Introduction

The clinical and economic burden of community-acquired pneumonia (CAP) in Europe is substantial. A review of the burden of CAP in Europe demonstrated that the incidence of CAP and hospitalisations for CAP are rising [1]. The incidence of CAP was shown to be higher in men than in women and to increase with age. In addition to increasing age [1–3], several other risk factors for CAP have been established, including smoking [2], immunosuppression [3] and the presence of comorbid conditions [4–7]. With an ageing population in Europe, the clinical and economic burden of CAP is expected to continue to rise over time, placing increasing pressure on hospital resources and society [1, 8–11].

Streptococcus pneumoniae is widely accepted as being the most common pathogen causing CAP. However, the frequency at which it is identified varies considerably between studies across Europe [1, 12]. In addition to S. pneumoniae, several other pathogens cause CAP, including atypical pathogens such as Legionella pneumophila and Staphylococcus aureus, and Gram-negative bacilli, including Pseudomonas aeruginosa [13–18].

It is important to understand the emerging role of different pathogens in the aetiology of CAP to effectively guide appropriate antibiotic management [19]. Inappropriate antibiotic treatment in patients with CAP has been repeatedly linked with worse outcomes [20–23]. This literature review was conducted to generate up-to-date information on the aetiology of CAP and its antibiotic management in adults across Europe.

Methods

The search methodology for this literature review was the same as that described for a previous literature search and analysis [7], but with additional filters for the topics of interest to this review (Fig. 1).

Fig. 1
figure 1

Summary of the study selection procedure. (Adapted from Fig. 1 of Torres et al. [7], used under the Creative Commons—Attribution-NonCommercial (CC BY-NC 3.0) license. The original can be found here: http://thorax.bmj.com/content/68/11/1057/F1.large.jpg). CAP community-acquired pneumonia. *One study did not include the terms ‘risk’ or ‘co-morbidity’/‘comorbidity’ in either the title or the abstract and, so, was not identified in the PubMed searches; however, ‘risk factors’ was included in the list of MeSH terms for the article

Full size image

The PubMed database was searched using the following search string: pneumonia AND English AND 2005/01/01–2012/07/31 AND risk NOT clinical trial, phase I OR clinical trial, phase II OR clinical trial, phase III OR controlled clinical trial OR randomized controlled trial OR case reports OR practice guideline OR editorial OR review OR cost OR cost effectiveness OR efficacy OR immunogenicity OR economic OR nosocomial. Additional searches used the same search string, but replaced ‘risk’ with either ‘comorbidity’ or ‘co-morbidity’.

Articles were included in the initial literature search [7] if they reported observational studies performed in Western European countries (Austria, Belgium, Denmark, Finland, France, Germany, Greece, Ireland, Italy, The Netherlands, Norway, Portugal, Spain, Sweden, Switzerland, UK) and presented data from individuals >15 years of age on either the incidence of CAP in at-risk individuals, defined as those with underlying risk factors placing them at increased risk of CAP (as defined in [7]), or risk factors for CAP. The papers identified were further screened for data on pathogens identified in patients with CAP and/or antibiotic treatment in patients with CAP. Studies that focused on nosocomial or healthcare-acquired pneumonia were excluded.

The included articles were reviewed in full and data on the study setting and methodology, characteristics of the populations studied, pathogens and antibiotic treatments were extracted. If more than one paper reported different aspects of the same study, all relevant papers were included. Where the same data were reported in more than one paper, the first paper to be published was selected for inclusion. The analysis of the included papers was descriptive and no meta-analyses of data were performed. Unless otherwise stated, all data are reported as odds ratios (ORs) (95 % confidence intervals [CIs]).

Results

Included studies

As reported previously [7], a total of 3,331 articles published between January 2005 and July 2012 were identified, of which 3,240 could be excluded on the basis of the title, abstract or study location. Further screening of the PubMed results and full papers identified 60 references meeting the inclusion and exclusion criteria. We summarise data from 33 of these studies that reported on the pathogens identified in patients with CAP and/or antibiotic treatment in patients with CAP. The included studies were performed in Denmark (n = 1), France (n = 3), Germany (n = 3), Greece (n = 1), Italy (n = 4), Spain (n = 20) and the UK (n = 1). Details of the study designs and populations are summarised in Table 1.

Table 1 Methodology and patient demographics of the studies included in the review. (Adapted from Supplementary Table 1 of Torres et al. [7], used under the Creative Commons—Attribution-NonCommercial (CC BY-NC 3.0) license. The original can be found here: http://thorax.bmj.com/content/suppl/2013/10/15/thoraxjnl-2013-204282.DC1/thoraxjnl-2013-204282supp_tables.pdf)
Full size table

The majority of studies included adults of all ages, but three studies considered only elderly patients (age ≥65 years) [24–26]. Additionally, most of the studies considered pneumonia of any aetiology, but six were performed in patients with pneumonia due to L. pneumophila (n = 3) [27–29], Haemophilus influenzae (n = 1) [30], Gram-negative bacteria (n = 1) [31], or Enterobacteriaceae or P. aeruginosa (n = 1) [32]. Six studies were conducted in specific populations: five studies in patients with human immunodeficiency virus (HIV) [33–37] and one study in patients with chronic obstructive pulmonary disease (COPD) [38].

Pathogens identified in patients with CAP

The aetiology of CAP was investigated in 25 observational studies in Denmark (n = 1), France (n = 2), Germany (n = 2), Italy (n = 2), Spain (n = 17) and the UK (n = 1). Tables 2, 3 and 4 summarise these data to show the most common microbiological techniques, the overall frequency of isolation of pathogens and the frequency of isolation of pathogens specifically in HIV and COPD study cohorts. Full details of the microbiological techniques used and the pathogens isolated in each study are included in Supplementary Table 1.

Table 2 Microbiological techniques/samples used for the isolation of pathogens in patients with CAP
Full size table
Table 3 Patients with CAP and episodes of CAP with a pathogen identified
Full size table
Table 4 Prevalence of pathogens identified in patients with CAP with HIV or COPD
Full size table

Microbiological methodologies used to establish the aetiology of CAP were reported in 67 % of the studies (n = 22). These methodologies were similar across studies and included the assessment of blood, sputum, urine and pleural fluid samples and, less commonly, tracheobronchial, bronchoalveolar, transthoracic and nasopharyngeal samples. Blood cultures were performed in all 22 studies (Table 2) and all but one study reported using at least two different techniques. Other frequently used techniques for the isolation of pathogens were sputum culture (91 % of studies), urine antigen tests (specifically for the detection of S. pneumoniae and L. pneumophila; 86 % of studies), serology, for the detection of antibodies against specific pathogens, including L. pneumophila, Chlamydophila pneumoniae, Coxiella burnetii, Mycoplasma pneumoniae, Chlamydophila psittaci, Chlamydia trachomatis and respiratory viruses (68 % of studies), and pleural fluid culture (45 % of studies).

The percentages of patients and episodes of CAP in which a pathogen was not identified were 26.7–87.3 % and 44.2–77.0 %, respectively. In patients in whom a pathogen was identified, S. pneumoniae was the most commonly isolated and was identified in 12.0–85.0 % of patients within 19 studies (Table 3). Of the atypical bacteria, M. pneumoniae (up to 61.3 % of patients within ten studies), L. pneumophila (up to 20.1 % of patients within 12 studies) and C. pneumoniae (up to 9.9 % of patients within nine studies) were frequently identified in patients with CAP, whereas C. burnetii was isolated less frequently (up to 3.4 % of patients within six studies). Other pathogens isolated included S. aureus (up to 20.0 % of patients within 12 studies), P. aeruginosa (up to 16.8 % of patients within ten studies), Klebsiella pneumoniae (up to 5.0 % of patients within five studies) and Acinetobacter baumannii (isolated in 2.0 % of patients in one study that was performed in patients hospitalised with bacteraemic CAP due to Gram-negative bacteria [31]). CAP of mixed aetiology was reported in four studies in 0.4–19.9 % of patients [22, 26, 39, 40].

For studies with data available stratified by age (<65 years and/or ≥65 years) [23–26, 40, 41], the frequencies of pathogens were generally similar between age groups. However, S. pneumoniae (<65 years: 20.9–28.0 %; ≥65 years: 19.9–85.0 %), H. influenzae (<65 years: 4.1–6.4 %; ≥65 years: 2.9–29.4 %) and respiratory viruses (<65 years: 4.6–7.7 %; ≥65 years: 7.8–18.6 %) appeared to be more frequently isolated in elderly patients aged ≥65 years, and M. pneumoniae appeared to be more frequently isolated in younger patients (<65 years: 14.0–25.1 %; ≥65 years: 0.7–6.8 %).

Among the studies reporting on the aetiology of CAP in patients with HIV [33–37], the frequencies of isolated pathogens were similar to those found for the overall data. S. pneumoniae was the most commonly isolated pathogen (57.8–81.8 % of patients), and H. influenzae and L. pneumophila were also often identified (Table 4).

The aetiology of CAP was similar in patients with and without COPD [38, 40, 42], in whom S. pneumoniae, H. influenzae, L. pneumophila, M. pneumoniae and respiratory viruses were all commonly identified (Table 4). In one study, P. aeruginosa was reported in a significantly higher percentage of patients with COPD than in those without COPD (7.4 % vs. 0.9 %; p < 0.01) and L. pneumophila was found to be significantly lower in patients with COPD than in those without COPD (2.1 % vs. 7.8 %; p < 0.05) [42].

Antibiotic treatment in patients with CAP

The antibiotic treatment of patients with CAP was reported in 23 studies: France (n = 1), Germany (n = 3), Greece (n = 1), Italy (n = 4) and Spain (n = 14). Rates of antibiotic treatment with beta-lactams, macrolides and quinolones are summarised in Table 5. Full details of the antibiotic therapies for each study are provided in Supplementary Table 2.

Table 5 Antibiotic treatment in adults with CAP
Full size table

The rates of antibiotic treatments in patients with CAP were available in 13 studies [22, 28, 29, 36–38, 40, 41, 43–47]. The rate of monotherapy ranged from 16.0 to 94.7 % of patients and the rate of combination antibiotic therapy ranged from 5.0 to 84.0 % of patients (Table 5). The rate of antibiotic monotherapy with beta-lactams was higher than that for macrolides and quinolones. In one study, younger patients (<65 years) received fewer beta-lactams and more quinolones than older patients (≥65 years) (beta-lactams: 62.5 % vs. 81.3 %; quinolones: 28.2 % vs. 17.1 %, respectively), whereas macrolide use was similar between age groups (32.6 % vs. 31.4 %, respectively) [41]. For combination therapy, the most common combinations were beta-lactams combined with macrolides or quinolones. We found that the rate of combination antibiotic therapy was higher in patients in an intensive care unit (ICU; 84.0 %) and other hospitalised patients (31.8–69.0 %) than in outpatients (5.0–29.9 %) (Table 6). Three studies reported data on antibiotic treatment in populations with comorbidities (COPD [38] and liver disease [46]) or lifestyle risk factors for CAP (alcoholism [45]). Antibiotic treatments did not differ according to the presence or absence of COPD, liver disease or alcoholism.

Table 6 Antibiotic treatment in adults with CAP stratified by ICU patients, hospitalised patients and outpatients
Full size table

In the 14 studies that reported on appropriate versus inappropriate antibiotic therapy [22, 23, 27–31, 43, 44, 46–50], the majority of patients had received adequate initial antibiotic treatment. Inappropriate antibiotic therapy was reported in 0–39.0 % of patients [22, 23, 27–32, 43, 44, 46, 48–50]. One study showed that patients with polymicrobial CAP were significantly more likely than those with monomicrobial CAP (p < 0.001) to receive inappropriate antibiotic treatment (39.0 % vs. 10.0 %, respectively) [22]. Furthermore, inappropriate antibiotic therapy was found to be an independent predictor of mortality (univariate analysis: OR 11.23 [95 % CI 4.44–28.38], p < 0.001; multivariate analysis: adjusted OR 10.79 [3.97–29.30], p < 0.001) in one study [22].

The antibiotic resistance of pathogens responsible for CAP was described in four studies conducted in 1995–2008 [46], 2001 [45], 2001–2004 [43], 2001–2009 [44] and 2002 [30]. Penicillin resistance of S. pneumoniae was reported in 14.9–25.7 % of patients with CAP and in 8.4–20.7 % of isolates. Erythromycin resistance of S. pneumoniae was observed in 12.0–21 % of patients with CAP and in 14.7–17.1 % of isolates. In the two studies reporting on the antibiotic resistance of H. influenzae, beta-lactamase production was reported in 9.7 % [43] and 80.0 % [30] of isolates.

Discussion

This review provides a comprehensive overview of the aetiology of CAP and its antibiotic treatment in patients in Western Europe and builds on knowledge from earlier reviews of the incidence and risk factors for CAP among adults in this region [7], the burden of CAP in Europe [1] and a meta-analysis on the incidence of CAP in Europe by Rozenbaum et al. [12]. In addition, it provides important information to be taken into consideration in future updates to the European guidelines for the management of CAP.

Microbiological methodologies for the isolation of pathogens

The majority of studies were in patients hospitalised for CAP and, as may be expected, blood cultures were used for the isolation of pathogens. This is in line with current guidelines from the European Respiratory Society (ERS) and the European Society for Clinical Microbiology and Infectious Diseases (ESCMID) for the management of lower respiratory tract infections, which recommend that two sets of blood cultures are performed in patients hospitalised for CAP [18]. However, a study conducted between 2007 and 2011 in 14 countries in Europe found that blood cultures were performed in only 50 % of patients hospitalised with CAP [51]. This is similar to findings from a further retrospective, observational study in Europe conducted between 2010 and 2011 and published after the cut-off date for our search, in which blood cultures were performed in 55 % of patients hospitalised with CAP, suggesting that implementation of the guidelines across Europe is still incomplete [52]. It is also possible that some European hospitals have adopted the approach advocated within the most recent Infectious Diseases Society of America (IDSA)/American Thoracic Society (ATS) recommendations, which limit blood cultures to patients hospitalised in the ICU [53].

It is important to note that there are limitations in the methodologies used to evaluate the aetiology of CAP; for example, it is difficult to obtain all types of samples in all patients and many patients have received antibiotic treatment prior to sampling. Further limitations include the difficulty in obtaining a reliable sputum sample in the early stages of CAP in non-COPD patients and the technical limitations of diagnostic tests, such as a lack of sensitivity, or the poor ability of patients to form antibodies. Such limitations can lead to inaccurate estimations of the prevalence of pathogens [54]. Furthermore, there are often a substantial number of patients in which the aetiology of CAP cannot be identified. Therefore, the rates of isolated pathogens reported in studies could be under- or overestimated due to false-negative or false-positive results, or the inability to isolate a pathogen [54].

Aetiology of CAP in Europe

We found that S. pneumoniae was the most common pathogen isolated in patients with CAP in Europe across the studies included in our review. However, there was substantial variation in the incidence of this pathogen (12.0–85.0 % of patients), which is comparable with findings from Welte et al., which identified S. pneumoniae in 11.9–68.3 % of patients with CAP, and from the meta-analysis by Rozenbaum et al. on the incidence of CAP in Europe, which identified S. pneumoniae in 19.3 % of CAP episodes [1, 12].

Other frequently identified pathogens found to cause CAP across the included studies in our review were H. influenzae, Gram-negative enteric bacilli, respiratory viruses and M. pneumoniae. These pathogens were identified at rates similar to those found by Welte et al. [1], with the exception of M. pneumoniae, which was higher in our review (61.3 % vs. 32.4 % of patients). The high level of M. pneumoniae in our review was influenced by one study in particular, which consistently identified high M. pneumoniae rates in the age cohorts studied, particularly the younger cohorts [41]. A possible reason for these high rates of M. pneumoniae is the cyclical nature of M. pneumoniae outbreaks, which occur every 3–7 years; hence, the prevalence of this organism in a given study varies with the inclusion of such yearly epidemics. Secondly, the study methodology for isolating M. pneumoniae used polymerase chain reaction (PCR) from bronchoalveolar lavage and throat swab samples, which were available for almost all patients, as well as sputum samples, which were available in only approximately 40 % of patients. Therefore, the true denominator for calculating M. pneumoniae rates differed from that for other pathogens, which resulted in an overestimation of the M. pneumoniae rate by two-fold [41].

We found that the multidrug-resistant pathogens accounted for ≤20.0 % of CAP and that, of these pathogens, S. aureus and P. aeruginosa were more frequently isolated than K. pneumoniae or A. baumannii, which were rarely identified as the cause of CAP. In a European study of pathogens in hospitalised patients with CAP, multidrug-resistant pathogens were the cause of CAP in 3.3–7.6 % of patients in which a pathogen could be identified, with methicillin-resistant S. aureus being the most common multidrug-resistant pathogen [55]. The study also found that patients with CAP caused by multidrug-resistant pathogens typically presented with more severe pneumonia on admission to hospital and, correspondingly, multidrug-resistant pathogens were more prevalent among those patients admitted to an ICU than among those admitted to a general ward [55]. Overall, multidrug-resistant pathogens do not appear to be a major cause of CAP in Europe, but the severity of CAP caused by multidrug-resistant pathogens highlights the importance of routine testing for these pathogens. Probabilistic scores, such as the Aliberti and Shorr scores, can be useful for predicting the presence of multidrug-resistant pathogens in hospitalised patients and could help physicians to prescribe appropriate treatments without overprescribing broad-spectrum antibiotics [55].

The frequency of pneumonia of mixed aetiology varied across the four studies in which it was identified, from 0.4 to 19.9 % of patients. One important factor likely to contribute to this variation is the diagnostic methods used to identify pathogens. It is probable that a higher percentage of polymicrobial infections will be identified using newer molecular techniques. These techniques are also likely to contribute to an increase in the percentage of infections in which a causative pathogen can be identified, reducing underdiagnosis and increasing the accuracy of diagnoses, which will potentially lead to improvements in the accuracy of treatment.

Differences in the groupings of pathogens between studies (e.g. Gram-negative enteric bacilli ± P. aeruginosa) may have led to underestimations in the prevalence of some pathogens, such as P. aeruginosa. As for S. pneumoniae, we generally found large ranges in the frequency of pathogens isolated across studies. Differences in the isolation rates of pathogens between studies could be due to many factors, including the severity of CAP, healthcare settings (e.g. patients treated in the community versus in the ICU), populations studied (e.g. age, comorbidities, risk factors) and diagnostic tests used (e.g. traditional methods versus new technology). For example, in a meta-analysis of the prevalence of S. pneumoniae in Europe, S. pneumoniae was more likely to be detected in studies that used PCR assays compared with studies that used other diagnostics tests (OR 2.49 [95 % CI 1.39–4.46]) [12]. To establish the aetiology of CAP in Europe more accurately, improvements are needed in the sensitivity and specificity of diagnostic tests used to isolate pathogens. Furthermore, a more standardised approach to the diagnostic tests used will make comparisons across different studies more valid.

When looking at the aetiology of CAP stratified by age, we found a trend for S. pneumoniae, H. influenzae and respiratory viruses to be more frequent in elderly patients aged ≥65 years, and M. pneumoniae to be more frequent in those aged <65 years. Similar age-related trends have been observed previously in a study of the microbial aetiology of CAP in adults in Finland [54], in which S. pneumoniae infections were more frequent in adults aged ≥60 years than in those aged <60 years (48 % vs. 35 %, p = 0.04) and infections with M. pneumoniae were more frequent in individuals aged 15–44 years compared with older adults (24 % vs. 3 %, p < 0.001). The study in Finland also found viruses to be the cause of CAP in a higher proportion of older adults than younger adults; however, this trend was not significant. No consistent age-related trend was observed for H. influenzae [54]. A study of the microbial patterns of CAP in patients aged ≥65 years found that S. pneumoniae was the most frequent pathogen in all age groups over 65 years and that age did not influence the microbial cause of CAP [6]. In patients with COPD, we found that the aetiology of CAP was similar to that in patients without this condition. This was also observed in patients with HIV (in those who were non-severely immunocompromised or receiving treatment), which is in line with the results from studies examining the impact of HIV on the clinical outcomes of CAP in the highly active antiretroviral therapy era. Non-severely immunocompromised patients with HIV have been shown to have similar clinical outcomes in terms of the time to clinical stability, length of hospital stay and mortality rate when compared with individuals without HIV [56, 57].

Antibiotic treatment in patients with CAP

Current ERS/ESCMID guidelines (2011 edition) for the treatment of CAP [18] recommend one of the following for the treatment of CAP in hospitalised patients:

  • Aminopenicillin ± macrolide

  • Aminopenicillin beta-lactamase inhibitor ± macrolide

  • Non-antipseudomonal cephalosporin III

  • Cefotaxime or ceftriaxone ± macrolide

  • Levofloxacin

  • Moxifloxacin

  • Penicillin G ± macrolide

The ERS/ESCMID guidelines suggest that combination therapy should be restricted to patients with severe presentation of CAP, with combination therapy being the treatment of choice for patients with severe CAP being treated in the ICU [18]. In these patients, non-antipseudomonal cephalosporin III plus a macrolide, or moxifloxacin or levofloxacin ± non-antipseudomonal cephalosporin III are recommended in those patients without risk factors for P. aeruginosa, whereas in patients with risk factors for P. aeruginosa, antipseudomonal cephalosporin, or acylureidopenicillin beta-lactamase inhibitor or carbapenem, plus ciprofloxacin or plus macrolide plus aminoglycoside is preferred. When a specific pathogen has been identified, antibiotic therapy can be targeted against that pathogen. For example, for CAP caused by Legionella species, respiratory quinolones are recommended [18]. In our study, we found that beta-lactams and macrolides were more frequently prescribed than quinolones. This was not unexpected, as some treatment guidelines reserve quinolones for when initial empirical therapy has failed, or specifically for the treatment of CAP caused by Legionella species (which was identified in up to 20.1 % of patients in this review) [58].

Many European countries have their own national guidelines for the treatment of CAP [54, 58–63], which are derived from the European guidelines and take into account the local epidemiology and aetiology of CAP, as well as the national resistance rate against antibiotics, such as penicillins and macrolides. This means that, although some similarities exist and the European guidelines provide a good framework for guidance, there are variations in the antibiotic management of CAP throughout Europe, depending on the specific requirements of each country.

A limited amount of data regarding antibiotic resistance were reported in the studies included in this review. By contrast, Welte et al. found several studies with data on antibiotic resistance [1], possibly due to the use of different literature search criteria (e.g. differences in the time periods reviewed, databases searched, search terms used and inclusion/exclusion criteria applied). Antibiotic resistance against S. pneumoniae is the main clinical concern, due to its dominance in the aetiology of CAP. We found that pneumococcal resistance against penicillin was slightly higher than might be expected (8.4–20.7 % of isolates [44, 46]) when compared with that reported in Europe by the European Antimicrobial Resistance Surveillance Network (EARS-Net) in 2011 (8.8 % of isolates non-susceptible and 2.3 % resistant) [64]. However, this may be because all of the studies in this review reporting penicillin resistance were carried out in Spain, which has one of the highest levels of penicillin resistance of S. pneumoniae in Europe [64]. Pneumococcal resistance against erythromycin was similar to that reported by the EARS-Net for macrolides (14.7–17.1 % of isolates [43–46] vs. 14.1 % of isolates) [64]. The literature review by Welte et al. highlighted a trend for increased antibiotic resistance of CAP-related pathogens in Europe, including S. pneumoniae, which showed an increase in resistance to commonly prescribed antibiotics [1]. Globalisation and developments in healthcare may contribute to the changing pattern of the aetiology and antibiotic resistance of CAP. Understanding these changes is essential to guide best practices in the antibiotic management of CAP and to safeguard against the failure of empiric antibiotic treatment. The implementation of global surveillance systems would provide a means for guidelines to be adapted more rapidly in response to such changes.

Strengths and limitations

This literature review was based on a review of published data from Europe that aimed to capture as many studies as possible from the past 7 years. The main strength of this review is that many of the included publications were case–control studies performed with large numbers of patients drawn from registries or primary care databases, rather than small, single-centre studies, thus giving reassurance that they provide a good representation of CAP in European populations. However, this review also has some limitations. There was a lack of a well-defined diagnostic protocol in many of the studies and the percentage of patients or episodes of CAP in which a specific causative pathogen was not identified was high in some studies (26.7–87.3 % and 44.2–77.0 %, respectively). The majority of the included studies were based on patient populations in Spain (20 of 33 studies) and this could potentially limit the validity of extrapolating the data from this review to other European populations.

Conclusion

In conclusion, this review has highlighted that Streptococcus pneumoniae is the most common pathogen responsible for community-acquired pneumonia (CAP) in adults in Europe and that beta-lactams are the most frequently prescribed class of antibiotics for the treatment of CAP. Understanding the aetiology of CAP and the changing pattern of antibiotic resistance in Europe, together with an increased awareness of the risk factors for CAP, will help clinicians to identify those patients most at risk of developing CAP and provide guidance on the most appropriate treatment.