Quality Assurance Measures and Mortality After Stroke

A Retrospective Cohort Study

Background: Quality assurance for acute in-hospital care in Germany is based on compulsory comparisons between institutions, so-called external quality assurance (EQA). The effectiveness of EQA has not yet been adequately studied. The purpose of the QUASCH project, which is supported by the Innovation Fund of the Federal Joint Committee, is to investigate the association between EQA and health care outcomes, specifically with respect to stroke.

Methods: The analyses were based on data from 379 825 patients insured by the AOK health insurance fund who were acutely admitted to a hospital because of stroke over the period 2007–2017. Data on 47 659 patients were derived from EQA documentation in the state of Hesse, in which stroke EQA had already been introduced in 2003; data on the remaining 332 166 patients were from other federal states, where 117 734 of these patients had been treated under EQA conditions. The association of EQA with mortality over the period of observation was analyzed by multivariate Cox regression, with the following covariates: age, sex, comorbidities, time period of occurrence, nursing care level, type of stroke, socio-economic deprivation in the region of origin, and treatment in a stroke unit.

Results: Compared to treatment without EQA, mortality risk under EQA in the state of Hesse was significantly lower (hazard ratio [HR]: 0.93; 95% confidence interval: [0.92; 0.95]). The reduction in mortality risk with EQA was somewhat lower in the other federal states (HR: 0.96 [0.95; 0.97]). Treatment in a stroke unit was associated with a mortality risk that was lower still (HR: 0.86 [0.85; 0.87]). Mortality risk rose with age, comorbidities, and need for nursing care; it was lower in women and in persons whose stroke occurred in a later period.

Conclusion: Quality assurance measures are associated with lower mortality risk after stroke. The concentration of care in specially qualified institutions is associated with stronger effects than EQA alone.

The external quality assurance (EQA) system, which covers all health care institutions, plays a prominent role among the measures used in Germany to promote quality in health care. According to Section 135a of the German Social Code Part V (SGB V), all health care providers are obliged to “participate in inter-institutional quality assurance measures aimed in particular at improving the quality of outcomes”. As part of EQA, care providers will document data on health care processes and outcomes that will be collated, analyzed, and reported back in the form of care provider comparisons on a regional or national basis. The comparative data should prompt improvement processes that will ultimately benefit patients. In the acute inpatient sector alone, quality assurance documentation was produced for approximately three million cases in 2019, or 16% of all hospital cases (1), requiring from the care providers an average of 15 minutes working time per case (2). Furthermore, the work input by the evaluating bodies and the service providers to deal with the data must also be considered, with the result that, on the whole, EQA involves a great amount of time and effort.

It is therefore all the more surprising that evidence of any benefit from EQA for patients in Germany has so far not been systematically reported (3, 4). Instead, only improvement trends have been observed, which can hardly be distinguished from health care improvements resulting from other measures (5). In terms of the benefits from external quality assurance, international evidence can be cited for two of the methods used in the context of EQA. One of these is quality reporting, for which a 2018 update of a Cochrane review concluded that only a few studies found, at most, a small impact on selection decisions made by patients, admitting or referring physicians, and funders, or on health outcomes (6). Secondly, a Cochrane review acknowledges moderate effects of audit and feedback methods on conformity with a desired clinical practice; however, the evidence for effects on health care outcomes is not clear (7).

With this in mind, the project “QUASCH – Outcomes of Quality Assured Stroke Care – Hesse compared with the rest of Germany”, funded by the Innovation Committee of the Federal Joint Committee, has investigated the association between EQA with stroke mortality. Stroke care is an appropriate subject for study for several reasons:

• Strokes are a very relevant cause of death or disability, with approximately 304 000 cases in Germany in 2019, and are becoming increasingly important due to demographic aging (8, 9).
• Although a concept for this was created as early as 2015 (10), a nation-wide EQA for stroke has so far not been developed in Germany. Instead, only the state of Hesse (2003), and later Baden-Wurttemberg, Hamburg, Rhineland-Palatinate and Bavaria have so far introduced mandatory EQA for stroke – combined with further measures such as structured dialogue.
• Meanwhile, the care of stroke patients has changed significantly in all German states over the years. This includes the introduction of stroke units, thrombolysis and thrombectomy, the development of guidelines and quality indicators, and the establishment of regional stroke networks and registries, some of which have indeed been shown to improve quality (11, 12, 13, 14, 15, 16, 17).

Stroke care, therefore, provides a unique opportunity for investigating the association of external quality assurance with relevant healthcare outcomes at a time when various other quality-enhancing measures were being introduced.

Methods

The present study is designed as a retrospective cohort study based on secondary data. A comparison was made of adult stroke patients (discharge diagnoses ICD-10 I60, I61, I63, I64, G45) admitted as acute inpatients in an emergency setting and insured by one of the AOK health insurance companies, who had suffered a primary event during the period between 2007 and 2017 and had been treated either in Hesse (the state with the longest EQA experience) or in the rest of Germany. Data sources included the documentation of stroke patients anonymously registered during the observation period at the Quality Assurance Office Hesse (GQH), which is responsible for EQA in Hesse, and anonymized data sets provided by hospitals on the basis of Section 301 SGB V to the AOKs and from there to the Scientific Institute of the AOK (WIdO). The WIdO links the data with the patients’ other benefit and master data, so that information on any long-term care requirements (nursing care level) and on death were also available. The WIdO and GQH datasets were linked using the variables sex, month and year of birth, the hospital’s institution code, and date of admission. The GQH data (n = 47 659 patients) also include 962 patients from other federal states who had been treated in hospitals in Hesse.

After excluding any duplicates, recurrent cases, health insurance provider switchers, and cases with negative follow-up time, the analysis dataset comprised a full survey of all AOK stroke patients from Hesse (n = 56 749, of whom n = 46 697 were registered with GQH) and a random sample of AOK stroke patients from the other German states (n = 332 166), which was structurally identical with regard to age, sex, and stroke diagnosis distribution. The difference of 10 052 patients between AOK-Hesse and GQH results from incomplete case reports at GQH, especially at the beginning of the observation period, as well as cases that could not be combined using the linkage variables. All patients had had an initial event during the observation period and were followed up to the end of 2017 at the latest. The number of cases was based on the condition that a maximum of 500 000 AOK patient records could be made available for the study by the WIdO.

The primary outcome measure of the study was the overall mortality risk of the stroke patients during the observation period in relation to treatment under EQA conditions in Hesse compared with treatment without EQA or under EQA conditions in the rest of Germany.

Furthermore, mortality within 10, 30, and 90 days and 1-, 2-, 3-, 5-year mortality after cerebral infarction (ICD-10-I63) during the observation period were also examined in relation to EQA. To assess the outcome measures, the survival time of all stroke patients (in days) after the initial event during the observation period, the average mortality risk in the subgroups, and the respective proportion of deceased patients within the different time periods were calculated. Differences in mortality risk were analyzed using Cox regression models and Kaplan-Meier curves, including log rank tests. The methodological procedure and the included covariates are explained in more detail in the eMethods section.

Results

The analysis dataset included 379 825 patients with a first stroke between 2007 and 2017. Under EQA conditions in Hesse, i.e., GQH-registered, there were 47 659 patients, including 962 patients with residence in bordering federal states. A total of 332 166 non-GQH-registered patients were from other federal states. Because other states also had EQA in place during the observation period, a total of n = 174 483 cases, of which n = 117 734 were outside of Hesse, were treated under EQA conditions at the time of their initial event. Table 1 presents the characteristics of the entire study population in comparison with the GQH and non-GQH samples. The distribution of characteristics of GQH patients versus non-GQH patients were matched with respect to almost all parameters, such as average age (75.6 years) or proportion of female patients (53.7%). GQH patients had a slightly lower comorbidity rate according to the Elixhauser index (65.3% with index ≤3); however, they were found to have a 2.4% higher number of cases with a nursing care level in place before their stroke. During the observation period, the proportion of GQH patients treated in a stroke unit was 61.2%, about 10% higher than non-GQH patients, and the proportion of deceased patients among GQH patients was 49.1%, about 2% lower. On average, the observation period was 3.3 years after the initial event for all groups.

Table 1

Characteristics of the study population

The overall mortality risk of GQH stroke patients treated under EQA conditions in hospitals in Hesse was statistically significantly (p <10^–5) reduced compared with treatment without EQA in other federal states (hazard ratio [HR]: 0.93; 95% confidence interval: [0.92; 0.95]) (Table 2). Similarly, treatment under EQA conditions in other states was associated with a lower mortality risk compared with states without EQA (HR: 0.96 [0.95; 0.97]). A greater reduction of the mortality risk was shown for treatment in stroke units; after adjustment for all other independent variables and the presence of EQA, the HR here was 0.86 [0.85; 0.87].

Table 2

Association of external quality assurance and covariates – including stroke unit treatment – with adjusted overall mortality risk of stroke patients

The mortality risk was also reduced in women as compared with men (HR: 0.82 [0.81; 0.83]) and in patients with a later initial event; in comparison with the period from 2007 to 2010, the hazard ratios were 0.95 [0.94; 0.96] in 2011–2013 and 0.89 [0.88; 0.91] in 2014–2017.

With the exception of the regional German Index of Socioeconomic Deprivation (GISD), the other factors included in the model were each associated with an increased risk of death compared with the reference; age, prestroke nursing care level, comorbidity index, and stroke diagnosis (transient ischemic attacks versus embolic and hemorrhagic infarcts) increased the mortality risk statistically significantly (p<10^–5) (Table 2).

Further stratified time analysis of the association of adjusted mortality risk with EQA in Hesse (GQH), EQA in other German states (EQA-FS), and the covariates, differentiated for treatment with/without stroke unit, initially showed that the mortality risk for patients with stroke unit treatment—considering all patients treated with or without EQA conditions—decreased constantly (2007–2010: HR: 0.82 [0.81; 0.83]; 2011–2013: HR: 0.80 [0.79; 0.82]; 2014–2017: HR: 0.72 [0.71; 0.74]) (Table 3, eTable 1). Looking only at those patients who were treated under GQH and EQA-FS conditions but not in a stroke unit, a significantly lower mortality risk was found over all time periods compared with patients without EQA (Table 3, eTable 2). On the other hand, the relative mortality risk for patients treated in stroke units under GQH or EQA-FS conditions increased over time and was also statistically different from treatment without EQA only in the first time period (Table 3, eTable 3).

eTable 2

Complete Cox regression model assessing overall mortality risk of patients under external quality assurance conditions in Hesse (Quality Assurance Office Hesse) or in other German states, stratified according to time periods

Table 3

Overall mortality risk of the patients under Quality Assurance Office Hesse or under external quality assurance conditions in other federal states without/with stroke unit treatment over time

eTable 1

Complete Cox regression model assessing overall mortality risk of patients under external quality assurance conditions in Hesse (Quality Assurance Office Hesse) or in other German states, stratified according to time periods

eTable 3

Complete Cox regression model assessing overall mortality risk of patients under external quality assurance conditions in Hesse (Quality Assurance Office Hesse) or in other German states, stratified according to time periods

The survival time analyses in the form of Kaplan-Meier curves in Figure 1 (treatment without stroke unit [SU]) and Figure 2 (treatment in a SU) also highlight the advantages of EQA. Patients treated in a region without EQA and not in an SU survived a median of 3.99 years. In contrast, the survival time of patients in other federal states with EQA was statistically significant longer (4.57 years, p <0.0001). Patients survived the longest under GQH conditions for 4.59 years (Figure 1). Patients treated in a region without EQA but in an SU survived a median of 5.28 years. On the other hand, SU patients survived statistically significantly longer under GQH conditions (5.65 years, p <0.0001) and SU patients in other federal states with EQA survived the longest at 5.98 years (Figure 2).

Figure 1

Treatment without stroke units in the period 2007–2017

Figure 2

Treatment in stroke units in the period 2007–2017

There were also a number of advantages associated with GQH care over other federal states with regard to short and long-term mortality after cerebral infarction (ICD-10: I63). Whereas, for example, the risk-adjusted proportion of those who died within ten days after the initial event during the period between 2007 and 2010 was 6.5% [6.1; 7.0] under GQH conditions, this proportion was 7.8% [7.6; 8.0] in other federal states (p <0.001). The beneficial connection continued for up to two years after the stroke: GQH deceased persons 33.4% [32.5; 34.2], other federal states 35.2% [34.9; 35.6] (p <0.001). There was no mortality advantage for GQH patients when treatment in stroke units was considered in isolation. Furthermore, there was no advantage evident any more during the last observation period between 2014 and 2017 (eTables 4, 5).

eTable 4

Proportion of deceased stroke patients after 10, 30 and 90 days under external quality assurance conditions in Hesse (GQH) as compared with other federal states, taking into account stroke unit care

eTable 5

Proportional of deceased stroke patients after 1, 2, 3 and 5 years under external quality assurance conditions in Hesse (GQH) as compared with other federal states, taking into account stroke unit care

Discussion

Taking stroke care as an example, the present study shows that quality assurance measures may be associated with a reduced short and long-term mortality risk. EQA in Hesse was associated with an average reduction in overall mortality risk by about 7% in comparison with treatment without EQA during the period between 2007 and 2017. The mortality risk in other federal states with EQA was reduced by 4% as compared with states without EQA. This was the first long-awaited, and ultimately successful, evaluation of the costly, mandatory external quality assurance (EQA) in Germany based on long-term, relevant health care outcomes (2, 3, 4). Furthermore, the study enriches, at least indirectly, the international findings on audit and feedback methods and on quality reporting, for which so far little association with health care outcomes had been demonstrated (6, 7).

The only downside with regard to the potential effects of EQA was the results for stroke unit treatment. This revealed that stroke-unit patients demonstrated twice the reduction in mortality risk (14%) over those not treated in a stroke unit. Part of the advantage of EQA in Hesse as compared with other federal states with EQA may therefore be explained by the fact that, over the years, about 10% more stroke patients in Hesse were treated in stroke units.

The study confirms the results regarding the short-term survival benefits of stroke units (11, 16) even for longer periods. Patients treated for stroke in a stroke unit in other federal states with EQA had the greatest survival benefit (median survival six years) in comparison with patients not treated in a stroke unit in states without EQA (median survival four years).

The subanalyses reveal that the positive associations of EQA were primarily noted from 2007 to 2010, and partially still from 2011 to 2013, and from 2014 to 2017 applied only for patients not treated in stroke units. An additional effect of EQA, evident when looking at the whole observation period for the mortality risk analyses as well as the short and long-term mortality analyses, was no longer apparent in isolation from 2014 to 2017 for stroke unit patients. This outcome confirms the findings from the Cochrane review on structured outcome audit and feedback (18), according to which the basic principle of EQA achieves an effect – and here in particular on health care processes (12) – when there is still great potential for optimization at the outset, although the overall effectiveness is low.

Limitations

Based on an observational study, the results cannot be interpreted causally. Instead, only associations between different quality promotion measures – EQA in Hesse and other states, stroke unit treatment – and mortality risk in stroke patients are presented. The associations can only be interpreted to mean that the measures had a high probability of impacting care outcomes. The study is confined to mortality as an aspect of outcome quality that can be illustrated using routine data, which to some extent has attracted criticism (19). Other relevant outcomes, such as functionality after stroke as assessed using the International Classification of Functioning, Disability, and Health, are not reported as no routine data are available on this. This limitation also applies to risk adjustment based on routine data, which means that potentially relevant determinants, such as stroke severity on admission, were not taken into account. The study also only covers patients insured with the AOK health insurance fund. However, with a share of insured persons of 32% of the German population, the risk of selection bias can be assumed to be low. In addition, matching of the characteristics of the different study cohorts indicates that the risk-adjusted differences in mortality risk are convincing.

Conclusion

Despite the limitations mentioned above, it can be concluded that measures of structured quality promotion for stroke patients—and in particular, stroke unit treatment—may be positively associated with relevant care outcomes. But based on the study results, it should be questioned whether it makes sense to have quality assurance with the legally specified goal of improving the outcome quality that relies preferentially on EQA—i.e., relies on care providers to initiate quality improvement measures as part of their internal quality management when they perform comparatively poorly. Instead, better outcomes of stroke units seem to be generated by their active components, including material and personnel-related structural requirements, process standardization, internal quality management, and internal and external collaborations.

In the end, patient care should always be provided in appropriate institutions. In the case of stroke care, this would mean care provided preferably in stroke units. Given the results of the present study, it would seem appropriate for the German Federal Joint Committee to also review existing quality assurance measures in other service areas with regard to patient-relevant effects and to assess care in specially qualified institutions as a quality promotion measure for other services as well.

Acknowledgments

This publication is based on a project funded by the Innovation Fund of the Federal Joint Committee under the grant number 01VSF18041.

Conflict of interest statement
The authors declare that no conflict of interest exists.

Manuscript received on 9 July 2021, revised version accepted:
20 July 2021.

Translated from the original German by Dr. Grahame Larkin, MD

Corresponding author
Prof. Max Geraedts, MD, M. San.

Institute for Health Services Research and Clinical Epidemiology

Department of Medicine

Philipps University of Marburg

Karl-von-Frisch-Straße 4, 35043 Marburg

Cite this as
Geraedts M, Ebbeler D, Timmesfeld N, Kaps M, Berger K, Misselwitz B, Günster C, Dröge P, Schneider M: Quality assurance measures and mortality after stroke—a retrospective cohort study. Dtsch Arztebl Int 2021; 118: 857–63. DOI: 10.3238/arztebl.m2021.0339

►Supplementary material

eReferences, eMethods, eTables:
www.aerzteblatt-international.de/m2021.0339

References

1. IQTIG – Institut für Qualitätssicherung und Transparenz im Gesundheitswesen. Qualitätsreport 2020. https://iqtig.org/qs-berichte/qualitaetsreport/ (last accessed on 28 June 2021).

2. Petzold T, Steinwitz A, Schmitt J, Eberlein-Gonska M: Evaluation der externen Qualitätssicherung gemäß § 137 SGB V am Universitätsklinikum Carl Gustav Carus Dresden. Z Evid Fortbild Qual Gesundhwes 2013; 107: 541–7

3. Glattacker M, Jäckel WH: Evaluation der Qualitätssicherung – aktuelle Datenlage und Konsequenzen für die Forschung. Gesundheitswesen 2007; 69: 277–83

4. Khan C, Ollenschläger G: Wirksamkeit von Qualitätsprogrammen in der stationären Versorgung in Deutschland – eine Literaturanalyse. Z Evid Fortbild Qual Gesundhwes 2014; 108: 576–86

5. Rückle F, Stausberg J: Entwicklung der Qualität der stationären Versorgung zwischen 2008 und 2011: eine Studie aus Daten der externen Qualitätssicherung. Dtsch Med Wochenschr 2015; 140: e36–42

6. Jamtvedt G, Flottorp S, Ivers N: Audit and feedback as a quality strategy. In: Busse R, Klazinga N, Panteli D, Quentin W (eds.): Improving healthcare quality in Europe: characteristics, effectiveness and implementation of different strategies. WHO Regional Office for Europe, Copenhagen (Denmark) 2019; 265–86.

7. Metcalfe D, Rios Diaz AJ, Olufajo OA, et al.: Impact of public release of performance data on the behaviour of healthcare consumers and providers. Cochrane Database Syst Rev 2018; 9: CD004538

8. Diagnosedaten der Krankenhäuser Eckdaten der vollstationären Patienten und Patientinnen. www.gbe-bund.de/gbe/!pkg_olap_tables.prc_set_page?p_uid=gast&p_aid=14368157&p_sprache=D&p_help=2&p_indnr=550&p_ansnr=57248869&p_version=3&D.001=1000001&D.946=12920&D.011=44302 (last accessed on 3 May 2021).

9. Busch M, Kuhnert R: 12-Monats-Prävalenz von Schlaganfall oder chronischen Beschwerden infolge eines Schlaganfalls in Deutschland. Journal of Health Monitoring 2017; 2: 70–5.

10. Meyer S, Willms G, Broge B, Szecsenyi J: Themenerschließung für die sektorenübergreifende Qualitätssicherung der Versorgung bei Schlaganfall und TIA. Z Evid Fortbild Qual Gesundhwes 2016; 115–116: 1–9

11. Nimptsch U, Mansky T: Trends in acute inpatient stroke care in Germany--an observational study using administrative hospital data from 2005–2010. Dtsch Arztebl Int 2012; 109: 885–92

12. Wiedmann S, Heuschmann PU, Hillmann S, et al.: The quality of acute stroke care- an analysis of evidence-based indicators in 260 000 patients. Dtsch Arztebl Int 2014; 111: 759–65

13. Mader FM, Schwenke R: Schlaganfall. S3-Leitlinie. www.awmf.org/leitlinien/detail/ll/053-011.html (last accessed on 4 October 2021).

14. Otten K, Kugler C, Geraedts M: Qualitätsindikatoren in der Schlaganfallversorgung – eine Übersicht aus der internationalen Literatur. Aktuelle Neurologie 2004; 31: 411–9

15. Haas K, Rücker V, Hermanek P, et al.: Association between adherence to quality indicators and 7-day in-hospital mortality after acute ischemic stroke. Stroke 2020: 51: 3664–72

16. Pross C, Berger E, Siegel M, Geissler A, Busse R: Stroke units, certification, and outcomes in German hospitals: a longitudinal study of patient-based 30-day mortality for 2006–2014. BMC Health Serv Res 2018; 18: 880

17. Hillmann S, Wiedmann S, Rücker V, et al.: Stroke unit care in Germany: the German stroke registers study group (ADSR). BMC Neurol 2017; 17: 49

18. Ivers N, Jamtvedt G, Flottorp S, et al.: Audit and feedback: effects on professional practice and healthcare outcomes. Cochrane Database Syst Rev 2012: CD000259

19. Nacimiento W, Töpper R, Erbguth F, Höfling W, Voltz R, Brassel F: Schlaganfallmedizin: Mortalitätsrate allein kein Kriterium für eine gute Versorgung. Dtsch Arztebl 2016; 113: A 595–8.

e1. Elixhauser A, Steiner C, Harris DR, Coffey RM: Comorbidity measures for use with administrative data. Med Care 1998; 36: 8–27

e2. Kroll LE, Schumann M, Hoebel J, Lampert T: Regional health differences—developing a socioeconomic deprivation index for Germany. Journal of Health Monitoring 2017; 2: 98–113.