What is "EVIDENCE BASED MEDICINE"

[medking]

---

Members do not see this ad.

They aren't examples -- they are why your example is wrong. In fact they are exactly why anecdotal evidence is not accepted. You may not like them, but we don't treat based on anecdotal evidence in the allo world exactly because of these shortcomings (confounding effects, placebo effects, true-true-but unrelated issues).

I think for the most part we agree.

I'm just saying that there are some things right now that western medicine cannot help patients with (more so non-acute/chronic stuff). I see no problem in a person trained as a physician that has knowledge of herbal mixes to use whichever he believes will work best for his patients.

Again, this doctor is not some hopeful that will treat cancer with herbal mixes. I'm sure he will think that is absurd; he knows his limits.

 

[Law2Doc]

... I see no problem in a person trained as a physician that has knowledge of herbal mixes to use whichever he believes will work best for his patients....

Then you fundamentally disagree with the principles of allopathic medicine, and are on the wrong board. Sorry, but there isn't really the middle ground you are seeking. If there is no evidence to support it, you shouldn't be prescribing it as an allopathic physician. You can say -- there's no treatment so go try what you want, but you can't be the one pushing the herbs. Sorry. (and FWIW your insurance won't be covering you if you do this anyhow).

 

[HPSPpayissues]

Evidence-based medicine simply comes down to P-values and confidence intervals.

 

[pietachok]

didn't read through the whole thread, and I think the definition of EBM has already been fleshed out, but if you want to see examples of issues of EBM in medical practice/education, read more about OB/GYN. OB/GYN as a field that has had some great benefits and detriments b/c the historical need to deal w/ taboo parts of the body balanced w/ the extreme pressure to improve fetal and maternal mortality has caused it more than other fields (arguably) to accept more non-EBM...which gets carried on for years and becomes a hard-to-break habit.

One great boon to OB/GYN that wasn't initially EBM--washing hands and limiting insertions during delivery...maternal mortality that was originally thought to be a necessary risk of delivery turned out to be contamination from arrogant physicians who took a long time to admit they could be part of the problem.

In contrast, the dorsal lithotomy position for delivery has been used for so long that many dr's assume it is the EBM best. It's only been studied in recent decades, and there's still disagreement. Some studies show that in this position: since the baby is heading unnaturally upwards at the end of the delivery births may be complicated by slowing down delivery as the baby fights gravity, increasing pressure on the vena cava and other blood supplies lowers the blood oxygen of the baby and increases the risk of thrombosis for the mom, and contractions/pushing have reduced forcefulness/efficacy. The real benefit may simply be that it is the most convenient for the physician and attaching monitors...but whether this benefit improves outcomes sufficiently to limit women to this position remains debated until EBM submits thorough proof...enough to do the extra burden of fighting tradition.

 

[pietachok]

Then you fundamentally disagree with the principles of allopathic medicine, and are on the wrong board. Sorry, but there isn't really the middle ground you are seeking. If there is no evidence to support it, you shouldn't be prescribing it as an allopathic physician. You can say -- there's no treatment so go try what you want, but you can't be the one pushing the herbs. Sorry. (and FWIW your insurance won't be covering you if you do this anyhow).

Well, Law2Doc, IMO, I think both you and MedKing are too far on the extremes, here. You don't just act blind to a patient who expresses a desire to try an herbal remedy before resorting to a known therapy--that is not safest for your patient or GCP. Example: St. John's Wort, Gingko, Ginseng, and Dong Quai ALL have MD initiated studies that can confirm their coagulant/anticoagulant properties, b/c they were investigated for interfering in clotting disorders during surgery...You aren not better off ingoring that a patient expressed a desire to take one of them before trying coumadin or aspirin therapy! You should bring the patient in for regular assessments of their condition, and make sure that they know that even though there may be many studies proving some "blood thinning" properties of the herb, there are no studies (to my knowledge) that show they are good enough blood thinners to replace known FDA-approved therapies.

This is most relevant for an individual patient who has expressed a desire to pursue non-FDA-approved treatment before using known drugs b/c that's how you get a "one patient" sample size. Under our federal regulations, you are allowed as a physician to oversee the treatment of ONE patient on an unsubstantiated therapy, with proper informed consent, as part of personalized clinical practice. ONLY IF YOU MAKE THE SAME ATTEMPT WITH MORE THAN ONE PATIENT, do you need to present it as a clinical investigation to a registered IRB (which will review the informed consent) and obtain an IND number from the FDA. Otherwise, we'd never have new drugs! Obviously, you can still obtain insurance as a physician who does clinical investigations.

Of course for an allopathic physician to discuss non-FDA-approved remedies you simply must discuss that their benefits/risks are unknown through informed consent...as long as proper informed consent is obtained and the process is thoroughly documented in the chart, you can participate in a non-FDA-approved options to one patient. Of course, you must explain that although these are options might have annecdotal claims of efficacy, you can't recommend it in lieu of approved treatment and you must present the approved treatments. You should discuss the extent to which an attribute of an herb has been established through research and the fact that the herb not been evaluated for the treatment of the relevant symptoms/disease even if it might be known to have a certain property. You must acknowledge any known risks of the herb and that there may be unknown risks.

The worse option is to ignore the patients' desire, b/c then you risk a non-compliant patient who takes a prescription but doesn't fill it...and you end up stumped as to why they aren't improving. Better to open the trusting relationship with the skeptical patient to ensure the best care!

This was kind of on a tangent, but I thought it merited pointing out.

 

[medking]

Then you fundamentally disagree with the principles of allopathic medicine, and are on the wrong board. Sorry, but there isn't really the middle ground you are seeking. If there is no evidence to support it, you shouldn't be prescribing it as an allopathic physician. You can say -- there's no treatment so go try what you want, but you can't be the one pushing the herbs. Sorry. (and FWIW your insurance won't be covering you if you do this anyhow).

Really? I don't know about that.

 

[medking]

Well, Law2Doc, IMO, I think both you and MedKing are too far on the extremes, here. You don't just act blind to a patient who expresses a desire to try an herbal remedy before resorting to a known therapy--that is not safest for your patient or GCP. Example: St. John's Wort, Gingko, Ginseng, and Dong Quai ALL have MD initiated studies that can confirm their coagulant/anticoagulant properties, b/c they were investigated for interfering in clotting disorders during surgery...You aren not better off ingoring that a patient expressed a desire to take one of them before trying coumadin or aspirin therapy! You should bring the patient in for regular assessments of their condition, and make sure that they know that even though there may be many studies proving some "blood thinning" properties of the herb, there are no studies (to my knowledge) that show they are good enough blood thinners to replace known FDA-approved therapies.

This is most relevant for an individual patient who has expressed a desire to pursue non-FDA-approved treatment before using known drugs b/c that's how you get a "one patient" sample size. Under our federal regulations, you are allowed as a physician to oversee the treatment of ONE patient on an unsubstantiated therapy, with proper informed consent, as part of personalized clinical practice. ONLY IF YOU MAKE THE SAME ATTEMPT WITH MORE THAN ONE PATIENT, do you need to present it as a clinical investigation to a registered IRB (which will review the informed consent) and obtain an IND number from the FDA. Otherwise, we'd never have new drugs! Obviously, you can still obtain insurance as a physician who does clinical investigations.

Of course for an allopathic physician to discuss non-FDA-approved remedies you simply must discuss that their benefits/risks are unknown through informed consent...as long as proper informed consent is obtained and the process is thoroughly documented in the chart, you can participate in a non-FDA-approved options to one patient. Of course, you must explain that although these are options might have annecdotal claims of efficacy, you can't recommend it in lieu of approved treatment and you must present the approved treatments. You should discuss the extent to which an attribute of an herb has been established through research and the fact that the herb not been evaluated for the treatment of the relevant symptoms/disease even if it might be known to have a certain property. You must acknowledge any known risks of the herb and that there may be unknown risks.

The worse option is to ignore the patients' desire, b/c then you risk a non-compliant patient who takes a prescription but doesn't fill it...and you end up stumped as to why they aren't improving. Better to open the trusting relationship with the skeptical patient to ensure the best care!

This was kind of on a tangent, but I thought it merited pointing out.

Of course 🙂

 

[BigRedBeta]

Ok lets talk about treatments not rigorously studied in prospective trials and the standard of care.

A patient presents to the er s/p head trauma. The patient is obtunded and is moving the right side less than the left; CT demonstrates 3.5 cm L sub dural hematoma with midline shift. I can tell you that you will not find a prospective randomized study to guide your treatment of this situation but you will be in breech of the standard of care if the subdural collection is not drained.

Agreed with neuro here. Only questions where then answer isn't clear get trials anyway. There are some things that are simply too "duh" in nature to warrant investigation. Are you ever going to see a randomized, double blind, placebo controlled trial for antibiotic use in bacterial meningitis? Of course not. neurolddoc gives another example.

Pediatrics certainly is a field which lags behind in the use of EBM because of the barriers to research on children and the fact that for a great many serious childhood conditions, patient numbers aren't at levels that generate study designs with sufficient statistical power. There are swaths of pediatric care where the reasoning behind the standard of care is simply "that's the way we've always done it".

 

[GH253]

Critics of "allopathy" aren't necessarily antagonistic toward the concept of evidence-based medicine. Instead, they question whether mainstream medicine really adheres to an evidence-based ideology. Most people take for granted that mainteam medicine is scientific, but there is a case to be made that "evidence-based medicine" has to a large extent been replaced by "corporate corruption-based medicine." I am sympathetic toward these criticisms.

 

[GH253]

Then you fundamentally disagree with the principles of allopathic medicine, and are on the wrong board. Sorry, but there isn't really the middle ground you are seeking. If there is no evidence to support it, you shouldn't be prescribing it as an allopathic physician. You can say -- there's no treatment so go try what you want, but you can't be the one pushing the herbs. Sorry. (and FWIW your insurance won't be covering you if you do this anyhow).

I strongly disagree. This ideology requires faith in authority- it assumes the validity of information fed to you by educators and regulators. In reality, there are compelling reasons to believe that many currently accepted treatments are not based on a valid body of evidence. Conflicts of interest are so rampant in medical research that nobody can even tell which studies are valid and which are corrupt. In the absense of objective, reliable evidence that is free from conflicts of interest, there is a place for independent thought in medicine. Remember that all drugs are, at some point, tried on humans without prior research. It is wrong to award this privilege solely to FDA regulators and even more wrong to assume that their published results are authoritative.

 

[Law2Doc]

I strongly disagree. This ideology requires faith in authority- it assumes the validity of information fed to you by educators and regulators. In reality, there are compelling reasons to believe that many currently accepted treatments are not based on a valid body of evidence. Conflicts of interest are so rampant in medical research that nobody can even tell which studies are valid and which are corrupt. In the absense of objective, reliable evidence that is free from conflicts of interest, there is a place for independent thought in medicine. Remember that all drugs are, at some point, tried on humans without prior research. It is wrong to award this privilege solely to FDA regulators and even more wrong to assume that their published results are authoritative.

The bold is never true under the US law. All drugs are required to go through clinical trials on animals first at a minimum. Your argument that doctors should "question authority" isn't so much a disagreement with my statement that if you don't follow EBM you are violating the principles of allopathic medicine. You are taking a position totally foreign to allopathic medicine.

 

[Law2Doc]

...

Of course for an allopathic physician to discuss non-FDA-approved remedies you simply must discuss that their benefits/risks are unknown through informed consent...as long as proper informed consent is obtained and the process is thoroughly documented in the chart, you can participate in a non-FDA-approved options to one patient. Of course, you must explain that although these are options might have annecdotal claims of efficacy, you can't recommend it in lieu of approved treatment and you must present the approved treatments. You should discuss the extent to which an attribute of an herb has been established through research and the fact that the herb not been evaluated for the treatment of the relevant symptoms/disease even if it might be known to have a certain property. You must acknowledge any known risks of the herb and that there may be unknown risks.

The worse option is to ignore the patients' desire, b/c then you risk a non-compliant patient who takes a prescription but doesn't fill it...and you end up stumped as to why they aren't improving. Better to open the trusting relationship with the skeptical patient to ensure the best care!

This was kind of on a tangent, but I thought it merited pointing out.

Well, most allopathic physicians would actually avoid advising patients to try alternative medicine options, although they would be cognizant of the fact that patients may resort to these options, and will do their best to reign the patient in by having them come to them with their questions and try to get a good list of what herbs they are thinking of taking. A lot of these things, although not FDA approved, are drugs and have VERY dangerous interactions with FDA approved drugs. At any rate, most medmal insurance policies aren't going to protect you when you start recommending things that aren't FDA approved or the accepted standard of care, so you are really walking out on a plank with your livelihood if you start pushing these. If it's not EBM, it's not accepted allopathic medical practice. You can tell patients -- I can't do anything for you. You can tell patients, that herb you are taking hasn't been shown to do anything in any medical study. But you can't tell a patient -- try this, as an allopath.

 

[GH253]

The bold is never true under the US law. All drugs are required to go through clinical trials on animals first at a minimum. Your argument that doctors should "question authority" isn't so much a disagreement with my statement that if you don't follow EBM you are violating the principles of allopathic medicine. You are taking a position totally foreign to allopathic medicine.

My point is not merely that we should question authority. I'm saying that a large amount what is currently accepted as the standard of practice in medicine is not evidence based, and that you are making an almost totallly arbitrary distinction between FDA-approved products and unproven herbal ones. Furthermore, given the seriousness of the ideological problems that currently exist in mainstream medicine, it is not necessarily wrong to take a position that goes against the grain.

By the way, it's beside the point, but allopathy is a derogatory term coined by the inventor of homeopathy, and no self-respecting western doctor should ever adopt it.

 

[medking]

Well, most allopathic physicians would actually avoid advising patients to try alternative medicine options, although they would be cognizant of the fact that patients may resort to these options, and will do their best to reign the patient in by having them come to them with their questions and try to get a good list of what herbs they are thinking of taking. A lot of these things, although not FDA approved, are drugs and have VERY dangerous interactions with FDA approved drugs. At any rate, most medmal insurance policies aren't going to protect you when you start recommending things that aren't FDA approved or the accepted standard of care, so you are really walking out on a plank with your livelihood if you start pushing these. If it's not EBM, it's not accepted allopathic medical practice. You can tell patients -- I can't do anything for you. You can tell patients, that herb you are taking hasn't been shown to do anything in any medical study. But you can't tell a patient -- try this, as an allopath.

Yea, because western medicine is not familiar with many of these things. No background. The history isn't there. Of course they wouldn't recommend it. I would rather them not recommend it.

Now I question, have you lived in the U.S. your whole life?

 

[oldbearprofessor]

Well, most allopathic physicians would actually avoid advising patients to try alternative medicine options,

Well, if most = 29% (of pediatricians, according to this article) who wouldn't refer for CAM. BTW, there are numerous well-established CAM based pediatric practices in the US of allo/osteopathic physicians. I'd be curious (note that I'm not doubting you, just curious), at what point in the referral/treatment practice they are actually violating their malpractice coverage rules. That is, if a child comes in with a headache and they recommend therapeutic massage/magnets and miss a brain tumor, is their malpractice insurance going to not cover them compared to if they recommended aspirin and missed the tumor - a situation in which they surely would be covered. I'm also curious if members of the AAP's section on CAM are at legal risk for advising each other on CAM therapies for children and on CAM practitioners? Just wondering.

BMC Complement Altern Med. 2007 Jun 4;7:18.
Pediatricians' attitudes, experience and referral patterns regarding Complementary/Alternative Medicine: a national survey.

Department of Adolescent Medicine, Children's Hospital of Michigan, Wayne State University School of Medicine, Detroit, MI

BACKGROUND: To assess pediatricians' attitudes toward & practice of Complementary/Alternative Medicine (CAM) including their knowledge, experience, & referral patterns for CAM therapies. METHODS: An anonymous, self-report, 27-item questionnaire was mailed nationally to fellows of the American Academy of Pediatrics in July 2004.648 of 3500 pediatricians' surveyed responded (18%). RESULTS: The median age ranged from 46-59 yrs; 52% female, 81% Caucasian, 71% generalists, & 85% trained in the US. Over 96% of pediatricians' responding believed their patients were using CAM. Discussions of CAM use were initiated by the family (70%) & only 37% of pediatricians asked about CAM use as part of routine medical history. Majority (84%) said more CME courses should be offered on CAM and 71% said they would consider referring patients to CAM practitioners. Medical conditions referred for CAM included; chronic problems (headaches, pain management, asthma, backaches) (86%), diseases with no known cure (55.5%) or failure of conventional therapies (56%), behavioral problems (49%), & psychiatric disorders (47%). American born, US medical school graduates, general pediatricians, & pediatricians who ask/talk about CAM were most likely to believe their patients used CAM (P < 0.01). CONCLUSION: Pediatricians' have a positive attitude towards CAM. Majority believe that their patients are using CAM, that asking about CAM should be part of routine medical history, would consider referring to a CAM practitioner and want more education on CAM.

 

[Transmogrifier]

There are actually some decent studies that suggest a better than placebo result, to the point that a lot of allo med schools are now teaching this as one of the few viable forms of alternative medicine. It is still obviously controversial, but making gains.

Do you have any references I can check out? I'd be happy to accept acupuncture if the data back it up. It just seems that the bigger and better the study the more it looks like placebo from what I've read.

 

[shindotp]

Maybe some things are above the understanding of current science, or perhaps the framework just might not be enough to test certain things.

Maybe, but homeopathic medicine and naturopathic medicine are definitely not above the understanding of current science.

 

[Rendar5]

EBM is using the best available evidence to guide our diagnostics, managements, and treatments. Best available is key because for a lot of treatments there is little to no evidence (nitrates in ACS). Or stuff cannot be conclusively studied at the best available evidence level (for example, pronated sleep and smoking as risk factors for SIDS). A lot of medicine is based on theoretical pathophys models (say Na-channel and K-channel blockers post-MI) and anectdotal reports (Wellbutrin for nicotine addiction) long before they become established or disproven.

As has been said, schools that espouse this, as most schools have been doing for some time already (eye roll at Obama) include EBM really as an incorporation of the various important studies and ways of reading CLINICAL studies. You will (supposedly) learn not just whether a study is good or bad, but exactly what flaws there are (there are very few without flaws, it doesn't mean not to listen to research with holes) and how it should influence your incorporation of it into your practice.

Alternative medicine should undergo research just as much as standard of care medicine. And a lot of thought should be kept in mind as to the power of the placebo effect. The majority of the effect of many drugs such as anti-depressants is a pure placebo effect. Albuterol in asthmatics also has significant placebo effect. The placebo effect does have a proven effect on mortality (although the study I saw may have had a correlation and not cause-effect), so someone who has had 20 years of constipation, gets some herbs and is suddenly better? Yes, even that could be anywhere from 10-100% placebo effect even with a direct time correlation. There are more requirements to show a cause-effect relationship than just that, even if it makes no sense how it could be nothing else.


EDIT: Not everything you're told is proven is actually scientific fact. Think second hand smoking is dangerous to non-asthmatic adults...? Find me 5 good large-scale studies. Tell me what they say 🙂

 

[LizzyM]

That's a bad example for that particular subject. Here's a chapter with more than five studies (far more) about the dangers of second hand smoke (I believe these studies deal with lung cancer risk, it's 92 pages of summary about the studies): http://monographs.iarc.fr/ENG/Monographs/vol83/mono83-7B.pdf. There are quite a few topics that studies have not explored, smoking is not one of them.

I didn't read the entire monograph but many, many studies found no risk or had a confidence interval that could not rule out no increase in risk.

Furthermore, case control studies run the risk of recall bias. Those who are sick recall and report exposures differently than those who are well. A better design is a nested case-control where a large group is followed prospectively with exposures measured at the start and periodically thereafter. When a subset develop the disease of interest they are matched on age, sex, etc to others in the cohort who do not have disease. The exposures were measured before the disease was diagnosed therefore getting around the issue of recall bias.

 

[LizzyM]

Other studies were done that addressed a variety of the systematic problems and variables that potentially influenced many of the older smaller studies (and recently I know that nested case-control studies have been done are are being undertaken with non-asthmatic populations). They found that lung cancer can be correlated and probably causally associated with prolonged exposure to environmental smoke. I think we should remember that the tobacco industry spent enormous sums of money to prevent large scale studies from being done in the first place and tried to discredit those studies that were done. That probably contributed to the disordered studies that accumulated over the years. Enough evidence has been gathered over the last 50 years to show very strong correlations between second hand smoke and detrimental conditions.

With EBM, a placebo controlled trial is generally the gold standard. Alas, for studies of etiology, such studies are unethical or impossible.

Rather than studies of etiology, let's talk about studies of treatments or diagnostic techniques where what's at stake is the delivery of medical care. How about smoking cessation? Drugs? counseling? acupuncture? OTC patches or gum? I'm not particularly interested in any specific problem/treatment but just saying that that's where docs should be most intersted when considering the evidence for clinical care... second hand smoke is a public health/public policy issue more than a medical care issue.

 

[Rendar5]

That's a bad example for that particular subject. Here's a chapter with more than five studies (far more) about the dangers of second hand smoke (I believe these studies deal with lung cancer risk, it's 92 pages of summary about the studies): http://monographs.iarc.fr/ENG/Monographs/vol83/mono83-7B.pdf. There are quite a few topics that studies have not explored, smoking is not one of them.

Thank you for that monograph (I was not aware of it before), it seems to support my point that the data on second hand smoke exposure related to lung cancer is not particularly good, even though it has been studied significantly and at length. I read tthe first couple of pages and have since glanced through the first 45 pages I know that the monograph is trying to say that this has been extensively studied and boldly in support of carcinogenic second-hand smoke. But look at the data: The vast majority show no significant increased risk of cancer with a smoking spouse. I know you're thinking "but a lot of those show a relative risk greater than 1", but look at the confidence intervals...notice how most of those cross 1? When that monograph says that most of the studies show an increased relative risk, it's being a bit disingenuous

Now maybe if you grouped all those studies together you'd have something statistically significant (though now looking at the RR's and CI's of the meta-analyses, I'm not sure it would). I could totally grant you that. But when you have to have such a large n to show a slightly increased risk (which is what most of those studies implied), that seems kind of clinically insignificant. what's 1.3 times a 1:1,000,000 risk? a 1.3:1,000,000 risk.

Now even if you take the data to point in the direction of a cause-effect relationship, my point is more that what is unabashedly taken as proven and quite a risk increaser by the american public and I'd say a large portion of the medical community is in fact really something that may or may not exist, with multiple sources unable to show a statistically significant effect, and as a whole, point towards a mild risk modifier for a relatively rare disease: Lung cancer in the non-smoker (10% of the lung cancer population on a guess??). Now if you take this ane make it worldwide, you may have a public health burden. Maybe on the order of male breast cancer? It's important that it's dealt with, but the real target should be smoking, and perhaps smoking around kids (though I don't know the asthma data, and the SIDS data was retrospective and should remain that way for ethical reasons), not second-hand smoke in general.

 

[limarodessa]

.Where I can find the standards of clinical trials?.
.I searched here:.

Clinical trial
ClinicalTrials.gov.
.
.b..ut I could not find.

.My research is not research of drugs. I wish to collect the statistical information about critically ill patients with certain concomitant disorders and compare this data with data received from observing on the critically ill patients without such concomitant disorders..

 

[LizzyM]

.Where I can find the standards of clinical trials?.
.I searched here:.

Clinical trial
ClinicalTrials.gov.
.
.b..ut I could not find.

.My research is not research of drugs. I wish to collect the statistical information about critically ill patients with certain concomitant disorders and compare this data with data received from observing on the critically ill patients without such concomitant disorders..

What you propose is not a clinical trial, it is an observational study.

In a clinical trial, an intervention is applied to some or all of the subjects of the study and outcomes are measured. In a Phase I clinical trial, an intervention (often a drug) is administered either to healthy people or to seriously ill people for whom all other treatments have failed. The purpose of the study is to determine the safety of the product. The Phase II trial is most often controlled (one group receives the experimental treatment and the other does not), randomized (subjects are assigned by luck of the draw to experimental intervention or not) sometimes placebo controlled (a placebo is administered to some of the subjects while others get the active compound), and often double blind (neither the subjects nor those measuring the outcomes know who is getting the experimental intervention). The purpose is to determine the safety and efficacy ("does it work") of the treatment. Phase III is most often a study of a larger sample, often for a loner period of time to gather more data on safety and efficacy before marketing. Phase IV is the further collection of data after the product is approved for marekting.

 

[limarodessa]

What you propose is not a clinical trial, it is an observational study...

Thanks. Are anywhere in the NET the international (or USA) standarts for observational study?

 

[LizzyM]

Thanks. Are anywhere in the NET the international (or USA) standarts for observational study?

I don't know of anything on the net but you could try this book:

Methods in Observational Epidemiology Second Edition by
Jennifer L. Kelsey , Alice S. Whittemore, Alfred S. Evans, W. Douglas Thompson

 

[limarodessa]

... you could try this book:

Methods in Observational Epidemiology Second Edition by
Jennifer L. Kelsey , Alice S. Whittemore, Alfred S. Evans, W. Douglas Thompson

Thanks