Prognostic factors of patients admitted in a medical intermediate care unit : a prospective observational study

Results: Two hundred and eighty-eight patients were included. lntCU and in-hospital mortality was 9,38 and 17,71 %, respectively. All the scores applied, concerning comorbidity, functional status, acute illness severity and nurse workload were significantly associated to mortality. Simplified Acute Physiology Score II (SAPS II) was the better predictor of mortality followed by Nursing Activities Score (NAS).


Background
Medical Intermediate Care Units (IntCU's) are high-dependency units intended for the treatment of unstable patients who do not meet the criteria for admission to intensive care units (ICU's) but require a higher level of care than that provided on a general ward [1].IntCU's operate as transitional unit [2], working as a step-down unit of patients from ICU [3,4] and as step-up unit of patients from general wards or emergency department [4].
On the other hand IntCU's may relieve the general medicine ward of the most unstable patients [15].Some authors have also suggested that chronically critically ill patients who might not benefit of intensive care should be managed in alternative places to ICU [25,27,31], such as IntCU's.
Clinical effectiveness of IntCU's has also been demonstrated.Continuous monitoring carried out in IntCU permits early identification of physiologi-cal deterioration, before organ dysfunction occurs, and timely intervention, avoiding need for ICU admission [4,32] and improving outcome [21].As a step-down unit IntCU might also improve outcome of patients discharged from ICU [21,33] as some deaths may occur following premature discharge to general wards unable to provide the level of care still necessary to a recovering patient.Consequently, some authors have found that IntCU's prevented ICU readmissions and reduced hospital mortality.[33,34,56].IntCU's also improve patient and family satisfaction as its physical environment is less agressive than the ICU [1].
Several studies have focused on prognostic factors of patients admitted in ICU's, demonstrating utility of severity scores as prognostic factors [35][36][37][38][39][40].However, very few studies were performed in IntCU's [41][42][43]56].Some authors suggested using ICU specific acute illness severity scores to identify low-risk patients eligible to IntCU's [8,44,45].So far there are no specific severity scores for intermediate care, and efficiency of intensive care severity scores has not been fully established in intermediate care setting.
The primary aim of this study was evaluation of mortality and analysis of acute illness severity, nurse workload, comorbidity and previous functional status as prognostic factors of patients admitted to a medical intermediate care unit.

Methods
This study was a prospective, single center, observational study conducted during a 32 months period in the Medical Intermediate Care Unit (Int-CU) of São Francisco Xavier Hospital.São Francisco Xavier Hospital is a central and university hospital of Lisbon, that belongs to a Hospital Centre of 900 beds, serving a population of about 935.000 people as a tertiary referral center.
The IntCU is a 4-bed unit that mainly admits medical patients from the emergency department, from medical and surgical wards and from the intensive care unit.During 32 months (April 2008-December 2010) 377 patients were admitted in the IntCU (mean age 67.1±19.3years, mortality 10,6%).Two hundred and eighty eight patients were enrolled to this study, excluded those with data incomplete and those who stayed in the IntCU for a period of less than 24 hours (Figure 1).
Patients data were recorded in a database including age, gender, previous functional status, previous comorbidity, length of stay, primary diagnosis on admission, severity of acute illness, nurse workload and outcome.All data were obtained using standardized instruments.

Previous Functional Status
The previous functional status was determined applying the Barthel index of basic activities of daily living (BADL) [46], classifying patients as independent, slightly dependent, moderately dependent, highly or totally dependent in BADL, according to the number of basic care skills in which patients require physical assistance.Information was obtained directly through interview to patient and/or caregiver and clinical records of the origin departments.

Previous Comorbidity
Comorbidity was determined applying the Charlson comorbidity index [47], which is an instrument performed to predict 10-year mortality in longitudinal studies, according to comorbid diseases that patient presents.Existence of comorbid conditions was assessed through research of previous hospital archives and/or interview to patient and/or caregiver.

Primary Diagnosis
Fifteen categories of primary diagnosis were established: Heart Failure, Arrhythmias, Myocardial infarction or ischemia, Pulmonary Thromboembolism, Pulmonary Disease, Sepsis, Neurological disease, Hepatic Disease, Metabolic or Endocrinologic disease, Digestive bleeding, other Gastroenterologic Disease, Intoxication, Pancreatitis, Kidney Diseases and Others.

Severity of Acute Illness
Severity of acute illness was determined applying intensive care scores: Acute Physiology and Chronic Health Evaluation II (APACHE II) [37], the Simplified Acute Physiology Score II (SAPS II) [35] and the Sequential Organ Failure Assessment (SOFA) [41], which were calculated in the first 24 hours of admission.The main purpose of these three scores is to stratify patients according to in-hospital mortality risk and higher scores represent higher mortality risk.

Nursing Workload
Nursing workload, which can be an indicator of illness severity, was measured through two intensive care scores: Therapeutic Intervention Scoring System-28 (TISS-28)[48] and the Nursing Activities Score (NAS) [49], which were calculated in the first 24 hours of admission.IntCU nurse: patient ratio in daytime was 0,5.

Outcome
The main outcome was mortality in the IntCU and the secondary one was in-hospital mortality.In-hospital mortality was obtained reviewing hospital electronic records.According outcome, two groups of patients were established: Survivors and Non-Survivors.

Statistical analysis
Continuous variables were expressed as mean ± standard deviation, median and 95% confidence interval (IC) for the median.Comparison of continuous variables between Survivors and Non-Survivors was performed using the T test.Non-parametric Test (Kruskal Wallis) was used to compare differences between independent multiple groups, such as NAS variation concerning origin of patients.
Bivariate and multiple logistic regression with forwards stepwise selection were used to identify prognostic factors of IntCU and in-hospital mortality.The entry criterion for the multivariate model was p ≤ 0,05.The receiver operating characteristics (ROC) area under the curve (AUC) was used to assess models discrimination.
Tests were two-tailed and reported statistically significant at p<0,05.
SPSS version 18 was used to statistical analysis.
Comparison of patients enrolled in this study was performed according to their outcome: Survivors (n=237) and Non Survivors (n=51) (Table 1).
Survival was significantly associated to younger age, lower comorbidity, better functional status, lower acute illness severity and nursing workload.
The origin of these patients can be a bias for survival once the authors rejected the hypothesis that there is an equal distribution of NAS according to origin (p=0.003).The most significant difference of NAS variation was between outpatient clinic and others (non ICU, emergency department or medical ward).Concerning these last three departments, medical ward presented the worst mean of NAS and so the necessity of a higher nurse:patient ratio for these patients (Figure 2).
No statistically significant differences were observed between IntCU and in-hospital non-survivors.
The overall mortality was 17,71% (9,38% in IntCU and 8,33% after IntCU discharge).Mean length of stay in IntCU was 10,18 ± 9,07 days.These data are presented in table 2. Bivariate logistic regression analysis (Table 3) revealed that Charlson index, the three acute illness severity scores, and the two nursing workload scores applied were associated with IntCU and in-hospital mortality.Barthel index revealed to be a protective factor for death.Age was also associated to both IntCU and in-hospital mortality, though it was a weaker mortality predictor compared to acute illness severity, comorbidity and functional status.Multiple logistic regression revealed that SAPS II and NAS were the strongest predictors of IntCU and in-hospital mortality and both models showed good discriminant performance (AUC 0,826, 95% CI: 0,749-0,903; AUC 0,811, 95% CI: 0,751-0,871, respectively).

Discussion
This study revealed that outcome of patients admitted in IntCU depends on several factors, such as previous functional status, comorbidity, acute illness severity and also age.To classify each variable we applied scores validated in other levels of care, as tools specific of intermediate care are inexistent.
Previous functional status was inversely associated to mortality, being higher independence in BADL a protective factor for death (odds-ratio (OR) 0,981 (0,970-0,992), p = 0,001), similarly to data presented by Torres et al [43].Previous studies in Medical Wards and ICU revealed that functional impairment is a predictive factor of mortality [50].
Our results revealed that acute illness severity and nursing workload scores validated in intensive care setting are also good mortality predictors in intermediate care setting.Indeed, all the scores applied were significantly associated to mortality.
Concerning the acute illness severity scores, SAPS II revealed to be the strongest predictor of both IntCU and in-hospital mortality (p<0,001).Previous studies in IntCU have already suggested that SAPS II was reliable to assess severity of illness of patients admitted to an IntCU [41].We believe that better performance of SAPS II in prediction of IntCU mortality, in comparison to APACHE II and SOFA, might be explained by integration of distinct clinical data in the score, including not only age and data related to organ dysfunction but also previous diseases generally associated to poor outcome, such as neoplastic diseases.
Regarding the nursing workload scores, both NAS and TISS-28 revealed to be good predictor of outcome, but NAS appeared to be superior to TISS-28.Previous studies have reported TISS-28 as a predictor of short-term mortality in IntCU [43] and a potential tool to differentiate between ICU and high-dependency unit patients [51].
Our results revealed a lower IntCU mortality (9,38%) than that predicted by SAPS II, as the mean score obtained corresponds to a predicted mortality of about 15%.Nevertheless, in-hospital mortality was similar to that predicted by SAPS II.Porath et al [30] have conducted a similar study and have found identical in-hospital mortality (17,6% vs 17,71% in our study) though mean APACHE II (12,9) and TISS (12,6) were lower than ours.
Unlike previous studies [52], age revealed to be a predictor of outcome in IntCU, though the weaker one among all the other variables we have studied.
Focusing on multiple logistic regression analysis we might consider that among the variables we have studied the strongest mortality predictor was SAPS II, followed by NAS.
Limitations of our study are mainly related to a relatively small sample size and being a single-centre study.Multicentric studies in intermediate care setting are difficult to perform because of lack of standardization of structure and procedures performed and heterogeneity of patients admitted.Bias might be introduced concerning topics inquired to caregivers or preadmission-related, namely previous functional status [53,54].Nevertheless, all data obtained were statistically significant.

Conclusions
We consider that our data demonstrate the importance of several variables in patient outcome, including not only acute physiologic variables but also previous comorbidity and functional status.
Our results establish the usefulness and reliability of acute illness severity scores validated in ICU in predicting mortality of patients admitted to intermediate care units.
Identification of prognostic factors of IntCU patients is necessary to appropriate and reasonable selection of patients that would benefit from admission in IntCU.Otherwise, absence of admission criteria in IntCU may lead to overutilization of IntCU beds, similar to that observed in ICU setting [3].
Consequently, a comprehensive assessment of patients admitted in IntCU is mandatory to reliably predict their outcome, as previously described [43].According our results we believe that a possible standard admission assessment would include the SAPS II and NAS calculated in the first 24 hours of admission, the Barthel and Charlson Indexes.Age might be considered a prognostic factor, though inferior than other scores applied.
Nonetheless, we believe that individual patient decisions should not be standardized and exclusively based on scores.Indeed, previous studies have revealed limited usefulness of scores in taking individual patient decisions [40], and clinical judgement should not be substituted for physiological scores [55].Instead, scores should help and support the individual clinical decision.
Future studies should focus on development of assessment tools integrating physiologic variables, comorbidity and previous functional status, in order to support individual patient decisions.In addition, an effort should be made to implement multicentric studies including similar centers, rather than single-center.Publish with iMedPub http://www.imed.pub

Figure 1 :
Figure 1: Flow diagram of patients included and their outcome

48.
Miranda DR, de Rijk A, Schaufeli W, Simplified Therapeutic Intervention Scoring System: the TISS-28 items--results from a multicenter study.Crit Care Med 1996, 24(1):64-73.49.Miranda DR, Nap R, de Rijk A, Schaufeli W, Iapichino G, Nursing activities score.Crit Care Med 2003, 31(2):374-382.50.Bo M, Massaia M, Raspo S, Bosco F, Cena P, Molaschi M et al, Predictive factors of in-hospital mortality in older patients admitted to a medical intensive care unit.J Am Geriatr Soc 2003, 51(4):529-533.51.Pirret AM, Utilizing TISS to differentiate between intensive care and high-dependency patients and to identify nursing skill requirements.Intensive Crit Care Nurs 2002, 18(1):19-26.52.Hood E, Bhangu A, Pandit D, Michael A, Is age a predictor of mortality in a UK medical high dependency unit?Br J Anaesth 2011, 107(2):186-192.53.Nelson LM, Longstreth WT, Jr. Koepsell TD, van Belle G, Proxy respondents in epidemiologic research.Epidemiol Rev 1990, 12:71-86.54.Mundt DJ, Gage RW, Lemeshow S, Pastides H, Teres D, Avrunin JS, Intensive care unit patient follow-up.Mortality, functional status, and return to work at six months.Arch Intern Med 1989, 149(1):68-72.55.Bion J, Outcomes in intensive care.BMJ 1993, 307(6910):953-954.56.Capuzzo M., Volta C.A., Moreno R.P., Valentin A., Guidet B., Lapichino G. et al, Hospital mortality of adults admitted to Intensive Care Units in hospitals with or without Intermediate care Units: a multicentre European Cohort study.Critical Care 2014, 18:551.Where Doctors exchange clinical experiences, review their cases and share clinical knowledge.You can also access lots of medical publications for free.Join Now! http://medicalia.org/Comment on this article:International Archives of Medicine is an open access journal publishing articles encompassing all aspects of medical science and clinical practice.IAM is considered a megajournal with independent sections on all areas of medicine.IAM is a really international journal with authors and board members from all around the world.The journal is widely indexed and classified Q1 in category Medicine.

Table 3 .
Results of the bivariate and multiple logistic regression analyses (OR odds ratio)