Factors related to interrupted inpatient stroke rehabilitation due to acute care transfer or death

1 Introduction

Stroke is a leading cause of death and disability in Thai population (Bundhamcharoen, Odton, Phulkerd, & Tangcharoensathien, 2011). Statistics from 14 rehabilitation centers in Thailand reported that 37.5% of patients in rehabilitation wards were stroke patients, which makes stroke the second most common condition after spinal cord disorders (Kuptniratsaikul et al., 2016). The goals of rehabilitation medicine are to help patients achieve the highest level of independence and the best quality of life that they can in spite of their physical disabilities and impairments. Inpatient rehabilitation programs provide patients with intensive training from a multidisciplinary care team that includes physiatrists, physical therapists, occupational therapists, speech therapists, rehabilitation psychologists and recreational therapists. This multidisciplinary approach seeks to address and improve deficiencies in all aspects of physical function and activities of daily living. A major obstacle in inpatient rehabilitation settings is severe medical complications that require transfer to acute care facilities. The rates of medical complications in stroke inpatients were reported to range from 44% to 75% (Doshi, Say, Young, & Doraisamy, 2003; Hung, Tsay, Chang, Leong, & Lau, 2005; Kitisomprayoonkul, Sungkapo, Taveemanoon, & Chaiwanichsiri, 2010; Kuptniratsaikul, Kovindha, Suethanapornkul, Manimmanakorn, & Archongka, 2009; Roth, Lovell, Harvey, Heinemann, Semik & Diaz, 2001), and 3.2 to 19% of stroke inpatients suffered medical complications that required immediate treatment that was then followed by patient transfer to an acute care facility (Doshi et al., 2003; Hung et al., 2005; Kitisomprayoonkul et al., 2010; Roth et al., 2001; Stineman, Ross, Maislin, Fiedler, & Granger, 2003). Reported causes of transfer or death included pneumonia (Doshi et al., 2003; Hung et al., 2005; Kitisomprayoonkul et al., 2010; Roth et al., 2001), cardiac diseases (Hung et al., 2005; Kitisomprayoonkul et al., 2010; Roth et al., 2001), cerebrovascular diseases (Hung et al., 2005; Kitisomprayoonkul et al., 2010; Roth et al., 2001), deep vein thrombosis (Kitisomprayoonkul et al., 2010; Roth et al., 2001), and upper gastrointestinal bleeding (Roth et al., 2001).

Medical complications are associated with multiple disadvantages that adversely affect patients, families, and hospitals. For patients, medical complications interrupt the rehabilitation process, and this has the effect of slowing or stalling patient recovery. In addition, prolonged hospitalization increases the risk of hospital-acquired infection and deconditioning, both of which can cause further complications. Families are also affected because they need to devote more time and money to the patient’s recovery. The downside of these complications for hospitals includes extended occupancy of inpatient beds and the increased costs associated with extended care. Improved understanding of these complications, their prevalence, and factors related to their occurrence may help to minimize the incidence of medical complications, reduce patient transfers, and optimize both healthcare costs and patient outcomes.

A 2001 study identified the following factors as being associated with medical complications that require transfer to acute care facilities during stroke rehabilitation: those were a National Institutes of Health Stroke Scale (NIHSS) score less than 5; presence of feeding tube; presence of Foley catheter; high white blood cell count; abnormal hemoglobin level; history of cardiac arrhythmia; history of pneumonia; and, onset to rehabilitation admission interval greater than seven days (Roth et al., 2001). A study that was published in 2003 investigated factors that were specifically associated with medical complications that required transfer (Stineman et al., 2003). They identified cognitive problems, greater number/degree of disabilities/disability at admission, and duration of onset to rehabilitation admission greater than two weeks as being significantly associated with increased rate of transfer (Stineman et al., 2003).

Few studies have investigated association between clinical factors and interrupted rehabilitation due to acute care transfer or death during stroke rehabilitation, and none of those studies were performed in Thailand. Accordingly, the aim of this study was to identify medical complications leading to patient transfer to acute care facilities and to determine clinical factors that independently predict interruption of rehabilitation due to acute care transfer or death.

2 Materials and methods

This retrospective chart review included stroke patients that were admitted to the Department of Rehabilitation Medicine, Faculty of Medicine, Siriraj Hospital, Mahidol University, Bangkok, Thailand for inpatient rehabilitation during the January 2012 to December 2017 study period (we began our study in 2012, because medical records were completely recorded since that year). Siriraj Hospital is Thailand’s largest university-based national tertiary referral hospital. To be eligible for inclusion, subjects had to be aged older than 18 years, and had to have been diagnosed with either ischemic or hemorrhagic stroke by a neurologist. The protocol for this study was approved by the Siriraj Institutional Review Board (SIRB) (COA no. Si 473/2017) and was in accordance with the ethical principles stated in the latest version of the Declaration of Helsinki (WMA, 2018). The study was written according to the STROBE criteria.

Patients were divided into either the case or control group. The case group included all stroke patients with an interrupted rehabilitation program due to acute care transfer or death. The control group consisted of randomly selected patients admitted for inpatient stroke rehabilitation during the same month and who did not have an interruption of rehabilitation program. The case to control ratio established for this study was 1 to 5. Control group patients were randomly selected using a simple computerized randomization technique.

3 Results

A retrospective chart review of patients admitted to the inpatient rehabilitation unit at Siriraj Hospital from 2012 to 2017 showed that 995 of 1,827 patients (54.5%) were stroke patients. Thirty-four of those stroke inpatients developed medical complications interrupting their rehabilitation. Of those, 32 patients required transfer to an acute care facility for treatment and two patients died. As there were 34 patients in the case group, we recruited 170 control group subjects in accordance with the case to control ratio of 1 to 5 that we established for this study. Therefore the total number of included patients was 204.

Patient demographic and clinical data compared between groups are presented in Table 2. Stroke characteristics, including type of stroke (ischemic vs. hemorrhagic), side of weakness, and stroke episode (first vs. recurrent), were similar between groups. All but three patients had a caregiver. The gender distribution was comparable between groups. The mean age was significantly lower in the control group than in the case group (64.3±12.0 vs. 75.7±9.8 years; p < 0.001).

The number of patients with interrupted inpatient stroke rehabilitation in this study was 34. The types and frequencies of medical complications interrupting stroke rehabilitation are shown in Table 3. Respiratory tract infection was found to be the most commonly occurring complication (29.4%), followed by intracranial hemorrhage (11.8%), recurrent ischemic stroke (8.8%), ischemic heart disease (8.8%), and seizure (8.8%). Two of the 34 patients died, with both deaths caused by intracranial hemorrhage.

Of all four cases with intracranial hemorrhage, two cases were primarily diagnosed with ischemic stroke and two were with hemorrhagic stroke. Three of the four cases developed new hemorrhagic stroke that was not related with previous etiology and one suffered hemorrhagic transformation.

The results of univariate and multivariate analyses are shown in Table 4. Univariate analysis revealed that the following seven factors had a p-value less than 0.2: history of pneumonia crude odds ratio (OR) 2.06 95% CI: 0.79, 5.37), presence of feeding tube crude odds ratio (OR) 4.35; 95% CI: 2.02, 9.37), presence of anemia crude odds ratio (OR) 3.33; 95% CI: 1.54, 7.19), age crude odds ratio (OR) 1.10; 95% CI: 1.05, 1.14), BI_ad crude odds ratio (OR) 0.88; 95% CI: 0.80, 0.96), TMSE score crude odds ratio (OR) 0.91; 95% CI: 0.85, 0.98), and serum albumin level crude odds ratio (OR) 0.17; 95% CI: 0.06, 0.47).

Of the seven factors from the univariate analysis, only history of pneumonia, presence of feeding tube, presence of anemia, age and BI_ad were included in multivariate analysis. TMSE score and serum albumin level were excluded from further analysis due to a high percentage of missing data. Multiple logistic regression analysis revealed presence of feeding tube [adjusted odds ratio (AOR): 3.94, 95% CI: 1.30–11.96; p = 0.015)], presence of anemia (AOR: 2.62, 95% CI: 1.04–6.57; p = 0.040) and age (AOR: 1.08, 95% CI: 1.04–1.13; p < 0.001) to be independently associated with the interrupted stroke rehabilitation. BI_ad did not maintain its significant association in multivariate analysis (p = 0.75).

Authors performed test of interaction between presence of a feeding tube and respiratory tract infection (excluding one case of influenza A infection, and one case of nontuberculous mycobacterial infection) in order to investigate for its association. This study revealed patients with a feeding tube to be at significantly greater risk for developing respiratory tract infection (OR: 7.50, 95% CI: 1.47–38.28; p = 0.005) (data not shown).

This is the first study from Thailand that investigated factors significantly associated with interruption of inpatients stroke rehabilitation. Rate of interruption of rehabilitation program in this study is 3.4% which considered to be at the low end when compared with the 3–19% range reported by previous studies (Doshi et al., 2003; Hung et al., 2005; Kitisomprayoonkul et al., 2010; Roth et al., 2001; Stineman et al., 2003). Difference in rate of acute care transfer between each study could be from multiple factors. One of the main factors is the type rehabilitation setting factor. A higher rate of transfer was reported for dedicated rehabilitation centers (Kitisomprayoonkul et al., 2010; Roth et al., 2001), and a lower rate was reported for rehabilitation units located within hospitals (Hung et al., 2005; Doshi et al., 2003).

Hung et al. found a transfer rate (including one death case) of 3.2% (Hung et al., 2005) which is almost identical to the rate found in our study. Both our study and the study by Hung et al. were conducted in a medical center where onsite medical physicians were available. In this setting, patients with active medical problems would normally receive treatments from consulting physicians during their admission in the rehabilitation unit. This factor seems to increase the threshold relative to why and when to transfer a patient to an acute care unit. In this study, we found that almost all referred patients had severe or life-threatening conditions (e.g., on vasoactive drugs, impending intubation, and/or was already intubated) that prevented them from continuing to attend their rehabilitation program. The next possible factors could be from patient-related factors. This refers to patient’s baseline status when admitted to rehabilitation unit including level of dependency, nutritional status, and co-morbidities. Low FIM score and higher comorbidity score were related to transfer to acute care facility (Civelek et al., 2016) and low albumin level was found to be related with medical complications (Roth et al., 2001). This patient-related factors would depend largely on the admission criteria and admission screening process of each hospital. The third factors could be from physician-related factors such as judgement and experience on when patient should be transferred to acute care facility.

In those transferred cases, respiratory tract infection (which included pneumonia, tracheaobronchitis, and aspiration pneumonia) was the most common type of complication causing transfers. This is consistent with many previous studies that found pneumonia to be among the most common causes of patient transfer to acute care (Doshi et al., 2003; Hung et al., 2005; Kitisomprayoonkul et al., 2010; Roth et al., 2001). A prospective study in stroke patients (most with ischemic type) reported that one-third of patients suffered from pneumonia, and 34% of those pneumonia patients succumbed to their conditions (Sellars et al., 2007). Not surprisingly, pneumonia was reported to be the most common cause of mortality of all medical complications following stroke (Heuschmann et al., 2004).

Four patients in this study had intracranial hemorrhage (4/34, 11.8%), and three patients had recurrent ischemic stroke (3/34, 8.8%). Two of the patients with intracranial hemorrhage died. One of those two cases was diagnosed with hypertensive emergency with intracranial hemorrhage. Interestingly, three of the four cases with intracranial hemorrhage had short interval of onset to rehabilitation admission (i.e., 5, 11, and 16 days), which may be a factor that warrants closer investigation. Although this study did not find significant association between onset to admission interval and severe medical complications, close observation and blood pressure monitoring in stroke patient undergoing rehabilitation may help to prevent hemorrhagic stroke.

In addition to identifying the most common complications that interrupted rehabilitation and precipitated the transfer of a patient to an acute care facility, we investigated for factors that were independent predictors of interrupted stroke rehabilitation. Multivariate analysis revealed presence of feeding tube, presence of anemia and old age to be significantly independently associated with interrupted stroke rehabilitation due to acute care transfer or death.

The first one is nasogastric tube. Tube feeding is generally indicated in patients with severe swallowing problems, and in patients with decreased level of consciousness who were at risk of malnutrition or aspiration. Feeding tube was found to be one of clinical factors associated with patient transfer to an acute care facility (Roth et al., 2001). Respiratory tract infection in stroke patients was commonly thought to be caused by swallowing-related problems and aspiration (Langdon, Lee, & Binns, 2009). In this study, we found that stroke patients with tube feeding were 3.94 times more likely to develop complications interrupting their rehabilitation and as high as 7.5 times when considering respiratory problems as a complication alone. Of the three significant factors identified in our study, feeding tube seems to have the strongest association with interrupted stroke rehabilitation. Stroke patients with a feeding tube should be carefully assessed and managed to prevent further complications.

In order to prevent respiratory tract infection, a screening tool was proposed to identify patients at risk for developing pneumonia (Sellars et al., 2007). The tool contains the following five risk factors: age over 65 years, dysarthria or aphasia, score on modified Rankin Scale ≥4, score on Abbreviated Mental Test <8, and difficulty with the water swallowing test. Patients with two or more of these risk factors would likely be at the increased risk of the occurrence of pneumonia, with 90.9% sensitivity and 75.6% specificity (Sellars et al., 2007). Good oral hygiene care, prevention of reflux (Langdon et al., 2009), and good hand washing technique by healthcare workers (Boyce, White, Spruill, & Wall, 1985) are also highly recommended as factors that prevent respiratory tract infection.

The next independent predictor was presence of anemia. It was found to be one of several risk factors associated with the development of severe medical complications that interrupted rehabilitation (Roth et al., 2001). Another previous study also found that patients with anemia had higher rate of complications (Chan & Ganasekaran, 2015). The prevalence of anemia tends to increase with age (Culleton et al., 2006; Woodman, Ferrucci, & Guralnik, 2005), and the majority of stroke patients are of older age. In the present study, we found a prevalence of anemia of 36% in the control group, and 65% in the case group. Anemia was also reported to be independently associated with higher morbidity and mortality among the elderly (even in community-based elderly), and it could increase the risk of death (Culleton et al., 2006). Additionally, the more severe the anemia, the higher the hazard ratio of all-cost mortality (Culleton et al., 2006). The mechanism of how anemia influences mortality remains unknown. Culleton, et al. proposed an optimal hemoglobin range in older people of 13–15 mg/L in women and 14–17 mg/L in men. Though no definite consensus has been established regarding the optimal level of hemoglobin in stroke patients undergoing intensive rehabilitation program, we recommend that physicians perform an overall assessment and then provide appropriate treatment, in addition to careful monitoring of anemic patients.

The last clinical factor that we identified as an independent predictor is old age. Not surprisingly, people with old age are at higher risk for developing severe medical complications. However, not all studies that investigated a relationship between age and the development of complications found a significant association (Civelek et al., 2016; Dromerick & Reding, 1994; Hung et al., 2005; Kitisomprayoonkul et al., 2010; Kuptniratsaikul et al., 2009). Another previous study found no relationship between age and transfer rate (Stineman et al., 2003). Yet, Davenport et al. reported a relationship between increasing age and medical complications (Davenport, Dennis, Wellwood, & Warlow, 1996), and Stineman and colleague reported association between increasing age and death rate (Stineman et al., 2003). Older people normally have reduced body reserves and develop frailty (Clegg, Young, Iliffe, Rikkert and Rockwood, 2013), and this may contribute to their increased susceptibility to medical problems, and to subsequent progression to more severe complications. Our study revealed old age as one of three independent risk factors but with the least odds ratio when compared with the other two factors; presence of feeding tube and anemia. With an odds ratio of only 1.08, old age may not have clinical significance. However, it would still be beneficial that older patients should be closely monitored so that early interventions can be initiated to prevent progression of complication from non-severe to severe.

Although BI_ad was not found to be independently related with severe medical complications in this study, it was found to be significant in univariate analysis. Previous studies reported low ADL performance to be a risk factor for both medical complications (Dromerick & Reding, 1994; Kitisomprayoonkul et al., 2010) and transfers to acute care facilities (Stineman et al., 2003). There are many different ways to measure patient functional limitations. Some studies measured patient ability to perform daily activities using BI_ad (Dromerick & Reding, 1994; Kuptniratsaikul et al., 2009) or ADLs score (Stineman et al., 2003). Other studies measured neurological impairment using NIHSS score (Roth et al., 2001) or limb movement and hand functions (Hung et al., 2005). Patients with greater neurological deficits usually had more severe functional limitations. Roth et al. reported that the degree of neurological impairment at rehabilitation unit admission has the greatest ability in predicting the occurrence of both medical complications and transfers to acute care facilities (Roth et al., 2001). Patients who presented with more severe neurological impairment tended to have more frequent and more severe medical complications (Roth et al., 200). The possible reason that BI_ad was not found to have significant association in this study might be from low sample size that limit power to detect significant association.

We concluded that some of the severe complications that were observed in this study were preventable, either from happening in the first place or from progressing to a severe complication during rehabilitation period. An example of the aforementioned is respiratory tract infection –particularly pneumonia. Patients with feeding tube or swallowing problems should receive careful assessment and proper management to prevent aspiration pneumonia. The other factor is anemia. Early detection and management of anemia could possibly reduce rate of severe complications.

This study has some mentionable limitations. First, the retrospective nature of this study made it vulnerable to missing or incomplete data. Second, the actual transfer rate that we reported may be low due to the fact that our rehabilitation unit is within Thailand’s largest national tertiary referral hospital. As such, our center receives severe or complicated cases that are referred from hospitals located across Thailand, causing hospital beds in acute care facilities to be fully occupied, which meant that rehabilitation patients in need of transfer to acute care could not be transferred. Third and last, the size of our study group was relatively low, and this may have limited the statistical power of our study to identify all significant differences or associations. A study with a larger patient population is needed to confirm the findings of this study, to identify additional risk factors, and to better understand the variables that contribute to the development of severe complications in this vulnerable patient population during inpatient rehabilitation.

5 Conclusion

The rate of interrupted inpatients stroke rehabilitation due to acute care transfer or death was 3.4% and respiratory tract infection was the most common cause. Presence of feeding tube, anemia and old age were identified as independent predictors of interrupted inpatient stroke rehabilitation due to acute care transfer or death. Interventions to reduce the occurrence of medical complications should be implemented in order to continue the rehabilitation process and to reduce the patient transfer rate.

Conflict of interest

None of the authors declare a personal or professional conflict of interest.

Funding

This study was funded by a grant from the Faculty of Medicine Siriraj Hospital, Mahidol University (grant no. R016131001).