• Tidak ada hasil yang ditemukan

Supplemental Table 3: Posterior classification table of the 2-class model

N/A
N/A
Protected

Academic year: 2023

Membagikan "Supplemental Table 3: Posterior classification table of the 2-class model"

Copied!
14
0
0

Teks penuh

(1)

Supplemental Material

Supplemental Summary 1. CKD-REIN clinical sites and investigators, by region Supplemental Table 1: Definitions of the operational variables used in the study.

Supplemental Table 2: Summary table of the statistical criteria for fitting and classification of the different n-class models tested.

Supplemental Table 3: Posterior classification table of the 2-class model.

Supplemental Table 4: Comparison of participants included in the analysis (n=2,787) and those not included (n=246).

Supplemental Figure 1: Observed individual symptoms trajectories of 50 randomly selected participants a posterior classified in each class.

Supplemental Figure 2: Time course of individual symptoms according to trajectories.

Supplemental Figure 3: Proportion of severe symptoms (a lot or very much) at baseline according to trajectory.

Supplemental Item 1: Main analysis strategies

(2)

Supplemental Summary 1. CKD-REIN clinical sites and investigators, by region

Alsace: Prs T. Hannedouche et B. Moulin (CHU, Strasbourg), Dr A. Klein (CH Colmar);

Aquitaine: Pr C. Combe (CHU, Bordeaux), Dr J.P. Bourdenx (Clinique St Augustin, Bordeaux),

Dr A. Keller, Dr C. Delclaux (CH, Libourne), Dr B. Vendrely (Clinique St Martin, Pessac), Dr B.

Deroure (Clinique Delay, Bayonne), Dr A. Lacraz (CH, Bayonne); Basse Normandie: Dr T.

Lobbedez (CHU, Caen), Dr I. Landru (CH, Lisieux); Ile de France: Pr Z. Massy (CHU, Boulogne–

Billancourt), Pr P. Lang (CHU, Créteil), Dr X. Belenfant (CH, Montreuil), Pr E. Thervet (CHU, Paris), Dr P. Urena (Clinique du Landy, St Ouen), Dr M. Delahousse (Hôpital Foch, Suresnes);

Languedoc–Roussillon: Dr C. Vela (CH, Perpignan); Limousin: Pr M. Essig, Dr D. Clément (CHU,

Limoges); Lorraine: Dr H. Sekhri, Dr M. Smati (CH, Epinal), Dr M. Jamali, Dr B. Hacq (Clinique Louis Pasteur, Essey-les-Nancy), Dr V. Panescu, Dr M. Bellou (Polyclinique de Gentilly, Nancy), Pr Luc Frimat (CHU, Vandœuvre-les-Nancy); Midi-Pyrénées: Pr N Kamar (CHU, Toulouse);

Nord-Pas-de-Calais: Prs C. Noël et F. Glowacki (CHU, Lille), Dr N. Maisonneuve (CH,

Valenciennes), Dr R. Azar (CH, Dunkerque), Dr M. Hoffmann (Hôpital privé La Louvière, Lille);

Pays-de-la Loire: Pr M. Hourmant (CHU, Nantes), Dr A. Testa (Centre de dialyse, Rezé), Dr D.

Besnier (CH, St Nazaire) Picardie: Pr G. Choukroun (CHU, Amiens), Dr G. Lambrey (CH, Beauvais); Provence-Alpes–Côte d’Azur: Pr S. Burtey (CHU, Marseille), Dr G. Lebrun (CH, Aix- en-Provence), Dr E. Magnant (Polyclinique du Parc Rambot, Aix-en-Provence); Rhône-Alpes: Pr M. Laville, Pr D. Fouque (CHU, Lyon-Sud) et L. Juillard (CHU Edouard Herriot, Lyon), Dr C.

Chazot (Centre de rein artificiel Tassin Charcot, Ste Foy-les-Lyon), Pr P. Zaoui (CHU, Grenoble), Dr F. Kuentz (Centre de santé rénale, Grenoble).

(3)

Supplemental Table 1: Definitions of the operational variables used in the study.

Variables Definitions Depression

score

Assessed by the Center for Epidemiologic Studies Depression Scale (CES-D) depression symptoms index, short Boston form. Variable score standardized on a scale from 0 to 100, with higher scores indicating the presence of more depression.

Physical activity Assessed by the Global Physical Activity Questionnaire (GPAQ) and classified according to WHO physical activity levels:

- High: intense physical activity at least 3 days per week, resulting in an energy expenditure of at least 1500 MET-min/week OR at least 7 days of walking and moderate or intense physical activity until reaching a minimum of 3000 MET-min per week.

- Moderate: not meeting the criteria in the previous category but meeting one of the following criteria: at least 20 min of vigorous physical activity per day for ≥ 3 days per week OR at least 30 min of moderate physical activity or walking per day for ≥ 5 days per week OR at least 5 days of walking and moderate or vigorous physical activity, until a minimum of 600 MET-min per week is achieved.

- Low: not meeting the above criteria.

Chronic kidney disease stages (eGFR)

Classification of CKD into 5 stages based on eGFR by Kidney Disease Improving Global Outcomes (KDIGO) 2012. Stage 3a: eGFR 45–59 ml/min/1.73m2, stage 3b: eGFR 30–44 ml/min/1.73m2, stage 4: eGFR 15–

29 ml/min/1.73m2, and stage 5: eGFR <15 ml/min/1.73m2. eGFR was estimated by Chronic Kidney Disease Epidemiology Collaboration 2009.

Chronic kidney disease stages (ACR)

Classification of CKD according to the ratio of urine albumin to creatinine (ACR). Stage A1: ACR < 30 mg/g, A2: ACR 30 to 299 mg/g, and A3: ACR

≥ 300 mg/g (KDIGO 2012).

Dropout The variables initiating kidney replacement therapy (KRT) and death before KRT were considered dropouts. KRT was defined as the initiation of dialysis (hemodialysis/peritoneal dialysis) or kidney transplantation.

(4)

Deaths and KRT events (dialysis or pre-emptive transplantation) were tracked from study enrolment (July 2013 to May 2016) until December 31, 2020 (censoring date). Deaths were identified from medical records, or reported by family members, or traced through record linkage with the national death registry. KRT (dialysis or pre-emptive transplantation) events were identified from patient interview or by record linkage with the national REIN registry of KRT. Less than 5% of the participants ended CKD care at the nephrology clinics before December 31, 2020 and were not identified in the KRT of death registry. Their follow-up was censored at the date of last visit.

BMI Body mass status was classified according to the WHO. Underweight: BMI

< 18.5 kg/m2, healthy weight: 18.5 to <25 kg/m2, overweight: 25.0 to <30 kg/m2, and obese: ≥ 30.0 kg/m2.

Diabetes mellitus

Presence of any of the following criteria: 1) treatment with insulin or oral antidiabetics, 2) fasting blood glucose ≥ 7 mmol/L, 3) random blood glucose ≥ 11 mmol/L, and 4) HbA1c level ≥ 6.5%.

Cardiovascular history

History of stroke, ischemic heart disease or heart failure

Anemia Hemoglobin level < 12.0 g/dl in women and < 13.0 g/dl in men Serum calcium

level

Low: < 8.4 mg/dl High: > 10.4 mg/dl Serum

potassium level

Low: < 3.5 mmol/L High: > 5.3 mmol/L

(5)

Supplemental Table 2: Summary table of the statistical criteria for fitting and classification of the different

n-class models tested. The 2-class model with a matrix common to all classes was retained. The statistical criteria of the 5-class model were not presented because number of iterations (200 and a grid search (30 iterations and 100 replications)) reached without convergence. Abbreviations: BIC, Bayesian Information Criterion; ICL, Integrated Classification Likelihood; Postprob, posterior probability.

Model npm BIC ICL Entro

-pie

% of participants in each class Class

1

Class 2

Class 3

Class 4

Class 5 1 class 16 77161.1 77161.1 1,00 100

2 classes 24 77045.8 72019.6 0,65 31.39 68.60

3 classes 32 77064.4 72754.0 0.57 35.67 11.45 52.89

4 classes 40 77089.4 72863.4 0.56 23.36 7.97 15.97 52.71 5 classes 48 Number of iterations reached without convergence

(6)

Supplemental Table 3: Posterior classification table of the 2-class model.

Class

Mean of posterior probabilities in each class

Class 1 Class 2

Class 1 0.8430 0.1570

Class 2 0.0714 0.9286

(7)

Supplemental Table 4: Comparison of participants included in the analysis (n=2,787) and those excluded (n=246). Chronic Kidney Disease-Renal Epidemiology and Information Network (CKD- REIN) cohort.

Characteristics Overall, N = 3,033

Included N = 2,787

Excluded N = 246

p-value

Age (years) 67 ± 13 67 ± 13 65 ± 15 0.066

Age group (years) <0.001

18-44 206 (7) 174 (6) 32 (13)

45-64 917 (30) 845 (30) 72 (29)

65-74 853 (28) 785 (28) 68 (28)

≥ 75 1,057 (35) 983 (35) 74 (30)

Male sex 1,982 (65) 1,829 (66) 153 (62) 0.3

Currently married 1,695 (56) 1,690 (61) 5 (2) <0.001

BMI (kg/m2) 29 ± 6 29 ± 6) 29 ± 6 0.2

Body weight status 0.008

Underweight 46 (2) 41 (2) 5 (2)

Healthy weight 789 (26) 734 (26) 55 (22)

Overweight 1,050 (35) 956 (34) 94 (38)

Obesity 1,082 (36) 1,003 (36) 79 (32)

Diabetes mellitus 1,307 (43) 1,172 (42) 135 (55) <0.001 Cardiovascular history 1,594 (53) 1,457 (52) 137 (56) 0.6 Charlson Comorbidity

Index ≥ 5

2,303 (76) 2,126 (76) 177 (72) 0.030

KRT 768 (25) 690 (25) 78 (31) 0.016

Died before KRT 771 (25) 681 (24) 90 (37) <0.001

Number of drugs 8 ± 4 8 ± 4 9 ± 4 0.017

PCS score 42 ± 10 42 ± 10 38 ± 13 0.5

MCS score 48 ± 7 48 ± 7 44 ± 5 0.2

Burden score 74 ± 24 74 ± 24 44 ± 16 0.003

Effect of kidney disease score

81 ± 18 81 ± 18 58 ± 21 0.023

Depression score (CES-D) 25 ± 17 25 ± 17 34 ± 23 0.3

Physical activity (GPAQ) 0.9

Intense 689 (27) 687 (27) 2 (33)

Moderate 625 (25) 624 (25) 1 (17)

Low 1,214 (48) 1,211 (48) 3 (50)

eGFR (ml/min/1.73m2) 33 ± 12 33 ± 12 32 ± 13 0.041

CKD stages (eGFR) 0.087

Stage 2 65 (2) 59 (2) 6 (2)

Stage 3 1,601 (53) 1,487 (53) 114 (46)

Stage 4 1,249 (41) 1,138 (41) 111 (45)

Stage 5 118 (4) 103 (4) 15 (6)

(8)

Urine albumin-creatinine ratio (mg/g)

0.2

< 30 767 (25) 711 (26) 56 (23)

30 to 299 861 (28) 798 (29) 63 (26)

≥ 300 1,130 (37) 1,034 (37) 96 (39)

Not reported 275 (9) 244 (9) 31 (13)

Anemia* 1,143 (38) 1,038 (37) 105 (43) 0.11

Serum albumin < 4.0 g/dL 245 (8) 213 (8) 32 (13) <0.001 All data are presented as n (%) or mean ± SD.

Abbreviations: BMI, body mass index; eGFR, estimated glomerular filtration rate; MCS, mental component summary; PCS, physical component summary; CKD, chronic kidney disease; KRT, kidney replacement therapy; GPAQ, Global Physical Activity Questionnaire;

CES-D, Center for Epidemiologic Studies Depression Scale

A higher score indicates the presence of more depression or best quality of life (PCS, MCS, burden, and effect)

* Hemoglobin level < 12.0 g/dL in women and < 13.0 g/dL in men.

(9)

Supplemental Figure 1: Observed individual symptom score trajectories of 50 randomly selected

participants a posterior classified in each class. “Worse symptom score and worsening trajectory” is represented in red and “Better symptom score and stable trajectory” in green. More than half of the participants in the “Worse symptom score and worsening trajectory” had an event (KRT or death before KRT) at 3 years of follow-up, unlike those in the “Better symptom score and stable trajectory”. Chronic Kidney Disease-Renal Epidemiology and Information Network (CKD-REIN) cohort.

(10)

Supplemental Figure 2: Evolution of KDQOL-36 symptom items over time, by sub-group of symptom

score trajectories. Each point represents the mean of the responses (1 to 5) available for each symptom at each time point. A) Participants (n=875) classified in the “Worse symptom score and worsening trajectory”.

B) Participants (n=1912) classified in the “Better symptom score and stable trajectory”. The higher the score, the more severe the symptom. Scores 1 to 5 represent the 5 Likert-type response options (1=not at all; 2= a little; 3=Moderately; 4=a lot; 5: very much).

(11)

Supplemental Figure 3: Prevalence (%) of severe symptoms (a lot or very much) at baseline by sub-group of symptom trajectory. Severe symptoms were more common in the sub-group with

“Worse symptom score and worsening trajectory”.

0 5 10 15 20 25 30 35

Fatigue

Muscle pain and soreness

Cramps

Shortness of breath

Dry skin

Numbness in hands or feet

Itchy skin

Faintness or dizziness

Lack of apetite

Nausea or upset stomach

Chest pain

Better symptom score and stable trajectory Worse symptom score and worsening trajectory

(12)

Supplemental Item 1: Main analysis strategies

Modeling the best link function was the first step of the main analysis. A joint 1-class empty model with no covariates outside of time tested the best link function between the observed symptom dimension scores and the latent process because the distribution of scores does not follow a normal distribution. We tested the following link functions: linear, cumulative distribution function of the beta distribution (beta-CDF), natural splines with three knots at quantiles, natural splines with five knots at quantiles, and natural splines with seven knots at quantiles. These six models were compared by Akaike information criterion (AIC) and number of parameters. The link function with the smallest AIC and the simplest was selected. This step allowed us to retain the beta-CDF link function given its AIC and its number of parameters. Thus, the beta-CDF function is used in all other models.

In the second step, several combinations of risk and time functions with an unstructured or diagonal variance-covariance matrix (idiag=TRUE or FALSE) were tested. Twenty models (20 combinations) with 1-class were tested and compared by AIC and number of parameters. This step resulted in the selection of the model combining a natural spline with one interior knot at the median and a Weibull hazard function for the two events with an unstructured variance-covariance matrix.

Using a beta-CDF link function, a natural spline with one interior knot at the median of the time, a Weibull hazard function for the two competitive events (KRT and death before KRT), and an unstructured variance- covariance matrix, we tested one- to multi-class models with no covariates outside of time by using the following strategy:

(13)

- Systematically set the initial parameters from the 1-class model (mod1).

> Script code R for two-class JLCMM

summary(mod2<- jlcmm(symp~ ns(time, knots = 1.3989, Boundary.knots =c(0,5.7123 )),

random= ~ ns(time, knots = 1.3989, Boundary.knots =c(0,5.7123 )), mixture = ~ns(time, knots = 1.3989, Boundary.knots =c(0,5.7123 )),

survival = surv(Tevent, event)~cause1(1)+cause2(1),

hazard = c(“Weibull”,”Weibull”),maxiter = 200,B=random(mod1), link=“beta”, ng=2,hazardtype= c(“Specific”,”Specific”),

subject=“id_ckdrein”, data=table_mcj))

- A grid search with 30 replications and 60 iterations was systematically performed for the models with more than one class.

> Script code R for two-class JLCMM with grid search

summary(grid_mod2<- gridsearch(rep = 30,maxiter = 60,minit = mod1,

jlcmm(symp~ ns(time, knots = 1.3989, Boundary.knots =c(0,5.7123 )), random= ~ ns(time, knots = 1.3989, Boundary.knots =c(0,5.7123 )), mixture = ~ns(time, knots = 1.3989, Boundary.knots =c(0,5.7123 )), survival = surv(Tevent, event)~cause1(1)+cause2(1),

hazard = c(“Weibull”,”Weibull”),maxiter = 200,

(14)

link=“beta”, ng=2,hazardtype= c(“Specific”,”Specific”), subject=“id_ckdrein”, data=table_mcj)))

- For each model, we systematically tested a variance covariance matrix common to the classes and then specific to each class.

Referensi

Dokumen terkait