Candida Score – The Forgotten (Child of) Scoring System(s)

Disseminated candida infections kill. It shouldn’t be terribly surprising. In the ICU, ongoing fevers despite antibiotic therapy is usually taken as a sign of invasive candidiasis and is the impetus for antifungal therapy. Risk stratifying patients for candidal infections can be difficult, given the myriad of comorbid conditions that are associated with these infections. Despite this, there has been attempts at generating scores or prediction tools, especially in the ICU setting. Many of these take into consideration length of hospital stay or candida colonization. For instance, colonization of one or more sites in an ICU setting was associated with increased risk of invasive candidiasis (1):

The risk was dependent upon the time at which colonization was detected and the density of candida in each given site (ratio of number of sites colonized by Candida spp to the number of body sites surveyed), with higher densities being associated with higher risk early on:

Colonization was also associated with higher risk of mortality, with multifocal colonization having higher risk of death compared to unifocal colonization (3, 4):

A meta-analysis also found that patients with longer ICU stays (2.82 difference in means, 95% CI 1.42-4.22) were more likely to have invasive candidiasis (2). While these are helpful risk to think about, there have been attempts at creating a score to risk stratify non-neutropenic, ICU patients and decide the use of early antifungals. One of the most widely known risk scores is the Candida score, which was developed by a Spanish group back in 2006. The initial score was developed in a prospective, observational, multicenter cohort study that evaluated over 1700 patients admitted to medical-surgical ICUs (3). Univariate analysis found that TPN, hemodialysis, severe sepsis, surgery on admission, and candida colonization was associated with proven candidal infection. Multivariate analysis found that surgery, TPN, severe sepsis, and candida colonization were associated with increased risk of candida infection:

Using 65% of the sample, the authors calculated the Candida score using logistic regression model, and used the remaining 35% of the patients to validate the score: 

Patients with a score >2.5 had a RR of 7.7 (95%CI 4.74-12.66). Another prospective study from the same group used a “rounded Candida score” and compared it to the candida colonization index (ratio of number of positive surveillance sites to total number of sites cultured) with cutoffs of greater than 3 and <0.5, respectively (4). Surgical procedures, abdominal surgery, secondary peritonitis, mechanical ventilation, TPN, and broad-spectrum antibiotics were associated with invasive candida infection:

The incidence of invasive candidiasis was 2.3% in those with scores ❤ while it was 13.8% in those whose scores >3, with a statistically significant linear association between increasing values of the  candida score and rate of invasive infection:

Indeed, scores >3, while having poor PPV, had good NPV and it was better in predicting IC than the colonization index of >0.5:

The score was further validated in a prospective observational cohort study in 5 French ICUs, which included 94 patients (5). The overall incidence of invasive candidiasis was 5.3% (5 patients), with rising scores being associated with higher rates of invasive candidiasis, with 4 of the patients having a score >3 (rate of 0% in those whose score was 2 or 3, 17.6% in those who had a score of 4 and 50% in those whose score was 5). While having a good negative predictive value (and allowing you to forego antifungal therapy if the scores are <3), one thing to note is the fact these scores were obtained and validated from a mix of surgical-medical ICUs and one of the points is awarded for surgery on ICU admission. Indeed, a 10 year retrospective analysis (6)  of 139 patients with Candidemia in a medical ICU found that only 37% of patients got a score >2.5:

A significant percentage of patients who had candidemia had scores <2. Understandably, this is a single center that applied the score to all patients with candidemia and not necessarily an ICU population, however the applicability of the score to a non-surgical setting may limit its use in the medical ICU. Indeed, the patients in this analysis were more likely to have chronic liver disease or renal disease and compared to the initial Candida score study, their rates of parenteral nutrition (87.6% vs 6%) and recent surgery (52.6% vs 22%) were much lower. 

Different approaches and scoring systems have been proposed for this dilemma. For instance, a retrospective, single-center study compared the Candida score to BD glucan level and colonization index in a cohort of 95 patients (7). BD glucan had a higher sensitivity than the candida score, though the combination of both CS and BD glucan yielded a sensitivity and specificity of 100% (95% CI 76 to 100) and 93.5 (95% CI 73.5-90.9%) respectively (PPV 52 and NPV 100):

Further, the found that aside from a positive Candida score and TPN, chronic renal replacement therapy and a positive BD glucan were found to be associated with Candidemia, suggesting these are also significant risk factors:

A more complicated scoring system was compared to the Candida score in a prospective study of 114 patients (8). 16 factors were evaluated for the candidemia index, of which parenteral nutrition, severe sepsis, surgery, renal failure, and high APACHE II scores being associated with candidemia:

Each of the above factors were worth one point for a maximum of 16. A cut-off of 8.5 for the candidemia index had a sensitivity, specificity, PPV and NPV of 77%, 70%, 30%, and 94%, respectively, though this was quite cumbersome. The authors also included a clinical predictor index, which required both the use of broad spectrum antibiotics and central line along with another risk factor (parenteral nutrition, surgery, pancreatitis, immunosuppressants, or steroid use). Comparing the Candida score to the clinical predictor index, the latter performed better and was simpler to calculate:

A similar score to that of the clinical predictor index was evaluated in a multicenter retrospective study of 2890 patients (9). 3 different prediction rules were evaluated at different time frames; he use of antibiotics and central lines alone was associated with a RR of 4.71 (95% CI 2.45-9.06), with the addition of other risk factors such as pancreatitis or steroid use having similar RRs:

All this to say that, while the Candida score is a useful screening tool for bedside assessment and to rapidly decide if antifungals need to be started, it may fall short in the medical ICU population and other factor such as the presence of renal and liver disease may need to be taken into consideration. Indeed, the sensitivity of just using “antibiotics + central line” to guide antifungal initiation may be greater, though you risk overtreating. Another criticism of this score is its use of the old Sepsis 2.0 criteria to define severe sepsis. The use of the Sepsis 3.0 criteria was evaluated in a retrospective cohort of over 17,000 patients (10). Similar to the CS-2009 score (the original score), higher scores were associated with higher number of invasive candida cases:

Comparison of both CS-2009 and CS-3.0 revealed they performed similarly:

Given how cumbersome it is to calculate Sepsis 3.0, I would stick to the original score. 

Are scores really useful?

I think from a clinical perspective, they can be useful in certain scenarios where you are not sure as to starting antifungals. As highlighted above, positive scores are not terribly sensitive or have a high positive predictive value, but a negative score can help you rule out invasive candidal infections. There are shortcomings with the Candida score, some of which were already discussed and include the possible lack of applicability to the medical setting, not accounting for chronic renal or liver disease or central line usage, and no insights into immunosuppression. Regardless, I think clinical gestalt in combination with the scoring system can allow you to risk stratify ICU patients and decide who needs antifungals.

References:

1. Lau AF, Kabir M, Chen SC, Playford EG, Marriott DJ, Jones M, Lipman J, McBryde E, Gottlieb T, Cheung W, Seppelt I, Iredell J, Sorrell TC. Candida colonization as a risk marker for invasive candidiasis in mixed medical-surgical intensive care units: development and evaluation of a simple, standard protocol. J Clin Microbiol. 2015 Apr;53(4):1324-30. doi: 10.1128/JCM.03239-14. Epub 2015 Feb 11. PMID: 25673797; PMCID: PMC4365204.
2. Zhang Z, Zhu R, Luan Z, Ma X. Risk of invasive candidiasis with prolonged duration of ICU stay: a systematic review and meta-analysis. BMJ Open. 2020 Jul 12;10(7):e036452. doi: 10.1136/bmjopen-2019-036452. PMID: 32660950; PMCID: PMC7359383.
3. León C, Ruiz-Santana S, Saavedra P, Almirante B, Nolla-Salas J, Alvarez-Lerma F, Garnacho-Montero J, León MA; EPCAN Study Group. A bedside scoring system (“Candida score”) for early antifungal treatment in nonneutropenic critically ill patients with Candida colonization. Crit Care Med. 2006 Mar;34(3):730-7. doi: 10.1097/01.CCM.0000202208.37364.7D. PMID: 16505659.
4. León C, Ruiz-Santana S, Saavedra P, Galván B, Blanco A, Castro C, Balasini C, Utande-Vázquez A, González de Molina FJ, Blasco-Navalproto MA, López MJ, Charles PE, Martín E, Hernández-Viera MA; Cava Study Group. Usefulness of the “Candida score” for discriminating between Candida colonization and invasive candidiasis in non-neutropenic critically ill patients: a prospective multicenter study. Crit Care Med. 2009 May;37(5):1624-33. doi: 10.1097/CCM.0b013e31819daa14. PMID: 19325481.
5. Leroy G, Lambiotte F, Thévenin D, Lemaire C, Parmentier E, Devos P, Leroy O. Evaluation of “Candida score” in critically ill patients: a prospective, multicenter, observational, cohort study. Ann Intensive Care. 2011 Nov 30;1(1):50. doi: 10.1186/2110-5820-1-50. PMID: 22128895; PMCID: PMC3247094.
6. Laine ME, Flannery AH, Moody B, Thompson Bastin ML. Need for expanded Candida Score for empiric antifungal use in medically critically ill patients? Crit Care. 2019 Jul 4;23(1):242. doi: 10.1186/s13054-019-2525-3. PMID: 31272491; PMCID: PMC6609394.
7. Posteraro B, De Pascale G, Tumbarello M, Torelli R, Pennisi MA, Bello G, Maviglia R, Fadda G, Sanguinetti M, Antonelli M. Early diagnosis of candidemia in intensive care unit patients with sepsis: a prospective comparison of (1→3)-β-D-glucan assay, Candida score, and colonization index. Crit Care. 2011;15(5):R249. doi: 10.1186/cc10507. Epub 2011 Oct 22. PMID: 22018278; PMCID: PMC3334800.
8. Gaspar GG, Menegueti MG, Auxiliadora-Martins M, Basile-Filho A, Martinez R. Evaluation of the predictive indices for candidemia in an adult intensive care unit. Rev Soc Bras Med Trop. 2015 Jan-Feb;48(1):77-82. doi: 
9. Ostrosky-Zeichner L, Sable C, Sobel J, Alexander BD, Donowitz G, Kan V, Kauffman CA, Kett D, Larsen RA, Morrison V, Nucci M, Pappas PG, Bradley ME, Major S, Zimmer L, Wallace D, Dismukes WE, Rex JH. Multicenter retrospective development and validation of a clinical prediction rule for nosocomial invasive candidiasis in the intensive care setting. Eur J Clin Microbiol Infect Dis. 2007 Apr;26(4):271-6. doi: 10.1007/s10096-007-0270-z. PMID: 17333081. 10.1590/0037-8682-0292-2014. Epub 2015 Jan 1. PMID: 25860468.
10. Li D, Zhang J, Han W, Bai G, Cheng W, Cui N. Evaluation of the updated “Candida score” with Sepsis 3.0 criteria in critically ill patients. Ann Transl Med. 2020 Aug;8(15):917. doi: 10.21037/atm-20-995. PMID: 32953717; PMCID: PMC7475415.