Development and Validation of a Nomogram for the Prediction of Unfavorable Treatment Outcome Among Multi-Drug Resistant Tuberculosis Patients in North West Ethiopia: An Application of Prediction Modelling

Figures & data

Figure 1 Flowchart of participant selection for the development of nomogram for prediction of unfavorable treatment outcome among MDR-TB patients, and reasons for exclusion, North West Ethiopia, September 2010 to July 2020.

Table 1 Socio-Demographic and Behavioral Characteristics of MDR-TB Patients in North West Ethiopia, 2010 to 2020 (N = 517)

Characteristics	Frequency (%)
Gender
Male	321 (62.1)
Female	196 (37.9)
Age (Median ± IQR)	30±17
Occupation
Government employee	33 (6.4)
Self-employed	69 (13.4)
Farmer	217 (42)
Unemployed	8 (1.5)
Student	75 (14.5)
Daily laborer	92 (17.8)
Others*	23 (4.4)
Educational status
No formal education	193 (37.4)
Primary	192 (37.1)
Secondary	90 (17.4)
Tertiary	42 (8.1)
Marital status
Single	144 (27.9)
Married	236 (45.7)
Divorced	83 (16.0)
Widowed	14 (2.7)
Others**	40 (7.7)
Religion
Orthodox	483 (93.4)
Muslim	30 (5.8)
Others***	4 (0.8)
Residence
Rural	264 (51.1)
Urban	253 (48.9)
Treatment supporter
Yes	435 (84.1)
No	82 (15.9)
Baseline smoking status
No	493 (95.4)
Yes	24 (4.6)
Baseline alcohol use
No	373 (72.1)
Yes	144 (27.9)
Baseline khat use
No	499 (96.5)
Yes	18 (3.5)

Notes: Others *Preschool children, Others **Children below 18 years, Others ***Protestant and catholic.

Table 2 Clinical Characteristics of MDR TB Patients in North West Ethiopia, 2010 to 2020 (N = 517)

Characteristics	Frequency (%)
Type of drug regimen
Long term regimen	475 (91.9)
Short term regimen	42 (8.1)
Registration group
Previously treated	422 (81.6)
New	95 (18.4)
Model of care
Hospitalized	483 (93.4)
Ambulatory	34 (6.6)
Site of the disease
Pulmonary	461 (89.2)
Extra-pulmonary	56 (10.8)
Comorbidity
No	345 (66.7)
Yes	172 (33.3)
HIV co-infection
No	83 (74.1)
Yes	134 (25.9)
Major adverse event
No	446 (86.3)
Yes	71 (13.7)
Complications
No	390 (75.4)
Yes	127 (24.6)
Type of resistance
Mono	315 (60.9)
MDR	163 (31.5)
Poly	39 (7.5)
Fluoroquinolone resistance
No	512 (99.0)
Yes	5 (1.0)
Baseline culture
Positive	368 (71.2)
Negative	149 (28.8)
Baseline sputum smear
Positive	316 (61.1)
Negative	201 (38.9)
Baseline anemia
No	270 (52.2)
Yes	247 (47.8)
Radiological abnormalities
Cavitary lesions	139 (26.9)
Non-cavitary lesions	291 (56.3)
Normal	87 (16.8)
Body mass index
<18.5kg/m^2	367 (71.0)
≥18.5kg/m^2	150 (29.0)

Figure 2 Treatment outcomes among MDR-TB patients in North West Ethiopia, September 2010 to July 2020.

Table 3 Multivariable Logistic Regression Analysis and Model Reduction Using Potential Predictors of Unfavorable Treatment Outcome in Patients with Multidrug-Resistant Tuberculosis, in North West Ethiopia, 2010–2020 (N = 517)

Predictors Selected by Lasso Algorithm	Multivariable Analysis		Model Reduction
	Coef (95% CI)	p-value	p-value
Anemia^ᵅ
No	0
Yes	1.08[0.6432, 1.5242]	<0.001***	0.0023**
Sex
Male	0
Female	−0.33 [−0.7959, 0.1265]	0.159181	0.3135
Age^ᵅ
< 45 years	0
≥45 years	0.82 [0.2146,1.4308]	0.01**	0.0082**
Residence
Urban	0
Rural	−0.17[−0.6518, 0.3109]	0.488892	0.162
Treatment supporter^ᵅ
Yes	0
No	0.97[0.4374, 1.5139]	<0.001	0.016*
Major adverse event^ᵅ
No	0
Yes	1 [0.4382, 1.5673]	<0.001	0.046*
Baseline smoking
No	0
Yes	0.71[−0.2957, 1.7147]	0.165743	0.2212
Comorbidity^ᵅ
No	0
Yes	0.71[0.2579, 1.1643]	0.003	0.105
Baseline sputum smear
Negative	0
Positive	0.32[−0.2772,0.9209]	0.297561	0.2606
Educational status
Literate	0
No formal education	0.47[−0.0218, 0.9702]	0.061392	0.9995
Baseline alcohol use
No	0
Yes	0.09[−0.4102, 0.6004]	0.698426	0.251
Marital statusᵅ
Married	0
Un married	0.75[0.2699,1.2509]	0.003	0.124
Intercept	−2.98 [−3.8193, −2.2016]	<0.001

Notes: ^ᵅVariables included in the final simplified model, ***P-value ≤0.001, **P-value ≤0.01, *P-value<0.05.

Abbreviations: Coef, coefficients; CI, confidence interval.

Figure 3 Nomogram for the prediction of unfavorable treatment outcome among MDR-TB patients in North West Ethiopia, September 2010 to July 2020.

Figure 4 Observed versus predicted unfavorable treatment outcome probabilities in a sample.

Table 4 Risk Classification of Unfavourable Treatment Outcome Using a Nomogram (n = 517)

Risk Category*	Unfavourable Treatment Outcome Prediction Nomogram
	Number of Patients (%)	Incidence of Unfavourable Treatment Outcome
Low (<0.298)	245 (47.4%)	35 (14.3%)
Intermediate (0.298–0.62)	225 (43.5%)	100 (44.4%)
High (≥0.62)	47 (9.1%)	22 (46.8%)
Total	517 (100%)	157 (30.4%)

Note: Risk category *The risk probability calculated using the nomogram.

Table 5 Sensitivity, Specificity, PPV, and NPV of the Nomogram for Several Cut-off Points

Cut-Off Points*	Sensitivity	Specificity	PPV	NPV	LR+	LR-
>0.32	59.2%	77.2%	53%	81.3%	2.6	0.53
>0.3	64.9%	71.7%	50%	82.4%	2.3	0.5
>0.298	73.9%	63.9%	47.2%	84.9%	2.1	0.4
>0.25	77.7%	58.3%	44.8%	85.7%	1.9	0.4
>0.24	82.2%	55.3%	44.5%	87.7%	1.8	0.3

Note: Cut-off points *Risk probabilities.

Abbreviations: PPV, positive predictive value; NPV, negative predictive value; LR+, likelihood ratio positive; LR-, likelihood ratio negative.

Figure 5 ROC curve of the model after internal validation using the bootstrapping method.

Figure 6 Decision curve plot showing the net benefit of the developed model for carrying out an intervention measure in MDR-TB patients at risk of unfavorable treatment outcome compared to all or none scenarios.