades (about 2 times higher than flat routes)
construction (about 2 times higher than no construction)

Interesting. I take a major street on my commute to work, with bike lanes. The only general hazards I see on a daily basis are:
1) Right hooks (common, almost daily risk).
2) Passing too close (usually city buses!)
3) Occasional red light runners (not much of a risk for me, since I'm not exactly jumping through lights on my bike, more indirect if through traffic gets hit, they might come into the bike lane).

This study may be fine if the goal is counting up number totals of "accidents" but is worthless if the intent is to evaluate the effect of various safety equipment or facility infrastructure on cycling risk. Risk cannot be properly evaluated without measuring, or at least considering, the severity level of the accident events.

Any and all conclusions about cycling risk from this study are fatally flawed by the absence of any data about the severity of the injuries incurred by the study subjects. The term "risk" should not be associated with this study. Trips to the emergency does not filter out minor injuries, nor does it distinguish them from catastrophic injury. An accident resulting in a scratched knee or broken finger from a "fall' is tallied the same as catastrophic injuries suffered from a collision with a high speed automobile.

Any and all conclusions about cycling risk from this study are fatally flawed by the absence of any data about the severity of the injuries incurred by the study subjects.

It must be lonely being the smartest being in a universe full of idiots. How do you cope?

We found that route infrastructure does affect the risk of cycling injuries. The most commonly observed route type was major streets with parked cars and no bike infrastructure. It had the highest risk. In comparison, the following route types had lower risks (starting with the safest route type):


cycle tracks (bike lanes physically separated from motor vehicle traffic) alongside major streets (about 1/10 the risk)
residential street bike routes (about 1/2 the risk)
major streets with bike lanes and no parked cars (about 1/2 the risk)




This is really interesting, given that the common wisdom among cycling cognoscenti is that separated bike lanes are a bad idea. Also that major streets with bike lanes have accident rates similar to residential streets (I would expect the major street accident rates to be higher, bike lanes or no.)

It must be lonely being the smartest being in a universe full of idiots. How do you cope?

+1

It must be lonely being the smartest being in a universe full of idiots. How do you cope?

If you are happy being ignorant about evaluating risk, join the crowd who spout BS conclusions about risk from BS studies. You have plenty of company.
Your confusion about a bogus term such as "accident rates" which considers all accidents as equal, being relevant to risk comparisons/evaluation is typical. Enjoy being blissfully ignorant about risk evaluation.

From a scientific standpoint, you would indeed require more information to properly assess risk. Practically speaking, accident rates give you a pretty good handle on risk.

If you are happy being ignorant about evaluating risk, join the crowd who spout BS conclusions about risk from BS studies. You have plenty of company.
Your confusion about a bogus term such as "accident rates" being relevant to risk comparisons/evaluation is typical.

How about doing your own study, as you seem to be the resident expert on what is risky or not. You have denied reports/stories/opinions here on BF on everything from Styrofoam hats to cell phone distracted drivers to cycling facilities, to headphone use, and even Forester's "opinions."

We have come to see that whatever it is, ILTB will show up on here and tell us why something is invalid due to "safety nannies over reach."

But you have become like the boy that cried wolf... no one values your opinion as all it is, is negative, and all it does is blast the studies and opinions of others... while offering nothing constructive, nor do you point out specifics.

You just cry WOOF... (yeah a pun.)

Perhaps we should change P&R to "Why ILTB thinks you are wrong."

Certainly the studies discussed here may be flawed... but they are the best data available. Now perhaps you should contact all these organizations and tell them how they should do their studies. Apparently no one ever taught them... according to ILTB.

Oh and it is only a matter of time before Forester shows up, and does his own brand of "the study is flawed."

Meanwhile cyclists here are hoping to learn something from what ever data may be available, and is relatively current.

From a scientific standpoint, you would indeed require more information to properly assess risk. Practically speaking, accident rates give you a pretty good handle on risk.

Hardly. One can conclude that riding a bicycle in the park is riskier than riding in high speed traffic if more accidents occur. "Works for risk evaluation" only if you ignore the severity of the resulting accidents.

Certainly the studies discussed here may be flawed... but they are the best data available. Now perhaps you should contact all these organizations and tell them how they should do their studies. Apparently no one ever taught them... according to ILTB.
"Certainly the studies discussed here may be flawed..." Exactly what I said and causes your bowels to get in uproar.

"best data available" is no excuse for drawing conclusions based on ignorance of the necessary data. You sound exactly like Forester now.

The problem is that all the studies quoted on BF about bicycling risk are substituting bastard terminology such as "crash rates" and "accident rates" without any of the necessary severity data essential for evaluation of risk. And it isn't worth a dang. No matter how much you wish it were so. It ain't smart to be proud of being ignorant.

From a scientific standpoint, you would indeed require more information to properly assess risk. Practically speaking, accident rates give you a pretty good handle on risk.

Practically speaking, lots of BF posters have what they consider "a pretty good handle on risk" without any information or data, but their own opinion, conjecture and/or guesswork. And it may be just as useful as half-assed "studies" that neglect to gather or consider essential data measurements.

Hardly. One can conclude that riding a bicycle in the park is riskier than riding in high speed traffic if more accidents occur. "Works for risk evaluation" only if you ignore the severity of the resulting accidents.

Severity sure matters if I'm the one wrecking.

I want to know where all of the riders who don't get injured are riding. I guess it's harder to get data from them.

From a scientific standpoint, you would indeed require more information to properly assess risk. Practically speaking, accident rates give you a pretty good handle on risk.

Not really. I-like-to-bike is right. However, if we pair this study with a Dutch study (which I can't find at the moment) showing that the fatality risk is much higher between intersections when biking in car lanes than IN intersections when biking on bike paths I think it gets interesting.

How about doing your own study, as you seem to be the resident expert on what is risky or not. You have denied reports/stories/opinions here on BF on everything from Styrofoam hats to cell phone distracted drivers to cycling facilities, to headphone use, and even Forester's "opinions."

We have come to see that whatever it is, ILTB will show up on here and tell us why something is invalid due to "safety nannies over reach."

But you have become like the boy that cried wolf... no one values your opinion as all it is, is negative, and all it does is blast the studies and opinions of others... while offering nothing constructive, nor do you point out specifics.

You just cry WOOF... (yeah a pun.)

Perhaps we should change P&R to "Why ILTB thinks you are wrong."

Certainly the studies discussed here may be flawed... but they are the best data available. Now perhaps you should contact all these organizations and tell them how they should do their studies. Apparently no one ever taught them... according to ILTB.

Actually, it's just a matter of time before someone settles the discussion by telling us that all those on-street accidents are the result of not taking the lane and/or being competent.

"Certainly the studies discussed here may be flawed..." Exactly what I said and causes your bowels to get in uproar.

"best data available" is no excuse for drawing conclusions based on ignorance of the necessary data. You sound exactly like Forester now.

The problem is that all the studies quoted on BF about bicycling risk are substituting bastard terminology such as "crash rates" and "accident rates" without any of the necessary severity data essential for evaluation of risk. And it isn't worth a dang. No matter how much you wish it were so. It ain't smart to be proud of being ignorant.

But the problem is you are complaining to the folks drinking the kool-aid that the kool-aid is not made properly... you should be talking to the kool-aid makers.

We are just the consumers. Of course no doubt you will tell us to just back away from the kool-aid... but how, when it is the only thing available?

Hey Seattle Forest -- thanks for posting this.

BTW, the second "elsewhere" link you provided links to "Personal and trip characteristics associated with safety equipment use by injured adult bicyclists: a cross-sectional study" and I couldn't find the route-specific information. It seemed to be more focused on helmet and light usage.

Putting aside the question of severity, do you know what the authors mean by "risk"? Is it the probability that a cyclist using a particular route will have an accident? Or that an accident on a particular route is more likely to result in an emergency room injury?

I don't understand why they are comparing to "control" sites or concerned about "observer bias" if they are measuring rates (which is why I'm asking about how they define risk). Seems like this would be an empirical discussion and that they would be measuring traffic volume on the various routes.

Another methodology link

http://cyclingincities.spph.ubc.ca/files/2011/10/HarrisTRBposter.pdf

Another methodology link

http://cyclingincities.spph.ubc.ca/files/2011/10/HarrisTRBposter.pdf

Exactly what I was looking for, and addresses the concern I had but was unable to articulate well: Accurately characterizing the denominator of persons at risk at each site (e.g. number of people riding past the location). So it seems that the study probably assessed the likelihood of an injury accident occuring on particular types of routes. And my question regarding "observer bias" is answered -- it has to do with how routes are classified. It seems they have also addressed the issue of confounding, which had not occurred to me.

ILTB, you raise the issue of failing to account for injury severity. I had two thoughts on that -- (1) given that all the cases involve an ER visit, the range of severity is somewhat narrowed (acknowledging that there still remains a wide gap between road rash and paralysis), and (2) I would presume that the raw interview data is still be available, from which severity could be discerned. It also occurs to me that the interview scenarios could be further analysed to identify outliers (e.g., he was lucky he only had a skinned elbow, because he could have died, etc.).

I would presume that the raw interview data is still be available, from which severity could be discerned.
Probably so, but the the people who put this report together chose not to use any such data. Which means they are ignorant of how to measure/evaluate risk (unlikely) or their agenda/conclusion could be "confirmed" without.

given that all the cases involve an ER visit, the range of severity is somewhat narrowed (acknowledging that there still remains a wide gap between road rash and paralysis)
Given the advice often given to accident victims, especially if there is any possibility the head had any impact, to get themselves "checked out", the range of "injuries" seen at an ER can range from none at all to fatal. But for this study, there is no distinction , all accidents are the same.

With this study's methodology, if the "accident rate" in the park is an identical number to the "accident rate" on the road, both scenarios are equally "risky" and it would be immaterial if the worse accidents occurring in the park were minor boo-boos from simple falls and all the accidents on the road were high speed collisions resulting in permanent paralyzing injuries.

Actually, it's just a matter of time before someone settles the discussion by telling us that all those on-street accidents are the result of not taking the lane and/or being competent.

Unless they are struck while in the lane, then they were not "competent cyclists." They obviously failed at some other tenet of competent cycling.

By the power of circular reasoning, Vehicular Cycling™ Brand "competent cyclists" are never involved in traffic mishaps because they never make mistakes and are never in the "wrong" place at the "wrong" time. If a cyclist has an accident or collision he/she obviously was not a competent cyclist.

In addition to the weaknesses outlined by ILTB, this study has several other significant problems. Among those are; self reporting by the accident victim of all of the questions and lack of an external control group. The former leads to the datum, that only about 11% of the accident victims had any alcohol in the six hours prior to the accident. I wonder how similar the result would be if one were to ask folks involved in a DUI accident if they had any alcohol in the six hours prior to the accident... And the problem with the latter is illustrated by the self-selection bias, ie, only those involved in an accident are included in the control group...

I am also concerned by the removal of nearly 40% of the potential bike accident victims as 'ineligable' for a number of reasons. One of which seems inexplicable; not being a resident. The other is also somewhat concerning; not being injured in the city... Both factors would seem to introduce unnecessary biases.

In addition to the weaknesses outlined by ILTB, this study has several other significant problems.
The significant problems that you outlined compound the lack of validity of the conclusions about risk from this so-called study. But none of the weaknesses or errors mean a thing to self described "cycling cognoscenti" who pride themselves on being know-nothing morons about evaluating or measuring cycling risk. They have the answer they were looking for - mission accomplished.

Perhaps we should change P&R to "Why ILTB thinks you are wrong."


I snorted.

I snorted.

Beats thinkin', don't it, eh?

Beats thinkin', don't it, eh?

What an odd thing to say.

I'd like to see the whole study as the information from the abstracts is more or less useless. Reading the information in the two links we have determined:

- Fewer bicycle accidents happen on quiet streets where there are bike lanes and no parked cars
- Bike paths separated from motor vehicle lanes are safer for the cyclist
- Wearing bright colored clothing may help prevent accidents
- People use bike lights more at night than during daylight hours
- More riders wear a helmet when it is mandated by law
- More experienced riders are less likely to be in an accident

It appears this study cast a very wide net for very non-specific data. They data mined ER patient records for the information and "recruited" riders who had been in the ER due to a cycling accident. The complete report might shed more light on their methodology but if the abstract is any indication, their cross section is not likely an accurate representation and their analysis is questionable. For example, the total number of accidents on city streets is compared directly to the number of accidents on bike paths without reference to the total number of miles traveled on each. The fact that half as many accidents happened on bike paths becomes irrelevant if only half as many miles were traveled on them. The more relevant expression of the data would be accidents per 100 miles traveled or something similar. If you don't place the data in proper context (apples to apples) and include all appropriate data sets, it really isn't useful and can be outright misleading.

While I decline to join in the pizzing contest, I have to agree that type and severity of injuries cross referenced with contributing factors would be an important part of the big picture. For example: If there were 50% more head injuries on busy city streets is it because a) there were more collisions with motor vehicles which are more likely than other types of accidents to result in head injury, b) the average speed was higher resulting in harder blows to the head, or c) fewer riders were wearing helmets on the city streets than were wearing them on the MUP? Another example would be to subgroup the injuries from each category of infrastructure (ie. street, bike lane, bike path, etc) by severity such as minor, moderate, severe, and life threatening. Casting a broad net to cover all types of injuries related to bicycles under multiple conditions with multiple contributing factors makes for an unweildy amount of data from which it is difficult, if not impossible, to draw direct correlations. Basic research methodology requires that you limit the number of variables in a given study to those that can be accurately isolated and measured. Just from the "conclusions" offered I can see they were dealing with numerous variables. I look at the conclusions presented in the abstracts which are linked to the OPs post and my first response is "duh". Generalized data results in generalized observations which is not good science.


99.5% of all people who ate carrots in 1915 are now dead, therefore we conclude that there must have been something wrong with the carrots served in 1915.

I'd like to see the whole study as the information from the abstracts is more or less useless.
Attached for your reading pleasure.
http://www.biomedcentral.com/content/pdf/1471-2458-12-765.pdf

Another major problem with the methodology, related to the researcher's wacky filtering out of subjects to interview. They didn't include injured cyclists with injuries so severe that they couldn't describe/remember the particulars of the accident.

Originally Posted by Daves_Not_Here http://www.bikeforums.net/images/buttons/viewpost-right.png (http://www.bikeforums.net/showthread.php?p=14872462#post14872462) I would presume that the raw interview data is still be available, from which severity could be discerned.



Probably so, but the the people who put this report together chose not to use any such data. Which means they are ignorant of how to measure/evaluate risk (unlikely) or their agenda/conclusion could be "confirmed" without.

The study's interview form did ask a very relevant question: "Were you admitted to hospital, in other words, did you stay overnight in a ward other than the Emergency Department? Yes or No"

The results could have been used to draw relevant conclusions about injury severity and its relationship to the various scenarios, but they chose to ignore the relevant data on injury severity that they gathered.

Interesting to note is the absence of what the risk factor is with bike lanes on major streets, alongside parked cars, especially when a BL is located on the driver's side of a motor vehicle.

This is really interesting, given that the common wisdom among cycling cognoscenti is that separated bike lanes are a bad idea. Also that major streets with bike lanes have accident rates similar to residential streets (I would expect the major street accident rates to be higher, bike lanes or no.)

I blame dogma for the cognoscenti's commonly accepted wisdom.

People are making wild pronouncements in this thread, which is no surprise. There are also questions, like exactly what "risk" means in this study. There's contact info, including an email address, in the link to the article in the American Journal of Public Health.

:troll:

And how many articles have you had published in prestigious journals like this?

I blame dogma for the cognoscenti's commonly accepted wisdom.

People are making wild pronouncements in this thread, which is no surprise. There are also questions, like exactly what "risk" means in this study. There's contact info, including an email address, in the link to the article in the American Journal of Public Health.

I know what risk is. I developed the course and taught hazard analysis techniques for several years in an industrial setting. No valid or credible conclusions can be drawn from the crippled methodology described in the study. There is nothing to ask this group. The interesting sidenote is that the so-called studies refered to by the dogmatic cognoscenti, whose commonly accepted wisdom came to different conclusions about "crash rates" is just as crippled as this study (no injury severity considered) and is just as worthless for drawing risk evaluation conclusions. But since it fits your agenda, as it obviously does the "researchers," keep believing in BS.

Attached for your reading pleasure.
http://www.biomedcentral.com/content/pdf/1471-2458-12-765.pdf

Another major problem with the methodology, related to the researcher's wacky filtering out of subjects to interview. They didn't include injured cyclists with injuries so severe that they couldn't describe/remember the particulars of the accident.

Just to be clear, this is not the study that started this thread. That one, in the American Journal of Public Health, isn't available to non-subscribers yet (even through university databases). Gotta wait for the next issue to be released.

Attached for your reading pleasure.
http://www.biomedcentral.com/content/pdf/1471-2458-12-765.pdf

Another major problem with the methodology, related to the researcher's wacky filtering out of subjects to interview. They didn't include injured cyclists with injuries so severe that they couldn't describe/remember the particulars of the accident.

But where hte location could still be determined, right?

What bothers me far more however is that as far as I can see they only looked at riders who had accidents. How many rode the same route and did not have accidents would seem a rather obvius piece of data that is needed to make a meaningful analysis.

But where hte location could still be determined, right?
As best as I can tell accidents weren't tallied or considered unless the "researchers" got an interview with the cyclist involved, and the injured cyclist fit into the population they wished to include in the survey.

Just to be clear, this is not the study that started this thread. That one, in the American Journal of Public Health, isn't available to non-subscribers yet (even through university databases). Gotta wait for the next issue to be released.
What makes you think this is a different study? The PDF for this study is found on the URL furnished in the OP: http://www.biomedcentral.com/1471-2458/12/765/abstract

(http://www.biomedcentral.com/1471-2458/12/765/abstract)"The complete article is available as a provisional PDF (http://www.biomedcentral.com/content/pdf/1471-2458-12-765.pdf). The fully formatted PDF and HTML versions are in production. "

All I can say is that I'm very happy everybody likes the word cognoscenti so much.

All I can say is that I'm very happy everybody likes the word cognoscenti so much.
Nice for a change from Forester Brand disciples/acolytes.

And how many articles have you had published in prestigious journals like this?

Not even one, ILTB???

Maybe you're spending too much time in A&S, when you could be out educating the researchers. :lol:

What makes you think this is a different study? The PDF for this study is found on the URL furnished in the OP: http://www.biomedcentral.com/1471-2458/12/765/abstract

(http://www.biomedcentral.com/1471-2458/12/765/abstract)"The complete article is available as a provisional PDF (http://www.biomedcentral.com/content/pdf/1471-2458-12-765.pdf). The fully formatted PDF and HTML versions are in production. "



Yes, and that study is not the one that everybody's talking about. That one is here:
http://www.theatlanticcities.com/commute/2012/10/dedicated-bike-lanes-can-cut-cycling-injuries-half/3654/
http://ajph.aphapublications.org/doi/abs/10.2105/AJPH.2012.300762?journalCode=ajph

Seattle Forrest asked the following with respect to the recent article in the American Journal of Public Health:


And how many articles have you had published in prestigious journals like this?

The prestige of a journal is largely created by its editors and its reviewers. This prestige, of course, also determines the number and quality of the papers submitted to it, the raw material from which its editors and reviewers work. However, the prestige of a journal is limited to only the field in which its editors and reviewers work. The authors of the paper being discussed (Teschke, in AJPH) nearly all work in the statistics and epidemiology of public health; one has to expect that the editors and reviewers of AJPH also concentrate only in this field. The prestige of AJPH in the field of traffic engineering is zero. Teschke's paper is an investigation into the crash rates associated with particular types of traffic engineering facilities, which is a field entirely outside of the expertise of the authors and of the editors and reviewers of AJPH.

Yes, and that study is not the one that everybody's talking about. That one is here:
Are you being silly? I referred to the URL cited in the OP and downoaded the specific report discussed. Only you are referring to some other study. Both "reports" are different sets of stats from the same 690 events analyzed by the same researchers and both suffer from the same methodology failures - compete lack of consideration for injury severity and biased sampling as previously discussed by myself and other posters..

I'd like to see the whole study as the information from the abstracts is more or less useless. Reading the information in the two links we have determined:

- Fewer bicycle accidents happen on quiet streets where there are bike lanes and no parked cars
- Bike paths separated from motor vehicle lanes are safer for the cyclist
- Wearing bright colored clothing may help prevent accidents
- People use bike lights more at night than during daylight hours
- More riders wear a helmet when it is mandated by law
- More experienced riders are less likely to be in an accident

It appears this study cast a very wide net for very non-specific data. . .
Studies like this produce generalized results like you see above. If you want more than generalized results then the study gets a lot more expensive.

This and similar studies tend to copy the format of the classic "Hurt Report." They follow that format because it has been the authoritative motorcycle safety study for about thirty years. Many researchers allow their hubris to lead them to seek to become as well known as Harry Hurt (what a descriptive name for a safety researcher).

While there is nothing wrong with seeking fame through excellence, in a recent interview, even Dr. Hurt expressed misgivings about The Hurt Report. He did have concern for the failure to, meaningfully, include non-crashed riders into the data. He also expressed concern for the impact that producing an "authoritative" report had on the field. Once his report was finished the topic was not re-looked at for almost thirty years.

Primarily I mention the Hurt report because it is the model for reports like the one being discussed here. All of the issuers being raised here have already been raised for decades. In many ways it is good that a true, authoritative, report has not been developed, in regard to bicycle safety. I say good because it may suffer the same fate as motorcycle safety research after Dr. Hurts report. The researchers are seldom stupid; however, to produce the best possible report requires more funding than researchers, generally, have available.

I do suggest becoming familiar with The Hurt Report. If for no other reason, familiarity with the methods and procedures, used by Hurt, will make these bicycle reports more useful and help the reader to understand the flaws and avoid misuse of the data.

Are you being silly? I referred to the URL cited in the OP and downoaded the specific report discussed. Only you are referring to some other study. Both "reports" are different sets of stats from the same 690 events analyzed by the same researchers and both suffer from the same methodology failures - compete lack of consideration for injury severity and biased sampling as previously discussed by myself and other posters..

I think there are 2 different reports referenced and linked -- the first one, at American Journal of Public Health (http://ajph.aphapublications.org/doi/pdf/10.2105/AJPH.2012.300762) is Route Infrastructure and the Risk of Injuries to Bicyclists: A Case-Crossover Study and does not appear to be downloadable at this time. There is only an abstract. The second one at elsewhere (http://www.biomedcentral.com/1471-2458/12/765/abstract) is
Personal and trip characteristics associated with safety equipment use by injured adult bicyclists: a cross-sectional study and is downloadable.

Maybe I'm missing something. How are you able to access the first study? Could you post the link you are accessing?

Perhaps we should change P&R to "Why ILTB thinks you are wrong."


http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect

The Internet: Allowing people with no scientific credentials whatsoever dispute scientific studies since 1991.

http://en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect

The Internet: Allowing people with no scientific credentials whatsoever dispute scientific studies since 1991.

:thumb:

I think there are 2 different reports referenced and linked -- the first one, at American Journal of Public Health (http://ajph.aphapublications.org/doi/pdf/10.2105/AJPH.2012.300762) is Route Infrastructure and the Risk of Injuries to Bicyclists: A Case-Crossover Study and does not appear to be downloadable at this time. There is only an abstract. The second one at elsewhere (http://www.biomedcentral.com/1471-2458/12/765/abstract) is
Personal and trip characteristics associated with safety equipment use by injured adult bicyclists: a cross-sectional study and is downloadable.

Maybe I'm missing something. How are you able to access the first study? Could you post the link you are accessing?

Both reports are reporting results from the same data set selected by the researchers from the same selected 690 accidents. One report associates the accidents with the equipment used, the other the same accidents with the street/road/path type.

The methodology problems with faulty sampling and ignoring injury severity data essential for determining cycling RISK, apply to both reports since they share the same biased sample and data omissions. They are essentially 2 chapters of the same novel about cycling risk. No credible conclusions about cycling risk associated with either safety equipment or road/path use can be drawn from such a faulty data set.