Methodological and statistical problems in the construction of composite measurement scales: A survey of six medical and epidemiological journals

Coste, Joël; Fermanian, Jacques; Venot, Alain

doi:10.1002/sim.4780140402

Cited by 124 publications

(84 citation statements)

References 39 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Reliability is the property of a method of measurement that ensures that it is repeatable. 13 Reliability is at best difficult and perhaps impossible to determine for pain scales 1,19,20 because http://www.pediatrics.org/cgi/content/full/110/3/e33pain intensity does not remain constant long enough to allow inter-or intraobserver comparisons. Pain is a dynamic state and changes in intensity over brief periods of time.…”

Section: Discussionmentioning

confidence: 99%

Validation of 2 Pain Scales for Use in the Pediatric Emergency Department

Bulloch

Tenenbein

2002

Pediatrics

View full text Add to dashboard Cite

ABSTRACT. Objective. To determine the construct, content, and convergent validity of 2 self-report pain scales for use in the untrained child in the emergency department (ED).Methods. A prospective study was conducted of all children who presented to an urban ED between 5 and 16 years of age inclusive after written informed consent was obtained. Children were excluded if they were intoxicated, had altered sensorium, were clinically unstable, did not speak English, or had developmental delays. Children marked their current pain severity on a standardized Color Analog Scale (CAS) and a 7-point Faces Pain Scale (FPS). They were then asked whether their pain was mild, moderate, or severe. Children were then administered an analgesic at the discretion of the attending physician and asked to repeat these measurements. For assessing content validity, the scales were also administered to age-and gender-matched children in the ED for nonpainful conditions. Convergent validity was assessed by determining the Spearman correlation coefficient between the 2 pain scales.Results. A total of 60 children were enrolled, 30 with pain and 30 without, with a mean age of 9.3 ؎ 3. ABBREVIATIONS. ED, emergency department; CAS, Color Analog Scale; FPS, Faces Pain Scale; IQR, interquartile range; CI, confidence interval; ACCS, Analog Continuous Chromatic Scale. I n assessing a child's pain, a measuring tool must take into account a child's age, cognitive level, type of pain, and the situation in which the pain is occurring. 1,2 No single measure is useful for all children with all types of pain, which can be acute, chronic, or recurrent. However, it should be possible to have a practical, valid, and reliable measure to evaluate a child's pain for any particular setting.The most effective method of pain measurement is self-report. [1][2][3][4][5][6][7][8] Children as young as 3 years have been shown reliably to self-report pain intensity with the use of several assessment tools. 1,2 However, pain management has been inadequately studied in the acute setting. 9 Any assessment of pain management in children requires a valid tool. There are no published data regarding the validity of pediatric pain scales in the untrained child in the emergency department (ED). Agreement among measures in one context does not necessarily suggest agreement in another context. 10 The primary objective of this study was to determine the construct validity of 2 self-report pain scales in the untrained pediatric ED patient. The secondary objectives were to determine the content and convergent validity of these 2 tools. METHODSThis was a prospective, descriptive study conducted in an urban ED at a children's hospital, with an annual census of 36 000. All children who presented to the ED between 5 and 16 years of age inclusive were eligible for enrollment. The minimum age of 5 years was chosen, despite some original research showing that children aged 3 and 4 years also can use these measures, because self-report measures of pain severity have been widely used in t...

show abstract

Section: Discussionmentioning

confidence: 99%

Validation of 2 Pain Scales for Use in the Pediatric Emergency Department

Bulloch

Tenenbein

2002

Pediatrics

View full text Add to dashboard Cite

show abstract

“…13,33. 34 Logistic regression methods require a fixed follow-up period (e.g. 10 years) and the outcome is reduced to whether or not the subject developed CHD during the follow-up period.…”

Section: Discussionmentioning

confidence: 99%

Multivariate risk estimation for coronary heart disease: the Busselton Health Study

Knuiman

Bartholomew

1998

Australian and New Zealand Journal of Public Health

View full text Add to dashboard Cite

ortality from coronary heart disease (CHD) inAustralia has been M falling for more than 20 years, but the rates remain high and a cause for concern. Risk factors associated with increased morbidity and mortality from CHD include hypertension, smoking, elevated serum total cholesterol, low high-densitylipoprotein (HDL) cholesterol, diabetes, obesity, left ventricular hypertrophy and previous history of CHD. Moderate levels of alcohol consu.mption are associated with decreased morbidity and mortality from CHD.CHD is a multifactorial disease and risk should be estimated by assessing all cardiovascular risk factors simultaneously. A simple method for obtaining an integrated or composite risk score is to categorise an individual as being 'at risk' or 'not at risk' according to whether the risk factor exceeds some arbitrary threshold, and then to add up the number of factors with 'at risk' values. This method is easily applied in practice and will identify high-risk subjects.However, CHD occurs in individuals with moderate elevations of several risk factors as well as in individuals who have extreme elevations of few risk factors, and therefore this simple composite score may fail to identify high-risk subjects with multiple risk factors at moderately elevated values which do not exceed the thresholds.A more effici'ent approach is to use a continuous multivariate risk score which does not use thresholds and will recognise that an individual with borderline elevation of two or three risk factors may be at the same risk as another individual with a serious elevation of one of them. ' Investigators from the Framingham Heart Study in the United States have developed risk scores for predicting the development of CHD4.5 and handbooks and programs for calculators have been de~eloped.~.' The best known of these scoring systems calculates risk based on age, sex, systolic or diastolic blood pressure, plasma total cholesterol, ECG evidence of left ventricular hypertrophy, diabetes and smokmg status. More recently published equations have also included HDL cholesterol.* A number of other studies have produced risk scoring systems for men.'-'* There are few risk scores developed for women. No CHD risk scores have been developed from Australian data.The development of several risk scores has been partly motivated by the belief that risk scores developed from one population may not be applicable to another population. A review of 15 published estimated logistic risk functions for CHD morbidity/mortality found that there were differences in the odds ratios between studies and the authors concluded that the 'use of a scoring system derived from logistic regression to produce a measure of risk for a different population may be more misleading than advantageous'.13 However, some evaluations of risk scores derived from one population and applied in other settings produced findings that do not support this vie^.'^'^-'^ Nevertheless, locally derived scores would be more acceptable to clinicians and general practitioners and thus more lik...

show abstract

“…The number of factors retained was determined from eigenvalues greater than or equal to 1 [9]. Factor analysis with rotation enabled identification of the dimensions in the questionnaire, and showed that the items selected measured domains that were identical to those hypothesised at the outset.…”

Section: Test-retest Reliabilitymentioning

confidence: 99%

“…This concept, which cannot be measured by direct observation of the care provided, involves the identification of expectations, needs, perceptions, past experiences, opinions and attitudes of patients [4,7,8]. Several authors have thus considered that the assessment of satisfaction required an operational formalisation of the concept into dimensions with their constituent items making up questionnaires [7], and hence this rapidly entailed the need to assess the psychometric properties of such instruments [9].…”

Section: Introductionmentioning

confidence: 99%