Sixteen-week versus standard eight-week prednisolone therapy for childhood nephrotic syndrome: the PREDNOS RCT

Notes

Permissions

Chapter 1 Introduction

Idiopathic nephrotic syndrome (INS) is the most common glomerular disorder of childhood, with an incidence of 2 per 100,000 child population in the UK. The disease presents at a median age of 2 to 3 years and is twice as common in boys than in girls. 1,2 There is ethnic variability in the disease incidence, with a fourfold to sixfold higher incidence in the UK South Asian population. 1,3,4

The onset of INS is characterised by the acute onset of heavy proteinuria, resulting in the development of hypoalbuminaemia and generalised oedema. There is not infrequently a delay in diagnosis, with the child having being treated for allergy prior to eventual presentation to a paediatrician or paediatric nephrologist as an emergency. The disease pathogenesis is poorly understood; however, both in vitro and in vivo experiments have identified the immune system to be dysregulated at the time of disease onset. 5 The presence of nephrotic syndrome places the child at increased risk of a number of complications, including thromboembolic disease and infection, particularly with Streptococcus pneumoniae. Prior to the development of adequate antibiotic and remission-inducing therapy, the mortality rate from INS was of the order of 50%, the majority of deaths being related to infection. 6,7

In excess of 90% of children who present with INS will respond to a course of high-dose corticosteroid therapy. 8 For this reason, the majority are treated empirically with a course of corticosteroids without a renal biopsy being performed. Those who respond to the treatment are given the diagnostic label of having steroid-sensitive nephrotic syndrome (SSNS). Only those with atypical features at presentation (age < 12 months or over 12 years, persistent hypertension or impaired renal function, gross haematuria, low plasma C3, hepatitis B or C virus positivity) and those who do not respond to this initial course of corticosteroid therapy undergo renal biopsy. 9 This is in contrast to practice in adult patients with nephrotic syndrome, in whom the causes of nephrotic syndrome are diverse and biopsy at presentation is routinely performed to establish a histological diagnosis and to guide subsequent therapy. Little emphasis is placed upon histological diagnosis in children with SSNS, as it has been shown that corticosteroid sensitivity rather than histology is the key prognostic indicator. 10 Those children who respond to corticosteroids generally have a good long-term prognosis with a low risk of developing chronic kidney disease; in contrast, those who are corticosteroid unresponsive suffer significant morbidity, and around 50% will progress to end-stage kidney failure, necessitating dialysis and kidney transplantation over a 15-year period. 11 The majority of these children will have focal segmental glomerulosclerosis on renal biopsy. In an early seminal study, conducted by the International Study of Kidney Disease in Children (ISKDC),2 a large cohort of children underwent renal biopsy at presentation prior to the commencement of corticosteroid therapy. The majority of those who responded to corticosteroids were noted to have minimal change disease (MCD) histology, so called because the appearance of the kidney tissue at light microscopic level is essentially normal. Somewhat confusingly, in much of the published literature, the terms MCD and SSNS are used interchangeably, although this is not strictly correct, as a small number of children with MCD do not respond to corticosteroids and, similarly, a small number of corticosteroid-sensitive children have a histological diagnosis other than MCD.

Following initial successful treatment with corticosteroids, around 80% of children with SSNS develop disease relapses necessitating further courses of high-dose prednisolone, and around 50% develop frequently relapsing nephrotic syndrome (FRNS), which is defined as two or more relapses within the first 6 months following presentation or four relapses within any 12-month period, or steroid-dependent nephrotic syndrome (SDNS), which is defined as relapses occurring within 14 days of discontinuation of corticosteroid therapy. 12 Similar to the presenting episode, nephrotic syndrome relapses are associated with a risk of significant complications, including sepsis, thrombosis, dyslipidaemia and malnutrition. 7 The treatment of relapses with repeated courses of high-dose prednisolone is associated with major adverse effects, including hip avascular necrosis, growth failure, hypertension, obesity, diabetes and behavioural problems. 13,14 Furthermore, children frequently have to miss school during relapses, resulting in impaired academic performance and parental absence from work.

When complications of repeated courses of corticosteroids develop, or when they are expected, a range of immunosuppressive strategies are employed in an attempt to reduce the frequency of disease relapses. These include the use of long-term, low-dose, alternate-day prednisolone, as well as a range of non-corticosteroid immunosuppressive agents, including levamisole, cyclophosphamide, ciclosporin, tacrolimus, mycophenolate mofetil and rituximab.

Early follow-up studies8,15 suggested that the long-term prognosis for children with SSNS was excellent, with all retaining normal kidney function and over 90% achieving long-term remission, with complete cessation of relapses by the end of puberty. However, subsequent studies have reported higher rates of relapsing disease persisting beyond childhood, with 19% of UK patients suffering ongoing relapses into early adult life. 16 In the majority of patients who achieve permanent remission during childhood, relapse occurs at 13–16 years of age. 17 However, by this time, the majority of affected children will have received a significant cumulative corticosteroid dose, and many will have been exposed to other immunosuppressive agents. The role of the paediatric nephrologist is to maintain the child with SSNS as being well and free from relapses, while at the same time minimising the adverse effects of exposure to corticosteroids and other immunosuppressive therapies, thus ensuring that they emerge as healthy adults free from relapses and with no significant long-term treatment-related morbidity.

The ideal initial corticosteroid regimen for use at presentation of childhood INS should rapidly induce urinary remission (defined as 3 consecutive days of zero or trace proteinuria) with resolution of oedema. It must be sufficient to prevent frequent relapses necessitating the use of alternative immunosuppressive agents, although not so intensive that serious corticosteroid-related AEs develop. The first standardised corticosteroid treatment regimen was introduced by the ISKDC in the 1960s and consisted of a dose of 60 mg/m² of prednisone (maximum 80 mg) given daily for 4 weeks followed by 40 mg/m² (maximum 60 mg) on 3 consecutive days out of 7 for a total of 4 weeks. 18 Many centres made a minor modification whereby 40 mg/m² was administered on alternate days rather than on 3 days out of 7 during the second 4-week period. Centres in the UK adopted the use of prednisolone rather than prednisone, as this was and remains the corticosteroid in routine use in UK paediatric practice, although children in the USA and other parts of Europe have continued to receive prednisone. These two agents are very closely related, with prednisone being metabolised to the active prednisolone following absorption.

Following the introduction of the ISKDC regimen, there has been significant debate regarding the optimal prednisolone regimen at the time of presentation of SSNS, and a number of randomised controlled trials (RCTs)19–28 have investigated whether giving a more or less intensive corticosteroid regimen at first presentation of INS affects the number of children suffering both disease relapses and adverse effects of corticosteroid therapy. One single RCT19 has shown that, compared with the ISKDC regimen, a less intensive regimen comprising a daily dose of 60 mg/m² of prednisone only until the urine was negative for 3 consecutive days (urinary remission, median time 14 days) followed by 4 weeks of 40 mg/m² of prednisone on alternate days resulted in a higher rate of disease relapse. No further studies have investigated this therapeutic strategy. By 2005, when the PREDnisolone in NephrOtic Syndrome (PREDNOS) study was being planned, a total of six RCTs20–25 had compared 2 months of prednisolone using the ISKDC regimen with a variety of different regimens of ≥ 3 months in duration. These regimens intensified the initial prednisolone regimen by increasing the duration of both the daily and alternate daily prednisolone phases. A Cochrane review29 of these six studies concluded that intensification of the initial corticosteroid therapy at disease presentation significantly reduced the rate of relapse at 12–24 months [risk ratio 0.7, 95% confidence interval (CI) 0.58 to 0.84]. There was an inverse linear relationship between treatment duration and risk of relapse [relative risk (RR) 1.26–0.112 duration; p = 0.03]. Furthermore, there was a significant reduction in the number of frequent relapsers and the mean relapse rate per participant per year. In addition to these six studies comparing the ISKDC regimen with longer duration corticosteroid regimens, a further four studies23,26–28 compared 3 months’ treatment with prednisolone with 6 months’ treatment; two were published only in abstract form. 27,28 Longer therapy duration resulted in a significantly decreased risk of relapse at 6 and 12 months (risk ratio 0.48, 95% CI 0.35 to 0.64, and risk ratio 0.57, 95% CI 0.45 to 0.71, respectively). Furthermore, the number of participants who developed FRNS was also lower in the 6-month group than in the 12-month group (risk ratio 0.55, 95% CI 0.38 to 0.80). Further analysis suggested that the benefits of more intensive corticosteroid therapy were more likely to be related to the increased duration of treatment than the higher cumulative dose; however, collinearity between treatment duration and dose prevented the Cochrane group29 from drawing definitive conclusions.

However, significant concerns were raised on a number of issues relating to the six studies contributing to the meta-analysis comparing 2 months of treatment with treatment for ≥ 3 months. The total number of participants was small, at only 520 participants across all six studies, and concerns were expressed about the quality of a number of these trials. One was (and remains) unpublished,22 with data available only in abstract form. None was placebo controlled or blinded in any way, and only two were at low risk of bias for allocation concealment. Trials with inadequate allocation concealment can exaggerate the efficacy of the experimental treatment by 30–40%, and meta-analysis of low-quality trials may overestimate the benefit of therapy. 30,31 Furthermore, only one of these trials was analysed on an intention-to-treat (ITT) basis; however, this same study indicated, in the discussion, that parents could exert some influence on which treatment group their child was allocated to, implying that the randomisation process was flawed. 23 It was also unclear whether or not there was a clinically useful reduction in the incidence of steroid-dependent disease and the use of second-line immunosuppressive agents. The studies also reported somewhat different corticosteroid-related adverse events, making interpretation of the impact that increased duration of corticosteroid therapy had on adverse effect profile difficult. Therefore, the authors of the Cochrane review29 (Dr Elisabeth Hodson and Professor Jonathan Craig) concluded that further well-designed and adequately powered RCTs were required to establish the optimum dose and duration of treatment, and were consulted from an early stage regarding the design of the PREDNOS study.

There has continued to be great debate regarding what the ideal corticosteroid regimen at disease presentation should be, and there is considerable variation in the treatment regimens used. Kidney Disease Improving Global Outcomes guidelines32 published in 2013 supported the conclusions of the Cochrane review,29 recommending that INS in children be treated initially with 60 mg/m² or 2 mg/kg of prednisone or prednisolone for at least 12 weeks (4–6 weeks daily followed by 40 mg/m² or 1.5 mg/kg every other day), followed by a slow tapering of dose over the next 2–5 months. Despite these recommendations, in the UK the majority of centres have continued to use the 8-week ISKDC regimen, as is the case in Canada, Nigeria and South Korea. In contrast, Germany, the Netherlands, Spain and other European nations have adopted a longer treatment regimen, as proposed by the Arbeitsgemainschaft für Pädiatrische Nephrologie (APN, the German Society for Paediatric Nephrology) and investigated in its RCT. 20 This consists of 6 weeks of daily prednisolone at a dose of 60 mg/m² followed by 6 weeks of alternate daily prednisolone at 40 mg/m². In France, a longer 18-week course of prednisone is in routine use. 33 A questionnaire survey reported significant heterogeneity in the regimens used in centres in the USA, with 13% using the ISKDC regimen, 7% using the APN regimen and many using either of these with a subsequent corticosteroid taper. 34 This genuine clinical equipoise confirmed the importance of conducting a high-quality RCT to determine whether or not extending the course of prednisolone beyond that recommended by the ISKDC was associated with improved clinical outcomes in UK children. We chose time to first relapse as our primary outcome measure and, following consultation with the British Association for Paediatric Nephrology and our patient group advisers from the Nephrotic Syndrome Trust (NeST) and the Renal Patient Support Group, we selected secondary outcome measures that were felt to be of clinical importance, including the incidence of FRNS and SDNS and the need for alternative, potentially more potent, immunosuppressive therapies. Given the paucity of high-quality information on the adverse effect profiles of standard course (SC) and extended course (EC) treatment courses of prednisolone, we also aimed to collect comprehensive adverse effect data, including abnormal behaviour, which was assessed through the use of the Achenbach Child Behaviour Checklist (ACBC). Abnormal behaviour is one of the most commonly reported adverse events (AEs) in routine clinical practice; however, it was rarely reported on in previously conducted clinical trials. Finally, we aimed to perform a detailed cost-effectiveness analysis to determine the relative cost and efficacy of the two regimens in quality-adjusted life-year (QALY) terms.

Since the commencement of the PREDNOS study, three further studies have reported their findings. 33,35,36 A well-conducted double-blind placebo-controlled RCT33 performed in the Netherlands aimed to ascertain whether the apparently better outcomes associated with prolonged prednisolone treatment occurred as a result of the increased duration of treatment or the higher cumulative dose of prednisolone administered. One hundred and fifty Dutch children were randomised to receive 3 months of prednisolone followed by 3 months of placebo or 6 months of prednisolone; both groups received equal cumulative doses of prednisolone (3360 mg/m², the same dose as that administered in the APN regimen) and were followed up for a median of 47 months. One hundred and twenty-six children commenced trial medication. A primary end point of the development of FRNS was selected and no difference was detected (45% with 3 months of prednisolone vs. 50% with 6 months of prednisolone). There was no difference in the number of participants who developed relapses (77% vs. 80%), the number requiring alternative immunosuppressive agents or the number of AEs. The authors concluded that the reduced relapse rate associated with longer prednisolone regimens observed in previous studies most likely occurred as a result of the increased cumulative prednisolone dose administered rather than the lengthening of the duration of the treatment course.

More recently, two high-quality studies,35,36 published alongside one another in Kidney International with an accompanying editorial by Hoyer,37 reported outcomes that differed significantly from those reported in the Cochrane review. 13 Sinha et al. 35 from New Delhi and four other Indian centres enrolled 181 children aged 1–12 years presenting for the first time with INS. Participants were treated with a dose of prednisone 2 mg/kg daily for 6 weeks followed by 1.5 mg/kg on alternate days for a further 6 weeks, and were then randomised in a double-blind manner to receive either placebo or prednisone in decreasing doses for a further 3 months. The total dose of prednisone received was 3530 mg in the 6-month group and 2792 mg in the 3-month group. There was no difference between the two groups in the chosen primary end point [the number of relapses per 12 months of follow-up (1.26 vs. 1.54, respectively; p = 0.21)] or the percentage of participants with relapses or frequent relapses. There was no significant difference in the mean time to first relapse. The authors concluded that extending initial prednisolone treatment from 3 to 6 months did not influence the course of illness in children with SSNS. The second study36 randomised 255 Japanese children presenting with INS to either the ISKDC regimen (total dose of 2240 mg/m²) or a 6-month prednisolone regimen comprising 4 weeks of daily prednisolone followed by 20 weeks of tapering alternate-day prednisolone (total dose of 3885 mg/m²). Median follow-up was 36.7 months in the 2-month group and 38.2 months in the 6-month group. The chosen primary end point was the time to development of FRNS and was similar in both groups [hazard ratio (HR) 0.86, 90% CI 0.64 to 1.16]. The time to first relapse was also similar in both groups, as was the number of relapses per year. The frequency and severity of AEs were similar in both groups, despite the 6-month group receiving a significantly higher median cumulative dose of prednisolone over 2 years. Yoshikawa et al. 36 concluded that prolongation of the initial corticosteroid regimen from 2 to 6 months did not improve patient outcomes.

Following the publication of these studies, in 2015 the Cochrane group performed an update of their systematic review and meta-analysis. 13 They reported that the addition of these three well-designed studies had changed the conclusion of their review. They noted that studies of long versus shorter duration of corticosteroid treatment had heterogeneous treatment effects, with the older studies that were rated as having a higher risk of bias tending to overestimate the effect of longer-course therapy compared with more recently published studies rated as having a low risk of bias. Among the studies rated as having a low risk of bias, the group found that there was no significant difference in the risk of FRNS between those given prednisolone for 2 or 3 months and those receiving treatment for longer durations or a higher total dose, indicating that there is no benefit of increasing the duration of prednisolone beyond 2 or 3 months in the initial episode of SSNS. 13

Chapter 2 Methods

Trial-related information, including the protocol, study information sheets, consent and assent forms and the case report forms, is available at the PREDNOS website (www.birmingham.ac.uk/prednos; accessed 17 August 2017).

Interventions

Treatment groups

Participants were randomised to receive either SC prednisolone therapy [the ISKDC regimen: a dose of 60 mg/m²/day of prednisolone (maximum dose 80 mg) for 4 weeks followed by 40 mg/m² (maximum dose 60 mg) on alternate days for a further 4 weeks] or EC prednisolone therapy [a dose of 60 mg/m²/day of prednisolone (maximum 80 mg) for 4 weeks followed by 60 mg/m² (maximum 60 mg) on alternate days for 2 weeks, with a subsequent gradual reduction in dose over a total of 12 weeks (tapering by 10 mg/m² every 2 weeks), resulting in a total course of prednisolone of 16 weeks]. The trial schema is shown in Figure 1.

FIGURE 1.

Trial schema.

All participants received the initial 4 weeks of prednisolone as open-label treatment prior to recruitment and randomisation into the study and the dispensed study medication prescribed was of 12 weeks’ duration (4 weeks of prednisolone and 8 weeks of placebo in those randomised to the SC group and 12 weeks of prednisolone in those randomised to the EC group). The treatment schedule and the prednisolone dose administered at each time point is outlined in Table 1. Matching placebo tablets were used to maintain the double blind at each time point.

TABLE 1 - Prednisolone regimens in the SC and EC groups

Placebo was added here to maintain the double-blinding, so that there was no difference in the number of tablets taken between the SC and EC groups.

Time (weeks)	Therapy prednisolone dose
	SC	EC
Open-label routine clinical treatment
0–4	60 mg/m2/day (maximum 80 mg)	60 mg/m2/day (maximum 80 mg)
Randomised phase
5–6	40 mg/m2/day (+placeboa) on alternate days	60 mg/m2/day on alternate days
7–8	40 mg/m2/day (+placeboa) on alternate days	50 mg/m2/day on alternate days
9–10	Placebo on alternate days	40 mg/m2/day on alternate days
11–12	Placebo on alternate days	30 mg/m2/day on alternate days
13–14	Placebo on alternate days	20 mg/m2/day on alternate days
15–16	Placebo on alternate days	10 mg/m2/day on alternate days

The entire course of study drug to be administered from weeks 5 to 16 was supplied in a blister pack by the central clinical trials pharmacy at the Birmingham Children’s Hospital (Figure 2). Prednisolone was supplied as 5-mg tablets alongside matching placebo, so that participants in both treatment groups received the same number of tablets at any time point in the study. The central pharmacy controlled allocation concealment and distributed trial medication once informed consent was obtained and randomisation had occurred. Both the study drug and the placebo were manufactured by Essential Nutrition Ltd (Brough, UK). Participants who were unable to swallow tablets whole were allowed to crush the study drug using a tablet crusher, which was supplied upon request. Parents were instructed to administer the study drug to participants first thing in the morning in keeping with routine clinical practice with prednisolone therapy.

FIGURE 2.

Blister pack of study drug as supplied to participants.

Delivered with the study drug, every participant also received a standard study pack containing a participant diary and a bottle of urinalysis sticks (Albustix^®, Siemens Healthcare Limited, Frimley, UK; Bayer Diagnostics, Tarry Town, NY, USA) for daily morning testing for proteinuria.

Blinding

All those involved in treating the participant, the participant and their parents/guardians were masked as to the randomised treatment allocation.

Chapter 3 Results

Recruitment

The PREDNOS study opened to recruitment in July 2011 and the first participant was recruited into the trial on 2 August 2011. Two hundred and thirty-seven participants were recruited and randomised, the last entering the study on 7 October 2014. The rate of recruitment was relatively constant with no evidence of seasonal variation (Figure 3). One hundred and eighteen participants were randomised to the SC group and 119 to EC group. The 237 participants were recruited from 86 (69%) of the 124 recruiting study sites; recruitment at the 86 sites varied between 1 and 19 participants. Individual site recruitment data are shown in Appendix 1. Participants had completed at least 2 years’ follow-up by October 2016.

FIGURE 3.

Recruitment of participants into the study.

Participant flow

Of the 237 participants randomised into the study, 44 (19%) were withdrawn from the trial (Figure 4). Following consent and randomisation, 14 participants (6%; nine in the SC group and five in the EC group) who had initially responded to open-label prednisolone, suggesting that they were corticosteroid sensitive, developed proteinuria (with time from commencing open-label prednisolone to withdrawal in these participants ranging from 26 to 35 days). These participants were deemed to be corticosteroid resistant and were withdrawn from the study as per the protocol. Their data were not included in any of the subsequent analysis, and so the ITT population was based on 223 participants.

FIGURE 4.

A Consolidated Standards of Reporting Trials diagram of participant flow through the trial. a, Two patients had discontinued their study medication. b, Two patients had discontinued their trial medication owing to relapse. Time points are at times post start of open-label treatment.

During the trial, 15 participants (6%) had their consent to participate in the study withdrawn (12 in year 1, one in year 2 and two in year 3), 11 participants (5%) became lost to follow-up (one in year 1, one in year 2, five in year 3 and four in year 4) and four participants (2%) withdrew for other reasons (two emigrated, one was withdrawn at the principal investigator’s discretion and one participant was randomised in error prior to entering remission with open-label treatment). For these 30 participants, the data collected up until the time of their withdrawal from the study were included in the analysis. Withdrawn participants were distributed evenly between the two groups (Table 2).

TABLE 2 - Participant withdrawal

Reason for withdrawal	Group (n)		Total (N = 237), n (%)
	SC	EC
Corticosteroid resistant	9	5	14 (6)
Withdrew consent	9	6	15 (6)
Lost to follow-up	1	10	11 (5)
Other reason	1	3	4 (2)
Exclusions (total)	20	24	44 (19)

Discontinuation of study medication

Eighty-six (39%) out of the 223 participants did not complete their course of study medication. The number of participants discontinuing study medication was higher in the SC group than in the EC group (SC 50% vs. EC 28%; p = 0.001). The predominant reason for discontinuation was the development of relapse during the 12-week period when double-blind study medication was being administered (n = 79, 35%). When relapses developed during this period of study drug administration, the protocol stated that study medication was to be discontinued and relapse treatment was to be commenced. The incidence of study drug discontinuation was comparable in the two groups up until week 8 (week 4 of randomised study drug), when participants in both groups were receiving active prednisolone. However, thereafter, there was an increase in the number of discontinuations of study medication in the SC group (Figure 5). The number of participants who discontinued owing to relapse was higher in the SC group (n = 50) than in the EC group (n = 29). Participants in the SC group were scheduled to receive active prednisolone until week 8 (week 4 of randomised study drug) followed by placebo for weeks 8–16 (weeks 4–12 of randomised study drug), whereas those in the EC group were scheduled to receive active prednisolone right through until week 16 (week 12 of randomised study drug). The majority of participants in the SC group (38/50) who discontinued the study drug because of relapse over this 16-week period did so during weeks 9–16, once active prednisolone had been discontinued and they were receiving placebo. In contrast, in the EC group, discontinuations owing to relapse were spread over the 12-week study drug period.

FIGURE 5.

Time to discontinuation of study medication.

The other seven participants (3%; SC, n = 4, vs. EC, n = 3) who prematurely discontinued study medication did so in a deviation from the study protocol. In four participants, it proved impossible to administer the study medication, including in crushed form; one participant refused to take the study medication and two families withdrew consent during the 12-week period of study drug administration (one participant refused to take their medication and the other participant’s parent stopped the trial medication without consulting the site).

Primary outcome

The number of participants who reported a relapse during the trial was 179: 88 out of 109 (81%) in the SC group and 91 out of 114 (80%) in the EC group. There was no significant difference in time to first relapse between the SC and EC groups (Figure 6a; unadjusted HR 0.87, 95% CI 0.65 to 1.17; log-rank p = 0.3). Analyses adjusting for the minimisation variables gave similar results (HR 0.86, 95% CI 0.64 to 1.16). The median [interquartile range (IQR)] time to first relapse was 87 (64.5–134) days for the SC group and 139 (90–179) days for the EC group. The time to first relapse using the actual relapse date is shown in Figure 6b.

FIGURE 6.

Time to first relapse. (a) When using time to relapse as 18 weeks if relapse occurred before 18 weeks; and (b) using actual date of relapse.

Secondary outcomes

The number of relapses experienced by participants in each treatment group is shown in Table 4. The number of relapses that participants experienced during the trial ranged from 0 to 15; there were eight participants in the SC group and nine in the EC group who experienced 10 or more relapses. The mean number of relapses did not differ (SC 3.61 vs. EC 3.98; IRR 1.09, 95% CI 0.86 to 1.39; p = 0.5) (see Table 4).

TABLE 4 - Secondary outcome measures

Total dose of prednisolone received during the study (following completion of study medication).

A ratio of less than one favours the EC group. A negative mean difference favours the EC group.

Secondary outcome measures	Group		Estimate (95% CI)	p-value
	SC (N = 109)	EC (N = 114)
Relapses
Number of relapses	394	454
Number of participants experiencing a relapse (%)	88 (81)	91 (80)	HR 0.87 (0.65 to 1.17)	0.3
Mean (SD) number of relapses per participant	3.61 (3.25)	3.98 (3.30)	IRR 1.09 (0.86 to 1.39)	0.5
Number of participants who developed FRNS (%)	55 (50)	60 (53)	RR 1.04 (0.81 to 1.35)	0.7
Number of participants who developed SDNS (%)	48 (44)	48 (42)	RR 0.96 (0.71 to 1.29)	0.8
Second-line immunosuppressive agents
Number of participants who received second-line immunosuppressants (%)	61 (56)	62 (54)	RR 0.97 (0.77 to 1.23)	0.8
Type of immunosuppressant received
Ciclosporin	6 (6%)	4 (4%)
Tacrolimus	8 (7%)	18 (16%)
Levamisole	35 (32%)	34 (30%)
Cyclophosphamide	31 (28%)	29 (25%)
Mycophenolate mofetil	13 (12%)	15 (13%)
Rituximab	5 (5%)	1 (1%)
Corticosteroid dose	n = 90	n = 94
Mean (SD) total prednisolone dose (mg)a	5474.6 (3697.3)	6674.1 (4998.2)	Mean difference 1199.5 (–83.8 to 2482.8)	0.07

There were no significant differences between the two treatment groups in the proportion of participants developing FRNS (SC 50% vs. EC 53%, RR 1.04, 95% CI 0.81 to 1.35; p = 0.7) or SDNS (44% vs. 42%, RR 0.96, 95% CI 0.71 to 1.29; p = 0.8) (see Table 4). The median time to the development of FRNS was 129 days for the SC group and 173 days for the EC group (HR 0.93, 95% CI 0.64 to 1.34). There was no difference between groups in the proportion of participants requiring alternative immunosuppressive agents (56% vs. 54%; p = 0.8). The most common immunosuppressive agents used were levamisole and cyclophosphamide. The total dose of prednisolone received during the study (following completion of study medication) was larger in the EC group (6674.1 mg) than in the SC group (5474.6 mg) but this was not statistically significant (p = 0.07).

Serious adverse events

There were 67 SAEs reported in 46 participants (21%): 39 SAEs in 27 participants (25%) in the SC group and 28 SAEs in 19 participants (17%) in the EC group (p = 0.1). The most common reasons for SAE reporting were admission for disease relapse or bacterial infection. Five SAEs in the SC group and six SAEs in the EC group were felt by the principal investigator to be related to the study drug, although none resulted in study drug discontinuation. There was one accidental death that was unrelated to the trial.

Adverse events

The most common AEs reported were increased appetite, poor behaviour (parent reported), Cushingoid facies, hypertrichosis and abdominal pain (Table 5). In the first 16 weeks of the trial, increased appetite was reported in 87% of participants (SC 87% vs. EC 86%), poor behaviour in 83% (SC 90% vs. EC 75%), Cushingoid facies in 67% (SC 66%, vs. EC 68%), hypertrichosis in 26% (SC 23%, vs. EC 30%) and abdominal pain in 26% (SC 28%, vs. EC 25%). By 24 months, these had increased to 94% (SC 94%, vs. EC 93%) for increased appetite, 87% (SC 93%, vs. EC 82%) for poor behaviour, 72% (SC 72%, vs. EC 73%) for Cushingoid facies, 39% (SC 38%, vs. EC 39%) for hypertrichosis and 45% (SC 47%, vs. EC 43%) for abdominal pain. At 16 weeks and at 6, 12 and 24 months, there were no significant differences between the groups in the cumulative number of participants reporting any of the AEs, except for poor behaviour, which was lower in the EC group than in the SC group. In the first 16 weeks, 90% in the SC group reported poor behaviour, compared with 75% in the EC group (RR 0.85, 95% CI 0.76 to 0.96). Differences were also seen at 6 months (91% vs. 81%, RR 0.90, 95% CI 0.82 to 1.00), 12 months (92% vs. 82%, RR 0.90, 95% CI 0.82 to 0.98) and 24 months (93% vs. 82%, RR 0.90, 95% CI 0.82 to 0.98).

TABLE 5 - Adverse event data: cumulative number of participants reporting AEs

Ophthalmoscopy was performed annually. Data on cataract were only available on 57 and 52 participants in the SC group and 60 and 54 participants in the EC groups at 1 and 2 years, respectively.

A ratio of less than one favours the EC group.

AE	Time point
	Week 16			6 months			12 months			24 months
	Group, n reporting AEs (%)		RR (95% CI)	Group, n reporting AEs (%)		RR (95% CI)	Group, n reporting AEs (%)		RR (95% CI)	Group, n reporting AEs (%)		RR (95% CI)
	SC (N = 109)	EC (N = 114)		SC (N = 109)	EC (N = 114)		SC (N = 109)	EC (N = 114)		SC (N = 109)	EC (N = 114)
Cushingoid facies	72 (66)	77 (68)	1.04 (0.87 to 1.25)	75 (69)	79 (69)	1.03 (0.87 to 1.22)	76 (70)	81 (71)	1.05 (0.89 to 1.23)	78 (72)	83 (73)	1.02 (0.88 to 1.19)
Striae	3 (3)	8 (7)	2.69 (0.73 to 9.87)	4 (4)	11 (10)	2.72 (0.90 to 8.27)	6 (6)	11 (10)	1.80 (0.69 to 4.67)	7 (6)	14 (12)	1.92 (0.81 to 4.54)
Hypertrichosis	25 (23)	34 (30)	1.35 (0.87 to 2.09)	30 (28)	40 (35)	1.28 (0.87 to 1.89)	37 (34)	42 (37)	1.12 (0.80 to 1.59)	41 (38)	45 (39)	1.05 (0.77 to 1.45)
Acne	3 (3)	6 (5)	2.02 (0.52 to 7.86)	6 (6)	9 (8)	1.49 (0.55 to 4.02)	7 (6)	11 (10)	1.52 (0.62 to 3.77)	7 (6)	12 (11)	1.64 (0.68 to 3.99)
Increased appetite	95 (87)	98 (86)	0.99 (0.89 to 1.09)	98 (90)	100 (88)	0.98 (0.90 to 1.07)	102 (94)	104 (91)	0.99 (0.92 to 1.07)	103 (94)	106 (93)	1.00 (0.94 to 1.07)
Poor behaviour	98 (90)	86 (75)	0.85 (0.76 to 0.96)	99 (91)	92 (81)	0.90 (0.82 to 1.00)	100 (92)	93 (82)	0.90 (0.82 to 0.98)	101 (93)	94 (82)	0.90 (0.82 to 0.98)
Glycosuria	10 (9)	9 (8)	0.92 (0.39 to 2.16)	11 (10)	13 (11)	1.24 (0.59 to 2.61)	12 (11)	17 (15)	1.49 (0.76 to 2.91)	14 (13)	19 (17)	1.34 (0.72 to 2.48)
Cataracta	–	–	–	–	–	–	1 (1)	0 (0)	–	1 (1)	1 (1)	0.96 (0.06 to 15.00)
Abdominal pain	31 (28)	28 (25)	0.92 (0.60 to 1.42)	35 (32)	38 (33)	1.08 (0.74 to 1.56)	46 (42)	44 (39)	0.95 (0.70 to 1.29)	51 (47)	49 (43)	0.91 (0.69 to 1.20)

Achenbach Child Behaviour Checklist

Quantitative data on behaviour were collected using the ACBC. There were no significant differences in the ACBC t-scores (p = 0.9) or total scores (p = 0.3) (Figure 7a and b; see also Table 21 in Appendix 2). There were also no differences in the proportion of participants reporting normal ACBC scores (see Table 22 in Appendix 2).

FIGURE 7.

ACBC scores. (a) t-score; and (b) total score. Higher scores indicate more abnormal behaviour.

Other outcomes

Growth data

After the 16-week study medication period, the mean height z-score (Figure 8a) and percentile (see Figure 8b) scores gradually increased over time. At 12, 18 and 24 months, the mean height z-scores were –0.06 (SD 1.11), –0.06 (SD 1.13) and 0.12 (SD 1.09), respectively, in the SC group and –0.02 (SD 0.95), 0.05 (SD 0.93) and 0.02 (SD 0.99), respectively, in the EC group.

FIGURE 8.

Height. (a) z-score; and (b) percentile. a, Randomisation counted as week 0.

The mean weight z-scores at 8 weeks (when the corticosteroid dose between the two treatment arms are the most different, i.e. after 4 weeks of study medication) were 0.64 (SD 1.09) in the SC group and 0.98 (SD 1.03) in the EC group. At 12, 18 and 24 months, the mean weight z-scores were 0.64 (SD 1.04), 0.61 (SD 1.19) and 0.70 (SD 1.19), respectively, in the SC group and 0.85 (SD 1.09), 0.80 (SD 1.06) and 0.76 (SD 1.12), respectively, in the EC group (Figure 9a). The mean weight percentile data are shown in Figure 9b.

FIGURE 9.

Weight. (a) z-score; and (b) percentile. a, Randomisation counted as week 0.

The mean BMI z-scores at 8 weeks (when the corticosteroid dose between the two treatment arms are the most different, i.e. after 4 weeks of study medication) were 1.24 (SD 1.27) in the SC group and 1.53 (SD 1.07) in the EC group. At 12, 18 and 24 months, the mean BMI z-scores were 0.82 (SD 1.21), 0.78 (SD 1.31) and 0.69 (SD 1.39), respectively, in the SC group and 1.20 (SD 1.22), 1.11 (SD 1.10) and 0.99 (SD 1.20), respectively, in the EC group (Figure 10a). The mean BMI percentile data are shown in Figure 10b.

FIGURE 10.

BMI. (a) z-score; and (b) percentile. a, Randomisation counted as week 0.

Blood pressure

The mean systolic blood pressure z-scores at 4 weeks (at the end of the open-label prednisolone period) were 1.25 (SD 0.86) in the SC group and 1.30 (SD 0.92) in the EC group. At 12, 18 and 24 months, the mean systolic blood pressure z-scores were 0.75 (SD 0.96), 0.67 (SD 1.03) and 0.70 (SD 0.97), respectively, in the SC group and 0.67 (SD 1.06), 0.53 (SD 0.96) and 0.53 (SD 0.89), respectively, in the EC group (Figure 11a). The mean systolic blood pressure percentile data are shown in Figure 11b.

FIGURE 11.

Systolic blood pressure. (a) z-score; and (b) percentile.

The mean diastolic blood pressure z-scores at 4 weeks (at the end of the open-label prednisolone period) were 1.12 (SD 0.93) in the SC group and 1.12 (SD 0.91) in the EC group. At 12, 18 and 24 months, the mean systolic blood pressure z-scores were 0.61 (SD 0.99), 0.48 (SD 0.93) and 0.58 (SD 1.02), respectively, in the SC group and 0.57 (SD 0.96), 0.37 (SD 1.03) and 0.42 (SD 0.83), respectively, in the EC group (Figure 12a). The mean diastolic blood pressure percentile data are shown in Figure 12b.

FIGURE 12.

Diastolic blood pressure. (a) z-score; and (b) percentile.

Chapter 4 Pilot study

Prior to submission of the application for NIHR funding for the PREDNOS study, a pilot study was conducted to test the proposed study methodology. At the time, the large majority of UK district general hospital paediatric departments had little or no experience in the conduct of RCTs involving an investigational medicinal product, and investigators were keen to ascertain that patients could be successfully recruited and followed in accordance with the study protocol. This was of great importance given that the majority of patients with nephrotic syndrome present to district general hospitals rather than to tertiary paediatric nephrology centres. This pilot study, which was funded jointly by Kidney Research UK and Kids Kidney Research, used an identical methodology to that of the proposed main study, randomising participants to either an 8-week SC or a 16-week EC of prednisolone therapy, and aimed to:

• provide ‘proof of principle’ of successful recruitment of participants and collaboration in district general hospitals

• develop a network of investigators, initially in the North West of the UK, although then extending beyond this region

• provide information on recruitment rates

• provide further evidence on the incidence of trial outcomes that could be used to inform the definitive trial design; these outcomes included sustained remission rate at 6 and 12 months, time to relapse, incidence of FRNS, incidence of SAEs and incidence of need for other immunosuppressive medications.

This pilot trial was carried out under a Clinical Trial Authorisation carried over from a Doctors and Dentist Exemption (reference number MF8000/13293), in accordance with the Medicines for Human Use Clinical Trials regulations. 43 Ethics approval for the study was granted by the Thames Valley Multi-centre Research Ethics Committee (reference number 04/12/025). The trial was sponsored by Great Ormond Street Hospital for Children NHS Foundation Trust (reference number 03/NU/13). Both study drug and placebo were manufactured by Essential Nutrition Ltd. The study was registered on ClinicalTrials.gov as reference number NCT00308321 and EudraCT as reference number 2004-001813-33.

The pilot study recruited its first participant in August 2006. Trial recruitment was assisted significantly by the development of the NIHR Medicines for Children Research Network (MCRN), which formally adopted the study into its portfolio in March 2007. A successful collaborative trial network was established, with principal investigators appointed at a total of 37 sites. By study completion, 26 sites were fully set up, and, by June 2008, 18 sites had recruited 55 participants. A further 13 sites had expressed active interest in participation in the study.

Of the 55 participants recruited, one was recruited and randomised in error before site set-up and never received the study drug or entered the trial and two proved resistant to corticosteroid therapy after providing informed consent, although prior to starting trial medication. These three participants were excluded from the trial and any analyses. Two experienced difficulties in taking solid tablets and withdrew from the study and two participants changed their area of residence and were lost to follow-up. Therefore, the ITT cohort consisted of 52 participants. The mean age was 6.1 (SD 3.0) years; 31 (60%) were male, 38 (73%) were white and 10 (19%) were South Asian. The median BMI percentile was 77.8 and the mean dose of prednisolone administered in the open-label phase was 60.3 mg/m²/day (Table 7).

There were 35 relapses in 52 participants. There were no suspected unexpected serious adverse reactions, although three participants experienced a SAE (trapped finger stitched in theatre, hospital admission for abdominal pain and hospital admission for viral wheeze).

A decision was made to not unblind the pilot data prior to the commencement of the main study, but to allow these data to be added to the results of the main study using meta-analysis methods. One minor modification was made to the protocol for the main study following the pilot study, in that the visit schedule between months 6 and 12 was reduced from monthly to 2-monthly; monthly visits were felt to be too onerous for participants and in excess of the visit schedule for routine clinical care for the large majority of patients with SSNS at this stage post presentation.

The data from the 223 participants in the PREDNOS study and the 52 participants in the PREDNOS pilot study were combined using random-effects meta-analysis methods. In the pilot study, date of relapse was not recorded, so the date that relapse treatment was commenced was used for the relapse date in the time to first relapse analysis here. This meta-analysis included 275 participants (136 in SC group and 139 in EC group) and showed no significant difference in time to first relapse between the two treatment groups (pooled HR 0.76, 95% CI 0.50 to 1.17; p = 0.21) (Figure 13).

FIGURE 13.

Meta-analysis of time to first relapse for the PREDNOS and PREDNOS pilot study. df, degrees of freedom; IV, inverse variance.

Chapter 5 Economic analysis: the mapping exercise

The economic analysis is organised into two chapters. Chapter 5 provides detail on the mapping exercise that was conducted to inform the outcomes for the main economic evaluation. Chapter 6 describes the economic evaluation to estimate the cost-effectiveness of extending prednisolone therapy over a 16-week period compared with the standard 8-week regimen for treating children with SSNS.

Results

Sample characteristics

There were 643 observations across the five data collection time points from participants who were aged ≥ 5 years. These observations were randomised into groups A (n = 321) and B (n = 322). The longitudinal nature of the study meant that the number of missing data in the groups varied across the data collection points. After removing missing items, 279 observations with pairs of valid PedsQL and CHU-9D index scores in the first group formed the estimation sample, while the 284 observations in the second group formed the validation sample. The estimation and validation samples constitute the total mapping sample (n = 563). Figure 14 shows the distribution of the CHU-9D and PedsQL scores in the estimation and validation samples.

FIGURE 14.

Distribution of CHU-9D and PedsQL scores in the estimation and validation samples. (a) Estimation sample, CHU-9D; (b) validation sample, CHU-9D; (c) estimation sample, PedsQL; and (d) validation sample, PedsQL.

Table 8 reports the descriptive statistics for each sample at each time point. Overall, it seems that the randomisation ensured a balanced distribution of demographic characteristics between the estimation and the validation samples.

TABLE 8 - Demographic characteristics of estimation and validation sample by data collection time point

Characteristic	Time point
	Baseline (4 weeks)	16 weeks	12 months	24 months	36 months	48 months
Estimation sample	N = 55	N = 47	N = 58	N = 54	N = 39	N = 26
Age (year)
Mean (SD)	7 (2.1)	7.6 (2.1)	7.4 (2.1)	7.2 (1.9)	7.3 (1.9)	8.1 (2.0)
Median (IQR)	6 (3)	7 (4)	7 (3)	7 (2)	7 (3)	8 (3)
Range	5–12	5–12	5–12	5–12	5–12	5–12
Gender
Male, n (%)	35 (63.6)	33 (70.2)	36 (62.1)	32 (59.2)	23 (58.9)	18 (69.2)
CHU-9D score
Mean (SD)	0.940 (0.063)	0.929 (0.103)	0.941 (0.080)	0.950 (0.068)	0.922 (0.081)	0.937 (0.077)
Median (IQR)	0.952 (0.106)	0.952 (0.100)	0.952 (0.081)	0.968 (0.073)	0.931 (0.108)	0.967 (0.107)
Range	0.786–1.000	0.534–1.000	0.509–1.000	0.68–1.000	0.702–1.000	0.697–1.000
PedsQL score
Mean (SD)	77.11 (16.16)	82.4 (16.8)	81.94 (15.91)	84.24 (14.31)	78.49 (20.58)	80.85 (17.72)
Median (IQR)	79.35 (28.26)	89.13 (29.35)	87.5 (20.65)	88.04 (18.48)	82.61 (30.43)	82.61 (29.35)
Range	40.22–100.00	45.65–100.00	41.3–100.00	43.48–100.00	31.52–100.00	39.13–100.00
Validation sample	N = 36	N = 46	N = 50	N = 70	N = 56	N = 26
Age
Mean (SD)	6.9 (1.8)	7.1 (1.9)	7.3 (2.0)	7.6 (2.2)	7.4 (2.2)	8 (1.9)
Median (IQR)	7 (3)	7 (2)	7 (3)	7 (3)	7 (3)	8 (2)
Range	5–11	5–12	5–12	5–12	5–13	5–13
Gender
Male, n (%)	25 (69.4)	30 (65.2)	32 (64.0)	44 (62.9)	29 (51.8)	15 (57.7)
CHU-9D score
Mean (SD)	0.924 (0.081)	0.945 (0.067)	0.941 (0.075)	0.938 (0.076)	0.951 (0.067)	0.945 (0.06)
Median (IQR)	0.952 (0.1)	0.96 (0.079)	0.96 (0.081)	0.952 (0.102)	0.967 (0.071)	0.959 (0.097)
Range	0.711–1	0.69–1	0.739–1	0.65–1	0.712–1	0.828–1
PedsQL score
Mean (SD)	75.88 (16.91)	81.35 (14.53)	78.28 (19.01)	80.6 (17.25)	83.13 (19.11)	81.68 (20.3)
Median (IQR)	77.72 (27.36)	83.7 (18.48)	83.7 (28.26)	86.96 (27.17)	91.85 (27.17)	90.76 (20.65)
Range	42.39–97.83	41.3–100	21.74–100	33.7–100	40.22–100	29.35–100

The mean CHU-9D utility score across all time points was 0.9374 (SD 0.0790) and 0.9409 (SD 0.0717) in the estimation and validation samples, respectively. The corresponding mean PedsQL score across all time points was 80.93 (SD 16.76) in the estimation sample and 80.31 (SD 17.79) in the validation sample. Within each sample, the mean PedsQL total score was lower than the mean CHU-9D utility score when both scores were standardised on a 100-point scale. Although both HRQoL measures were negatively skewed, the ceiling effect was more prominent with CHU-9D. Tables 23 and 24 in Appendix 3 summarise the CHU-9D responses for the estimation and validation samples across all data collection time points. Level 1 or ‘no problem’ always had the highest proportion of responses, hence the observed ceiling effect for the CHU-9D index score.

Performance and validation

Table 25 in Appendix 3 summarises the performance measures for all the model specifications, for both the estimation and the validation samples. Within the estimation sample, the models were able to reasonably predict the mean CHU-9D value (0.93742, SD 0.07898). Of the 18 models, 12 were able to predict the precise mean value by up to 1/10,000th of a QALY. The exceptions were the six tobit models. However, within the validation sample, the models were less able to predict the mean CHU-9D value (0.94094, SD 0.07174). The model GLM 2 had lowest mean predicted value (0.93409), whereas the model Tobit_3 had the highest mean predicted value (0.96575), giving a difference between the observed and predicted mean values of 0.0069 and 0.0245, respectively. These differences were below the threshold of 0.03, for which differences smaller than this level are considered to be a minimally important difference. 70,71 A further observation was that some OLS models and all the tobit models had maximum predicted values beyond the upper limit of the CHU-9D utility scale (0.33–1.00). However, none of the models predicted a utility value below the lower limit of the CHU-9D utility scale.

Models were initially assessed in terms of their ability to predict the mean value in the estimation sample. All GLM and OLS models were consequently shortlisted for further comparison and progress to ‘step 2’. The two models (GLM 6 and OLS 3) that had the ‘best’ performance in terms of MAE, in both the estimation and validation samples, were selected for a final comparison: ‘step 3’. Table 9 contains the model performance results for both of these models.

TABLE 9 - Model performance of the two best-fitting models

< 0.03 AE (%), percentage with absolute error below 0.03; < 0.05 AE (%), percentage with absolute error below 0.05; < 0.10 AE (%), percentage with absolute error below 0.10; CV, coefficient of variation; P₂₅, 25th percentile; P₅₀, 50th percentile; P₇₅, 75th percentile.

Statistics	Sample
	Estimation			Validation
	Observed	GLM 6	OLS 3	Observed	GLM 6	OLS 3
Mean	0.937419	0.937419	0.937419	0.940941	0.937612	0.939018
SD	0.078978	0.051926	0.047318	0.071737	0.054762	0.046323
CV	0.084251	0.055393	0.050477	0.076240	0.058406	0.049331
Minimum value	0.509400	0.660930	0.812068	0.650000	0.705160	0.788717
P 25	0.907600	0.910639	0.900076	0.912300	0.914908	0.905183
P 50	0.952100	0.957229	0.946303	0.952100	0.958496	0.950063
P 75	1.000000	0.978902	0.980433	1.000000	0.977276	0.977413
Maximum value	1.000000	0.989350	0.995221	1.000000	0.985504	0.993891
MSE	–	0.00353	0.00398	–	0.00345	0.00310
MAE	–	0.04078	0.04245	–	0.04182	0.03981
< 0.03 AE (%)	–	53.40	51.61	–	55.89	53.17
< 0.05 AE (%)	–	72.04	70.25	–	73.23	70.77
< 0.10 AE (%)	–	92.27	90.32	–	91.20	93.31

For the GLM, a logit transformation of the variable containing the CHU-9D utility scores was applied before the variable was used as the dependent in the prediction equation. As such, any predicted value from that equation will be a transformed value and, therefore, requires a back-transformation to estimate utility values. The information on the back-transformation step is as follows. Given that GLM 6 has a logit link, the CHU-9D utility values are calculated as shown below:

⁽²⁾ CHU-9D utility score [ GLM ] = e CHU- 9 D utility value 1 + e CHU- 9 D utility value

For the final models in step 3, in addition to assessing how accurately the models estimated the mean CHU-9D score in the validation sample, the distribution of the predicted score was also examined (see Figure 17, Appendix 3). GLM 6 had a wide range of predicted CHU-9D scores compared with OLS 3.

Approximately 56% of the predicted values from GLM model 6 in the validation sample had absolute errors less than the minimally important difference value of 0.03; the corresponding values for the OLS model 3 was 53%. GLM model 6 remained the preferred model specification when the error threshold was extended to 0.05.

Although the prediction accuracy of the mean scores was similar in both models, the accuracy level was not uniform across the CHU-9D utility range, as shown in Table 10. The number of observations with utility score of < 0.7 was small; therefore, the comparison between the best two models was restricted to observations with higher utility values. GLM 6 was superior to OLS 3 in the estimation sample; however, in the validation sample there were diverging results. OLS 3 had a better prediction accuracy when utility values were > 0.8, but less than full health, while the GLM 6 was superior at predicting full health and utility values between 0.7 and 0.8. So, although OLS 3 had a better prediction accuracy in the validation sample overall, it was found to be only marginally better than GLM 6.

TABLE 10 - Distribution of errors by observed CHU-9D range

CHU-9D range	n	Model
		GLM 6		OLS 3
		MSE	MAE	MSE	MAE
Estimation
0.5 ≤ value < 0.6	3	0.09443	0.30095	0.11168	0.33058
0.6 ≤ value < 0.7	6	0.01873	0.12096	0.02823	0.16497
0.7 ≤ value < 0.8	11	0.00726	0.07674	0.00988	0.09602
0.8 ≤ value < 0.9	49	0.00301	0.04441	0.00259	0.04034
0.9 ≤ value < 0.10	111	0.00154	0.02853	0.00155	0.03015
Full health	102	0.00242	0.03847	0.00279	0.03899
Validation
0.6 ≤ value < 0.7	3	0.05502	0.23049	0.05277	0.22693
0.7 ≤ value < 0.8	12	0.01185	0.09583	0.01376	0.10958
0.8 ≤ value < 0.9	47	0.00468	0.05691	0.00329	0.04316
0.9 ≤ value < 0.10	115	0.00187	0.03057	0.00131	0.02942
Full health	107	0.00223	0.03593	0.00237	0.03643

In summary, relative to GLM 6, OLS 3 lacked the ability to predict the wider range of CHU-9D values (0.7–1), and a higher proportion of its predicted values had absolute errors above the minimally important difference. Furthermore, it was less able to predict full health, which is particularly important for utility data that tend to have ceiling effects. Taking all these factors into account, the GLM 6 model was selected as the preferred model for mapping from PedsQL to CHU-9D. Table 11 shows the coefficients for generating deterministic and probabilistic utility predictions using the GLM 6 model. Coefficients for OLS 3 have also been presented in situations in which this might be desired.

TABLE 11 - Coefficients for the two best-fitting models

*p < 0.05, **p < 0.01, ***p < 0.1.

EF, emotional functioning; FU, school functioning; PF, physical functioning; SE, standard error; SF, social functioning.

Covariate	Model
	GLM 6		OLS 3
	Coefficient	SE	Coefficient	SE
PedsQL PF squared	0.0001615	0.000103	–	–
PedsQL EF squared	0.0004766***	0.000127	–	–
PedsQL SF squared	–0.0000402	0.000145	–	–
PedsQL FU squared	–0.0001646	0.000101	–	–
PedsQL PF × EF	–0.0001103	0.000147	–	–
PedsQL PF × SF	–0.000114	0.000173	–	–
PedsQL PF × FU	0.0000371	0.000143	–	–
PedsQL EF × SF	–0.0002461	0.000209	–	–
PedsQL EF × FU	–0.0001158	0.000167	–	–
PedsQL SF × FU	0.0004356***	0.00013	–	–
PedsQL PF	–	–	0.0007133*	0.000297
PedsQL EF	–	–	0.0016477***	0.000228
PedsQL SF	–	–	–0.00011	0.000383
PedsQL FU	–	–	0.000261	0.000276
Age	0.0279345	0.039717	–	–
Sex	–0.0546336	0.146341	–	–
Constant	0.7135215	0.399623	0.7422337***	0.028841

Chapter 6 Economic evaluation

Chapter 6 describes the economic evaluation alongside the PREDNOS trial. The overall aim was to estimate the cost-effectiveness of an extended prednisolone regimen over a 16-week period compared with a standard 8-week regimen for treating children with SSNS. The primary evaluation was a cost–utility analysis, with the outcome measured in terms of QALYs. All costs were expressed in 2015 prices.

Missing data were addressed using multiple imputation techniques. Uncertainty surrounding the cost-effectiveness estimates was examined using probabilistic sensitivity analysis75 and presented using cost-effectiveness acceptability curves (CEACs). The analysis of cost-effectiveness was conducted according to current best practice methods for conducting economic evaluations alongside clinical trials. 76

Results

Impact of extended prednisolone therapy on health-related quality of life

A total of 207 participants had been followed up to the 24-month primary end point, which was 92.8% of the ITT study population. The proportion of the study population that returned the study questionnaires at each time point gradually declined over the 24 months but did not fall below 90%, as shown in Table 13. Complete items on the entire eight CHU-9D domains were required for the calculation of the CHU-9D index score in older participants. The four PedsQL domain scores were also required for indirect estimation of CHU-9D score for younger participants. After excluding missing items, 90% of the participants within the study had a valid CHU-9D score at baseline. This proportion declined to 85% at month 24. One hundred and sixty-eight (75.3%) participants had a valid CHU-9D index score at baseline, at week 16 and at months 12 and 24.

TABLE 13 - Number of follow-ups, returned questionnaires and valid CHU-9D utility scores by data collection time point for health-related quality-of-life instruments (PedsQL and CHU-9D)

Time point	Followed up, n (%)	Followed up and returned questionnaire, n (%)	Valid CHU-9D index score calculated, n (%)	No valid CHU-9D index score (as % of participants in treatment group), n (%)
				SC (N = 109)	EC (N = 114)
Week 4	223 (100.0)	222 (99.6)	201 (90.1)	13 (11.9)	9 (7.9)
Week 16	215 (96.4)	211 (94.6)	199 (89.2)	12 (11.0)	12 (10.5)
Month 12	209 (93.7)	207 (92.8)	202 (90.6)	9 (8.3)	12 (10.5)
Month 24	207 (92.8)	203 (91.0)	190 (85.2)	13 (11.9)	20 (17.5)
Complete case	–	–	168 (75.3)	30 (27.5)	25 (21.9)

Table 14 describes the mean utility value at each time point for each treatment group. It further classifies the data into utility scores for participants aged ≥ 5 years (collected using the CHU-9D questionnaire), and for all the participants in the study with a valid CHU-9D utility score – both mapped and direct CHU-9D utility scores. The mean utility at week 4 (baseline) for all participants across all age groups was 0.9242. At baseline, the mean utility for the SC group was 0.9231 compared with 0.9252 for the EC group. This baseline difference was adjusted for within the main cost–utility analyses. 78

TABLE 14 - Mean CHU-9D index score per participant follow-up time-point by treatment group

Time point	Group				Whole sample
	SC		EC
	n	Mean score (SD)	n	Mean score (SD)	n	Mean CHU-9D score (SD)
Direct CHU-9D estimation
Baseline (week 4)	42	0.9289 (0.0752)	58	0.9301 (0.0686)	101	0.9296 (0.0710)
Week 16	47	0.9394 (0.0665)	54	0.9353 (0.0984)	101	0.9372 (0.0846)
Month 12	56	0.9442 (0.0703)	63	0.9396 (0.0826)	119	0.9418 (0.0768)
Month 24	69	0.9341 (0.0784)	70	0.9548 (0.0666)	139	0.9445 (0.0732)
Direct and indirect CHU-9D estimation
Baseline	96	0.9231 (0.0675)	105	0.9252 (0.0656)	201	0.9242 (0.0664)
Week 16	97	0.9333 (0.0567)	102	0.9391 (0.0760)	199	0.9363 (0.0672)
Month 12	100	0.9383 (0.0639)	102	0.9425 (0.0688)	202	0.9405 (0.0663)
Month 24	96	0.9316 (0.0708)	94	0.9492 (0.0624)	190	0.9403 (0.0672)

Table 15 describes the unadjusted and the adjusted mean QALYs for each treatment group of the study. Participants in the study were required to have been followed up for at least 24 months. At month 24, the mean adjusted QALY difference between the groups was 0.0162 in favour of EC therapy; however, this difference was not significant at the 5% level. The lack of significance was attributable to the way the QALYs were calculated using the ‘area-under-the-curve’ method, which adjusts for imbalances in the CHU-9D score at baseline (week 4).

TABLE 15 - Mean CHU-9D score at each follow-up time point and mean QALY per participant over 24 months by treatment group

Adjusted values have controlled for age, gender and baseline utility score.

Time point	Group		Bootstrapped difference (95% CI)
	SC	EC
Unimputed CHU-9D score, mean (SD)
Baseline	0.9231 (0.0675)	0.9252 (0.0656)	0.0020 (–0.0162 to 0.0203)
Week 16	0.9333 (0.0567)	0.9391 (0.0760)	0.0058 (–0.0130 to 0.0246)
Month 12	0.9383 (0.0639)	0.9425 (0.0688)	0.0042 (–0.0141 to 0.0225)
Month 24	0.9316 (0.0708)	0.9492 (0.0624)	0.0176 (–0.0016 to 0.0368)
Unadjusted QALY over 24 months, mean (SE)	1.7940 (0.0105)	1.8046 (0.0105)	0.0107 (–0.0183 to 0.0396)
Adjusted QALY over 24 months, mean (SE)	1.7901 (0.0094)	1.8027 (0.0095)	0.0125 (–0.0137 to 0.0388)
Imputed CHU-9D score, mean (SD)
Baseline	0.9228 (0.0640)	0.9258 (0.0632)	0.0031 (–0.0139 to 0.0200)
Week 16	0.9315 (0.0555)	0.9404 (0.0725)	0.0089 (–0.0083 to 0.0261)
Month 12	0.9393 (0.0614)	0.9429 (0.0652)	0.0036 (–0.0135 to 0.0208)
Month 24	0.9306 (0.0672)	0.9488 (0.0594)	0.0181 (0.0013 to 0.0349)
Unadjusted QALY over 24 months, mean (SE)	1.7906 (0.0085)	1.8072 (0.0083)	0.0166 (–0.0067 to 0.0398)
Adjusted QALY over 24 months, mean (SE)	1.7908 (0.0076)	1.8070 (0.0074)	0.0162 (–0.0047 to 0.0372)

Resource use and cost

Participants in the EC group had fewer hospital admissions than participants in the SC group, and admissions were of shorter duration (Table 16). The average length of stay in the SC and EC groups over 24 months was 5.8 days and 2.5 days, respectively, which equates to a mean per-patient cost of £1539 and £691, respectively (Table 17). A breakdown of length of stay and cost from baseline to month 24 is shown in Tables 26 and 27 in Appendix 4. In addition to having fewer hospital admissions, participants in the EC group were also reported to have fewer hospital emergency visits than participants in the SC group (see Table 16). This led to a per-patient mean cost saving of £411 over the 24 months (see Table 17). In general, over the 24 months, participants in the EC group had fewer hospital outpatient visits than those in the SC group, and this led to a per-patient mean cost saving of £382 (see Table 17). Tables 16 and 17 also outline the disaggregated mean (SD) number of primary care visits and the corresponding mean per-patient cost by treatment group. Tables 28 and 29 in Appendix 4 show the hospital emergency visits and hospital outpatient visits at each follow-up time point until month 24. Overall, participants in the EC group had fewer primary care visits than participants in the SC group.

TABLE 16 - Mean resource use per participant over 24 months

Resource use	Group		Bootstrapped difference (95% CI)
	SC (n = 92)	EC (n = 94)
Primary care, n, mean (SD)
GP visits	0.576 (1.030)	0.383 (0.893)	–0.193 (–0.474 to 0.088)
Practice nurse visits	1.554 (6.716)	0.564 (1.535)	–0.991 (–2.369 to 0.388)
Other primary care staff visit	0.065 (0.440)	0.011 (0.103)	–0.055 (–0.148 to 0.039)
Secondary care, n, mean (SD)
Hospital admission bed-days	5.826 (28.682)	2.526 (5.307)	–3.299 (–8.742 to 2.144)
Hospital emergency visits	0.630 (1.155)	0.319 (0.832)	–0.311 (–0.594 to –0.029)
Hospital outpatient visits	7.120 (11.607)	5.383 (5.187)	–1.737 (–4.243 to 0.770)
Medication, mean (SD)
Ciclosporin (number of days)	8.685 (51.922)	0.638 (6.189)	–8.046 (–19.088 to 2.995)
Cyclophosphamide (number of days)	14.380 (30.161)	9.883 (23.684)	–4.497 (–12.972 to 3.977)
Levamisole (number of days)	11.935 (44.831)	13.234 (46.566)	1.299 (–10.991 to 13.589)
Mycophenolate mofetil (number of days)	6.696 (31.042)	1.011 (9.799)	–5.685 (–12.541 to 1.171)
Rituximab (number of days)	0.272 (2.204)	0.000 (0.000)	–0.272 (–0.761 to 0.218)
Tacrolimus (number of days)	0.500 (2.956)	7.670 (46.942)	7.170 (–2.617 to 16.957)
Prednisolone (mg consumed)	3851.83 (2918.22)	4274.81 (3311.25)	422.988 (–499.172 to 1345.148)

Reported figures are for participants who returned study questionnaire at all time points over the 24-month period.

TABLE 17 - Mean per participant cost over 24 months by treatment group

Cost component	Group, mean cost (£) (SD)		Bootstrapped difference (95% CI)
	SC (n = 92)	EC (n = 94)
Primary care visits
GP	28.24 (50.47)	18.77 (43.79)	–9.47 (–23.00 to 4.07)
Practice nurse	14.46 (62.46)	5.24 (14.28)	–9.21 (–21.99 to 3.57)
Other primary care staff	0.49 (3.30)	0.08 (0.77)	–0.41 (–1.11 to 0.30)
Secondary care
Hospital admission	1539.36 (7603.47)	691.13 (1308.60)	–848.23 (–2253.65 to 557.20)
Hospital emergency	832.80 (1525.99)	421.60 (1099.49)	–411.21 (–804.67 to –17.75)
Hospital outpatient	1566.30 (2553.53)	1184.26 (1141.15)	–382.05 (–946.98 to 182.88)
Medications
Ciclosporin	14.76 (88.27)	1.09 (10.52)	–13.68 (–32.21 to 4.85)
Cyclophosphamide	19.99 (41.92)	13.74 (32.92)	–6.25 (–17.43 to 4.93)
Levamisole	4.44 (16.68)	4.92 (17.32)	0.48 (–4.01 to 5.06)
Mycophenolate mofetil	2.49 (11.56)	0.38 (3.64)	–2.11 (–4.42 to 0.191)
Rituximab	31.64 (256.64)	0.00 (0.00)	–31.64 (–82.07 to 18.80)
Tacrolimus	4.09 (24.19)	62.77 (384.17)	58.68 (–19.51 to 136.88)
Prednisolone	44.30 (33.56)	49.16 (38.08)	4.86 (–5.04 to 14.77)

Extending the course of prednisolone over a longer duration meant that all participants in the EC group incurred a higher prednisolone drug cost over 24 months. The bootstrapped cost difference in Table 17 shows that mean cost difference (mean £4.86; 95% CI –£5.04 to £14.77) between the treatment groups was not significant. Ciclosporin, cyclophosphamide, levamisole, mycophenolate mofetil, rituximab and tacrolimus were the second-line immunosuppressants used in the treatment pathway. Table 16 shows that participants in the EC group were given levamisole and tacrolimus for longer than participants in the SC group and, overall, across the 24-month follow-up period, the per-patient mean cost difference attributed to just second-line immunosuppressant therapy was £11 lower in the SC group than in the EC group. The mean cost for other prescribed medications was £50 lower in the EC group.

The per-patient total mean cost was made up of costs associated with prednisolone prescriptions, hospital admissions, hospital outpatient visits, emergency visits, primary care visits, second-line immunosuppressants and other prescribed medications. To account for missing data, multiple imputation (5000 iterations) was applied at each follow-up time point. To estimate the difference in the per-patient total mean cost between the two treatment groups, adjustments were made for age, gender and the baseline (week 4) utility score. Over the 24-month study period, participants in the EC group incurred less cost than those in the SC group, by an average of £1673 per patient (Table 18).

TABLE 18 - Mean incremental cost per participant over 24 months

Values are means unless stated otherwise. Adjusted values have controlled for age and gender.

Imputed cost	Group, mean cost (£) (SD)		Bootstrapped cost difference (95% CI)
	SC (n = 109)	EC (n = 114)
Baseline	45.54 (121.89)	78.94 (202.20)	35.24 (–7.57 to 78.05)
Week 8	132.10 (391.97)	116.56 (458.72)	–21.53 (–137.43 to 94.38)
Week 12	401.43 (1285.11)	56.54 (189.02)	–361.14 (–610.11 to –112.16)
Week 16	235.45 (410.97)	73.74 (217.51)	–162.19 (–247.61 to –76.77)
Month 5	301.15 (700.77)	265.89 (526.68)	–38 (–204.59 to 128.59)
Month 6	193.07 (435.69)	255.04 (648.82)	63.86 (–91.67 to 219.38)
Month 8	398.85 (624.72)	255.98 (522.45)	–139.29 (–289.89 to 11.32)
Month 10	887.64 (5827.01)	287.69 (519.73)	–653.39 (–1758.95 to 452.17)
Month 12	239.91 (434.46)	258.11 (500.91)	32.65 (–91.40 to 156.70)
Month 18	791.95 (1558.03)	566.72 (880.52)	–225.15 (–569.61 to 119.32)
Month 24	666.18 (2103.73)	466.64 (803.76)	–235.16 (–649.90 to 179.59)
Total imputed cost over 24 months
Unadjusted, mean (SE)	4462.21 (748.13)	2607.51 (731.46)	–1854.70 (–3959.44 to 250.03)
Adjusted, mean (SE)	4369.20 (748.13)	2696.43 (731.37)	–1672.77 (–3455.06 to 109.53)

Cost–utility analysis

The incremental difference in costs was then offset against the difference in QALYs for the SC group versus the EC group. The EC group dominated the SC group, which means that, on average, participants incurred less cost and gained more QALYs. Cost–utility analysis combines the incremental costs with the incremental QALYs to produce an ICER, or a cost per additional QALY gained. In this situation, as costs were lower and QALYs gained higher, the mean ICER values were not calculated. The EC group was both more effective and cheaper (Table 19).

TABLE 19 - Incremental cost and QALY per participant over 24 months

Imputed values	Group		Bootstrapped difference (95% CI)
	SC (n = 109)	EC (n = 114)
Mean QALYs (SE)	1.7908 (0.0076)	1.8070 (0.0074)	0.0162 (–0.0047 to 0.0372)
Mean cost, £ (SE)	4369.20 (748.13)	2696.43 (731.37)	–1672.77 (–3455.06 to 109.53)

Values are mean (SE) unless stated otherwise. QALYs have been adjusted for age, gender and baseline utility value. Costs have been adjusted for age and gender. All costs are in Great British pounds (£).

Sensitivity analysis

To account for the uncertainty around the point estimates, an incremental cost-effectiveness plane and a CEAC were constructed using the net monetary benefit approach. The plane (Figure 15) shows 5000 jointly bootstrapped cost–QALY difference pairs scattered across all four quadrants.

FIGURE 15.

Cost–utility plane for the comparison of extended prednisolone therapy with standard prednisolone therapy, based on 5000 bootstrapped cost–effect pairs.

The majority of the points are in the south-east quadrant, indicating that the extended prednisolone course was less expensive and more effective than the standard prednisolone course. The dashed line represents the willingness-to-pay threshold of £20,000 per QALY and the solid line represents the willingness-to-pay threshold of £30,000 per QALY. Points to the right of both lines indicate situations when the extended prednisolone course is the preferred option, while points to the left indicate instances in which the SC is recommended. Figure 16 shows that the probability of extended prednisolone course being cost-effective is 98.8% and 99.9% at £20,000 and £30,000 per QALY, respectively.

FIGURE 16.

Cost-utility acceptability curve comparing extended and standard prednisolone therapy, based on 5000 bootstrapped cost–effect pairs.

The EC group remained dominant at 24 months when the analysis was restricted to participants with complete cost and outcome data (Table 20). The main difference from the base-case analysis was the smaller incremental cost at 24 months (£582 vs £1673 cost saving). Overall, the probability of EC being cost-effective compared with SC, remained high, at 89%.

TABLE 20 - Incremental cost and QALYs by age category, ethnicity, and completeness of data

Incremental costs and QALYs	Group, mean (SE)		Bootstrapped difference (95% CI)
	SC (n = 109)	EC (n = 114)
Base case
QALYs	1.7908 (0.0076)	1.8070 (0.0074)	0.0162 (–0.0047 to 0.0372)
Cost (£)	4369.20 (748.13)	2696.43 (731.37)	–1672.77 (–3455.06 to 109.53)
Complete case
QALYs	1.7931 (0.0098)	1.8022 (0.0010)	0.0091 (–0.0169 to 0.0350)
Cost (£)	3078.10 (362.24)	2495.61 (358.26)	–582.49 (–156.12 to 379.40)
Ethnicity
South Asian
QALYs	1.7784 (0.0168)	1.8290 (0.0154)	0.0507 (–0.0073 to 0.0941)
Cost (£)	2713.04 (527.78)	1737.51 (484.10)	–975.52 (–2227.23 to 497.65)
White
QALYs	1.7862 (0.0010)	1.7968 (0.0095)	0.0106 (–0.0147 to 0.0359)
Cost (£)	3724.13 (376.31)	3024.11 (362.57)	–700.02 (–1576.34 to 553.46)
Other
QALYs	1.8088 (0.0160)	1.8117 (0.0161)	0.0029 (–0.0402 to 0.0460)
Cost (£)	12,235.10 (4816.98)	4494.64 (4847.47)	–7740.48 (–19,198.18 to 4017.14)
Age (years)
≤ 5
QALYs	1.7982 (0.0128)	1.8138 (0.0122)	0.0156 (–0.0073 to 0.0386)
Cost (£)	2381.26 (1621.93)	377.27 (1546.25)	–2003.99 (–4235.41 to 427.12)
≥ 6
QALYs	1.7627 (0.0204)	1.7779 (0.0211)	0.0152 (–0.0253 to 0.0557)
Cost (£)	3534.97 (757.87)	2690.46 (783.96)	–844.50 (–2400.28 to 629.32)

Subgroup analysis

Participants from South Asian backgrounds had the highest QALY gains (see Table 20). Although the mean cost of the EC group was highest for the ‘other’ ethnicities category (non-white and non-South Asian), the intervention was associated with a considerable cost saving for this group. This group were also least likely to gain QALYs from extended prednisolone treatment.

The base-case result was robust to subgroup analysis according to age group, with the extended prednisolone therapy group retaining its dominance in participants who were ≤ 5 years at baseline and in older participants. It was cheaper and more effective to treat the younger age group than older participants, thereby making this population the more beneficial group to treat actively with extended prednisolone course.

Chapter 7 Discussion

The PREDNOS study has not shown any clinical benefit for a 16-week EC of prednisolone compared with the 8-week SC as described by the ISKDC in UK children presenting with SSNS. There was no significant difference between the two treatment groups in time to first relapse of nephrotic syndrome or in any other of the clinically important secondary end points, including the number of relapses experienced, the proportion of participants who went on to develop FRNS or SDNS or the requirement for alternative non-corticosteroid immunosuppressive therapies. However, despite showing no clinical benefit, the cost-effectiveness analysis suggested that EC therapy may be cheaper, with the possibility of a small QALY benefit.

These findings differ from the six studies published prior to the commencement of the PREDNOS study, which had compared the ISKDC regimen with prednisolone regimens of > 3 months’ duration. A Cochrane review29 of these studies performed in 2005 showed a benefit of longer-course prednisolone therapy, with a lower rate of relapse at 12–24 months, and a significant reduction in the number of children with FRNS. Based upon this, a recommendation was made that children presenting with SSNS should be treated with a minimum of 3 months of prednisolone therapy. This did not, however, lead to international consensus, and significant clinical equipoise and variation in practice persisted.

More recent studies have reported results that are consistent with PREDNOS. A Japanese study of 255 participants that compared the ISKDC regimen with a 6-month course of prednisolone found no benefit associated with longer-duration prednisolone therapy. 36 Yoshikawa et al. 36 chose a primary end point of time to the development of FRNS, a clinically important end point identifying those children who have developed a complicated disease course and who are likely to develop disease- and treatment-related morbidity and, therefore, require alternative, more potent, immunosuppressive therapies. There was no difference in FRNS, and the time to first relapse and the incidence of adverse effects were also similar in the two groups. An Indian study comparing 3 and 6 months of prednisolone did not show any benefit associated with increased duration of prednisolone,35 and neither did the Dutch study by Teeninga et al. ,33 which also compared 3 and 6 months of therapy. The inclusion of three well-designed studies in a 2015 update of the Cochrane review13 resulted in a change in the overall conclusions. It was noted that these studies of longer versus shorter duration of corticosteroids had heterogeneous treatment effects, with the older, higher risk of bias studies tending to overestimate the effect of longer-course therapy compared with the more recently published, lower risk of bias studies. Among studies at low risk of bias, there was no significant difference in the risk of FRNS between participants given prednisone for 2 or 3 months and those receiving therapy of longer duration or higher total dose, indicating that there is no benefit of increasing the duration of prednisone beyond 2 or 3 months in the initial episode of SSNS. However, when the meta-analysis was restricted to those addressing the same question as the PREDNOS study, comparing the 8-week ISKDC regimen with regimens of ≥ 3 months (i.e. adding the study of Yoshikawa et al. 36 to the six studies reported in the original Cochrane review), there remained a benefit for the longer (≥ 3 months) treatment regimen, although this only just reached statistical significance. The risk of FRNS was significantly lower (RR 0.68, 95% CI 0.47 to 1.00), as was the number of participants relapsing by 12–24 months (RR 0.8, 95% CI 0.64 to 1.00). It is likely that once the results of the PREDNOS study are added to the Cochrane review that the overall result will show no difference in outcome between the ISKDC and longer treatment regimens.

The data reported in our study are similar to those reported in previous studies. The proportion of participants experiencing a relapse was 80.3% (179/223) over a median follow-up of 37 months, which is comparable to the rate of 60–90% reported in the literature. 13,87 Teeninga et al. 33 reported a relapse rate of 78.6% (99/126) in a European population with a median follow-up of 47 months, with a median time to first relapse of 6 months for the 3-month prednisolone group and 8 months for the 6-month group. Sinha et al. 35 reported a lower relapse rate of 57.8% (104/180); however, participants were followed up for only 12 months. In the Japanese study, the overall relapse rate was not stated; however, the median time to relapse was 242 days and 243 days in the 2-month and 6-month groups, respectively, significantly longer than the 87 days for the SC group and 139 days for EC group observed in the PREDNOS study. It is noteworthy that Yoshikawa et al. used urine dipstick values of ++ or higher as their definition of relapse, although, if anything, this would have overdiagnosed relapse and reduced the time to first relapse. 36

We also found rates of FRNS similar (50% and 53%) to those previously reported. Using the same ISKDC definition as used in PREDNOS, Sinha et al. 35 reported a rate of FRNS at last follow-up of 50.4% in the 6-month group and 60.4% in the 3-month group. The time to FRNS was 23.0 months and 17.6 months, respectively. Teeninga et al. 33 reported FRNS in 45% of the 3-month prednisolone group and in 50% of the 6-month prednisolone group, commenting that this was higher than expected. In previous studies, FRNS has been reported in 32–78% of participants who received a 2-month course of prednisolone15,19–21,47,88 and in 18–44% of those who received prednisolone for 3 months. 20,26,47 It has been proposed that this variation may, in part, be explained by regional differences or variations in definitions of FRNS, length of observation and relapse treatments.

Interestingly, although we found no clinical benefit for the EC prednisolone treatment regimen, we did find that this regimen was cheaper and more effective in QALY terms. The cost analysis showed that, over a 24-month period, the EC treatment regimen cost less because of a lower rate of hospital admission, a shorter duration of hospital stay, fewer hospital emergency visits and fewer outpatient and primary care visits; therefore, on average, it was cheaper than the SC treatment regimen by £1673 per patient. Furthermore, the EC treatment regimen produced more QALYs than the SC treatment regimen. Using commonly applied threshold values for how much society is willing to pay for a QALY gain, the EC treatment regimen is cost-effective. At first glance, this result may seem surprising, as the clinical outcomes have shown little or no benefit of extending prednisolone treatment, yet the heath economics reveals possible evidence of cost-effectiveness. These differences, in part, relate to the differences in cost and QALYs, but also to the different methods of analysis adopted in the health economic and clinical evaluations.

Unlike the objectives of the clinical evaluation, which are about testing whether or not extending prednisolone therapy leads to an improved patient outcome relative to a control group, the objectives of the economic evaluation are to provide an estimation of the value of the extended therapy reflecting both efficiency and equity, and, thus, an estimate of whether or not the difference in cost between the treatment groups is worth the difference in effect, taking into account the opportunity cost of that investment and the fact that the resources could have been invested elsewhere across all parts of the NHS.

The key thing to note is that small insignificant clinical benefits can be cost-effective. The health economics focus is about comparing two things: costs and effects. For the health economic analysis, the effects are measured using QALYs, which reflect societal values and incorporate preferences for domains of QoL. These values are measured using preference-based QoL instruments, for example the CHU-9D questionnaire that was used in this study. Measuring cost differences between different treatment therapies has no meaning until these are offset against differences in effects: it is the simultaneous consideration of costs and effects and, therefore, the joint density of cost and effect differences76 that is the focus of a health economic evaluation.

Within the PREDNOS study, the clinical analysis quantified the difference in time to first relapse, at the individual participant level, using statistical inference. The economic analysis compared the per-participant cost of EC versus SC therapy, and found that the cost was, on average, £1673 lower in the EC group, and the QALYs gained were, on average, 0.0162 higher in the EC group. When the costs and QALY differences are assessed separately, these differences are not statistically significant; however, when assessed simultaneously, the ICER (the ratio of the mean cost and the mean QALY difference) produces a cost-effective result, as the EC regimen is cheaper and produces more QALYs, on average. Therefore, it is dominant, as it is not only more effective in QALY terms but also saves health-care resources, relative to the SC group.

Furthermore, there are different methods within the health economic analysis for representing the uncertainty in the cost and QALY differences. QALYs and cost data tend to have unusual distributional properties and are often skewed, exhibiting ceiling effects or having a bimodal distribution; for this reason, the stochastic bootstrapping method was applied. Bootstrapping generates multiple samples of joint cost and effect estimates from the same trial data, and these cost and effect pairs are then represented on a scatterplot on an incremental cost-effectiveness plane. Figure 15 in Chapter 6 presents the bootstrapped cost and QALY pairs from the PREDNOS study. It shows that most of the pairs lie in the south-east quadrant, indicating that there are cost-savings and QALY gains from the extended therapy versus the standard therapy; however, there are some points spread within the north-east quadrant (indicating that the extended therapy is more costly) and the south-west quadrant (indicating that the extended therapy leads to a QALY loss). This reflects some uncertainty regarding the cost savings and QALY gains to be achieved from extended therapy compared with standard therapy, which is consistent with the finding of a non-significant difference for both costs and effects, when considered independently, between the two treatment groups.

To account for the uncertainty in the cost and effect pairs, the proportions of points falling above and below a willingness-to-pay threshold line are simply counted and then the threshold line is varied to produce a CEAC. CEACs are regarded as an alternative method of calculating CIs and indicate the probability that the extended therapy is cost-effective, compared with standard therapy, for different threshold values of willingness to pay for a QALY gain. Using the PREDNOS trial data, the probability that the extended therapy is cost-effective, at the commonly applied threshold value of £20,000 per QALY, is 0.988. Therefore, despite there being no statistically significant differences in costs and effects for extended therapy compared with standard therapy, the CEAC shows that there is very little uncertainty, from a cost-effectiveness perspective, about the choice to treat patients with EC therapy compared with SC therapy. It is also worth noting that, regardless of benefit measured in QALYs, parents and children value avoidance of hospital admission. This is also valued by clinicians and reduces demand pressures on the NHS.

Previous studies have been somewhat inconsistent in their reporting of the adverse effects associated with using corticosteroids; however, the most recent (2015) Cochrane review13 found no significant differences in the risk of adverse events between extended duration and 2 or 3 months of prednisolone. We found no differences in the adverse effect profiles between the two treatment groups, with the exception of parentally reported poor behaviour, which was significantly more common in the SC group. At 24 months, the cumulative incidence of poor behaviour was 93% in the SC group, compared with 82% in the EC group (RR 0.90, 95% CI 0.82 to 0.98). There was no difference in the incidence of any other adverse effects including Cushingoid facies, striae, hypertrichosis, acne, increased appetite, glycosuria, cataract and abdominal pain. These findings are broadly comparable with those of multiple other larger-36 and smaller-scale20,21,23,25 trials addressing this same clinical question. The large majority of these have found no significant difference in the incidence of adverse effects; however, there was significant heterogeneity in the extent to which these were monitored. Most adverse events were transient and occurred relatively early on during the course of treatment, when the prednisolone dose was at its highest.

We were particularly interested in the impact that the two prednisolone regimens had on behaviour, as expert clinical opinion and advice from our patient and public involvement group indicated that this was the adverse effect of greatest prevalence and significance to families. When the PREDNOS study was designed, no previous study had objectively and systematically investigated this using a quantitative measure. In PREDNOS, we collected quantitative data on behaviour using the ACBC. Although parentally reported poor behaviour was significantly more common in the SC group, when behaviour was assessed objectively through the ACBC questionnaire completed by the parents, there was no significant difference in either the total behaviour score or t-score. The proportion of participants assessed as having abnormal behaviour by the ACBC was also not different between the two groups and varied between 21% and 31% at different time points throughout the study. The proportion of participants whose parents reported poor behaviour was higher than the proportion whose scores were outside the normal ACBC range. This provides some reassurance to parents that perceived poor behaviour is generally within normal bounds and is not greatly impacted by corticosteroid treatment, a finding of relevance in other paediatric conditions treated with corticosteroids. Teeninga et al. 33 assessed behaviour using visual analogue scales. Compared with baseline, participants scored significantly higher on eating, overactive behaviour and aggressive behaviour at 3 months’ follow-up (p < 0.01); however, these scores returned to baseline within 1 year in both groups. Scores for happiness temporarily dropped in the first 6 months, while scores for sleeping remained relatively stable over the entire observation period.

Subgroup analyses showed that there was no clear evidence to suggest that the treatment effect differed according to ethnicity, age or gender, although we were not powered to detect differences in subgroups. For age, there may be some suggestion that the time to first relapse was extended in those in the EC group in participants aged ≤ 5 years, with no difference between the two groups in participants aged ≥ 6 years. This remains a topic of some debate, as a number of studies have reported young age at diagnosis to be associated with an increased risk of FRNS and/or corticosteroid dependence,8,16,47,89 whereas others have not reported this association. 12,90,91 In a post hoc analysis of Sinha et al. ’s study,40 Cox regression suggested that participants aged ≤ 3 years benefited from prolonged therapy, with a reduction in the risk of a first relapse, but not of frequent relapses, and Poisson regression confirmed a higher relative relapse rate in younger participants. Other reports have strongly argued that age may be a predictor of disease severity, including FRNS, corticosteroid dependence and response to cyclophosphamide therapy. 92,93 A few non-randomised studies have investigated the role of gender in the disease course and have reported males to be at a disadvantage. 16,47 Cox regression analysis in the study of Teeninga et al. 33 did not identify boys as being at significantly greater risk of developing FRNS. We found also no evidence of a difference in treatment effect according to gender.

Systolic and diastolic blood pressure z-scores were similar in both treatment groups throughout the course of the study. z-scores were relatively high at the time of the week 4 visit, presumably as a result of the high dose of prednisolone being administered at this point. However, the z-scores decreased progressively during study follow-up. These observations are entirely consistent with the findings of other similar studies.

Interestingly, over the course of the study, following an initial slight fall during the first 16 weeks, the height z-scores increased in both treatment groups. This is an interesting observation, given the fact that these participants received multiple courses of prednisolone for treatment of relapses, a treatment that is known to have a negative impact on linear growth. Previous studies have also reported a fall in height velocity during the first few months of high-dose prednisolone treatment, with a subsequent return to baseline by 12 months. 33 Others have described a dose-dependent effect of corticosteroids on growth in children with SSNS. 94–96 A small number of studies have noted the baseline height SDS to be relatively low in children presenting with SSNS, although no satisfactory explanation has been found for this observation. 33,97 We did not observe this in PREDNOS study. Weight z-scores were relatively constant throughout the study, and BMI z-scores decreased over time.

The main strength of the PREDNOS study is its randomised, double-blind, placebo-controlled design. This ensured a low risk of selection, performance, detection and selective reporting bias. Inclusion criteria were defined to ensure that the study population was representative of the population of children presenting with SSNS in the UK. Outcomes were assessed using internationally accepted ISKDC definitions. Our primary outcome measure of time to first relapse was felt by UK clinicians to be of clinical importance, and previous studies88 have shown a link between timing of first relapse and subsequent clinical course. Baseline features were well balanced and there was a low rate of attrition. Regular safety assessment was ensured through regular clinical review. Prior to the commencement of PREDNOS, previous studies had been small (largest 184 participants) and no previous double-blind placebo controlled RCTs had ever been conducted in children with SSNS. PREDNOS successfully recruited 237 participants from 124 sites throughout all regions of the UK into a double-blind, placebo-controlled RCT. Since then, the studies of both Teeninga et al. 33 and Sinha et al. 35 were double-blinded, although one further study comparing the ISKDC regimen with longer duration therapy was not blinded. 36 The authors acknowledged that this may have introduced preconception bias; however, they proposed that because their design was a non-inferiority trial with regular visits and with relapses being measured objectively, they could not assume positive placebo effects. It must be remembered that this would not have been the case with the reporting of AEs, for which there is considerable scope for bias.

The sample size calculations for the study were based on detecting an absolute difference of 20% in relapse rate at 1 year from 60% in the SC group to 40% in the EC group using a log-rank test (80% power, α = 0.05). The 1-year relapse rate observed in the SC group was 77%. This means that the study has > 85% power to detect an absolute difference of 20% between groups (i.e. from 77% to 57%) using a log-rank test. This makes the likelihood of our results being the result of a type 2 statistical error small.

An additional strength of the PREDNOS study is the generalisability of its findings. We recruited participants with a first presentation of INS from across the UK with broad inclusion criteria. We selected an age range of 1–14 years as this is the range in which the large majority of patients present and are treated empirically with a course of corticosteroids without recourse to renal biopsy. One of the key purposes of the PREDNOS study was to ascertain the optimal prednisolone treatment regimen for UK children by comparing the 8-week SC ISKDC regimen with a longer 16-week EC regimen. We chose the ISKDC SC regimen for one group because this was the regimen in use in the very large majority of UK centres at the time of the planning of the PREDNOS study and chose a 16-week EC regimen as a comparator because longer duration treatment regimens have previously been shown to result in lower rates of relapse and FRNS. The ethnic mix of the study population broadly represented that of the wider population of children with nephrotic syndrome, including significant representation from the South Asian community. We recruited 44 participants (20% of the study population) from families of South Asian origin and 31 (14%) from families recorded as other non-white ethnic origin. This is a very similar figure to that reported in the study of Teeninga et al. ,33 which included 35% of participants of non-Western European descent. This is an important achievement, as the UK South Asian community in particular is significantly over-represented in the SSNS population, the incidence being around six times greater than in the UK white population. Furthermore, the UK South Asian population is generally under-represented in clinical trials and recruitment poses a number of particular challenges. 98 Finally, although formal screening logs were requested, in keeping with other studies these were not kept well; however, based on known epidemiological data, we estimate that we have managed to include 34% of newly presenting patients over a recruitment period of 3 years and 2 months. This indicates a high level of acceptability of the trial among both families and clinicians.

One of the greatest challenges in setting up the PREDNOS study was facilitating the recruitment of participants in district general hospitals. Children with INS have traditionally been, and continue to be, investigated and managed within district general hospitals rather than tertiary paediatric nephrology centres. Referral to tertiary centres generally takes place only if the presenting features are atypical or when investigation or management proves problematic. For this reason, any study that recruited solely from tertiary centres would not reach the large majority of potential participants and would risk sampling a preselected, somewhat atypical group of participants. In the early 2000s, when the PREDNOS study was being planned, there was little paediatric experience in the conduct of RCTs involving an investigational medicinal product and little funded infrastructure to support this work. Our Kidney Research UK and Kids Kidney Research jointly funded pilot study confirmed that there was great willingness among principal investigators to participate in the study and similar interest in the study from participants from both the white and South Asian communities. The pilot study allowed the trial design and infrastructure to be tested, including aspects such as the provision of study medication from a single national clinical trials pharmacy with delivery direct to the participant, attendance rates for study visits and the completion of questionnaires and the study case report forms. The success of the pilot study was significantly enhanced by the development of the NIHR Clinical Research Network: Children (formerly the NIHR MCRN), which commenced operating in 2005. 99 The PREDNOS pilot study was one of the first studies to be adopted onto the MCRN study portfolio and the infrastructure put in place facilitated study set-up and participant recruitment in many sites. The main PREDNOS study was also adopted onto the study portfolio and similarly benefited.

Possible weaknesses of the study include the possibility that the choice and preparation of study drug might have influenced the age profile of the population under study, with a potential trend towards the relative overinclusion of older participants. Because the PREDNOS study drug was supplied in crushable tablet form rather than in suspension or in a soluble or dispersible form, this may have resulted in younger children not participating in the study because of inability or perceived inability to swallow the crushed tablets. The initial ISKDC studies reported that the median age at presentation of MCD nephrotic syndrome was 3 years,2 and a UK series from the county of Yorkshire reported the incidence to be greatest in the 1–4 years age group. 1 The mean age of participants in our study was 4.9 years, with 65% of the study participants being < 6 years of age. This rather suggests that there was a trend towards the recruitment of slightly older participants, perhaps as a result of this study medication formulation issue. However, our study participant age profile is broadly comparable to those reported in the three most recent RCTs of corticosteroid therapy in SSNS by Teeninga et al. 33 (median age 4.2 years, IQR 3.2 to 6.2 years), Yoshikawa et al. 36 (mean age 6.7 years, SD 4.1 years, in the SC group and mean age 6.3 years, SD 4.1 years, in the EC group) and Sinha et al. 35 (median 42.4 months in 3-month treatment group and median 44.2 months in the 6-month treatment group).

The parents of two participants recruited at the chief investigator’s site commented that they felt that they knew which group their child had been randomised to because their child had noticed slight differences in taste between the active prednisolone and placebo tablets. Prednisolone and other oral corticosteroids tend to have a somewhat bitter taste and it may be that other study participants noticed this same phenomenon. This was not, however, reported by other principal investigators and, therefore, it seems unlikely that this would have had a significant impact on the study results.

One further potential minor limitation is the fact that study participants were, in the majority of cases, observed and treated at their local hospital, where the study visits took place. As such, observation and scoring of adverse effects in the study was performed by multiple observers. To avoid this issue would have meant that all study participants would have had to travel to a single or small number of centres, which would have proved a very significant barrier to participation. Randomisation was not minimised by centre, as individual centre contributions were difficult to predict.

In designing the study, we wanted to ensure that we objectively and comprehensively collected prednisolone-related AEs to adequately compare the two treatment regimens. Earlier studies lacked consistency in the level of information that was recorded and none reported quantitative data regarding behavioural change. Our adverse event reporting was, however, somewhat less comprehensive than that of some more recent studies and this warrants some further discussion. Yoshikawa et al. 36 performed formal ophthalmological assessment, including measurement of intraocular pressure, which was found to be elevated in 32 out of 246 participants (13%). 36 In the study of Teeninga et al. ,33 participants underwent formal ophthalmological assessment at diagnosis and after 6 months, specifically looking for evidence of cataract and glaucoma. 33 No cases of glaucoma were detected and only one single case of posterior subcapsular cataract was detected in the 3-month prednisolone group. In the PREDNOS study, we did not ask for participants to have a formal ophthalmological review, although principal investigators screened participants for cataract on an annual basis. Only one case of cataract was detected in each group, a similar frequency to the single case in the study of Sinha et al. 35 On the basis of their observations, Teeninga et al. 33 commented that cataract and glaucoma have been reported with much greater frequency in cohorts of Japanese children than those from other races and that their findings indicate that routine ophthalmological screening was not indicated at an early stage in Dutch children.

Yoshikawa et al. additionally performed regular blood tests and detected minor abnormalities of liver function tests and plasma amylase in up to 21% of participants. 36 We elected not to perform regular blood tests as part of the study protocol, principally because this is not routine clinical practice in the UK; children with SSNS generally have very few blood tests performed unless they are commenced on alternative immunosuppressive therapies mandating monitoring of drug levels or adverse effects. Furthermore, we were of the opinion that the introduction of regular blood tests into the study protocol would have a negative impact on study participation. We did include a single episode of blood sampling for the purpose of collecting deoxyribonucleic (DNA) samples for a separate study aiming to identify potential genetic changes associated with SSNS. A recommendation was made that this be performed at the time of venepuncture for other clinical reasons if this occurred, although our Research Ethics Committee approval permitted us to perform a standalone venepuncture solely for this sample. Sampling was successfully performed in 173 study participants.

In the study by Teeninga et al. 33 of Dutch children, lumbar spine bone mineral density was measured using dual-energy X-ray absorptiometry (DEXA) at baseline and after 6 months. We did not perform this investigation, principally because there is little in the published literature that indicates that significant abnormality of bone mineral density is likely to occur within the first 24 months of treatment, particularly in an unselected cohort of newly presenting SSNS patients. Although our own work has reported a minor reduction in trabecular bone mineral density in adult ‘survivors’ of childhood relapsing nephrotic syndrome, such changes required the use of peripheral quantitative computerised tomography, which is not widely available and would not have been detected using DEXA alone. 100 Many of the district general hospitals participating in PREDNOS would have also experienced difficulties in providing DEXA services for paediatric study participants. Furthermore, a high-quality prospective study examined 60 children with INS and 195 control children and found no deficits in spine or whole body bone mineral content. 101 In the study of Teeninga et al. , no difference was detected in bone mineral density from baseline to 6 months in either group, and they were not able to achieve bone assessment in all participants. 33

Acknowledgements

The PREDNOS investigators would like to thank NIHR for funding this study and Kidney Research UK and Kids Kidney Research, who funded the pilot study. We thank all the parents and participants who agreed to enter the study, the investigators who contributed to the trial and the NIHR Clinical Research Network: Children for their support with study set-up and recruitment. We also thank the Service Users Group from the UK NeST and the Renal Patient Support Group, in particular Mrs Wendy Cook, Mrs Samantha Davies-Abbott and Dr Shahid Muhammad who provided valuable input regarding trial design and conduct and reviewed the plain English summary. The support of the British Association for Paediatric Nephrology and the Royal College of Paediatrics and Child Health is much appreciated.

References

1. McKinney PA, Feltbower RG, Brocklebank JT, Fitzpatrick MM. Time trends and ethnic patterns of childhood nephrotic syndrome in Yorkshire, UK. Pediatr Nephrol 2001;16:1040-4. https://doi.org/10.1007/s004670100021.
2. International Study of Kidney Disease in Children . Nephrotic syndrome in children: prediction of histopathology from clinical and laboratory characteristics at time of diagnosis. Kidney Int 1978;13:159-65. https://doi.org/10.1038/ki.1978.23.
3. Feehally J, Kendell NP, Swift PG, Walls J. High incidence of minimal change nephrotic syndrome in Asians. Arch Dis Child 1985;60:1018-20. https://doi.org/10.1136/adc.60.11.1018.
4. Sharples PM, Poulton J, White RH. Steroid responsive nephrotic syndrome is more common in Asians. Arch Dis Child 1985;60:1014-17. https://doi.org/10.1136/adc.60.11.1014.
5. Vivarelli M, Massella L, Ruggiero B, Emma F. Minimal change disease. Clin J Am Soc Nephrol 2017;12:332-45. https://doi.org/10.2215/CJN.05000516.
6. McCaffrey J, Lennon R, Webb NJ. The non-immunosuppressive management of childhood nephrotic syndrome. Pediatr Nephrol 2016;31:1383-402. https://doi.org/10.1007/s00467-015-3241-0.
7. Webb NJA, Molony D, Craig J. Evidence-Based Nephrology. Oxford: Wiley-Blackwell; 2008.
8. Trompeter RS, Lloyd BW, Hicks J, White RH, Cameron JS. Long-term outcome for children with minimal-change nephrotic syndrome. Lancet 1985;1:368-70. https://doi.org/10.1016/S0140-6736(85)91387-X.
9. Webb NJA, Rees L, Brogan PA, Bockenhauer D, Webb NJA. Pediatric Nephrology. Oxford: Oxford University Press; 2012.
10. Webb NJ, Lewis MA, Iqbal J, Smart PJ, Lendon M, Postlethwaite RJ. Childhood steroid sensitive nephrotic syndrome: does the histology matter?. Am J Kid Dis 1996;27:484-8. https://doi.org/10.1016/S0272-6386(96)90157-2.
11. Trautmann A, Schnaidt S, Lipska-Ziętkiewicz BS, Bodria M, Ozaltin F, Emma F, et al. Long-term outcome of steroid-resistant nephrotic syndrome in children. J Am Soc Nephrol 2017;28:3055-65. https://doi.org/10.1681/ASN.2016101121.
12. International Study of Kidney Disease in Children . Nephrotic syndrome in children: a randomised controlled trial comparing two prednisolone regimens in steroid responsive patients who relapse early. J Pediatr 1979;95:239-43.
13. Hahn D, Hodson EM, Willis NS, Craig JC. Corticosteroid therapy for nephrotic syndrome in children. Cochrane Database Syst Rev 2015;3. https://doi.org/10.1002/14651858.CD001533.pub5.
14. Hall AS, Thorley G, Houtman PN. The effects of corticosteroids on behavior in children with nephrotic syndrome. Pediatr Nephrol 2003;18:1220-3. https://doi.org/10.1007/s00467-003-1295-x.
15. Koskimies O, Vilska J, Rapola J, Hallman N. Long-term outcome of primary nephrotic syndrome. Arch Dis Child 1982;57:544-8. https://doi.org/10.1136/adc.57.7.544.
16. Lewis MA, Baildom EM, Davis N, Houston IB, Postlethwaite RJ. Nephrotic syndrome: from toddlers to twenties. Lancet 1989;1:255-9. https://doi.org/10.1016/S0140-6736(89)91266-X.
17. Fakhouri F, Bocquet N, Taupin P, Presne C, Gagnadoux MF, Landais P, et al. Steroid-sensitive nephrotic syndrome: from childhood to adulthood. Am J Kidney Dis 2003;41:550-7. https://doi.org/10.1053/ajkd.2003.50116.
18. Abramowicz M, Barnett HL, Edelmann CM, Greifer I, Kobayashi O, Arneil GC, et al. Controlled trial of azathioprine in children with nephrotic syndrome. A report for the international study of kidney disease in children. Lancet 1970;1:959-61.
19. Arbeitsgemeinschaft für Padiatrische Nephrologie . Short versus standard prednisone therapy for initial treatment of idiopathic nephrotic syndrome in children. Lancet 1988;1:380-3.
20. Ehrich JH, Brodehl J. Long versus standard prednisone therapy for initial treatment of idiopathic nephrotic syndrome in children. Arbeitsgemeinschaft für Pädiatrische Nephrologie. Eur J Pediatr 1993;152:357-61.
21. Bagga A, Hari P, Srivastava RN. Prolonged versus standard prednisolone therapy for initial episode of nephrotic syndrome. Pediatr Nephrol 1999;13:824-7.
22. Jayantha UK. Comparison of ISKDC regime with a 7 months steroid regime in the first attack of nephrotic syndrome. Pediatr Nephrol 2004;19.
23. Ksiazek J, Wyszyńska T. Short versus long initial prednisone treatment in steroid-sensitive nephrotic syndrome in children. Acta Paediatr 1995;84:889-93.
24. Norero C, Delucchi A, Lagos E, Rosati P. Initial therapy of primary nephrotic syndrome in children: evaluation in a period of 18 months of two prednisone treatment schedules. Chilean Co-operative Group of Study of Nephrotic Syndrome in Children. Rev Med Chil 1996;124:567-72.
25. Ueda N, Chihara M, Kawaguchi S, Niinomi Y, Nonoda T, Matsumoto J, et al. Intermittent versus long-term tapering prednisolone for initial therapy in children with idiopathic nephrotic syndrome. J Pediatr 1988;112:122-6.
26. Hiraoka M, Tsukahara H, Matsubara K, Tsurusawa M, Takeda N, Haruki S, et al. A randomized study of two long-course prednisolone regimens for nephrotic syndrome in children. Am J Kidney Dis 2003;41:1155-62.
27. Pecoraro C, Caropreso MR, Malgieri G, Ferretti AVS, Raddi G, Piscitelli A, et al. Therapy of first episode of steroid responsive nephrotic syndrome: a randomised controlled trial. Pediatr Nephrol 2004;19.
28. Sharma RK, Ahmed M, Gupta A, Gulati S, Sharma AP. Comparison of abrupt withdrawal versus slow tapering regimens of prednisolone therapy in the management of first episode of steroid responsive childhood idiopathic nephrotic syndrome. J Am Soc Nephrol 2000;11.
29. Hodson EM, Knight JF, Willis NS, Craig JC. Corticosteroid therapy for nephrotic syndrome in children. Cochrane Database Syst Rev 2005;1. https://doi.org/1002/14651858.CD001533.
30. Schulz KF, Chalmers I, Hayes RJ, Altman DG. Empirical evidence of bias. Dimensions of methodological quality associated with estimates of treatment effects in controlled trials. JAMA 1995;273:408-12.
31. Moher D, Pham B, Jones A, Cook DJ, Jadad AR, Moher M, et al. Does quality of reports of randomised trials affect estimates of intervention efficacy reported in meta-analyses?. Lancet 1998;352:609-13.
32. Lombel RM, Gipson DS, Hodson EM. Kidney disease: improving global outcomes. Treatment of steroid-sensitive nephrotic syndrome: new guidelines from KDIGO. Pediatr Nephrol 2013;28:415-26. https://doi.org/10.1007/s00467-012-2310-x.
33. Teeninga N, Kist-van Holthe JE, van Rijswijk N, de Mos NI, Hop WCJ, Wetzels JFM, et al. Extending prednisolone treatment does not reduce relapses in childhood nephrotic syndrome. J Am Soc Nephrol 2012;24:149-59.
34. Lande MB, Leonard MB. Variability among pediatric nephrologists in the initial therapy of nephrotic syndrome. Pediatr Nephrol 2000;14:766-9.
35. Sinha A, Saha A, Kumar M, Sharma S, Afzal K, Mehta A, et al. Extending initial prednisolone treatment in a randomised controlled trial from 3 to 6 months did not significantly influence the course of illness. Kidney Int 2015;87:217-24.
36. Yoshikawa N, Nakanishi K, Sako M, Oba MS, Mori R, Ota E, et al. A multicentre randomized trial indicates initial prednisolone treatment for childhood nephrotic syndrome for two months in not inferior to six-month treatment. Kidney Int 2015;87:225-32.
37. Hoyer PF. New lessons from randomized trials in steroid-sensitive nephrotic syndrome: clear evidence against long steroid therapy. Kidney Int 2015;87:17-9.
38. Banh TH, Hussain-Shamsy N, Patel V, Vasilevska-Ristovska J, Borges K, Sibbald C, et al. Ethnic differences in incidence and outcomes of childhood nephrotic syndrome. Clin J Am Soc Nephrol 2016;11:1760-8. https://doi.org/10.2215/CJN.00380116.
39. Takeda A, Matsutani H, Niimura F, Ohgushi H. Risk factors for relapse in childhood nephrotic syndrome. Pediatr Nephrol 1996;10:740-1.
40. Sinha A, Hari P, Sharma PK, Gulati A, Kalaivani M, Mantan M, et al. Disease course in steroid sensitive nephrotic syndrome. Indian Pediatr 2012;49:881-7.
41. Child Growth Standards. Geneva: World Health Organization; n.d.
42. National High Blood Pressure Education Program Working Group on High Blood Pressure in Children and Adolescents . The fourth report on the diagnosis, evaluation, and treatment of high blood pressure in children and adolescents. Pediatrics 2004;114:555-76.
43. Medicines for Human Use (Clinical Trials) Regulations 2004. London: The Stationery Office; 2004.
44. Data Protection Act 1998. London: The Stationery Office; 1998.
45. Human Tissue Act 2004. London: The Stationery Office; 2004.
46. Guidelines for Good Clinical Practice. Geneva: International Conference on Harmonisation; 1996.
47. Andersen RF, Thrane N, Noergaard K, Rytter L, Jespersen B, Rittig S. Early age at debut is a predictor of steroid-dependent and frequent relapsing nephrotic syndrome. Pediatr Nephrol 2010;25:1299-304. https://doi.org/10.1007/s00467-010-1537-7.
48. Varni JW, Burwinkle TM, Seid M, Skarr D. The PedsQL 4.0 as a pediatric population health measure: feasibility, reliability, and validity. Ambul Pediatr 2003;3:329-41.
49. Patrick DL, Deyo RA. Generic and disease-specific measures in assessing health status and quality of life. Med Care 1989;27:217-32.
50. Coons SJ, Rao S, Keininger DL, Hays RD. A comparative review of generic quality-of-life instruments. PharmacoEconomics 2000;17:13-35.
51. Mehrez A, Gafni A. Quality-adjusted life years, utility theory, and healthy-years equivalents. Med Decis Making 1989;9:142-9.
52. Brazier JE, Yang Y, Tsuchiya A, Rowen DL. A review of studies mapping (or cross walking) non-preference based measures of health to generic preference-based measures. Eur J Health Econ 2010;11:215-25. https://doi.org/10.1007/s10198-009-0168-z.
53. Gerard K, Mooney G. QALY league tables: handle with care. Health Econ 1993;2:59-64.
54. Nord E. The QALY – a measure of social value rather than individual utility?. Health Econ 1994;3:89-93.
55. Stevens KJ. Working with children to develop dimensions for a preference-based, generic, pediatric, health-related quality-of-life measure. Qual Health Res 2010;20:340-51. https://doi.org/10.1177/1049732309358328.
56. Griebsch I, Coast J, Brown J. Quality-adjusted life-years lack quality in pediatric care: a critical review of published cost-utility studies in child health. Pediatrics 2005;115:e600-14.
57. Canaway AG, Frew EJ. Measuring preference-based quality of life in children aged 6-7 years: a comparison of the performance of the CHU-9D and EQ-5D-Y – the WAVES pilot study. Qual Life Res 2013;22:173-83. https://doi.org/10.1007/s11136-012-0119-5.
58. Stevens K, Ratcliffe J. Measuring and valuing health benefits for economic evaluation in adolescence: an assessment of the practicality and validity of the child health utility 9D in the Australian adolescent population. Value Health 2012;15:1092-9. https://doi.org/10.1016/j.jval.2012.07.011.
59. Stevens K. Valuation of the Child Health Utility 9D Index. PharmacoEconomics 2012;30:729-47. https://doi.org/10.2165/11599120-000000000-00000.
60. Dakin H. Review of studies mapping from quality of life or clinical measures to EQ-5D: an online database. Health Qual Life Outcomes 2013;11. https://doi.org/10.1186/1477-7525-11-151.
61. Golub GH, Van Loan CF. An analysis of the total least squares problem. SIAM NUMER Anal 1980;17:883-93.
62. Crawley MJ, Crawley MJ. The R Book. Chichester: John Wiley &amp; Sons; n.d.
63. Kramer AA, Zimmerman JE. Assessing the calibration of mortality benchmarks in critical care: The Hosmer–Lemeshow test revisited. Crit Care Med 2007;35:2052-6. https://doi.org/10.1097/01.CCM.0000275267.64078.B0.
64. Hosmer DW, Lemeshow S. Applied Logistic Regression. New Jersey, NJ: John Wiley &amp; Sons; 2004.
65. Tobin J. Estimation of relationships for limited dependent variables. Econometrica 1958;26:24-36.
66. Reynolds A, Shonkwiler J. Testing and correcting for distributional misspecifications in the tobit model: an application of the information matrix test. Empir Econ 1991;16:313-23.
67. Khan KA, Petrou S, Rivero-Arias O, Walters SJ, Boyle SE. Mapping EQ-5D utility scores from the PedsQL™ generic core scales. PharmacoEconomics 2014;32:693-706. https://doi.org/10.1007/s40273-014-0153-y.
68. Hyndman RJ, Koehler AB. Another look at measures of forecast accuracy. Int J Forecast 2006;22:679-88.
69. Petrou S, Rivero-Arias O, Dakin H, Longworth L, Oppe M, Froud R, et al. The MAPS reporting statement for studies mapping onto generic preference-based outcome measures: explanation and elaboration. PharmacoEconomics 2015;33:993-1011. https://doi.org/10.1007/s40273-015-0312-9.
70. Drummond M. Introducing economic and quality of life measurements into clinical studies. Ann Med 2001;33:344-9.
71. Marra CA, Woolcott JC, Kopec JA, Shojania K, Offer R, Brazier JE, et al. A comparison of generic, indirect utility measures (the HUI2, HUI3, SF-6D, and the EQ-5D) and disease-specific instruments (the RAQoL and the HAQ) in rheumatoid arthritis. Soc Sci Med 2005;60:1571-82.
72. Boyer NR, Miller S, Connolly P, McIntosh E. Paving the way for the use of the SDQ in economic evaluations of school-based population health interventions: an empirical analysis of the external validity of SDQ mapping algorithms to the CHU9D in an educational setting. Qual Life Res 2016;25:913-23. https://doi.org/10.1007/s11136-015-1218-x.
73. Furber G, Segal L, Leach M, Cocks J. Mapping scores from the Strengths and Difficulties Questionnaire (SDQ) to preference-based utility values. Qual Life Res 2014;23:403-11. https://doi.org/10.1007/s11136-013-0494-6.
74. Chen G, Stevens K, Rowen D, Ratcliffe J. From KIDSCREEN-10 to CHU9D: creating a unique mapping algorithm for application in economic evaluation. Health Qual Life Outcomes 2014;12. https://doi.org/10.1186/s12955-014-0134-z.
75. Ramsey SD, Willke RJ, Glick H, Reed SD, Augustovski F, Jonsson B, et al. Cost-effectiveness analysis alongside clinical trials II-An ISPOR Good Research Practices Task Force report. Value Health 2015;18:161-72. https://doi.org/10.1016/j.jval.2015.02.001.
76. Guide to the Methods of Technology. London: National Institute for Health and Care Excellence; 2013.
77. Torrance GW, Drummond MF. Methods for the Economic Evaluation of Health Care Programmes. Oxford: Oxford University Press; 2005.
78. Manca A, Hawkins N, Sculpher MJ. Estimating mean QALYs in trial-based cost-effectiveness analysis: the importance of controlling for baseline utility. Health Econ 2005;14:487-96. https://doi.org/10.1002/hec.944.
79. National Schedule of Reference Costs: 2015–16. London: Department of Health and Social Care; 2015.
80. Curtis L, Burns A. Unit Costs of Health and Social Care 2015. Canterbury: Personal Social Services Research Unit, University of Kent; 2015.
81. PbR Tariff Information Spreadsheet for 2013 to 2014. London: Department of Health and Social Care; 2014.
82. Guidance on Actions and Uses of Drugs Prescribed in the UK. London: BMJ Group and Pharmaceutical Press; 2015.
83. Burton A, Billingham LJ, Bryan S. Cost-effectiveness in clinical trials: using multiple imputation to deal with incomplete cost data. Clin Trials 2007;4:154-61.
84. Rubin DB, Schenker N. Multiple imputation in health-care databases: an overview and some applications. Stat Med 1991;10:585-98.
85. O’Brien BJ, Briggs AH. Analysis of uncertainty in health care cost-effectiveness studies: an introduction to statistical issues and methods. Stat Methods Med Res 2002;11:455-68. https://doi.org/10.1191/0962280202sm304ra.
86. Thompson AJ, Newman WG, Elliott RA, Roberts SA, Tricker K, Payne K. The cost-effectiveness of a pharmacogenetic test: a trial-based evaluation of TPMT genotyping for azathioprine. Value Health 2014;17:22-33. https://doi.org/10.1016/j.jval.2013.10.007.
87. Tarshish P, Tobin JN, Bernstein J, Edelmann CM. Prognostic significance of the early course of minimal change nephrotic syndrome: report of the International Study of Kidney Disease in Children. J Am Soc Nephrol 1997;8:769-76.
88. International Study of Kidney Disease in Children . Early identification of frequent relapsers among children with minimal change nephrotic syndrome. A report of the International Study of Kidney Disease in Children. J Pediatr 1982;101:514-18.
89. Kabuki N, Okugawa T, Hayakawa H, Tomizawa S, Kasahara T, Uchiyama M. Influence of age at onset on the outcome of steroid-sensitive nephrotic syndrome. Pediatr Nephrol 1998;12:467-70.
90. Yap HK, Han EJ, Heng CK, Gong WK. Risk factors for steroid dependency in children with idiopathic nephrotic syndrome. Pediatr Nephrol 2001;16:1049-52. https://doi.org/10.1007/s004670100024.
91. Takeda A, Takimoto H, Mizusawa Y, Simoda M. Prediction of subsequent relapse in children with steroid-sensitive nephrotic syndrome. Pediatr Nephrol 2001;16:888-93.
92. Hoyer PF, Brodeh J. Initial treatment of idiopathic nephrotic syndrome in children: prednisone versus prednisone plus cyclosporine A: a prospective, randomized trial. J Am Soc Nephrol 2006;17:1151-7.
93. Vester U, Kranz B, Zimmermann S, Hoyer PF. Cyclophosphamide in steroid-sensitive nephrotic syndrome: outcome and outlook. Pediatr Nephrol 2003;18:661-4. https://doi.org/10.1007/s00467-003-1170-9.
94. Donatti TL, Koch VH. Final height of adults with childhood-onset steroid-responsive idiopathic nephrotic syndrome. Pediatr Nephrol 2009;24:2401-8. https://doi.org/10.1007/s00467-009-1301-z.
95. Simmonds J, Grundy N, Trompeter R, Tullus K. Long-term steroid treatment and growth: a study in steroid-dependent nephrotic syndrome. Arch Dis Child 2010;95:146-9. https://doi.org/10.1136/adc.2007.129957.
96. Hung YT, Yang LY. Follow-up of linear growth of body height in children with nephrotic syndrome. J Microbiol Immunol Infect 2006;39:422-5.
97. Schärer K, Essigmann HC, Schaefer F. Body growth of children with steroid-resistant nephrotic syndrome. Pediatr Nephrol 1999;13:828-34.
98. Macneill V, Nwokoro C, Griffiths C, Grigg J, Seale C. Recruiting ethnic minority participants to a clinical trial: a qualitative study. BMJ Open 2013;3. https://doi.org/10.1136/bmjopen-2013-002750.
99. Attar SG, Poustie V, Beresford MG. The medicines for children research network: building on current success as we move forward. Clin Invest 2014;4:399-405.
100. Hegarty J, Mughal MZ, Adams J, Webb NJ. Reduced bone mineral density in adults treated with high-dose corticosteroids for childhood nephrotic syndrome. Kidney Int 2005;68:2304-9.
101. Leonard MB, Feldman HI, Shults J, Zemel BS, Foster BJ, Stallings VA. Long-term, high-dose glucocorticoids and bone mineral content in childhood glucocorticoid-sensitive nephrotic syndrome. N Engl J Med 2004;351:868-75. https://doi.org/10.1056/NEJMoa040367.

Appendix 1 Site recruitment

Appendix 2 Achenbach Child Behaviour Checklist

Appendix 3 Economic analysis: the mapping exercise

FIGURE 17.

Distribution in the estimation and validation samples. Observed CHU-9D score, (a) estimation group and (b) validation group; GLM_6-predicted CHU-9D score, (c) estimation group and (d) validation group; and OLS_3-predicted CHU-9D score, (e) estimation group and (f) validation group.

Appendix 4 Economic evaluation

List of abbreviations

ACBC

Achenbach Child Behaviour Checklist

AE

adverse event

APN

Arbeitsgemainschaft für Pädiatrische Nephrologie

BMI

body mass index

CEAC

cost-effectiveness acceptability curve

CHU-9D

Child Health Utility 9D

CI

confidence interval

DEXA

dual-energy X-ray absorptiometry

EC

extended course

FRNS

frequently relapsing nephrotic syndrome

GLM

generalised linear model

GP

general practitioner

HR

hazard ratio

HRG

Healthcare Resource Group

HRQoL

health-related quality of life

ICER

incremental cost-effectiveness ratio

INS

idiopathic nephrotic syndrome

IQR

interquartile range

IRR

incidence rate ratio

ISKDC

International Study for Kidney Disease in Children

ITT

intention to treat

MAE

mean absolute error

MCD

minimal change disease

MCRN

Medicines for Children Research Network

MSE

mean squared error

NeST

Nephrotic Syndrome Trust

NIHR

National Institute for Health Research

OLS

ordinary least squares

PedsQL

Pediatric Quality of Life Inventory

PREDNOS

PREDnisolone in NephrOtic Syndrome

QALY

quality-adjusted life year

QoL

quality of life

RCT

randomised controlled trial

RR

relative risk

SAE

serious adverse event

SC

standard course

SD

standard deviation

SDNS

steroid-dependent nephrotic syndrome

SDS

standard deviation score

SE

standard error

SSNS

steroid-sensitive nephrotic syndrome