Clinical determinants of oxygen saturation and length of hospitalisation of COVID-19 patients: A cross-sectional study in Indonesia

Abstract

Aims

Indonesia was one of the countries with the highest COVID-19 positive cases. Understanding the length of hospitalisation is critical for anticipating bed demand and resource allocation, such as oxygen. This study aims to examine the determinants of oxygen saturation and the length of hospitalisation in Hermina Mekarsari Hospital, West Java, Indonesia.

Methods

This cross-sectional study uses medical records from June to August 2021. The inclusion criteria were: COVID-19 patients aged between18 and 65, fully conscious, and not using mechanical ventilation. Participants who passed away during hospital stay were excluded. We used demographic information, laboratory data, and the clinician's assessments of the patients admitted to the hospital. Linear regression was performed for oxygen saturation on day seven, while logistic regression analysis was conducted to predict the length of hospital stay.

Results

In total, 371 participants with an average age of 47.2 (standard deviation 15.8) years were included. Most participants were female (57.7%) and smoking (78.4%). The results indicated that decreasing oxygen saturation was reported in vomiting patients (β = 1.63, p-value = .001), hypertensive patients (β 1.18 with, p-value = .034), and patients with the increased respiratory rate (β = 0.28, p-value = .000). In the logistic regression, we found that respondents who experienced dyspnoea, headache, fever, an increasing number of D-Dimer and blood glucose, and those with diabetes mellitus were more likely to stay more than 14 days.

Conclusions

Oxygen saturation was influenced by vomiting, hypertension, and increasing respiratory rate. Length of hospitalisation of more than 14 days was influenced by dyspnoea, headache, fever, increased number of D-Dimer, blood glucose, and diabetes mellitus. Identifying the determinants of oxygen saturation and length of stay can inform health professionals in designing a suitable intervention to reduce mortality and length of stay among COVID-19 patients in Indonesia.

Introduction

The novel coronavirus disease 2019 (COVID-19) was first discovered in Wuhan in December 2019 before spreading worldwide. In Indonesia, the deaths related to COVID-19 reached 143,592 cases from a total of 4,249,323 positive cases.¹ In mid-2020, World Health Organization (WHO) reported that the number of cases in Indonesia increased by 50% or more.¹ From January to April 2022, the total number of cases was 6,015,748, with 155,164 deaths.¹

A previous study pointed out that 57% of deaths due to COVID-19 occurred in men aged between 18 and 59, with a cumulative mortality rate of 13 deaths in every 100,000 people.² WHO reported that Indonesia's health system was under tremendous strain due to the rising prevalence of COVID-19, including the supply of oxygen (WHO, 2021). Oxygen saturation ≤ 90% predicts mortality among COVID-19 patients with respiratory distress after oxygenation.² Therefore, it is crucial to determine the risk variables associated with oxygen saturation in the triage of patients.

Hospitalisation due to COVID-19 profoundly affected global health systems, i.e., overcrowding health facilities, demanding many health experts, causing burnout, and rising health finance and medical costs.³ A previous investigation has revealed that in Ghana, the presence of hypertension and diabetes mellitus influenced the length of stay among COVID-19 patients.⁴ However, their data is limited to only participants’ characteristics and comorbidities as the predictor. A more thorough understanding of the determinants of the length of hospitalisation is critical for anticipating bed demand and resource allocation, especially in resource-constrained settings. To fill the gaps and support the health care services, this study aims to identify the determinants of oxygen saturation and length of hospitalisation among COVID-19 patients in Indonesia.

Methods

This cross-sectional study uses data from the medical record in Hermina Mekarsari Hospital, West Java, Indonesia. West Java was the province where the first COVID-19 cases were confirmed.⁵ The data were collected in the second wave of the epidemic, between May and August 2022, when the number of COVID-19 cases increased from an average of 1170 cases per week to 6949 cases per week in West Java. The inclusion criteria were: (a) COVID-19 positive, (b) aged between 19 and 64, (c) fully conscious, and (d) not using mechanical ventilation. We excluded participants who passed away during their treatment at the hospital.

The sample size was calculated based on the previous study, reporting an odds ratio (OR) of 0.2 between comorbidities on length of stay⁶ to achieve the power of 80% and a significance level of 0.05. We set the final sample size to at least 371 patients. We extracted demographic and health characteristic data, signs and symptoms, comorbidities, clinical assessment, and laboratory indices from medical records when patients were admitted to the hospital. The study protocol was approved by The Ethical Review Board of Saint Carolus Health Science, Indonesia (No. 005/KEPPKSTIKSC/I/2022).

We analysed the data using the Statistical Package for the Social Sciences for Windows (version 26.0; SPSS, Chicago, IL). The descriptive statistics were analysed using percentage, frequency, mean, and standard deviation to report the distribution of the variables. In bivariate analysis, we used a t-test to compare the mean between oxygen saturation and some potential factors, for instance, demographic and vital signs, signs and symptoms at admission, comorbid, and laboratory indices. We used linear regression to examine the factors of oxygen saturation among the COVID-19 patients, while binary logistics was used to analyse the clinical determinants of length of stay.

Results

Characteristics of participants

Of the 371 participants recruited, all completed the survey. The characteristics of the participants are described in Table 1. The results indicated that most participants were female, smoking, had several symptoms for instance dry cough, fever, tachipnoea. Additionally, most of the participant also had comorbidity.

Table 1.

Participants’ characteristics.

Variables	Frequency (n = 371)	Percentage (%) (n = 371)
Age, Mean (SD)	47.2	15.8
Gender
Female	214	57.7
Male	157	42.3
Smoking
Yes	291	78.4
No	80	21.6
Haemoglobin, Mean (SD)	13.1	7.5
Thrombocyte, Mean (SD)	276,044.4	116,244.4
D-dimer, Mean (SD)	481.1	557.1
Blood glucose, Mean (SD)	152.5	116
Temperature, Mean (SD)	37.5	1.0
Heart rate, Mean (SD)	94.3	10.7
RR, Mean (SD)	27.2	3.7
Systole, Mean (SD)	124.8	15.9
Dry cough
Yes	328	88.4
No	43	11.6
Fever
Yes	328	88.4
No	43	11.6
Tachypnoea
Yes	306	82.5
No	65	17.5
Chest distress
Yes	60	16.2
No	311	83.8
Myalgia
Yes	122	32.9
No	249	67.1
Headache
Yes	123	33.2
No	248	66.8
Vomitus
Yes	74	19.9
No	297	80.1
Diarrhoea
Yes	35	9.4
No	336	90.6
Diabetes mellitus
Yes	92	24.8
No	279	75.2
Immunodeficiency
Yes	1	0.3
No	370	99.7
Hypertension
Yes	92	24.8
No	729	75.2
Pregnancy
Yes	58	15.6
No	313	84.4
Comorbidity
Yes	208	56.1
No	163	43.9

Note. N: frequency, SD: standard deviation, RR: respiratory rate, %: percentage.

Clinical determinants of oxygen saturation and length of stay

The determinants of oxygen saturation and length of hospitalisation are described in Tables 2 and 3. The results indicated that participant who had vomitus, hypertension, and also tachypnoea were significantly reported decreasing oxygen saturation. From analysis, we also found that patient who had tachypnoea.

Table 2.

Regression analysis of oxygen saturation among COVID-19 patients (n = 371).

Variable	M	SD	B	Standardised β	p-Value
Age
<30 years	303	81.7	−0.001	−0.002	0.968
≥30 years	68	18.3	Ref	Ref
Wheezing
Yes	29	7.8	−1.25	−0.09	0.132
No	342	92.2	Ref	Ref
Fever
Yes	262	70.6	−0.06	−0.01	0.901
No	109	29.4	Ref	Ref
Chest distress
Yes	60	16.2	0.55	0.05	0.383
No	311	83.8	Ref	Ref
Vomitus
Yes	74	19.9	−1.63	−0.17	0.001
No	297	80.1	Ref	Ref
Diabetes mellitus
Yes	92	24.8	0.22	0.02	0.710
No	279	75.2	Ref	Ref
Hypertension
Yes	92	24.8	−1.18	−1.22	0.034
No	279	75.2	Ref	Ref
Pregnancy
Yes	58	15.6	0.14	0.01	0.848
No	312	84.1	Ref	Ref
Comorbid
Yes	208	56.1	0.08	0.01	0.884
No	163	43.9	Ref	Ref
Headache
Yes	123	33.2	0.09	0.01	0.853
No	248	66.8	Ref	Ref
Haemoglobin
Abnormal	166	44.7	0.04	0.02	0.641
Normal	205	55.3	Ref	Ref
D-dimer
Abnormal	127	41.2	0.00	−0.06	0.264
Normal	181	58.8	Ref	Ref
Blood glucose
Abnormal	71	20.2	−0.002	−0.063	0.303
Normal	281	79.8	Ref	Ref
Tachypnoea
Abnormal	359	96.8	−0.28	−0.26	0.000
Normal	12	3.2	Ref	Ref
HR
Abnormal	76	20.5	−0.03	−0.08	0.118
Normal	295	79.5	Ref	Ref

Note: M: mean, SD: standard deviation, B: beta, RR: respiratory rate, HR: heart rate.

Bold: Significant value.

Table 3.

Regression analysis of length of hospitalisation more than 14 days (n = 371).

Variable	N	%	OR (95% CI)	SE	p-Value
Tachypnoea			3.80 (1.11–13.0)
Yes	306	82.5	Ref	0.63	0.033
No	65	17.5
Headache			1.23 (0.11–0.51)
Yes	123	33.2	Ref	0.40	0.000
No	248	66.8
Fever			2.24 (1.07–4.68)
Yes	262	70.6	Ref	0.38	0.032
No	109	29.4
D-dimer			1.00 (1.00–1.001)
Abnormal	127	41.2	Ref	0.00	0.006
Normal	181	58.8
Blood glucose			1.00 (1.00–1.005)
Abnormal	71	20.2	Ref	0.01	0.006
Normal	281	79.8
HR
Abnormal	76	20.5	0.58 (0.29–1.13)	0.34	0.110
Normal	295	79.5	Ref
Diabetes mellitus			2.23 (1.08–4.61)
Yes	92	24.8	Ref	0.37	0.030
No	279	75.2
Pregnancy
Yes	58	15.6	0.27 (0.06–1.21)	0.77	0.088
No	312	84.1	Ref
Comorbidity
Yes	208	56.1	0.93 (0.43–1.93)	0.40	0.847
No	163	43.9	Ref

Note: N: total number, %: percentage, OR: odd ratio, SE: standard error, HR: heart rate.

Bold: Significant value.

Discussion

In this study, tachypnoea has the strongest relationship with oxygen saturation and length of stay. This is in line with the previous study that revealed the manifestation of tachypnoea due to the distribution of the majority of Angiotensin-converting enzyme 2 (ACE2) receptors in lung epithelial cells, where then the SARS-CoV-2 virus enters and destroys the alveoli.^7,8 T helper CD4 cells and T cytotoxic CD8 cells begin to be trapped in the lung tissues. These cells play a role in fighting off the viruses, but they also play a role in the inflammation and subsequent lung injuries.⁹

It is important to note that vomiting significantly predicted lowered oxygen saturation. This aligns with the previous study with 27 autopsy cases, stating that vomiting followed by inhalation of gastric contents into the bronchial tract may cause pulmonary inflammation.¹⁰ The severity of lung injury increases significantly with the large volume of gastric contents aspirated and the low potential hydrogen (PH) of gastric contents.¹¹ Patients experiencing the aspiration of gastric contents may experience respiratory signs and symptoms, such as wheezing, coughing, dyspnoea, cyanosis, pulmonary oedema, and hypoxaemia, which may develop into severe ARDS.¹²

Our study also found that hypertension contributed to lowered oxygen saturation. This study also aligned with the previous study, revealing that COVID-19 patients who suffered from hypertension as comorbidity showed low oxygen saturation levels.¹³ The possible explanation for this is the Angiotensin-Converting Enzyme 2 (ACE2) receptor as the site of attachment for the SARS-CoV-2 virus on the cell surface to enter the cell.^14,15 The ACE2 catalyses the conversion of angiotensin II (a vasoconstrictor peptide) to angiotensin 1–7 (a vasodilator). This can bring influence the occurrence of hypertension in COVID-19 patients.

In addition, our study has shown that dyspnoea, fever, and diabetes mellitus (DM) are prognostic factors for a more extended hospitalisation. This result is consistent with the previous study, claiming that patients with more severe symptoms tend to be hospitalised longer than patients with less severe symptoms of the same diagnosis.¹⁶ In COVID-19 cases, dyspnoea and fever are signs of pneumonia, which could be classified as moderate symptoms of COVID-19.¹⁷ Meanwhile, studies on immunity showed that poorly controlled diabetes could inhibit the lymphocyte response and impair the function of monocytes, macrophages, and neutrophils,^18,19 which means a longer hospitalisation.

Finally, this study shows that high D-Dimer also influences the length of hospitalisation. An increase in D-Dimer is a sign of coagulopathy, indicating that severe sepsis has occurred, leading to a critical illness.²⁰ This suggests that the D-Dimer level is related to the disease severity, which has also been proven in a previous study describing that the D-Dimer level at admission is a risk factor for severe symptoms and death.⁶

This study had some limitations. First, our study recruits participants from one hospital, so the generalisation is low. Future study needs to include multiple centres and locations. Second, the causal relationship was difficult to build since we used a cross-sectional design. Longitudinal studies are required to explore the changes in oxygen saturation among COVID-19 patients. This study indicates important variables associated with oxygen saturation, including tachypnoea, vomiting, and hypertension. Health professionals should consider some variables, including dyspnoea, headache, fever, D-dimer, blood glucose, and diabetes mellitus, in reducing the length of hospitalisation of COVID-19 patients.

Funding

This research received no funding.

Conflict of interest

The authors declare no conflict of interest.

Acknowledgment

We thank Douglas S Umboh, MD, Director of Hermina Mekarsari Hospital, for the permission in conducting this research.