Patient characteristics associated with delayed neurological improvement following acute ischemic stroke: A systematic review and meta-analysis

Abstract

Background

Delayed neurological improvement (DNI) is a phenomenon that involves patient improvement in the absence of early neurological change following treatment for acute ischemic stroke. The patient characteristics associated with this condition are largely unexplored.

Methods

Following the PRISMA guidelines, a systematic review of the English language literature was conducted using PubMed, Embase, Web of science, and Scopus. We calculated pooled odds ratios (ORs), mean differences (MDs), and their corresponding 95% confidence intervals (CIs) to test the association between patient characteristics and achievement of DNI.

Results

Seven studies, with 3266 patients, were included in our analysis. All studies reported a different definition of DNI, with five studies focusing on rates of good functional outcome at 90 days post-treatment in the absence of early neurological improvement. Use of intravenous thrombolytics was associated with increased rates of DNI (OR 1.96, 95% CI 1.28 to 3.00; p = 0.002). Atrial fibrillation was associated with decreased rates of DNI (OR 0.69, 95% CI 0.57 to 0.82; p < 0.001), as was hypertension (OR 0.66, 95% CI 0.53 to 0.83, p < 0.001), and diabetes mellitus (OR 0.71, 95% CI 0.56 to 0.90; p = 0.005). On average, patients who achieved DNI were 6.30 years younger than their non-DNI counterparts (MD −6.30, 95% CI −9.19 to −3.41; p < 0.001). There were modest associations between male sex and DNI (OR 1.36, 95% CI 1.01 to 1.74, p = 0.042), and smoking and DNI (OR 1.28, 95% CI 1.03 to 1.59, p = 0.027).

Conclusions

DNI is a phenomenon that is not presently well understood. Lack of uniformity among definitions of DNI hinders efforts to explore DNI and the factors associated with its occurrence. Future studies should work to establish a consensus definition of DNI to determine its causes and significance more accurately.

Introduction

Patients can follow different clinical trajectories after treatment for acute ischemic stroke (AIS). The National Institutes of Health Stroke Scale (NIHSS) is the most frequently used tool to gauge AIS severity. The NIHSS is measured upon patient presentation to the hospital, 24 hours post-treatment, and occasionally at other, later time points. ¹ The modified Rankin Scale (mRS) is the most common tool used to categorize functional outcomes. The mRS is usually measured at 90 days post-stroke. ² Though these tools are both used to assess patient clinical status, the relationship between NIHSS scores and long-term mRS scores remains unclear.

Previous literature has mostly examined the utility of early neurological improvement (ENI) and early neurological deterioration as potential predictors of 90-day mRS scores.^3,4 Both ENI and END provide great clinical utility, as they can inform clinicians and patients about their long-term functional outcomes in a short-term timeframe. ENI and END definitions are anchored in the integer change in NIHSS of patients within the first 24 hours post-treatment. There is sparse literature on a patient trajectory that involves delayed neurological change, particularly delayed neurological improvement (DNI). DNI, occasionally referred to “stunned brain,” is different from ENI and END in that it is most often used as an outcome, rather than a predictor, of clinical status following treatment for AIS. ⁵ Furthermore, DNI is not as well studied as ENI and END, and there are multiple definitions of DNI.

To examine existing literature on DNI, we performed a systematic review and meta-analysis exploring different definitions of DNI and patient characteristics associated with this phenomenon.

Methods

Search strategy

On 8 August 2022, following the PRISMA guidelines for performing systematic reviews, a systematic literature review of the English language literature was conducted within the Nested Knowledge Autolit software per the drafted protocol, from inception, using PubMed, Embase, Web of Science, and Scopus. ⁶ Based on each database, different combinations of possible keywords and/or MeSH terms were used for that purpose. Keywords and MeSH terms included: “thrombectomy” AND “clinical improvement,” “functional outcomes,” and others. The full search strategy is provided in the Supplement. Moreover, we did an extensive manual search through the references of the included articles to retrieve any missed papers.

Screening process

We included all original studies fulfilling our pre-determined PICO. The population was patients with AIS, Intervention treatment with endovascular therapy, there was no Control group, and the Outcomes of interest were the presence or absence of DNI and associated patient characteristics. We excluded papers where patients did not have AIS, papers that did not report both the presence and absence of DNI, review articles, duplicate studies including the same patients presented in other included papers, case reports, case series with less than five patients, and conference abstracts.

Two authors performed the title and abstract screening against the pre-defined criteria. This was followed by a full-text screening of any retained studies of the first screening step. In both stages, the senior author was consulted to resolve any conflicts in the decisions.

Data extraction

Following a pilot extraction, an extraction sheet was built, and the extraction was performed by two authors. The extracted data included study characteristics, baseline data of the included patients, and the aforementioned outcomes of interest. After performing the extraction, a third author performed an extensive revision of the extracted data to avoid any prior mistakes. The data that support the findings of this study are available from the corresponding author upon reasonable request.

Risk of bias

The Newcastle Ottawa Scale was used to assess the risk of bias, with two independent reviewers evaluating all studies. ⁷ The following rating system was used: “Good” quality: 3 or 4 stars in the selection domain AND 1 or 2 stars in the comparability domain AND 2 or 3 stars in the outcome/exposure domain; “Fair” quality: 2 stars in the selection domain AND 1 or 2 stars in comparability domain AND 2 or 3 stars in outcome/exposure domain; “Poor” quality: 0 or 1 star in selection domain OR 0 stars in comparability domain OR 0 or 1 stars in outcome/exposure domain. Two authors evaluated the quality of each study and ratings were adjudicated by a third author, whenever needed.

Statistical analysis

Using R software version 4.1.2, we calculated pooled odds ratios (ORs), mean differences (MDs), and their corresponding 95% confidence intervals (CI). The random-effects model was used to pool the data in the presence of significant heterogeneity; otherwise, a fixed-effect model was adopted. Heterogeneity was assessed using the Q statistic and the I² test, where I² > 50% or P-value < 0.05 were considered significant. ⁸ Because the minimum per examined covariate in meta-regression is ten studies, ⁹ analogous to the traditional rule of thumb used to minimize the risk of overfitting in regression models, ¹⁰ we could not adjust for different co-founders due to the number of included studies.

Results

Search and screening results

Following the removal of 1038 duplicate records, we retrieved 1856 papers for further screening. We excluded 1850 records through the title and abstract screening stage, to retain seven records for full-text screening. One paper was identified via expert recommendation. Finally, seven papers were determined to satisfy our inclusion criteria with the appropriate report of outcomes of interest (Supplementary Figure S1).

DNI definitions

All seven included studies reported unique definitions of DNI. Bang et al. and Talavera et al. defined DNI as mRS 0-2 at 90 days post-treatment in the absence of ENI, and ENI had a different definition in both studies.^11,12 In Bang et al., ENI was defined as an improvement of 40% or greater in NIHSS score within 24 hours post-treatment, ¹¹ while in Talavera et al. ENI was defined as a decrease in NIHSS score ≥4 within 24 hours post-treatment or a 24-hour NIHSS score of 0-1 in patients with a baseline NIHSS score ≤4. ¹² Jadhav et al. defined DNI as mRS 0-2 at 90 days post-treatment in the absence of early functional independence. ¹³ Put et al. defined DNI as improvement in NIHSS score by 41% or greater at seven days post-treatment. ¹⁴ Desai et al. defined DNI as mRS 0-2 at 90 days post-treatment in the absence of mRS 0-2 at discharge. ¹⁵ Kniep et al. defined DNI as mRS 0-2 at 90-days with NIHSS score ≥ 15 at 24 hours. ¹⁶ Cai et al. defined DNI as improvement in NIHSS score ≥10 at seven days post-treatment, or NIHSS score at seven days post-treatment equal to 0-1. ¹⁷ For the purpose of our analysis, all definitions of DNI that included mRS 0-2 at 90 days were reported as “90-day mRS 0-2 with no ENI.”

Study characteristics and risk of bias

Of the seven included studies, six analyzed data that was prospectively collected,^{11–13,15–17} while Pu et al. analyzed data that was retrospectively collected. ¹⁴ Six studies were multicenter studies. The size of the included studies ranged from 45 patients ¹¹ to 1115 patients. ¹⁶ Five of the studies included in our analysis were determined to have “good” quality, while two of the studies were determined to have “fair” quality (Supplementary Table S1). Patient characteristics, such as age, baseline NIHSS, use of intravenous thrombolytics (IV tPA), and co-morbidities are detailed in Table 1. The intervention utilized in five of the included studies was mechanical thrombectomy,^{12,13,15–17} while in the remaining two studies, the intervention was broadly defined as endovascular therapy.^11,14

Table 1.

Characteristics of the included studies.

Author	Year	DNI status	Arm size	Age (SD), mean	Age (IQR), median	Male sex	Baseline NIHSS (SD), mean	Baseline NIHSS (IQR), median	Hypertension	Hyperlipidemia	Diabetes mellitus	Atrial fibrillation	Smoking	IV tPA	Prior stroke	CAD/CHF
Bang et al.	2011	DNI achieved	18	58.9 (16.5)		11	10.9 (6.1)		10		4	8	6	6	2
Bang et al.	2011	DNI not achieved	27	72.3 (13.9)		15	17.3 (5.9)		13		6	14	3	3	7
Pu et al.	2018	DNI achieved	285		63 (53–71)	181		15 (11–19)	173		51	107	82	101	40	62
Pu et al.	2018	DNI not achieved	283		66 (56–74)	174		16 (13–20)	185		54	115	75	87	44	66
Desai et al.	2020	DNI achieved	85	67 (15.4)		41	16.9 (6.4)		57	38	17	25	27	39	11	19
Desai et al.	2020	DNI not achieved	197	74 (13.8)		82	18.1 (5.5)		161	118	55	81	38	59	39	64
Talavera et al.	2021	DNI achieved	32	75.1 (9.8)		23		9 (8–10)	23	19	7	5	5	19
Talavera et al.	2021	DNI not achieved	110	77.7 (11.4)		45		8 (7–10)	72	50	21	39	12	27
Kniep et al.	2022	DNI achieved	61	63.4 (15.9)		35	16.7 (7)		39	16	12	18		42
Kniep et al.	2022	DNI not achieved	1054	76.4 (11.9)		468	17.7 (5.4)		847	358	264	483		490
Cai et al.	2022	DNI achieved	67	62.52 (13.16)		38		15 (9–21)	32		5	25	15	25	11	11
Cai et al.	2022	DNI not achieved	160	68.3 (11.56)		92		14 (7–21)	96		25	72	36	55	37	29
Jadhav et al.	2022	DNI achieved	308	67.2 (13)			16 (5.6)		226	141	58	98	58			72
Jadhav et al.	2022	DNI not achieved	579	71.7 (13.7)			17.8 (5.7)		469	293	166	235	92			144

DNI: delayed neurological improvement; NIHSS: National Institutes of Health Stroke Scale; IV tPA: intravenous thrombolytics; CAD/CHF: coronary artery disease/congestive heart failure; SD: standard deviation; IQR: interquartile range.

Contributory patient characteristics

Use of IV tPA was associated with increased rates of DNI (OR 1.96, 95% CI 1.28 to 3.00; p = 0.002); however, there was significant heterogeneity among studies (I² = 64%, p < 0.017; Figure 1). There was a significant association between atrial fibrillation and decreased rates of DNI (OR 0.69, 95% CI 0.57 to 0.82; p < 0.001; Figure 2). On average, patients who achieved DNI were 6.30 years younger than their non-DNI counterparts (MD −6.30, 95% CI −9.19 to −3.41; p < 0.001); however, there was significant heterogeneity among studies (I² = 75%, p < 0.001; Figure 3). There was a significant association between diabetes mellitus (DM) and decreased rates of DNI (OR 0.71, 95% CI 0.56 to 0.90; p = 0.005; Figure 4). There was a significant association between hypertension and DNI (OR 0.66, 95% CI 0.53 to 0.83; p < 0.001; Figure 5). Male sex was associated with higher odds of achieving DNI (OR 1.36, 95% CI 1.01 to 1.84; p = 0.042; Figure 6). Finally, there was a significant association between smoking and higher odds of achieving DNI (OR 1.28, 95% CI 1.03 to 1.59; p = 0.027; Figure 7).

Figure 1.

Forest plot of IV tPA. Results were stratified by definition of DNI.

IV tPA: intravenous thrombolysis; DNI: delayed neurological improvement.

Figure 2.

Forest plot of atrial fibrillation (Afib). Results were stratified by definition of DNI.

DNI: delayed neurological improvement.

Figure 3.

Forest plot of age. Results were stratified by definition of DNI.

DNI: delayed neurological improvement.

Figure 4.

Forest plot of diabetes mellitus (DM). Results were stratified by definition of DNI.

DNI: delayed neurological improvement.

Figure 5.

Forest plot of hypertension. Results were stratified by definition of DNI.

DNI: delayed neurological improvement.

Figure 6.

Forest plot of male sex. Results were stratified by definition of DNI.

DNI: delayed neurological improvement.

Figure 7.

Forest plot of smoking. Results were stratified by definition of DNI.

DNI: delayed neurological improvement.

Non-contributory patient characteristics

Seven studies reported baseline NIHSS scores. There was no significant association between baseline NIHSS score and rates of DNI (MD −1.11, 95% CI −2.29 to 0.08; p = 0.067); however, there was significant heterogeneity among studies (I² = 84%, p < 0.001; Supplementary Figure S2). Four studies reported rates of coronary artery disease (CAD) or congestive heart failure (CHF), hyperlipidemia, and prior stroke. There was no significant association between CAD/CHF and rates of DNI (OR 0.86, 95% CI 0.69 to 1.08; p = 0.192; Supplementary Figure S3). There was no significant association between hyperlipidemia and rates of DNI (OR 0.80, 95% CI 0.56 to 1.13; p = 0.196; Supplementary Figure S4). Finally, there was no significant association between prior stroke and rates of DNI (OR 0.74, 95% CI 0.53 to 1.03; p = 0.078; Supplementary Figure S5).

Discussion

In this systematic review and meta-analysis, we found that there is a high rate of heterogeneity in the definition of DNI. Additionally, we found that atrial fibrillation, age, DM, use of IV tPA, hypertension, male sex, and smoking were significantly associated with rates of DNI while other patient characteristics did not yield a significant association. Our findings are important because they highlight the inconsistency between DNI definitions, while nonetheless identifying characteristics that are associated or not associated with DNI.

Previous studies have examined DNI, though the literature is sparse on the topic. The data supporting ENI and END as predictor of outcomes are much stronger than for DNI.^3,4 This is reflected by the low number of studies that satisfied our inclusion criteria, but also due to the extreme heterogeneity in the definition of DNI. Overall, there were two overarching types of DNI definitions. One defines DNI as the good functional outcome (mRS 0-2) at 90 days in the absence of ENI and despite a high NIHSS score at 24 hours. Five studies in our analysis reported definitions along these lines.^{11–13,15,16} However, even among these studies, the definition of “ENI” varied. So, though the general theme of these definitions was the same, the specifics of each definition were unique. The other type of DNI definition was improvement of NIHSS score at day seven post-treatment. Two studies reported this definition, although the metrics used to define DNI between these two studies varied greatly, and the definitions were convoluted.^14,17

The variance in the definition of DNI hinders its research and clinical utility. A consensus definition of DNI will not be possible until ENI and END are consistently defined because DNI relies on the presence or absence of early changes. Reaching consensus definitions of ENI, END, and DNI is crucial to better examine the process of neurological recovery after reperfusion or, in the future, after the administration of neuroprotective or neuroreparative therapies. This should be a future research priority.

Similarly, there was heterogeneity in significant associations based on the definition of DNI. Among the contributory characteristics, atrial fibrillation, intravenous thrombolysis, DM, hypertension, male sex, and smoking rates were significant overall. However, when stratified by the definition of DNI, none of the aforementioned characteristics were significant for either of the definitions of DNI that focused on seven days post-treatment. At present, this is an interesting phenomenon, but it is difficult to comment on the significance of this observation due to the small sample size in each of the seven-day definitions of DNI. Consistency in definitions and reporting of outcomes will help to power future meta-analyses and determine the significance and associations in these definitions.

In this meta-analysis, DNI was more likely in younger patients, patients treated with intravenous thrombolysis, and patients without atrial fibrillation or DM. These four factors have also been found to influence the likelihood of achieving functional independence (mRS 0-2) at 90 days in previous studies,^18–21 which supports the validity of the associations identified in our analysis. They are also biologically plausible. The presumed pathophysiological substrate of DNI is the slow recovery of “stunned brain.” ⁵ Younger age and absence of atrial fibrillation (often accompanied by multiple vascular risk factors) and DM may identify patients with better brain reserve. Another more recently described potential mechanism is delayed reperfusion. ²² Meanwhile, intravenous thrombolysis increases the chances of reperfusion. It is interesting that male sex and smoking were associated with higher rates of DNI, as these two variables are generally associated with poorer outcomes.^23,24 Further data is warranted to explore these associations.

There is great clinical value in not only DNI, but also ENI and END. Understanding these phenomena and their associated characteristics would allow clinicians to better inform patients and their families regarding trajectories their long-term recovery could follow after endovascular therapy for AIS. Yet, fully understanding early changes in NIHSS score and their association with long-term mRS scores is not possible unless consensus definitions of the terms are reached.

Our meta-analysis has limitations. First, despite a thorough review of the literature, we were able to identify only seven studies reporting on the presence or absence of DNI; thus, our statistical power was low. Second, differences in the DNI definition hindered efforts to ascertain an overall rate of DNI and may have influenced some of the results by introducing heterogeneity. Lastly, we did not have access to patient-level data, limiting the analysis we were able to perform.

In conclusion, DNI is an insufficiently studied phenomenon that happens in a subset of patients with AIS. It is more common in younger patients who undergo thrombolysis and do not have atrial fibrillation, DM, or hypertension. It is also more common in males and those who smoke. Future studies should focus on establishing a homogenous definition of not only DNI, but also ENI and END.

Conclusions

In this meta-analysis of a systematic review of seven studies, DNI was shown to be associated with atrial fibrillation, age, DM, use of IV tPA, hypertension, male sex, and smoking status. Additionally, we found a high degree of heterogeneity in definitions of DNI, with every study reporting a different definition. Future studies should work to establish a consensus definition of DNI to determine its causes and significance more accurately.