OUP user menu

Does public reporting improve the quality of hospital care for acute myocardial infarction? Results from a regional outcome evaluation program in Italy

Cristina Renzi, Federica Asta, Danilo Fusco, Nera Agabiti, Marina Davoli, Carlo Alberto Perucci
DOI: http://dx.doi.org/10.1093/intqhc/mzu041 223-230 First published online: 15 April 2014

Abstract

Objective To evaluate whether public reporting of performance data was associated with a change over time in quality indicators for acute myocardial infarction (AMI) in Italian hospitals.

Design Pre–post evaluation of AMI indicators in the Lazio region, before and after disclosure of the Regional Outcome Evaluation Program, and a comparative evaluation versus other Italian regions not participating in the program.

Setting/data sources Nationwide Hospital Information System and vital status records.

Participants 24 800 patients treated for AMI in Lazio and 39 350 in the other regions.

Intervention Public reporting of the Regional Outcome Evaluation Program in the Lazio region.

Main Outcome Measure Risk-adjusted indicators for AMI.

Results The proportion of ST-segment elevation myocardial infarction (STEMI) patients treated with percutaneous coronary interventions (PCI) within 48 h in Lazio changed from 31.3 to 48.7%, before and after public reporting, respectively (relative increase 56%; P < 0.001). In the other regions, the proportion increased from 51.5 to 58.4% (relative increase 13%; P < 0.001). Overall 30-day mortality and 30-day mortality for patients treated with PCI did not improve during the study period. The 30-day mortality for STEMI patients not treated with PCI in Lazio was significantly higher in 2009 (29.0%) versus 2006/07 (24.0%) (P = .002).

Conclusions Public reporting may have contributed to increasing the proportion of STEMI patients treated with timely PCI. The mortality outcomes should be interpreted with caution. Changes in AMI diagnostic and coding systems should also be considered. Risk-adjusted quality indicators represent a fundamental instrument for monitoring and potentially enhancing quality of care.

  • acute myocardial infarction
  • percutaneous coronary intervention
  • public reporting
  • quality indicators

Introduction

Coronary heart disease (CHD) represents the most common cause of death in Europe (www.heartstats.org). Annual hospital admissions for ST-segment elevation myocardial infarction (STEMI) in Europe range between 44 and 142 cases per 100 000 inhabitants [1]. Since the 1990s, percutaneous coronary interventions (PCIs) increased. In Italy, ∼119 000 PCIs were performed in 2009 [2]. Despite remarkable advances in the treatment of acute myocardial infarction (AMI), optimal care is not always made available to all eligible patients [36].

Public reporting of hospital performance has become increasingly common and may influence hospital performance through two related pathways [7, 8]. In the first pathway, patients or general practitioners use performance data to choose better performing providers (which may motivate providers to improve performance). Evidence supporting this pathway is scant [9]. In the second pathway, providers respond to performance data because of professional pride, competitiveness and sensitivity to their reputation among peers and implement internal improvement projects [9, 10].

Overall, the evidence of the effect of public reporting on quality of care is limited [7] and only very few report on European countries [11]. In Italy, various studies have examined quality of care for AMI [6, 1215], but no study has addressed public reporting.

The Regional Outcome Evaluation Program (called P.Re.Val.E.) was launched in the Lazio region in 2005, and includes an evaluation of hospital care for AMI [16]. Lazio is located in central Italy and has ∼5 700 000 residents, mainly concentrated in the city of Rome (National Institute of Statistics, 2010). As previously described [16], the P.Re.Val.E. program has grown since 2005 to include 54 indicators of hospital care in various clinical areas, including cardiology, orthopedics, obstetrics, gastroenterology, respiratory and cerebrovascular diseases.

In January/February 2008, the P.Re.Val.E. hospital-specific results were communicated to individual hospitals in Lazio through professional meetings and were publicly released on the web (www.epidemiologia.lazio.it/vislazio/vis_index.php) and through patient/citizen associations and press conferences. Local and national newspapers and Italian medical journals covered the topic. Other Italian regions had no active communication of the performance evaluation program.

The objective of this study was to evaluate whether the reporting of the P.Re.Val.E. results in Lazio was associated with a change in care for AMI patients. In particular, we compared the change over time in AMI indicators in Lazio and in Italian regions, where no reporting of the performance evaluation had taken place. We hypothesized that public reporting in Lazio was associated with improved performance compared with other Italian regions.

Methods

We compared hospital care for AMI in Lazio during the years 2006/07 and 2009 (i.e. before and after the public release of the P.Re.Val.E. cardiac care performance data and excluding 2008, the year that public reporting began). Using available data regarding 30-day mortality, STEMI and non-STEMI, from Lazio and other regions, we compared changes during the study period in Lazio to those in Italian regions where no public reporting had taken place. We excluded regions such as Tuscany, which had a performance evaluation program with private and public reporting. Other regions were excluded due to issues of data availability. Regions included in the control group account for 13.2% of hospital admissions within the National Health System.

The P.Re.Val.E. program focused mainly on directly involving health professionals and managers. By providing them with timely, peer-hospital comparison, it emphasized the importance of sharing the interpretation of the results and offered the possibility of additional in-depth analyses, if requested by providers. In 2008, P.Re.Val.E. results were disseminated to hospitals in Lazio via in-person meetings with the managers of each hospital and its clinical staff. Hospital-specific performance and peer-hospital comparisons were presented, with the aim to promote a positive and constructive approach to quality improvement, rather than to identify the best or worst performances. Hospital staff was invited to suggest possible interpretations of unexpected findings. Participation of clinical staff was voluntary; ∼90% of cardiologists and cardiothoracic surgeons took part. The general public in Lazio was informed in general terms that a performance evaluation was underway without highlighting the results of specific indicators.

Study population and data source

We used the hospital information system (HIS), which includes all hospital admissions within the Italian National Health Service, and linked the database with vital status records through individual identification codes. We included all patients diagnosed with AMI between January 2006 and December 2007 and between January 2009 and November 2009 who fulfilled the following inclusion criteria: International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM) code 410.1–410.9; age 18–100 years; and discharged from any hospital within the National Health System. Main exclusion criteria: hospitalizations lasting fewer than 48 h and discharge at home (hypothesizing a misdiagnosis), previous hospital admission for AMI within the last 4 weeks and non-resident in the region. A detailed description of codes used to define the study cohorts is provided elsewhere [16].

In line with previous studies [17], we classified AMI according to the ICD-9-CM fourth-digit subcategories in STEMI (codes 410.1–410.6 and 410.8) and non-STEMI (code 410.7). Similar to previous studies [18] we did not include the code 410.9 in the STEMI category and examined this subgroup separately. According to the ICD-9-CM, code 410.9 includes ‘AMI of an unspecified site not otherwise specified’. Qualitative information provided by cardiologists has shown that code 410.9 is often used for cases that cannot be attributed to the two previous categories due to bundle branch block or pacemaker (data not shown). Other studies have included code 410.9 in the STEMI category [17].

Quality indicators

The P.Re.Val.E. project included 54 indicators, with 14 indicators regarding cardiac care [16]. In the present study, we focused on PCI. The following crude and risk-adjusted indicators were examined separately among patients with STEMI, non-STEMI, and AMI code 410.9: proportion of patients treated with PCI within 48 h; 30-day mortality; 30-day mortality after PCI performed within 48 h; and 30-day mortality among patients who have not been revascularized (no PCI).

In accordance with previous studies [3], we used the following procedure codes to define PCI: ICD-9-CM codes 00.66, 36.01, 36.02, 36.05, 36.06 and 36.07. The 48-h pre-PCI time interval was measured from the patient's first hospital access. The HIS only reports the date and not the exact time when a procedure was performed; in this study, PCI was considered to have been performed within 48 h if the procedure was carried out the same day as hospital access or within the next day. The HIS does not allow examining whether PCI was performed within a shorter time interval.

For the subgroup of STEMI patients in Lazio, we could also analyze PCI performed within 90 min and the proportion of patients treated with fibrinolysis. These data could only be obtained for the years 2008 and 2009, and were unavailable for previous years and for the other regions. Information about clinical characteristics, risk factors and comorbidities was obtained from the HIS on the basis of ICD-9-CM codes registered in the index hospitalization and in hospital admissions during the preceding 2 years. A validated automatic record linkage procedure was employed [19].

Statistical analysis

For each indicator, patient risk in the two geographic areas (Lazio and other regions) in the two study periods was obtained using the direct standardization method with the general population as a reference. A severity measure (a priori risk) was constructed for the study population to account for patient characteristics, such as age, sex, severity of disease and/or chronic comorbidities. A logistic regression model with no intercept, including centered covariates and an interaction term between geographic area and study period, was applied for each outcome. These models estimate group-specific (geographic areas) log odds of outcome. Adjusted risks were obtained for each level of interest by back-transforming parameter estimates with the following formulas:$$mathtex$$\hbox{Adj}\,\hbox{risk} = \left[ {\displaystyle{{\hbox{exp(estimate)}} \over {(1 + \hbox{exp(estimate)})}}} \right]*k,$$mathtex$$ where k is a correction coefficient introduced to take into account the non-linear nature of the logistic model. K is calculated as follows:$$mathtex$$K = \displaystyle{{\hbox{actual}\,\hbox{number}\,\hbox{of}\,\hbox{events}} \over {\sum\nolimits_{\,j = 1}^m {{\,p_j}} *{n_j}}},$$mathtex$$ where pj are the adjusted risks, nj is the group size, and m is the number of groups.

The outcome-specific predictive model included age and gender as a priori risk factors and comorbidities selected from all coexisting medical conditions potentially associated with the outcome by a bootstrap stepwise procedure. Stepwise analysis was performed with 500 replicated samples from the original data, and significance thresholds of 0.10 and 0.05 were imposed for input and removal, respectively. Only risk factors selected in at least 50% of the runs were included in the final models. Relative risks, absolute differences, and the corresponding P-values were estimated to compare the two study periods by geographic area.

In addition, we performed a validation study on a random sample of 2033 hospital admissions for CHD in Lazio during 2006/07 and we evaluated the agreement between diagnoses reported in the administrative database and on clinical records, considering the latter as the gold standard (details available on request). Trained personnel, who had no access to the administrative data, reviewed the clinical records.

Results

Overall, the study included 24 800 AMI patients in Lazio and 39 350 in the other regions. Among Lazio patients, 56% were STEMI, 35% were non-STEMI and 8% had AMI code 410.9 in 2006/07; in 2009, the surveyed population was 47% STEMI, 50% non-STEMI and 4% had AMI code 410.9. In the other regions, 45% were STEMI, 46% were non-STEMI and 10% were AMI code 410.9 in 2006/07; in 2009, 41% were STEMI, 50% non-STEMI and 9% AMI code 410.9. Patient characteristics were similar in 2006/07 and 2009 (Table 1).

View this table:
Table 1

Characteristics of the study population in the Lazio Region and other Italian regions in 2006/07 and 2009

Main characteristicsLazioOther regions
2006–200720092006–20072009
STEMINon-STEMI (code 410.7)AMI code 410.9STEMINon-STEMI (code 410.7)AMI code 410.9STEMINon-STEMI (code 410.7)AMI code 410.9STEMINon-STEMI (code 410.7)AMI code 410.9
%%%%%%%%%%%%
Total N9680610114453521374930412 52912 6962641475256871045
Age: mean (SD), years68 (13.5)72 (12.3)75 (11.6)67 (14.0)71 (12.7)78 (11.1)70 (13.5)74 (12.3)77 (11.8)70 (13.6)74 (12.2)77 (12.6)
Gender: women32.638.141.230.635.945.735.839.347.234.938.346.0
Previous myocardial infarction10.619.015.88.116.011.57.717.013.66.714.511.3
Hypertension10.115.119.07.714.018.19.616.717.58.316.716.9
Other coronary heart disease7.614.015.24.812.015.87.316.918.95.416.216.0
Chronic kidney diseases7.513.215.07.014.415.87.812.917.38.115.017.7
Diabetes6.811.614.05.010.114.55.39.711.14.710.410.9
Chronic cerebrovascular diseases6.29.513.45.17.711.86.710.113.96.58.813.5
Previous PCI5.07.75.96.19.77.910.712.96.412.315.56.3
Heart failure3.87.011.02.66.27.23.89.113.43.68.711.1
Vascular diseases4.67.09.23.36.08.95.18.69.54.07.77.1
Blood disorders4.26.610.84.57.612.25.18.110.65.19.112.3
Cardiac dysrhytmias4.26.210.32.85.99.53.77.19.13.26.98.5
Chronic obstructive pulmonary disease3.05.37.52.64.65.93.05.36.73.15.56.7
Malignant neoplasms4.14.87.53.74.67.64.65.68.14.75.87.8
Previous CABG1.74.73.31.45.74.01.86.43.61.76.42.2
Other cardiac disorders1.73.74.61.43.84.61.33.34.11.23.93.9
Hyperlipidemia2.73.63.61.82.73.32.34.52.51.64.62.3
  • SD, standard deviation; AMI, acute myocardial infarction; PCI, percutaneous coronary intervention; CABG, coronary artery bypass graft.

The risk-adjusted proportion of STEMI treated with PCI within 48 h in Lazio increased from 31.3% before public reporting to 48.7% after (relative increase 56%, P < 0.001) (Table 2). In the other regions, the proportion increased from 51.5 to 58.4% (relative increase 13%, P < 0.001). Among non-STEMI, PCIs within 48 h increased from 10.1 to 16.0% in Lazio (relative increase 57%, P < 0.001), and from 16.6 to 19.4% in the other regions (relative increase 16%, P < 0.001). In Lazio, 28.5% of STEMI patients received PCI within 90 min in 2009 and 25.1% in 2008. In 2009, 7.4% of STEMI in Lazio were treated with fibrinolysis and 5.3% with a combination of fibrinolysis and PCI. In 2008, 6.8% received fibrinolysis and 11.1% received a combination of fibrinolysis and PCI.

View this table:
Table 2

Change in the proportion of patients treated with PCI within 48 h among patients with STEMI, Non-STEMI and AMI code 410.9 in the Lazio region and in other Italian regions in the years 2006/2007 and 2009

 2006–20072009Change 2009 vs 2006–07
Crude proportionRisk-adjusted proportionTotal NCrude proportionRisk-adjusted proportionTotal NRRP-valueAbsolute difference
STEMI
 Lazio35.131.3968051.748.735211.56<0.00117.4
 Other regions50.751.512 52956.558.447521.13<0.0016.9
Non-STEMI (code 410.7)
 Lazio10.710.1610116.815.937491.57<0.0015.8
 Other regions16.116.612 69618.619.456871.16<0.0012.7
AMI code 410.9
 Lazio11.914.214459.512.13040.850.401−2.1
 Other regions15.119.9264121.127.310451.38<0.0017.5

The risk-adjusted 30-day mortality in 2009 after STEMI was 14.9% in Lazio and 14.7% in the other regions (Table 3); for non-STEMI in 2009, it was 8.1% in Lazio and 6.4% in the other regions. There was no significant change in overall 30-day mortality. Thirty-day mortality after PCI within 48 h also did not change significantly during the study period. In 2009, the mortality after STEMI treated with PCI was 5.4% in Lazio and 6.6% in the other regions; the 2009 mortality for non-STEMI treated with PCI was 2.3% in Lazio and 1.8% in the other regions.

View this table:
Table 3

Change in 30-day mortality after Acute Myocardial Infarction (AMI) in the Lazio region and in other Italian regions in the years 2006/2007 and 2009

 2006–20072009Change 2009 vs 2006–07
Crude rateRisk-adjusted rateTotal NCrude rateRisk-adjusted rateTotal NRRP-valueAbsolute difference
30-day mortality after AMI
 STEMI
  Lazio11.814.6968011.514.935211.020.7520.3
  Other regions13.414.712 52913.614.747521.010.9160.1
 Non-STEMI (code 410.7)
  Lazio9.28.961018.38.137490.920.239−0.7
  Other regions7.76.212 6968.06.456871.030.6460.2
 AMI code 410.9
  Lazio29.526.3144536.532.03041.220.0885.7
  Other regions28.922.3264132.024.810451.120.1382.6
30-day mortality after AMI treated with PCI within 48 h
 STEMI
  Lazio5.15.833984.85.418200.930.584−0.4
  Other regions6.26.063556.96.626851.100.3550.6
 Non-STEMI (code 410.7)
  Lazio2.02.06532.22.36311.150.7190.3
  Other regions2.01.720442.11.810551.060.8300.1
 AMI code 410.9
  Lazio18.616.617217.217.9291.080.8841.3
 Other regions11.010.740011.411.12201.040.8820.4
30-day mortality after AMI not treated with PCI
 STEMI
  Lazio20.724.0442625.629.011821.210.0025.0
  Other regions27.327.0451229.327.814971.030.5850.8
 Non-STEMI (code 410.7)
  Lazio13.114.1402312.914.021490.990.888−0.1
  Other regions12.011.1716412.711.631161.050.4690.5
 AMI code 410.9
  Lazio35.736.1107642.942.12471.160.1245.9
  Other regions35.231.3198741.035.67391.140.0534.3

Risk-adjusted 30-day mortality for STEMI not treated with PCI in Lazio was significantly higher in 2009 (29.0%) than in 2006/07 (24.0%); in the other regions, the mortality for STEMI not treated with PCI was 27.8% in 2009, with no significant change during the study period. Among non-STEMI patients not treated with PCI, 30-day mortality was 14.0% in Lazio and 11.6% in the other regions, with no significant change over time. The small subgroup with code 410.9 was characterized by a higher mortality compared with STEMI and non-STEMI. There was a decrease in the number of code 410.9 in Lazio accompanied by an increase in timely PCI for these patients in the other regions.

The validation study on a random sample of CHD admissions in Lazio, comparing diagnoses in the administrative database with clinical records, has shown that the respective positive predictive values of a diagnosis of STEMI or non-STEMI based on administrative data were 80.3 (935/1165) and 85.3% (238/279), compared with the gold standard of clinical records. Sensitivity was 95.1% (935/983) for STEMI and 46.6% (238/511) for non-STEMI. Approximately 16% of patients coded as STEMI in the administrative database had a diagnosis of non-STEMI on clinical records. Among code 410.9 patients, 62.9% had a diagnosis of non-STEMI and 27.4% of STEMI on clinical records.

Discussion

Our results show an increase in the proportion of STEMI treated with PCI within 48 h during the study period in Lazio compared with smaller increases in other regions offering support for our hypothesis that reporting of the P.Re.Val.E. results in Lazio may have contributed to improvements in cardiac care. Lazio started from a lower baseline proportion of timely PCI in 2006/07 compared to other regions (31.3 and 51.5%, respectively); the higher baseline percentage in the control regions might partially explain their relatively small increase in timely PCI. Although the gap between Lazio and the other regions has been reduced, further improvement is warranted.

A multinational study [20] revealed an increase in STEMI treated with PCI from 15 to 44% between 1999 and 2006. We previously observed an increase of AMI treated with PCI throughout Italy, with Lazio demonstrating greater improvement after 2007 [21]. However, our previous observation was based on only one indicator for cardiac care, and the available data did not allow to distinguish between STEMI and non-STEMI. The present study includes a more detailed analysis showing improvements for STEMI in particular.

International guidelines for STEMI recommend a 90-min door-to-balloon time or that PCI be performed as soon as possible, as timely PCI decreases the risk of death after STEMI; any reperfusion delay is associated with a continuous non-linear increase in mortality [5, 22, 23]. There is less evidence for non-STEMI, but invasive reperfusion within 12–24 h reduces mortality also for high-risk non-STEMI patients; shorter reperfusion time showed no additional benefit [23].

Even though PCI within 90 min in Lazio remained low (28.5%), we have shown some improvement. Unfortunately, this information was unavailable for the control group. For STEMI patients, timely primary PCI (defined as angioplasty and/or stenting without prior or concomitant fibrinolysis [24]) is preferred over thrombolytic therapy [25]. However, the 90-min door-to-balloon time is difficult to achieve, even in well-organized systems [26]. In a multinational study [20], PCI was performed outside the recommended time window in >40% of reperfused patients. Rapid pharmacological reperfusion followed by PCI within 24 h can represent a valid alternative [26].

Our data did not allow to distinguish between primary PCI, PCI combined with pharmacological reperfusion and ‘rescue’ PCI (i.e. after failed pharmacological reperfusion) [24]. According to the data for Lazio, fibrinolysis was performed in a minority of STEMI patients and it decreased in 2009 versus 2008, paralleled by an increase in timely PCI. Other authors also reported a decrease in fibrinolytic therapy from 41 to 16% between 1999 and 2006, in favor of PCI [20].

Additional aspects suggest that public reporting may be responsible for our results. Increases in timely PCI preceded public reporting, but improvements accelerated in Lazio immediately after reporting of the P.Re.Val.E. results [21]. Also, improvements have been observed in Lazio following public reporting in other areas of care addressed by the P.Re.Val.E. program, such as orthopedics [21, 27].

Some factors other than public reporting are plausible explanations for the increases we observed, but these are present in both Lazio and the comparison regions. A reorganization of hospital services for AMI took place during the last decade in various Italian regions. Throughout Italy, health districts have expanded the number of centers offering interventional procedures and promoted greater integration between hub and spoke centers and emergency medical systems within local networks providing acute care for STEMI [28]. These initiatives are not homogeneously distributed in all regions and, in some cases, are still in an experimental phase [29]. No comprehensive data are available on these initiatives, but the reorganization of AMI services appears to be occurring in both the Lazio region and the control group, making it unclear whether this might be an alternative explanation for our results. The introduction of a systematic approach of performance measurement ‘per se’ might have influenced hospitals to introduce quality improvement initiatives already before and independently of the communication of performance results [30, 31]. Independent quality improvement initiatives might have taken place in specific hospitals of some regions, but these would have the effect of reducing the differences we observed. The release of treatment guidelines and scientific publications might have led to a heightened awareness among clinicians, but this would have occurred throughout Italy.

We did not find an improvement in overall 30-day mortality after AMI or in 30-day mortality for patients treated with PCI. In Lazio, we observed an increase in 30-day mortality among STEMI patients not treated with PCI. The mortality outcomes should be interpreted with caution, taking the possible changes in the diagnostic and coding systems of AMI during the study period into account. In particular, in Italy, ICD-9-CM code 410.7 has been increasingly used in recent years for non-STEMI [17], while in the past some of these less severe cases were included in the STEMI category, which might have led to dilution of mortality in the STEMI category in the past.

The introduction, during the past decade, of more sensitive and specific biomarkers of myocardial necrosis (troponin test) and new diagnostic criteria have probably contributed to the increase of non-STEMI shown by international and Italian studies [17, 32] and by our data.

The lack of improvement in mortality, despite improved PCI performance, may also be explained in part by residual confounding. Data were lacking on some important AMI severity measures at hospital admission (e.g. ejection fraction) and on smoking history, which can influence AMI severity and case fatality [33]. Moreover, mortality is affected by numerous factors, in addition to timely PCI. Previous studies have shown that the decline in coronary disease mortality in the USA has been influenced in equal proportions by reductions in major risk factors and by evidence-based therapies [33]. Quality of care explains only a small proportion of observed differences in mortality between hospitals [34].

Some authors worry that public reporting may reduce performance of PCI in higher-risk patients [3537]. A decline in patients with cardiogenic shock treated with PCI has been shown following public reporting, with a simultaneous increase in in-hospital mortality among non-revascularized patients [35]. Enhancing physicians' confidence in the risk-adjustment methodology might limit risk avoidance behavior.

In general, studies about the effect of public reporting on quality of care have been inconclusive [7, 38]. Some studies have demonstrated improved outcomes in cardiac care [39, 40]. A Canadian study [30] reported a reduction in 30-day mortality among STEMI patients after public release of hospital-specific quality indicators (from 11.5 to 8.3%); however, no significant change was observed for non-STEMI and only 1 in 12 AMI process-of-care indicators improved significantly.

Limitations of our study include the need of more detailed preprocedural information regarding AMI severity to better control for extremely high-risk cases. The low prevalence of some comorbidities suggests underreporting of detailed clinical information. However, underreporting did not differ among the examined years and regions. A previous study demonstrated that administrative data provided similar league tables in cardiac surgery as a more complex database [14].

In conclusion, our results suggest that public reporting may have contributed to increasing the proportion of STEMI patients treated with timely PCI; however, 30-day mortality after AMI and for patients treated with PCI did not improve. Thirty-day mortality increased among STEMI patients not treated with PCI in Lazio. To establish whether cardiac care has improved overall, other quality indicators should be considered, in addition to those examined here. For example, non-mortality outcomes, such as ischemic outcomes, can be indicative of technical expertise [35]; moreover, health-related quality of life and patient satisfaction can shed light on aspects not evaluated by clinical indicators. Process measures, including door-to-balloon time and discharge on appropriate medical therapy can provide a different insight in quality of care.

Funding

This work was supported by the National Outcome Program, Ministry of Health, Italy.

Acknowledgements

We thank Paola Colais, Mirko Di Martino, Mariangela D'Ovidio, Luigi Pinnarelli, Chiara Sorge and Martina Ventura for their support. We are also grateful to Daniela D'Ippoliti, Massimo Stafoggia and Roberto Ricci for their critical revision of the manuscript.

References

View Abstract