Calculate Z-scores for single individual

[tamuri]

• measurement-to-sds

• age for gestation

  • see vba code for exact algorithm for adjustment

• input age as

  • date of birth + date of measurement
  • age in days + weeks + months + years

• note interaction between input boxes

  • e.g. height+weight automatically calculated BMI
  • but adjusting BMI doesn't adjust height, weight (can be inconsistent)

• available measurements:

  • determined by what's available in reference
  • current LMSgrowth allow for max 6, selected by user
    • default from british1990 growth reference
      • height, weight, BMI, head, sitting height, leg length
  • calculate Z-scores for multiple measurements:
    • SDS,
    • centile,
    • % predicted (of mean value),
    • predicted (the mean value),
    • % CV ([S] * 100),
    • skewness ([L] value),
    • LLN5 (5th centile of distribution at this age

• input unit of measurement are fixed by the application (metric)

[tamuri]

Input form from LMSgrowth add-in

[statist7]

The shiny calculator looks good. Obviously it's early days, but I've a couple of comments, one general and the other specific.

The specific one is that leg length is calculated as the difference between height and sitting height. So if you have height and sitting height then you also have leg length. This means you can treat leg length like BMI, i.e. calculate it as soon as height and sitting height are known.

The more general point is the presentation of numerical results. I hate presenting more decimal places than necessary, and I've written guidelines - see https://adc.bmj.com/content/100/7/608 - about this. For example z-scores never need more than 2 decimal places, and often 1 is enough. It would be good to have a general function that simplified the process of rounding, controlling both decimal places and significant digits as per my recommendations. My z2cent function has this in mind.

[tamuri]

Thanks, Tim. It was our intention to implement rounding after discussion with you, so that's no problem.

We'll add calculation of sitting height/leg length.

[statist7]

Hi Asif,

I’m slightly concerned that not much seems to be happening with the project.

Could we meet to discuss timescales, so we are on the same page?

Thanks, Tim

[tamuri]

@statist7 - given the relationship between height, sitting height and leg length. If some given height and sitting height are both 50th centile, do we expect the leg length to then also be 50th centile? I assumed so, but maybe not? e.g. I noticed using uk90 that 1 yr old boy, height = 75.51 (50th), sitting height = 48.64 (50th) means leg length is 26.87 (19th centile)

[statist7]

You’re right in theory but not in practice 😀 The uk90 height reference uses different data from the sitting height and leg length references, so they don’t match exactly.

Bw Tim

[tamuri]

@giordano Just found these on issue #7, note on correction for gestational age:

Separately we need an option to adjust age for gestational age in preterm babies. The calculation is:

adjusted age = postnatal age + gestational age - 40 weeks

Preterm is gestational age < 37 weeks, while term is 37 to <42 weeks, with 40 weeks the expected term gestational age.

The adjustment is applied up to 2 years postnatal age but not beyond. As an example, a baby born at 27 weeks and measured at 52 postnatal weeks has an adjusted (or post-term) age of 39 weeks.

[tamuri]

From Tim (https://github.com/UCL/LMSgrowth2/pull/43#issuecomment-602607350):

The IOTF reference is for BMI, but it doesn't work with the sample dataset as there is no BMI column. There needs to be a way to say, calculate BMI from weight and height. And ideally it would use a user-specified formula that could be extended to other constructed measures such as sitting height divided by leg length.

[tamuri]

• [ ] Fix required: if age of child is outside of growth reference range, throws error