Cost-Utility Analysis of Mycophenolate Mofetil versus Azathioprine Based Regimens for Maintenance Therapy of Proliferative Lupus Nephritis

Table 2

(a) Base-case inputs of probability parameters in lifetime model (Cochrane data). (b) Lifetime model inputs of direct and indirect costs. (c) Lifetime model inputs of utility parameters.

(a)

Probability parameters (1-year cycle)	Base-case value	Range (95% CI)	Probability distribution	Sources
Remission in AZA group
Probability of lupus-related death during remission	Age-dependent	n/a	n/a	Bernatsky et al. 2006 [27]; Arias 2011 [28]; Cochrane 2012 [13]
Probability of ESRD during remission	0.0061	0.0012–0.0317	Beta (35.9, 11,927.1)	Cochrane 2012 [13]
Probability of relapse during remission	0.0716	0.0463–0.1140	Beta (5106.9, 135,192.3)	Cochrane 2012 [13]
Remission in MMF group
Probability of lupus-related death during remission	Age-dependent	n/a	n/a	Bernatsky et al. 2006 [27]; Arias 2011 [28]; Cochrane 2012 [13]
Probability of ESRD during remission	0.0025	0.0005–0.0125	Beta (16.6, 6771.3)	Cochrane 2012 [13]
Probability of relapse during remission	0.0367	0.0244–0.0564	Beta (16.0, 419.5)	Cochrane 2012 [13]
Relapse in MMF group
Probability of lupus-related death during relapse	Age-dependent	n/a	n/a	Bernatsky et al. 2006 [27]; Arias 2011 [28]
Probability of ESRD during relapse	0.1183	0.0455–0.2910	Beta (491.1, 3670.9)	Cochrane 2012 [13]
Probability of complete and partial remissions	0.8319	0.6613–0.9313	Beta (45.7, 9.2)	Cochrane 2012 [13]
Relapse in AZA group
Probability of lupus-related death during relapse	Age-dependent	n/a	n/a	Bernatsky et al. 2006 [27]; Arias 2011 [28]
Probability of ESRD during relapse	0.1183	0.0455–0.2910	Beta (491.1, 3670.9)	Cochrane 2012 [13]
Probability of complete and partial remissions	0.8319	0.6613–0.9313	Beta (45.7, 9.2)	Cochrane 2012 [13]
ESRD due to lupus nephritis
Probability of death due to lupus nephritis ESRD	Age-dependent	n/a	n/a	USRDS 2012 [20]; Sule et al. 2011 [29]

AZA: azathioprine; MMF: mycophenolate mofetil; CYC: cyclophosphamide; ESRD: end stage renal disease; CI: confidence interval.
Probabilities from the data sources were reported over various follow-up durations. Probabilities were converted to rates and then to 6-month probabilities [15]. First, the probabilities were converted to yearly rates (event per patient per year) using the equation , where = rate; = time in years; = probability of an event occurring during time .
These annual rates were then converted to 6-month probabilities using the equation , where = one-year rate; = time in years; = probability of an event occurring during time .
Beta distributions are characterized by (, ).
Probability based on MMF for relapse in either AZA- or MMF-based regimen.
The age-specific annual mortality rate for the general dialysis population in 2011 [20] is multiplied by hazard ratio (HR) 1.7. In a USRDS study, Sule et al. found that adult patients with ESRD secondary to SLE were at increased risk of death compared with other adult patients (HR 1.7; 95% CI 1.2–2.7) [29]. Conversion between rates and probabilities as noted above.
In the relapse state for both MMF and AZA strategies, the rate of lupus-related death is derived from age-specific annual mortality rate in the general population [28] multiplied by a standardized mortality ratio (SMR) 7.9. In a cohort of 9,547 SLE patients, Bernatsky et al. estimated an SMR 7.9 in those with nephritis [27]. Conversion between rates and probabilities as noted above.
Values in (e) divided by 9.3, given that the relative risk of lupus-related death during relapse versus remission on MMF treatment is 9.3 [13].
Values in (f) × 0.58, given that the relative risk of lupus-related death during remission on AZA versus MMF is 0.58 [13].

(b)

Cost parameters (over 1-year period)	Mean costs ($)	Range ($)	Probability distribution	Sources/Comments
Direct costs × 1 year (nonpharmaceuticals)
Remission	3,368.34	1263.13–2105.21	Gamma (31.5, 0.019)	Clarke et al. 2004 [18]; Clarke et al. 2008 [17]; range assumed to be ±25% mean
Relapse	6,486.85	2432.57–4054.29	Gamma (29.2, 0.009)	Clarke et al. 2004 [18]; Clarke et al. 2008 [17]; range assumed to be ±25% mean
Indirect costs × 1 year
Remission	16,066.38	6024.89–10041.49	Gamma (16.1, 0.002)	Panopalis et al. 2007 [19]; Clarke et al. 2008 [17]; range assumed to be ±25% mean
Relapse	17,128.13	6423.05–10705.09	Gamma (18.3, 0.002)	Panopalis et al. 2007 [19]; Clarke et al. 2008 [17]; range assumed to be ±25% mean
ESRD/dialysis: mean cost per person × 1 year	86,608	n/a	Gamma (75.0, 0.002)	USRDS 2012 [20]

ESRD: end stage renal disease; USRDS: United States Renal Data System.
Direct costs included care provided by specialists, nonspecialists, nonphysician healthcare professionals, laboratory studies, imaging studies, emergency room visits, and outpatient surgery and hospitalizations [18].
Indirect costs included time lost from labor and nonlabor (i.e., household work) market activity as well as time that a caregiver spent helping the patient receiving healthcare services and the time the caregiver spent doing housework [19].
Gamma distributions are characterized by ; , , where = mean; = variance.

(c)

Utility parameters	Base-case mean	Range	Probability distribution	Sources/Comments
Utility of ESRD on dialysis	0.67	0.54–0.85	Beta (14.1, 6.9)	Liem et al. 2008 [22], based on TTO method
Utility of remission, on MMF or AZA	0.70	0.65–0.80	Beta (14.0, 6.0)	Grootscholten et al. 2007 [23], Clarke et al. 2008 [17], based on VAS method
Utility of relapse, on MMF or AZA	0.60	0.50–0.70	Beta (13.8, 9.2)	Grootscholten et al. 2007 [23], Clarke et al. 2008 [17], based on VAS method
Utility of death	0.00	n/a	n/a	Drummond et al. 2005 [26]

AZA: azathioprine; MMF: mycophenolate mofetil; ESRD: end stage renal disease; TTO: time trade-off; VAS: visual analog scale.
Based on 95% confidence interval or standard deviation.
Beta distributions are characterized by (α, β).