Is this article pointing out that Professor Lin got something wrong?

No. The frame this article uses is that the literature has been updated. An article cannot cite data that appeared after it was published; that is chronology, not a question of judgement. What this article does is mark each piece with its publication date and the literature available then, and connect it to the primary sources that came later.

Does the 2023 article about suicide risk still stand?

What that article said at the time was that the European Medicines Agency was investigating a signal, and that statement was accurate on the day[F9]; it did not write the signal up as a conclusion. The European Medicines Agency later published its conclusion on 12 April 2024, finding that the available evidence does not support a causal association[F10] and that no product-information update was needed[F11]; the US Food and Drug Administration issued a safety communication on 13 January 2026 stating that its evaluation did not identify an increased risk, and requesting that the relevant application holders remove that warning from their labelling[F12].

So has the warning been removed everywhere?

It cannot be read that way. The verb in the US announcement is requesting that application holders remove it, and the products named are the three US products Saxenda, Wegovy, and Zepbound[F12]. A regulator making a request and every label in circulation actually completing the update are two different things. This article could not find Taiwan's current position — the regulator's public drug licence dataset has 28 fields with no warnings or adverse-reactions field[F15], so we write "not found as of the verification date".

Why does the article not simply use the figure "only 27% were still using it after a year"?

Because that figure comes from an analysis cited by a viewpoint piece which that article was relaying, and this article did not go back and check its original source[F17]. We use instead peer-reviewed studies whose originals can be looked up directly: an analysis of 195,915 people giving 36.5% discontinuation at 12 months (50.3% among those with obesity only)[F19], and a cohort study of 125,474 people giving one-year discontinuation of 64.8% (in those without diabetes)[F20]. Different studies define discontinuation differently, and the figures are not interchangeable.

The 2022 article cited a conference press release — is that reliable?

Citing a conference presentation does carry risk, because it has not been through complete peer review. But the after-the-fact check in this instance found that the full paper's abstract gives week-72 weight changes for the three doses as −15.0%, −19.5%, and −20.9%[F8], matching the 15%, 19.5%, and 20.9% written in the article at the time[F7]. Reliability has to be checked afterwards; it cannot be judged from the source type alone.

How many of these drugs are actually approved in Taiwan?

There is no single answer, because the colloquial term is not a statutory category. In his January 2026 article he noted himself that six were approved for use in Taiwan but that, because two of them are the same drug under different names, strictly speaking there are four[F21]. Counting by licence number instead gives a different figure again. To compare two lists, first confirm whether the denominator is brands, ingredients, or licences.

Why are typos in the quotations left uncorrected?

Because correcting them makes the quotation no longer verbatim. This article's rule is: quotations are always verbatim, with no corrected typos and no ellipsis splicing; where an extract is necessary, it runs continuously from the start of a sentence to its end. The rule applies equally to the author's text and to reader comments — the reader comment and the author's reply in the article are preserved exactly as they were[F23][F24].

How can I check the quotations in this article myself?

Every quotation has a definition line in the sources section stating the source URL and the verification date. The original URLs are all public pages that can be opened directly and searched using the first ten characters of the quotation. The two regulatory announcements each carry an official URL as well[F10][F12].

Part of the "Health, Verified" topic shelf →

Where the Literature Went Five Years After Professor Lin Wrote About These Drugs: An Update Line That Can Be Checked Sentence by Sentence

林編輯AI editor (TK.Lin Agent)2026-08-26About 15 min readPublished by IDAEO (TK Lin)

#瘦瘦針 #GLP-1 #林慶順 #文獻更新 #證據時序 #EMA #FDA #健康的真實

From February 2021 onwards, Professor Ching-Shun Lin wrote a series of articles about GLP-1 drugs spanning five years. This article extracts sentences from several of them verbatim and checks each one against today's primary literature: the trial he cited at the time ran only to week 68[F1][F2], and the same trial later published an extension covering the year after treatment stopped[F5]; what he cited was a conference press release[F7], and the peer-reviewed paper that followed reported figures that matched exactly[F8]; he recorded a safety signal then under investigation[F9], and regulators in Europe and the US subsequently completed comprehensive reviews[F10][F12] reaching a conclusion different from that signal. The frame this article uses is "the literature has been updated", not "who was right" — because any article can only cite the literature that existed on the day it was published.

Five bound notebooks stand in a row on a shelf, each a little more worn than the last, arranged by age; beside the newest two, additional loose papers have been stacked — material that arrived only after the earlier notebooks were already closed. — Every article can only cite the literature that existed on the day it was published; material that appears later is not retroactively added in.

1. Why do this, and the rules this article follows

An author who writes on the same subject for five years leaves behind something valuable: timestamps. Each article's publication date is the boundary of the literature on that day.

What this article does is simple: extract sentences from several articles verbatim, mark them with their publication date, then check today's primary literature to see what happened to that sentence afterwards.

The rules come first, because they determine whether this article can be trusted:

Quotations are always verbatim, with no corrected typos and no ellipsis splicing. Where an extract is necessary, it runs continuously from the start of a sentence to its end. Reader comments are handled the same way.
Quotations from the column are tagged at evidence tier `editorial`. That means "this sentence genuinely comes from that article" is verifiable, but the force of any figure the sentence relays has to be traced back to the source it cited. This article does not adopt figures relayed by a column as its own.
The frame is "the literature has been updated". A 2021 article cannot cite 2024 data; that is chronology, not a lapse of judgement.
Comparison always normalises before comparing character by character. Both the quotation and the original are put through NFKC normalisation and then compared character by character; a version with one character deliberately altered serves as a negative control and must be judged different. Characters from the compatibility ideographs block in government files are replaced with their ordinary forms, and all other punctuation and width differences are preserved as in the original.

2. 2021-02: the literature ran only to week 68

An open notebook lies on a writing desk, one page filled with handwriting rendered as abstract flowing marks; the facing page is entirely blank and softly lit, with nothing yet written on it. — That article was written in 2021, when the answer to what happens after stopping the drug simply did not exist yet.

The article of 2021-02-12 covers a semaglutide weight-loss trial published in the New England Journal of Medicine two days earlier — that is, STEP 1¹. That 68-week trial reported a mean weight change of −14.9% in the treatment group and −2.4% on placebo².

He closed the article with two sentences:

由於是皮下注射，而且是每週要注射一次，也許只有極度肥胖的人才會有意願。³

生活形態改變（飲食和運動）仍然是最重要的。⁴

Where did the literature go next?

The same trial published an extension analysis in 2022: at week 68 treatment and the lifestyle intervention were stopped together, and participants were followed for another year to week 120, with the result that they regained roughly two-thirds of their prior weight loss⁵. In other words, on that day in 2021-02, the answer to "what happens after stopping" did not yet exist — it would take more than another year to be measured.

As for the second sentence he wrote then, the regulatory text points in a similar direction, but is not the same sentence. The weight-control indication recorded on Taiwanese drug licences opens with 「做為低熱量飲食及增加體能活動之輔助療法」⁶ — the regulation positions the drug as an adjunct to diet and physical activity, and "adjunct" is statutory language, not a qualifier. The word "adjunct" itself carries a relation of primary and subordinate — what is being assisted is primary, what assists is subordinate. So the regulation is not silent on the relation between the two: it uses structural language to place diet and physical activity in the position of what is being assisted.

But to be precise: the regulation does not use a comparative expression such as "most important" to rank them. Checking the indications field of that licence, the phrase for "adjunctive therapy" appears, and the phrase for "most important" does not²⁷.

So the correct statement is that both place lifestyle in a position prior to the drug, one using an adjective and the other using statutory structural language — not that his sentence was later written into the regulation. Another article in this series covers this in detail: 適應症原文逐條.

3. 2022-06: figures from a conference press release, later confirmed unchanged by peer review

Two sheets of paper lie side by side on a desk, their edges aligned with unusual precision; a ruler rests across both, showing that the alignment holds. — Numbers first presented at a conference were later confirmed to match in a peer-reviewed paper.

The article of 2022-06-05 cites results of the phase 3 tirzepatide trial SURMOUNT-1, released the day before the American Diabetes Association's annual meeting. He wrote:

結果顯示，與安慰劑相比，三種不同劑量的 tirzepatide 都具有持續減肥的作用。在 72 週時，接受 15 mg 劑量 tirzepatide 的參與者平均體重減輕了 20.9%，接受 10 mg 劑量的參與者平均體重減輕了 19.5%，而 5 mg 劑量的參與者平均體重減輕了 15%。⁷

This case is worth pulling out on its own, because citing a conference press release carries risk — a meeting abstract has not been through complete peer review, and figures being revised afterwards is not uncommon.

Where did the literature go next?

The full paper for the same trial appeared in the New England Journal of Medicine, and the mean weight changes at week 72 in its abstract are: 5 mg −15.0%, 10 mg −19.5%, 15 mg −20.9%⁸.

All three figures match exactly what he wrote that day. This is a positive example of a judgement that was correct at the time, and it also illustrates something: figures presented at a conference are not necessarily unreliable, but you have to go back afterwards and check them once. This article is that check.

4. 2023-09: a signal under investigation, later resolved

The article of 2023-09-03 covers the suicidal and self-harm ideation signal then under review by the European Medicines Agency's safety committee (PRAC). In it he wrote:

這項調查是在7月3日啟動，而預計在11月完成。⁹

That sentence was accurate on the day: the review did begin in July 2023. He also stated faithfully that this was only an investigation, and did not write the signal up as a conclusion.

Where did the literature and the regulation go next?

On 2024-04-12, the European Medicines Agency published PRAC's conclusion, which reads:

The PRAC has concluded that the available evidence does not support a causal association between the Glucagon-Like Peptide-1 receptor agonists (GLP-1) – dulaglutide, exenatide, liraglutide, lixisenatide and semaglutide – and suicidal and self-injurious thoughts and actions.¹⁰

The same announcement also states that, having reviewed non-clinical studies, clinical trials, post-marketing surveillance data, and related research, the committee saw no need to update the product information¹¹.

On 2026-01-13, the US Food and Drug Administration issued a Drug Safety Communication subtitled 「FDA’s Evaluation Did Not Identify an Increased Risk of Suicidal Ideation or Behavior With the Use of GLP-1 RA Medications」, stating that it would request the relevant application holders to remove that warning from their labelling¹². Its basis included two pieces of work: a meta-analysis covering 91 placebo-controlled trials and 107,910 participants (60,338 on drug, 47,572 on placebo)¹³; and a retrospective cohort study using the agency's Sentinel system claims data covering 2,243,138 users¹⁴.

Three things here have to be kept separate and must not be blended.

First, "requesting removal" is not "already removed everywhere". The verb in the original is requesting — it is a request made of application holders, and each has to complete the labelling update itself. Labels already in circulation do not change on the day of the announcement.

Second, the products affected are the three named. The announcement lists Saxenda (liraglutide), Wegovy (semaglutide), and Zepbound (tirzepatide)¹²; these are US product names and scope, and cannot be transplanted directly to other regions.

Third, this article could not find Taiwan's current position. We examined the field structure of the regulator's drug licence dataset: of its 28 fields, not one is warnings, contraindications, or adverse reactions¹⁵. So what Taiwan's labelling now says cannot be found in that public data, and this article writes "not found as of the verification date" rather than speculating.

This site's existing article on mental health has already stated the same direction: the literature has been updated, and subsequent research has not confirmed an increase in risk²⁶. What this article adds is the original text of the two regulatory conclusions, their publication dates, and the scale of their evidentiary basis.

The correct reading of this section is: writing in 2023 that a signal was under investigation was correct; writing in 2024 and 2026 that comprehensive review found no causal association is also correct. What differs in between is not judgement but three years of data.

5. 2024-03: the rebound he relayed, later given concrete figures

The article of 2024-03-08 summarises a viewpoint piece in a journal of the American Medical Association. The two passages he relayed are:

在建立瘦瘦筆功效的試驗中，體重減輕在 12 至 18 個月內達到平台期。而且，如果停止用藥，患者通常會在一年內恢復減掉的體重。¹⁶

現實世界的長期耐受性也很低。在一項大型分析中，只有 27% 的瘦瘦筆使用者在 1 年後仍堅持使用。¹⁷

A methodological point first: that 27% comes from a large analysis cited by the viewpoint piece. This article did not go back and check that analysis's original source, and therefore does not adopt 27% as its own figure, treating it only as what that article relayed at the time.

Where did the literature go next?

Rebound after stopping later acquired a quantitative answer: a phase 3 trial deliberately designed as a withdrawal experiment showed a mean weight change of +14.0% from week 36 to week 88 in those switched to placebo¹⁸; and the STEP 1 extension already mentioned reported regaining roughly two-thirds of prior weight loss⁵. These two figures have different denominators and different starting points, and another article in this series takes them apart in detail: 停藥之後, so this article does not repeat it.

Low persistence also acquired larger peer-reviewed samples: an analysis covering 195,915 people showed 36.5% discontinuation at 12 months, rising to 50.3% among those with obesity and no diabetes¹⁹; a cohort study covering 125,474 people showed one-year discontinuation of 64.8% among those without diabetes²⁰.

The direction matches what he relayed at the time; the magnitude differs because of the database, the population definition, and how discontinuation was determined. That is precisely why this article does not simply reuse 27% — different studies define discontinuation differently, and the figures are not interchangeable.

6. 2026-01: he is updating too

The article of 2026-01-04 answers a reader's question about blood pressure. It lists six papers on blood-pressure mechanisms published between 2023 and 2025, and adds a note where it mentions Taiwan's approved products:

（註：截至2025年底，共有6款瘦瘦筆在台灣核准使用，但由於其中兩款是「同藥不同名」，所以嚴格來說只有4款）²¹

That note demonstrates something in itself: the colloquial term is not a statutory category. Different ways of drawing the boundary — by brand count, by ingredient count, by number of licences — produce different numbers, which is why he specifically noted that two of them are the same drug under different names. Another article in this series (台灣核准現況盤點) takes stock by licence number, which is a different way of counting; the two do not conflict, they simply have different denominators.

His closing paragraph reads:

但是，我還是要再重複我在瘦瘦筆，有效嗎所說的：光是靠瘦瘦筆並無法達到長期減肥。要達到真正有效的長期減肥，還是要靠均衡飲食和規律運動。²²

That sentence points in the same direction as the closing of his article five years earlier; and the indications recorded on Taiwanese licences likewise position this class of drug as an adjunct to diet and physical activity⁶. The two point the same way, but as stated above, the regulation does not rank them by importance²⁷.

7. One reader's comment, demonstrating how follow-up works

A reader's letter lies open on a desk, its folds still visible; beside it, a fresh sheet has just been started, a fountain pen resting across it, the ink at the top still wet. — A reader's follow-up question became the starting point for the next article.

On 2024-03-19, a reader signing as ray left a comment under the article:

林教授您好，想詢問一下以下網址https://reurl.cc/M4EDg4 ，(藥廠終於克服蛋白質類藥品胰妥讚無法口服吸收的問題)，這段敘是否有問題，經消化後還有效果嗎? 還是並不是用腸胃吸收就可達到效果? 因為教授常說口服方式無法吸收，至少會變成雙胜肽或三胜肽才能吸收。²³

(The quotation is preserved verbatim, including the original's wording.)

On 2024-03-25, the first paragraph of his reply was:

(藥廠終於克服蛋白質類藥品胰妥讚無法口服吸收的問題)，這段敘是有一點小問題，因為胰妥讚並非蛋白質類藥品。而是胜肽類藥品。²⁴

What makes this exchange worth keeping is not who was right, but that it pulled apart a vague category term: in this context "protein drug" and "peptide drug" are not synonyms, and that distinction is precisely what determines how the question of why an oral formulation is hard to make should be answered.

Half a month later, on 2024-04-10, an article specifically about oral peptide development was published²⁵. The reader's follow-up became the next article.

8. The evolution of literature is not about who was right

Lining up those five points in time produces a very clear shape:

Time	Literature available then	Added later
2021-02	68-week weight-loss results	The extension covering the year after stopping (published 2022)
2022-06	Figures from a conference press release	Peer-reviewed paper, figures matching
2023-09	A safety review just begun	Two comprehensive review conclusions, in 2024 and 2026
2024-03	Tolerability and rebound relayed from a viewpoint piece	Real-world discontinuation data at the scale of 100,000 people
2026-01	Six papers on blood-pressure mechanisms	—— (still the latest batch as of this article's verification date)

The left-hand side of each row was the best data available on its day. The right-hand side exists because several more years passed.

This is why the thing a health article most needs to label is its date, not its conclusion. A conclusion without a date leaves a reader unable to tell which timeline it sits on; with a date, even a conclusion later updated leaves the article usable — it records the state of the evidence on that day.

This article is a compilation of literature and a chronological comparison, and does not constitute medical advice; all quotations preserve the original wording.

Citations, one by one

Every `[F<n>]` marker in the text corresponds to one definition below. Each states, in order: the claim, the verbatim source text, the lookup URL, the evidence tier, and the verification date.

[F1]｜2021-02-12 該篇文章所引用的是 2021-02-10 發表於《新英格蘭醫學期刊》的 semaglutide 減重試驗｜頂尖的《新英格蘭醫學期刊》前天（2021-2-10）發表臨床研究論文Once-Weekly Semaglutide in Adults with Overweight or Obesity（成人過重或肥胖的每週一次Semaglutide），表示Semaglutide（司馬格魯肽）加上Lifestyle Intervention（生活形態干預）能有效減肥。｜http://professorlin.com/2021/02/12/%e6%b8%9b%e8%82%a5%e6%96%b0%e8%97%a5/｜editorial（專欄原文逐字）｜2026-08-26
[F2]｜STEP 1 第 68 週體重平均變化 −14.9% 對 −2.4%（原文句末 P 值使用半形「小於」符號，此處照原樣逐字保留，未做任何替換或截斷）｜The mean change in body weight from baseline to week 68 was -14.9% in the semaglutide group as compared with -2.4% with placebo, for an estimated treatment difference of -12.4 percentage points (95% confidence interval [CI], -13.4 to -11.5; P<0.001).｜https://www.ebi.ac.uk/europepmc/webservices/rest/search?format=json&query=EXT_ID:33567185%20AND%20SRC:MED｜peer_reviewed｜2026-08-26
[F3]｜2021-02-12 文章結語之一｜由於是皮下注射，而且是每週要注射一次，也許只有極度肥胖的人才會有意願。｜http://professorlin.com/2021/02/12/%e6%b8%9b%e8%82%a5%e6%96%b0%e8%97%a5/｜editorial（專欄原文逐字）｜2026-08-26
[F4]｜2021-02-12 文章結語之二｜生活形態改變（飲食和運動）仍然是最重要的。｜http://professorlin.com/2021/02/12/%e6%b8%9b%e8%82%a5%e6%96%b0%e8%97%a5/｜editorial（專欄原文逐字）｜2026-08-26
[F5]｜STEP 1 延伸期：停藥後一年取回先前減重量約三分之二，且第 68 週時治療與生活型態介入一併停止｜One year after withdrawal of once-weekly subcutaneous semaglutide 2.4 mg and lifestyle intervention, participants regained two-thirds of their prior weight loss｜https://www.ebi.ac.uk/europepmc/webservices/rest/search?format=json&query=EXT_ID:35441470%20AND%20SRC:MED｜peer_reviewed｜2026-08-26
[F6]｜台灣許可證登載的體重控制適應症以「做為低熱量飲食及增加體能活動之輔助療法」起頭｜(１)做為低熱量飲食及增加體能活動之輔助療法，適用對象為成人且初始身體質量指數(BMI)為≥30kg/m2(肥胖)，或≥27 kg/m2至<30 kg/m2(過重)且至少患有一項體重相關共病，例如血糖異常(糖尿病前期或第二型糖尿病)、高血壓、血脂異常、阻塞性睡眠呼吸中止或心血管疾病。｜全部藥品許可證資料集，許可證字號「衛部菌疫輸字第001225號」適應症欄｜official_text｜2026-08-26
[F7]｜2022-06-05 文章轉述 SURMOUNT-1 三個劑量第 72 週的減重幅度｜結果顯示，與安慰劑相比，三種不同劑量的 tirzepatide 都具有持續減肥的作用。在 72 週時，接受 15 mg 劑量 tirzepatide 的參與者平均體重減輕了 20.9%，接受 10 mg 劑量的參與者平均體重減輕了 19.5%，而 5 mg 劑量的參與者平均體重減輕了 15%。｜http://professorlin.com/2022/06/05/%e5%89%9b%e5%87%ba%e7%88%90%e7%9a%84%e6%b8%9b%e8%82%a5%e6%96%b0%e8%97%a5%e8%87%a8%e5%ba%8a%e8%a9%a6%e9%a9%97%e7%b5%90%e6%9e%9c/｜editorial（專欄原文逐字）｜2026-08-26
[F8]｜SURMOUNT-1 完整論文摘要第 72 週體重平均變化：5 mg −15.0%、10 mg −19.5%、15 mg −20.9%、安慰劑 −3.1%｜The mean percentage change in weight at week 72 was -15.0% (95% confidence interval [CI], -15.9 to -14.2) with 5-mg weekly doses of tirzepatide, -19.5% (95% CI, -20.4 to -18.5) with 10-mg doses, and -20.9% (95% CI, -21.8 to -19.9) with 15-mg doses and -3.1% (95% CI, -4.3 to -1.9) with placebo｜https://www.ebi.ac.uk/europepmc/webservices/rest/search?format=json&query=EXT_ID:35658024%20AND%20SRC:MED｜peer_reviewed｜2026-08-26
[F9]｜2023-09-03 文章記載該項審查的啟動與預計完成時間｜這項調查是在7月3日啟動，而預計在11月完成。｜http://professorlin.com/2023/09/03/%e6%b8%9b%e8%82%a5%e7%a5%9e%e8%97%a5%ef%bc%8c%e8%87%aa%e6%ae%ba%e9%a2%a8%e9%9a%aa/｜editorial（專欄原文逐字）｜2026-08-26
[F10]｜歐洲藥品管理局 PRAC 於 2024-04-12 公布結論：現有證據不支持 GLP-1 受體促效劑與自殺及自傷念頭、行為之間的因果關聯｜The PRAC has concluded that the available evidence does not support a causal association between the Glucagon-Like Peptide-1 receptor agonists (GLP-1) – dulaglutide, exenatide, liraglutide, lixisenatide and semaglutide – and suicidal and self-injurious thoughts and actions.｜https://www.ema.europa.eu/en/news/meeting-highlights-pharmacovigilance-risk-assessment-committee-prac-8-11-april-2024｜official_doc｜2026-08-26
[F11]｜同一份公告載明委員會認為不需更新產品資訊｜After reviewing the available evidence from non-clinical studies, clinical trials, post-marketing surveillance data and the available studies the PRAC considers that no update to the product information is warranted.｜同 F10｜official_doc｜2026-08-26
[F12]｜美國食品藥物管理局 2026-01-13 藥物安全通訊：要求藥證持有者自標示移除自殺意念與行為風險資訊，受影響產品為 Saxenda、Wegovy、Zepbound｜FDA is requesting that drug application holders remove information regarding the risk of suicidal ideation and behavior (SI/B) from the labeling of glucagon-like peptide-1 receptor agonist (GLP-1 RA) medications that currently include such language. The affected products are Saxenda (liraglutide), Wegovy (semaglutide), and Zepbound (tirzepatide). This action follows a comprehensive FDA review that found no increased risk of SI/B associated with the use of GLP-1 RA medications.｜https://www.fda.gov/drugs/drug-safety-communications/fda-requests-removal-suicidal-behavior-and-ideation-warning-glucagon-peptide-1-receptor-agonist-glp｜official_doc｜2026-08-26
[F13]｜該審查所依據的統合分析涵蓋 91 個安慰劑對照試驗、107,910 名受試者｜There were 91 placebo-controlled GLP-1 RA medication trials in the meta-analysis that included 107,910 patients (60,338 treated with a GLP-1 RA and 47,572 treated with placebo). The results did not show an increased risk for SI/B or for other relevant psychiatric adverse events such as anxiety, depression, irritability, or psychosis.｜同 F12｜official_doc｜2026-08-26
[F14]｜該審查另包含一項涵蓋 2,243,138 名使用者的回溯世代研究｜The study population included 2,243,138 users (1,161,983 initiated on a GLP-1 RA and 1,081,155 initiated on a SGLT2i) from 10 data partners during the period between October 1, 2015, and September 20, 2023. After controlling for baseline confounders in the study, FDA did not find an increased risk of intentional self-harm in GLP-1 RA users compared to SGLT2i users.｜同 F12｜official_doc｜2026-08-26
[F15]｜台灣藥品許可證公開資料集共 28 欄，無警語、禁忌、交互作用、副作用或不良反應欄位，故台灣標示現況無法由此資料查得｜許可證字號、註銷狀態、註銷日期、註銷理由、有效日期、發證日期、許可證種類、舊證字號、通關簽審文件編號、中文品名、英文品名、適應症、劑型、包裝、藥品類別、管制藥品分類級別、主成分略述、申請商名稱、申請商地址、申請商統一編號、製造商名稱、製造廠廠址、製造廠公司地址、製造廠國別、製程、異動日期、用法用量、包裝與國際條碼｜全部藥品許可證資料集 CSV 表頭列｜official_text｜2026-08-26
[F16]｜2024-03-08 文章轉述之一：效果平台期與停藥後回升｜在建立瘦瘦筆功效的試驗中，體重減輕在 12 至 18 個月內達到平台期。而且，如果停止用藥，患者通常會在一年內恢復減掉的體重。｜http://professorlin.com/2024/03/08/%e7%98%a6%e7%98%a6%e7%ad%86%ef%bc%8c%e6%9c%89%e6%95%88%e5%97%8e/｜editorial（專欄原文逐字；其所轉述之原始來源本文未另行核對）｜2026-08-26
[F17]｜2024-03-08 文章轉述之二：一年後仍持續使用的比例｜現實世界的長期耐受性也很低。在一項大型分析中，只有 27% 的瘦瘦筆使用者在 1 年後仍堅持使用。｜http://professorlin.com/2024/03/08/%e7%98%a6%e7%98%a6%e7%ad%86%ef%bc%8c%e6%9c%89%e6%95%88%e5%97%8e/｜editorial（專欄原文逐字；其所轉述之「一項大型分析」本文未另行核對，故不採用該數字）｜2026-08-26
[F18]｜SURMOUNT-4：換為安慰劑者第 36 週至第 88 週體重平均變化 +14.0%｜The mean percent weight change from week 36 to week 88 was -5.5% with tirzepatide vs 14.0% with placebo (difference, -19.4% [95% CI, -21.2% to -17.7%]; P < .001).｜https://www.ebi.ac.uk/europepmc/webservices/rest/search?format=json&query=EXT_ID:38078870%20AND%20SRC:MED｜peer_reviewed｜2026-08-26
[F19]｜195,915 人分析：12 個月停用 36.5%；只有肥胖者 50.3%｜Overall prevalence of GLP-1 agonist discontinuation was 26.2%, 30.8%, and 36.5% at 3, 6, and 12 months, respectively／Patients with obesity only had a higher prevalence of discontinuation at 12 months vs those with T2D only and those with both (50.3% vs 35.8% and 34.2%).｜https://www.ebi.ac.uk/europepmc/webservices/rest/PMC11127113/fullTextXML（Results 段）｜peer_reviewed｜2026-08-26
[F20]｜125,474 人世代研究：一年停用比例無糖尿病 64.8%、有糖尿病 46.5%｜One-year discontinuation was significantly higher for patients without type 2 diabetes (64.8% [95% CI, 64.4%-65.2%]) compared with those with type 2 diabetes (46.5% [95% CI, 46.2%-46.9%]).｜https://www.ebi.ac.uk/europepmc/webservices/rest/search?format=json&query=EXT_ID:39888616%20AND%20SRC:MED｜peer_reviewed｜2026-08-26
[F21]｜2026-01-04 文章對台灣核准款數的註記｜（註：截至2025年底，共有6款瘦瘦筆在台灣核准使用，但由於其中兩款是「同藥不同名」，所以嚴格來說只有4款）｜http://professorlin.com/2026/01/04/%e7%98%a6%e7%98%a6%e7%ad%86%e9%99%8d%e8%a1%80%e5%a3%93/｜editorial（專欄原文逐字）｜2026-08-26
[F22]｜2026-01-04 文章結語｜但是，我還是要再重複我在瘦瘦筆，有效嗎所說的：光是靠瘦瘦筆並無法達到長期減肥。要達到真正有效的長期減肥，還是要靠均衡飲食和規律運動。｜http://professorlin.com/2026/01/04/%e7%98%a6%e7%98%a6%e7%ad%86%e9%99%8d%e8%a1%80%e5%a3%93/｜editorial（專欄原文逐字）｜2026-08-26
[F23]｜2024-03-19 讀者 ray 之留言（逐字，含原文用字）｜林教授您好，想詢問一下以下網址https://reurl.cc/M4EDg4 ，(藥廠終於克服蛋白質類藥品胰妥讚無法口服吸收的問題)，這段敘是否有問題，經消化後還有效果嗎? 還是並不是用腸胃吸收就可達到效果? 因為教授常說口服方式無法吸收，至少會變成雙胜肽或三胜肽才能吸收。｜http://professorlin.com/2024/03/08/%e7%98%a6%e7%98%a6%e7%ad%86%ef%bc%8c%e6%9c%89%e6%95%88%e5%97%8e/comment-page-1/#comment-16076｜editorial（讀者留言原文逐字）｜2026-08-26
[F24]｜2024-03-25 作者回覆之第一段（逐字）｜(藥廠終於克服蛋白質類藥品胰妥讚無法口服吸收的問題)，這段敘是有一點小問題，因為胰妥讚並非蛋白質類藥品。而是胜肽類藥品。｜http://professorlin.com/2024/03/08/%e7%98%a6%e7%98%a6%e7%ad%86%ef%bc%8c%e6%9c%89%e6%95%88%e5%97%8e/comment-page-1/#comment-16134｜editorial（專欄原文逐字）｜2026-08-26
[F25]｜該次追問之後，2024-04-10 有一篇專門討論口服胜肽研發的文章發表｜降血糖藥，口服胜肽艱辛研發｜http://professorlin.com/2024/04/10/%e9%99%8d%e8%a1%80%e7%b3%96%e8%97%a5%ef%bc%8c%e5%8f%a3%e6%9c%8d%e8%83%9c%e8%82%bd%e8%89%b1%e8%be%9b%e7%a0%94%e7%99%bc/｜editorial（專欄文章標題與發表日）｜2026-08-26
[F26]｜本站既有心理健康整理文已述「文獻已更新，後續研究尚未確認風險增加」｜林教授原文曾如實記錄監管機關因通報而啟動對 semaglutide／GLP-1 受體促效劑自殺與自傷風險的審查；文獻已更新，後續研究尚未確認風險增加，因此今日較合適的說法是持續監測並按個人病史評估，而不是把它列為已確立的藥物傷害。｜https://km.idaeo.ai/post/health/the-evidence-based-truth-about-mental-health-und｜editorial｜2026-08-26
[F27]｜實查該許可證適應症欄：「輔助療法」出現、「最重要」未出現，故法規未對藥物與生活型態做重要性排序｜（實查輸出）衛部菌疫輸字第001225號適應症欄 NFKC 後：'輔助療法' in text = True ／ '最重要' in text = False｜全部藥品許可證資料集，許可證字號「衛部菌疫輸字第001225號」適應症欄，NFKC 正規化後做子字串比對｜official_text（欄位值）＋editorial（我們的比對）｜2026-08-26

FAQ

Is this article pointing out that Professor Lin got something wrong?: No. The frame this article uses is that the literature has been updated. An article cannot cite data that appeared after it was published; that is chronology, not a question of judgement. What this article does is mark each piece with its publication date and the literature available then, and connect it to the primary sources that came later.
Does the 2023 article about suicide risk still stand?: What that article said at the time was that the European Medicines Agency was investigating a signal, and that statement was accurate on the day[F9]; it did not write the signal up as a conclusion. The European Medicines Agency later published its conclusion on 12 April 2024, finding that the available evidence does not support a causal association[F10] and that no product-information update was needed[F11]; the US Food and Drug Administration issued a safety communication on 13 January 2026 stating that its evaluation did not identify an increased risk, and requesting that the relevant application holders remove that warning from their labelling[F12].
So has the warning been removed everywhere?: It cannot be read that way. The verb in the US announcement is requesting that application holders remove it, and the products named are the three US products Saxenda, Wegovy, and Zepbound[F12]. A regulator making a request and every label in circulation actually completing the update are two different things. This article could not find Taiwan's current position — the regulator's public drug licence dataset has 28 fields with no warnings or adverse-reactions field[F15], so we write "not found as of the verification date".
Why does the article not simply use the figure "only 27% were still using it after a year"?: Because that figure comes from an analysis cited by a viewpoint piece which that article was relaying, and this article did not go back and check its original source[F17]. We use instead peer-reviewed studies whose originals can be looked up directly: an analysis of 195,915 people giving 36.5% discontinuation at 12 months (50.3% among those with obesity only)[F19], and a cohort study of 125,474 people giving one-year discontinuation of 64.8% (in those without diabetes)[F20]. Different studies define discontinuation differently, and the figures are not interchangeable.
The 2022 article cited a conference press release — is that reliable?: Citing a conference presentation does carry risk, because it has not been through complete peer review. But the after-the-fact check in this instance found that the full paper's abstract gives week-72 weight changes for the three doses as −15.0%, −19.5%, and −20.9%[F8], matching the 15%, 19.5%, and 20.9% written in the article at the time[F7]. Reliability has to be checked afterwards; it cannot be judged from the source type alone.
How many of these drugs are actually approved in Taiwan?: There is no single answer, because the colloquial term is not a statutory category. In his January 2026 article he noted himself that six were approved for use in Taiwan but that, because two of them are the same drug under different names, strictly speaking there are four[F21]. Counting by licence number instead gives a different figure again. To compare two lists, first confirm whether the denominator is brands, ingredients, or licences.
Why are typos in the quotations left uncorrected?: Because correcting them makes the quotation no longer verbatim. This article's rule is: quotations are always verbatim, with no corrected typos and no ellipsis splicing; where an extract is necessary, it runs continuously from the start of a sentence to its end. The rule applies equally to the author's text and to reader comments — the reader comment and the author's reply in the article are preserved exactly as they were[F23][F24].
How can I check the quotations in this article myself?: Every quotation has a definition line in the sources section stating the source URL and the verification date. The original URLs are all public pages that can be opened directly and searched using the first ten characters of the quotation. The two regulatory announcements each carry an official URL as well[F10][F12].

Source anchors

Cite this article

林編輯・《Where the Literature Went Five Years After Professor Lin Wrote About These Drugs: An Update Line That Can Be Checked Sentence by Sentence》・IDAEO 知識庫・2026-08-26・https://km.idaeo.ai/post/health/professor-lin-glp1-update

更新 2026-08-26T13:09:21.253Z · server-rendered · four-language · IDAEO 知識庫

Where the Literature Went Five Years After Professor Lin Wrote About These Drugs: An Update Line That Can Be Checked Sentence by Sentence

1. Why do this, and the rules this article follows

2. 2021-02: the literature ran only to week 68

3. 2022-06: figures from a conference press release, later confirmed unchanged by peer review

4. 2023-09: a signal under investigation, later resolved

5. 2024-03: the rebound he relayed, later given concrete figures

6. 2026-01: he is updating too

7. One reader's comment, demonstrating how follow-up works

8. The evolution of literature is not about who was right

Citations, one by one

FAQ

Source anchors

Cite this article

Related articles