A cost-minimisation analysis comparing different antibiotic regimens used in treating all-cause bacterial pneumonia in Hong Kong

Abstract

Background: To find out the antibiotic treatment regimens with the lowest cost for all-cause bacterial pneumonia, a study to compare the costs of different antibiotic regimens in the treatment of patients diagnosed with all-cause bacterial pneumonia who required hospitalisation was carried out.

Methodology: This was a multicentre, retrospective study of patient medical records. The primary aim was to examine whether the initial choice of antibiotic had affected the total cost of treatment, while the secondary aim was to find out whether the initial choice of antibiotic had affected the initial treatment failure rates and death rates. A cost-minimisation analysis (CMA) from a public hospital perspective was employed.

Results: A total of 333 patient medical case notes were reviewed. The most commonly prescribed antibiotic regimen was amoxycillin-clavulanate (AC) followed by amoxycillin-clavulanate plus macrolide (ACM) and quinolone (Q). In the study population, no statistical significance could be detected between the mean cost of the three regimens. In the subgroup analysis of patients with a history of chronic obstructive pulmonary disease (COPD) and patients with a history of smoking, the Q regimen appeared to be the least expensive.

Conclusion: In the study population, no significant difference could be identified between the mean cost of the three antibiotic regimens. In a special populations such as patients with a history of COPD and patients with a history of smoking, the Q regimen appeared to be superior. Further studies in these areas are needed.

Keywords::

• antibiotics
• cost-minimisation analysis
• all-cause bacterial pneumonia

Introduction

Disease overview and prevalence

Pneumonia is an acute infection of the lung parenchyma including alveolar spaces and interstitial tissue¹. It is a serious public health problem which brings about a very large number of hospital admissions each year². Pneumonia ranks sixth among all disease categories as a cause of death and is the most common lethal nosocomial infection in the United States¹. In 1999, the age-adjusted death rate due to influenza and pneumonia was 23.6 per 100,000 persons³ and estimates of the incidence of community-acquired pneumonia (CAP) range from 4–5 million cases per year, with about 25% requiring hospitalisation⁴. In Hong Kong, the number of patients suffering from pneumonia requiring hospitalisation in 2005/06 was more than 43,000 and among these more than 4,000 died⁵.

Bacterial pneumonia is the most common pneumonia in adults. In the management of bacterial pneumonia, antibiotics are one of the essential weapons. Although many antibiotics are marketed for the treatment of bacterial pneumonia, inappropriate use of antibiotics may result in increased morbidity, mortality and healthcare costs⁶. Therefore, the selection of an appropriate antibiotic as empirical treatment has a significant impact on the healthcare budget.

In Hong Kong, the current treatment recommendations for different types of bacterial pneumonia, according to the third edition of the Inter-hospital Multi-disciplinary Programme on Antimicrobial ChemoTherapy (IMPACT) Guideline published by the Hospital Authority (HA) of Hong Kong in 2005⁶, are summarised in Table 1.

Objective of the study

In order to identify the antibiotic treatment regimens with the lowest cost for bacterial pneumonia, the study team compared the cost of different antibiotics regimens in the treatment of patients diagnosed with all-cause bacterial pneumonia that required hospitalisation.

It was hoped that the study findings would provide insight into the selection of the most appropriate antibiotic regimen in the treatment of bacterial pneumonia in order to achieve an effective use of healthcare resources.

Methodology

Study design

This was a multicentre, retrospective study of patient medical records. The primary aim of this study was to see whether the initial choice of antibiotic had affected the total costs of treatment for patients who had been diagnosed with all-cause bacterial pneumonia that required hospitalisation, while the secondary aim was to see whether the initial choice of antibiotic had affected the initial treatment failure rates and death rates. In this study, a cost-minimisation analysis (CMA) from a hospital perspective was employed in which the study team assumed the treatment outcomes of different patients were the same (i.e., eradication of infection upon discharge)⁷.

Study site

This was a multicentre study and three hospitals (hospitals A, B and C) under the management of the Hospital Authority (HA) of Hong Kong were involved. Approval was obtained from the ethics committees of the three hospitals before the study was commenced. The Hospital Authority is a statutory body established to manage all public hospitals in Hong Kong⁸. About 92.8% of secondary and tertiary care and 100% of extended and long-term care services are provided by the HA⁸.

Antibiotic regimens evaluated

This study was designed to evaluate the most common antibiotic regimens used in Hong Kong for the treatment of patients with bacterial pneumonia that required hospitalisation. For simplicity of subsequent data analysis, the antibiotics that the study team planned to evaluate were placed into four groups, based on their chemical composition. The antibiotic groups are shown in Table 2.

Study population

Approximately 350 patients were recruited. In order to generate a list of patients, the Hospital Authority's internal computer system was used to retrieve data from the patients' electronic medical records.

Inclusion criteria

Any patient aged over 18 at the time of hospitalisation ands who attended any of the three selected hospitals during the period May 2005 to April 2006 inclusive with a principal discharge diagnosis of all-cause bacterial pneumonia was regarded as fulfilling the inclusion criteria of the study.

Exclusion criteria

Patients with the concomitant use of systemic antimicrobial(s) during the study period and/or had been admitted with a diagnosis of pulmonary embolism, septic shock and/or patients with cystic fibrosis and/or those who were immunocompromised (e.g., with HIV infection, on anti-cancer drugs, on immunosuppressants), and/or patients who had unexplained changes of antibiotics during the review period, together with patients with untraceable medical records, were excluded from the study.

Data collection

The patient list was generated with the inclusion and exclusion criteria as stated above and medical records of selected patients reviewed. Data collected included demographic characteristics, such as age, gender, source of admission, history of smoking and history of chronic obstructive pulmonary disease (COPD); name, dose and duration of medications (i.e., both antibiotics and non-antibiotic drugs) received; number of diagnostic tests (e.g., haematological and microbiological laboratory tests) performed and number of visits by allied-health professionals and duration of hospital stay (i.e., admission and discharge dates).

To ensure the completeness of the data collected, the study team validated the data collected from the electronic medical records with paper-based records for each patient. The study team also checked the validity of the study assumption (i.e., eradication of infection upon discharge) to ensure that when a patient was discharged (with or without a course of antibiotics at discharge), infection was indeed eradicated and no more medical follow up for residual infection was needed.

Outcome measurement

The primary outcome measure of this study was the actual cost per infection cured for the antibiotic regimens evaluated, while the secondary outcome measures were the initial treatment failure and death rates arising from the antibiotic regimens evaluated.

Initial treatment failure

Initial treatment failure was defined as signs or symptoms relevant to the original infection persisting or progressing after at least 3 days of first antibiotic therapy, a change of antimicrobial agents being necessary, or the fact that patients died due to pneumonia. The number of patients who experienced initial treatment failure with one particular antibiotic regimen was divided by the total number of patients receiving that regimen as initial choice of antibiotic to obtain the initial treatment failure rate of that particular antibiotic regimen.

Data analysis

Computer software was used in the analysis of study data. The data collected was first categorised by using Microsoft Excel 2003, while for the subsequent statistical analysis, the Statistical Package for Social Sciences (SPSS) Version 13.0 for Windows was used.

Statistical analysis methods

Different statistical analysis methods (e.g., ANOVA, chi-square test, independent samples t-test, etc) were employed depending on the nature of the data^9,10. In this study, a probability of type I error of 0.05 was used to determine the statistical significance.

Cost reference

The reference cost of medications was extracted from the Hospital Authority's internal database as of January 2007; while the costs for consultation of different healthcare professionals and various examinations were based on a gazette published by the government of the Hong Kong Special Administrative Region in 2003¹¹.

Results

Demographic data

During the medical case note review period from July 2006 to October 2006, 333 patient medical case notes were reviewed (hospital A: 131, hospital B: 102, hospital C: 100). Patient demographics were generally similar between individual hospitals. A detailed account of patient demographic data is summarised in Table 3.

Antibiotic regimens employed

The most commonly prescribed antibiotic regimen was amoxycillin-clavulanate (AC), which accounted for 63.66 % (212 of 333) of the medical case notes reviewed. The two other commonly prescribed regimens were amoxycillin-clavulanate plus macrolide (ACM) and quinolone (Q), which accounted for 17.42% (58 of 333) and 8.41% (28 of 333) of the medical case notes reviewed, respectively.

During the medical case note review period, the study team also encountered antibiotic regimens such as ampicillin-sulbactam, cephalosporins (e.g., cefuroxime, ceftriaxone, cefotaxime and cefoperazone-sulbactam) and macrolides (e.g., erythromycin, clarithromycin). However, the sample sizes for the above antibiotics were too small and so no meaningful results during initial data analysis could be obtained. As a result, these antibiotic regimens were grouped as ‘others’ and were not involved in subsequent analysis. A brief summary of different antibiotic regimens encountered during the medical case note review period is shown in Table 4.

Initial treatment failure and death rates

The number of patients who experienced initial treatment failure and the number patient who died during treatment were used to calculate the ‘initial treatment failure rate’ and the ‘death rate’, respectively. The study team used chi-square tests to compare the difference in death rates between the three antibiotic regimens and found that the ACM regimen death rate was statistically lower than that of both the AC regimen (p=0.0029) and the Q regimen (p=0.0032), while for the AC and Q regimens, no statistical difference could be found (p=0.9302).

Adjusted total cost of treatment

For the three antibiotic regimens analysed, total costs of treatment were adjusted by dividing the original costs by the respective survival rates (i.e., 1 – death rate) of that particular regimen. Adjusted total cost of treatment was used in subsequent analysis because it could better reveal the actual unit cost of treatment per infection cured and avoid any potential overestimation of treatment cost. A formula of the adjusted total cost is as follow:

The adjusted total cost of treatment for the remaining three groups was analysed by using an analysis of variance (ANOVA) model. The results of the ANOVA model were as shown in Figure 1.

Extra hospital stay

The differences in mean duration of hospital stay between success cases and failure cases for the three antibiotic regimens were compared by independent samples t-tests. In the t-tests, all the death cases were excluded as they could not reveal the actual duration of hospital stay per infection.

The test results were as shown in Table 5. The mean differences for the AC and ACM regimens revealed that failure cases had a statistically longer mean duration of hospital stay (p=0.0008 and p=0.0472, respectively), while for the Q regimen, a difference of 9.14 days was still observed, but this difference was not statistically significant – probably because the sample size of the Q regimen was too small.

Subgroup analysis

Patients with a history of COPD and those with a history of smoking were selected for subgroup analysis. For these two special populations, the initial treatment failure rates, death rates and adjusted total cost of treatment were analysed.

Patients with a history of COPD

The initial treatment failure and death rates of this subgroup were analysed by chi-square test. Results are shown in Table 6. The initial treatment failure rate was statistically higher in patients with a history of COPD (p=0.0197), whereas for the death rate, no statistical difference could be found (p=0.5792).

The adjusted total cost of treatment for the different antibiotic regimens in this subgroup were compared by ANOVA. Results are shown in Figure 1. As shown in the figure, the mean total cost (Hong Kong dollars, HK$) of treatment for the Q regimen was HK$52,353.49, which was much lower than that of the AC (HK$68,470.66) and ACM (HK$76,149.15) regimens, although no statistical difference between the mean costs could be seen (p=0.3050 for AC and p=0.2229 for ACM regimens).

The initial treatment failure and death rates of different antibiotic regimens in this subgroup are shown in Table 7. The initial treatment failure rates for the AC, ACM and Q regimens were 42.42% (28 of 66 patients), 50.00% (6 of 12 patients) and 11.11% (1 of 9 patients), respectively, while the death rates for the AC, ACM and Q regimens were 16.67% (11 of 66 patients), 0.00% (0 of 12 patients) and 0.00% (0 of 9 patients), respectively.

Patients with a history of smoking

The number of patients with a history of smoking was calculated as the sum of smokers and ex-smokers. Analysis to this subgroup was similar to that of patients with a history of COPD. The initial treatment failure rate and death rate of the non-smoking subgroup were analysed by chi-square test. Results are shown in Table 6. As shown in the table, neither initial treatment failure rate nor death rate had any statistical difference (i.e., initial treatment failure rate p=0.8935, death rate p=0.0607).

The adjusted total cost of treatment for different antibiotic regimens in the non-smoking subgroup was compared using an ANOVA model. Results are shown in Figure 1. As shown in the figure, the mean total cost of treatment for Q regimen was HK$51,646.17, which was much lower than that of AC (HK$80,780.83) and ACM (HK$76,484.20) regimens, although a statistical difference in mean cost could only be shown between the Q and AC regimens (p=0.0449).

The initial treatment failure and death rates of different antibiotic regimens in the non-smoking subgroup are shown in Table 7. The initial treatment failure rate for the AC, ACM and Q regimens were 33.33% (30 of 90 patients), 29.03% (9 of 31 patients) and 13.33% (2 of 15 patients), respectively, while the death rates for the AC, ACM and Q regimens were 10.00% (9 of 90 patients), 0.00% (0 of 31 patients) and 6.67% (1 of 15 patients), respectively.

Discussion

Characteristics and differences among the three hospitals

The data for this study came from three different hospitals (A, B and C). Although the patient characteristics of the three hospitals were generally similar, differences could still be identified. The nature of the hospital specialisation might help to explain the difference.

Hospital C is an acute hospital; most cases admitted were referred by its own A&E department – the A&E department would admit patients with different backgrounds, which explains why hospital A admitted patients with a wider age range.

Compared with hospital C, hospital A is a small-scale regional hospital that specialises in the management of various chest diseases; thus it is understandable why hospital A had the highest proportion of patients with a history of COPD (45%).

Hospital B is a small scale regional hospital which mainly provides sub-acute and rehabilitation services. There were a few old age homes nearby and this explained why hospital B had a higher proportion of patients who were admitted from old age home (41.2%), and had the highest patient average age (mean 79.19) among the three hospitals.

Total cost of treatment among different regimens

In this study, it was found that average total cost of treatment with Q regimen (mean HK$51,236.43) was slightly lower than that of AC regimen (mean HK$54,542.13) or ACM regimen (mean HK$53,476.71) in general population (Figure 1). Unlike other similar studies conducted overseas, the study team was unable to detect any significant difference in the total cost of treatment among the three antibiotic groups. In a US study involving 3,000 patients with pneumonia, those treated with a β-lactam plus a macrolide had reduced mortality compared with the group received a β-lactam alone¹². Another US prospective study involving 203 patients also concluded that quinolone (gatifloxacin) monotherapy for CAP patients requiring hospitalisation provided an economic advantage compared to cephalosporin with or without a macrolide¹³.

In this study, the failure rates of empirical antibiotic treatment with the Q and ACM regimens were about 10% lower than that with the AC regimen (i.e., 21.43% for the Q regimen, 22.41% for the ACM regimen and 33.96% for the AC regimen) (Table 4). In similar studies conducted overseas, lower failure rates of empirical antibiotic treatment usually resulted in a lower total treatment cost. In this study, the study team also found the total treatment cost was lower in the regi-mens with lower failure rates, but the differences were not statistically significant. One of the reasons behind may be the small sample size in the ACM and Q regimens. According to the IMPACT Guideline, IV/PO amoxycillin-clavulanate or ampicillin-sulbactam with or without a macrolide are recommended as the first-line empirical therapy for CAP, while the use of quinolones is not (Table 1); Although the result of a meta-analysis study suggested that quinolones were as effective as a β-lactam plus macrolide regimen¹⁴, a US guideline also recommended the use of either quinolones or β-lactam plus macrolide as a first-line empirical regimen for CAP that requires hospitalisation¹⁵. However, in Hong Kong, the use of quinolones is reserved only for patients who are allergic to alternative agents (e.g., patients who are allergic to penicillin) or for patients with documented infection due to pneumococci with high penicillin resistance (i.e., MIC 4μg/ml)⁶. As a result, the treatment arm of the AC regimen dominated the study. Of the 333 patient medical case notes evaluated by the study team, only 28 (8.41%) and 58 (17.42%) of them were treated with the Q and ACM regimens, respectively (Table 4). As a small sample size usually contributes to type II errors – although there might be clinical differences between the groups – the study team was unable to detect them statistically¹⁰.

Being a retrospective study, the study team was unable to randomise the study samples and this might have affected the study results. As suggested by the IMPACT Guideline, the quinolones are not the drug of choice for patients with mild-to-moderate pneumonia but are reserved for patients with more severe pneumonia⁶. Those patients with more severe pneumonia might be infected by a resistant strain bacteria and they would probably need more intensive treatment. This may be the reason why patients treated with the Q regimen had a higher failure rate (21.43%) compared with similar studies conducted overseas. In the above-mentioned US prospective study, the failure rates of quinolone (gatifloxacin) monotherapy were 0% for mild-to-moderate pneumonia and 4% for severe pneumonia¹³.

Subgroup analysis – patients with a history of COPD

The observations above could be explained by subgroup analysis of patients with different disease modifying factors. Disease modifying factors, as defined by the IMPACT Guideline, are factors to be considered in choosing the empirical therapy for the management of infection⁶. A history of COPD is considered to be disease modifying factor, as patients with a history of COPD are more prone to be infected by drug-resistant Streptococcus pneumoniae ⁶. In the current study, there were 87 patients (n = 298) with a history of COPD (29.19%) (Table 6). In patients with a history of COPD, the mean total cost of treatment for the Q regimen was HK$16,117.16 and HK$23,795.66 lower than that of the AC and ACM regimens, respectively (Figure 1). In this subgroup, the initial treatment failure rate for the Q regimen was 11.11% (1 of 9 patients), which was much lower than that of the AC (42.42%, 28 of 66 patients) regimen and ACM (50%, 6 of 12 patients) regimens (Table 7). Regarding the death rate, the study team also found that all the 11 patients who died in this subgroup were treated with the AC regimen as the initial choice of antibiotic (Table 7). The above results suggest that the Q regimen is a better choice than the AC regimen or the ACM regimen in this patient subgroup. Unfortunately, due to small sample size in each treatment arm, these data were not statistically significant.

When comparing the data of patients with a history of COPD with those without it, the study team found that the initial treatment failure rate of patients with a history of COPD was much higher (40.23%, 35 of 87 patients) than those without the history (26.54%, 56 of 211 patients), which had a statistical difference of p=0.0197 (Table 6). This high initial treatment failure rate was mainly due to the failure of the AC regimen. Therefore, the above-suggested AC regimen may not be the best choice for this patient subgroup in terms of either total cost of treatment, initial treatment failure rate or death rate.

Subgroup analysis – patients with a history of smoking

Smoking has identified as a risk factor for Streptococcus pneumoniae pulmonary infections. This is because abnormalities have been found in smokers' cilia, which impair tracheo-bronchial clearance and thus allow adherence of Streptococcus pneumoniae to the buccal epithelia¹⁶.

In this study, 162 of 298 evaluated patients were non-smokers (Table 6). In the non-smoking subgroup, the mean total costs of treatment for patients treated with the AC (HK$64,270.93), ACM (HK$60,230.36) and Q (HK$60,782.26) regimens were similar. However, the figures were different in patients with a smoking history.

For the 136 patients who had a history of smoking (including smokers and ex-smokers), the mean total costs of treatment for patients treated with the AC (HK80,780.83) and ACM (HK$76,484.20) regimens were much higher than that of the Q regimen (HK$51,646.17) (Figure 1), the difference between the AC and Q regimens being statistically significant (p=0.0449) (Figure 1). The initial treatment failure rates of the AC (33.33%, 30 of 90 patients) and ACM (29.03%, 9 of 31 patients) regimens were also higher than those of the Q regimen (13.33%, 2 of 15 patients) (Table 7). These findings might be explained by the possibility that multidrug-resistant strains of Streptococcus pneumoniae were present that were not susceptible to the AC and ACM regimens. Multidrug-resistant strains of Streptococcus pneumoniae have been found worldwide. In the US, they had been found in more than 25% of cases, with some areas having an even higher incidence¹⁷. In this study, the total cost of treatment and the initial treatment failure rate for the Q regimen were much lower, which suggests that the Q regimen might be a better choice for this subgroup of patients.

Implications of this study

The IMPACT Guideline is an in-house guideline widely used in all HA institutions in Hong Kong. In this study, the study team have evaluated more than 300 different pneumonia cases and found that the choice of antibiotic regimen usually followed the recommendations of the guideline.

In this study, the total treatment cost and initial treatment failure rate of the AC, ACM and Q regimens were similar to the study population, which suggests that none of the antibiotic regimens was superior to other agents. However, use of the macrolides is associated with more side-effects and drug interactions¹⁸, and frequent use of the quinolones might also increase the chance of developing drug resistance⁶; therefore, use of the AC regimen for the treatment of pneumonia is an acceptable choice, which is consistent with the recommendations of the IMPACT Guideline.

The IMPACT Guideline suggested that disease modifying factors (e.g., patients with a history of COPD, old people's home residents and elderly patients, etc.) should be considered in choosing empirical antibiotic therapy. The study team found that the AC regimen might not be the best choice for some subgroups (e.g., patients with a history of COPD).

To conclude, the study team observed that the recommendations of the IMPACT Guideline might help to strike a balance between minimising the total treatment cost and preventing the emergence of drug resistance in treating pneumonia in Hong Kong.

Limitations

This study had several limitations in that it was a retrospective medical case notes review study.

First of all, the patients recruited in this study might have suffered from pneumonia of different severities (i.e., those patients might have different Pneumonia Severity Indexes, PSIs)¹⁹, which might have affected the initial choices of antibiotic used by physician. However, as is local practice in Hong Kong, physicians generally do not write down the PSIs of patients in medical case notes; therefore, the study team was unable to carry out data analysis by considering this potentially important information.

Secondly, this study had no randomisation of samples. Without it, the study team could not guarantee that the patients recruited were similar in terms of demographics and PSIs. This limitation is inevitable as the study was a retrospective one.

In addition, it is possible that some data might have been missed during the data collection period. Although the study team had validated the completeness of data, similar to other retrospective studies, it was not possible to guarantee the completeness of data.

Finally, this study was a multicentre trial and patients were recruited from three different hospitals with different specialties (i.e., hospitals A and B were sub-acute hospitals while hospital C was an acute hospital). The difference might have affected the demographics of patient recruited and might subsequently have affected the quality of the data collected and results of the analysis performed.

Conclusion

From the data collected in this study, the study team found that in the study population, the three antibiotic regimens, when used as the initial choice of antibiotic in the treatment of pneumonia, had no significant difference in terms of total cost of treatment. Hence, the study team could not recommend any of these regimens as the drug of choice for patients admitted to hospital with pneumonia.

In some special populations, such as patients with a history of COPD and patients with a history of smoking, the study team found that the Q regimen appeared to be superior to the AC and ACM regimens. However, whether the Q regimen should be employed as empirical treatment for these patient subgroups in Hong Kong remains to be established because other factors such as bacterial resistance should also be considered.

Figure 1. Adjusted total costs (Hong Kong dollars, HK$) with treatment of different antibiotic regimens.

Table 1. Recommendations of empirical antibiotic regimens by IMPACT Guideline.

Type of bacterial pneumonia	Preferred empirical regimens	Alternative regimens
Community-acquired pneumonia (CAP)
CAP, hospitalised in general ward	IV/PO amoxycillin-clavulanate or ampicillin-sulbactam ± a macrolide	Cefotaxime or ceftriaxone ± a macrolide
CAP, hospitalised in intensive care unit (ICU) or serious pneumonia	IV piperacillin-tazobactam or cefotaxime or ceftriaxone + a macrolide	Cefepime + a macrolide
Hospital-acquired pneumonia (HAP)
HAP, onset <4 days after admission + no previous antibiotics	IV/PO ampicillin-sulbactam or amoxycillin-clavulanate	Cefuroxime if patient is allergic to penicillins
HAP, onset <4 days after admission + had antibiotics recently OR onset <5 days after admission OR mechanical ventilation	IV ticarcillin-clavulanate or piperacillin-tazobactam ± an aminoglycoside	Cefoperazone-sulbactam or cefepime ± an aminoglycoside
Aspiration pneumonia (AP)	IV penicillin G + PO metronidazole or PO clindamycin	Amoxycillin-clavulanate or ampicillin-sulbactam

Table 2. Antibiotic regimens evaluated.

Antibiotic regimen	Examples
Penicillins	Amoxycillin-clavulanate
	Ampicillin-sulbactam
Cephalosporins	Cefuroxime
	Ceftriaxone
	Cefoperazone-sulbactam
Macrolides	Erythromycin
	Clarithromycin
	Azithromycin
Quinolones	Ciprofloxacin
	Ofloxacin
	Levofloxacin

Table 3. Demographic data.

Demographic characteristic		Hospital A (n=131)	Hospital B (n=102)	Hospital C (n=100)	Total (n=333)
Gender	Male	76 (58.02%)	51 (50.00%)	58 (58.00%)	185 (55.56%)
	Female	55 (41.98%)	51 (50.00%)	42 (42.00%)	148 (44.44%)
Age (years)	Range	29–95	26–103	19–102	19–103
	Mean (SD)	73.03 (13.58)	79.19 (14.15)	72.11 (18.42)	74.64 (15.76)
	≤60	19 (14.51%)	8 (7.84%)	21 (21.00%)	48 (14.41%)
	61–90	108 (82.44%)	79 (77.45%)	68 (68.00%)	255 (76.58%)
	>90	4 (3.05%)	15 (14.71%)	11 (11.00%)	30 (9.01%)
Smoking status	Non-smoker	53 (40.46%)	66 (64.71%)	66 (66.00%)	185 (55.56%)
	Smoker	14 (10.69%)	8 (7.84%)	10 (10.00%)	32 (9.61%)
	Ex-smoker	64 (48.85%)	28 (27.45%)	24 (24.00%)	116 (34.83%)
Old people's home resident		26 (19.85%)	42 (41.18%)	29 (29.00%)	97 (29.13%)
History of COPD		59 (45.04%)	27 (26.47%)	8 (8.00%)	94 (28.23%)

Table 4. Antibiotic regimens encountered.

Antibiotic regimen	Number of patients	Number of patients who experienced initial treatment failure (%)	Number of patients who died from treatment (%)
AC	212	72 (33.96%)	29 (13.68%)
ACM	58	13 (22.41%)	0 (0.00%)
Q	28	6 (21.43%)	4 (14.29%)
Others	35	7 (20.00%)	2 (5.71%)
Total	333	98 (29.43%)	35 (10.51%)

AC, amoxycillin-clavulanate; ACM, amoxycillin-clavulanate plus macrolide; Q, quinolones

Table 5. Extra hospital stay for treatment failure cases.

Antibiotic regimen	Initial treatment	Mean duration of stay (days)	p-value
AC	Success (n=140)	8.49	0.0008
	Failure (n=43)	13.16 (+55.12%)
ACM	Success (n=45)	9.24	0.0472
	Failure (n=13)	19.08 (+106.36%)
Q	Success (n=22)	8.36	0.5834
	Failure (n=2)	17.5 (+109.24%)

Table 6. Subgroup analysis: initial treatment failure rate and death rate.

Subgroup		No history of COPD (n=211)	History of COPD (n=87)	p-value
COPD	Initial treatment failure rate	56 (26.54%)	35 (40.23%)	0.0197
	Death rate	22 (10.43%)	11 (12.64%)	0.5792
		No history of smoking (n=162)	History of smoking (n=136)	p-value
Smoking	Initial treatment failure rate	50 (30.86%)	41 (30.15%)	0.8935
	Death rate	23 (14.20%)	10 (7.35%)	0.0607

Table 7. Subgroup analysis: initial treatment failure rate and death rate of different antibiotic regimens.

Subgroup	Antibiotic regimen		Initial treatment failure rate	Death rate
COPD (n=87)	AC	(n=66)	28 (42.42%)	11 (16.67%)
	ACM	(n=12)	6 (50.00%)	0 (0.00%)
	Q	(n=9)	1 (11.11%)	0 (0.00%)
Smoking (n=136)	AC	(n=90)	30 (33.33%)	9 (10.00%)
	ACM	(n=31)	9 (29.03%)	0 (0.00%)
	Q	(n=15)	2 (13.33%)	1 (6.67%)

Acknowledgements

Declaration of interest: The authors have no conflicts of interest that are directly relevant to the content of this manuscript.

Notes