Advertisement

Patient-Level Hospital Costs and Length of Stay After Conventional Versus Minimally Invasive Total Hip Replacement: A Propensity-Matched Analysis

Open ArchivePublished:September 14, 2012DOI:https://doi.org/10.1016/j.jval.2012.06.008

      Abstract

      Objectives

      A current trend in total hip replacement (THR) is the use of minimally invasive surgery. Little is known, however, about the impact of minimally invasive THR on resource use and length of stay. This study analyzed the effect of minimally invasive surgery on hospital costs and length of stay in German hospitals compared with conventional treatment in THR.

      Methods

      We used patient-level administrative hospital data from three German hospitals participating in the national cost data study. We conducted a propensity score matching to account for baseline differences between minimally invasively and conventionally treated patients. Subsequently, we estimated the treatment effect on costs and length of stay by conducting group comparisons, via paired t tests and Wilcoxon signed-rank tests, and regression analyses.

      Results

      The three hospitals provided data from 2886 THR patients. The propensity score matching led to 812 matched pairs. Length of stay was significantly higher for conventionally treated patients (11.49 days vs. 10.90 days; P < 0.05), but total costs did not differ significantly (€6018 vs. €5986; P = 0.67). We found a difference in the allocation of costs, with significantly higher implant costs for minimally invasively treated patients (€1514 vs. €1375; P < 0.001) in contrast to significantly higher staff and overhead costs for conventionally treated patients.

      Conclusions

      Minimally invasive surgery was compared with conventional THR and was found to be associated with a reduced length of stay. Total hospital costs, however, did not differ between the two treatment groups, because of higher implant costs for minimally invasively treated patients.

      Keywords

      Introduction

      Total hip replacement (THR) has been described as “the operation of the century” [
      • Learmonth I.D.
      • Young C.
      • Rorabeck C.
      The operation of the century: total hip replacement.
      ]. It is a high-volume surgical procedure [
      • Kurtz S.M.
      • Ong K.L.
      • Schmier J.
      • et al.
      Primary and revision arthroplasty surgery caseloads in the United States from 1990 to 2004.
      ] that is considered to be successful for the treatment of diseases such as coxarthrosis [
      • Ethgen O.
      • Bruyère O.
      • Richy F.
      • et al.
      Health-related quality of life in total hip and total knee arthroplasty: a qualitative and systematic review of the literature.
      ]. With approximately 200,000 procedures each year, hip replacement is one of the most frequent kinds of surgery in German hospitals [
      Federal Statistical Office
      Die 50 häufigsten Operationen der vollstationären Patientinnen und Patienten in Krankenhäusern.
      ]. Because of demographic change and the increasing use of this procedure in older age groups, the demand for hip replacement is expected to increase further [
      • Culliford D.J.
      • Maskell J.
      • Beard D.J.
      • et al.
      Temporal trends in hip and knee replacement in the United Kingdom 1991 to 2006.
      ,
      • Kurtz S.
      • Ong K.
      • Lau E.
      • et al.
      Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030.
      ,
      • Kurtz S.M.
      • Lau E.
      • Ong K.
      • et al.
      Future young patient demand for primary and revision joint replacement: national projections from 2010 to 2030.
      ,
      • Pulido L.
      • Parvizi J.
      • Macgibeny M.
      • et al.
      In hospital complications after total joint arthroplasty.
      ]. In recent years, there is a trend of using minimally invasive (MI) surgery approaches in THR [
      • Learmonth I.D.
      • Young C.
      • Rorabeck C.
      The operation of the century: total hip replacement.
      ]. MI surgery approaches were developed on the basis of conventional approaches and are supposed to reduce pain, postoperative blood loss, rehabilitation time, and length of hospital stay [
      • Sculco T.P.
      Minimally invasive total hip arthroplasty: in the affirmative.
      ]. Similar to conventional procedures in THR, a variety of different MI approaches exist [
      • de Verteuil R.
      • Imamura M.
      • Zhu S.
      • et al.
      A systematic review of the clinical effectiveness and cost-effectiveness and economic modelling of minimal incision total hip replacement approaches in the management of arthritic disease of the hip.
      ] but a coherent definition of MI THR is lacking [
      • Pospischill M.
      • Kranzl A.
      • Attwenger B.
      • Knahr K.
      Minimally invasive compared with traditional transgluteal approach for total hip arthroplasty: a comparative gait analysis.
      ]. In some articles, an MI THR is defined by the length of incision (<10 cm) [
      • Dorr L.D.
      • Maheshwari A.V.
      • Long W.T.
      • et al.
      Early pain relief and function after posterior minimally invasive and conventional total hip arthroplasty: a prospective, randomized, blinded study.
      ,
      • Goosen J.
      • Kollen B.
      • Castelein R.
      • et al.
      Minimally invasive versus classic procedures in total hip arthroplasty: a double-blind randomized controlled trial.
      ,
      • Ogonda L.
      • Wilson R.
      • Archbold P.
      • et al.
      A minimal-incision technique in total hip arthroplasty does not improve early postoperative outcomes: a prospective, randomized, controlled trial.
      ,
      • Pour A.E.
      • Parvizi J.
      • Sharkey P.F.
      • et al.
      Minimally invasive hip arthroplasty: what role does patient preconditioning play?.
      ], whereas in other articles, it is defined by the minimization of tissue and muscle dissection [
      • Sendtner E.
      • Boluki D.
      • Grifka J.
      Current state of doing minimal invasive total hip replacement in Germany, the use of new implants and navigation – results of a nation-wide survey.
      ]. Although a number of studies have been published that compare MI THR with conventional THR, evidence about its relative merits is still limited [
      • de Verteuil R.
      • Imamura M.
      • Zhu S.
      • et al.
      A systematic review of the clinical effectiveness and cost-effectiveness and economic modelling of minimal incision total hip replacement approaches in the management of arthritic disease of the hip.
      ,
      • Mahmood A.
      • Zafar M.S.
      • Majid I.
      • et al.
      Minimally invasive hip arthroplasty: a quantitative review of the literature.
      ,
      • Cheng T.
      • Feng J.
      • Liu T.
      • Zhang X.
      Minimally invasive total hip arthroplasty: a systematic review.
      ,
      • Smith T.
      • Blake V.
      • Hing C.
      Minimally invasive versus conventional exposure for total hip arthroplasty: a systematic review and meta-analysis of clinical and radiological outcomes.
      ]. In particular, the direct costs of an MI procedure in THR for hospitals have hardly been studied [
      • Duwelius P.J.
      • Moller H.S.
      • Burkhart R.L.
      • et al.
      The economic impact of minimally invasive total hip arthroplasty.
      ] despite the fact that, given the high number of THR procedures, even slight changes in direct costs can be expected to be important for the respective hospitals.
      In this study, we assessed the effect of MI THR surgery on direct hospital costs and length of stay (LOS) in German hospitals compared with the effect of convention surgery for THR.

      Material and Methods

      We used patient-level administrative hospital data from German hospitals participating in the national cost data study conducted by the Institute for the Hospital Remuneration System. The data include sociodemographic, medical, and treatment information, as well as cost data. Hospitals participating in the national cost data study use a standardized cost accounting approach, reporting direct hospital costs in 99 cost categories. Treatment information includes type of treatment, provided by the German procedure codes (Operationen- und Prozedurenschlüssel [OPS]), and date of treatment. Medical information is given by International Statistical Classification of Diseases, 10th Revision, German Modification (ICD-10-GM) (Version 2008), including principal and secondary diagnoses recorded during hospital admission, along with conditions acquired or developed during the hospital stay. A distinction between diagnosis at admission and conditions acquired at the hospital is only partially possible as the administrative data in Germany do not formally make this differentiation.
      From approximately 250 hospitals, 31 that participated in the national cost data study provided patient-level data from the year 2008. We identified patients with a recorded primary THR by using the first five digits of the OPS codes 5-820.0x and 5-820.2x. As no specific OPS code exists for MI THR, we classified patients as MI treated if the OPS code 5-986 “minimally invasive technique” was recorded. Patients who had also undergone a revision hip arthroplasty, identified by OPS code 5-821.xx, were excluded from the analysis. Hospitals with either no conventional THR or no MI THR were excluded, as we aimed to compare MI and conventional cases from the same hospitals.
      We contacted the remaining hospitals and asked 1) whether they coded MI THR consistently via the OPS code 5-986 and 2) which treatment patterns were a prerequisite for coding a patient as MI THR. After consultation with the relevant professional medical association, we included only those hospitals in the sample that consistently coded MI THR via the OPS code 5-986 given a surgical approach that minimizes tissue and muscle dissection. The hospitals kept in the sample were recontacted and were asked to provide data from 2009.
      For our outcome variable “total cost”, we summed up all reported costs. To allow further analysis, we grouped the 99 cost categories into the following 7 categories: staff costs—medical service; staff costs—nursing service; staff costs—medical-technical service; pharmaceutical costs; implant costs; costs for further medical devices; and overhead costs.
      We calculated the LOS as the difference between the date of admission and the date of discharge. We also divided the LOS into further categories: preoperative LOS (pre-LOS), the difference between the date of admission and the date of surgery, and postoperative LOS (post-LOS), the difference between the date of surgery and the date of discharge.
      Using observational studies for the analysis of treatment effect suffers from one drawback in comparison with randomized controlled trials: Individuals who receive treatment (MI surgery) are likely to differ in various baseline characteristics from those who do not (conventional surgery). These differences may affect the outcome, leading to biased estimates of the treatment effect [
      • D'Agostino R.B.
      Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group.
      ]. Nevertheless, using administrative data has some advantages: it allows for treatment to be examined as it occurs in routine clinical care [
      • Motheral B.
      • Brooks J.
      • Clark M.A.
      • et al.
      A checklist for retrospective database studies—report of the ISPOR Task Force on Retrospective Databases.
      ] and includes relatively large sample sizes [
      • Sørensen H.T.
      • Sabroe S.
      • Olsen J.
      A framework for evaluation of secondary data sources for epidemiological research.
      ]. Since individuals in routine clinical care are not randomly assigned for treatment, methods adjusting for the missing randomization have to be applied [
      • McKee M.
      • Britton A.
      • Black N.
      • et al.
      Methods in health services research: interpreting the evidence: choosing between randomised and non-randomised studies.
      ]. To address the treatment-selection bias, we applied propensity score matching [
      • Rubin D.B.
      Estimating causal effects from large data sets using propensity scores.
      ]. In the propensity score method, all covariates that might predict treatment and influence the outcome are reduced into a single score, which represents the probability of treatment assignment conditional on observed background covariates [
      • Rosenbaum P.R.
      • Rubin D.B.
      The central role of the propensity score in observational studies for causal effects.
      ]. Assuming that no other confounders exist, matching on propensity scores mimics a randomized treatment assignment, with matched treated and untreated individuals having the same probability of being treated.
      We estimated the probability of selection for MI surgery (the propensity scores) by using multivariate logistic regression analysis. In the model we included all potential confounders, that is, factors that had been reported to be associated with both treatment selection and outcome (costs and LOS). We identified confounders on the basis of a literature search, leading to a total of four confounding factors for our propensity score model: age [
      • Cookson R.
      • Laudicella M.
      Do the poor cost much more? The relationship between small area income deprivation and length of stay for elective hip replacement in the English NHS from 2001 to 2008.
      ,
      • Dall G.F.
      • Ohly N.E.
      • Ballantyne J.A.
      • Brenkel I.J.
      The influence of pre-operative factors on the length of in-patient stay following primary total hip replacement for osteoarthritis: a multivariate analysis of 2302 patients.
      ,
      • Foote J.
      • Panchoo K.
      • Blair P.
      • Bannister G.
      Length of stay following primary total hip replacement.
      ,
      • Fry D.E.
      • Pine M.
      • Jones B.L.
      • Meimban R.J.
      Adverse outcomes in surgery: redefinition of postoperative complications.
      ,
      • Müller M.
      • Tohtz S.
      • Dewey M.
      • et al.
      Muskeltrauma in der primären Hüftendoprothetik unter Berücksichtigung von Alter und BMI sowie in Abhängigkeit vom Operativen Zugangsweg.
      ], sex [
      • Cookson R.
      • Laudicella M.
      Do the poor cost much more? The relationship between small area income deprivation and length of stay for elective hip replacement in the English NHS from 2001 to 2008.
      ,
      • Dall G.F.
      • Ohly N.E.
      • Ballantyne J.A.
      • Brenkel I.J.
      The influence of pre-operative factors on the length of in-patient stay following primary total hip replacement for osteoarthritis: a multivariate analysis of 2302 patients.
      ,
      • Fry D.E.
      • Pine M.
      • Jones B.L.
      • Meimban R.J.
      Adverse outcomes in surgery: redefinition of postoperative complications.
      ,
      • Müller M.
      • Tohtz S.
      • Dewey M.
      • et al.
      Muskeltrauma in der primären Hüftendoprothetik unter Berücksichtigung von Alter und BMI sowie in Abhängigkeit vom Operativen Zugangsweg.
      ], obesity [
      • Oinuma K.
      • Eingartner C.
      • Saito Y.
      • Shiratsuchi H.
      Total hip arthroplasty by a minimally invasive, direct anterior approach.
      ,
      • Azodi O.S.
      • Bellocco R.
      • Eriksson K.
      • Adami J.
      The impact of tobacco use and body mass index on the length of stay in hospital and the risk of post-operative complications among patients undergoing total hip replacement.
      ,
      • Epstein A.M.
      • Read J.L.
      • Hoefer M.
      The relation of body weight to length of stay and charges for hospital services for patients undergoing elective surgery: a study of two procedures.
      ,
      • Chimento G.F.
      • Pavone V.
      • Sharrock N.
      • et al.
      Minimally invasive total hip arthroplasty: a prospective randomized study.
      ], and the diagnosis indicating the THR (e.g., coxarthrosis or osteonecrosis) [
      • Cookson R.
      • Laudicella M.
      Do the poor cost much more? The relationship between small area income deprivation and length of stay for elective hip replacement in the English NHS from 2001 to 2008.
      ]. Since according to previous studies various factors are associated with LOS and hospital costs and an extended LOS has been associated with an increase in resource use [
      • Foote J.
      • Panchoo K.
      • Blair P.
      • Bannister G.
      Length of stay following primary total hip replacement.
      ,
      • Fry D.E.
      • Pine M.
      • Jones B.L.
      • Meimban R.J.
      Adverse outcomes in surgery: redefinition of postoperative complications.
      ,
      • Rana A.J.
      • Iorio R.
      • Healy W.L.
      Hospital economics of primary THA decreasing reimbursement and increasing cost, 1990 to 2008.
      ,
      • Shah A.N.
      • Vail T.P.
      • Taylor D.
      • Pietrobon R.
      Comorbid illness affects hospital costs related to hip arthroplasty: quantification of health status and implications for fair reimbursement and surgeon comparisons.
      ], we used the same propensity scores and consequently the same matched sample for both variables of interest (total costs and LOS).
      In addition, we included dummy variables indicating the hospital of treatment, because this approach has been reported suitable to account for a hierarchical data structure (in our study, patients treated in hospitals) [
      • Arpino B.
      • Mealli F.
      The specification of the propensity score in multilevel observational studies.
      ]. To avoid confounding by different years of the data, we conducted a subgroup matching. We first calculated the propensity scores and conducted the matching for each year separately. Subsequently, we merged the 2008 and 2009 data.
      The matching was conducted by using a one-to-one caliper matching with replacement. On the basis of previously published studies, we chose a caliper width of 0.2 of the SD of the logit of the propensity score [
      • Austin P.C.
      Some methods of propensity-score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations.
      ] and assessed the matching quality by using standardized differences [
      • Austin P.C.
      Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples.
      ]. We rated standardized differences of up to 10% between the covariates as adequately balanced [
      • Normand S.T.
      • Landrum M.B.
      • Guadagnoli E.
      • et al.
      Validating recommendations for coronary angiography following acute myocardial infarction in the elderly: a matched analysis using propensity scores.
      ]. After the matching, we excluded all nonmatched cases from the sample and conducted all further statistical analysis by using the matched sample.
      The effect of the treatment strategy was analyzed in two steps. In a first step, differences between the two treatment groups in our outcome variables (costs and LOS), as well as in the respective subgroups, were assessed by using paired t test and, as a nonparametric alternative, Wilcoxon signed-rank test.
      In a second step, we applied generalized linear models (GLMs) to estimate the effect of an MI treatment on total costs and LOS. We used generalized estimating equations to account for the matched data structure. In the GLMs, we controlled for additional factors that had been associated with hospital costs and LOS: further treatment strategies, such as acetabular roof construction; type of implant (cementless, cemented, or hybrid) [
      • Foote J.
      • Panchoo K.
      • Blair P.
      • Bannister G.
      Length of stay following primary total hip replacement.
      ,
      • Stargardt T.
      Health service costs in Europe: cost and reimbursement of primary hip replacement in nine countries.
      ,
      • Yates P.
      • Serjeant S.
      • Rushforth G.
      • Middleton R.
      The relative cost of cemented and uncemented total hip arthroplasties.
      ]; type of admission (emergency or elective) [
      • Sams J.D.
      • Milbrandt J.C.
      • Froelich J.M.
      • et al.
      Hospital outcome after emergent vs elective revision total hip arthroplasty.
      ]; and comorbidities assessed through the Charlson index [
      • Shah A.N.
      • Vail T.P.
      • Taylor D.
      • Pietrobon R.
      Comorbid illness affects hospital costs related to hip arthroplasty: quantification of health status and implications for fair reimbursement and surgeon comparisons.
      ]. We included these factors only in the GLMs but not in the propensity score matching either because they were not related to treatment selection or because they could not be determined before treatment. In the GLMs with LOS as the dependent variable, we specified a model with a Poisson distribution and a log-link. For the dependent variable total cost, we specified a model with a normal distribution and the natural log of the costs as the dependent variable.
      In the propensity score model and the GLMs, we included all factors using dummy variables, except for the continuous variable age. The value of the dummies was set as 1 if the treatment or the diagnosis had been reported and as 0 if it had not been reported. We retrieved all necessary information from the sociodemographic and medical information included in the administrative electronic patient files. Obese patients (body mass index > 30 kg/m2) were identified by using the ICD-10 code E66. To include the diagnosis leading to the THR, we grouped the cases according to their main diagnosis. If a case had a primary diagnosis not related to THR, we screened all secondary diagnoses and grouped according to those. Finally, we grouped the diagnosis into the five dummy variables: coxarthrosis, fracture, arthritis, osteonecrosis, and others.
      In addition, we applied the enhanced ICD-10–based version of the Charlson index [
      • Charlson M.E.
      • Pompei P.
      • Ales K.L.
      • MacKenzie C.R.
      A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
      ,
      • Quan H.
      • Sundararajan V.
      • Halfon P.
      • et al.
      Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.
      ] to account for comorbidities. Following this approach, we first calculated a weighted global Charlson index score by identifying the relevant ICD-10 codes recorded as secondary diagnoses and by overweighting the 6 most severe among the 17 dimensions of comorbidity proposed by Charlson: the “Hemiplegia/Paraplegia,” “Renal disease,” and “Cancer (any malignancy)” comorbidities are weighted by a coefficient 2, cases of “Moderate or severe liver diseases” by a coefficient 3, and “Metastatic solid tumor” and the “AIDS/HIV” cases by a coefficient 6 (see Charlson et al. [
      • Charlson M.E.
      • Pompei P.
      • Ales K.L.
      • MacKenzie C.R.
      A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
      ] for details). We then used the Charlson index score to group patients into two dummy variables: Charlson1 (patients suffering from one single nonsevere comorbidity; Charlson score = 1) and Charlson2 (patients suffering from at least one severe or two nonsevere comorbidities; Charlson score > 1).
      In all statistical analysis, a P value of ≤0.05 was considered statistically significant. All statistical analyses were performed with SAS 9.2.

      Results

      The 31 hospitals providing data for our study had an average of 448 beds. Most hospitals were private not-for-profit hospitals (n = 19), followed by public hospitals (n = 10) and private for-profit hospitals (n = 2). From the 31 hospitals in our data, we had to exclude 28 hospitals because of no THR cases, no MI cases, or inconsistent coding of MI cases.
      The remaining three hospitals provided the additional data from 2009, leading to a final sample of three hospitals and 2886 cases. Compared with the total sample, the remaining hospitals had fewer beds on average (mean = 300) and they were all private not-for-profit hospitals.
      In the three hospitals, MI THR was defined by reduced tissue and muscle dissection. The sample characteristics are displayed in Table 1. A total of 812 (28.14%) patients were treated with MI surgery. The average age of the patients was 67 years, and 60.33% were women. Coxarthrosis was the most frequent diagnosis (90.61%), followed by arthritis (4.23%), osteonecrosis (2.22%), and fractures (1.56%). It was found that 1.39% of the patients had none of these diagnoses and was summed up in the group “other diagnosis.” In the sample, 18.71% of the patients were coded as obese, 18.02% had a Charlson score of 1, and 4.44% had a Charlson score above 1. In more than 80% of the cases, cementless prostheses were used, followed by hybrid prostheses (15.59%) and cemented prostheses (2.08%). Acetabular roof reconstruction was found in 1.14% of the patients, and osteosynthesis equipment was used in 0.62% of the cases.
      Table 1Sample characteristics (n = 2886).
      VariableMean (SD)/% (number of cases)
      Patient characteristics
       Age67.12(11.10)
       Sex male39.67%(1,145)
       Obesity18.71%(540)
       Coxarthrosis90.61%(2,615)
       Arthritis4.23%(122)
       Osteonecrosis2.22%(64)
       Fracture1.56%(45)
       Other diagnosis1.39%(40)
       Charlson118.02%(520)
       Charlson24.44%(128)
       Emergency1.91%(55)
      Treatment
       Minimally invasive28.14%(812)
       Cementless prosthesis82.33%(2,376)
       Cemented prosthesis2.08%(60)
       Hybrid prosthesis15.59%(450)
       Acetabular roof reconstruction1.14%(33)
       Osteosynthesis equipment0.62%(18)
      SD, standard deviation.
      The average LOS was 11.85 ± 3.86 days with a minimum LOS of 2 days and a maximum LOS of 75 days. The average pre-LOS was 1.27 ± 2.13 days, and the average post-LOS was 10.66 ± 3.35 days.
      The total direct hospital costs ranged from €3,364 to €27,960 with a mean of €6,146 ± €1,340. The overhead costs were responsible for the largest share of total costs, followed by implant costs, staff costs (nursing service, medical service, and medical-technical service), costs for medical devices, and pharmaceutical costs.
      Both logistic regression models (for the years 2008 and 2009) calculating the propensity scores had a high ability of predicting treatment type (2008: c-statistic = 0.869; 2009: c-statistic = 0.852). Matching on the propensity scores resulted in a total of 812 matches; thus, every MI case could be matched to a conventional case. Differences in the unmatched sample indicate that conventionally treated patients were older, were more likely women, and were more likely obese. MI treated patients were more likely to have coxarthrosis as the main diagnosis. After matching the patients along covariates and cluster variables, the standardized differences reduced substantially, from between 0.47% and 137.39% in the unmatched sample to between 0.25% and 6.65% in the matched sample (see Appendix Table 1 in Supplemental Materials found at http://dx.doi.org/10.1016/j.jval.2012.06.008). According to the predefined threshold of 10%, we rated the matched sample as adequately balanced.
      After the matching, the differences between the conventionally and the MI treated cases consisted in further treatment strategies and comorbidities. The biggest difference was regarding the kind of prosthesis, as MI patients were more likely to have a cementless prosthesis (82.39% vs. 90.39%) while conventional patients were more likely to receive a hybrid prosthesis (12.07% vs. 7.76%). Conventionally treated patients also experienced more severe comorbidities; 11.08% of the patients had a Charlson score of 1 and 7.27% had a Charlson score above 1, compared with 13.18% and 4.10% of the MI treated patients, respectively.
      The results of the group comparisons are shown in Table 2, Table 3. In the matched sample, conventionally treated patients who were compared with MI treated patients had a significantly longer LOS (11.49 days vs. 10.90 days) and post-LOS (10.29 days vs. 9.71 days). In the unmatched sample, the difference in LOS between conventionally and MI treated patients was higher than in the matched sample (12.22 days vs. 10.90 days).
      Table 2Comparison of LOS (in days) of MI and conventionally treated patients (n = 1624).
      VariableMeanMedianMinMaxSD95% CIP
      LOS
       Conventional11.49117463.3311.26–11.72<0.01
      Paired t test.
       MI10.90105754.3910.60–11.20
      Pre-LOS
       Conventional1.2410331.941.10–1.370.34
      Wilcoxon signed-rank test.
       MI1.2610302.291.10–1.42
      Post-LOS
       Conventional10.29106242.7110.11–10.48<0.001
      Paired t test.
       MI9.7194663.629.46–9.96
      LOS unmatched (n = 2886)
       Conventional12.22122473.5612.07–12.37<0.001
      Unequal variance t test.
       MI10.90105754.3910.60–11.20
      CI, confidence interval; LOS, length of stay; Pre-LOS, preoperative length of stay; Post-LOS, postoperative length of stay; MI, minimally invasive; SD, standard deviation.
      low asterisk Paired t test.
      Wilcoxon signed-rank test.
      Unequal variance t test.
      Table 3Comparison of hospital costs (in €) of MI and conventionally treated patients (n = 1624).
      VariableMeanMedianMinMaxSD95% CIP
      Total
       Conventional6,0185,8854,19725,4911,5215,913–6,1230.67
      Paired t test.
       MI5,9865,8513,72627,9601,5275,881–6,091
      Staff—Medical service
       Conventional9258324535,645369900–951<0.001
      Wilcoxon signed-rank test.
       MI8818213705,406305860–902
      Staff—Nursing service
       Conventional9258614434,248227899–9510.05
      Wilcoxon signed-rank test.
       MI89582076,847422866–924
      Staff—Medical-technical service
       Conventional6225552242,993227607–638<0.001
      Wilcoxon signed-rank test.
       MI5935453012,849197579–606
      Implant
       Conventional1,3751,206102,9514831,342–1,408<0.001
      Wilcoxon signed-rank test.
       MI1,5141,445183,6895141,479–1,550
      Pharmaceuticals
       Conventional109104238367399–1180.94
      Paired t test.
       MI10997233,074136104–114
      Medical devices
       Conventional4384441182,952201424–4520.19
      Paired t test.
       MI4264181062,545180414–439
      Overhead
       Conventional1,6241,5079367,7615101,588–1,659<0.05
      Paired t test.
       MI1,5681,45762410,0005501,531–1,606
      Total unmatchted (n = 2886)
       Conventional6,2085,9963,36425,4911,2556,154–6,262<0.001
      Unequal variance t test.
       MI5,9865,8513,72627,9601,5275,881–6,091
      CI = confidence interval; MI, minimally invasive; SD, standard deviation.
      low asterisk Paired t test.
      Wilcoxon signed-rank test.
      Unequal variance t test.
      While we found a slight but nonsignificant difference in the total cost between MI (€5986) and conventional (€6018) cases in the matched sample, we found significant differences in the other cost categories. For conventionally treated patients, significantly higher costs were reported for overhead costs and all staff cost categories, that is, medical service, nursing service, and medical-technical service. Only the reported implant costs were significantly higher for the MI cases. In the unmatched sample, however, we found a bigger difference in the total cost between MI and conventional cases (€6208 vs. €5986). It has to be considered though that differences in the P value between the matched and unmatched group comparisons are also determined by the different sample sizes.
      Table 4 displays the allocation of costs for conventional and MI cases. In both groups, overhead costs account for the highest share of total costs, with 27% in the conventional surgery group and 26% in the MI surgery group, followed by implant costs, which account for 23% and 25%, respectively.
      Table 4Allocation of costs for conventionally and MI treated patients (n = 1624).
      Cost categoryConventionalMinimally Invasive
      Mean (€)% allocation of costsMean (€)% allocation of costs
      Medical service92515.38%88114.72%
      Nursing service92515.37%89514.95%
      Medical-technical service62210.34%5939.90%
      Pharmaceuticals1091.80%1091.82%
      Implant1,37522.85%1,51425.29%
      Other medical devices4387.28%4267.12%
      Overhead1,62426.98%1,56826.20%
      Total costs6,018100%5,986100%
      The results from the GLMs (Table 5, Table 6) are consistent with the prior results from the group comparisons: an MI treatment significantly reduced LOS, but there was no significant effect on total costs.
      Table 5Results from the GLM, dependent variable: length of stay (n = 1624).
      ParameterEstimateSE95% CIP
      Intercept2.38080.01202.3572–2.4044<0.0001
      Acetabular roof construction−0.05190.0192−0.0894–−0.0143<0.01
      Osteosynthesis equipment0.14100.0737−0.0036–0.28550.0559
      CementlessRef.
      Cemented0.34390.09950.1489–0.5390<0.001
      Hybrid0.04500.0278−0.0095–0.09960.1057
      No comorbiditiyRef.
      Charlson10.06830.02340.0224–0.1142<0.01
      Charlson20.39270.08210.2319–0.5535<0.0001
      Emergency0.37610.07570.2277–0.5245<0.0001
      MI−0.04420.0165−0.0765–−0.0120<0.01
      CI, confidence interval; GLM, generalized linear model; MI, minimally invasive; Ref., reference category; SE, standard error.
      Table 6Results from the GLM, dependent variable: Log of total costs (n = 1624).
      ParameterEstimateSE95% CIP
      Intercept8.64500.00748.6304–8.6596<0.0001
      Acetabular roof construction0.08170.01410.0541–0.1093<0.0001
      Osteosynthesis equipment0.26440.04540.1754–0.3533<0.0001
      CementlessRef.
      Cemented0.20770.06590.0786–0.3368<0.01
      Hybrid0.05480.01560.0242–0.0853<0.001
      No comorbidityRef.
      Charlson10.00740.0125−0.0171 to 0.03190.55
      Charlson20.17830.04070.0984–0.2581<0.0001
      Emergency0.15390.05200.0521–0.2558<0.01
      MI0.00980.0083−0.0064 to 0.02600.23
      CI, confidence interval; GLM, generalized linear model; MI, minimally invasive; Ref., reference category; SE, standard error.

      Discussion

      A total of 812 matched pairs resulted from the propensity score matching. Thus, every MI THR case was matched with a conventional THR control and no MI THR cases had to be excluded.
      The results of the group comparisons show that MI treated THR patients have a significantly shorter LOS and post-LOS, controlling for diagnosis, age, sex, obesity, hospital, and year of treatment. We found no significant difference in the total cost between MI and conventional cases, but we found differences in the cost categories, with implant costs as the only cost category significantly higher for MI treated patients. Differences in implant cost are partly explained by different implants for conventional and MI THR but are also because MI patients more frequently were given the more expensive cementless prostheses [
      • Stargardt T.
      Health service costs in Europe: cost and reimbursement of primary hip replacement in nine countries.
      ,
      • Yates P.
      • Serjeant S.
      • Rushforth G.
      • Middleton R.
      The relative cost of cemented and uncemented total hip arthroplasties.
      ]. Staff costs and overhead costs were significantly higher in conventionally treated patients. The allocation of costs shows that overhead as well as implant costs account for the majority of the total costs. Thus, our results suggest that despite differences in terms of LOS, the lack of significant total cost difference between the two treatment groups is due to the higher implant costs for MI patients. This finding is consistent with a previous study showing that implant costs are a major cost driver in THR [
      • Rana A.J.
      • Iorio R.
      • Healy W.L.
      Hospital economics of primary THA decreasing reimbursement and increasing cost, 1990 to 2008.
      ].
      The results of the GLMs indicate that the type of treatment (MI compared with a conventional surgery) in THR has a significant effect on LOS, with MI patients having a reduced LOS after controlling for diagnosis, age, sex, obesity, comorbidities, hospital and year of treatment. Holding all other variables constant, an MI treatment is associated with a decrease in LOS by 4.3%. With an intercept of 10.81 days, this relates to a change of approximately −0.5 days and thus is very similar to the results from the univariate group comparison. In the GLMs with hospital cost as the depending variable, we could not find an effect of an MI treatment.
      Therefore, the results of our analysis suggest that when compared with a conventional approach to THR, an MI approach leads to a reduced LOS with no differences in total costs. On the one hand, from a hospital management perspective, considering that there is no difference in the reimbursement of MI THR and conventional THR, an MI approach in THR is attractive as a possible way of saving resources through shorter LOS. On the other hand, the MI approach is associated with higher implant costs. Incurring higher implant costs may lead to pressure to reduce costs in other cost categories to avoid the higher overall costs for MI THR compared with conventional THR. Still, with a reduced LOS and no cost differences, an MI THR approach is more attractive than a conventional approach from a hospital management perspective. Overall, the differences between the MI and the conventional approach are rather small. Thus, a clear recommendation for or against MI THR is not possible, especially as we cannot estimate the differences in medical outcome. Further studies that assess the quality of the different treatment approaches are needed.
      To our knowledge, this is the first study that analyzes the effect of MI treatment for THR on hospital costs and LOS by using observational data to ensure a relatively large sample size and estimates from routine operations. If observational data are used to compare different treatment groups, however, there is a high chance for wide baseline imbalances, likely due to treatment-selection bias. To account for baseline imbalances, we implemented a propensity score matching. Because German hospital data do not make it possible to distinguish between diagnosis at admission and conditions acquired during the hospital stay, we carefully chose the diagnoses for inclusion in the propensity score model and included only those diagnoses that had very likely been recorded before the hip replacement. As a propensity score matched-pair sample does not consist of independent subjects, suitable methods for matched data have to be applied [
      • Austin P.C.
      Propensity-score matching in the cardiovascular surgery literature from 2004 to 2006: a systematic review and suggestions for improvement.
      ]. We did this by using paired t test and Wilcoxon signed-rank test for the group comparisons and GLMs with generalized estimating equations for the regression analysis [
      • Austin P.C.
      A critical appraisal of propensity-score matching in the medical literature between 1996 and 2003.
      ]. The differences between the matched and the unmatched group comparisons clearly indicate how the baseline imbalances between MI and conventionally treated patients influence the results. A careful consideration of possible baseline imbalances is necessary when observational data are analyzed to assess the difference between MI and conventional THR.
      Our study suffers from several limitations. Using hospital records for our analysis, the quality of the data and our results are highly determined by the coding quality of the participating hospitals. In particular, an inconsistent coding of the treatment strategy would bias the results and possibly minimize statistical differences between the two treatment groups. Because hospitals in Germany do not have a financial incentive for coding an MI technique of THR because reimbursement for MI and conventional THR does not differ, we assumed that not all hospitals in Germany code MI THR regularly and also do not code MI THR if a conventional THR had been conducted (i.e., there is no “upcoding”: coding of a more expensive procedure even though the cheaper alternative is used). To ensure interpretability of our results, we included only those hospitals that affirmed a consistent coding of MI THR, resulting in a sample of only three hospitals. Hence, the sample is not representative of all hospitals in Germany, and the generalizability of our results is limited. In addition, it is not possible to make inferences about causality based on our study. Randomized controlled trials provide much stronger evidence for causality because the randomization process ensures that the groups are comparable on observed and unobserved factors. Thus, further studies, especially randomized controlled trials analyzing the economic impact of MI surgery in THR, are still needed.
      Furthermore, our study suggests that future research based on observational data needs to carefully distinguish among different definitions of THR MI surgery because we find that only 3 of 23 hospitals with THR procedures consistently coded MI surgeries by using the OPS code 5-986. In fact, although 77% of the German surgeons report using MI surgery [
      • Sendtner E.
      • Boluki D.
      • Grifka J.
      Current state of doing minimal invasive total hip replacement in Germany, the use of new implants and navigation – results of a nation-wide survey.
      ], it lacks a clear definition. Both professional medical associations and policymakers in Germany need to clarify relevant definitions and procedure codes.
      Source of financial support: The data collection for this analysis was funded through the Seventh Framework Programme (FP7) of the European Commission under Grant Agreement Number 223300.

      Supplemental Materials

      References

        • Learmonth I.D.
        • Young C.
        • Rorabeck C.
        The operation of the century: total hip replacement.
        Lancet. 2007; 370: 1508-1519
        • Kurtz S.M.
        • Ong K.L.
        • Schmier J.
        • et al.
        Primary and revision arthroplasty surgery caseloads in the United States from 1990 to 2004.
        J Arthroplasty. 2009; 24: 195-203
        • Ethgen O.
        • Bruyère O.
        • Richy F.
        • et al.
        Health-related quality of life in total hip and total knee arthroplasty: a qualitative and systematic review of the literature.
        J Bone Joint Surg Am. 2004; 86: 963-974
        • Federal Statistical Office
        Die 50 häufigsten Operationen der vollstationären Patientinnen und Patienten in Krankenhäusern.
        (Accessed July 2, 2011)
        • Culliford D.J.
        • Maskell J.
        • Beard D.J.
        • et al.
        Temporal trends in hip and knee replacement in the United Kingdom 1991 to 2006.
        J Bone Joint Surg Br. 2010; 92B: 130-135
        • Kurtz S.
        • Ong K.
        • Lau E.
        • et al.
        Projections of primary and revision hip and knee arthroplasty in the United States from 2005 to 2030.
        J Bone Joint Surg Am. 2007; 89A: 780-785
        • Kurtz S.M.
        • Lau E.
        • Ong K.
        • et al.
        Future young patient demand for primary and revision joint replacement: national projections from 2010 to 2030.
        Clin Orthop. 2009; 467: 2606-2612
        • Pulido L.
        • Parvizi J.
        • Macgibeny M.
        • et al.
        In hospital complications after total joint arthroplasty.
        J Arthroplasty. 2008; 23: 139-145
        • Sculco T.P.
        Minimally invasive total hip arthroplasty: in the affirmative.
        J Arthroplasty. 2004; 19: 78-80
        • de Verteuil R.
        • Imamura M.
        • Zhu S.
        • et al.
        A systematic review of the clinical effectiveness and cost-effectiveness and economic modelling of minimal incision total hip replacement approaches in the management of arthritic disease of the hip.
        Health Technol Assess. 2008; 12: 1-244
        • Pospischill M.
        • Kranzl A.
        • Attwenger B.
        • Knahr K.
        Minimally invasive compared with traditional transgluteal approach for total hip arthroplasty: a comparative gait analysis.
        J Bone Joint Surg Am. 2010; 92: 328-337
        • Dorr L.D.
        • Maheshwari A.V.
        • Long W.T.
        • et al.
        Early pain relief and function after posterior minimally invasive and conventional total hip arthroplasty: a prospective, randomized, blinded study.
        J Bone Joint Surg Am. 2007; 89: 1153-1160
        • Goosen J.
        • Kollen B.
        • Castelein R.
        • et al.
        Minimally invasive versus classic procedures in total hip arthroplasty: a double-blind randomized controlled trial.
        Clin Orthop. 2011; 469: 200-208
        • Ogonda L.
        • Wilson R.
        • Archbold P.
        • et al.
        A minimal-incision technique in total hip arthroplasty does not improve early postoperative outcomes: a prospective, randomized, controlled trial.
        J Bone Joint Surg Am. 2005; 87: 701-710
        • Pour A.E.
        • Parvizi J.
        • Sharkey P.F.
        • et al.
        Minimally invasive hip arthroplasty: what role does patient preconditioning play?.
        J Bone Joint Surg Am. 2007; 89A: 1920-1927
        • Sendtner E.
        • Boluki D.
        • Grifka J.
        Current state of doing minimal invasive total hip replacement in Germany, the use of new implants and navigation – results of a nation-wide survey.
        Z Orthop Unfall. 2007; 145: 297-302
        • Mahmood A.
        • Zafar M.S.
        • Majid I.
        • et al.
        Minimally invasive hip arthroplasty: a quantitative review of the literature.
        Br Med Bull. 2007; 84: 37-48
        • Cheng T.
        • Feng J.
        • Liu T.
        • Zhang X.
        Minimally invasive total hip arthroplasty: a systematic review.
        Int Orthop. 2009; 33: 1473-1481
        • Smith T.
        • Blake V.
        • Hing C.
        Minimally invasive versus conventional exposure for total hip arthroplasty: a systematic review and meta-analysis of clinical and radiological outcomes.
        Int Orthop. 2011; 35: 173-184
        • Duwelius P.J.
        • Moller H.S.
        • Burkhart R.L.
        • et al.
        The economic impact of minimally invasive total hip arthroplasty.
        J Arthroplasty. 2011; 26: 883-885
        • D'Agostino R.B.
        Propensity score methods for bias reduction in the comparison of a treatment to a non-randomized control group.
        Stat Med. 1998; 17: 2265-2281
        • Motheral B.
        • Brooks J.
        • Clark M.A.
        • et al.
        A checklist for retrospective database studies—report of the ISPOR Task Force on Retrospective Databases.
        Value Health. 2003; 6: 90-97
        • Sørensen H.T.
        • Sabroe S.
        • Olsen J.
        A framework for evaluation of secondary data sources for epidemiological research.
        Int J Epidemiol. 1996; 25: 435-442
        • McKee M.
        • Britton A.
        • Black N.
        • et al.
        Methods in health services research: interpreting the evidence: choosing between randomised and non-randomised studies.
        BMJ. 1999; 319: 312-315
        • Rubin D.B.
        Estimating causal effects from large data sets using propensity scores.
        Ann Intern Med. 1997; 127: 757-763
        • Rosenbaum P.R.
        • Rubin D.B.
        The central role of the propensity score in observational studies for causal effects.
        Biometrika. 1983; 70: 41-55
        • Cookson R.
        • Laudicella M.
        Do the poor cost much more?.
        Soc Sci Med. 2011; 72: 173-184
        • Dall G.F.
        • Ohly N.E.
        • Ballantyne J.A.
        • Brenkel I.J.
        The influence of pre-operative factors on the length of in-patient stay following primary total hip replacement for osteoarthritis: a multivariate analysis of 2302 patients.
        J Bone Joint Surg Br. 2009; 91B: 434-440
        • Foote J.
        • Panchoo K.
        • Blair P.
        • Bannister G.
        Length of stay following primary total hip replacement.
        Ann R Coll Surg Engl. 2009; 91: 500-504
        • Fry D.E.
        • Pine M.
        • Jones B.L.
        • Meimban R.J.
        Adverse outcomes in surgery: redefinition of postoperative complications.
        Am J Surg. 2009; 197: 479-484
        • Müller M.
        • Tohtz S.
        • Dewey M.
        • et al.
        Muskeltrauma in der primären Hüftendoprothetik unter Berücksichtigung von Alter und BMI sowie in Abhängigkeit vom Operativen Zugangsweg.
        Orthopäde. 2011; 40: 217-223
        • Oinuma K.
        • Eingartner C.
        • Saito Y.
        • Shiratsuchi H.
        Total hip arthroplasty by a minimally invasive, direct anterior approach.
        Oper Orthop Traumatol. 2007; 19: 310-326
        • Azodi O.S.
        • Bellocco R.
        • Eriksson K.
        • Adami J.
        The impact of tobacco use and body mass index on the length of stay in hospital and the risk of post-operative complications among patients undergoing total hip replacement.
        J Bone Joint Surg Br. 2006; 88B: 1316-1320
        • Epstein A.M.
        • Read J.L.
        • Hoefer M.
        The relation of body weight to length of stay and charges for hospital services for patients undergoing elective surgery: a study of two procedures.
        Am J Public Health. 1987; 77: 993-997
        • Chimento G.F.
        • Pavone V.
        • Sharrock N.
        • et al.
        Minimally invasive total hip arthroplasty: a prospective randomized study.
        J Arthroplasty. 2005; 20: 139-144
        • Rana A.J.
        • Iorio R.
        • Healy W.L.
        Hospital economics of primary THA decreasing reimbursement and increasing cost, 1990 to 2008.
        Clin Orthop. 2011; 469: 355-361
        • Shah A.N.
        • Vail T.P.
        • Taylor D.
        • Pietrobon R.
        Comorbid illness affects hospital costs related to hip arthroplasty: quantification of health status and implications for fair reimbursement and surgeon comparisons.
        J Arthroplasty. 2004; 19: 700-705
        • Arpino B.
        • Mealli F.
        The specification of the propensity score in multilevel observational studies.
        Comput Stat Data Anal. 2011; 55: 1770-1780
        • Austin P.C.
        Some methods of propensity-score matching had superior performance to others: results of an empirical investigation and Monte Carlo simulations.
        Biometr J. 2009; 51: 171-184
        • Austin P.C.
        Balance diagnostics for comparing the distribution of baseline covariates between treatment groups in propensity-score matched samples.
        Stat Med. 2009; 28: 3083-3107
        • Normand S.T.
        • Landrum M.B.
        • Guadagnoli E.
        • et al.
        Validating recommendations for coronary angiography following acute myocardial infarction in the elderly: a matched analysis using propensity scores.
        J Clin Epidemiol. 2001; 54: 387-398
        • Stargardt T.
        Health service costs in Europe: cost and reimbursement of primary hip replacement in nine countries.
        Health Econ. 2008; 17: S9-S20
        • Yates P.
        • Serjeant S.
        • Rushforth G.
        • Middleton R.
        The relative cost of cemented and uncemented total hip arthroplasties.
        J Arthroplasty. 2006; 21: 102-105
        • Sams J.D.
        • Milbrandt J.C.
        • Froelich J.M.
        • et al.
        Hospital outcome after emergent vs elective revision total hip arthroplasty.
        J Arthroplasty. 2010; 25: 826-828
        • Charlson M.E.
        • Pompei P.
        • Ales K.L.
        • MacKenzie C.R.
        A new method of classifying prognostic comorbidity in longitudinal studies: development and validation.
        J Chronic Dis. 1987; 40: 373-383
        • Quan H.
        • Sundararajan V.
        • Halfon P.
        • et al.
        Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data.
        Med Care. 2005; 43: 1130-1139
        • Austin P.C.
        Propensity-score matching in the cardiovascular surgery literature from 2004 to 2006: a systematic review and suggestions for improvement.
        J Thorac Cardiovasc Surg. 2007; 134: 1128-1135
        • Austin P.C.
        A critical appraisal of propensity-score matching in the medical literature between 1996 and 2003.
        Stat Med. 2008; 27: 2037-2049