External validation of the HANDOC score in a Spanish cohort—Analysis of blood culture scoring and time to positivity

Abstract

Introduction

Bacteremia caused by gram-positive cocci (GPC) remains challenging, particularly in assessing the risk of infective endocarditis (IE). Various scoring systems have been developed to guide the use of echocardiography. The HANDOC score was specifically designed for non-β-hemolytic streptococci (NBHS). This study aimed to validate the HANDOC score in a cohort across diverse geographic settings.

Methods

A retrospective study enrolled patients with NBHS bacteremia from 2017 to 2021 at two Spanish hospitals. Cases of IE were defined according to European Society of Cardiology 2015 modified Duke criteria. Patient characteristics were extracted from medical records for the analysis of HANDOC score validation in our cohort.

Results

Among 280 patients diagnosed with NBHS bacteremia, 31 met the modified Duke criteria for infective endocarditis (11.1%). Using a cutoff of ≥3, the HANDOC score demonstrated a sensitivity of 95%, specificity of 74% and a negative predictive value of 98%. The same metrics were analyzed with an adapted score based on positive blood culture vials, yielding similar results. Time to positivity (TTP) was analyzed with different cutoffs or by each NBHS group showing no statistically significant difference.

Conclusions

The HANDOC score is a valuable tool for decision-making in NBHS bacteriemia in a Spanish cohort. Scoring by vials may be employed for blood culture item in different clinical settings. Time-to-positivity did not show a significant difference that would justify its potential inclusion in the score.

Resumen

Introducción

La bacteriemia causada por cocos grampositivos sigue siendo un desafío para la evaluación del riesgo de endocarditis infecciosa (EI). Se han desarrollado varias escalas de puntuación para guiar el uso de la ecocardiografía. La escala o score HANDOC fue diseñada específicamente para estreptococos no β-hemolíticos (ENBH). Este estudio tiene como objetivo validar el score HANDOC en una cohorte con un emplazamiento geográfico diferente.

Métodos

Estudio retrospectivo que incluye a pacientes con bacteriemia por ENBH entre 2017 y 2021 en dos hospitales en España. Los casos de EI se definieron según los criterios modificados de Duke por la Sociedad Europea de Cardiología en 2015. Las características de los pacientes se extrajeron de las historias clínicas para el análisis.

Resultados

De los 280 pacientes diagnosticados con bacteriemia por ENBH, 31 cumplieron con los criterios de Duke modificados para EI (11,1%). Utilizando un punto de corte de ≥3, el score HANDOC demostró una sensibilidad del 95%, una especificidad del 74% y un valor predictivo negativo del 98%. Se analizaron los mismos valores con una puntuación adaptada en base a viales de hemocultivos positivos, siendo similares. El tiempo de positivización de hemocultivos con diferentes puntos de corte y para cada grupo de estreptococos no demostró diferencias estadísticamente significativas.

Conclusiones

El score HANDOC es una herramienta útil para la toma de decisiones en bacteriemia por ENBH en una cohorte española. La puntuación de los hemocultivos en el score puede adaptarse a diferentes situaciones clínicas. El tiempo de positivización no mostró una diferencia significativa como para valorar de momento una posible introducción en el score.

Introduction

Gram-positive cocci (GPC) remain the main microorganisms causing infective endocarditis (IE).^1,2 However, not all patients with GPC bacteremia are equally at risk of developing IE.^3,4 Echocardiography is the preferred method for diagnosing IE. Transesophageal echocardiography (TEE) has higher sensitivity compared to transthoracic echocardiography (TTE). However, TEE is a semi-invasive and resource-demanding procedure, limiting its availability in all hospitals. In recent years, TTE has been increasingly used in GPC bacteremia to exclude IE due to heightened clinical awareness. Nevertheless, sometimes the demand is unaffordable, potentially leading to underdiagnosis or delayed identification of other high-risk clinical situations.

Risk stratification scores have been developed to aid clinical decisions regarding the necessity of undergoing echocardiography in GPC bacteremia. Currently, the PREDICT,^5,6 VIRSTA,^7,8 and POSITIVE⁹ scores have been developed for Staphylococcus aureus bacteremia, while the NOVA^10,11 and DENOVA scores¹² were designed for Enterococcus spp. bacteremia.

The HANDOC score¹³ (2018) (Fig. 1), developed specifically for non-beta-hemolytic streptococci (NBHS) bacteremia with a cutoff point ≥ 3 points, achieved a sensitivity (Se) of 100% (95% confidence interval [CI], 88–100%) and a specificity (Sp) of 73% (95% CI, 67–88%), with an area under the receiver operating characteristic (ROC) curve of 0.96.

Fig. 1.

HANDOC score items. Adapted from Sunnerhagen T.¹⁰ Validated cutoff point: 3 points.

Despite these promising initial results and subsequent validations in Northern European cohorts,^14–16 these scores are not widely adopted in routine clinical practice. Furthermore, HANDOC score had not been validated in different geographical settings until recently.¹⁷ The primary objective of this study is to externally validate the HANDOC score in a Spanish cohort.

Methods

Study design and participants

A retrospective observational study was conducted on all episodes of non-beta-hemolytic streptococci (NBHS) bacteremia collected from two secondary-level hospital centers in southern Spain: Hospital Clínico San Cecilio in Granada (with 540 beds) and Hospital Vega Baja in Alicante (with 330 beds). The study protocol was approved by the ethics committee at both hospitals (TF-Strepto-2022). Consecutive streptococci bacteremia cases from January 1, 2017 to December 31, 2021, were identified using the microbiology laboratory records of both hospitals (IntraLab® and GestLab®).

Patients aged 16 years and older were included, and clinical, epidemiological, microbiological and outcome variables were collected. Comorbidity was assessed using Charlson's comorbidity index (CCI).¹⁸ A one-year virtual follow-up of medical records was established to monitor mortality and identify potential delayed diagnoses of IE.

Streptococci were initially identified in both hospitals by blood culture system: BD BACTEC™ FX. Most positive blood cultures were analyzed using MALDI-TOF MS (matrix-assisted laser desorption ionization time-of-flight mass spectrometry) specifically with MALDI Biotyper® sirius System.

Episodes of beta-hemolytic streptococci bacteremia were excluded according to the HANDOC score criteria, as episodes of S. pneumoniae bacteremia due to its low incidence of IE.¹⁹ Additionally, episodes without specific identification of streptococcal species were excluded, unless categorized as ‘Streptococcusviridans’, which were included as such.

Patients whose medical records were unavailable or who were terminally ill with hospital stays < 72 h and managed conservatively (without antibiotics) were also excluded, unless there was clinical suspicion of IE.

Definitions

IE was defined according to the 2015 modified Duke criteria by ESC (European Society of Cardiology).²⁰ The ‘Definite IE’ group was categorized based on major and/or minor criteria. Patients were classified into the ‘No endocarditis’ (or ‘No IE’) group if they did not meet the 2015 Duke-ESC criteria or the criteria established in the initial HANDOC score cohort¹³; this included cases where transesophageal echocardiography (TEE) showed no signs of endocarditis, or patients received less than 14 days of intravenous antibiotics and survived for at least six months without bacteremia relapse. No autopsies were performed. Patients in whom IE could not be confirmed or ruled out by 2015 Duke-ESC criteria or the aforementioned criteria were categorized as ‘Possible IE’, corresponding to the ‘Unknown group’ in previous articles.^13,15

The 2023 modified Duke-ESC criteria²¹ and the 2023 modified Duke-ISCVID (International Society for Cardiovascular Infectious Diseases) criteria²² were not applied because, at the time clinicians categorized cases as ‘Definite IE’ or ‘Possible IE’, new criteria had not been published yet. Moreover, the HANDOC score was performed according to the 2015 Duke-ESC definitions.

Bacteremia was defined as the isolation of streptococci from any vial of blood culture (BC) sets. Most BC culture collections consist of two sets, each containing two vials or flasks. However, seriously ill patients may occasionally have more vials collected as part of a single sample. ‘Repeated BC’ sets are considered those which were drawn spanning > 12 h up to 72 h, with positive results considered ‘Persistent BC’.

In our cohort, to apply the HANDOC score variable ‘Number of positive blood cultures ≥ 2’, we defined positivity according to the HANDOC score definition and subsequent validation cohorts: scoring requires 2 or more different positive blood cultures sets of the same sample collection. In a subsequent analysis, we propose adapting this criterion to account for the positivity at least two positive vials, regardless of whether they were obtained from a single or multiple BC sets, in order to enhance its applicability to different clinical settings.

TTP (Time to Positivity) is defined as the number of hours required for streptococci to grow in blood cultures from the moment they were introduced into the blood culture system (BACTEC™ FX), as identified by microbiology services.

Only information available at the time of blood culture result receipt was used to score each patient. None of the scores were applied in clinical practice for any of our patients, as they are not routinely used in our centers. The score's cutoff point was established as 3 points based on the original analysis.

Statistics

A baseline descriptive analysis was conducted on the characteristics of NBHS bacteremia. The Shapiro–Wilk test was used to assess the normality of quantitative variables, revealing that all variables were non-normally distributed. A p-value < 0.05 was considered statistically significant.

Secondly, a Chi-squared test was performed on qualitative dichotomous variables to calculate univariate odds ratios between ‘No IE’ and ‘Definite IE’ groups only, and the non-parametric Mann–Whitney U test was employed to compare independent samples. Kruskal–Wallis test was performed for comparing median TTP for species within the NBHS group. Due to the non-normal distribution and the corresponding residuals for the regression, linear regression could not be performed. Therefore, TTP was log-transformed to conduct a logistic regression, with ln(TTP) as the outcome variable, using different cutoff values.

For external validation of the HANDOC score, the ‘Possible IE’ group was combined with the ‘Definite IE’ group, as patients in the former group typically require ruling out IE with at least one echocardiographic study in clinical practice. ROC curve analysis was conducted, and sensitivity, specificity, positive predictive value, negative predictive value, and AUC (area under the curve) were calculated. All analyses for both stages were performed using IBM SPSS Statistics Version 25, and the ROC curve analysis was conducted using R-software® 2021.²³

Results

Socio-demographical, clinical and microbiological characteristics

A total of 631 episodes of streptococcal bacteremia were recorded, out of which 280 NBHS met the inclusion criteria for in the study (Fig. 2). The mean number of NBHS positive blood cultures per year was 56 ± 1.37. The prevalence of NBHS IE was 11.1%. The mean age was 70.6 years with a standard deviation of ± 14.9 years. Patients had a median Charlson's comorbidity index (CCI) of 5 points, with 70% having a CCI ≥ 4 points. At least 52 patients (18.6%) had a history of cardiac disorders, with 15% of these being of valvular origin. Additional clinical and epidemiological characteristics of the cohort are detailed in Supplementary Material Appendix 1-1.

Fig. 2.

Flow chart of the exclusion criteria and final results.

Regarding microbiological data, 55 episodes (19.6%) were identified as S. viridans. Streptococcus mitis group (32.9%) was the most commonly reported specific species and, along with Streptococcus gallolyticus group, the most frequent causative agents of IE (45.2% and 25.8% each one). Four out of eight isolates of Streptococcus sanguinis were involved in IE.

In our cohort, in a considerable portion of our patients (98 out of 280 patients – 35%) only one set of blood cultures had been collected at the initial assessment, amounting to 2 vials. However, repeated BC sets were obtained in 117 patients. Other microbiological details are provided in Supplementary Material Appendices 1-2 and 1-3.

Regarding the echocardiographic study, in the total sample, 36.8% underwent TTE and 16.4% underwent TEE. In the ‘Definite IE’ group, 100% underwent TTE and 90.3% underwent TEE. In the ‘Possible IE’ group, 88.2% underwent TTE and 70.6% underwent TEE. Number needed to screen to detect 1 case of IE was 2.2. Among all IE cases, 19 (61%) were detected by TTE. Additionally, 11 cases (35.4%) were identified by TEE. All TEEs were preceded by TTEs. The median time until echocardiography was performed was 6 days for the ‘No endocarditis’ group and 4 days for ‘Possible IE’ and ‘Definitive IE’.

Detailed information is provided in Supplementary Material Appendix 2.

Risk factors for IE

Results comparing ‘No endocarditis’ and ‘Definitive IE’ groups are presented in Table 1. Our study introduced several new risk variables analyzed in the comparison between ‘No IE’ and ‘Definitive IE’, in addition to those already present in the modified Duke criteria and HANDOC score: Variables such as low-grade fever (>37.2 °C) for 3 days despite appropriate antibiotic treatment and persistent positivity of BC were statistically associated with ‘Definitive IE’. CRP levels (mean and >100 mg/dL for 3 days) did not show statistically significant differences, along with penicillin–susceptible strains (categorized as susceptible or MIC < 0.125 or 0.125–0.250).

Table 1.

Results of the comparative analysis on patients with and without endocarditis, excluding possible or suspected cases.

	No endocarditis 232 N° (%)	Definite IE31 N° (%)	OR (Odds ratio)	P valor
Age (mean)	70.65 ± 15.15	68.69 ± 13.56		0.397
Charlson (mean)	4.72 ± 2.38	4.16 ± 2.63		0.166
Charlson ≥ 4	164 (70.7)	18 (58.1)	0.574 [0.267; 1.237]	0.078
Men	137 (59.1)	20 (64.5)	0.793 [0.363; 1.732]	0.560
Women	95 (40.9)	11 (35.5)
High risk of heart disease
Prior endocarditis	1 (0.4)	0	-	0.714
Diseased native valve	20 (8.6)	10 (32.3)	5.047 [2.090,12.189]	<0.001
Prosthetic valve	4 (1.7)	8 (25.8)	19.826 [5.543; 70.918]	<0.001
Cardiac device	4 (1.7)	3 (9.7)	6.107 [1.299; 28.703]	0.011
Any	23 (9.9)	13 (41.9)	6.562 [2.852; 15.1]	<0.001
Acquisition
Community	167 (72.0)	30 (96.8)	11.676 [1.560; 87.396]	0.008
Nosocomial	55 (23.7)	1 (3.2)
Healthcare-associated	10 (4.3)	0
Source of bacteriemia
Otorhinolaryngology	4 (1.7)	1 (3.2)
Respiratory	69 (29.7)	0
Articular / Trauma	7 (3.0)	0
Abdominal in general	73 (31.4)	0
Urinary	19 (8.2)	0
Cutaneous	12 (5.2)	0
Catheter	10 (4.3)	0
Stroke / Heart attack	7 (3.1)	0
Meningitis	3 (1.3)	0
Unknown	28 (12.1)	28 (90.3)	68 [19.395; 238.408]	<0.001
Species
S. viridans	50 (21.5)	2 (6.4)
S.mitis	72 (31.0)	14 (45.2)	1.793 [0.839; 3.834]	0.066
S.bovis	25 (10.)	8 (25.8)	3.014 [1.215; 7.479]	0.008
S.anginosus	60 (25.9)	4 (12.9)	0.424 [0.143; 1.264]	0.061
S.salivarius	19 (8.2)	2 (6.5)	0.773 [0.171; 3.492]	0.369
S.mutans	6 (2.6)	1 (3.2)	1.255 [0.146; 10.789]	0.418
MIC Sensible(n=228)^a	139/180 (77.2)	28/31 (90.3)	2.753 [0.796;9.518]	0.054
Other streptococcus	10 (4.3)	0 (0)	-	-
Other microorganism	79 (34.1)	3 (9.7)	0.207 [0.061; 0.704]	0.005
Persistent BC	11 (4.7)	17 (54.8)	24.396 [9.615; 61.901]	<0.001
Temp. > 37.2&#¿; for 3 days	40 (17.3)	22 (71.0)	11.733 [5.030; 27.371]	<0.001
Sepsis	39 (16.8)	5 (16.1)	0.951 [0.344; 2.631]	0.924
PCR (mean)	151.2 ± 97.57	114.5 ± 74.71		0.047
PCR > 100mg/dl for 3 days(n=257)*	92/210 (43.8)	10/30 (30.0)	0.641 [0.286; 1.437]	0.280
Murmur	15 (6.5)	21 (67.7)	30.38 [12.143; 76.003]	<0.001
Embolism	12 (5.2)	16 (51.6)	19.555 [7.847; 48.734]	<0.001
Spondylodiscitis	1 (0.4)	1 (3.2)	7.700 [0.469; 126.344]	0.076
Symptoms >7days	29 (12.5)	26 (83.9)	[12.954; 102.282]	<0.001

New added variables in highlighted rows. Quantitative variables in mean ± standard deviation.

MIC: minimum inhibition concentration; CRP: C-reactive-protein.

a

Available data.

Another microorganism: if other microorganism different than streptococcus has grown in BC.

Persistent BC: new positive BC drawn spanning > 12 h up to 72 h.

Time to positivity

No significant differences were found in the TTP of blood cultures between both hospitals. Neither, no significant differences were found in the median TTP of blood cultures between IE (14.9 h) and non-IE (17.9 h) (p = 0.053) (Table 2). There were no significant differences between shorter TTP and IE (<10 h, <11 h, <12 h, <13 h, <14 h) in the logistic regression carried out with different cutoff values of the TTP log-transformed (lnTTP) (Supplementary Material Appendix 3). We also compared the median TTP of each streptococcal group for IE and non-IE but found no significant differences in any of the groups (Table 2).

Table 2.

Time to positivity median comparison (hours) (median ± interquartile range) by Streptococcus species.

TTP median by Streptococcus species
	TotalMedian ± IQR	NO IEMedian	IEMedian	Sig
All NBHS		17.90 ± 13.01	14.90 ± 10.72	0.053
S. viridans	16.94 ± 11.54	17.46	12.50	0.121
S. mitis/oralis	16.77 ± 12.76	16.67	17.29	0.665
S. bovis	9.41 ± 3.87	8.88	11.00	0.270
S. anginosus	24.89 ± 15.27	25.41	14.67	0.091
S. salivarius	14.04 ± 5.98	14.32	11.50	0.468
S. mutans	18.14 ± 10.94	14.33	41	–

HANDOC score validation study

The diagnostic accuracy of the HANDOC score, using a standard cutoff point of ≥3 and standard scoring with ≥2 positive in different culture sets, showed a sensitivity of 0.95 (95% CI, 0.90–1), specificity of 0.74 (95% CI, 0.69–0.80), PPV of 0.44 (95% CI, 0.34–0.53), NPV of 0.98 (95% CI, 0.97–1), and an AUC of 0.915. On the other hand, the diagnostic accuracy of subsequent analysis using blood culture scoring with at least ≥2 positive vials, showed a sensitivity of 0.99 (95% CI, 0.98–1), specificity of 0.68 (95% CI, 0.62–0.74), PPV of 0.39 (95% CI, 0.30–0.48), NPV of 0.99 (95% CI, 0.98–1), and an AUC of 0.936 (Table 3). The distribution of cases associated with each score rating and the ROC curves are depicted in Figs. 3 and 4.

Table 3.

Summary of the results after applying HANDOC score in the cohort scoring by ≥2 cultures (standard) or scoring by ≥2 vials (adapted).

Scoring	Se	Sp	PPV	NPV	AUC
HANDOC≥2 cultures	0.95(0.902; 1)	0.74(0.691; 0.801)	0.44(0.342; 0.533)	0.98(0.973; 1)	0.915
HANDOC≥2 vials	0.99(0.983; 1)	0.68(0.625; 0.745)	0.39(0.309; 0.485)	0.99(0.987; 1)	0.936

Sensitivity (Se), specificity (Sp), positive predictive value (PPV), negative predictive value (NPV) with confidence interval [CI] and area under the curve (AUC).

Fig. 3.

ROC curve of HANDOC score in the cohort (AUC of 0.915) with standard scoring and number of cases associated with each score rating.

Fig. 4.

ROC curve of HANDOC score in the cohort (AUC of 0.936) scoring with ≥2 positive vials and number of cases associated with each score rating.

Discussion

In our cohort, the prevalence of IE in NBHS bacteremia is 11.1%, higher than the approximately 7–7.5% reported in some previous series.^1,4,24,25 This difference may be attributed to geographical variations compared to previous studies conducted solely in northern Europe, as well as the complexity of our patient population (70% with Charlson index ≥ 4). Furthermore, increased awareness in clinical settings and the establishment of specific units for monitoring bacteremia may have contributed to the increased detection of IE along the years. There could not be substantial differences by criteria as we applied the same 2015 modified Duke criteria as previous series.

Validation score

Using a cutoff value of ≥3, the HANDOC score demonstrated a sensitivity of 95% in our cohort, similar compared with the primary study,¹³ but consistent with the findings of subsequent validations.^15–17 It exhibited a specificity of 74%, comparable to the 73% reported in the initial results,¹³ but higher to 62%¹⁵ and 66%¹⁶ observed in validation cohorts.

It is noteworthy that recently the NPV is increasingly recognized as the most suitable item for evaluating scores, as it helps categorize patients as low-risk, where detailed investigations with echocardiography to rule out IE may not be required. A generic threshold has been established where an NPV of at least 98% is considered safe for excluding endocarditis.^17,26 In our cohort, we have obtained an NPV of 98%.

In our cohort, a considerable portion (35%) of our patients had acquired only one blood culture set at the initial assessment, despite repeated BC sets were obtained later: 98 patients were initially positive with only 2 vials collected. In this regard, we analyzed the application of the HANDOC scoring with ≥2 positive vials, regardless of whether they were obtained from a single or multiple blood culture sets. Undoubtedly, this reduces the score specificity as we report, but it increases sensitivity and, more important, NPV. This scenario is representative of many daily clinical situations, further highlighting the score's usefulness as a tool for either ruling out or suspecting IE. Moreover, obtaining 2 positive different sets of blood cultures and persistent positivity of blood cultures are substantial clinical events in suspecting IE and are major Duke criteria itself. Therefore, the validation of the score through the positivity of 2 vials, rather than diminishing the study's robustness, enhances the score's applicability across diverse clinical settings, preserving its validity. In 6 out of 31 cases of IE (5 of them without ‘Repeated BC’ collected or negative ‘Persistent BC’ confirmed) and 7 out of 17 possible IE cases had microbiological yield with a single set of blood cultures, that could have been missed with the standard HANDOC scoring system.

Echocardiography

Considering the percentage of cases where echocardiography was performed as a measure of quality and reliability within our cohort, our study achieved satisfactory results: 36.8% TTE and 16.4% TEE, which are higher compared to previous cohorts. For cases classified as ‘Possible IE’, the rates were notably high at 88.2% for TTE and 70.6% for TEE (only two patients categorized as it did not undergo an ultrasound and had low-risk profile). Among all IE cases, 19 (61%) were detected by TTE, a proportion higher than that observed in cases of enterococcal endocarditis.¹² Additionally, 11 cases (35.4%) were identified by TEE, although 9 of these cases were missed by TTE but identified by TEE (6 of them scoring > 4 points) and one case only after PET/CT imaging.

If we had to consider only the cutoff of ≥3 points as the only criterion for performing an ultrasound could suggests that the HANDOC score might lead to an increased number of performed ultrasounds.¹⁶ However, in our cohort 30% of echocardiography studies were performed in patients scoring < 3 points, with 10.8% of TEE studies accumulated in that group. If echocardiography had been performed only in cases with a HANDOC score of ≥3, the total number of investigations would have decreased from 149 to 103. In the initial HANDOC cohort,¹³ the ‘No endocarditis’ group had a very high rate of echocardiography performed, which contributed to increase the estimation of TEEs that could have been avoided in that cohort. In our study, the number needed to screen to detect 1 case of IE was 2.2, lower than in the initial HANDOC cohort but similar to the second validation cohort.¹⁵

Timeliness until echocardiography should be improved in the high-suspicion group. Among patients with ≥3 points, 52% of echocardiography studies were conducted within the first three days, although 4 patients experienced delays of more than seven days. It could be considered that the score may help categorize the priority in performing echocardiography studies.

Risk variables – time to positivity

To date, the association between the time to positivity of blood cultures and infective endocarditis has been established, particularly for S. aureus.⁹ However, for NBHS an association with the previously proposed growth cutoff points (15 h) has not been confirmed.²⁷

In our cohort, although there were no significant differences between the overall median times for IE and non-IE, we attempted to find differences at lower cutoff points (<10 h, <11 h, etc.) without detecting significant differences. This may primarily be due to the heterogeneity in growth times of the NBHS as observed in table results. However, no significant differences were found when analyzing median times for each streptococcal group. These results, along with those previously presented by Krus D et al.,²⁷ suggest that blood culture TTP might be a challenging factor to consider in the HANDOC score.

Uncertainty group

The ‘Possible IE’ group is a highly heterogeneous category, referred in other cohorts as ‘Unknown’,^13,15 encompassing patients with greater diagnostic or therapeutic uncertainty. Despite having a clear risk profile, this patient meets only minor criteria as per the 2015 modified Duke criteria,²⁰ potentially leading to underdiagnosis. 16 out of 17 patients categorized as ‘Possible IE’ regimen had a HANDOC score of 3 or more points, scoring mainly 3 and 4 points (12 patients). IE cases scoring mainly 5 and 6 points (18 patients); thus, the score generates a trend beyond the statistical values calculated through the cutoff point. All 9 out of these 17 patients treated with a four-week regimen had a HANDOC score of 3 or more points, so the score also appears to support therapeutic decision-making in suspected cases.

However, our study has limitations. Firstly, not all species were specifically identified using MALDI-TOF MS, as this technology was not implemented during the early years of our cohort, meaning it was not applied in 30% of cases. In this regard, some cases were excluded because the species were not specifically identified; however, no cases of IE were lost. When S. viridans was identified, it was retained due to its historical association with IE, and because this technology is not already available in all hospitals nowadays. This may have influenced the distribution of certain species and subspecies. Nevertheless, not as much as 23% of microbiological findings without the use of MALDI-TOF were not specifically categorized into subspecies. Secondly, the 2023 modified Duke-ESC criteria and the 2023 modified Duke-ISCVID criteria definitions were not applied because, at the time clinicians categorized cases as ‘Definite IE’ or ‘Possible IE’, the new criteria had not yet been published. Indeed, if the 2023 criteria had been applied, none of our patients would have moved to the ‘Definitive IE’ group. Only 4 patients would have been reclassified into the ‘Possible IE’ group, mainly due to changes in predisposition, echocardiographic findings, and new murmur onset as minor criteria.

Present and future

Scores are becoming an increasingly valuable tool in daily clinical practice. Some new validations are being published^15,16,22 coinciding with the conclusion of our study, showing similar results to those previously reported. The flexible application of modified Duke criteria can be challenging, and it may not always resolve certain clinical scenarios, particularly in suspected or possible cases. Scores are not meant to replace clinical judgment but rather to guide clinical practice²⁸ and potentially enhance modified Duke criteria in the future, along with recent publications in 2023 Duke-ESC criteria and modified 2023 Duke-ISCVID criteria.

Moreover, items should be re-explored, such as those addressed by the 2023 Duke-ISCVID, including the incorporation of Streptococcus agalactiae and Streptococcus dysgalactiae as typical IE pathogens, or the proposed HANDOC variation scoring by vials in some clinical settings, especially to avoid missing cases of IE when only one set of blood cultures is available, offsetting the stringency of the microbiological item.

Blood culture TTP might be a challenging factor to consider in the HANDOC score, or at least it appears difficult to incorporate it in a rapid and practical manner without species-level disaggregation.

In conclusion, the HANDOC score proves valuable for decision-making in NBHS bacteremia within a Spanish cohort. It helps prioritize the need for echocardiography studies and supports therapeutic decisions. Blood culture item scoring could be adapted to different clinical settings based on vial counting. Median time-to-positivity did not show a significant difference between episodes with IE and those without IE, regardless of different cutoff points or NBHS group. Further exploration of additional variables could enhance the score.

Declaration of generative AI and AI-assisted technologies in the writing process

During the preparation of this work the author(s) used Chatgpt in order to translate. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the publication.

Conflict of interest

The authors declare that they have no conflict of interest.