Thirty-day outcomes and predictors of mortality following acute myocardial infarction in northern Tanzania: a prospective observational cohort study

Abstract

Objective:

There is a rising burden of myocardial infarction (MI) within sub-Saharan Africa. Prospective studies of detailed MI outcomes in the region are lacking.

Methods:

Adult patients with confirmed MI from a prospective surveillance study in northern Tanzania were enrolled in a longitudinal cohort study after baseline health history, medication use, and sociodemographic data were obtained. Thirty days following hospital presentation, symptom status, rehospitalizations, medication use, and mortality were assessed via telephone or in-person interviews using a standardized follow-up questionnaire. Multivariate logistic regression was performed to identify baseline predictors of thirty-day mortality.

Results:

Thirty-day follow-up was achieved for 150 (98.7%) of 152 enrolled participants. Of these, 85 (56·7%) survived to thirty-day follow-up. Of the surviving participants, 71 (83·5%) reported persistent anginal symptoms, four (4·7%) reported taking aspirin regularly, seven (8·2%) were able to identify MI as the reason for their hospitalization, and 17 (20·0%) had unscheduled rehospitalizations. Self-reported history of diabetes at baseline (OR 0·32, 95% CI 0·10–0·89, p = 0·04), self-reported history of hypertension at baseline (OR 0·34, 95% CI 0·15–0·74, p = 0·01), and antiplatelet use at initial presentation (OR 0·19, 95% CI 0·04–0·65, p = 0·02) were all associated with lower odds of thirty-day mortality.

Conclusions:

In northern Tanzania, thirty-day outcomes following acute MI are poor, and mortality is associated with self-awareness of comorbidities and medication usage. Further investigation is needed to develop interventions to improve care and outcomes of MI in Tanzania.

1. Introduction

Myocardial infarction (MI) is a leading cause of death in adults globally.¹ In Sub-Saharan Africa (SSA), the burden of MI is presumed to be rising as the region proceeds through the epidemiologic transition,² but data regarding MI incidence or prevalence in SSA remain limited.³ A recent study in Tanzania found that acute MI was common among patients presenting to the emergency department, but that MI cases were frequently missed by providers and the use of evidence-based MI treatment was rare.⁴ Against this backdrop, little is known about patient-level MI outcomes in SSA. A handful of studies have investigated in-hospital or thirty-day mortality rates in SSA.^5–8 These studies, which were conducted in advanced centers with capacity for cardiac catheterization, reported in-hospital mortality rates ranging from 9 to 17%.^5–8 However, advanced cardiac centers are scarce in SSA;^7,9 in 2017, for example, there were only 12 cardiac catheterization facilities within three countries in East and Central Africa⁷, a region containing 29 countries. In a recent study at a Tanzanian hospital without cardiac catheterization capacity, a 43% thirty-day mortality rate was observed,⁴ suggesting that MI outcomes may be worse in settings without access to specialized cardiac care.

Beyond the limited crude mortality data described above, much remains to be learned about MI outcomes in SSA. Specifically, to our knowledge, no published studies have reported healthcare utilization following discharge, use of evidence-based therapies for secondary prevention, re-hospitalization rates, or patient-perceived barriers to care. Furthermore, to our knowledge, there are no data regarding predictors of death following acute MI or symptom status among MI survivors in SSA. Obtaining such data would aid clinicians, researchers, and policy-makers in identifying gaps in care and targets for strengthening existing systems of care.

The aim of this prospective observational study was to report several thirty-day outcomes including healthcare utilization, predictors of death, and mortality among adult patients with confirmed MI in the Kilimanjaro Region of northern Tanzania.

2. Methods

2.1. Dating Sharing

All reasonable requests for data sharing will be honored; please contact the corresponding author to request data access.

2.2. Setting

This study was conducted in the Kilimanjaro Region of northern Tanzania. In 2014, the community prevalence of hypertension among adults living in Kilimanjaro was 28%,¹⁰ and the prevalence of diabetes was 6%.¹¹ The dominant local tribe is the Chagga tribe. The local tertiary care center is Kilimanjaro Christian Medical Centre (KCMC), which was not staffed with a trained cardiologist or equipped with a cardiac catheterization lab at the time of this study.

2.3. Participant selection

The participants in this study were enrolled from a prospective MI surveillance study in northern Tanzania, with methods described elsewhere.⁴ The surveillance study was conducted in the KCMC emergency department (ED) in 2019. Adult patients (age ≥ 17 years) presenting with acute chest pain or shortness of breath to the ED were eligible for enrollment in the surveillance study and underwent testing for acute MI with electrocardiogram and point-of-care troponin-I testing (Abott iSTAT cTnI assay, Abbott Point of Care, Princeton, New Jersey, United States). Patients in the surveillance study who were found to have acute MI were enrolled into the longitudinal follow-up study. For study purposes, acute MI was defined according to the Fourth Universal Definition of MI criteria:¹² any patient with pathologic ST elevation in contiguous leads or pathologically elevated troponin (>0·08ng/ml) were considered to have an acute MI. Three external independent physician adjudicators reviewed electrocardiograms to determine pathologic ST elevations and reviewed ECG and troponin levels to determine presence or absence of MI. Troponin and ECG results were shared immediately with the clinical team, but for study purposes determination of MI was performed by independent adjudicators. ST elevation MI (STEMI) was defined as pathologic ST elevation in at least two contiguous leads, and non-STEMI (NSTEMI) was defined as pathologically elevated troponin in the absence of STEMI.

2.4. Study procedures

Neither study participants nor the public were involved in the design, or conduct, or reporting, or dissemination plans of this study. At initial presentation, trained research assistants administered a standardized questionnaire to all participants, eliciting information about past medical history, medication use, and sociodemographic information. Blood pressure, weight, and height were measured at time of initial enrollment by research assistants. Thirty days following presentation to the ED, participants were contacted via telephone and a standardized follow-up questionnaire was administered by research assistants. Participants were asked to provide information about their symptom progression, rehospitalizations, medication use, and barriers to care subsequently encountered. They were also asked to identify their diagnosis in their own words. If participants were unreachable by telephone, they were called on five separate days. If they remained unreachable, they were visited in their home by a member of the study team to complete the follow-up questionnaire face-to-face. If the participant had died, the survey was administered to a relative. In case of participant death, a brief verbal autopsy based on the World Health Organization (WHO) 2016 verbal autopsy instrument was administered to available relatives.¹³

2.5. Study definitions

The primary outcome was death from any cause. Secondary outcomes included death from MI, location of death, resolution of symptoms, rehospitalization, numbers of days hospitalized, prescribed medications, follow-up rates, and knowledge about their disease. Causes of death were adjudicated by a committee of physicians from Tanzania and the United States. The adjudicating committee reviewed the verbal autopsy data, documented clinical diagnoses, patient demographics, vital signs at hospital presentation, electrocardiograms, laboratory data including troponin values, and all other available clinical data such as presenting symptoms and symptom duration prior to hospital presentation. After reviewing these data, the adjudicating committee ascribed one of the following causes of death to each case: recurrent myocardial infarction, heart failure, stroke, renal failure, or other. Deaths due to acute left heart failure or cardiogenic shock were both categorized as deaths due to heart failure. If the adjudicating committee did not feel there was enough data to determine a cause of death, the cause of death was reported as “indeterminant.” The adjudicating committee also reviewed participants’ self-identified diagnoses to determine whether or not the patient accurately identified MI as their diagnosis. Any patient description that was felt by the committee to describe MI (for example, “heart attack” or “heart problem”) was characterized as an accurate identification of the diagnosis. When patients did not know their diagnosis or provided a diagnosis unrelated to MI (for example, “pneumonia” or “malaria”), they were categorized as being unable to identify MI as their diagnosis. Disagreements among committee members were resolved by consensus. Health behaviors were explored using international standards. Sedentary lifestyle was defined as less than 150 minutes of moderately vigorous exercise per week, in accordance with WHO guidelines.¹⁴ Poor diet was defined as not eating vegetables and fruits daily, consistent with the known association between daily fruit and vegetable consumption and reduced risk of cardiovascular disease.¹⁵ Body mass index (BMI) was calculated directly from measured height and weight, and alcohol and tobacco use were defined by patient self-report. Baseline medical co-morbidities, such as hypertension and diabetes, were defined by patient self-report. Dual anti-platelet therapy was defined as self-report of taking both aspirin and another anti-platelet agent, such as clopidogrel.

2.6. Statistical analyses

Statistical analyses were performed in the R Suite (Ver 1·3·1056·1). Descriptive statistics were used to describe participant characteristics and outcomes; continuous variables are presented as medians (interquartile ranges) and categorical variables are presented as frequencies and proportions. We secondarily sought to identify predictors of thirty-day mortality. A pool of potential predictor variables was developed based on biologic plausibility and known predictors of mortality in high-income settings.¹⁶ Univariate analyses were performed to assess associations between predictor variables obtained at enrollment and mortality, using Pearson’s chi-square for categorical variables and Welch’s t-test for continuous variables. Odds ratios and corresponding 95% confidence intervals were calculated directly from two-by-two contingency tables. Multivariate logistic regression was then performed to identify predictors of thirty-day mortality. Any variable with evidence of univariate association with mortality (p < 0·1) were included in the model; age and sex were also forced into the model.

2.7. Ethics

All participants provided written informed consent prior to study participation. This study received ethics approval from Duke Health, KCMC, and the Tanzanian National Institute for Medical Research.

2.8. Role of the Funding Source

The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

3. Results

Of 681 screened patients, a total of 152 participants with acute MI (61 STEMIs and 91 NSTEMIs) were identified and enrolled in this study. Baseline sociodemographic characteristics and medical comorbidities of participants are presented in Table 1. The median (IQR) age of participants was 61 (49, 76) years and the median BMI was 23·2 (20·8,26·4) kg/m², and 91 participants (59·9%) were male. At presentation, 94 (61·8%) participants reported a known history of hypertension and 27 (17·8%) reported a known history of diabetes. Eighteen (11·8%) participants reported taking an antiplatelet medication regularly prior to presentation. None of the participants received fibrinolysis during their index ED presentation.

Table 1.

Characteristics of adults presenting with acute myocardial infarction to a Tanzanian emergency department, 2019 (N=152)

	Median	IQR
Age	61	49·0,76·0
Body mass index (kg/m2)	23·2	20·8,26·4
Troponin (ng/ml)	0·12	0·02,0·42
Systolic blood pressure at initial presentation (mmHg)	145	113–168
Diastolic blood pressure at initial presentation (mmHg)	85	68–101
	Number of Participants	(%)
Sex
Male	91	59·9%
Female	61	40·1%
Education
None	10	6·6%
Primary	88	57·9%
Secondary	20	13·2%
Post-Secondary	34	22·4%
History of prior myocardial infarction	4	2·6%
Current tobacco use	16	10·5%
Current alcohol use	54	35·5%
Poor diet	140	92·1%
Sedentary lifestyle	91	59·9%
Given aspirin in the ED	35	23·0%
Received fibrinolysis in the ED	0	0%
Medical History, self-reported
Diabetes	27	17·8%
Hyperlipidemia	8	5·3%
Hypertension	94	61·8%
HIV	5	3·3%
Chronic kidney disease	16	10·5%
Taking antiplatelet medication at presentation	18	11·8%

ED: emergency department

HIV: human immunodeficiency virus

Thirty-day follow-up was achieved for 150 (98·7%) participants; two participants were lost to follow-up. Of the 150 for whom follow-up was obtained, 65 (43·3%) died. Table 2 summarizes outcomes of the 85 participants surviving to thirty-day follow-up. Of the 85 participants who survived to thirty-day follow-up, 14 (16·5%) reported their symptoms resolved, 53 (62·4%) reported symptom improvement, and 18 (21·2%) reported their symptoms had worsened or were unchanged. Among surviving patients, four (4·7%) reported taking aspirin daily and seven (8·2%) were able to identify MI as their diagnosis. In the 30 days following initial hospital presentation, 36 (42·4%) reported receiving a scheduled follow-up appointment and 17 (20·0%) reported having an unscheduled rehospitalization for recurrent MI symptoms.

Table 2.

Thirty day outcomes among patients with acute myocardial infarction surviving to thirty-day follow-up, northern Tanzania, 2019 (N=85)

All surviving patients N=85*	Number of Participants	(%)
Symptom Progression at 30 days
Resolved	14	16.5%
Improved	53	63.4%
Worsened or unchanged	18	21.2%
Received a prescription at hospital discharge	81	95.3%
Reports taking medications as prescribed (N=81)	75	92.6%
Reports taking aspirin	4	4.7%
Reports taking dual anti-platelet therapy	0	0%
Reports taking beta-blocker	1	1.2%
Reports taking angiotension-converting enzyme (ACE) inhibitor or angiotension receptor blocker (ARB)	3	3.5%
Reports taking statin	1	1.2%
Identified acute MI as their diagnosis	7	8.2%
Feels they understand their treatment	64	75.3%
Given 30-day follow-up appointment	36	42.4%
Attended follow-up clinic appointment (N=36)	23	63.9%
Unscheduled hospitalizations for recurrent chest pain/shortness of breath after discharge	17	20.0%

^*2 lost to follow up

MI: myocardial infarction

The adjudicated causes of death are summarized in Table 3. Of the 65 deaths occurring within thirty days, 17 (26·2%) were due to myocardial infarction, 17 (26·2%) were due to heart failure, and 10 (15·4%) were due to renal failure. The majority of deaths (53, 81·5%) occurred in a hospital, and the median (IQR) time interval between hospital presentation and death was three (1, 16) days. Table 4 presents univariate analyses of predictors of thirty-day mortality. The following baseline characteristics were found to be potentially significant predictors of lower thirty-day mortality (p < 0·10) and were included in the multivariate model: self-reported history of diabetes (OR 0·40, 95% CI 0·15–0·98, p = 0·044), self-reported history of hypertension OR 0·26, 95% CI 0·18–0·72, p = 0·003, taking antiplatelet medication at presentation OR 0·34, 95% CI 0·09–1·03, p = 0·054, higher BMI (p = 0·014) and lower troponin level (p = 0·053).

Table 3.

Causes of death among adults dying within 30 days of acute myocardial infarction, northern Tanzania, 2019 (n=65)

	Number of participants	(%)
Location of death
Hospital	53	81·5%
Home	6	9·2%
Other facility	6	9·2%
Physician-adjudicated cause of death
Recurrent myocardial infarction	17	26·2%
Heart failure	17	26·2%
Renal failure	10	15·4%
Indeterminate	21	32·3%
	Median	IQR
Days from hospital presentation to death	3	1,16

Table 4.

Univariate and Multivariate analysis of predictors of thirty-day mortality following acute myocardial infarction, northern Tanzania, 2019

	Participants surviving to 30 days n(%)	Participants dying within 30 days n(%)	Univariate OR (95% CI)	P	Multivariate OR (95% CI)	p
Male sex	51(60·0%)	38(58·5%)	0·94(0·48,1·82)	0·849	0·86 (0·41,1·80)	0·693
Post-primary education	40(47·1%)	13(20·0%)	0·23(0·05,0·94)	0·025*	0·20 (0·03,1·06)	0·062
Poor diet	77(90·6%)	61(93·8%)	1·55(0·46,6·27)	0·466
Sedentary lifestyle	50(58·8%)	40(61·5%)	1·12(0·58,2·18)	0·737
Self-reported History of Diabetes	20(23·5%)	7(10·8%)	0·40(0·15,0·98)	0·044*	0·32 (0·10,0·89)	0·037*
Self-reported History of hypertension	62(72·9%)	32(50·0%)	0·36(0·18,0·72)	0·003*	0·34 (0·15,0·74)	0·008*
Self-reported History of hyperlipidemia	7(8·2%)	1(1·5%)	0·20(0·01,1·18)	0·138	··	··
Self-reported History of chronic kidney disease	9(10·6%)	7(10·8%)	1·02(0·34,2·95)	0·972	··	··
Current Alcohol use	25(43·8%)	27(51·9%)	1·37(0·65,2·96)	0·400	··	··
Current Tobacco use	8(25·8%)	7(31·8%)	1·34(0·38,4·60)	0·632
Given Aspirin in the ED	23(27·1%)	12(18·5%)	0·62(0·27,1·34)	0·217	··	··
Taking antiplatelet at presentation	14(16·5%)	4(6·2%)	0·34(0·09,1·03)	0·054	0·19 (0·04,0·65)	0·016*
	Mean (s.d)	Mean (s.d.)		p
Age (years)	60·08(16·0)	63·34(21·17)		0·303	1·02 (1·00,1·04)	0·054
Troponin level (ng/ml)	0·43 (1·32)	1·70(5·05)		0·053	1·18 (1·02,1·53)	0·118
Body mass index (kg/m2)	24·57(4·29)	22·83(4·26)		0·014*	0·92 (0·84,1·01	0·072

ED: Emergency department

The results of a multivariate analysis of predictors of death are summarized in Table 4. Self-reported history of diabetes (OR 0·32, 95% CI 0·10–0·89, p = 0·04), self-reported history of hypertension (OR 0·34, 95% CI 0·15–0·74, p = 0·01), and taking antiplatelet at presentation (OR 0·19, 95% CI 0·04–0·65, p = 0·02) were associated with lower odds of 30-day mortality.

4. Discussion

To our knowledge, this is one of the first studies to assess multiple thirty-day outcomes following acute MI in sub-Saharan Africa. In northern Tanzania, we found that thirty days following MI, the majority of patients had died over three-quarters of those surviving reported persistent anginal symptoms. Few patients understood their diagnosis, and one in five surviving patients was re-hospitalized. Furthermore, use of evidence-based secondary prevention medications such as antiplatelet therapy was low. These findings call attention to an urgent need to develop patient-centered interventions to improve MI care and outcomes in Tanzania.

As discussed elsewhere, the thirty-day mortality rate within this cohort is among the highest reported globally.⁴ The majority of deaths occurred within three days of hospital presentation and occurred in-hospital. Recent quantitative and qualitative studies from Tanzania have identified potential contributors to poor MI outcomes, including inadequate physician training, lack of diagnostic capacity, frequent MI misdiagnosis, and an inefficient referral system.^4,17 Our findings suggest that short term MI outcomes in Tanzania are substantially worse than in other world regions, and our data identify multiple potential explanations for these poor outcomes: low rates of ED aspirin administration, lack of local percutaneous coronary intervention capacity, low uptake of secondary prevention therapies, inconsistent follow-up appointments, and poor patient understanding.

Even among the patients who survived to thirty days, the majority (84%) reported persistent anginal symptoms, which may be due in part to the inability to obtain revascularization in these patients. This stands in contrast to what has been observed in high-income settings: in the United States for example, approximately 17% of patients report persistent angina thirty days following MI.¹⁸ In addition to high prevalence of persistent anginal symptoms, patient understanding of their disease appeared to be low: only a minority were aware that they had a recent MI. Improved understanding of disease and risk factors leads to improved health outcomes in a wide variety of settings, including in Tanzania.^19,20 Perhaps most concerningly, few patients reported taking secondary prevention medications such as antiplatelet medications, statins, and antihypertensives, which are known to reduce morbidity and mortality following acute MI.^21–23 The 5% of patients observed to be taking antiplatelet therapy following acute MI stands in stark contrast to data from other settings: in high-income settings like Australia and the United States, the vast majority of patients are taking antiplatelets and other recommended secondary prevention therapies.^24,25 Even in low- and middle-income countries outside of SSA, a recent WHO study found that 81% of patients with recent MI were taking aspirin.²⁶ The reasons for low uptake of secondary prevention therapies warrant further investigation. Possible explanations for the low usage rates of these therapies include physician failure to prescribe these medications at discharge, cost of medications, and patient non-adherence, among others. Our study methods did not include observation of the hospital discharge process, and therefore we are unable to determine whether these medications were in fact prescribed at discharge. Further study is needed to detail specific barriers to use of aspirin and other therapies following hospitalization.

Furthermore, few patients reported receiving 30-day follow-up and one in five were re-hospitalized for recurrent AMI symptoms. The 20% unscheduled 30-day rehospitalization rate observed in our study is also worse than what has been reported in other settings: recent studies from China and the United States reported 30-day rehospitalization rates of 6% and 14%, respectively.^27,28 Statistically significant predictors of thirty-day survival included self-reported history of diabetes, self-reported history of hypertension, and taking an antiplatelet agent on presentation. These results are in contrast to prior studies of MI outcomes in other settings: a review of MI outcomes in Europe, for example, found that diabetes and hypertension are significant predictors of death.²⁹ The apparent protective effect of a self-reported hypertension and diabetes in our cohort was unexpected. However, in the sub-Saharan African context, where a large majority of patients with hypertension and diabetes are unaware of their diagnoses,^30–32 a self-reported history of diabetes and hypertension may be a proxy for health awareness, engagement with the healthcare system, and pre-existing use of secondary prevention therapies. This finding emphasizes the critical need for expanded screening, treatment, and education for hypertension and diabetes in SSA. The association between ongoing use of antiplatelet agent and improved MI mortality is unsurprising; treating MI patients with aspirin was recently cited by the WHO as a “best buy” for reducing noncommunicable disease morbidity and mortality.^{33, 34} Given the low numbers of MI patients observed to be treated with aspirin in the ED or taking aspirin on thirty-day follow-up, there is an urgent need to develop interventions to increase aspirin use in this patient population. Unfortunately, a recent systematic review of the WHO recommendations found no published studies of interventions to increase aspirin therapy for MI patients in low-income countries.³⁵

4.1. Study Limitations

The findings of this study must be interpreted in light of its limitations. A major limitation of this study is that only patients presenting to a referral center hospital with typical MI symptoms were enrolled. Those who did not survive to hospital chose not to present hospital at all, went to a different facility, or had an atypical presentation were not included in our study. These patients likely had worse outcomes and perhaps were even less likely to be taking appropriate preventative medications. Furthermore, we relied on patient self-report to identify their comorbidities, such as hypertension and diabetes, which likely led to an underestimation of the prevalence of comorbid disease. Similarly, we relied on self-report for medication use at thirty-day follow-up. To maximize the accuracy of these data, research assistants asked the patients to read their prescriptions to them over the phone and when in-home follow-up visits were done, research assistants directly looked at the prescriptions. Nonetheless, it is possible that in some cases patients misidentified or forgot certain medications, which would have led to an underestimation of the proportion of patients on appropriate preventative therapy. Furthermore, due to resource limitations at the study site, we were unable to obtain coronary angiography or echocardiography on our participants and therefore cannot describe the severity of our participants’ underlying coronary artery disease or their ejection fractions, which may have been important unexplored predictors of death. Our study may have been under-powered to detect other important associations with death. Finally, as discussed elsewhere⁴, a major limitation of the parent surveillance study from which our participants were enrolled was that we were unable to completely exclude other causes of troponin elevation, such as pulmonary embolism or myocarditis. Due to resource limitations at the study site, we were unable to obtain coronary angiography, echocardiography, or serial troponins to confirm coronary atherothrombosis in our participants. We attempted to mitigate this limitation by excluding patients with self-reported fever and only enrolling patients with symptoms suggestive of acute MI. Moreover, as discussed elsewhere⁴, renal dysfunction was uncommon among our patient population and excluding patients with renal dysfunction did not substantially affect the number of participants meeting the study definition for MI or the thirty-day mortality rate among those with MI. Nonetheless, it remains possible that some participants without acute MI were included in our MI cohort; as advanced cardiac diagnostic testing becomes more widely available in SSA, additional study will be needed to confirm our findings.

In conclusion, in northern Tanzania, thirty-day outcomes following acute MI are poor and there are multiple opportunities to improve post-MI care. Among this cohort, self-reported hypertension, self-report diabetes, and taking antiplatelet medications were associated with lower thirty-day mortality, likely reflecting prior preventive care and health knowledge. Further investigation is needed to understand these findings. Interventions are needed to improve care and outcomes of MI in Tanzania.

Highlights.

• There is a rising burden of myocardial infarction (MI) within sub-Saharan Africa.

• In northern Tanzania, thirty-day outcomes following acute-MI are poor.

• Mortality following acute-MI is associated with comorbidities and medication usage.