How should cost data in pragmatic randomised trials be analysed?

Thompson, Simon G.

doi:10.1136/bmj.320.7243.1197

Cited by 694 publications

(539 citation statements)

References 21 publications

(19 reference statements)

Supporting

Mentioning

522

Contrasting

Unclassified

Order By: Relevance

“…Resolution of such infiltrates is usually rapid and complete. 2,4 However, in our case significant stromal scarring limited the final visual acuity. The addition of topical corticosteroids in severe cases seems to shorten the disease period and hasten resolution.…”

Section: Commentcontrasting

confidence: 54%

Sterile corneal ulcer with ring infiltrate and hypopyon after recurrent erosions

Papathanassiou

Gartry

2006

Eye

View full text Add to dashboard Cite

Section: Commentcontrasting

confidence: 54%

Sterile corneal ulcer with ring infiltrate and hypopyon after recurrent erosions

Papathanassiou

Gartry

2006

Eye

View full text Add to dashboard Cite

“…A. & Thompson, S. G., 2000. Analysis of cost data in randomised trials: an Willingness to pay fo r unit reductio n in alco ho l co nsumptio n (£) …”

Section: Discussionmentioning

confidence: 99%

“…Initially, traditional statistical tests for differences in costs were undertaken. Despite the skewed distribution of the cost data, parametric tests were used on untransformed costs, because this method enables inferences to be made about the arithmetic mean (Thompson & Barber, 2000). Non-parametric bootstrapping was used to assess the robustness of confidence intervals to non-normality of the cost distribution (Barber & Thompson, 2000).…”

Section: Methodsmentioning

confidence: 99%

Cost-effectiveness of screening and referral to an alcohol health worker in alcohol misusing patients attending an accident and emergency department: A decision-making approach

Barrett

Byford

Crawford

et al. 2006

Drug and Alcohol Dependence

View full text Add to dashboard Cite

Telephone: 020 7848 5091 Fax: 020 7701 7600 2 We present the cost and cost-effectiveness of referral to an alcohol health worker (AHW) and information only control in alcohol misusing patients. The study was a pragmatic randomised controlled trial conducted from April 2001 to March 2003 in an accident and emergency department (AED) in general hospital in London, England. A total of 599 adults identified as drinking hazardously according to the Paddington Alcohol Test were randomised to referral to an alcohol health worker who delivers a brief intervention (n=287) or to an information only control (n=312). Total societal costs, including health and social care costs, criminal justice costs, accommodation costs and productivity losses, and clinical measures of alcohol consumption were measured. Levels of drinking were observably lower in those referred to an AHW at twelve months follow-up and statistically significantly lower at six months followup. Total costs were not significantly different at either follow-up. Referral to AHWs in an AED produces favourable clinical outcomes and does not generate a significant increase in cost. A decision-making approach revealed that there is at least a 70% probability that referral to an AHW is more cost-effective than the information only control in reducing alcohol consumption among AED attendees with a hazardous level of drinking.3

show abstract

“…Thompson and Barber [2] however argued strongly that transformation of cost data is not appropriate. They recommended the use of the sample mean as an estimator, with confidence limits derived either from standard asymptotic theory, or from non-parametric bootstrapping of the sample mean.…”

Section: Introductionmentioning

confidence: 99%

Parametric modelling of cost data: some simulation evidence

et al. 2005

View full text Add to dashboard Cite

SummaryRecently, commentators have suggested that the distributional form of cost data should be explicitly modelled to gain efficiency in estimating the population mean. We perform a series of simulation experiments to evaluate the usual sample mean and the mean estimator of a lognormal distribution, in the context of both theoretical distributions and three large empirical datasets. The sample mean is always unbiased, but is somewhat less efficient when the population distribution is truly lognormal. However the lognormal estimator can perform appallingly when the true distribution is not lognormal. In practical situations, where the true distribution is unknown, the sample mean generally remains the estimator of choice, especially when limited sample size prohibits detailed modelling of the cost data distribution. Copyright

show abstract

How should cost data in pragmatic randomised trials be analysed?

Cited by 694 publications

References 21 publications

Sterile corneal ulcer with ring infiltrate and hypopyon after recurrent erosions

Sterile corneal ulcer with ring infiltrate and hypopyon after recurrent erosions

Cost-effectiveness of screening and referral to an alcohol health worker in alcohol misusing patients attending an accident and emergency department: A decision-making approach

Parametric modelling of cost data: some simulation evidence

Contact Info

Product

Resources

About