Peri-operative mortality after hemiarthroplasty for fracture of the hip

1 downloads 0 Views 354KB Size Report
H. S. Gosal,. G. Holt. From Royal. Cornwall Hospital,. Truro, Cornwall,. United Kingdom ...... Emery RJ, Broughton NS, Desai K, Bulstrode CJ, Thomas TL. Bipolar ...
 HIP

Peri-operative mortality after hemiarthroplasty for fracture of the hip DOES CEMENT MAKE A DIFFERENCE? R. G. Middleton, C. E. Uzoigwe, P. S. Young, R. Smith, H. S. Gosal, G. Holt From Royal Cornwall Hospital, Truro, Cornwall, United Kingdom  R. G. Middleton, MBBS, Speciality Doctor Trauma and Orthopaedics Royal Cornwall Hospital, Truro, Cornwall TR1 3LJ, UK.  C. E. Uzoigwe, MRCS, Orthopaedic Surgeon, Speciality Registrar Trauma and Orthopaedics Leicester Royal Infirmary, Leicester, Leicestershire, LE1 5WW, UK.  P. S. Young, MRCS, Speciality Registrar Trauma and Orthopaedics Southern General Hospital, Glasgow, Lanarkshire G51 4TF, UK.  R. Smith, PhD, Senior Information Analyst Scottish Hip Fracture Audit, Edinburgh EH12 9EB, UK.  H. S. Gosal, FRCS(Orth), Consultant Trauma and Orthopaedics Cheltenham General Hospital, Cheltenham, Gloucsestershire GL53 7AN, UK.  G. Holt, FRCS(Orth), Consultant Trauma and Orthopaedics Crosshouse Hospital, Kilmarnock, East Ayrshire, KA2 0BE, UK. Correspondence should be sent to Mr R. G. Middleton; e-mail: [email protected] ©2014 The British Editorial Society of Bone & Joint Surgery doi:10.1302/0301-620X.96B9. 33935 $2.00 Bone Joint J 2014;96-B:1185–91. Received 8 February 2014; Accepted after revision 23 May 2014

We aimed to determine whether cemented hemiarthroplasty is associated with a higher post-operative mortality and rate of re-operation when compared with uncemented hemiarthroplasty. Data on 19 669 patients, who were treated with a hemiarthroplasty following a fracture of the hip in a nine-year period from 2002 to 2011, were extracted from NHS Scotland’s acute admission database (Scottish Morbidity Record, SMR01). We investigated the rate of mortality at day 0, 1, 7, 30, 120 and one-year post-operatively using 12 case-mix variables to determine the independent effect of the method of fixation. At day 0, those with a cemented hemiarthroplasty had a higher rate of mortality (p < 0.001) compared with those with an uncemented hemiarthroplasty, equivalent to one extra death per 424 procedures. By day one this had become one extra death per 338 procedures. Increasing age and the five-year co-morbidity score were noted as independent risk factors. By day seven, the cumulative rate of mortality was less for cemented hemiarthroplasty though this did not reach significance until day 120. The rate of re-operation was significantly higher for uncemented hemiarthroplasty. Despite adjusting for 12 confounding variables, these only accounted for 15% of the observed variability. The debate about the choice of the method of fixation for a hemiarthroplasty with respect to the rate of mortality or the risk of re-operation may be largely superfluous. Our results suggest that uncemented hemiarthroplasties may have a role to play in elderly patients with significant co-morbid disease. Cite this article: Bone Joint J 2014;96-B:1185–91.

Fracture of the hip is the most common serious orthopaedic injury to affect the elderly and is expected to increase in incidence during the next two decades.1 Displaced, intra-capsular fractures in the elderly are most commonly treated by hemiarthroplasty.2 Several national guidelines including the Scottish Intercollegiate Guideline Network (SIGN)3 and the National Institute for Health and Care Excellence (NICE)4 recommend the use of a cemented hemiarthroplasty in favour of an uncemented prosthesis. However, uncemented designs continue to be used, particularly in patients who have significant co-morbidities, due to the perceived mortality and morbidity associated with the use of polymethylmethacrylate (PMMA) cement.5-9 This issue remains controversial with conflicting outcomes reported in the literature.10-17 Few patients, however, die in the immediate post-operative period, and studies investigating this particular aspect are generally limited by small sample sizes. In this study we investigate the effect of the use of cement at the time of hemiarthroplasty on the rate of mortality and of re-operation.

VOL. 96-B, No. 9, SEPTEMBER 2014

Patients and Methods The source of the hospital data for this study was NHS Scotland’s Scottish Morbidity Record linked dataset (SMR), which records all inpatient and day case discharges from nonobstetric and non-psychiatric specialties in National Health Service hospitals in Scotland (SMR01). These are routinely combined with the registration of death to form continuous patient profiles. We selected any patient who had been admitted to hospital after sustaining a fracture of the hip (ICD codes S72.0, S72.1 or S72.2). The data were extracted in accordance with NHS Scotland’s Information Governance Standards,18 and analysed at an aggregated and non-identifying level that did not require specific ethical consent. Preliminary investigation of the data indicated that different designs of hemiarthroplasty are used in different geographical areas of Scotland. Some units use predominantly cemented hemiarthroplasties (perhaps reserving uncemented hemiarthroplasties for patients with limited life expectancy or significant medical co-morbidity).7,19 Conversely, 1185

1186

R. G. MIDDLETON, C. E. UZOIGWE, P. S. YOUNG, R. SMITH, H. S. GOSAL, G. HOLT

Table I. Multivariable regression analysis case-mix variables Variable

Criteria

Age Deprivation quintile Gender Residence

Divided into cohorts of < 50, 50 to 59, 60 to 69,70 to 79, 80 to 89 and 90+ 1 to 5 Male or female Own home, Institution (e.g. nursing home), Temporary residence, Transfer from other NHS Unit, Transfer within unit or Other S72.0, S72.1 or S72.2 diagnosis codes First specialty treating the patient (Surgical or Medical) Main condition (first listed ICD10 diagnosis category on SMR01) during hip fracture admission, grouped by HSMR for all admissions into 26 conditions, but the vast majority (94%) for this hip fracture analysis were ‘Trauma 1’ (including hip fractures). However, 6% had one of 25 other mainly medical categories as the ‘Main condition’, so these were grouped separately Excluding current hip fracture admission (0, 1, 2, 3, or 4+)

Type of hip fracture Admitting specialty Primary Diagnosis

Number of emergency admissions in previous year One year co-morbidity score

Co-morbidity score for conditions noted on admissions during the previous year (excluding current admission); Charlson Index Weightings (1 to 6) given to different conditions as modified by ISD Scotland’s HSMR work; multiple conditions are added together; see section on Prior Morbidity and Table B in ISD Scotland 2013 for further details Co-morbidity score for conditions noted on admissions during the previous five years (as above but added over five years) Jan to Mar, Apr to Jun, Jul to Sep, Oct to Dec 2002 to 2005, 2005 to 2008, 2008 to 2011

Five year co-morbidity score

Patients who had died (%)

Season Period

14 12 10 Hemi - uncemented

8 6

Hemi - cemented

4 Other recognised HF operation (not hemi)

2 0 0

5

10

15

20

25

30

35

40

45

50

Days after operation

Fig. 1 Cumulative mortality by operation type (HF, hip fracture).

other units use uncemented hemiarthroplasties for most fractures. An analysis of mortality is likely to be skewed in the favour of cemented hemiarthroplasties as these will tend to be performed on patients with relatively fewer medical comorbidities.7,19 We selected patients who underwent surgical treatment for a fracture of the hip after 1st April 2002. In order to control for confounding variables we formed a fiveyear pre-fracture co-‘morbidity variable’ as the International Classification of Diseases (ICD-10)20 diagnosis coding which had been incorporated into the SMR01 dataset on 1st April 1997. This variable was included in a multivariable logistic regression analysis. In order to adjust for potential change in technique over time, the data were divided into three cohorts of three years duration each, so that the last patients included were those who underwent surgery on 31st March 2011. The rate of mortality was calculated for each cohort from the day of surgery to 365 days (days 0, 1, 7, 30, 120 and one year). Patients with non-Scottish postcodes were excluded as it is unlikely NHS Scotland would have recorded postoperative deaths or re-admissions.

Statistical Analysis. A total of 12 case-mix variables were used to create a multivariable logistic regression analysis model. Of these, ten were derived directly from ISD Scotland’s Hospital Standardised Mortality Ratios (HSMR) methodology,21 but we also split the data into four annual quarters (Table I). All 12 case-mix variables were subsequently entered into logistic regression analyses on the rates of mortality at 0, 1, 7, 30, 120 days and one year. A separate analysis of the rates of mortality was performed between day 0 and day one, thus excluding patients who had died on day 0. Similar analyses were undertaken: between days one and day seven, excluding patients who had died by day one, between days seven and 30; between days 30 and 120 and between days 120 and one year. A p-value of 0.05 was used to determine statistical significance.

Results A total of 64 979 patients were identified who were coded as having sustained a fracture of the hip within the period of the study. Of this group, 22 387 (34%) had undergone hemiarthroplasty; of which 13 280 (59%) were cemented and 6389 (29%) were uncemented with the remaining 2718 (12%) not having a fixation technique specified. A further large group of 32 763 (50%) patients undergoing the three other most common operations for a fracture of the hip (Office of Population Censuses and Surveys (OPCS))22 codes (W19 - Primary open reduction of fracture of bone and intramedullary fixation, W24 - closed reduction of fracture of bone and internal fixation, and W37 – total prosthetic replacement of hip joint using cement) are shown in Figure 1 and Table II for comparative purposes. Those 2718 (4%) patients where cementing was not specified were excluded from further analysis, 1922 (3%) had THE BONE & JOINT JOURNAL

PERI-OPERATIVE MORTALITY AFTER HEMIARTHROPLASTY FOR FRACTURE OF THE HIP

1187

Table II. The numbers of patients by type of treatment or reason for exclusion Treatment

OPCS Code

Percentage (%)

Number

Cemented hemiarthroplasty Uncemented hemiarthroplasty Internal fixation Hemiarthroplasty – cement unknown Other ‘W’ operations Conservative treatment* Total

W461 W471 W19, W24, W37* W48 Other ‘W’ Nil or not ‘W’

20 10 50 4 3 12†

13280 6389 32763 2718 1922 7907 64979

* This also includes trochanteric fractures, fracture which did not require surgery and patients who were considered unfit for surgery † W37 = cemented total hip replacement (THR) (n = 2249, 3%). Other THR codes are included in Other ‘W’ operations

Table III. Unadjusted mortality rates by type of operation. Uncemented hemiarthroplasty

Cemented hemiarthroplasty

Time of death

Number of patients

Mortality number (%)

Number of patients

Mortality number (%)

χ21

p-value

On Day 0 By Day 1 By Day 7 By Day 30 By Day 120 By Day 365

6389 6389 6389 6389 6389 6389

11 (0.2) 39 (0.6) 180 (2.8) 567 (8.9) 1411 (22.1) 2141 (33.5)

13280 13280 13280 13280 13280 13280

51 (0.4) 112 (0.8) 323 (2.4) 937 (7.1) 2231 (16.8) 3579 (27.0)

6.2 3.1 2.6 20.2 79.8 90.0

0.013 0.08 0.11 < 0.001 < 0.001 < 0.001

28 (0.4) 141 (2.2) 387 (6.2) 844 (14.5) 730 (14.7)

13229 13168 12957 12343 11049

61 (0.5) 211 (1.6) 614 (4.7) 1294 (10.5) 1348 (12.2)

0.05 9.2 18.9 61.4 18.5

0.83 0.002 < 0.001 < 0.001 < 0.001

other coded operations not falling into any of the above categories, and 7907 (12%) were treated conservatively. The figure for conservative management was derived from routinely collected data and was an overestimate based on more detailed data collected by the Scottish Hip Fracture Audit prior to 2008.23 The unadjusted, cumulative rate of mortality was initially higher in the cemented cohort on days 0, 1, 2 and 3. The differences were, however, small with 0.2% of the uncemented cohort dying on day 0 versus 0.4% of the cemented cohort. This equates to an extra death on the day of surgery for every 250 patients treated. By day four, the unadjusted cumulative rate of mortality for uncemented hemiarthroplasty matched that for cemented hemiarthroplasty and exceeded it from that point on (Fig. 1, Table III). However, as uncemented fixation is often used in patients with significant co-morbidities, we repeated the analysis after assessing and adjusting for case-mix variables. Of the 12 case-mix variables which were examined (Table I), age was the only statistically significant univariate predictor of mortality on day 0. Even in the most susceptible group, those aged > 90 years, the overall rate of mortality on day 0 remained low (0.3%; Fig 2, Table IV). However, cemented hemiarthroplasties, when compared with their uncemented counterparts carried a significantly VOL. 96-B, No. 9, SEPTEMBER 2014

0.5 Patients (%)

On day 1 6378 Between Days 1 and 7 6350 Between Days 7 and 30 6209 Between Days 30 and 120 5822 Between Days 120 and 365 4978 Tests are univariate chi-square

0.4 0.3 0.2 0.1 0.0 < 60

60 to 69

70 to 79

80 to 89

90+

Age (yrs) Fig. 2 Plot graph showing mortality on day of operation in relation to age.

increased risk of mortality on day 0 (0.7% vs 0.2%, Table IV) in those aged > 90 years. After day 0, many other variables entered the models and the comparison of the rate of mortality in the cemented and uncemented hemiarthroplasties was carried out by comparing their effect on the best-fit models after all case-mix variables were entered (Table IV). Although the unadjusted rate of mortality increased to the same level in the uncemented and cemented cohorts

1188

R. G. MIDDLETON, C. E. UZOIGWE, P. S. YOUNG, R. SMITH, H. S. GOSAL, G. HOLT

Table IV. Mortality in patients aged > 90 years on day 0 versus operation (p = 0.005) Survival status day 0 Operation

Survived

Died

Total

Uncemented hemiarthroplasty (%) Cemented hemiarthroplasty (%) Other types of operation (not hemiarthroplasty) (%) Total (%)

1202 (98.8) 1749 (99.3) 4939 (99.8) 7890 (99.7)

3 (0.2) 13 (0.7) 11 (0.2) 27 (0.3)

1250 (100) 1762 (100) 4950 (100) 7917 (100)

Table V. Adjusted rates of mortality; logistic regression models Significance of type of operation when added to full model*

Relative risk* uncemented vs cemented (95% CI)

p* comparison of cemented vs uncemented

< 0.001 < 0.001 < 0.001

0.41 (0.21 to 0.78) 0.63 (0.43 to 0.91) 1.02 (0.84 to 1.23)

0.008 0.014 0.85

0.01 < 0.001 0.002

1.08 (0.97 to 1.21) 1.17 (1.08 to1.26) 1.13 (1.06 to 1.21)

0.15 < 0.001 0.001

Age, 5-year co-morbidity, 1-year no. of emergencies < 0.001 0.02 Age, gender, admitted from, primary diagnosis, 5year co-morbidity, Surg/Med, type of fracture, deprivation All except Surg/Med and deprivation 0.26

0.80 (0.51 to 1.26) 1.22 ( 0.98 to 1.52)

0.34 0.07

1.12 (0.98 to 1.28)

0.10

All except type of fracture

0.001

1.20 (1.09 to 1.32)

< 0.001

All except type of fracture

0.37

1.06 (0.96 to 1.18)

0.24

Significant case-mix variables Death on day 0 Death by day 1 Death by day 7

Death by day 30 Death by day 120 Death by day 365 Death on day 1 Death between day 1 and day 7 Death between day 7 and day 30 Death between day 30 and day 120 Death between day 120 and day 365

Age, period Enter base model at p< 0.05 Age, 5year co-morbidity Age, gender, admitted from, primary diagnosis, 5year co-morbidity, Admitting Speciality, type of fracture All except Surg/Med and deprivation All All except season

* Relative risks calculated from odds ratios in logistic regression models when adding type of operation to the model after all 12 other case-mix variables were force entered

within four days (Fig. 2), case-mix adjusted rates became similar sometime just before day seven (Relative Risk was 1.02 on day 7). However, the cumulative difference did not become significant until sometime after day 30 and before day 120 (Table V). The case-mix adjusted rate of mortality was significantly lower in the uncemented compared with the cemented cohort on day 0 and up to day one. However, the patients with a cemented hemiarthroplasty had a lower rate of mortality on day one (i. e., excluding those who had died on day 0), though not significantly so. This trend was also observed for mortality between days one and seven, as well as between days seven and 30 (Table V). It was only for death between days 30 and 120 that the differences reached statistical significance (Table V). Thereafter, having excluded those who had already died, no difference in mortality rate was found, though cumulative mortality was significantly higher at days 120 and 365 in patients who had received an uncemented implant. As re-operation and re-admission are also likely to be associated with an immediately higher rate of mortality, we removed this potential confounding factor (already reported

in Table IV) by only looking at rates of re-operation and re-admission for patients who were alive one year after operation. Of those alive, a higher proportion of those with an uncemented than cemented hemiarthroplasty had undergone further surgery (Table VI, p = 0.005, relative risk 1.28 after adjusting for case-mix in a logistic regression model). There was no overall difference between the two groups in the rate of re-admission during the year.

Discussion We found that cemented hemiarthroplasty was associated with a small but significant higher cumulative rate of mortality than uncemented hemiarthroplasty on day 0 and up to day one. The difference persisted after adjustment was made for potential confounders. The risk of mortality on day 0 was 2.3 times higher for patients who received a cemented hemiarthroplasty than for those who received an uncemented hemiarthroplasty (p = 0.008). The adjusted risk of mortality by day one was 1.7 times higher for patients receiving an uncemented hemiarthroplasty than those receiving a cemented hemiarthroplasty (p = 0.014). THE BONE & JOINT JOURNAL

PERI-OPERATIVE MORTALITY AFTER HEMIARTHROPLASTY FOR FRACTURE OF THE HIP

1189

Table VI. Difference in rate of re-operation and re-admission at one year Operation Uncemented Cemented p†

Total number of patients

Number (%) alive at one year post-operation

If alive, number (%) that had a re-operation*

If alive, number (%) that had been re-admitted*

6389 13280

4248 (66) 9701 (73)

227(5.3) 430 (4.4) 0.005‡

1616 (38) 3647 (38) 0.51

* Re-admissions are for any cause, re-operations were for re-operations on the hip following the original surgery † As with previous analyses we have adjusted for case-mix by providing the p associated with model improvement after type of operation (uncemented or cemented hemiarthroplasty) was added to a logistic regression model already including all the other measured case-mix factors ‡ Relative risk of re-operation within 365 days in uncemented hemiarthroplasty versus cemented = 1.28 (1.08 to 1.52 CI, calculated from Odds Ratios in logistic regression model adding Type of Operation to the model after all 12 other case-mix variables were force entered)

However, no significant difference was noted between the methods of fixation for mortality on day one, (excluding those who had died on day 0). The 30-day rate of mortality was similar for both groups. However, those undergoing uncemented hemiarthroplasty had a higher rate of mortality at 120 days and one year (relative risk 1.17 and 1.13, respectively), with a higher rate of re-operation (relative risk 1.28) if they were alive at one year following surgery. Costain et al7 performed an analysis of 25 000 patients from the Australian National Joint Registry and reported that patients undergoing a cemented hemiarthroplasty had a statistically significant higher rate of mortality at day one compared with those undergoing an uncemented hemiarthroplasty. Unfortunately, due to limitations in their registry data, they were only able to adjust for age and gender. In a sub group analysis involving cemented versus uncemented monoblock hemiarthroplasty they found that the relative risk of mortality at day one was 1.7 times higher in the cemented cohort. This figure is identical to our findings. At day seven and day 30, the trend had reversed and mortality for uncemented monobloc hemiarthroplasty was higher, becoming statistically significant again at one year following surgery. They also found that patients aged > 80 years were at an increased risk of mortality at day one with cemented fixation. Talsnes et al19 looked at 11 210 patients from the Norwegian Registry. They corrected for five confounding variables including the American Society of Anesthesiologists (ASA)24 score, cognitive impairment, age, time to surgery and gender. They looked specifically at death by day one, which was 2.3 times higher in those receiving a cemented hemiarthroplasty with a number needed to harm (NNH) of 116 operated per death. Again, similar to both ours and Costain’s7 study, Talsnes et al19 found increasing age to be a significant risk factor. They used the ASA score as a surrogate of comorbidity. Mirroring our findings with respect to the five-year co-morbidity score, they found the ASA score to be a significant risk factor with the NNH increasing from 811 for ASA grades 1 and 2 to 86 for ASA 3 and 33 for ASA 4 and 5. However, they found no difference in the rate of mortality between day two to the end of follow-up, though all the other variables were found to be independent and significant risk factors for mortality. In contrast to the two previously mentioned studies, the SMR database allowed us to perform a more extensive VOL. 96-B, No. 9, SEPTEMBER 2014

adjustment for potential confounding variables, using a total of 12. We believe this is important to allow for the possibility of surgeons selecting or deselecting the use of cement on the basis of patient factors, which may act as surrogate markers for mortality. The fact that almost all the case-mix variables enter the later models suggests that the case-mix variables are appropriate. The NICE Guidance advocates the use of cement for all patients receiving a hemiarthroplasty.4 This followed a meta-analysis that found no difference in the rate of mortality between cemented and uncemented hemiarthroplasties. The 2010 Cochrane review came to the same conclusion.10 Unfortunately, both of these are limited by small numbers and both preceded this study and the findings of both Costain et al7 and Talsnes et al.19 One other study of note is that of Costa et al6 who audited 16 496 patients from the England and Wales national hip fracture database in 2011. Having adjusted for potential confounding variables, they found that cement was associated with a small but significant survival benefit (Odds Ratio 0.83) with respect to death at the time of discharge from hospital. However, they did not analyse death within the immediate post-operative period. The Scottish SIGN guidance makes no global appraisal of the data but notes that cemented arthroplasty is associated with peri-operative morbidity.3 The guidance is more cautious and less didatic than NICE and makes the recommendation that “Cement should be used when undertaking hemiarthroplasty, unless there are cardiorespiratory complications, particularly in frail older patients.”3 The evidence from our study, in conjunction with that from the Norwegian19 and Australian7 Registries, lends support to this position. We found the use of cement, increasing age and five-year co-morbidity score to be independent predictors of mortality by the first post-operative day. However, the excess mortality attributable to cement was low. Using the relative risk of 0.63 for mortality by the first postoperative day and the absolute risk for the use of cement of 0.8% for mortality within the same time frame, we calculate a NNH of 340. For those aged > 90 years the NNH falls to 240. Talsnes et al19 reported an even lower NNH figure of 116 for mortality by the first operative day with the use of cement. This falls to 33 for those with an ASA grade > 3. Given our findings and those of Talsnes et al19 and Costain et al,7 there may be some merit in a guideline based on the

1190

R. G. MIDDLETON, C. E. UZOIGWE, P. S. YOUNG, R. SMITH, H. S. GOSAL, G. HOLT

stratification of risk as proposed by SIGN rather than the single unqualified recommendation of NICE. However, the situation may be more complex. Studies published so far have failed to report coefficients of correlation (r-values) within the regression models. The r value is a measure of how well the variation of one variable explains the variation of the other, where 1.0 is a perfect correlation and -1.0 is a perfect negative correlation. Despite adjustment for 12 confounding variables, analysis of our case-mix variables shows that they only explain a relatively small amount of the variation in the rates of mortality; for example, up to 15% for rate of mortality at 120 days. Thus, many variables are ignored by both this and other studies which undoubtedly play a role. For example, it has been suggested that surgical techniques such as rigorous lavage and the use of low viscosity cement may be protective.9,25,26 However, the magnitude of these effects and whether they translate into a reduction in rate of mortality has yet to be determined. As in the NICE meta-analysis and the Cochrane review, we observed no difference in the rate of mortality at 30 days between recipients of cemented and uncemented hemiarthroplasties. However, and in contrast, we did observe an improved 120-day and one-year survival for patients with cemented hemiarthroplasties. It is difficult to explain these findings conclusively. It is unlikely that cement had a protective effect and such results help to emphasise the role of the myriad unknown variables highlighted by the r value analysis. Indeed, such results may serve to highlight the innate ability of the operating surgeon to predict mortality and choose the method of fixation accordingly. Echoing the findings of Jameson et al27 analysis of the Hospital Episodes Statistics (HES) data, but in contrast to the NICE and Cochrane meta-analyses, our study shows considerable difference in the rate of re-operation at one year. The figure is much higher for those with an uncemented hemiarthroplasty (5.3% vs 4.4%). The calculated NNH is 110. However, the re-operation data are crude and do not code for the side of surgery or reason for surgery. For example, re-operation for infection rather than pain is more likely to reflect frailty of the patients rather than failure of fixation. Although not recorded specifically within the SMR data, we know from the SHFA that almost all the uncemented hemiarthroplasties used are of the older AustinMoore monobloc design. Modern uncemented designs have been reported to offer superior functional outcomes, equal to their cemented counterparts.16 If such implants were used, a lower rate of re-operation might have been recorded. Indeed, an improved functional outcome might conceivably allow greater mobility and hence better prevention of attendant morbidity and associated mortality. We acknowledge the limitations of our study. It was retrospective raising the possibility of some episodes being missed. However, the sample size was large and we are confident that it is representative of the population of patients with a fracture of the hip. Much information was obtained

by coding. The accuracy of this system has been called into question.28 However, most of the weaknesses in coding operations reflect the complexity of the procedure rather than the recording of co-morbidity.29 Although we used a number of variables to determine co-morbidity, it is possible that elements of co-morbid disease were missed. The abbreviated mental score30 is a strong predictor of mortality and was not adjusted for in the regression analysis. However, our extensive analysis of co-factors may have included surrogate markers of cognitive ability including the source of admission and one and five year co-morbidity scores. Although there is evidence that delay to surgery increases the rate of mortality,31 a previous analysis of the SMR01 dataset32 concluded that this was not a variable of significance and so it was not included in our analysis. Other factors such as the serum lactate33 and haemoglobin34 levels on admission have also been shown to be associated with survival. However, the SMR01 does not record these. Immediate peri-operative mortality was found to be higher in patients with a fracture of the hip who undergo cemented rather than uncemented hemiarthroplasty. This difference persisted even after adjustment for confounding variables. Increasing age and co-morbidity were found to be risk factors for an increased rate of mortality by day one after surgery. However, no differences between the methods of fixation were noted after day one and those treated with a cemented hemiarthroplasty had a higher 120-day and oneyear survivorship, with a lower rate of revision at one year. Our results highlight the fact that, that despite adjustment for 12 confounding factors, these only explained 15% of the observed variation and the situation is more complex than a simple cement/no-cement dichotomy. Based on our analysis of the SMR01 database, we concur with the SIGN guidance that the use of uncemented hemiarthroplasty in the elderly patient with significant co-morbid disease is appropriate. No benefits in any form have been received or will be received from a commercial party related directly or indirectly to the subject of this article. This article was primary edited by G. Scott and first proof edited by J. Scott.

References 1. Holt G, Smith R, Duncan K, Hutchison JD, Reid D. Changes in population demographics and the future incidence of hip fracture. Injury 2009;40:722–726. 2. Parker M, Johansen A. Hip fracture. BMJ 2006;333:27–30. 3. No authors listed. Scottish Intercollegiate Guidelines Network (SIGN). Guideline No. 111. 2009. http://www.sign.ac.uk/guidelines/fulltext/111/ (date last accessed 27 May 2014). 4. No authors listed. National Institute for Health and Care Excellence (NICE). Guideline CG124 Hip Fracture: the management of hip fracture in adults, 2011. http://publications.nice.org.uk/hip-fracture-cg124 (date last accessed 27 May 2013). 5. No authors listed. National Hip Fracture Database Annual report 2013: http:// www.nhfd.co.uk/003/hipfractureR.nsf/4e9601565a8ebbaa802579ea0035b25d/ 566c6709d04a865780257bdb00591cda/$FILE/onlineNHFDreport.pdf (date last accessed 27 May 2013). 6. Costa ML, Griffin XL, Pendleton N, Pearson M, Parsons N. Does cementing the femoral component increase the risk of peri-operative mortality for patients having replacement surgery for a fracture of the neck of femur? Data from the National Hip Fracture Database. J Bone Joint Surg [Br] 2011;93-B:1405–1410. 7. Costain DJ, Whitehouse SL, Pratt NL, et al. Perioperative mortality after hemiarthroplasty related to fixation method. Acta Orthop 2011;82:275–281. THE BONE & JOINT JOURNAL

PERI-OPERATIVE MORTALITY AFTER HEMIARTHROPLASTY FOR FRACTURE OF THE HIP

8. Donaldson AJ, Thomson HE, Harper NJ, Kenny NW. Bone cement implantation syndrome. Br J Anaesth 2009;102:12–22. 9. Timperley AJ, Whitehouse SL. Mitigating surgical risk in patients undergoing hip arthroplasty for fractures of the proximal femur. J Bone Joint Surg [Br] 2009;91B:851–854. 10. Parker MJ, Gurusamy K. Arthroplasties (with and without bone cement) for proximal femoral fractures in adults. Cochrane Database Syst Rev 2010;6:CD001706. 11. Faraj AA, Branfoot T. Cemented versus uncemented Thompson's prostheses: a functional outcome study. Injury 1999;30:671–675. 12. Emery RJ, Broughton NS, Desai K, Bulstrode CJ, Thomas TL. Bipolar hemiarthroplasty for subcapital fracture of the femoral neck: a prospective randomised trial of cemented Thompson and uncemented Moore stems. J Bone Joint Surg [Br] 1991;73-B:322–324. 13. Azegami S, Gurusamy KS, Parker MJ. Cemented versus uncemented hemiarthroplasty for hip fractures: a systematic review of randomised controlled trials. Hip Int 2011;21:509–517. 14. Parker MJ, Pryor G, Gurusamy K. Cemented versus uncemented hemiarthroplasty for intracapsular hip fractures: a randomised controlled trial in 400 patients. J Bone Joint Surg [Br] 2010;92-B:116–122. 15. Taylor F, Wright M, Zhu M. Hemiarthroplasty of the hip with and without cement: a randomized clinical trial. J Bone Joint Surg [Am] 2012;94-A:577–583. 16. Figved W, Opland V, Frihagen F, et al. Cemented versus uncemented hemiarthroplasty for displaced femoral neck fractures. Clin Orthop Relat Res 2009;467:2426– 2435. 17. Khan RJ, MacDowell A, Crossman P, Keene GS. Cemented or uncemented hemiarthroplasty for displaced intracapsular fractures of the hip: a systematic review. Injury 2002;33:13–17. 18. No authors listed. Information Services Division. Research Governance Framework V2.0 July 2013 http://www.isdscotland.org (date last assessed 4 July 2914). 19. Talsnes O, Vinje T, Gjertsen JE, et al. Perioperative mortality in hip fracture patients treated with cemented and uncemented hemiprosthesis: a register study of 11,210 patients. Int Orthop 2013;37:1135–1140. 20. No authors listed International Statistical Classification of Diseases and Related Health Problems, 10th Revision. World Health Organization 2010. http:// www.who.int/classifications/icd/ICD10Volume2_en_2010.pdf (date last accessed 02 July 2014).

VOL. 96-B, No. 9, SEPTEMBER 2014

1191

21. No authors listed. National Services Scotland: Hospital Standardised Mortality Ratios. http://www.isdscotland.org/Health-Topics/Quality-Indicators/Publications/ 2013-02-26/2013-02-26-HSMR-Report.pdf (date last accessed 27 May 2013). 22. No authors listed. OPCS Classification of Interventions and Procedures Version 4.5. 2009. London, TSO. http://www.hpa.org.uk/webc/HPAwebFile/HPAweb_C/ 1215501690196 (date last accessed 02 July 2014). 23. No authors listed. Scottish Hip Fracture Audit. http://www.shfa.scot.nhs.uk/Annual Report/SHFA_Report_2007.pdf (date last accessed 2 July 2014). 24. Owens WD, Felts JA, Spitznagel EL Jr. ASA physical status classifications: a study of consistency of ratings. Anesthesiology 1978;49:239–243. 25. Christie J, Robinson C, Singer B, Ray D. Medullary lavage reduces embolic phenomena and cardiopulmonary changes during cemented hemiarthroplasty. J Bone Joint Surg [Br] 1995;77-B:456–459. 26. Rothberg DL, Kubiak EN, Peters CL, Randall RL, Aoki SK. Reducing the risk of bone cement implantation syndrome during femoral arthroplasty. Orthopedics 2013;36:463–467. 27. Jameson SS, Jensen CD, Elson DW, et al. Cemented versus cementless hemiarthroplasty for intracapsular neck of femur fracture: a comparison of 60,848 matched patients using national data. Injury 2013;44:730–734. 28. Dalal S, Roy B. Reliability of clinical coding of hip fracture surgery: implications for payment by results? Injury 2009;40:738–741. 29. 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. 30. Hodkinson HM. Evaluation of a mental test score for assessment of mental impairment in the elderly. Age Ageing 1972;1:233–238. 31. Uzoigwe CE, Burnand HG, Cheesman CL, et al. Early and ultra-early surgery in hip fracture patients improves survival. Injury 2013;44:726–729. 32. Holt G, Smith R, Duncan K, Finlayson DF, Gregori A. Early mortality after surgical fixation of hip fractures in the elderly: an analysis of data from the scottish hip fracture audit. J Bone Joint Surg [Br] 2008;90-B:1357–1363. 33. Uzoigwe CE, Venkatesan M, Smith R, et al. Serum lactate is a prognostic indicator in patients with hip fracture. Hip Int 2012;22:580–584. 34. Maxwell MJ1, Moran CG, Moppett IK. Development and validation of a preoperative scoring system to predict 30 day mortality in patients undergoing hip fracture surgery. Br J Anaesth 2008;101:511–517.