Nutritionists and other Professionals use statistics to help decide what is best for the general public. This is a very important paper in Lancet setting out the principles that are necessary for good practice. (Murray Lancet 2007, vol 369, 862-73)
Towards good practice for health statistics: lessons from the Millennium Development Goal health indicators
Health statistics are at the centre of many global health controversies. Several factors are influencing the supply and demand for high quality health information. The health-related Millennium Development Goals give a good example.
Thousands of indicators are recommended but of them are well measured. The international health community needs to focus its efforts on improving measurement of a small set of priority areas. Priority indicators should be selected on the basis of public-health significance and measurability.
Health statistics can be divided into three types: crude, corrected, and predicted.
Health statistics are prerequisites for planning and strategic decision making, programme implementation, monitoring progress towards targets, and assessment of what works and what does not.
Crude statistics that are biased have no role in any of these steps; corrected statistics are preferred. For strategic decision making, when corrected statistics are unavailable, predicted statistics can play an important part.
To monitor progress towards agreed targets and assessment of what works and what does not, however, predicted statistics should not be used.
The most effective method to reduce the controversies over health statistics and to encourage better primary data collection and the development of better analytical methods is a strong commitment to provide an explicit data audit trail.
This initiative would make available the primary data, all post-data collection adjustments, models including covariates used for farcasting and forecasting, and necessary documentation to the public.
Definitions of technical terms
A variable measured to monitor progress or assess what works and what does not.
Validity refers to the extent to which a measurement is capturing what it is intended to measure. There are different types of validity such as face validity, content validity, criterion validity (denoting predictive validity and concurrent validity), and construct validity (denoting convergent and discriminant validity).
Reliability refers to the repeatability or consistency of a set of measurements or measuring instrument, for example, test-retest reliability where a test and a retest are compared.
Measurements arc comparable if the same value means the same thing in the settings being compared. Two thermometers, one in Farenheit and one in Celsius, can both be valid and reliable but they do not give comparable results.
Prediction about ranges of values that are not in the investigator’s sample (ie, that the investigator’s data set does not cover).
Prediction about individuals, populations, etc, in time outside
the time range of the investigator’s sample.
Forecasting is the process of estimation in unknown situations. Predicting is a moie general term and connotes estimating for any time series, cross-sectional, or longitudinal data Forecasting is commonly used when discussing time series data.
Farcasting is trying to predict the value of a variable in a place that may be far away but is not a future value.
The prior is a reflection of some information the investigator has before the observations in the data set (the investigator should state explicitly how the information on the prior was obtained). The prior is the sum of what is known about the relationship under study.
Murray Lancet 2007, vol 369, 862-73
- Martin Eastwood