Page 97

Proefschrift binnenwerk Manon Ernst_DEF.indd

T H E   U S E   O F   A   P R E D I C T I O N   M O D E L   T O   P R E V E N T   A B N O R M A L  W E I G H T   I N   C H I L D H O O D   I N   T H E   Y O U T H   H E A L T H   C A R E   P R A C T I C E     To   compare   our   model   with   other   methods   (using   the   weight   for   length   growth   chart   and   calculation   of   PI)   to   estimate   the   growth   of   children,   we   calculated   the   sensitivity  and  specificity  of  these  methods,  all  applied  at  the  age  of  approximately   1  year  and  using  +  2SD  as  cut-­‐off  for  overweight,  shown  in  Table  1-­‐3.     The  sensitivity  of  our  model  is  46%  (50/109).  In  other  words,  46%  of  children  with   a  PIm  >  PIe     +  2SD  at  the  age  of  1  year  developed  overweight  at  the  age  of  5  years.   The  specificity  is  84%  (93/111).  In  other  words  84%  of  the  children  with  a  PIm  <  PIe     +   2SD   at   the   age   of   1   year   did   not   develop   overweight   at   the   age   of   5   years.     The   likelihood  ratio  for  positive  outcomes  (sensitivity  /  1-­‐specificity)  is  2.8  and  for  neg-­‐ ative  outcomes  (1-­‐sensitivity  /  specificity)  0.6.  The  diagnostic  odds  ratio  (true  posi-­‐ tive  *  true  negative  /  false  positive  *  false  negative)  is  4.4.       The   accuracy   to   predict   overweight   at   the   age   of   5   years,   based   on   the   prediction   model  applied  at  the  age  of  1  year,  is  65%  (143/220).     Table  1:   Sensitivity  (true  positive)  and  specificity  (true  negative)  of  the  prediction  model  (applied  at  1   year,  for  prediction  of  overweight  at  the  age  of  5  years).     Overweight   Not  overweight   TOTAL   PIm  >  PIe    +  2SD   50  (46%)   True  positive     18  (16%)   False  positive   68   PIm  <    PIe    +  2SD   59  (54%)   False  negative     93  (84%)   True  negative     152   TOTAL   109   111   220   Se:  46%;  Sp:  84%;  LR  +:  2.88;  LR  -­‐:  0.64;  DOR:  4.4   PIe=  Expected  Ponderal  Index,  PIm  =  Measured  Ponderal  Index,     SD  =  Standard  deviation.   Se=   Sensitivity,   Sp=   Specificity,   LR   +=   Positive   likelihood   ratio,   LR   -­‐=   Negative   likelihood   ratio,   DOR=   Diagnostic  odds  ratio.     The  sensitivity  of  the  weight  for  length  growth  chart  at  the  age  of  1  year  and  the  PI   calculation  at  the  age  of  1  year  is  16%  and  28%  respectively.  The  specificity  values   are  95%  and  89%  respectively.  Likelihood  ratios  for  positive  outcomes  are  3.2  re-­‐ spectively  2.5;  those  for  negative  outcomes  are  0.8  for  both  models.  The  diagnostic   ratios  are  3.6  respectively  3.1.     The  accuracy  of  the  weight  for  length  growth  chart  at  the  age  of  1  year  and  PI  calcu-­‐ lation  at  the  age  of  1  year  is  56%  and  58%  respectively.       95  


Proefschrift binnenwerk Manon Ernst_DEF.indd
To see the actual publication please follow the link above