1 Introduction

Quality improvement in medical practice is the cornerstone of high standards healthcare delivery. The benefits of external quality documentation programs have been documented for several surgical disciplines.¹ The National Surgical Quality Improvement Program (NSCIP) and the Veteran Affairs Surgical Quality Improvement Program are among the largest scale initiatives for monitoring and comparing the quality of care. Surgical quality registers have been only recently constructed in the majority of European countries. The German arthroplasty register (Endoprothesenregister Deutschland, EPRD)² has been launched since November 2012.

The aim of quality assurance is not to reveal the medical errors of the individual physician, but rather the analyzation of processes and outcomes aiming at development and improvement in healthcare. Quality assurance (QA)differs substantially from quality control. The term quality control is used for that “part of quality management focused on fulfilling quality requirements”. On the contrary, QA focuses on the “part of quality management focused on providing confidence that quality requirements will be fulfilled”.³ The role of quality assurance in orthopedic practice should not be limited to evaluation and recording of clinical and radiological examinations or information about the prostheses used, but should be extended by identifying defects in the provision of medical care, solving issues that are difficult to manage and encouraging change in the patient care process.

QA could be divided in two different categories: internal and external QA. Both types are based on a systematic documentation process. The purpose of documentation as part of internal QA is the institutional self-evaluation, recognition of problems and implementation of changes, whereas the documentation as part of external QA aims at comparing the outcomes across different hospitals or make a national or international assessment of healthcare delivery. Thus, the data collected from different institutions should be comparable using the same standards. Internal QA is essential as self-evaluation tool for an institution. After regular data evaluation, the healthcare providers could take appropriate measures to reduce the adverse event rates and improve the provided quality of care.

To our knowledge, no systematic research exists in the field of hip arthroplasty comparing the data documented for the external quality assurance with prospectively collected data on complications.

The primary aim of our study was to compare two different documentation ways: external documentation (quality assurance) vs. internal documentation after primary hip arthroplasty. Secondary aims were to evaluate the changes in the complication rates across the period under study and the potential association of patient's gender, duration of operation, duration of hospital stay and patient age with the complication rate.

2 Methods

Data of patients subjected to primary total hip arthroplasty in the orthopedic university clinic in Essen-Werden from 2004 until 2014 were prospectively collected using two different methods: internal documentation and external documentation (for quality assurance). Both documentations were performed by physicians. The study has been approved by the ethics committee of the Duisburg-Essen University.

The external documentation was performed through an electronic form according to §137 of Part V of the German Social Security Code (SGB V). Incomplete electronic submission of the documentation was not permitted. The quality assurance data from the German orthopedic departments should be submitted yearly until 15th of March. The external documentation was conducted usually within 4 weeks from the date of discharge either from the first surgeon or quality assurance manager. The records have been maintained and evaluated by the Federal Office for Quality Assurance (Bundesgeschäftsstelle Qualitätssicherung, BQS)⁴ from 2001 to 2008 and by the AQUA-Institute (Institutfürangewandte Qualitätsförderung und Forschung im Gesundheitswesen)⁵ from 2009 to 2014.Data from the hospitals in North Rhine-Westphalia are first collected by the office quality assurance NRW (Geschäftsstelle Qualitätssicherung Nordrhein-Westfalen, qs-nrw).⁶ The quality assurance was launched and implemented in our orthopedic department since 2001.

The internal documentation was conducted with pen and paper by physicians in our department on the day of discharge. The records have been kept in the orthopedic department and were thoroughly scrutinized in the case of suspected increase in complication rates. The internal records served for the control of in-hospital complications. The detailed documentation forms are available upon request.

Statistical analyses were performed using Pearson's chi-squared test or Fisher's exact test to compare internal with external data, and the incidence of complications observed during the years 2005–2009 and 2010-2014. The Mann-Whitney U test was used to compare the continuous outcomes and age of patients in the aforementioned periods. The complications of the year 2004 were excluded in order to create two uniform 5-year periods. The first year (2004) was excluded because the internal documentation was in the initial stage. A two-tailed value of p < 0.05 was considered significant. Furthermore, binary logistic regression analyses were performed on SPSS 22.0 (SPSS Inc., Chicago, Illinois, USA). Dependent variable was considered any surgical complication (occurrence/non-occurrence) and independent variables were considered the patient's gender, the normalized duration of operation, normalized length of hospital stay and normalized patient age. The transformation of continuous variables to normal was indicated according to Kolmogorov–Smirnov test, as data did not follow the normal distribution, and it was conducted through a two-step process in SPSS described by Templeton.⁷Ad hoc binary logistic regression analyses were conducted with ASA score as independent binary variable. Normal healthy patients (ASA-score 1) and patients with mild systemic disease (ASA-score 2) formed a group and patient with severe systemic disease (ASA-score 3 and 4) formed a second group. Statistical significance was set at a p-value of 0.05.

3 Results

The total number of primary hip arthroplasties was 3395. The median age of the patients who sustained a complication was 70 years (range: 20-90) and 32.7% of them were male. The number of patients experiencing at least one postoperative complicationwas197 and 169, internal and external, respectively (after exclusion of patients experiencing postoperatively ambiguously defined changes in the mobility). The median postoperative length of stay of patients experiencing a complication was 15 days (range: 1-72) and the median duration of operation was 96 min (range: 39-298).

The absolute number of complications observed in our internal and external quality assurance dataset are represented in a bar diagram (Fig. I). The most frequent complications were wound healing disorders (21%, internal data), reinterventions (15.6%, external data), hematoma or postoperative hemorrhage (14.9, external data) and cardiovascular complications (13%, internal data) [Insert Fig. I].

Close

Fig. 1 Number of patients that experienced internally or externally documented complications. In the bar diagram are represented the absolute number of complications that were observed in our internal and external quality assurance dataset.

Export to PPT

A statistically significant difference between the two datasets (internal vs. external) was detected in the “inability to walk at discharge” (p < 0.0001) and the occurrence of “postoperative wound infection” (p = 0.0038, OR 7.28, CI 1.6–32.3).

Furthermore, a significant reduction of the complication rate was observed during the period 2010-2014 according to the internal and external documented “total complication rate” (p < 0.0001). A significant difference between internal and external documentation was found for the parameters “hematoma and postoperative hemorrhage”, “cardiovascular complications” and “reintervention” (p < 0.01). The number of complications presented in Table 1 produced statistically significant differences between the two time periods under investigation. The Mann-Whitney U test identified a significant difference (p < 0.0001) in the postoperative length of hospital stay between the two study periods (median of 15 vs. 11 days, for the periods 2005-2009 and 2010-2014 respectively).

The conducted binary logistic regression analysis after removal of insignificant independent variables demonstrated that the variable “normalized length of hospital stay” more commonly predicted independently the occurrence of complications. The variable “normalized duration of operation” was the second most common variable associated with the occurrence of complications. Table 2 shows the results of the binary logistic regression analyses.

Finally, the ad hoc binary logistic regression analyses, which was conducted with ASA score as independent binary variable (ASA-score 1-2 or ASA-score 3-4) resulted in non-statistically significant results.

4 Discussion

The main finding of this study is a substantial difference between internal and external documentation in two key parameters of quality assurance, namely the “inability to walk at discharge” (p < 0.0001) and the occurrence of “postoperative wound infection”. Furthermore, a significant reduction of the complication rate was observed during the period 2010-2014according to internal and external data. Hematoma and postoperative hemorrhage, cardiovascular complications and reintervention were significantly different in the two periods of time. The length of hospital stay is been statistically significantly reduced during the second time period (2010-2014). Finally, the length of hospital stay was an independent factor predicting the risk of complications.

The observed reduction of surgical complication rate between 2010 and 2014 may be associated with the cumulative experience, the high specialization and the stability of personnel (senior surgeons) in our department. The reduction of the incidence of general complications may be influenced by the implementation of improved clinical pathways.

The “inability to walk at discharge” demonstrated a statistically significant difference between internal and external data probably because of a difference in the definition of this parameter. According to the external data, more patients have been considered to be unable to walk at discharge. The mobilization with the assistance of a physiotherapist and the use of advanced mobility aids is probably interpreted as an inability to walk, because of the dependence of the patient on the aforementioned aids, whereas internal assessment may provide a more modest judgment.

A study⁸ conducted in Germany (2010) on cholecystectomy patients compared the data documented for the external quality assurance with the prospectively recorded data on complications. The authors found a sensitivity of 46.6% and a specificity of 98.3% of the internal documentation method. The missing complications were both minor and major complications. A study from the same researchers was published in 2014⁹ after implementation of system reminding them to complete data on complications. They have found an improved sensitivity of 70% and a specificity of 98%. The over-documentation was attributed to erroneous documentation or documentation of events, which did not meet the criteria of a complication. The under-documentation was due to missing documentation of major and minor complications. These very factors may have played a role in the discordance of documentation in our cohort, however the lack of documentation of medical chart data in our study does not allow exploration of the sensitivity and specificity of the documentation methods.

A recent publication¹⁰ compared for quality assurance purposes the postoperative complications recorded during the hospitalization with the complications recorded during the follow-up period in 4 different medical fields (knee replacement, appendectomy, percutaneous coronary intervention and prostate operation). The post-discharge follow-up data derived from administrative AOK data (the largest German health insurance company). The follow-up interval for total knee arthroplasty patients was 1 year. The authors have found that an important part of the complications occurred after hospital discharge (7.7–92.6%) and a weak relationship existed between the aforementioned variables, as indicated by the correlation coefficient. The aforementioned study included only patients after total knee arthroplasty. Nevertheless, the results may be partially transferable to our primary hip arthroplasty patients. Thus, the documented complications may represent only partially the number of complications in case of longer follow-up.

A research team¹¹ investigated the differences between two time periods (1993-1997 and 2005-2009) in the quality assurance datasets of patients with femoral neck fractures in Germany. The authors have found a statistical significant reduction of length of hospital stay between the above mentioned study periods (27.73 days versus 18.49 days, p < 0.001), which could have a possible positive influence on the complication rate. Characteristic for the study was that the leading complication (24) was the postoperative hemorrhage in both study periods, due to the fact that the patients have suffered a hip fracture. In our study, the operations were conducted electively and none of them was performed after a current femoral fracture. Hematoma and postoperative hemorrhage were also common complications in our study (Fig. I), although the exact number of patients with postoperative hemorrhage could not be extracted. Furthermore, the first study period lies far back in the past making a comparison of the results with the present study impossible.

Several publications examined the role of patient characteristics on morbidity after total hip or total joint replacement. A systematic review¹² found that the revision rate after total joint arthroplasty was higher in younger and male patients. The mortality rate was higher in older males. Especially in patients undergoing a hip replacement surgery worse function was recorded in older females. In the analysis of the New Zealand Joint Registry¹³ was found a higher rate of revision surgery in patients undergoing total hip replacement with a higher ASA score. In a retrospective study¹⁴ obesity (BMI >50) and younger age are identified apart from diabetes as risk factors for deep infections in knee and hip joint arthroplasty patients. Higher risk of infection among total joint replacement patients with diabetes (preoperative glucose level of ≥124 mg/dL) and obesity was found in anotherstudy.¹⁵ Finally, a publication¹⁶ found that patients with uncontrolled diabetes mellitus had a higher complication rate after hip or knee arthroplasty. Due to the fact that comorbidities were not documented in our study, no assumption could be made regarding their potential influence on the complication rate. The ASA-score has not proved to be statistically significantly related to the complication rate in the present study.

Several authors investigated the role of patient characteristics on the mortality after total joint replacement. In the analysis of the New Zealand Joint Registry¹³ was founda higher mortality rate in patients undergoing major joint replacement with a higher ASA score. Two researchers¹⁷ showed additionally a higher rate of complications among the deceased patients. Furthermore, a study¹⁸ supported the opinion that the rarity of the death in the field of elective surgery makes morbidity a more suitable postoperative measure. Likewise, in the present study the mortality rates were rare (2 out of 3395 operated patients). This fact makes further statistical analysis in multicenter studies more appropriate.

A systematic review,¹⁹ published in 2011, found no consistent evidence that the behavior of the patient would be influenced or the provided healthcare results would be improved after publicly releasing the performance data. Lack of evidence regarding the potential influence on the healthcare personal or organizations has also been reported.

Recent research work²⁰ described that the median time of venous thromboembolism was eleven days after hip arthroplasty and the three days after knee arthroplasty. As a result half of the venous thromboembolic events would not have been captured in USA if the observation period would not be extended until 30 days postoperatively. This fact would significantly change the hospital ranking, which was based on the VTE rate, potentially influencing erroneously the patients' behavior and physicians’ referrals. The fact that the length of hospital stay in our department has been reduced in the period 2010-2014 (median: 11 days) may have caused a potential reduction to the documented in-hospital VTE rate after primary total hip replacement.

The influence of preoperative risk factors such as gender, increased age, and patients’ weight on the patient outcomes after primary hip replacement has been investigated in a maximum care university hospital.²¹ The gender did not influence statistically significantly the outcome. The increased age of the patient caused a rise of the complication rate, an increase of the operation duration and length of hospital stay. The outcomes were better in normal weight patients (BMI between 25 and 34.9 kg/m^∧2). The influence of patient age and ASA-Score on the length of hospital stay was shown in two recent studies.^22,23 The weight and BMI was not documented in our study. The age was a risk factor for prosthesis dislocation (internal and external data), pneumonia (external data) and cardiovascular complications (external data). The gender constituted a risk factor for the occurrence of wound healing disorders and delirium (internal data).

The arthroplasty registries play a great role in a special part of QA, the control of the implant survivorship and the complications related to the specific implant, since detailed data on implanted prostheses are recorded and assessed for a long time. Two publications from the Nordic Arthroplasty Register Association (NARA) study group in the context of improving the statistical analysis and presentation of registry results elucidated the problems arising in a major joint arthroplasty register.^24,25

The present study has been conducted in one of the associated departments of the university clinic for orthopaedics in Essen. It could be assumed that the results could be generalized to German orthopedic departments performing 300-600 total joint arthroplasties per year (departments with similar case mix, surgical experience and surgical specialization). The comorbidities were not analyzed. As a result, a possible negative influence on the observed complication rate could not be explored. The parameters or the definition of complications, which were documented in the internal and external dataset, demonstrate some differences, making their direct comparison difficult. Furthermore, the normalization of continuous parameters (duration of operation, length of hospital stay and patient age) through a two-step process⁷ made a back-transformation of data not possible. The conduction of a medical chart review would provide the opportunity to test the sensitivity and specificity of internal and external data, using a widely accepted gold standard documentation method. Finally, the lack of blinding of outcome assessment, possible incompleteness of outcome data and selective reporting may introduce detection, attrition and reporting bias, respectively, which have an unknown effect on effect estimates.

5 Conclusions

Further improvements should be made in the documentation processes of total hip arthroplasties. Possible factors associated with the reduction of the incidence of complications through time may be accumulated experience, the improvement of clinical pathways and the impact of continuous monitoring of complications through internal and external quality assurance.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.