Wednesday, October 26, 2011

NEJM Audio Summary - Oct 20, 2011

TIME TABLE
00:45 | Earlier versus Later Start of Antiretroviral Therapy in HIV-Infected Adults with Tuberculosis
02:29 | Timing of Antiretroviral Therapy for HIV-1 Infection and Tuberculosis
04:08 | Integration of Antiretroviral Therapy with Tuberculosis Treatment
05:41 | When to Start Antiretroviral Therapy in HIV-Associated Tuberculosis
06:15 | Brief Report: Inflammatory Skin and Bowel Disease Linked to ADAM17 Deletion
07:29 | Neighborhoods, Obesity, and Diabetes — A Randomized Social Experiment
08:54 | Panretinal Photocoagulation for Proliferative Diabetic Retinopathy
10:42 | Case 32-2011 — A 19-Year-Old Man with Recurrent Pancreatitis
12:31 | Stalking Influenza Diversity with a Universal Antibody
14:20 | Defining Essential Health Benefits — The View from the IOM Committee
16:07 | Medical Device Innovation — Is “Better” Good Enough?
17:58 | The Supply-Side Economics of Abortion
19:50 | Multiple Intracranial Tuberculomas
20:52 | Jejunal Diverticular Bleeding
NEJM BLOGGERS
EXCERPTED SCRIPTS
00:45 | "Earlier versus Later Start of Antiretroviral Therapy in HIV-Infected Adults with Tuberculosis" by François-Xavier Blanc from CHU Bicêtre, Le Kremlin-Bicêtre, France. When to initiate antiretroviral therapy in patients with newly diagnosed HIV infection and tuberculosis has been debated. This study from Cambodia determine whether the earlier initiation of ART (2 weeks after the onset of tuberculosis treatment), as compared with later initiation (8 weeks afterward), could reduce mortality among patients with advanced immunodeficiency. The risk of death was significantly reduced in the group that received ART earlier, with 59 deaths among 332 patients (18%), as compared with 90 deaths among 329 patients (27%) in the later-ART group. The risk of tuberculosis-associated immune reconstitution inflammatory syndrome was significantly increased in the earlier-ART group. Irrespective of the study group, the median gain in the CD4+ T-cell count was 114 per cubic millimeter, and the viral load was undetectable at week 50 in 96.5% of the patients. Initiating ART 2 weeks after the start of tuberculosis treatment significantly improved survival among HIV-infected adults with CD4+ T-cell counts of 200 per cubic millimeter or lower.
02:29 | "Timing of Antiretroviral Therapy for HIV-1 Infection and Tuberculosis" by Diane  Havlir, from University of Califonia, San Francisco. This international study involving 809 patients with HIV and TB coinfection compared earlier therapy for both infections with waiting 8 to 12 weeks to initiate antiretrovirals after anti-TB therapy.In the earlier-ART group, 12.9% of patients had a new AIDS-defining illness or died by 48 weeks, as compared with 16.1% in the later-ART group. Among patients with screening CD4+ T-cell counts of less than 50 per cubic millimeter, 15.5% of patients in the earlier-ART group versus 26.6% in the later-ART group had a new AIDS-defining illness or died. Tuberculosis-associated immune reconstitution inflammatory syndrome was more common with earlier ART than with later ART (11% vs. 5%). The rate of viral suppression at 48 weeks was 74% and did not differ between the groups.
Overall, earlier ART did not reduce the rate of new AIDS-defining illness and death, as compared with later ART. In persons with CD4+ T-cell counts of less than 50 per cubic millimeter, earlier ART was associated with a lower rate of new AIDS-defining illnesses and death.
04:08 | "Antiretroviral Therapy Initiation during Tuberculosis Treatment" by Salim Abdool Karim from the Centre for the AIDS Programme of Research in South Africa, Durban. This study from South Africa determined the optimal time for initiation of ART in patients with HIV infection and tuberculosis. The incidence rate of the acquired immunodeficiency syndrome (AIDS) or death was 6.9 cases per 100 person-years in the earlier-ART group (18 cases) as compared with 7.8 per 100 person-years in the later-ART group (19 cases) . However, among patients with CD4+ T-cell counts of less than 50 per cubic millimeter, the incidence rates of AIDS or death were 8.5 and 26.3 cases per 100 person-years, respectively. The incidence rates of the immune reconstitution inflammatory syndrome (IRIS) were 20.1 and 7.7 cases per 100 person-years, respectively. Early initiation of ART in patients with CD4+ T-cell counts of less than 50 per cubic millimeter increased AIDS-free survival. Deferral of the initiation of ART to the first 4 weeks of the continuation phase of tuberculosis therapy in those with higher CD4+ T-cell counts reduced the risks of IRIS and other adverse events related to ART without increasing the risk of AIDS or death.
05:41 |  In Editorial M. Estée Török from Addenbrooke's Hospital, Cambridge, UK, writes that the results of these three trials provide important evidence to guide clinicians who are treating patients with HIV-associated tuberculosis. The evidence, including the study results provides support for the earlier initiation of ART in patients coinfected with HIV and tuberculosis who have advanced immunosuppression, apart from those who present with tuberculous meningitis.

Thursday, October 20, 2011

NEJM Audio Summary - Oct 13, 2011

TIME TABLE
00:50 | A Bird's-Eye View of Fever
01:14 | Incidence of Adenocarcinoma among Patients with Barrett's Esophagus
03:02 | The Problems with Surveillance of Barrett's Esophagus
04:02 | Somatic SF3B1 Mutation in Myelodysplasia with Ring Sideroblasts
05:29 | Breast-Cancer Adjuvant Therapy with Zoledronic Acid
06:47 | Oil-in-Water Emulsion Adjuvant with Influenza Vaccine in Young Children
08:21 | Adult Primary Care after Childhood Acute Lymphoblastic Leukemia
10:09 | Case 31-2011 — A 55-Year-Old Man with Oligometastatic Lung Cancer
11:56 | “Pay for Delay” Settlements of Disputes over Pharmaceutical Patents
13:42 | Bariatric Surgery in Adolescents
15:34 | The Coming Explosion in Genetic Testing — Is There a Duty to Recontact?
17:21 | Rethinking Health Care Labor
19:20 | The New Language of Medicine
21:05 | Tropical Calcific Pancreatitis
22:03 | Myotonia of the Tongue
NEJM BLOGGERS
EXCERPTED SCRIPT
"The New Language of Medicine", a perspective article by Pamela Hartzband from Beth Israel Deaconess Medical Center, Boston.
In the new language of medicine, patients are “customers” or “consumers”; doctors and nurses are “providers.”  These descriptors have been widely adopted in the media, medical journals, and even on clinical rounds. Yet the terms are not synonymous. The word “patient” comes from patiens, meaning suffering or bearing an affliction. Doctor is derived from docere, meaning to teach, and nurse from nutrire, to nurture. These terms have been used for more than three centuries.
The words we use to explain our roles are powerful. They set expectations and shape behavior. This change in the language of medicine has important and deleterious consequences. The relationships between doctors, nurses, or any other medical professionals and the patients they care for are now cast primarily in terms of a commercial transaction. The consumer or customer is the buyer, and the provider is the vendor or seller. To be sure, there is a financial aspect to clinical care. But that is only a small part of a much larger whole, and to people who are sick, it's the least important part. The words “consumer” and “provider” are reductionist; they ignore the essential psychological, spiritual, and humanistic dimensions of the relationship — the aspects that traditionally made medicine a “calling,” in which altruism overshadowed personal gain. Reducing medicine to economics makes a mockery of the bond between the healer and the sick. 
COMMENT
医学と経済学は非常に似ていると思うことがある。使用者のプリンシプルによって益にも害にもなり得る。例えば、彼の国では2008年にノーベル経済学賞受賞のポール・クルーグマンは、反ブッシュの先鋒であって、New York Timesへの投稿"Patients Are Not Consumers"がスクリプトの元のアメリカ版「患者様問題」を扱った俯瞰記事でも引かれている。顧みて、此岸の国では社会的な責任を果たそうとする経済学者が希少なのか、マスコミにシャットアウトされているのか、耳目にする機会が少ないように思う。参考までに、他のいくつかのポール・クルーグマンの論文は、山形浩生さんのページで和訳を読むことが出来る。

Tuesday, October 18, 2011

英語のヒアリング練習に最適"PlayItSlowly + gPodder"

先に「 英語のヒアリングにUMPlayer + gPodderが最適」と書いたばかりなのだが、もっと素敵なものを見つけた。"PlayItSlowly"というソフト。スピードが可変なのは一緒だが、繰り返す部分の指定もできるし、スピードで変わったピッチを変えることも出来る。そして何より、軽いことこの上ない。ただ、gPodder側から設定するときにちょっとしたコツがいる。下は、gPodderのCtrl+Pで設定(Preferences)を開いたところだが、なぜかメニューには加わらないので、鉛筆ボタンを押して、"playitslowly %f"とコマンドを手打ちしてやる。こんなちょっとしたひと手間で定期購読のポッドキャストが、最高のヒアリング反復の教材に変身するのです。

Monday, October 17, 2011

Gnome3のキーボードショートカット

デスクトップ環境を Gnome3にしたら、断然使いやすい。そこでキーボードショートカットを覚えるためのwallpaperを作ってみた。LibreOffice Impressのスライドショーをプリントスクリーンしただけのもの。MacのKeynoteに慣れると、Impressはこんなことぐらいでしか使う気にならないな。


Sunday, October 16, 2011

ubuntu 11.10に最初にインストールした10のソフト


ubuntu新版リリース時恒例の忘備録を書いておく。
  1. Chromium: UbuntuソフトウェアセンターでもデフォルトのFirefoxより人気のようだ。
  2. mozc: Google日本語入力。Ubuntuソフトウェアセンターからインストール。
  3. Dropbox: オンラインストーレジサービスだが、ローカルにもファイルをおいてくれるので、OSを問わず複数のPC間でファイルを同期することができる。サイトからパッケージをダウンロードしてインストール。Android版もあるし、USBメモリーは殆ど使わなくなりました。
  4. UMPlayer + gPodder: 音楽・動画プレーヤとポッドキャスト・アグリゲータ。前者は、サイトから、後者はセンターにある。UMPlayerは再生速度を変えられるので、gPodderで適当な英語ポッドキャストを購読して、前者をプレーヤに設定すると、英語のヒアリング訓練に最適。
  5. Mendeley: 文献管理ソフト。サイトにUbuntuへのパッケージも用意されている。iPhone版もあるが、Android版は未リリース。
  6. R + RStudio: 統計環境とそのIDE。前者はセンターから、後者はサイトからインストール。最近話題になっているggplot2を使った描画が出来るGUI環境"Deducer"を試したかったが、なぜかインストールに失敗する。今後の課題。
  7. Gephi: ネットワーク分析のPhotoshopの謳い文句。サイトからインストール。
  8. GNU Backgammon: サイトからインストール。いつの間にか不具合がなくなっている。
  9. Quarry + GnuGo: 碁盤ソフト+囲碁思考エンジン。両方ともUbuntuソフトウェアセンターからインストール。これが結構強い。
  10. AMOR: UNIXやMacを古くから使っている人には懐かしいプログラム。キャラクターがデスクトップを駆けまわるだけのものだが、ラテン語の「愛」という意味の単語と"Amusing Misuse Of Resources"の略を掛けた命名がハードボイルド風の諧謔に溢れている。
残念なのは、日本郵便の「はがきでデザインキット」のプラットフォームAdobeAirがLinuxのサポートを止めたこと。Synapticパッケージマネージャがデフォルトで入ってないこと。インストールはソフトウェアセンターからってことなんだろう。試しに、センターで「愛」を探してみると…


悲しそうなスマイリーがつれない返事。決定的にubuntuに欠けているのはユーモアの精神とアフォーダンスへの配慮って気がする。一方、日本のチームの素早いローカリゼーションには、いつも感謝。今回はまだのようだが、「アップグレード後にJapanese Teamのリポジトリを有効にする方法」の記事がある。

Wednesday, October 12, 2011

NEJM Audio Summary - Oct 6, 2011


TIME TABLE
00:51 | Adjuvant Trastuzumab in HER2-Positive Breast Cancer
02:44 | Steady Progress against HER2-Positive Breast Cancer
03:15 | Oral Acyclovir Suppression and Neurodevelopment after Neonatal Herpes
04:53 | Neonatal Herpes Simplex Infection and the Three Musketeers
06:01 | Randomized Trial of Oral Teriflunomide for Relapsing Multiple Sclerosis
07:39 | Adverse Health Outcomes in Women Exposed In Utero to Diethylstilbestrol
09:47 | Mechanisms of Disease: Filaggrin Mutations Associated with Skin and Allergic Diseases
11:30 | It's Not All in Your Head
13:36 | The Uncertain Future of Medicare and Graduate Medical Education
15:15 | Can Age-Associated Memory Decline Be Treated?
16:55 | Routine HIV Screening — What Counts in Evidence-Based Policy?
18:48 | Reforming Provider Payment — The Price Side of the Equation
20:31 | Critical Moments — Doctors and Patients
22:27 | Patent Urachus
23:22 | Traumatic Diaphragmatic Hernia

Sunday, October 9, 2011

Gephi 0.8 beta、リリース

グラフ界のPhotoshopをうたうネットワーク可視化ソフトGephiは、10月4日に0.8 beta版がリリースされた。ローカリゼーションの日本語訳をお手伝いさせて頂いたので感無量。プログラマでもなく、社会ネットワーク分析の専門家でもない私が身の程知らずに関わり、誤訳も多いと思うので、使い続けて改訂の責任を果たしたい。スクリーンショットは、文献1のTable 1をCalcで入力、CSVで出力したものをGephiにインポートしてみたところ。R+igraphとの使い分け、グラフの細かな設定など今後、試行錯誤でやっていこう。
文献
  1. A glossary of terms for navigating the field of social network analysis J Epidemiol Community Health. 2004 December; 58(12): 971–975.

Saturday, October 8, 2011

Vermont州の医療改革 ー NEJM podcastから


EXCERPTED SCRIPTS
March 31, 2011   
16:33 | "State-Based Single-Payer Health Care — A Solution for the United States?", a PERSPECTIVE article by William Hsiao, from Harvard School of Public Health, Boston.  
In Vermont, the status quo in health care has become untenable. The health system analysts at the Harvard School of Public Health were commissioned by the Vermont Legislature to develop and evaluate options for health system reform. The analysts found that Vermont faces a $150 million budget shortfall. They found that the system capable of producing the greatest potential savings and achieving universal coverage was a single-payer system. This system reduces the rate of cost increases over time by insulating major decisions about health care spending from politics, as well as by paying providers through capitation rather than fee for service, promoting delivery-system integration, and reducing the practice of defensive medicine by implementing a no-fault medical malpractice system. All told, the analysts estimated that Vermont could save 25% in health care expenditures over 10 years. Now,the governor of Vermont, wishing to implement systemic health care reform to solve the intertwined problems of high and escalating health care costs and access to care for the swelling ranks of the uninsured, is turning toward a single-payer system. [->Article]
August 18, 2011 
16:28 | "Single Payer Ahead — Cost Control and the Evolving Vermont Model",   a PERSPECTIVE article by Anya Rader Wallack, special assistant to the governor for health reform, Montpelier, Vermont. 
Governor Peter Shumlin of Vermont recently signed into law ambitious health care reform legislation that puts Vermont on course to implement a single-payer health care system and includes key provisions for controlling health care costs.The law creates a Health Benefit Exchange, consistent with the federal Affordable Care Act, and anticipates using it as the administrative structure for a publicly funded program of health insurance coverage for all Vermonters. The Shumlin administration believes that a single-payer system offers the greatest promise for reducing administrative waste and hassle in our health care system, guaranteeing coverage to everyone in the state, and relieving employers of the increasingly crushing burden of health insurance costs.Vermont's plan would include many features common to single-payer systems around the world: a global budget for health care expenditures, guaranteed coverage that is not linked to employment, and a single system of provider payments and administrative rules. The health care delivery system would remain privately owned, however, and the state could contract with a private insurer to perform some functions required of the single payer.Vermont is a small state (population about 625,000), but even on that scale, plans to launch a U.S. single-payer system seem noteworthy. [->Article]
COMMENT
英国のNHSは、南ウェールズのTredegarという炭鉱街発のものであった。 台湾をはじめ実績を重ねているハーバード大学公衆衛生学William C. Hsiao (蕭慶倫) 教授の手腕や如何に。Vermont州は合衆国のTredegarとなり得るのだろうか。そして、韓国や台湾に続き、アメリカにまで我が国の医療システムは医療保険制度の一元化において遅れをとってしまうことになるのだろうか?
REFERENCES


NEJMからの返信

前回投稿の終わりに、NEJM編集部に疑問をメールした旨を書いた。私が編集部にメールしたのは、2日であったが、編集部では著者に転送してくれて、返信の転送を受け取ったのが6日、予想外の迅速な返信に感動した次第。

肝心な内容だが、2011年9月29日号の「Cystisineの禁煙効果」の論文においてprimary outcomeの検定がフィッシャーの正確確率検定になっていないのではないかという疑問を指摘したのだが、著者の返信によると、私のメールを統計学者と検討してくれた上で、検定はフィッシャーの正確確率検定で行い、信頼区間や相対リスクはロジステイック回帰を用いたとのこと。正直、なぜ、わざわざロジステイック回帰を用いたのか、どのように計算したのかは、私には理解不能。しかし、著者側にも、査読側にも統計の専門家がいて、先日私がうっかり記したような「凡ミス」が起こり難い体制になっていることは納得できたし、一読者のつまらない疑問に対しても編集部のレスポンスが速いことには感服させられた一件でした。

Thursday, October 6, 2011

Lehrの公式(2)

比率の検定の場合、Lehrの公式をどのように使うかというのが主題。NEJMの9月29日号の論文"Placebo-Controlled Trial of Cytisine for Smoking  Cessation"を例にとる。METHODSのStatistical Analysisに以下のようにある。
With the use of previous trial data as a guide, we estimated that we would need to enroll 740 participants (370 in each group) to detect a between-group difference of 6 percentage points (6% vs. 12%) for the primary outcome, with 80% power and at an alpha level of 0.05.
プラセボ群と介入群の禁煙成功率をそれぞれ6%、12%と見積もって、検出力を80%、あわてて帰無仮説を棄却してしまう危険率の水準を0.05と標準的な設定している。t検定の場合、標本サイズ=16÷(δ/σ)2で計算したが、比の検定の場合は、右掲書によると、「標準化した差=δ/σ」の部分に下の値を入れて計算するそうだ。





Google電卓では、16 / ((0.06 / sqrt(0.09 * 0.91))^2) = 364となる。Rで確かめてみると、
> power.prop.test(p1=.12, p2=.06, power=0.8)
     Two-sample comparison of proportions power calculation
              n = 355.9428
             p1 = 0.12
             p2 = 0.06
      sig.level = 0.05
          power = 0.8
    alternative = two.sided
 NOTE: n is number in *each* group
で妥当な数字。

おまけで、当該論文の結果の比の検定をしてみた。
> prop.test(c(31,9), c(370,370), correct=F)
        2-sample test for equality of proportions without continuity
        correction
data:  c(31, 9) out of c(370, 370)
X-squared = 12.7914, df = 1, p-value = 0.0003482
alternative hypothesis: two.sided
95 percent confidence interval:
 0.02715792 0.09176099
sample estimates:
    prop 1     prop 2
0.08378378 0.02432432 
あれれ?論文の数字といっしょ! 論文には、"The absolute percentage-point difference between participants who met the criteria for abstinence in the two groups was tested with the use of Fisher's exact test."って書いてあるのに。一応、Fisher's exact testもしてみたが、やはり論文の検定結果とは異なる。結論がひっくり返るほどの間違いではないし、NEJMの査読でもこんな凡ミスを見逃すんだと思うと微笑ましいが、一応、指摘のメールをしておいた。返事あるかな?

Wednesday, October 5, 2011

NEJM Audio Summary - Sep 29, 2011

TIME TABLE
00:54 | Genomewide Association between GLCCI1 and Response to Glucocorticoid Therapy in Asthma
02:40 | A Step toward Personalized Asthma Treatment
03:27 | Changes in Forced Expiratory Volume in 1 Second over Time in COPD
05:15 | Variable Loss of Lung Function in COPD
05:42 | Placebo-Controlled Trial of Cytisine for Smoking Cessation
06:49 | Treatment of Neonatal Sepsis with Intravenous Immune Globulin
07:53 | End-of-Life Transitions among Nursing Home Residents with Cognitive Issues
09:27 | Treating Smokers in the Health Care Setting
11:12 | Case 30-2011 — A 62-Year-Old Woman with Renal Failure
12:57 | Restrictions on the Use of Prescribing Data for Drug Promotion
14:38 | Reducing Unnecessary Hospitalizations of Nursing Home Residents
16:26 | Who Owns Federally Funded Research? The Supreme Court and the Bayh–Dole Act
17:53 | Providing for Those Who Have Too Little
19:40 | Ectopic Tooth in the Maxillary Sinus
20:19 | Orbital and Cerebral Arteriovenous Malformations
NEJM BLOGGERS' REMARKS
EXCERPTED SCRIPT
05:42 | "Placebo-Controlled Trial of Cytisine for Smoking Cessation" by Robert West from University College London in United Kingdom.
Cytisine is a low-cost medication that is potentially beneficial in smoking cessation. In this trial participants were randomly assigned to receive cytisine or matching placebo for 25 days. The rate of sustained 12-month abstinence was 8.4% (31 participants) in the cytisine group as compared with 2.4% (9 participants) in the placebo group. The 7-day point prevalence for abstinence at the 12-month follow-up was 13.2% in the cytisine group versus 7.3% in the placebo group. Gastrointestinal adverse events were reported more frequently in the cytisine group. Cytisine was more effective than placebo for smoking cessation. The lower price of cytisine as compared with that of other pharmacotherapies for smoking cessation may make it an affordable treatment to advance smoking cessation globally.  
07:53 | "End-of-Life Transitions among Nursing Home Residents with Cognitive Issues" by Pedro Gozalo from Brown University, Providence, Rhode Island.Health care transitions in the last months of life can be burdensome and potentially of limited clinical benefit for patients with advanced cognitive and functional impairment.
In this study, we attempted to define patterns of transition among persons with advanced cognitive impairment who were in a nursing home 120 days before death. Among 474,829 nursing home decedents, 19.0% had at least one burdensome transition (range, 2.1% in Alaska to 37.5% in Louisiana). In adjusted analyses, blacks, Hispanics, and those without an advance directive were at increased risk. Nursing home residents in regions in the highest quintile of burdensome transitions (as compared with those in the lowest quintile) were significantly more likely to have a feeding tube (adjusted risk ratio, 3.38), have spent time in an ICU in the last month of life (adjusted risk ratio, 2.10), have a stage IV decubitus ulcer (adjusted risk ratio, 2.28), or have had a late enrollment in hospice (adjusted risk ratio, 1.17). Burdensome transitions are common, vary according to state, and are associated with markers of poor quality in end-of-life care. 
09:27 | "Treating Smokers in the Health Care Setting", a CLINICAL PRACTICE article by Michael Fiore from University of Wisconsin School of Medicine and Public Health, Madison.Tobacco use is directly responsible for about one fifth of all deaths in the United States. Approximately 70% of smokers in the United States see a primary care physician each year; thus, there is a good opportunity to deliver effective interventions for smoking cessation. Many smokers lack the motivation to attempt to quit, but various interventions can increase the number of smokers who make an attempt and the likelihood of success. One such intervention is motivational interviewing, in which the clinician uses nonconfrontational counseling to resolve the patient's ambivalence about quitting by encouraging choices that are consistent with the patient's long-term goals. Another approach is to encourage and instruct unwilling smokers to substantially and persistently reduce their daily smoking, while they are receiving nicotine-replacement therapy. Extensive research provides support for the effectiveness of counseling and pharmacologic interventions, alone or in combination, in increasing smoking-cessation rates among patients who are willing to attempt to quit. However, many smokers will not engage in counseling. Therefore, a smoker's willingness to try to quit should guide the management plan. In this article strategies to facilitate smoking cessation are reviewed, including documentation of smoking status for all patients, regular assessment of readiness to quit, counseling, and medications. An audio version of this article is available at NEJM.org.
14:38 | "Reducing Unnecessary Hospitalizations of Nursing Home Residents", a PERSPECTIVE article by Joseph Ouslander from Florida Atlantic University, Boca Raton.More than 1.6 million Americans live in nursing homes. Hospitalizations are common in this population; in 2006, 23.5% of the people admitted to a post-acute-care skilled-nursing facility were rehospitalized within 30 days. Studies suggest that many hospitalizations of US nursing home residents are inappropriate, avoidable, or related to conditions that could be treated outside the hospitals — and they cost over $4 billion per year. But the causes of preventable hospitalizations in this population are complex. One fundamental problem is not clinical but financial, stemming from a misalignment of Medicare and Medicaid: state Medicaid programs do not benefit from savings that Medicare accrues from prevented hospitalizations of nursing home residents, even though the nursing home incurs expenses when managing changes in condition without hospital transfer. In addition, nursing homes have a financial incentive to hospitalize residents who have Medicaid coverage, because after a 3-day inpatient stay, the resident may qualify for Medicare Part A payment for post-acute care in the nursing home at three to four times the daily rate paid by Medicaid. Multifaceted strategies will be needed to address the current incentives for hospitalization if we are to improve nursing home care and prevent unnecessary hospitalizations, with their related complications and costs.

Saturday, October 1, 2011

分散均一性(homoskedasticity)


前回のおまけのスクリプトでは、分散均一性が分かっていたので、その検定を端折っていた。不明な場合は、F検定で分散均一性を検定する必要がある。その上で分散が同じと言えないようであれば、Welch–Satterthwaiteの式を適用した方法を用いることになる。
> s <- round(rnorm(17, 162, 5.13), digits = 1)
> m <- round(rnorm(17, 156, 5.07), digits = 1)
> var.test(s, m)
F test to compare two variances
data:  s and m
F = 1.1808, num df = 16, denom df = 16, p-value = 0.7437
alternative hypothesis: true ratio of variances is not equal to 1
95 percent confidence interval:
 0.4276064 3.2605437
sample estimates:
ratio of variances
          1.180775
> qf(0.95, 16, 16)
[1] 2.333484
> t.test(s, m, var.equal=TRUE, paired=FALSE)
Two Sample t-test
data:  s and m
t = 4.199, df = 32, p-value = 0.0001996
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 3.331689 9.609488
sample estimates:
mean of x mean of y
 162.7176  156.2471
> qt(0.975, 32)
[1] 2.036933
> s <- round(rnorm(17, 162, 9.13), digits = 1)
> m <- round(rnorm(17, 156, 4.07), digits = 1)
> var.test(s, m)
F test to compare two variances
data:  s and m
F = 4.2636, num df = 16, denom df = 16, p-value = 0.006053
alternative hypothesis: true ratio of variances is not equal to 1
95 percent confidence interval:
  1.544013 11.773261
sample estimates:
ratio of variances
          4.263575
> qf(0.95, 16, 16)
[1] 2.333484
> t.test(s, m, var.equal=FALSE, paired=FALSE)
Welch Two Sample t-test
data:  s and m
t = 2.4679, df = 23.114, p-value = 0.02142
alternative hypothesis: true difference in means is not equal to 0
95 percent confidence interval:
 0.8204995 9.3089123
sample estimates:
mean of x mean of y
 162.2059  157.1412
> qt(0.975, 32)
[1] 2.036933