Multicentric, observational study which included all the adult patients (≥18 years) admitted consecutively to ICU of Clinica Olivos, Sanatorio Anchorena and Clinica Bazterrica, between April 2011 and January 2012, who required MV for more than 24h. The area where the study was conducted was 3 monovalent intensive care units that have intensive care physicians 24hours a day, respiratory therapist, 12hours a day and a nurse every two patients. It should be noted that these institutions receive patients only from the private health system (prepaid health system). The patients readmitted to the ICU were not been considered as new patients, but they were included in the study protocol and all data was analyzed together with data collected in the previous hospitalization.

Patients who received MV for more than 24hours in a previous health center at admission, patients with do not resuscitate (DNR) before or after receiving MV, patients with previous cognitive impairment or unconsciousness (with diagnosis recorded in medical record or family interview) were excluded, and also those patients who did not sign the informed consent, pregnant women and patients with neoplastic or degenerative disease with life expectancy less than a year.

Every day, respiratory therapists were in charge of recollecting the following data. Demographic variables in eligible patients were evaluated: cause of admission to ICU, reason for MV requirement,1 severity of illness at admission (APACHEII score), comorbidities (Charlson index), sedation (Richmond agitation-sedation scale, RASS16), weaning process,17 daily doses of sedatives, analgesics and neuromuscular blockers (NMB), ICUAW (Medical Research Council scale, MRC18: cutoff point 48), delirium (CAM-ICU Spanish version19) and site at discharged from hospitalization: home (with or without home care), or chronic care centers (long term acute care hospital=LTACH).

QOL was assessed by EQ-5D questionnaire, a simple and generic instrument validated in Argentina and valued in Argentinian general population (general population=GP).20 This is a self-administered questionnaire, although phone evaluation has also been recommended21,22 and reliability for measurement by close relatives has been proven (indirectly).23 An index (from 0 to 1) and a (visual analogic scale=VAS) visual analog scale that assesses the overall health status of the patient is obtained from the questionnaire. A complete description of the questionnaire is shown in the electronic supplement. It was delivered to the patient (or family cohabiting in case the patient was unable to complete it) once the informed consent was signed; the results of this first evaluation were referred to four weeks prior to admission. All patients who were able to communicate countersigned the questionnaire later. Three months and one year after being discharged from ICU, the same questionnaire was reassessed by phone to the patient or family cohabitant (proxy), preferably, the same family member who performed it at admission to ICU, in case the patient cannot refer it. The proxy is defined as the person in daily contact with the patient before admission to the ICU. Since our study includes a small number of patients and was not supported by any pharmaceutical/medical firm, the use of the questionnaire was free of charge. Functional status was assessed by independence in activities of daily life (activities of daily living=ADL) not only basic ones, but also instrumental ones that determine people's autonomy. We have used Barthel index since it is one of the most popular generic measures, studied and spread worldwide, assigning different scores according to the level of independence regarding basic activities of daily life. This test has been validated in Argentina24 and for phone use. All Phone calls were made by the first author. A detailed description of the characteristics of ICUs and procedures performed during the study can be found in the electronic supplement.

This study was approved by the Ethics Committee of the three Institutions in accordance with the ethical standards laid down in the 1964 Declaration of Helsinki and its later amendments and also recommended and approved Informed Consent used.

Continuous variables are expressed as median and interquartile range (interquartile range (IQR)) represented by the 25–75 percentile or mean and standard deviation (SD), and categorical variables, as absolute value and percentage. For data analysis, parametric (t test, ANOVA) and nonparametric (Mann Whitney) tests, according to the frequency distribution of continuous variables were used, and Chi square, for categorical variables. Wilcoxon signed rank test was used to compare repeated measurements (EQ-5D and VAS). Logistic regression was performed, the entry criteria for multivariate analysis was p=0.10. A p value less than or equal to 0.05 were considered significant.

This is a prospective cohort study, but the main analysis corresponds to a design of cases and case–control study (nested).

Paired samples are studied; therefore, risk factors will be analyzed based on a conditional logistic regression model. According to this criterion, the sample size was calculated according to the classic recommendation of Freeman (n=10×(K+1). It is estimated that the inclusion of 110 patients will require six months of recruitment. To classify quality of life prior to admission as “good” or “poor”, the study of quality of life in healthy people of Argentina's population was used. It was considered “poor quality of life on admission” when baseline EQ-5D index value was less than lower the confidence interval (CI=confidence interval), and “good” when the index value was equal to or higher. It was considered worsening quality of life when the value of the EQ-5D index at 3 months was less than the lower limit of the confidence interval of the baseline index. The same criterion was used to categorize the QOL one year after discharging from ICU.

In the mentioned period, 181 patients receiving MV for more than 24hours were admitted at 3 ICUs (Fig. 1); 52 patients were excluded. Of the 129 eligible patients, eight of them refused to sign the consent, 37 died in ICU, and 84 patients who were discharged from the ICU, formed the study sample. At 3 months of follow-up, 13 patients had died (5 of them in the hospital), 71 were alive but only 70 could be evaluated. After one year of follow-up, 12 more patients had deceased (total of 25; 30%), 56 were alive (55 were tested) and 3 were lost at follow up.

Figure 1.

Flowchart. Flowchart of enrolled patients with more than 24h of MV and follow-up at 3 and 12 months after discharging from the ICU. MV: Mechanical ventilation.

(0.18MB).

The characteristics of the patients who entered the study are listed in Table 1. The level of education of the patients was 22.6% primary education, 47.6% secondary education and 29.8% college education; 75% of patients were active from the labor point of view (65.5% employee or independent, household activities 6%, student 3.6%) while 25% were retired. Most patients were referred by hospital emergency (60%), while the rest came from the hospitalization area (33%) or another institution (7%). The drugs most used to achieve sedo-analgesia were midazolam in 85% of patients and fentanyl in 88%, although to a much lesser extent, they were also used: remifentanyl in 15%, dexmedetomidine in 17%, lorazepam in 17% and propofol in 13%. The NMB were only used in 8 patients and for short periods of time. Of all the patients using dexmedetomidine (n=14), over 70% used it during NIV (noninvasive ventilation=NIV) after extubation (n=10). The presence of delirium was observed in 57 patients (72%) (5 of them were in a coma, therefore they could not be evaluated) during their stay in ICU; the median duration was [3 days (IQR25–75% 2–6)]; except for 7 patients, all were free of delirium at discharge from ICU. All patients received early mobilization and were evaluated according to their functional achievements; 71% managed to sit on the edge of the bed, 51% did so in a chair and 33% stood up before being discharged from ICU. The ICUAW measured by the MRC score in the wake of the MV was observed in 50 patients (59%); 30 of them (36%) continued with a score <48 at discharge from ICU. The median baseline QOL (corresponding to four weeks prior to admission) showed an index of [0.831 (0.527–0.931 IQR25–75%)], VAS [70 (50–80 IQR25–75%)] and Barthel [100 points (95–100 IQR25–75%)], that is, most of our patients had a good quality of life and functionality before admission to ICU but less than the GP by age (mean 0.880 CI (0.872–0.888) p<0.001). Patients who survived (n=79) were discharged to their homes (n=36), or their homes with home care (n=20) or LTACH (n=22); 1 patient was lost at follow-up.

At 3 months of follow-up, 70 patients were included in the analysis. The interview was direct in 62% of cases and 38% through relatives (proxies). After being discharged from ICU, 57% of patients received functional rehabilitation from 1 to 3 months, while only 29% received psychological treatment. Although 20% had to be readmitted for a median time of 8 days (1–50), none required MV. After a year of follow up, 55 interviews were included, 73.6% directly and the rest, through relatives (proxies).

The results of the EQ-5D index are shown in Table 2. The 73% (n=51) could not perform activities of daily life as they did before admission to ICU (study, work, household activities, etc.) and 53%, after one year of follow up.

The EQ-5D index at 3 months was [0.513 median (0.245–0.838 IQR25–75%)] much lower than the Argentina GP by age (p<0.001). When compared baseline index [0.831 median (0.612–0.931 IQR25–75%)] with index at 3 months, a significant difference p<0.001 was observed. The EQ-5D index after a year was [0.850 median (0.573–1.00 IQR25–75%)] and differed significantly from GP (p=0.0021) and the index at 3 months (p<0.001) but it was similar to baseline index (p0.857) (Fig. 2). EQ-5D VAS value at 3 months had a median of [70 (IQR25–75%) 55–80] similar to the value on admission (p=0.808); after a year was [80 (IQR70–90)] and showed a significant difference between the results at 3 months (p0.011) and on admission to ICU (p0.012). The Barthel Score was [median 97.5 (53.7–100 IQR25–75%)] at 3 months, significantly less than on admission (p<0.001). After one year, Barthel median was [100 (IQR90–100)], significantly higher than the value at 3 months (p0.006), but still lower than the value reported on admission (p0.014). Another way of presenting the results is in relation to independence in ADL (Table 3). It should be pointed out that at 3 months and at one year of follow up, 51% and 32% of patients had some degree of dependence compared to 27% of them recorded on admission.

Figure 2.

Box-Plot of EQ-5D index. Box plot of EQ-5D baseline index, 3 months and 1 year follow-up. *p<0.001 between baseline and 3 months EQ-5D index. #p<0.001 between 3 months and 1 year EQ-5D index.

(0.07MB).

The EQ-5D baseline index of each patient and the index that showed the same patient at 3 months after being discharged were compared. Of the 70 patients who survived at 3 months, 47 (67.1%) showed deterioration in quality of life score with respect to the one on admission [median baseline index 0.883 (0.651–0.931 IQR25–75%)] and at 3 months [median 0419 (0.05–0.453 IQR25–75%)]; however 23 patients (32.9%) were included into the confidence interval of baseline index [baseline index 0.651 (0.480–0.883 IQR25–75%)] and at 3 months [median 1.0 (IQR25–75 0.831–1.0%)]. According to these results, we can distinguish two subgroups and compare variables. In univariate analysis LHS, stay in ICU, age, days in delirium, previous QOL, persistent ICUAW and site after being discharged were significant factors (Table 4). Other variables such as gender, severity of illness on admission (APACHEII), presence of comorbidities (Charlson score), cause of MV, days of MV or reason for admission had no significance. Post ICU rehabilitation (patients discharged with home care or to LTACH) was prescribed according to medical decision. Most of these patients were dependent on DLA. We hypothesize that rehabilitation was finally indicated to more severe patients at discharge from hospital; that may be the reason why it showed an increase on the OR for worsening QOL (Table 4).

In the multivariate model (Table 5) it was observed that the variables that proved to be independent predictors of deterioration of quality of life were: LHS≥21 days, age≥50, previous poor quality of life and persistent ICUAW. The same individual analysis can be performed taking into account the results at one year after discharge. When comparing individual differences between baseline index and the index after a year we found that 41.8% (n=23) continued with worsening QOL [median baseline index 0.883 (IQR0.742–1.0)] and median after a year [0.559 (IQR0.229–0.770)], while the rest (58.2%) had recovered their previous QOL [median baseline index 0.692 (IQR0.486–0.931) and median at one year 1.0 (IQR1.0–1.0)]. The only two predictor of worsening QOL after a year were LHS≥21 days [OR12.55 (3.08–51.24)] and previous poor QOL [OR0.23 (0.06–0.90)].

If we analyze the results of Barthel index individually, in the same way we did with the score of quality of life (Barthel on admission – Barthel at discharge) we found that 44.3% of patients worsened their dependence on Basic activities of daily life at 3 months after discharge, while 55.7% had no changes. Risk factors for worsening Barthel score at 3 months after discharge can be observed in Table 6 and they are similar to those described for worsening of QOL. One year after discharge, 15 patients (27%) had a lower value relative to Barthel score regarding admission, while 40 (72.7%) had regained independence in ADL. The only predictor of worsening Barthel score after a year was LHS≥21 days [OR10.83 (2.15–54.70)].