Aspiration Pneumonia in Adults Hospitalized With Stroke at a Large Academic Hospital in Zambia

Abstract

Background and Objectives

Preventing complications of stroke such as poststroke aspiration pneumonia (PSAP) may improve stroke outcomes in resource-limited settings. We investigated the incidence and associated mortality of PSAP in Zambia.

Methods

We conducted a prospective cohort study of adults with stroke at University Teaching Hospital (Lusaka, Zambia) between December 2019 and March 2020. NIH Stroke Scale, Glasgow Coma Scale, and Modified Rankin Scale scores and 9 indicators of possible PSAP were collected serially over each participant's admission. PSAP was defined as ≥4 indicators present, and possible PSAP as 2%–3% present. T tests and χ² tests were used to compare clinical parameters across PSAP groups. Logistic regression was used to assess the relative effects of age, sex, PSAP status, and initial stroke severity on inpatient mortality.

Results

We enrolled 125 participants. Mean age was 60 ± 16 years, 61% were female, 55% of strokes were ischemic, and the baseline NIH Stroke Scale score was 19.7 ± 8.7. Thirty-eight (30%) had PSAP, and 32 (26%) had possible PSAP. PSAP was associated with older age and more adverse stroke severity scores. Fifty-nine percent of participants with PSAP died compared with 39% with possible PSAP and 8% with no PSAP. PSAP status independently predicted inpatient mortality after controlling for age, sex, and initial stroke severity. Swallow screening was not performed for any participant.

Discussion

PSAP is common and life threatening in Zambia, especially among older participants with severe stroke presentations. PSAP was associated with significantly increased mortality independent of initial stroke severity, suggesting that interventions to mitigate PSAP may improve stroke outcomes in Zambia and other resource-limited settings.

Stroke is a leading cause of morbidity and mortality worldwide.¹ Its incidence is rising in sub-Saharan Africa, where recent epidemiologic shifts have increased rates of hypertension, diabetes, tobacco use, and obesity. Although recent advances have transformed management of stroke in high-income settings, stroke units that provide protocolized care directed at mitigating poststroke complications have not been widely implemented in many low- and middle-income countries.^2,3 Understanding the burden of preventable poststroke complications in resource-limited settings may guide innovations in care that improve outcomes from this increasingly prevalent disease.

Stroke poses particular risk for aspiration pneumonia secondary to dysphagia and altered mental status. In the United States, poststroke aspiration pneumonia (PSAP) occurs in as many as 13% of cases and underlies as much as 10% of stroke-related mortality.⁴ Although data on PSAP in sub-Saharan African are limited, published studies from this region suggest that its incidence ranges from 13% to 34% and accounts for as much as 45% of inpatient stroke-related mortality.^5-8 PSAP prolongs hospital stays, causes fevers and systemic inflammation, and can progress to sepsis and acute respiratory failure.^9-12 Risk factors for PSAP in high-income settings include age, male sex, underlying pulmonary and cardiac disease, dysphagia, and increased stroke severity.^13-15 PSAP risk factors have been less well studied in sub-Saharan Africa.

We investigated the relationship between PSAP and clinical outcomes in adults with stroke admitted to the University Teaching Hospital (UTH), the national referral hospital in Lusaka, Zambia. We assessed the incidence of inpatient PSAP in this cohort and investigated differences in demographics, clinical severity, and mortality between participants with and without PSAP.

Methods

Setting

We conducted a prospective cohort study of adults with stroke admitted to UTH in Lusaka, Zambia. UTH is a 1,700-bed facility affiliated with the University of Zambia and is the country's largest academic medical center. It is staffed by 4 attending neurologists and 5 neurology residents and is the only hospital in Zambia with an inpatient neurology service and full-time neurologists. Adults with stroke are cared for on the adult medical admissions and inpatient wards by the inpatient neurology team and medical nurses with much of the bedside care provided by family members. Most beds on the inpatient wards do not allow head-of-bed (HOB) elevation, although this can be achieved by placing a box, large rolled blanket, or other similar item under the head of the mattress. There is no established protocol for bedside swallowing evaluation. Available diagnostics include CT and MRI scanners, although the patient and/or their family must pay for these investigations before they can be obtained. The out-of-pocket costs are often prohibitive, and logistical delays in imaging studies often render them unavailable for acute decision making. Nasogastric tubes and enteral feeding formulas are commercially available at out-of-pocket cost to the patient. The hospital's intensive care unit (ICU) has 13 beds and 8 ventilators.

Participant Enrollment and Study Procedures

All adults (aged 18 years or older) presenting with a clinical stroke syndrome (as determined by the inpatient neurology team on the day of admission) whose onset was within the prior 7 days were enrolled after informed consent was obtained from the patient or primary caregiver. All participants were enrolled within 24 hours of admission. Participants were removed from the study if neuroimaging later revealed an alternative nonstroke etiology for their presentation. Questionnaire data were collected on paper forms by research staff and entered into a RedCap database.¹⁶ As per standard of care, enrolled participants were reassessed daily on multiple parameters of clinical progress including level of alertness and aspiration risk. Neurology residents completed the NIH Stroke Scale (NIHSS), Glasgow Coma Scale (GCS), and modified Rankin Scale (mRS) scores at admission, on day 4 of the hospitalization (if the participant was still hospitalized), and at discharge.

PSAP Determination

Seven clinical variables pertaining to features of aspiration pneumonia were assessed daily by chart review and bedside examination for the first 10 days of each participant's hospitalization. These included fever (temperature >38°C), tachypnea (respiratory rate >20), hypoxemia (oxygen saturation <92%), cough, rhonchi, witnessed aspiration event, and initiation of antibiotics for clinically suspected aspiration pneumonia. In addition, leukocytosis (WBC >10,000) on complete blood count (CBC) or opacity on chest X-ray was also recorded when these studies were obtained during the first 10 days of admission (CBC is typically obtained on admission but not trended thereafter; X-ray is not consistently obtained for cases of suspected pneumonia). These variables were selected to minimize reliance on diagnostic testing and were based on established clinical features of aspiration pneumonia.^5,17

Participants were categorized as having PSAP if ≥ 4 of these variables (including leukocytosis and X-ray findings) were present and as having possible PSAP if 2 or 3 were present. Because no consensus criteria for the diagnosis of aspiration pneumonia have been established, the diagnosis is usually made by assimilating multiple clinical data points. Therefore, we sought to create a classification system that maximized the use of objective criteria that could be easily assessed at the bedside on serial evaluations in a resource-limited setting. While acknowledging the inherent arbitrary nature of choosing a cutoff in the absence of empirical evidence, we felt that any combination of 4 or more factors constituted a compelling clinical picture of PSAP, whereas 2 to 3 factors signaled a high likelihood of PSAP but with a broader differential diagnosis (e.g., a participant with only fever and leukocytosis).

Route of nutrition (oral or via nasogastric tube), whether a patient was nil per os (NPO), the person feeding the participant (participant, health care worker, or family member), presence/absence of a documented swallow screening, and HOB position at the time of the daily study assessment were also noted on a daily basis.

Statistical Analysis

Descriptive statistics were generated to describe cohort demographics and clinical characteristics, assess which hospital day participants met the criteria for PSAP, and describe the percentage of participants with each individual PSAP clinical criterion. Analysis of variance was performed to assess for differences across PSAP groups (PSAP, possible PSAP, and no PSAP) in age, admission NIHSS score, admission mRS score, and admission GCS score. Chi-square tests were performed to assess for PSAP group differences in mortality and stroke type. The Kruskal-Wallis test was used to assess group differences in admission level of alertness, expressed as an ordinal scale (1 = alert, 2 = lethargic, 3 = stuporous, and 4 = comatose). Logistic regression was performed to model the effects of PSAP status (PSAP, possible PSAP, and no PSAP), age, sex, admission NIHSS score, admission mRS score, and admission GCS score on inpatient mortality. Descriptive statistics were generated among participants meeting the PSAP criteria to assess whether aspiration precautions (i.e., HOB elevation, NPO, and NGT feeding) were implemented for at least 2 days before participants met the criteria for PSAP. For this analysis, participants with ≥4 PSAP criteria on admission were excluded. Statistical analyses were performed using Stata 15.0 (College Station, TX). p Values <0.05 were considered statistically significant.

Standard Protocol Approvals, Registrations, and Patient Consents

This study was approved by the Zambia ERES Converge Institutional Review Board, the Zambia National Health Research Authority, and the Johns Hopkins Institutional Review Board.

Data Availability

Deidentified data are not available in a public repository due to legal statues in Zambia regarding data sharing. Data will be shared on request after required approvals are obtained from a certified Zambian institutional review board and the Zambia National Health Research Authority.

Results

Study Population

Between December 2019 and March 2020, 125 adults met the inclusion criteria and consented to participate in the study. Mean age was 60 ± 16 years, and 61% were female. The mean baseline NIHSS score was 19.7 ± 8.7. Stroke etiologies were categorized as ischemic (55%), hemorrhagic (27%), and unknown (18%). Unknown strokes were defined as clinically suspected strokes when confirmatory neuroimaging was unavailable. Only 2 participants received care in the ICU, with the remainder being cared for on the admissions and general medical inpatient wards. Twenty-nine participants (23%) were people with HIV (Table 1).

Table 1.

Key Demographic and Clinical Characteristics

PSAP Group Assignments

Thirty-eight participants (30%) met the criteria for PSAP (Figure 1A), with an additional 32 participants (26%) meeting the criteria for possible PSAP. Within the PSAP group, 13 participants (34%) met the criteria at the time of admission, with the remaining 25 participants developing PSAP over hospital days 2–9 (Figure 1B). Fever, cough, tachypnea, and antibiotics for presumed PSAP were the most common PSAP clinical criteria. Leukocytosis and pneumonia on chest X-ray were less commonly reported, but these data were also less commonly available. The frequencies of each PSAP clinical criterion in the PSAP group are displayed in Figure 1C.

Figure 1. Clinical Features of Aspiration Pneumonia Among Participants With PSAP.

Characteristics of aspiration pneumonia among participants with poststroke aspiration pneumonia (PSAP), including (A) histogram of the number of PSAP criteria met by participants in this group (4 being the minimum number of criteria for inclusion in the PSAP group), (B) histogram of the day of hospitalization that participants met the criteria for PSAP, and (C) histogram depicting the frequency of each of the 9 PSAP criteria in the PSAP group. Bars in each plot represent the absolute number of participants. *Complete blood count was obtained for 103 participants, of whom 27 were found to have leukocytosis, defined as white blood cell count >10K. **Chest X-rays were obtained for 4 participants, of whom all were found to have radiographic evidence of pneumonia.

PSAP Risk Factors

Participants with PSAP were significantly older than those with possible or no PSAP (68 years vs 57 years vs 55 years, respectively; p < 0.001; Figure 2A). No association was found between the PSAP group and sex or HIV status (data not shown). The PSAP group was significantly associated with admission level of alertness (p < 0.001; Figure 2B), admission NIHSS score (PSAP mean 26.2 ± 6.5; possible PSAP mean 21.0 ± 8.6; no PSAP mean 13.6 ± 5.7; p < 0.001; Figure 2C), admission GCS score (PSAP mean 7.9 ± 3.3; possible PSAP mean 10.1 ± 4.2; no PSAP mean 13.4 ± 2.7; p < 0.001; Figure 2D), and admission mRS score (PSAP mean 4.9 ± 0.3; possible PSAP mean 4.7 ± 0.5; no PSAP mean 3.9 ± 1.3; p < 0.001; Figure 2E). PSAP occurred more frequently in patients with hemorrhagic than ischemic stroke, but this difference did not meet statistical significance (p = 0.051).

Figure 2. Demographic, Clinical, and Mortality Data Across PSAP Groups.

Demographic and clinical characteristics across poststroke aspiration pneumonia (PSAP) groups, including (A) mean participant age, (B) the percentage distribution in level of alertness at presentation across groups, with level of alertness categorized as awake (blue), lethargy (orange), stupor (gray), and coma (yellow), (C) mean admission NIH Stroke Scale score, (D) mean admission Glasgow Coma Scale score, and (E) mean admission modified Rankin Scale score, and (F) rate of inpatient mortality. In panels A, C, D, and E, bars represent group means, and error bars represent SDs.

Mortality

In-hospital death occurred in 22 participants with PSAP (60%), in 11 participants with possible PSAP (39%), and in 4 participants without PSAP (8%; p < 0.001; Figure 2F). PSAP status remained significantly associated with mortality after controlling for age, sex, admission NIHSS score, admission mRS score, and admission GCS score (OR = 3.07, 95% CI = 1.32–7.15, p = 0.009; Table 2). Lower admission GCS score also independently predicted inpatient mortality (OR = 0.81, 95% CI = 0.68–0.98, p = 0.031).

Table 2.

Results of Logistic Regression Modeling Risk for Inpatient Death as a Function of the PSAP Group (PSAP, Possible PSAP, and No PSAP), Age, Sex, Admission NIHSS Score, Admission GCS Score, and Admission mRS Score

Aspiration Precautions

Of the participants meeting the criteria for PSAP, 13 (34%) met the criteria for PSAP on admission. Of the remaining 25 PSAP participants, 9 (36%) had HOB elevated, 18 (72%) were kept NPO, and 17 (68%) were fed via NGT in the 2 days before meeting the PSAP criteria. Seven (18%) were fed by mouth before PSAP occurrence, 6 of whom were fed by family, and 1 of whom was self-fed. There were no significant PSAP group differences in adherence to HOB elevation, NPO status, or NGT tube feeding. There were no documented bedside swallow screenings for any participant in the cohort.

Discussion

PSAP is a common and highly morbid complication of acute stroke in Zambia, with nearly 60% inpatient mortality among the 30% of participants meeting the criteria for PSAP. Although these participants tended to be older and to present with more severe stroke syndromes, the development of PSAP was an independent predictor of inpatient mortality after controlling for initial stroke severity. We found inconsistent adherence to common aspiration precautions and a universal lack of bedside swallow screening.

Our findings are consistent with those of prior studies demonstrating that PSAP is a common complication of stroke and that it may be especially prevalent in resource-limited settings.^5-8 Although inpatient stroke centers in the United States and other high-resourced settings commonly use protocols to minimize the risk of PSAP, such as early swallow screening, NPO, and NGT feeding in patients with prohibitive dysphagia, resource-limited settings often lack sufficient nursing care and swallowing therapists to ensure safe nutrition after an acute neurologic insult that impairs mental status or oropharyngeal coordination.^18,19 Additional factors in sub-Saharan Africa, including limited access to NGTs, enteral feeding formulas, and food texture modification, inability to elevate the HOB in many hospital settings, and aggressive oral feeding by bedside family caregivers, may further contribute to aspiration risk.

We also confirm the heavy mortality burden associated with PSAP. Although patients presenting with greater neurologic impairment are at higher risk for both PSAP and death, PSAP remained an independent risk factor for death after controlling for age and initial stroke severity. Prior work in the United States has reported as much as a 3-fold increased risk of death after stroke in patients with PSAP after adjusting for initial stroke severity.⁴ The 60% mortality rate among those with PSAP in our sample is higher than most published data from high-income countries,^9,20,21 likely reflecting the resource limitations of the Zambian setting, which result in delayed diagnosis of aspiration pneumonia, limited antibiotic options, scarce availability of advanced airway and sepsis management, and limited access to ICU care.

Because PSAP is common, predictive of poor outcomes, and in many cases preventable, it may prove to be a high-yield target for intervention in settings where resource-intensive acute stroke interventions (e.g., thrombolysis and endovascular interventions) remain unavailable. We found a low rate of HOB elevation in patients with PSAP in the days preceding pneumonia onset. Data are limited on the benefits of HOB elevation in nonintubated patients with stroke, although there is evidence that it does help to reduce pneumonia in mechanically ventilated patients.²² A study of flat vs 30-degree HOB position on stroke outcomes revealed in nonintubated patients showed no significant difference in rates of aspiration pneumonia,²³ although the median NIHSS score in that series was 4, considerably lower than the initial stroke severity in our cohort and likely indicating a lower baseline PSAP risk. It is possible that protocols standardizing HOB elevation in all patients with stroke could prevent a significant proportion of PSAP cases in our population, although this intervention alone is unlikely to be sufficient as there was no difference in rates of HOB elevation between PSAP groups in our cohort.

In addition, we found a universal absence of bedside swallow screening, which has been proven to significantly reduce the incidence of PSAP.^20,24 Protocolized swallow screening during inpatient admission workflows is likely to improve identification of at-risk patients, but this would necessarily need to be followed by better access to and improved feasibility of implementation of precautionary measures, including beds that allow HOB elevation, NGT and enteral formulas, and bedside oral care, all of which are known to reduce incidence of PSAP.²⁵ In addition, the materials needed for routine bedside swallowing evaluation, including disposable cups and straws and potable water, are not always easily accessible in many low-resourced settings. As the impact of swallow screening and other interventions to prevent PSAP have not been previously assessed in low-resource settings, further prospective studies are needed to assess their effect on outcomes in stroke survivors in these regions.

It is worth noting that although bedside aspiration precautions may prevent many cases of PSAP, they are unlikely to be effective in patients with severe airway compromise. The mean GCS in our PSAP group was 7.9, below the threshold at which most patients would typically be intubated for airway protection in higher-resourced settings. Access to intensive care and mechanical ventilation is highly constrained in our hospital setting (only 2 participants in our cohort were cared for in the ICU) and is nonexistent in numerous countries throughout sub-Saharan Africa.²⁶ Efforts to improve outcomes from stroke and other neurologic emergencies in resource-limited settings must also focus on critical care capacity building. In the nearer term, dedicated stroke units that cluster inpatients to optimize delivery of protocolized nursing care and rehabilitation offers promise for the prevention of PSAP and other subacute complications of stroke^3,27 and are likely to benefit more patients and have a higher impact on reducing stroke-related mortality than more expensive and resource-intensive interventions like acute thrombolysis.²

Our study was limited by the challenges of collecting daily clinical data in a resource-limited hospital with no electronic medical record system. Despite the efforts of our research staff, limited documentation in patient charts, the inability to collect more than once daily clinical data on each participant, and often the difficulty of physically locating participants and their charts within busy hospital wards resulted in missing data points that likely obscured some of the complexity of participants' longitudinal clinical progress. Because participants were assessed only once daily, changes in aspiration precautions such as HOB position may not have been captured. In addition, it was not feasible to collect longitudinal data beyond the inpatient setting in participants who survived to hospital discharge, and we are therefore unable to assess the long-term clinical impact of PSAP on stroke recovery in this population. Finally, our sample size was limited by a cessation of data collection due to the coronavirus pandemic, and it is possible that a larger sample may have been powered to detect associations between the implementation of aspiration precautions and clinical outcomes that proved insignificant in our present cohort.

Overall, our findings provide compelling evidence that PSAP is widespread, associated with poor outcomes, and appears to be significantly more prevalent among adults with stroke in our Zambian hospital compared with reports on PSAP from high-income health care settings. PSAP predicts mortality independently of initial stroke severity in our cohort, suggesting that its prevention may improve outcomes after stroke, even in patients presenting with severe neurologic impairment. The prevalence of this often preventable complication highlights the need to implement data-driven aspiration prevention measures, develop and implement low-cost stroke units, and augment critical care resources in resource-limited settings.

Appendix. Authors

Study Funding

Funding was provided from the following grants: United States Department of State Fullbright Scholar Award (D.R. Saylor), NIH R21 NS118543-01 (D.R. Saylor), and NIH Fogarty International Center grant no. D43TW009340 (A. Nutakki).

Disclosure

The authors report no disclosures relevant to the manuscript. Full disclosure form information provided by the authors is available with the full text of this article at Neurology.org/cp.