代写POLITICAL SCIENCE 2120: “‘FOLLOW THE SCIENCE?’ THE POLITICS OF HEALTH” Midterm Exam代写Processing

日期：2024-09-15 11:18

POLITICAL?SCIENCE?2120:

“‘FOLLOW THE SCIENCE?’?THE?POLITICS OF HEALTH”

Midterm?Exam?— Take-Home?Questions

Colorectal cancer is the third-leading cause of cancer death?among both?men?and women?in?the?United?States. And it appears to have become more common?among younger?cohorts?since?the?mid-1990s.

There are several procedures currently used?to?screen?asymptomatic people?to?try?to?identity?this?disease?in its earliest stages, when it might?be more?treatable, with?colonoscopy being?the?most?common.

Colonoscopies can be both diagnostic?(used to detect?disease)?and?therapeutic?(used?to?treat) — if?precancerous polyps are found, they can be?removed?during?the?procedure,?potentially preventing?cancer from developing later.

The United States Preventive Services Task Force?(USPSTF) is?a volunteer,?government-sponsored?panel that?synthesizes the best available medical research?and?aims?to?provide?evidence-based?recommendations?for health procedures designed to prevent serious illness.?USPSTF?does?not?consider?costs in its recommendations — it?only looks to see whether?the?research?suggests?that?a?particular?procedure is likely to benefit patients, regardless of?the magnitude?of?the?benefit?or?the?financial?cost.

By law,?services recommended by the?USPSTF must be?covered by health?insurance?without?any?charge?to patients.

Currently, USPTSF currently recommends?that all adults?aged?45?and?older?undergo?colorectal?cancer?screening. While?several other cancer?screening methods have been examined?using?randomized controlled trials?(RCTs), colonoscopy had not — until now. Last?October,?the New England?Journal?of?Medicine published the?first RCT of a colonoscopy screening program?for?the general?population.?(Note:?We are not talking about?screening for high-risk populations,?such?as?people with?family?history?of?particular cancer.) That study is posted on Carmen. You will?need?to?read?it?before?answering?the??questions below. You are not expected to read?the?additional?documents hyperlinked?in?this?file;?I?included them as background for interested?students.

After you have read the?study, answer all?of?the?questions?described below.?Note?you?can?tackle?each???part?separately — there is no need to try to?combine your?answer into?a?single,?coherent?essay. Do?not?write more?(or less) than is necessary to answer?the?questions. The goal?is?to?have?clear?and?concise answers, without fluff or throat-clearing.

Part A:?In?your own?words,?briefly summarize?the?design?of?the study?and how it was?carried out. Also?briefly?summarize?its?overallfindings about the?effects?ofbeing randomly?assigned to?receive?a?referralfor screening.?Note?that?not everyone assigned to get?screened actually?followed?through,?a major?criticism?of?this?study?you’ll?consider below. Because it is?the?invitation?to?get screened?— rather than?actually?getting screened?—?that?is?randomly assigned, the experiment is able?to?estimate what?is?known?as?the?“intent-to-treat”?effect,?often?abbreviated as?ITT.

Part B: When evaluating medical interventions, it is important to understand the?difference?between?absolute?and?relative?effects. For example,?suppose that 2?percent?of?people?who?get?Covid-19?end?up passing away.?(This is not the true estimate,?I’m?just using?it?as?an?example.)?Suppose?further?that we?find a medication that reduces the?death rate?from?2?percent?to?1?percent. That?represents?a?relative?reduction in death of 50%?(1 is?50%?smaller than 2), but?an?absolute reduction?of only?1 percentage?point.

A quick note: In research, including?the?study you are?reading,?relative?risk is?often?reported?as?a?“risk?ratio.” The baseline risk?(e.g., for the control?group)?is?normalized?to?1,?and?then?the?increase?or decrease is measured relative to this baseline. As?an?example,?a?risk?ratio?of 1.15?corresponds?to?a?15%?increase above the baseline?(1.15-1=0.15 as a?decimal,?or?15%?as?a?percent).?A?risk?ratio?of 0.85?corresponds to a?15%?decrease?(0.85-1=-0.15, or?-15%).

Using the?intent-to-treat estimates?from?the?colonoscopy?study,?what are?absolute?and relative?effects?of?colonoscopy?screening?on?the?10y-ear risk?of?(1)?being diagnosed with?colorectal cancer;?(2)?death?from?colorectal cancer;?and (3)?death?from?any?causes. You can present these estimates in table?form,?if?that?is?easier.?Hint:?All?of?the?numbers?you?need??to calculate these effects are?found?in?Table?2?of?the?study.

Part C: There is much debate among medical?researchers?about?the?most?relevant?endpoint,?or?outcome??of?interest, for measuring the impacts of screening and other preventative?procedures.?Some?argue?that???we?should?focus?on?overall?death, regardless?of the?cause,?since ultimately that is what people care about and this end point can capture unintended?health?harms?of screening and?treatment.?Others?argue?that?we?should?focus?on?disease-specific death,?since everyone will die?of something at?some point.?Showing?that a treatment prevents you?from dying from cancer?only?to?die?from?a?heart?attack?a?few years?later?still means there is a meaningful impact of screening and?meaningful?gains?in?life?expectancy.

This debate is particularly heated in the context?of cancer?screening. For?breast?cancer,?for?example,?a????number of studies?have found that mammography reduces deaths?due?to?breast?cancer — but?does?not?appear to affect overall mortality?from?all?causes.

From?the?perspective?ofpolicy —?whether screenings should be?recommendedfor the?whole?population?by?the?government,?and?whether?the?government and?private?insurers should be?required to?pay?for them —?rather than?the?perspective?of?an?individualpatient,?which?endpoint?should?we?use?to?make?policy?decisions?

Part D: Some might be?surprised that?cancer?screenings?don’t always?prolong?people’s?lives.?This?could?happen for three reasons. First, diagnosing cancer?early?may?not?actually?help —?either?because?no effective treatments exist, or because the?cancer?that was?found was?unlikely to?kill?you.?(Prostate cancer,?for example, is usually very slow-growing. Many men?over the?age?of 70 who?die?from?some???other cause are found, upon autopsy,?to?have?had?a?cancer growing in?their?prostate?but likely?had?no?symptoms of?the?disease.)

Second, the screening procedure may itself?harm health, offsetting the?benefits.?Colonoscopy,?for?example, can cause bowel perforation and infection, bleeding,?severe?abdominal?symptoms,?and?even??heart attacks.?These unintended?side effects are quite rare?but?may?outweigh?the?benefits?of?proactive?screening at the population level if?the cancer is?also rare.

Third, the value of such tests depends on?their?accuracy,?and?all?clinical?tests?are?imperfect.?Clinical?tests?are?evaluated using two main metrics:?sensitivity and?specificity.?Sensitivity is?the?percent?of actual?cancers that the test accurately detects?(in other words,?the?probability?that?the?test will?accurately?detect?true cases of disease). Specificity is the percent?of?patients without?cancer?that?the?test?accurately?identifies as being healthy?(rather than a?false-positive). Both?sensitivity?and?specificity?for?colonoscopy?are around?90 percent.

Now, many people think that 90 percent?sensitivity and?specificity imply?that, if?you?have?a?colonoscopy that comes back positive, it means you have?90 percent?chance?of?actually?having?cancer.?But that’s incorrect! To interpret clinical test results, you also?need?to?consider?the?prevalence?of?the disease in the population?(also known the?pre-test?probability that?a?person?has?a?disease).

For example,?suppose that only?1.5 percent of?the population?actually has?colorectal?cancer?—?a ballpark estimate based on the RCT?study you read?for?this?exam. Among?this?subset?of?people,?screening will accurately find 90 percent of?the?cancers?(sensitivity=90%),?or?about?1.35 percent?of?the??population?(1.5 percent of?population with cancer * 90%?sensitivity). But?among?the?other?98.5?percent?of?the population that is cancer-free, the colonoscopy is going?to indicate?cancer?10?percent?of?the?time?(100% – 90%?specificity?=?10% false-positive rate). That’s?9.85 percent of?the?total?population!

Add up those number —?1.35 percent of?the population that is true-positive?plus?9.85?percent?of?the?population that is?false-positive — and you can estimate that?just?12%?of?the positive?colonoscopies?are?real cancer cases?(1.35+9.85/1.35), while the remaining?88% are?false?positives?among?people who?don’t?have???cancer. Even with really accurate tests, there will?always be?many?false?positives?for?conditions?as?rare?as?colorectal cancer. And these?false positives can have real?harms,?including?unnecessary?anxiety?and worse. For example, both?suicides and heart attack?rates?have?been?found?to?increase?significantly?among men who are told that their prostate?screening indicates they have?cancer,?even?though?most?of?these diagnoses turn out to be?false positives?upon?further?testing.

Are?you?surprised by?the?true?vs.false?positive?calculation?above? How should these?figures?be?incorporated into?the?policy?decision?about whether?to?recommend cancer?screening?for?the?whole?population?

Part E: Many vocal supporters of colonoscopy — including professional associations?for doctors who?make a living performing them?— have been very critical?of?the population?colonoscopy?screening RCT you read. They have made two arguments.

First, they argue that the results?from?the?experiment?are?not?consistent with?earlier?observational ?(non-

RCT) research, which seemed to?show much bigger benefits?of colonoscopy. For?example,?a 2018?study?found that colonoscopy reduced colorectal cancer?deaths?by?67?percent?over?10?years?(this?is?a?relative?risk reduction). This?study used the case-control design we have?covered in?class:?Patients who?died?from colorectal cancer were matched based on?observable?characteristics, including?sex,?age,?and geography, to people who didn’t die from cancer,?and?then?the?rate?of colonoscopies between?the?two?groups was?compared.?Drawing on?whatyou learnedfrom?the?other examples we have?studies?so?far this semester,

explain?why?this?observational?design?is?likely?to?overstate?the?benefits?of?colonoscopy,?and?why?should?we?prefer to?calculate?benefits using an?RCT.?Be?sure to?discuss?specific readings or examples?from?class in your?answer.

Second, only 42 percent of?the participants randomly assigned?an?invitation?to?get?a?colonoscopy in?the?RCT actually received one. “A colonoscopy will?only work if a patient gets?one,”?a?gastroenterologist highly critical of?the?study told NPR. In addition to?the?main intent-to-treat?results,?the?authors?of?the???study present a?supplemental “per-protocol” analysis. In this?analysis,?only?the?42?percent?of?the treatment group that actually chose?(i.e.,?self-selected) to get?a?colonoscopy were?matched?to?observationally similar participants in the control group, and?the?authors?found?much?larger?benefits?of?screening in this per-protocol?analysis?than in their?overall?estimates.?Should?we?believe the?results?of?this?analysis? How could they?be?biased in?ways that the?intent-to-treat analysis is unlikely?to?be??Again, your answer?should mention?specific examples?from readings or?class.

Part F: Policymaking?is driven in part by evidence,?but?also?many?other?political?considerations,

including advocacy from interest groups — like colonoscopy providers, who responded “swiftly?and

unequivocally?to the media coverage” of?this?study — and?well-meaning?policy?activists.?The?latter

group includes cancer?survivors. People who have an asymptomatic?cancer?discovered and?removed?as?a result of?undergoing a routine?screening will naturally believe?that the?screening?saved?their?lives?and?become powerful public advocates.?(They assume, of course, that the?cancer would’ve?killed?them?had?it not been found; as we?saw in the prostate?cancer?example?above,?that?may not?be?true. And?they?ignore the health harms created by the?false positives?for?other?people who were?screened?and?didn’t?end up having?cancer.)

Another group of?activists is?family members and loved ones of?people?who?have?died?from?cancer.

They also advocate?for universal asymptomatic?screening in the hope?that it will?catch?disease?early

enough for treatment and help other?families?avoid?the?pain?they have?suffered.?(They?assume?that?early?detection will indeed lead to effective treatment, which is not?always?the?case, and?also?ignore?the?harms?resulting from?false positives.)

One?such advocacy group, the Colon Cancer?Coalition, put?out?a?statement?in?response?to?the?study you read?suggesting that it “may be misleading” and?arguing?that?“getting?screened?for?CRC?does?save?lives.”

What role?should these?kinds?ofadvocacy?and considerations have?in?the?decision?about whether to?recommend cancer

screening??What?should?policymakers?do?when?the?anecdotal?stories?from?cancer?survivors andfamilies?conflict with

evidence?provided by?rigorous RCTs??Overall,?how do?you?think?these kinds?ofpolitical?—?rather than?evidence-based?—?considerations affect?government?policy?

Part G: As noted in the introduction, the USPSTF?does not?consider?the?cost?of?treatments.?But?these?costs are real. For the average patient,?a?colonoscopy has?been?estimated?to?cost?$1,200. When?paid?for?by private insurance, this will increase the cost of?health insurance?for?everyone?and?make?it unaffordable?for?some. When paid for by the government?(e.g., through Medicare?and?Medicaid), it could take away resources from other government programs, including?programs?that?could?improve?people’s health and extend their lives?in?other ways.

Although USPSTF is not allowed to consider?costs?and?tradeoffs,?policymakers?can?do?so.?Suppose?you?are?working as an?adviser to?public officials and helping?policymakers?decide?whether the?government should recommend?that everyone?gets?screenedfor colorectal cancer by?undergoing a?colonoscopy.?Incorporating whatyou?learnedfrom?reading?the RCT and the?analysisyou?didfor the?questions above,?as well as any?other relevant consideration?that we may?have?notyet taken?into?account,?what advice?wouldyou?give?to?policymakers? Be sure?to?clearly?articulate?and explain?the?reasoning?behindyour recommendation.

?