Perioperative predictability of unsatisfactory functional outcomes 6 months after hip arthroplasty

Abstract

Background

Aiming at an early identification of patients with an unsatisfactory outcome after hip arthroplasty we investigated the pre-, intra- and early postoperative predictability.

Methods

Using logistic regression models at nine different time points we calculated the area under the curve of the receiver operating characteristic (ROC-AUC) to compare the predictability of an unsatisfactory outcome.

Results

The predictability of unsatisfactory outcomes increased significantly from a ROC-AUC (95% CI) of 0.73 (0.62–0.84) in the preoperative setting to 0.85 (0.77–0.94) on day 2 after surgery.

Conclusion

Most of the patients with unsatisfactory outcomes can be identified during the first days after surgery.

Highlights

• •The majority of patients with unsatisfactory functional outcomes can be identified during the first days after surgery.
• •A higher acute postoperative pain intensity and a lower analgesic consumption are risk factors for an unsatisfactory outcome.
• •An immediate and sufficient acute pain management might prevent from unsatisfactory functional outcomes.

1. Introduction

Total hip arthroplasty is a common procedure that leads to pain relief¹ and to a distinct improvement of the activity level in the majority of patients.² However, more than 10% of the patients are not satisfied with their postoperative long-term outcome, with satisfaction closely associated to functional outcome.³ Focusing measures on this group of patients requires an early prediction of unsatisfactory functional outcomes. Reported risk factors for an unsatisfactory outcome are higher age,⁴ female sex,⁴ limited preoperative function,⁵ and psychological factors such as pain catastrophizing,⁶ anxiety⁵ and depression.⁷

Based on risk factors, prediction models for an unsatisfactory hip replacement outcome have been constructed. However, despite the use of large registry data sets, the models have failed to show a clinically relevant performance.^8,9 One of the main reasons for this limited prediction accuracy might be the neglect of important confounding factors. For instance, acute postoperative characteristics like pain intensity and analgesic consumption have been frequently identified as risk factors for the development of persistent pain.¹⁰ Therefore, these early postoperative characteristics might also be risk factors for an unsatisfactory functional outcome since pain influences mobility and function.¹¹ Likewise, intraoperative characteristics such as the dosing of intraoperative analgesics might also be related to the development of persistent pain.¹² Hence, including intra- and early postoperative characteristics into a prediction model might allow for an improved prediction of an unsatisfactory functional outcome after orthopaedic surgery.

Therefore, we performed this prospective observational study using pre-, intra- and early postoperative characteristics to identify perioperative risk factors and to estimate the predictability of an unsatisfactory functional outcome after hip arthroplasty.

2. Methods

2.1. Ethics

This monocentric prospective observational study has been approved by the local ethics committee (Ethikausschuss 1, Charité – Universitätsmedizin Berlin, application number EA1/033/13, 4 April 2013, Chairperson Prof. R. Uebelhack) and registered at the German register for clinical studies (Deutsches Register Klinischer Studien, registration number DRKS00000665).

2.2. Objectives

The primary objective of the study was to investigate the predictability of the immediate postoperative pain using intraoperative nociceptive reflexes. This manuscript only reports the results in regard to one of the secondary objectives of the study, which was to investigate the predictability of the clinical outcome six month after surgery using the pre-, intra- and early postoperative characteristics. Data from this study in regard to the primary objective and to one other secondary objective have already been published. In regard to the primary objective we already published the results about the ability of nociceptive reflexes for intraoperative analgesia monitoring and to predict immediate postoperative pain.¹³ In another publication we reported the results of a post-hoc analysis of the correlation between intraoperative analgesia and persistent pain after surgery.¹²

In this third work on the data of the same population we focus on the long-term functional outcome, specifically the 6-months-postoperative Oxford Hip Score (OHS). The postoperative data of the days after the day of surgery and the 6-months-postoperative OHS data that are analyzed here have not been reported in the other two publications.

2.3. Study population

This study was conducted on patients scheduled to undergo total hip arthroplasty. Patients with American Society of Anesthesiologists (ASA) status > 3, body mass index (BMI) > 35, use of regional anaesthesia and participation in other clinical studies were excluded from the study. The sample size was determined by the primary objective of the study to investigate a significant correlation at a significance level of 0.05 of the intraoperative nociceptive reflexes and the immediate postoperative pain with a power of 0.8.¹³

2.4. Course of the study

Eligible patients were included in the study after obtaining written informed consent. Data were collected preoperatively approximately one week before surgery, on the day of surgery, on each day within six days after surgery as well as six months after surgery. Patients who underwent revision surgery during the first six months after surgery were excluded from this analysis.

2.5. Preoperative assessment

In the preoperative setting the patients’ age, sex and BMI were recorded. Additionally, the patients were asked to indicate whether they were diabetic and to indicate daily consumption of nicotine or alcohol. To assess preoperative pain the patients were asked to rate the average pain intensity at the affected hip on a numeric rating scale (NRS) from 0 to 10. Further, the Oxford Hip Score was used to assess hip related discomfort.¹⁴

To assess psychological factors like anxiety, depression and pain catastrophizing validated German versions of the State-Trait-Anxiety-Inventory (STAI),¹⁵ Hospital Anxiety and Depression Scale (HADS)¹⁶ and Pain Catastrophizing Scale (PCS)¹⁷ were used. Pain sensitivity was assessed using the Pain Sensitivity Questionnaire (PSQ).¹⁸

2.6. Surgical and anaesthetic technique

Surgical and anaesthetic treatment was performed following the hospital's guidelines independently of the study. All patients underwent total hip arthroplasty under general anaesthesia. The lateral transgluteal standard approach was used in all patients.¹⁹ For maintaining general anaesthesia a standard protocol with propofol and remifentanil was used. Piritramide was administered intraoperatively to reduce immediate postoperative pain.

2.7. Postoperative treatment

The postoperative treatment including analgesic treatment was performed independently of the study by the regular hospital staff according to the hospital's guidelines. For acute postoperative analgesia each patients received intravenous piritramide boluses by the staff nurses of the postoperative care unit (PACU) and later on via patient controlled analgesia (PCA) pumps, providing 2 mg piritramide boluses with 5 min blocking-time and a maximum of 60 mg within 4 h. The PCA pumps were left for 24–72 h postoperatively. After removal of the PCA pump, supplementary opioid analgesics were administered by the staff nurses if needed.

Standardly administered oral analgesics were metamizole and a non-steroidal anti-inflammatory drug (NSAID). Physiotherapeutic measures were performed independent of the study by the hospital's physiotherapist according to hospital's guidelines.

2.8. Postoperative assessment

To obtain data about the immediate postoperative pain the patients were asked to rate their pain intensity at the arrival in the PACU and 1 h after surgery. Further, the cumulative consumption of piritramide within the first hour and the first 24 h was recorded.

From day 1 until day 6 after surgery the patients were asked daily to rate their pain intensity at rest and in motion. Patients’ degree of mobility was recorded following a 6-item-scale from bed rest to climbing stairs.

The analgesic intake was scaled from 0 to 3, reflecting the intake of no analgesics (0), NSAIDs (1), weak opioids (2) and strong opioids (3).

2.9. Functional outcome measures

To assess hip related discomfort six months after surgery, the patients were asked to provide a completed OHS form. The OHS form and a short instruction were sent to them by mail. If no letter was returned, the patients were contacted via telephone and asked to answer the questionnaire. Primary functional outcome measure was the difference of the pre- and postoperative OHS using the revised scoring system.¹⁴ A difference of less than 14 points between the pre- and postoperative Oxford Hip Score was counted as an unsatisfactory functional result.²⁰

2.10. Statistical analysis

To identify risk factors at each time point and to investigate the predictability of an unsatisfactory outcome (defined as difference of less than 14 between the pre- and postoperative oxford hip score), logistic regression models at 9 different time points were calculated: Directly after the preoperative assessment (preoperative model), directly after surgery (intraoperative model), directly after the day of surgery (day 0 model) and after each postoperative day until day 6 (day 1–6 models).

At each time point, a logistic regression model was fit with a stepwise selection of those variable that were additionally available at the respective time point but not at the previous ones (variable inclusion/exclusion criteria: probability <0.20/> 0.40, respectively). Additionally, the model at each time point included all variables that were selected at all previous time points. To assess the predictive performance of each of the logistic regression models, the area under the receiver operating characteristic curve (ROC-AUC) was determined, a measure that takes values between 0.5, if the model has no predictive performance (i.e. is no better than chance) and 1.0, if the model perfectly predicts unsatisfactory outcomes. Further, to estimate the goodness of the model fit the Nagelkerke r² was calculated for each of the logistic regression models.

3. Results

122 patients were primarily included in this study, of which 101 patients (49 female/52 male) were included for statistical analysis (Fig. 1). The demographic and clinical characteristics are shown in Table 1. During the first three days after surgery we observed a decrease of the reported pain intensity at rest and during exercise, whereas from day 3 to day 6 the pain intensity remained on a steady level. The share of patients that use opioid analgesics also decreased from day 1 (90%) to day 3 (32%) after surgery. On day six after surgery 22% of the patients were still using opioid analgesics (Table 1). The functional performance also changed for the most part on the first three days. For instance, on day 2 only one third of the patients used the bathroom without any help, whereby on day 3 almost 90% achieved this goal.

Fig. 1.

Flow chart of patient inclusion.

Table 1.

Demographic and clinical characteristics.

Preoperative data
Included patients, n	101
Age, median (IQR)	72 (62–76) years
Sex, male/female	52/49
Body Mass Index, median (IQR)	26.7 (23.8–30.6) kg/m2
ASA Status [I-III], median (IQR)	2 (2–3)
Daily nicotine consumption, yes/no	21/80
Daily alcohol consumption, yes/no	10/91
Intake of analgesics [0-III], median (IQR)	0 (0–1)
Average pain intensity at affected hip [NRS 0–10], median (IQR)	5 (3–7)
Oxford Hip Score, median (IQR)	21 (14–25)
Hospital Anxiety and Depression Scale (HADS), median (IQR)	10 (5–14)
State-Trait-Anxiety-Inventory (STAI) state anxiety, median (IQR)	39 (32–45)
State-Trait-Anxiety-Inventory (STAI) trait anxiety, median (IQR)	36 (24–40)
Pain Catastrophizing Scale (PCS), median (IQR)	11 (5–19)
Pain Sensitivity Questionnaire (PSQ), median (IQR)	3.8 (3–4.6)
Intraoperative data
Duration of surgery, median (IQR)	80 (53–115) minutes
Average propofol rate, median (IQR)	6.7 (6.2–7.4) mg/kg/h
Average remifentail rate, median (IQR)	0.19 (0.14–0.24) μg/kg/min
Piritramide dose, median (IQR)	0.12 (0.09–0.17) mg/kg
Metamizole dose, median (IQR)	23.53 (20.5–28.4) mg/kg
Postoperative data
Pain intensity at the arrival in the PACU [NRS 0–10], median (IQR) [n = 106]	5 (2–6)
Cumulative piritramide consumption 1h after surgery, median (IQR)	0.04 (0–0.08) mg/kg
Cumulative piritramide consumption 24h after surgery, median (IQR)	0.34 (0.17–0.54) mg/kg
Postoperative Nausea or Vomiting, yes/no	8/93
Day 1
Pain intensity at rest [NRS 0–10], median (IQR)	1 (0–3)
Pain intensity during movement [NRS 0–10], median (IQR)	6 (4–8)
Mobility [n (%) of patients using bathroom without help]	1 (1%)
Cumulative piritramide i.v. consumption 24h after surgery, median (IQR)	0.34 (0.17–0.54) mg/kg
Patient using opioid analgesics, n (%)	91 (90.1%)
Day 2
Pain intensity at rest [NRS 0–10], median (IQR)	0 (0–2)
Pain intensity during movement [NRS 0–10], median (IQR)	5 (3–6)
Mobility [n (%) of patients using bathroom without help]	33 (32.7%)
Patient using opioid analgesics, n (%)	58 (57.4%)
Day 3
Pain intensity at rest [NRS 0–10], median (IQR)	0 (0–1)
Pain intensity during movement [NRS 0–10], median (IQR)	3 (2–5)
Mobility [n (%) of patients using bathroom without help]	87 (86.1%)
Patient using opioid analgesics, n (%)	32 (31.7%)
Day 4
Pain intensity at rest [NRS 0–10], median (IQR)	0 (0–1.5)
Pain intensity during movement [NRS 0–10], median (IQR)	3 (2–5)
Mobility [n (%) of patients using bathroom without help]	93 (92.1%)
Patient using opioid analgesics, n (%)	25 (24.8%)
Day 5
Pain intensity at rest [NRS 0–10], median (IQR)	0 (0–1.5)
Pain intensity during movement [NRS 0–10], median (IQR)	3 (2–5)
Mobility [n (%) of patients using bathroom without help]	99 (98%)
Patient using opioid analgesics, n (%)	22 (21.8%)
Day 6
Pain intensity at rest [NRS 0–10], median (IQR)	0 (0–2)
Pain intensity during movement [NRS 0–10], median (IQR)	3 (2–4)
Mobility [n (%) of patients using bathroom without help]	99 (98%)
Patient using opioid analgesics, n (%)	22 (21.8%)
6 months after surgery
Oxford Hip Score, median (IQR)	40.5 (33–48)

IQR = interquartile range; ASA = American Society of Anesthesiologists; PACU = post-anaesthesia care unit; NRS = numeric rating scale.

Six months after surgery the Oxford Hip Score was 42 (34–46), representing a satisfactory result for of the majority of the patients. For 30 patients (29.7%), the outcome did not allow for an increase of OHS of at least 14 points compared to the preoperative score (Table 1) which can be considered an unsatisfactory functional result.²⁰

Logistic regression models were calculated at nine different time points to address the stepwise possible increase of information to predict an unsatisfactory result, defined as an OHS increase of less than 14 (Figs. S1 and S2). To compare the predictive performance of the models we calculated the area under the curve of the receiver operating characteristics (ROC-AUC) for each model (Table 3, Fig. 2). To compare the model fit on base of the log-likelihood the Nagelkerke r² of each model is provided in Table 2. For the preoperative model, 7 of the eligible variables were selected allowing for a ROC-AUC of 0.73 (95% confidence interval: 0.62–0.84). For the second model the 7 selected variables from the first model were included and the intraoperative variables were additionally eligible. However, no further variables were selected, meaning that none of the intraoperative variables were able to improve the model.

Table 3.

Receiver operating characteristics for the prediction of an unsatisfactory functional outcome using the logistic regression models of 9 perioperative time points.

Logistic regression model	ROC-AUC	95% CI
Preoperative model	0.729	0.62–0.837
Intraoperative model	0.729	0.62–0.837
Day 0 model	0.777	0.672–0.881
Day 1 model	0.788	0.684–0.892
Day 2 model	0.854	0.768–0.941
Day 3 model	0.867	0.786–0.948
Day 4 model	0.867	0.786–0.948
Day 5 model	0.883	0.813–0.953
Day 6 model	0.892	0.824–0.96

ROC-AUC = area under the curve of the receiver operating characteristic; CI = confidence interval.

Fig. 2.

Curves of the receiver operating characteristics to predict an unsatisfactory functional outcome using the logistic regression models at nine different time points.

Table 2.

Logistic regression models to predict a difference of less than 14 between the pre- and postoperative Oxford Hip Score.

4	Constant model 0		Preoperative model		Intraoperative model		Day 0 model		Day 1 model		Day 2 model		Day 3 model		Day 4 model		Day 5 model		Day 6 model
	B	Sig	B	Sig	B	Sig	B	Sig	B	Sig	B	Sig	B	Sig	B	Sig	B	Sig	B	Sig
Constant	−0.86	0.00	−7.35	0.01	−7.35	0.01	−7.35	0.01	−6.37	0.02	−12.17	0.00	−13.90	0.00	−13.90	0.00	−15.87	0.00	−17.09	0.00
Preoperative data
Age [years]			0.05	0.10	0.05	0.10	0.06	0.06	0.07	0.03	0.09	0.01	0.10	0.01	0.10	0.01	0.11	0.01	0.12	0.01
Sex [male = 1/female = 0]			−1.18	0.04	−1.18	0.04	−1.47	0.01	−1.55	0.01	−1.74	0.01	−1.45	0.05	−1.45	0.05	−1.52	0.04	−1.67	0.03
Daily nicotine [yes = 1, no = 0]			1.04	0.18	1.04	0.18	1.32	0.11	1.50	0.08	1.84	0.06	2.15	0.03	2.15	0.03	2.79	0.02	2.99	0.02
Daily alcohol [yes = 1, no = 0]			−2.44	0.06	−2.44	0.06	−2.42	0.06	−2.29	0.08	−2.70	0.05	−2.79	0.06	−2.79	0.06	−3.26	0.04	−3.26	0.05
Oxford Hip Score [0–48]			0.07	0.06	0.07	0.06	0.06	0.10	0.06	0.08	0.07	0.10	0.09	0.05	0.09	0.05	0.12	0.02	0.13	0.02
STAI trait anxiety [20–80]			0.09	0.02	0.09	0.02	0.09	0.03	0.08	0.05	0.10	0.03	0.09	0.06	0.09	0.06	0.10	0.05	0.10	0.04
PCS [0–52]			−0.06	0.08	−0.06	0.08	−0.07	0.07	−0.05	0.17	−0.08	0.07	−0.07	0.10	−0.07	0.10	−0.06	0.19	−0.06	0.23
Intraoperative data
none
Postoperative data
Day of surgery
Cumulative piritramide dose 1h after surgery [mg/kg]							6.03	0.14	5.79	0.16	7.89	0.08	8.49	0.08	8.49	0.08	6.52	0.19	5.54	0.29
Cumulative piritramide dose 24h after surgery [mg/kg]							−2.14	0.07	−1.55	0.19	−1.93	0.14	−3.16	0.04	−3.16	0.04	−2.61	0.09	−2.06	0.21
Day 1
Intake of analgesics [WHO 0-III]									−0.65	0.14	−0.99	0.07	−0.92	0.10	−0.92	0.10	−1.22	0.05	−1.29	0.04
Day 2
Pain intensity during movement [NRS 0–10]											0.43	0.00	0.37	0.02	0.37	0.02	0.39	0.02	0.41	0.02
Day 3
Pain intensity at rest [NRS 0–10]													0.58	0.05	0.58	0.05	0.94	0.02	0.97	0.01
Intake of analgesics [WHO 0-III]													0.52	0.19	0.52	0.19	0.71	0.09	0.88	0.05
Day 4
none
Day 5
Pain intensity at rest [NRS 0–10]																	−0.43	0.12	−0.23	0.49
Day 6
Pain intensity at rest [NRS 0–10]																			−0.42	0.17
Nagelkerke r2			0.247		0.247		0.300		0.324		0.462		0.513		0.513		0.535		0.554

B = regression coefficient; Sig = probability of no model improvement by including the variable.

For the next models the variables from day 0 to day 6 were successively eligible and entered the model on base of the probability to improve the model (Table 3).

Inclusion of the analgesic consumption and pain intensity on day 0 and day 1 improved the model fit but the predictive performance was not significantly higher than that of the preoperative model (Table 2). In contrast, including additionally the data of day 2 improved not only the model fit but also significantly improved the predictive performance to a ROC-AUC (95% CI) of 0.85 (0.77–0.94). The subsequent data of the days 3–6 further improved the model fit to a Nagelkerke r² of 0.554 on day 6 but the ROC-ACU was not significantly increased compared to day 2.

The receiver operating characteristics are shown in Fig. 2, illustrating the stepwise increase of the predictive performance. The whole logistic regression models and eligible variables, also the ones that did not enter the models, are shown in supplementary tables.

4. Discussion

In this study we investigated the perioperative predictability of unsatisfactory functional outcomes six months after hip arthroplasty. Our results show that a prediction based on pre-, intra- and early postoperative clinical characteristics allows to identify the majority of patients with unsatisfactory functional outcomes six months already during the first week after the surgery.

Looking at outcome predictions at different time points, in line with the results of previous works,^8,9 our prediction model that included only preoperative data allowed for a limited model fit (Nagelkerke r² = 0.25) and predictive performance (ROC-AUC = 0.73 (0.62–0.84). (Table 2, Table 3). Including intraoperative characteristics in the prediction model did not improve its performance. In contrast, the analgesic consumption and pain intensity of the first two days after surgery improved the predictability of an unsatisfactory outcome significantly (Table 3). The model fit of day 2 showed a clear increase (Nagelkerke r² = 0.46) and the predictive performance was significantly better than in the preoperative setting, as the ROC-AUC was 0.85 (0.77–0.94) (Table 2, Table 3). Hereafter, the clinical data of the days 3–6 allowed no further significant improvement of the predictive performance (Table 3).

4.1. Preoperative predictive factors

Of the non-modifiable characteristics the patient's sex contributed to the predictability as female sex was associated with an unsatisfactory result (Table 2). This is in line with previous results which showed that the extent of improvement might be negatively affected by female sex,⁴ although female patients achieve good absolute results.²¹ Of the characteristics that can be possibly modified by choosing an earlier time point for surgery the patient's age and the preoperative Oxford Hip Score (OHS) were included in the our preoperative prediction model. This corroborates the results of previous works,^21,22 marking a higher age as a risk factor for an unsatisfactory outcome. However, a limited preoperative function was not associated with an unsatisfactory outcome, although frequently reported.^11,23 In our study a higher preoperative OHS was associated with an unsatisfactory outcome. Thus, a limited preoperative function might predict a limited absolute outcome but not a limited improvement. In our opinion, the possible effects of age and preoperative function should be considered when the ideal time point of surgery is discussed with the patient.

As further potential outcome predictors, we applied a set of validated questionnaires to assess psychological factors, of which trait anxiety (STAI) and pain catastrophizing (PCS) were included in the prediction model. A higher trait anxiety score was associated with higher probability of an unsatisfactory result which is in line with previous results suggesting anxiety as a risk factor.²⁴ Reportedly, pain catastrophizing is also a risk factor for acute and persistent postoperative pain.^6,25 However, in our study a higher score of the Pain Catastrophizing Scale was associated with a higher probability of a favorable outcome. This apparent contradiction to published results resembles the discrepancy regarding female sex and limited preoperative function since mostly absolute outcome parameter have been used in previous studies.²⁵ Considering the results of our study and previous results,²⁶ pain catastrophizing does not seem to be a consistent risk factor for a limited functional improvement after hip arthroplasty. Another frequently reported psychological risk factor for an unsatisfactory result have been preoperative symptoms of depression.^27,28 However, in our study the depression score (HADS) was not included in our preoperative prediction model. Potentially, this resulted from the inclusion of daily nicotine consumption. Nicotine abuse is not only known to be a risk factor for prosthetic complications²⁹ but also more often observed in patients who suffer from depression.³⁰ In contrast, the daily consumption of alcohol was associated with a favorable outcome. Possibly, the share of patients with a preoperative osteonecrosis of the femoral head was higher in those who reported favorable outcomes. The abuse of alcohol is a main risk factor for osteonecrosis and patients with osteonecrosis report excellent results after hip arthroplasty.^31,32

4.2. Intraoperative predictive factors

In our study none of the intraoperative characteristics improved the prediction model (Table 2).

This is in line with previous works that showed that a longer operative time may have some adverse effects, especially raising the risk of a periprosthetic infection,³³ but is not associated with a worse functional outcome.³⁴ Further, the intraoperative rate of analgesics did not improve the predictability of unsatisfactory results. This seems surprising since we previously reported an association between the dosing of intraoperative analgesics and the pain level after six months in the same population.¹² This finding suggests that the functional improvement after hip arthroplasty does not strongly correlate with the occurrence of persistent pain.

4.3. Postoperative predictive factors

The cumulative dose of piritramide within the first hour and the cumulative dose of piritramide within the first 24 h after surgery improved our prediction models on day 0 and day 1. A lower intake of piritramide within 24 h was associated with an unsatisfactory outcome. In contrast, a lower intake of piritramide in the first hour was associated with a favorable outcome. This controversy might be explained by the analgesic effect of the long-acting opioids that were applied at the end of surgery. A high analgesic effect of the long-acting opioids probably results in a lower need for analgesics within the first hour and might be associated with a favorable long-term outcome.¹² After the effect of the intraoperative opioids had diminished the level of analgesia was mostly determined by the piritramide application via PCA pump which was reflected by the cumulative dose within 24 h after surgery. In summary, little analgesia within the first 24 h correlated with an unsatisfactory functional outcome.

The correlation of analgesia and the functional outcome is underlined by the prediction models of the following days which were improved by either including the patient's pain intensity or the analgesic intake. This is in line with the results of previous studies that have shown that early postoperative pain and too little analgesic consumption are risk factors for persistent pain and limited function.^10,35,36 However, based on our data it is impossible to differentiate whether the acute postoperative pain intensity leads to the unsatisfactory outcome or whether a confounding factor like a large intraoperative trauma leads to both the acute postoperative pain intensity and the unsatisfactory long-term outcome. Nonetheless, it is possible that an immediate and adequate analgesia helps avoid an unsatisfactory outcome after hip arthroplasty.

The daily reported grade of mobility was not included into any of the prediction models (Table 2). Thus, the grade of mobility did not provide further information for the predictability of unsatisfactory results, underlining that mobility is mostly a function of pain.¹¹ Therefore, an improved pain management will probably be more effective than forcing the grade of mobility to avoid an unsatisfactory outcome.

4.4. Limitations

There are several limitations of this study that have to be taken in consideration. First, the population size was determined by the primary investigation and therefore the study may be underpowered to investigate the influence of many possible risk factors on the outcome six months after surgery.¹³ Second, this study was designed to investigate the perioperative predictability of an unsatisfactory long-term outcome. It was not designed to identify single independent risk factors for such an outcome. Therefore, it is not appropriate to conclude that the predictive factors used in the regression models are independent risk factors for an unsatisfactory outcome. Further studies are needed to investigate the influence of single factors on the long-term outcome.

Furthermore, there was no standardized treatment protocol after the patients were discharged from hospital until the follow-up six months later. We did not collect any information about the treatment in this time interval. In addition, it remains unclear if our outcome parameter was ideal to assess the outcome after hip arthroplasty. There is controversial data about which outcome parameter indicates best a satisfactory or an unsatisfactory outcome. However, the here used OHS is widely accepted and also allows for a prediction of the subsequent outcome.^20,37,38 Whether an absolute postoperative OHS score or as we used the difference to the preoperative score is more meaningful is also controversial.^20,39 To us, it appears reasonable to use the difference between pre- and postoperative OHS since it might be more sensitive to detect intra- and postoperative risk factors.

4.5. Conclusion

In this study, we investigated the perioperative predictability of an unsatisfactory functional outcome six months after surgery. Our results show that a prediction based on pre-, intra- and early postoperative clinical characteristics allows to identify the majority of patients with unsatisfactory functional outcomes after six months already during the first days after surgery. Considering the individual parameters on which the predictions are performed, preoperative risk factors like higher age and anxiety might be addressed by a presurgical treatment or choosing an earlier time point for surgery. Of the postoperative parameters, the patient's pain intensity and analgesic consumption of the first two days after surgery stand out. A higher pain intensity and a lower analgesic consumption during the first postoperative days seem to correlate quite well with worse long-term functional outcomes. However, based on our data we cannot differentiate whether the postoperative pain is actually a risk factor and thus a cause of worse outcomes or whether the postoperative pain is caused by a confounder which induces both the postoperative pain and the worse functional outcomes. To our knowledge, this is the first study that shows a correlation between pain during the first postoperative days and the long-term functional outcome.

Funding

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

Declaration of competing interest

None of the authors has commercial associations that might pose a conflict of interest in connection with the submitted article.

Appendix A. Supplementary data

The following are the Supplementary data to this article:

figs1.

figs2.