Motor vehicle crash (MVC) is a major public health problem worldwide and contributes to a large burden of death, disability, and economic loss (Nguyen et al., 2019). The consequences of MVC can range from minor sequelae to death, or physical disabilities, which burden the health care system with prolonged and costly hospitalizations, high rates of hospital readmission, followed by long-term rehabilitation programs (Paiva et al., 2015).
Readmissions are responsible for significant increases in costs for patients and for the public and private health care systems, in addition to higher hospital mortality rates (Oliveira et al., 2019), and are defined as a patient's admission to the same health care facility two or more times for a period of 1 year after discharge (Paiva et al., 2015).
It is estimated that psychological and socioeconomic factors and issues related to the health system are directly linked to the occurrence of hospital readmissions (Oliveira et al., 2019). In addition, these events function as indicators for the evaluation of care provided by health institutions, because they can be linked to the quality of care offered and the inadequate management of comorbidities during hospitalization, as well as reflect differences in access to health services and, consequently, inequalities between users (Khawaja et al., 2012; Ricci et al., 2016).
Hospital readmissions may be classified as planned, which are those readmissions necessary for continuity of treatment or diagnostic evaluation and, eventually, that are grouped into potentially avoidable and nonavoidable. Readmissions considered avoidable are commonly associated with poor care indicators, low resolution of the initial problem, inadequate postdischarge care, and unstable therapy at the time of discharge. Moreover, it is believed that the shorter the interval between the first and subsequent admissions, the greater the likelihood that readmission is due to an avoidable complication (Borges & Turrini, 2011; Ricci et al., 2016).
Although research reports that 12%-75% of readmissions in hospital settings can be avoided (Benbassat & Taragin, 2000; Tien-Ching et al., 2017), there is a scarcity of studies investigating these events, especially in MVC victims, with a methodological design that allows identifying their predictive factors. Therefore, this study was relevant because it can support the planning of health actions aimed at vulnerable individuals and those at a higher risk for the event, in order to reduce its occurrence and, consequently, its repercussions for the patient, the health system, and society in general.
KEY POINTS
* Of the 200 contacted, 50 (25.0%) patients reported hospital readmission.
* The length of hospital stay upon readmission ranged from 1 to 528 hr, with an average of 66.71 hr.
* Sociodemographic and pre- and post-hospital variables predict hospital readmission for MVC victims.
OBJECTIVE
The objective of this study was to identify the predictors of hospital readmission in MCV victims within 1 year after discharge.
METHODS
This was a prospective cohort study about the predictors of hospital readmission of individuals who suffered MVC and were admitted to a regional hospital located in the municipality of Jequie, Bahia, Brazil. This hospital was chosen as the study collection site because it is a reference in care for MVC victims in the health micro-region, which comprises 26 cities in the interior of the state.
The cohort consisted of 241 individuals recruited at baseline of the study, which took place from February to June 2019, who were followed up for 12 months prospectively after hospital discharge.
In this study, accident victims aged 16 years or older who agreed to participate in the longitudinal stage of the research were included. The following were excluded from the follow-up stage: those injured in the baseline who were only under observation, those whose cause of hospitalization was not current MVC injuries, and cases that had the following evolution: transfer, evasion, or death.
Data from the baseline were collected by interviewing the injured and consulting the medical records at the hospital. In the follow-up stage, which took place after the participant's discharge, the data were collected through interviews that took place every 2 months by telephone contact. Every 2 months, during the interview, the post-trauma repercussions were investigated, such as return to work, presence of sequelae, hospital readmission due to MVC, among others.
In this study, hospital readmission was considered as a dependent variable, which was estimated from the following questions to the injured person: "In the past 2 months, have you returned to any health care facility as a result of the MVC?" "In the past 2 months, when you returned to the health care facility, was there a need for observation or hospitalization?"
Readmission was considered both cases of readmission and those who stayed only for observation in a hospital unit, be it public or private, due to some complication of the accident. The return to the hospital only for medical review was not considered readmission.
The independent variables consisted of sociodemographic, accident, and clinical characteristics and those related to pre-, intra-, and post-hospital care that were organized in blocks for multivariable analysis, described as follows:
Block I distal: Sex (female; male), age (16-29 years; 30-59 years; 60 years or older), race/color (White; non-White), education (higher education; elementary to high school; illiterate), income Brazilian Real (>998.00; <=998.00), health insurance (yes; no), occupation (rural worker; merchant; construction worker; motorcycle taxi driver/driver; other), occupational attachment (yes; no).
Block II intermediate: Type of MVC (pedestrian; other; two-wheeled vehicle; four-wheeled vehicle), day of MVC (weekdays; weekends), shift of MVC (daytime; nighttime), speeding (no; yes), alcohol consumption (no; yes), multiple injuries (no; yes); severity of trauma (mild; moderate; severe).
Block III proximal: Pre-hospital care (yes; no), surgery (no; waiting; yes), number of surgeries (one; more than one), intensive care unit (ICU) (no; yes), length of stay (<=6 days; >6 days), postdischarge infection (no; yes), rehabilitation treatment (yes; no).
The severity of trauma was estimated from the severity of injuries, classified according to the Injury Severity Score (ISS). From the sum of the squares of the ISS values of the three most severe injuries, regardless of the body region, we obtained the New Injury Severity Score (NISS), which is the index that evaluates the severity of trauma, classifying it as follows: mild (<16), moderate (16-24), or severe (>=25) (Stevenson et al., 2001). In this study, this variable was dichotomized, grouping the moderate to severe cases as follows: mild (NISS <16) and severe (NISS >=16).
Data analysis was performed in three stages. Initially, the characterization of the cases was performed by means of descriptive statistics, which were presented in absolute and relative frequencies. Subsequently, the predictors of the outcome were verified through the crude and adjusted estimates of the relative risk (RR) and their respective 95% CIs.
In the multivariable analysis, logistic regression was initially used, where only the variables that presented in Pearson's chi-square test p<= .20 in the crude analysis were included, following the order of an established hierarchical model (Figure 1), which demonstrates that the variables of the higher (distal) levels interact and determine the variables of the lower (proximal) levels. The effect of each variable on the outcome was controlled by the same and higher level variables in the model.
Then, the adjusted RR and respective 95% CI of the variables that remained in the model were estimated and, only for the conversion of the measure of association, Poisson regression with robust variance was used, which is the technique indicated in this type of study (Coutinho et al., 2008). The estimation of the final model occurred on the basis of the statistical assumptions of significance (p < .05), as well as the theoretical importance of each variable for the model. For data analysis, we used the statistical program STATA, Version 12.0.
This research was submitted and approved by the Research Ethics Committee of the State University of Southwest Bahia (REC/UESB), under Opinion No. 2,416,824/2017.
RESULTS
In baseline, 245 hospitalizations of individuals who suffered MVC were identified, of which 241 met the criteria for the follow-up stage. Of the 241 followed up, 200 were contacted during the 12-month period after hospital discharge and comprised the population of this study.
Of the 200 patients contacted, 50 (25.0%) reported hospital readmission, of which 66.0% occurred in the first 60 days after discharge. The length of hospital stay at readmission ranged from 1 to 528 hr, with a mean of 66.71 hr (SD = 121.16).
Table 1 describes the characterization of the cases regarding sociodemographic, accident, and care aspects.
It was evidenced that both in the group of cases that were not readmitted and in those with new admission, the individuals were mostly male, aged 30-59 years, of non-White race/color, with schooling between elementary and high school, monthly income 998.00 or less, did not have health insurance, and had other professions as occupation. As for the occupational relationship, it was noted that the readmitted cases were mostly formal workers with employment relationship (52.6%), differing from the group without readmission.
In both groups, events involving two-wheeled vehicles were more frequent, occurred in a greater proportion on weekdays, during the day shift, and were reported not to be speeding or drinking alcohol. Most cases in both groups presented a single injury, that is, they did not have multiple injuries and the trauma was classified as mild severity.
Regarding the care aspects, it was observed that in the groups with and without readmission, there was a predominance of the injured who did not receive pre-hospital care, who were waiting for surgery, who underwent only one surgery, who did not stay in the ICU, who had a hospital stay of up to 6 days, and who did not have infection after discharge. Regarding rehabilitation treatment, it was found that most of the individuals who were readmitted were receiving the treatment (60.0%) different from the no readmission group, in which the largest proportion of individuals were not receiving the treatment (54.7%).
In the bivariate analysis, hospital readmission was statistically significant and a lower incidence of readmission in MVC involving individuals with schooling between elementary and high school (RR = 0.49; 95% CI [0.25, 0.96], p = .050). In events of moderate/severe severity (RR = 1.77; 95% CI [1.06, 2.96], p = .037), which victimized illiterate individuals (RR = 1.29; 95% CI [3.38, 4.88], p<= .050), who did not receive pre-hospital care (RR = 1.72; 95% CI [1.00, 2.95], p = .039), who were admitted to the ICU (RR = 2.38; 95% CI [1.30, 4.35], p = .028), and who had postdischarge infection (RR = 2.34; 95% CI [1.47, 3.72], p = .001), there was a higher incidence of readmission after discharge (Table 2).
The variables that presented p<= .20 and that followed for the hierarchical multivariate model were as follows: gender, education, health insurance, occupation, alcohol consumption, multiple injuries, trauma severity, pre-hospital care, number of surgeries, ICU, postdischarge infection, and rehabilitation treatment.
After verifying the best adjustments, and considering the theoretical importance of each variable, the variables sex, severity of trauma, postdischarge infection, and rehabilitation treatment remained in the final model and were shown to be predictors of hospital readmission in victims of MVC (Table 3).
It was evidenced that injured males have a 42% lower risk of being readmitted after discharge when compared with females (RR = 0.58; 95% CI [0.36, 0.95], p = .033).
Events classified with higher severity (moderate/severe) showed a 77% higher risk for readmission compared with mild trauma occurrences (RR = 1.77; 95% CI [1.03, 3.02], p = .036). Similarly, those who did not receive pre-hospital care showed 2.14 times higher risk for hospital readmission after discharge when compared with those who did (RR = 2.14; 95% CI [1.24, 3.69], p = .006).
Individuals who reported postdischarge infection showed 2.14 times higher risk of returning to the hospital unit compared with those who did not present this complication (RR = 2.14; 95% CI [1.37, 3.36], p = .001). Likewise, the injured who were receiving rehabilitation treatment showed a 64% higher risk of being readmitted compared with those who were not undergoing treatment (RR = 1.64; 95% CI [1.03, 2.62], p<= .001).
DISCUSSION
This study is one of the pioneers in investigating the predictors of hospital readmission in victims of MVC, in which it was found that individual, trauma, and care-related factors are associated with the event.
It was found that 25.0% of the injured patients were readmitted to the hospital during the period of 1 year after discharge. This incidence is higher than that estimated in a university hospital in the Minas Gerais triangle (17.4%) 1 year after discharge (Paiva et al., 2015) and in a study conducted in the United States 30 days after discharge (12.2%) (Parreco et al., 2018).
Readmission rates function as predictors for assessing the performance of hospital activities, the emergence of postdischarge complications (Reis et al., 2015), as well as reflecting the access to health services. However, these rates are not always associated with quality and access to services, as they may be linked to medical complexity and individual patient factors, such as socioeconomic conditions and disease severity. On the other hand, readmissions that occur in a planned manner are referred to as related to the quality of the hospital service provided (Fischer et al., 2015; Ziaeian & Fonarow, 2016).
Among the most commonly reported reasons in the literature for returning to hospital after discharge are the difficulties in accessing consultations at the primary level, as well as the idea that treatment is focused on the aggravation of the disease rather than its chronicity, which increases the demand for highly complex services (Mendes, 2010; Sousa et al., 2014). In the case of MVC, the high incidence of hospital readmission can be attributed to the injuries caused, which are mostly orthopedic traumas that require surgical procedures, with a scheduled return for reassessment, new surgery, or due to the emergence of some complication.
In this study, the predictors of hospital readmission after MVC were gender, trauma severity, pre-hospital care, access to rehabilitation treatment, and the occurrence of infection after discharge.
Being male was shown to be a protective factor for hospital readmission in MVC victims, with a 42% lower risk when compared with females. This finding differs from those found in studies conducted in Taiwan (Tien-Ching et al., 2017) and France (Roger et al., 2019), in which women had a significantly lower risk of being readmitted after hip or knee fracture.
It is believed that differences in lifestyle and behavior may explain this result, because culturally men are less likely to seek health services than women, and men are often the sole breadwinners in the family, so they return to work before they have fully recovered and do not seek health services again (Gomes et al., 2007).
Individuals involved in more severe MVC had a higher risk of being readmitted within 1 year after discharge compared with mild occurrences. Similar findings were found in a study in the United States, which found that MVC victims with higher severity injuries (ISS >15) had a higher risk of readmission (Parreco et al., 2018). In these cases, it is believed that the return to the hospital is related to the continuity of care, such as the need for a new surgical procedure, as well as the greater possibility of these individuals presenting complications after discharge.
Among the most frequent complications in victims of orthopedic trauma, which is a very common injury in MVC, it is worth mentioning surgical wound infection and osteomyelitis, which are serious clinical conditions that require surgical reoperation and influence patients' morbidity and mortality (Castro et al., 2013; Paiva et al., 2015). In this study, the presence of infection after discharge was shown to be a predictor of hospital readmission, similar to that evidenced in other studies (Paiva et al., 2015; Roger et al., 2019).
It is estimated that the occurrence of infection is related to the severity and complexity of the injuries, the number of surgeries performed at the injury site, and clinical risk factors such as preexisting diseases (Thu et al., 2005). This clinical condition is a serious complication for patients, health care professionals, and hospital institutions, as it doubles readmission rates, prolongs the length of stay, and increases care costs to more than 300%. Moreover, they cause important physical limitations, which significantly reduce the quality of life of those affected (Paiva et al., 2015).
The injured who did not receive pre-hospital care had a higher risk of being readmitted when compared with those who did. No studies were found that evaluated the influence of this variable on the incidence of readmission; however, it is believed that this result is justified by the benefits of complete and quality care to the trauma patient, which also involves pre-hospital care, thus avoiding the worsening of injuries, future complications, and, consequently, the need for new hospitalization.
The severity of trauma and the occurrence of complications, such as infections, may also explain access to rehabilitation treatment as a predictor of hospital readmission in MVC victims, because individuals who suffered more severe injuries and/or had complications after discharge usually require rehabilitation services (Roger et al., 2019), such as physical therapy, psychotherapy, nutrition therapy, and occupational therapy, leading them to return to the health service.
LIMITATIONS
A limitation of this study is the scarcity of studies that have also investigated the predictors of readmission in victims of MVC, which hindered a better comparison of the data found.
Therefore, and considering that readmissions in some situations are possible to prevent, it is worth emphasizing the role of the multiprofessional team, especially nursing in hospital discharge planning, as an indispensable tool for comprehensive care during hospitalization and after discharge (Paiva et al., 2015). In addition, health education is an essential strategy for guiding the care that will be provided to patients at home in order to avoid unplanned readmissions and facilitate the early identification of signs of postdischarge complications (Paiva et al., 2015).
CONCLUSION
The variables gender, severity of trauma, pre-hospital care, postdischarge infection, and rehabilitation treatment predict hospital readmission in individuals who suffered MVC within 1 year after discharge. Thus, being male is a protective factor, whereas trauma of greater severity, not having received pre-hospital care, infection after discharge, and access to rehabilitation treatment are risk factors for the occurrence of the event.
It is noteworthy that these results can be an important indicator of the quality of care provided to the victims of these events, as well as their costs. However, new studies should be conducted focusing on these aspects.
REFERENCES