Self-assessment risk scoring system for early screening of preterm birth in low resource settings

Abstract

Introduction

Preterm birth is a leading cause of neonatal mortality and morbidity globally. According to the World Health Organization (WHO), 15 million babies were born prematurely in 2010. Indonesia has the fifth-highest number of preterm births in the world, with 675,700 occurring every year. Early screening for preterm birth is crucial, but many health centers in low-resource settings lack the equipment required to do so. This study aims to develop a scoring system that can predict preterm birth based on socio-demographics, nutrition, lifestyle, previous pregnancy history, and maternal condition. The goal is to create a simple, fast, and inexpensive method that can be easily accessed for early screening in low-resource settings.

Methods

This case-control study was conducted between June 2021 and June 2022 at Cipto Mangunkusumo General Hospital, Jakarta. Subjects were categorized into preterm and control groups. Clinical data were obtained through history taking and medical records and analyzed using IBM SPSS 23. Significant variables in bivariate analysis underwent multivariable analysis through logistic regression. The receiver operating characteristics (ROC) indicate how accurately the risk-scoring system performs. The validity of the scoring formula was analyzed using the Hosmer-Lemeshow test. The performance of the score was assessed using the area under the curve parameter.

Results

Multivariate analysis showed that socioeconomic level, antenatal visits, history of preterm, hypertension, and weight gain during pregnancy were independent predictors of preterm birth, with an area under the curve of 0.844 (95% CI 0.802–0.885) and p < 0.001, indicating a good predictive ability. Hosmer-Lemeshow test showed a well-fitted model (p = 0.26). Based on the scoring system in this study, the cut-off ≥ 4 had a sensitivity and specificity of 81.3% and 82.2%, respectively, in predicting preterm birth.

Conclusion

Socioeconomic level, history of preterm delivery, antenatal visits, weight gain during pregnancy, and history of hypertension were independent predictors associated with preterm birth.

Introduction

According to the World Health Organization (WHO) report in 2010, preterm births accounted for 1 in 10 live births, equivalent to 15 million cases worldwide. Indonesia has the highest preterm birth rate in Southeast Asia, accounting for 15.5% of all preterm births, which is approximately 675,700 per year [1–5]. Half of all deliveries at Cipto Mangunkusumo General Hospital (CMGH) are preterm [6, 7]. Preterm delivery results in significant mortality and morbidity of neonates, including 1 million deaths annually [4]. Organ immaturity leads to impaired physiological and biochemical functions that manifest as hypothermia, hypoglycemia, intracranial hemorrhage, infection, hyperbilirubinemia, electrolyte imbalance, acute respiratory distress syndrome (ARDS), hyaline membrane disease, necrotizing enterocolitis, cerebral palsy, retinopathy of prematurity, and death [8]. It is crucial to keep in mind the enduring effects that may arise in the future, such as psychological, emotional, and behavioral impairments [3, 5, 9] Neonatal intensive care unit (NICU) admission is a major contributor to the skyrocketing healthcare costs. By preventing preterm delivery, these healthcare costs can be trimmed [10].

Despite its significant number of cases, complications, and health burden, its etiology and pathophysiology still remain unclear. The etiologies for the underlying pathogenesis, which involves inflammation, hypothalamus-pituitary-adrenal (HPA) axis imbalance, decidual hemorrhage, and uterine stretch, have not been specifically identified [4, 5]. However, some conditions serve as the contributing risk factors: (1) maternal characteristics (extreme age, lower education level, lower socioeconomic status, smoking, stress during pregnancy, and higher occupancy demand); (2) nutritional characteristics (pre-pregnancy body mass index/BMI that is considered as obese or underweight, < 10 kg of body weight gain during pregnancy, diet, micronutrient supplementation, consumption of coffee, artificial sweeteners, and water intake); and (3) obstetric characteristics (hypertension or diabetes in pregnancy, interpregnancy interval, preterm history, twin pregnancies, antenatal visit, maternal infection, anemia, abnormally abundant amniotic fluid, vaginal bleeding in the first and second trimester, abnormal uterine morphology, in-vitro fertilization, and nausea-vomiting during the second and third trimester) [11–22].

Predicting preterm delivery is critical for appropriate management using a simple, fast, and inexpensive instrument. This enables earlier screening from primary care. Salunkhe et al. [12] formula cannot be widely used since it does not consider nutrition and obstetric characteristics, and cannot be used by pregnant women. Indonesia is an archipelagic country, where midwives conduct 84.8% of antenatal care visits [23]. Early detection for preterm delivery screening is crucial to prevent delayed referral and improve outcomes. This research offers substantial benefits for obstetric care in low-resource settings, where access to advanced equipment for preterm birth prediction is limited. By developing a simple, low-cost, and easily applicable scoring system, this study provides a practical solution that supports early identification of preterm birth risk, enables timely referral, and enhances preventive care at primary healthcare levels. The system is based on a combination of maternal, obstetric, and nutritional characteristics that are recorded during history taking, physical exams, and supporting exams. This scoring system is designed to be used in limited-resource settings by healthcare providers and even pregnant women themselves.

Methods

Study design and setting

This case-control study was conducted between June 2021 and July 2022 at Cipto Mangunkusumo National Hospital, Jakarta. Data were obtained from medical records and standardized data collection forms containing sociodemographic, obstetric, nutritional, and maternal medical information. Following development of the scoring system, external validation was subsequently performed.

Study sampling

As this was an unmatched comparative study, we calculated the sample size using a rule of thumb based on event per variable. Pregnant women on the 22^+ 0- 36^+ 6 weeks of gestation (preterm groups) and ≥ 37^+ 0 weeks (control groups) who delivered live births were included. Participants were excluded if there were any congenital fetal anomalies. The ethical committee of the Faculty of Medicine at Universitas Indonesia has approved this study.

Outcome measures

A prepared form was used to record various data, which were grouped into maternal, nutritional, and obstetric characteristics. The independent variables were further classified into five domains, namely sociodemographic, obstetric characteristics, nutritional factors, lifestyle, and maternal medical condition, while delivery outcomes such as preterm or term were used as dependent variables. Sociodemographic characteristics included maternal age, education, socioeconomic status, and high occupancy demand (> 40 h/week or night shift). The obstetric characteristics recorded were interpregnancy interval, history of preterm birth, twin pregnancy, antenatal visit, abnormality of amniotic fluid, abnormal uterine morphology, and pregnancies via in vitro fertilization. Nutritional factors included pre-pregnancy BMI based on the WHO Asian BMI Classification, weight gain during pregnancy using IOM guidelines recommendation, diet, and micronutrient supplementation. Lifestyle factors included coffee and artificial sweeteners intake, water intake, and cigarette smoking. Maternal medical factors included diabetes mellitus, hypertension, infection, anemia, vaginal bleeding during the first and second trimesters, surgery history, and nausea and vomiting during the second and third trimesters.

Statistical analysis

Data was analyzed for normality using Kolmogorov-Smirnov. Bivariate analysis of categorical data was done for each independent variable using Chi-square or Fisher’s exact test. Significant variables in bivariate analysis were included for multivariate analysis through binary logistic (method: backward regression). Data were analyzed using IBM SPSS 23.0 with a confidence interval of 95%, and a p-value of < 0.05 was considered significant. The scoring system was developed through four steps: 1) dividing each prognostic factor’s coefficient B by its standard error (coefficient B/SE); (2) choosing the lowest B/SE value as a reference (3) dividing each B/SE value by the reference value; and (4) picking the rounded number nearest to the result from step 3. The validity of the scoring formula was analyzed using the Hosmer-Lemeshow test. Using the receiver operating characteristics (ROC), we tested our formula performance on predicting preterm delivery occurrence, sensitivity, and specificity.

Results

Study population’s characteristics

The study included 455 subjects, consisting of 226 preterm cases and 229 term controls. There were no significant differences in age, height, BMI, and parity status between groups. Table 1. shows the detailed subject characteristics across groups and detailed bivariate analysis results.

Table 1.

Demographic characteristics associated with preterm delivery on subjects

Variables	Preterm, n (%)	Term, n (%)	OR (95% CI)	P-value
Sociodemographic factors
Age groups			1.15 (0.71–1.88)	0.57c
35 y.o.*	170 (75.3)	190 (83)
>35 y.o	56 24.7	39 (17)
Socioeconomic status			2.24 (1.46–3.42)	< 0.001c
Low- Middle	132 (58.4)	96 (41.9)
High-Very high	94 (41.6)	133 (58.1)
Education			1.08 (0.75–1.56)	0.72c
12 years	120 (53.1)	117 (51.1)
>12 years	106 (46.9)	112 (48.9)
High occupancy demand			1.43 (0.93–2.20)	0.11c
Yes	67 (29.6)	65 (28.4)
No	159 (70.4)	164 (71.6)
Obstetric factors			0.76 (0.37–1.55)	0.45c
Interpregnancy interval
Never pregnant
<18 months	92 (40.7)	94 (41.0)
18 months	71 (59.3)	135 (59.0)
Preterm delivery history			37.57 (20.28–69.59)	< 0.001c
Yes	16 (7.1)	1 (0.4)
No	210 (92.9)	228 (99.6)
Num. of antenatal visit			2.70 (1.5–4.84)	0.001c
< 7	74 (32.7)	35 (15.3)
7	152 (67.3)	194 (84.7)
Twin pregnancy			4.26 (1.1-15.36)	0.02c
Yes	15 (6.6)	5 (2.2)
No	211 (93.4)	224 (97.8)
Abnormal amniotic fluid			1.26 (0.83–1.91)	0.24 c
Yes	166 (73.5)	178 (77.7)
No	60 (26.1)	51 (22.3)
Nutritional factors
Balance diet			1.06 (0.56–2.01)	0.84c
Yes	26 (11.5)	22 (9.6)
No	200 (88.5)	207 (90.4)
Micronutrient supplementation			0.93 (0.49–1.77)	0.86c
Yes	205 (90.7)	209 (91.3)
No	27 (9.3)	20 (8.7)
Maternal weight gain			2.25 (1.40–3.60)	0.001c
Less	86 (38.1)	64 (27.9)
Normal-Over	152 (67.2)	165 (72.1)
Underweight			3.95 (1.65–9.45)	0.001c
Yes	25 (11.1)	7 (3.1)
No	201 (88.9)	222 (96.9)
Lifestyle factors
Coffee			0.96 (0.94–0.99)	0.01c
Yes	3 (1.3)	4 (1.7)
No	223 (98.7)	225 (98.3)
Adequate water intake			0.72 (0.43–1.21)	0.22c
Yes	183 (81)	181 (79)
No	43 (19)	48 (21)
Cigarette smoking
Yes	0	0	N/A	N/A
No	226 (100)	229 (100)
Maternal medical factorsc
Diabetes mellitus			0.48 (0.17–1.33)	0.15c
Yes	13 (5.8)	10 (4.4)
No	213 (94.2)	219 (95.6)
Hypertension			2.61 (1.41–4.82)	0.001c
Yes	28 (12.4)	18 (7.9)
No	198 (87.6)	211 (92.1)
Infection during pregnancy			1.38 (0.80–2.38)	0.23c
Yes	184 (81.4)	188 (82.1)
No	42 (18.6)	41 (17.9)
Anemia			1.16 (0.63–2.13)	0.62c
Yes	35 (15.5)	46 (20.1)
No	191 (84.5)	183 (79.9)
1st and 2nd trimester vaginal bleeding			2.12 (0.88–5.09)	0.09c
Yes	26 (11.5)	10 (4.4)
No	200 (88.5)	219 (95.6)
Surgery history			N/A	N/A
Yes	0	0
No	226 (100)	229 (100)
2nd and 3rd trimester nausea-vomiting			1.22 (0.78–1.90)	0.39c
Yes	75 (33.2)	74 (32.3)
No	151 (66.8)	155 (66.7)

^cChi Square test

Bivariate analysis of sociodemographic, obstetric, nutritional, and medical factors

In the sociodemographic domain, only socioeconomic status (OR = 2.24, 95% CI: 1.46–3.42; p < 0.001) was significantly related to preterm birth. Prior history of preterm birth (OR = 37.57, 95% CI: 20.28–69.59; p < 0.001) had a significant effect on preterm birth. Antenatal visits (OR = 2.70, 95% CI: 1.5–4.84; p = 0.001) and multiple pregnancies (OR = 4.26, 95% CI: 1.1–15.36; p = 0.02) were considered significant contributing factors to preterm birth. Consumption of micronutrient supplementation (OR 0.93, 95% CI 0.49–1.77) and a balanced diet (OR = 1.06, 95% CI: 0.56–2.01) reduced preterm birth, although not statistically significant, while coffee intake was significantly related to preterm birth (OR = 0.96, 95% CI: 0.94–0.99; p = 0.01). Higher preterm birth was significantly associated with low weight gain during pregnancy (OR = 2.25, 95% CI: 1.40–3.60; p = 0.001) and low pre-pregnancy body mass index (OR 3.95, 95% CI 1.65–9.45; p = 0.001). In the maternal medical domain, it was found that the presence of hypertension (OR = 2.6, 95% CI: 1.41–4.82; p = 0.001) was a significant factor contributing to preterm birth.

The summary of the significant variables that were included in the multivariate analysis is portrayed in Table 2. For simplicity to develop the scoring system, we regroup several variables, such as socioeconomic status and body weight (BW) gain during pregnancy, into dichotomous variables.

Table 2.

The odds ratio of significant variables included in multivariate analysis

Variables	OR (95% CI)	p
Low-Middle Socioeconomic status	2.8 (1.21–6.45)	0.016
< 7 ANC antenatal visit	1.97 (1.02–4.49)	0.043
History preterm birth	48.70 (23-100.1)	0.001
Twin pregnancy	2.37 (0.98–5.33)	0.057
Coffee	2.12 (0.83–5.4)	0.114
Inadequate weight gain	2.99 (1.46–6.08)	0.03
Underweight	1.37 (0.43–4.33)	0.58
Hypertension	2.40 (1.95–6.1)	0.04

Multivariate analysis and development of the preterm delivery scoring system

The scoring formula was derived from Table 3, which had been assessed using the Hosmer-Lemeshow test (p = 0.26). Each “Yes” response to predictor factor scored 1, while “No” scored 0. Socioeconomic factors scored 2, and preterm history scored 5 for each “Yes” response. The score itself was acquired by calculating B/SE in each significant variable, which was then divided by the lowest B/SE number, and then rounded to the closest number. Thus, a total score of 13 in the scoring system was found. The calculation of the preterm delivery probability, Eq. 1 and Eq. 2 were used to calculate the value of y, then used to calculate the probability. Using the aforementioned formula (see Methods: statistical analysis), we determined the score for each significant variable for the preterm delivery prediction system as presented in Table 4.

Table 3.

Development of the preterm birth risk scoring system based on multivariable logistic regression

	Coeff. B	S.E.	Wald	p-value	OR (95% CI)	Scoring
Low-Middle Socioeconomic status	1.14	0.334	11.631	0.001	3.1 (1.6-6.021)	2
Less ANC duration	0.657	0.432	2.307	0.045	1.92 (1.02-4.49)	1
Preterm delivery history	3.832	0.357	114.941	<0.001	46.166 (22.91-93.04)	7
Inadequate weight gain	1.095	0.358	9.364	0.002	2.98 (1.48-6.02)	2
Hypertension	0.962	0.46	4.379	0.036	2.61 (1.06-6.44)	1

Table 4.

Scoring for preterm delivery prediction

Variables		Score
Socioeconomic status	Low-Middle	2
	High-Very high	0
Antenatal visit	< 7	1
	≥ 7	0
Preterm delivery history	Yes	7
	No	0
Hypertension	Yes	1
	No	0
BW gain	Less	2
	Adequate-over	0
Total score		13

Analysis using the ROC curve (Fig. 1) was done to determine the performance of our formula on predicting preterm delivery with AUC 0.844 (95% CI 0.802–0.885). A cut-off value of ≥ 4 gave the optimal sensitivity (81.3%) and specificity (82.2%) with 53.66% of preterm delivery probability. Detailed relationships between different cut-off points, sensitivity, and specificity, and probability are presented in Table 5.

Fig. 1.

ROC curve of scoring formula on various cut-offs; AUC 0.844 (95% CI 0.802–0.885)

Table 5.

Sensitivity, specificity, and probability analysis of the scoring system

Category	Total score	Sensitivity (%)	Specificity (%)	Probability (%)
Low risk	0	100	0
	1	98.4	38.3	21.58
	2	89.6	46.7	35.14
	3	86.8	76.1	41.13
High risk	4	81.3	82.2	53.66
	5	79.7	87.8	69.41
	6	62.6	90.6	84,21
	8	53.8	95	91.81
	9	31.3	96.1	96.91
	10	17	97.8	98.72
	11	8.2	98.9	99.48
	12	1.1	99.4	99.79
	13	0	100	99.92

External validation

We enrolled 139 subjects for external validation of the scoring tool. Of those, 68 had preterm delivery and 71 had term delivery. The performance of the preterm delivery risk scoring tool was assessed for calibration and discrimination results. As the calibration plot showed a coefficient of r = 0.911 (Fig. 2) and the Hosmer–Lemeshow test showed p = 0.464, the scoring system was considered to have good calibration. In Fig. 3, The AUC was 0.819, with 95% CI (0.746–0.891), showing excellent discrimination with no significant difference from the first part of the study.

Fig. 2.

The plot calibration diagram of the scoring system in external validation

Fig. 3.

The ROC curve of the scoring system in external validation

Discussion

According to our study, several variables were found to be significantly associated with higher preterm delivery rates. These variables include socioeconomic status, history of preterm delivery, number of antenatal visits, twin pregnancy, low body weight gain during pregnancy, coffee consumption, low BMI prior to pregnancy, and hypertension. No significant association was found between the adequacy (categorical) or amount (numerical) of various macronutrient and micronutrient types with preterm delivery. However, after adjusting the socioeconomic status, preterm birth history, number of antenatal visits, body weight gain during pregnancy, and hypertension, only 5 factors: (1) middle socioeconomic status, (2) preterm history; 3) < 7 antenatal visits, 4) low body weight gain during pregnancy, and 5) hypertension served as independent predictors of preterm delivery.

Our findings both confirmed and differed from previous reports. The association between maternal age and prematurity differed from Fuchs et al. [24]. In their study, the 30–34 age group had lower prematurity rates while the 20–24 age group had higher rates. Lawlor et al. [25] reported a “U-shaped” relation where the 24–30 age group had the lowest risk of preterm birth. A study in Finland determined the cut-off age of 28 years old as the threshold for increased risk of preterm delivery [26]. This inconsistency may be due to different inclusion criteria. The results of our study regarding pre-pregnancy BMI differ from those of Kosa et al. [27] who found a J-shaped association between BMI and preterm birth, with a BMI up to 24 kg/m2 being protective against preterm birth. Gao et al. [28] also discovered a significant increase in the risk of preterm delivery among overweight or obese mothers.

Whitehead et al. [29] found that lower socioeconomic status increases the risk of premature contraction, but not preterm delivery. Although the direct relation has not been established, Kramer et al. [30] found that bacterial vaginosis incidence is higher among those with lower socioeconomic status, cigarette smokers, and those with high physical occupancy demand. The latter was confirmed by Styliano-Riga et al. [31] as a risk factor for preterm birth along with maternal emotional stress. Our research findings concerning maternal education levels closely aligned with the study conducted by Facchinetti et al. [32], as well as with the comprehensive meta-analysis undertaken by Ruiz et al. [33] in 2015. This meta-analysis included data from 75,000 live births in Europe. The correlation between higher education and accessing superior healthcare and information was apparent.

McManemy et al. [34] reported that the risk of recurrent preterm birth increased by 33–57%, depending on the gestational age of the previous preterm delivery. Similarly, Pervin et al. [35] demonstrated a dose-dependent relationship between the number of antenatal visits and the risk of preterm birth. These findings are consistent with our results, in which both previous preterm birth and fewer antenatal visits remained independent predictors in the final scoring model. Furthermore, although previous obstetric history is not applicable to nulliparous women, it remains one of the strongest predictors of recurrent preterm birth among multiparous women [35].

Twin pregnancy was significantly associated with preterm birth in the bivariate analysis, consistent with previous reports showing substantially increased rates of preterm delivery among multifetal pregnancies [13, 34]. However, this variable was not retained in the final multivariable model after adjustment for other maternal and obstetric factors, suggesting that its predictive contribution may overlap with other variables included in the scoring system.

It is noteworthy that we found no correlation between macronutrient and micronutrient components and preterm delivery. The use of Food Frequency Questionnaire (FFQ) and food recall might be the source of some inaccuracies due to recall bias. Studies have shown that taking prenatal supplements, specifically folate and iron, 12 weeks before gestation can decrease the likelihood of preterm birth [36–38]. Insufficient folic acid serum levels could lead to abnormal T cell and neutrophil function, increased bacteriuria, and inflammatory markers in pregnancy [39, 40]. The addition of essential fatty acid supplementation, specifically in the form of fish oil, can regulate prostaglandin and cytokine production, reducing inflammation [41, 42]. A Cochrane review of 6 RCTs showed a reduction of preterm birth up to 46% (OR 0.54; 95% CI 0.30–0.98) using omega-3 fatty acid supplementation [43]. Another crucial micronutrient supplement is vitamin D, known for its role in modulating inflammation [44]. Aside from micronutrient supplementation, a gross parameter of adequate nutrition is body weight gain. A lower preterm rate was observed in adequate body weight gain, especially in the second and third trimesters, with a variation among different pre-pregnancy baseline BMI levels [45–47].

Recognition of hypertension’s role in preterm delivery has increased in the past two decades due to its links to fetal growth restriction and endothelial dysfunction. Premkumar et al. [48] and Catov et al. [49] conducted studies that demonstrate an increased risk of preterm delivery among mothers with hypertension. The study by Premkumar et al. [48] indicates risks 2.9 to 8.5 times higher, while the study by Catov et al. [49] shows a risk 5.5 times greater. Moreover, Berger et al. [50], in their investigation involving 500,000 pregnant women, uncovered a significantly heightened occurrence of preterm delivery within the context of preeclampsia (adjusted relative risk 45.42; 95% CI 39.69–51.99).

A comprehensive yet specific scoring system

The multifactorial etiology nature of preterm delivery makes it difficult to predict. A previous study by Tekesin et al. [51] produced CLEOPATRA I and II risk stratification with AUC of 0.811 and 0.692, respectively. Unfortunately, the usage of fetal fibronectin and cervical length is not widely applicable in Indonesia. Another scoring system by Salunkhe et al. [12] that includes high occupancy demand, < 4 portions of eating/day, < 2 h of sleep/day, education level, body weight gain less than10 kg, and < 4 antenatal visits have a sensitivity and specificity of 75.5% and 85.4% respectively (cut off ≥ 8). Beck et al. [52] introduced a graded scoring system that stratifies preterm risk according to the cumulative score assigned. Comparatively, our model outperforms this approach with an AUC of 0.844. Despite the favorable discrimination performance, the predictive value of the scoring system should be interpreted cautiously because the present study used a case–control design with an enriched proportion of preterm births. Consequently, positive and negative predictive values observed in routine obstetric populations may differ from those estimated in this study.

It is also important to recognize that multiple risk factors increase the likelihood of preterm delivery. For a scoring system to be effective, it must be accurate across diverse populations, including nulliparous and multiparous individuals, as well as varying gestational ages. It also should be able to differentiate between spontaneous and medically indicated preterm deliveries while also predicting the degree of prematurity. Thus, a practical scoring system is one that is favored for routine implementation [15].

Limitation

This study has several limitations. The case–control design and recruitment from referral hospitals may limit generalizability to primary or rural settings. Although the scoring system demonstrated good discrimination, caution is warranted when interpreting its predictive performance. Because this study employed a case–control design with an artificially enriched proportion of preterm births, the observed prevalence does not reflect that of the general obstetric population. While sensitivity and specificity are not affected by disease prevalence, positive and negative predictive values may differ substantially when the model is applied in routine clinical settings. Therefore, further validation in prospective cohort studies with population-based prevalence is required before the predictive performance of this scoring system can be fully established. Additional multicenter investigations with enhanced designs and larger sample sizes are needed to establish the findings of this case-control study, which was carried out at a national referral hospital. An adequate follow-up duration is crucial for accurately evaluating the effectiveness of risk factor identification in predicting preterm delivery.

Conclusion

This study successfully developed a simple self-administered scoring system for early screening of preterm birth in low-resource settings. The scoring tool incorporates five independent predictors, socioeconomic level, history of preterm birth, antenatal visit frequency, pregnancy weight gain, and hypertension, yielding strong predictive performance (AUC 0.844) with an optimal cut-off of ≥ 4.

This scoring system provides a practical and inexpensive method that can be used by midwives and pregnant women to identify high-risk individuals earlier, supporting timely referral and improving maternal–neonatal outcomes. Its implementation has the potential to shift clinical practice by enabling widespread, community-level preterm birth risk assessment without the need for specialized equipment.

Authors' contributions

All authors contributed to the study’s conception and design, and approved the final manuscript. LN, YBS, RI designed study and validate the data. LN, RH conducted the research, performed the analysis, interpreted of data. LN, YBS, RH, RI wrote revised and finalized the manuscript.

Funding

This research received no funding.

Data availability

The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

The study was approved by the Ethical Committee of the Faculty of Medicine Universitas Indonesia (Letter number S. 71/SPE/IX/2022). This was a retrospective study, thus only clinically relevant information from medical records were restricted. Written informed consent from all participants was waived by the Ethical Committee as this study also did not affect patients’ treatment and health.

Consent for publication

No consent was taken from patients during their enrollment into the study.

Competing interests

The authors declare no competing interests.