| Abstract | Cilj rada: Zatajivanje srca (ZS) je sindrom s velikim medicinskim, socijalnim i ekonomskim značenjem, koje proizlazi iz njegove učestalosti, troškova liječenja i visoke smrtnosti. Unatoč značajnom poboljšanju liječenja prognoza sindroma je loša. Predviđanje ishoda nakon postavljene dijagnoze ZS najnesigurniji je dio obrade ovih bolesnika. Većina dosadašnjih napora za poboljšanjem prognostičkih mogućnosti bila je usmjerena na dugoročni ishod u bolesnika s kroničnim ZS, a prognozi hospitalnog ishoda u akutnom ZS posvećeno je manje pažnje. Prognoza rizika smrti u bolesnika hospitaliziranih zbog akutnoga ZS važna je radi odluke o mjestu prijama, liječenju i intenzitetu nadzora nad bolesnikom. Posljednjih godina formirano je nekoliko modela za procjenu hospitalnog rizika smrti u ovih bolesnika, koji zbog zahtjevnosti u primjeni nisu stekli široku popularnost. Cilj ovoga rada bio je oblikovati jednostavan, jeftin, brz i lako primjenljiv model temeljen na jednostavnim pokazateljima rizika s već dokazanom prognostičkom snagom, a koji ne će zaostajati za dosadašnjim složenijim modelima, dakle, model sa značajkama koje bi mu omogućile široku primjenu i učinile procjenu rizika hospitalne smrti dijelom rutinske obrade kod prijama bolesnika.
Ispitanici i metode: Ovaj rad je izveden na Zavodu za kardiovaskularne bolesti Klinike za internu medicinu Kliničkoga bolničkog centra Rijeka. Obuhvatio je bolesnike koji su hospitalno liječeni zbog akutne dekompenzacije zatajivanja srca (ADZS) u razdoblju od 1. siječnja 2006. do 30. siječnja 2011. godine. ADZS definirana je kao progresivno pogoršanje ZS u bolesnika s ranije postavljenom dijagnozom kroničnoga ZS ili kao novonastala dekompenzacija uz uvjet isključenosti drugih oblika akutnoga ZS (kardiogeni šok, plućni edem, hipertenzivno ZS, izolirano zatajivanje desnoga srca, ZS u akutnome koronarnom sindromu). Derivacijska skupina od 654 bolesnika poslužila je za izvođenje formule rizika, a njezina pouzdanost provjerena je u validacijskoj skupini od 591 bolesnika. Tijekom hospitalizacije umro je 121 bolesnik derivacijske skupine, a ostali (533) su živi otpušteni. U ovoj skupini analizirana su 35 pokazatelja, od kojih je univarijatnom analizom izdvojeno 16 prediktora nepovoljnog ishoda (dob, hipertenzija, teška anemija, moždani udar/tranzitorna moždana ishemijska ataka, kronična bubrežna bolest, pušenje, sistolički i dijastolički krvni tlak, frekvencija srčane akcije, blok lijeve grane, procjenjena brzina glomerularne filtracije te serumske vrijednosti ureje, kreatinina, natrija, mokraćne kiseline i NT-proBNP-a). Njihova diskriminacijska snaga odreĊena je ROC-analizom, (engl. Receiver Operating Characteristic), nakon koje su iz daljnje analize isključene varijable s površinom ispod krivulje (AUC) <0,6. Zbog korelacije izmeĊu sistoličkoga i dijastoličkog tlaka te ureje, kreatinina i glomerularne filtracije izostavljeni su dijastolički tlak, kreatinin i glomerularna filtracija iz razloga da se zadovolji postavljeni cilj, a to je formula za brz i jednostavan način odreĊivanja rizika. Tako je za oblikovanje formule ostalo 6 pokazatelja: dob, sistolički krvni tlak, puls, natrij, ureja i mokraćna kiselina.
Rezultati: Dob bolesnika (D), frekvencija pulsa (P) i sistolički tlak (ST) sjedinjeni su u novi pokazatelj po već poznatoj formuli [(D/10)2xP/ST], koja je do sada bila korištena u druge svrhe, ali ne i za procjenu rizika u ZS. Ovaj izvedeni pokazatelj je pri ROC-analizi imao AUC 0,717 (95% granice pouzdanosti 0,662-0,772; P <0,001) i diskriminacijsku snagu veću od svake sastavnice (0,717 prema 0,614; P=0,001 u odnosu na dob, 0,717 prema 0,647; P=0,010 u odnosu na puls i 0,717 prema 0,658; P=0,034 u odnosu na sistolički tlak). Drugi pokazatelj izveden je iz razlike serumske ureje (U) i natrija (Na), čija je vrijednost dijeljena s 10 pa je tako formula imala oblik: U-Na/10. Za taj pokazatelj AUC je bila 0,694 (95% granice pouzdanosti 0,639-0,750; P <0,001). I ovaj je parametar imao veću diskriminacijsku snagu u odnosu na svoje sastavnice (0,694 prema 0,660; P=0,008 u odnosu na ureju i 0,694 prema 0,608, P=0,004 u odnosu na natrij). Mokraćna kiselina podijeljena sa 100 (MK/100) korištena je kao samostalan pokazatelj (AUC=0,647, 95% granice pouzdanosti 0,588-0,706, P <0,001). Kombiniranjem izvedenih pokazatelja i mokraćne kiseline oblikovana je nova formula [(D/10)2xP/ST]+(U-Na/10)+MK/100. Tako kombinirani pokazatelj imao je značajno veću AUC, 0,741, uz standardnu pogrješku od 0,027 te 95% granice pouzdanosti 0,706-0,774 i P <0,001, što je bio značajno bolji rezultat u usporedbi sa sve 3 njezine sastavnice [AUC u odnosu na (D/10)2xP/ST 0,741 prema 0,717 (P=0,004), u odnosu na (U-Na/10) 0,741 prema 0,694 (P=0,015) i u odnosu na MK/100 0,741 prema 0,647 (P=0,008]. Formula je provjerena u validacijskoj skupini (591 bolesnik, od kojih su umrla 64, tj., 10,8%). U njoj je AUC također bila 0,741 (95% granice pouzdanosti 0,701-0,776). Uz razdjelnicu od 53 pozitivna prediktivna vrijednost formule bila je 34,4%, negativna prediktivna vrijednost 94,8% i točnost 74,6%. U skupini s rezultatom formule <34 smrtnost je bila 3,3%, u bolesnika s rezultatom 34-53 bila je 7,0%, u onih s rezultatom 54-74 iznosila je 18,9% i u skupini s rezultatom >74 je bila 29,6%.
Zaključak: Dobivena formula koristi surogatne pokazatelje povećane neurohormonalne aktivnosti u ZS. Temelji se na parametrima koji su dio rutinske kliničke obrade bolesnika sa ZS i dostupni su kod prijama. Formula zadovoljava postavljene uvjete jednostavnosti i brze izvedljivosti, a pritom ne zaostaje za postojećim složenijim i zahtjevnijim modelima. Omogućava kvantitativnu procjenu hospitalnoga rizika smrti u bolesnika s ADZS i njihovo razvrstavanje u skupine rizika. Premda je pozitivna prediktivna vrijednost formule skromna, njezina velika negativna prediktivna vrijednost može biti korisna u prepoznavanju malog rizika smrti i omogućiti racionalniji postupak s bolesnikom. |
| Abstract (english) | Aim. Heart failure (HF) is a syndrome with major medical, social and economic significance, resulting from its prevalence, treatment costs and high mortality. Despite significant advances in the treatment of this syndrome, prognosis is poor. Predicting the outcome upon HF diagnosis is the least reliable part of the treatment of these patients. Most previous efforts to improve prognostic capabilities focused on long-term outcome in patients with chronic HF, while the prognosis of hospital outcomes in acute HF was devoted less attention. Considering of risk of death in patients hospitalized for acute HF is important in order to make decisions about the place of admission, treatment and intensity of supervision of the patient. In recent years, several models have been formed to estimate the risk of hospital death in these patients. However, the models were not widely used because their application was demanding. The aim of this study was to design a simple, inexpensive, quick and easy applicable model based on simple indicators of risk with already proven predictive strength, that will not lag behind current complex models, therefore, a model with features that would enable widespread use and would make a risk assessment of in-hospital mortality part of routine treatment in patients at admission.
Patients and Methods: This study was conducted at the Department of Cardiovascular Diseases, Department of Internal Medicine, Clinical Hospital Centre Rijeka. The patients who were included were those treated in hospital for acutely decompensated heart failure (ADHF) in the period from 01 January 2006 to 30 January 2011. ADHF is defined as a progressive deterioration of HF in patients with previously diagnosed chronic HF or a newly emerged decompensation provided exclusion of other forms of acute HF (cardiogenic shock, pulmonary edema, acute HF with arterial hypertension, isolated right heart failure, acute coronary syndrome with HF). The derivation group of 654 patients was used to perform risk formula and its reliability was tested in the validation group of 591 patients. During hospitalization, 121 patients in the derivation group died, and others (533) were discharged alive. In this group 35 indicators were analyzed, of which 16 predictors of adverse outcome (age, hypertension, severe anemia, stroke / transient cerebral ischemic attack, chronic renal disease, smoking, systolic and diastolic blood pressure, heart rate, left bundle branch block, estimated glomerular filtration rate and serum urea, serum creatinine, serum sodium, uric acid and NT-proBNP), were separated by univariate analysis. Their discriminative ability was determined by “Receiver Operating Characteristic”(ROC) analysis, after which variables with area under the curve (AUC) <0.6 were excluded. Because of the correlation between systolic and diastolic blood pressure, as well as between urea and creatinine, respectively estimated glomerular filtration rate, diastolic blood pressure, creatinine and glomerular filtration rate were omitted to satisfy the goal aimed at, and that was providing a formula for determining risk in a quick and easy way. Thus, 6 variables were included in the formula design: age, systolic blood pressure, heart rate, serum sodium, serumu urea, and uric acid.
Results: Patients age (A), heart rate (HR) and systolic blood pressure (SBP) were combined into a new indicator according to the already known formula [(A/10)2xHR/SBP], which has so far been used for other purposes, but not for the risk assessment in HF. In the ROC analysis, AUC for the derived indicator was 0.717 (95% confidence interval from 0.662 to 0.772, P <0.001) and it had a greater discriminative power in relation to each component (0.717 to 0.614, P=0.001 compared to age, 0.717 to 0.647, P=0.010 compared to heart rate, 0.717 to 0.658, P=0.034 compared to systolic blood pressure). Another indicator was derived from the difference in serum urea (U) and serum sodium (Na), whose value was divided by 10 and resulted in a formula: U-Na/10. For that indicator, the AUC was 0.694 (95% confidence interval from 0.639 to 0.750, P <0.001). This parameter had a greater discriminative power compared to its components (0.694 to 0.660, P=0.008 compared to serum urea and 0.694 to 0.608, P=0.004 compared to serum sodium). Uric acid (UA) divided by 100 (UA/100) was used as a stand-alone indicator (AUC=0.647, 95% confidence interval 0.588 to 0.706, P <0.001). Combining the derived parameters with the value for uric acid, a new formula was formed [(A/10) 2xHR / SBP]+(U-Na/10)+UA/100. This combined indicator had a significantly greater AUC (0.741), with a standard error of 0.027 and 95% confidence interval from 0.706 to 0.774 and P <0.001, which was significantly better compared to its three components [AUC compared to (A/10 )2xHR/SBP 0.741 to 0.717 (P=0.004), compared to (U-Na/10) 0.741 to 0.694 (p=0.015) and compared to UA/100 0.741 to 0.647 (p=0.008]. The formula was tested in the validation group (591 patients, of whom 64 died, i.e., 10.8%). In this group AUC was also 0.741 (95% confidence interval from 0.701 to 0.776). With the cut-off point of 53 the positive predictive value of the formula was 34.4%, negative predictive value of 94.8% and 74.6% accuracy. In the group with the formula result <34, mortality was 3.3%, in patients with result 34-53 mortality was 7.0%, in those with result of 54-74 was 18.9%, in the group with a result of > 74 was 29, 6%.
Conclusion: The resulting formula uses surrogate indicators of increased neurohormonal activity in HF. It is based on parameters that are part of the routine clinical management of patients with HF that are available at admission. The formula meets the requirements of simplicity and rapid feasibility, while not being inferior to the existing complex and demanding models. It allows a quantitative assessment of the risk of in-hospital mortality in patients with ADHF and their classification into risk groups. Although the positive predictive value of the formula is modest, its high negative predictive value may be helpful in identifying low mortality risk and allow a more rational procedure with the patient. |
