Why the scientific literature is important

Why the scientific literature is important

? New knowledge generated by RCTs takes 17 years to be incorporated into practice (Institute of Medicine)

? Hodgkin ? Prior to 1970: 5-10% for 4 year survival ? 1970 study: 80% complete remission ? 11 years before results were disseminated

? Testicular cancer ? Cure rate 80% ? 3 years before results were disseminated

Concordance between clinical practice and published evidence Findings from The National Dental Practice-Based Research Network JADA 2014

But many readers do not know how to assess the evidence critically, she said. "To be completely honest, probably it does have a big impact because most people who use the literature are not accustomed to doing critical analysis of it."

Medscape Medical News > Conference News Conflicts of Interest Prevalent in Dental Research Laird Harrison March 27, 2013

1

Levels of Evidence

1906: Pure Food and Drug Act

? Product Labeling

? Cocaine ? Opiates ? Alcohol

? Electricity



1938: Federal Food, Drug, and Cosmetic Act

? diethylene glycol led to over a hundred deaths ? mostly children

? "The Radium water worked fine until his jaw came off" WSJ, 1932



1962: Kefauver-Harris Amendment

1976: Medical Device Amendment

? Dalkon Shield? 3

? TMJ implants: "... a new era in (ProplastTeflon) TMJ reconstruction has begun, resulting in increased benefits to the patients whom we all serve."



"GRASSLEY SECURES INDEPENDENT REVIEW OF FDA

APPROVALS"

? Torcetrapib (2006) raises HDL

? Avandia (2007) controls blood glucose

? Vytorin (2008) lowers cholesterol

? Trasylol (2008) lowers bleeding

Senator Charles Grassley (R, Idaho) Picture from Washington Post

2

50% failure rate at 6 years "One of the biggest disasters in orthopedic history" BMJ

1

SR of RCTs

RCT

All-or-none

SR cohort studies

2

Cohort studies

Ecological studies

3

SR case-control studies Case-control study

Case-reports and case-series

Expert opinion, Biological plausibility, Animal evidence, Bench research

Perio No tx

Pre-term

+

-

280(12.6%) 1948

2228

261(11.7%) 1968 2229

Randomized Controlled Trial

CHX

R

Placebo

tooth loss No tooth loss

tooth loss No tooth loss

3

Randomized Trial on Tooth Loss

? 1101 elders ? Randomly assign to

CHX (550) or placebo (551) ? Followed for 4.4 years ? Results:

? CHX: 14 teeth/1000 tooth-years

? Placebo: 13.7/1000 tooth-years

RCT Pre-natal fluoride/caries

Pre-natal F

caries caries-free

30

368

398

7.5%

Control

36

364

400

9.0%

Measures of Association

NNT, OR, RR, HR, RD, AR, p -values

RCT

Lesion at day 6

+

-

+ 376 (30%) 878 1254

Penciclovir

- 526 (41%)

757 1283

Data adapted from K. Kjaerheim et al., 1998

376/1254

OR

(1- 376/1254) 526/1283

376x757 526x878

0.62

(1- 526/1283)

Data adapted from K. Kjaerheim et al., 1998

RCT

Lesion at day 6

+

-

+ 376

878 1254

Penciclovir

- 526

757 1283

Data adapted from K. Kjaerheim et al., 1998

4

Odds ratio = 0.62

The odds for a non-healed lesion at day 6 among individuals on penciclovir is only 62% of that for someone who was on placebo.

% odds decrease = |1 - 0.62| * 100%

The odds for a non-healing lesion among individuals on penciclovier decreased by 38%

The Atlantic 2010

1

SR of RCTs

RCT

All-or-none

SR cohort studies

2

Cohort studies

Ecological studies

3

SR case-control studies Case-control study

Case-reports and case-series

Expert opinion, Biological plausibility, Animal evidence, Bench research

5

Effect size & 95% confidence interval Vertical "no-effect" line

Effect size & 95% confidence interval Vertical "no-effect" line

Effect size & 95% confidence interval Vertical "no-effect" line

Study ID

0.09Fdaily Driscoll (1982) Heifetz (1981) Ringelberg (1982) Subtotal (I-squared = 22.1%, p = 0.277) . 0.09Fweekly Driscoll (1982) Heifetz (1981) Craig (1981) Ringelberg (1982) Horowitz (1971) Chikte (1996) Subtotal (I-squared = 0.0%, p = 0.622) . 0.09Fbiweekly Torell (1965) Poulsen (1983) Subtotal (I-squared = 90.7%, p = 0.001) . Overall (I-squared = 57.6%, p = 0.009)

NOTE: Weights are from random effects analysis

ES (95% CI)

% Weight

21.43 (-1.40, 44.26) 40.69 (23.39, 57.98) 22.75 (3.79, 41.72) 29.54 (16.81, 42.26)

7.93 10.27 9.50 27.70

36.61 (16.02, 57.20) 30.64 (12.12, 49.15) 30.77 (-6.52, 68.06) 20.36 (0.60, 40.12) 35.92 (19.13, 52.72) 15.34 (-6.09, 36.78) 28.77 (20.40, 37.14)

8.81 9.70 4.19 9.16 10.51 8.47 50.84

49.10 (39.83, 58.38) 4.37 (-20.71, 29.45) 28.32 (-15.41, 72.04)

14.32 7.14 21.46

29.81 (21.01, 38.60) 100.00

A Systematic Review of Class IIs and Herbst Appliances -

AUTHORS' CONCLUSIONS: It is not possible to provide any evidencebased guidance to recommend or discourage any type of orthodontic treatment to correct Class II division 2 malocclusion in children.

Cochrane Database Syst Rev. 2006

6

"Caries of teeth is restricted to people and animals who eat liberally of carbohydrate containing foods. Carnivorous man and animals do not suffer from this disease"

McCollum, 1941

1

SR of RCTs

RCT

All-or-none

SR cohort studies

2

Cohort studies

Ecological studies

3

SR case-control studies Case-control study

Case-reports and case-series

Expert opinion, Biological plausibility, Animal evidence, Bench research

What is "normal" Blood Pressure

? Blood pressure history

? Age 55 : 169/98 ? Age 59: 188/105 ? Age 63: 226/118

? Physician assessment:

"no more than normal for a man his age"

New England Journal of Medecine 1995 (332): 1038-1039

7

CHD and Lipids, October 1966

? Framingham

? 2022 males ? 1689 females

? Livermore

? 1961 males

Gofman et al., Circulation, 1966; 679

Cohort Study Smoking/Oral Cancer

? Follow-up a cohort of 10960 individuals; 8857 smokers, 2103 nonsmokers

? Enumerate the number of smokers and nonsmokers that develop oral cancer during a 20 year period

? Tabulate the data in a 2x2 table

Ischemic Heart Disease, Atherosclerosis, and Longevity

Parameters

Incident Heart Disease

Systolic blood pressure Sign. Elevated

Diastolic blood pressure Sign. Elevated

Weight

Sign. Elevated

Cohort Study

Oral carcinoma

+

-

+ 56 (0.6%) 8801

Smoking

- 4 (0.2%) 2099

8857 2103

Data adapted from K. Kjaerheim et al., 1998

56/8857

OR

(1- 56/8857) 4/2103

56x2099 4x8801

3.3

(1- 4/1203)

Data adapted from K. Kjaerheim et al., 1998

Cohort Study

Oral carcinoma

+

-

+ 56

8801

8857

Smoking

-4

2099 2103

Data adapted from K. Kjaerheim et al., 1998

8

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download