Skip to main content

Table 2 Multivariable logistic regression model relating drop-out between SPARCLE1 and SPARCLE2 to characteristics recorded in SPARCLE1

From: Predictors of drop-out in a multi-centre longitudinal study of participation and quality of life of children with cerebral palsy

Reason for drop-out: (i) Not traced n/N = 51/805 (6%) (ii) Death n/N = 32/754* (4%) (iii) Refusal n/N = 119/710**(17%) (iv) All non-response n/N = 214/804 (27%)
  n/N (%) OR (95%CI) p n/N (%) OR (95%CI) p n/N (%) OR (95%CI) p n/N (%) OR (95%CI) p
Child impairment: Walking ability                0.005      
I. Walks without limitation            51/225 (22%) 1.0 (0.7 to 1.4)       
II. Walks with limitation            15/147 (10%) 0.3 (0.2 to 0.6)       
III. Walks with assistive devices            25/126 (20%) 0.8 (0.5 to 1.3)       
IV. Unable to walk, limited self-mobility            15/100 (15%) 0.5 (0.3 to 0.9)       
V. Unable to walk, severely limited self-mobility            13/112 (12%) 0.4 (0.2 to 0.8)       
Child impairment: Feeding           <0.001           
By mouth       14/700 (2%) 1.0 -            
By tube (partially or completely)       18/54 (33%) 9.1 (4.0 to 21)            
Information missing, excluded from analysis       0/1 (0%) - -            
Child impairment: IQ           <0.001           
≥50       4/533 (1%) 1.0 -            
<50       28/221 (13%) 9.2 (2.9 to 29)            
Information missing, excluded from analysis       0/5 (0%) - -            
Parental educational qualifications      0.002           0.001      <0.001
Above university entry 5/201 (2%) 0.3 (0.1 to 0.7)        25/190 (13%) 0.6 (0.4 to 0.9)   36/201 (18%) 0.4 (0.3 to 0.7)  
Intermediate 28/413 (7%) 1.0 (0.7 to 1.5)        57/359 (16%) 1.0 (0.8 to 1.3)   111/412 (27%) 1.0 (0.8 to 1.2)  
None or lowest formal qualifications 18/191 (9%) 1.4 (0.8 to 2.4)        37/161 (23%) 1.9 (1.2 to 3.0)   67/191 (35%) 1.7 (1.2 to 2.4)  
Information missing, excluded from analysis 0/6 (0%) - -        1/4 (25%) - -   3/6 (50%) - -  
Parental stress                <0.001      <0.001
Below 75th percentile            73/517 (14%) 1.0 (0.7 to 1.3)   138/581 (24%) 1.0 (0.8 to 1.3)  
Above 75th percentile            36/166 (22%) 2.2 (1.5 to 3.3)   60/190 (32%) 1.8 (1.3 to 2.5)  
Not recorded            10/27 (37%) 4.3 (1.8 to 10.3)   16/33 (48%) 3.6 (1.7 to 7.6)  
Family structure                     <0.001
Married, living with partner                 135/567 (24%) 1.0 (0.8 to 1.2)  
Unmarried, living with partner                 35/81 (43%) 2.7 (1.7 to 4.3)  
Single or separated and living with parents                 10/18 (56%) 4.7 (1.7 to 13.0)  
Single, living alone                 34/138 (25%) 1.1 (0.7 to 1.6)  
Information missing, excluded from analysis                 0/1 (0%) - -  
Region      <0.001           <0.001      <0.001
North England 6/114 (5%) 1.0 (0.4 to 2.3)        22/102 (22%) 1.0 (0.6 to 1.7)   34/114 (30%) 1.0 (0.6 to 1.5)  
Northern Ireland 3/101 (3%) 0.5 (0.2 to 1.6)        8/93 (9%) 0.3 (0.2 to 0.7)   16/101 (16%) 0.4 (0.2 to 0.7)  
West Sweden 1/81 (1%) 0.3 (0.0 to 1.9)        6/74 (8%) 0.4 (0.2 to 1.0)   13/81 (16%) 0.4 (0.2 to 0.8)  
East Denmark 4/113 (4%) 1.1 (0.4 to 3.1)        26/103 (25%) 2.1 (1.3 to 3.5)   36/113 (32%) 1.9 (1.2 to 3.0)  
Southwest Ireland 6/96 (6%) 1.3 (0.6 to 3.0)        12/85 (14%) 0.7 (0.3 to 1.3)   23/96 (24%) 0.8 (0.5 to 1.3)  
Central Italy 16/85 (19%) 5.3 (3.0 to 9.2)        22/63 (35%) 2.6 (1.5 to 4.5)   44/85 (52%) 3.3 (2.1 to 5.2)  
Southeast France 8/67 (12%) 3.0 (1.4 to 6.3)        7/57 (12%) 0.5 (0.2 to 1.2)   17/67 (25%) 0.9 (0.5 to 1.7)  
Southwest France 6/74 (8%) 1.9 (0.8 to 4.5)        10/63 (16%) 0.7 (0.4 to 1.5)   21/73 (29%) 0.1 (0.7 to 1.9)  
Northwest Germany 1/74 (1%) 0.2 (0.0 to 1.8)        6/70 (9%) 0.3 (0.1 to 0.8)   10/74 (14%) 0.3 (0.2 to 0.7)  
  1. n = number of non-respondents; N = total number analysed; % = percentage non-responders; OR = odds ratio; CI = confidence interval; p = likelihood ratio test statistic p-value for removal of characteristic from model; OR above 1.0 indicate a greater risk of non-response in that category than in the reference category.
  2. For variables with more than two categories, 95%CI were based on quasi-variances [18].
  3. * The denominator for analysis of deaths was all traced families.
  4. ** The denominator for analysis of refusal was all traced families with live children, excluding those who had moved out of the region (13 families) or who had language problems (1 family).