Results
Baseline Data Set
Patients and Procedures The coding search identified 761 procedures. Forty-four were excluded, resulting in analysis of 361 THR and 356 TKR (Fig. 1). During the same period, our hospital reported 426 and 381 THR and TKR to the UK NJR. A total of 412 joint replacements were performed on female patients and 305 on males. Transfusion, admission, and discharge data were available for all patients in this study. Preoperative Hb results were successfully linked to operative records for 684 patients. The prevalence of preoperative anaemia was 24.3% (166/684; Table 1 and Table 2 ).
Relationship Between Baseline Variables and Outcome Preoperative anaemia and absolute preoperative Hb both independently predicted ABT while controlling for age, sex, operation site, and Hb loss. The ORs were 11.6 (6.2–25.7, P<0.001) for anaemia as a binary variable and 0.25 per 1 g dl increment (P<0.001) for absolute Hb, including in non-anaemic patients. Hb loss independently predicted ABT. The respective ORs were 2.66 (2.08–2.39, P<0.001) and 1.86 (1.55–2.23, P<0.001) when controlling for absolute preoperative Hb or yes/no for anaemia ( Table 3 ).
LOS was predicted by patient age (ES 0.13 day yr older, 95% CI 0.07–0.19 day, P<0.001) and preoperative Hb concentration (−0.7 day per 1 g dl increase in Hb, 95% CI −0.09 to −1.62 days, P=0.004; Table 4 ). ABT predicted all-cause hospital re-admission within 30 days (OR 2.94, 95% CI 1.27–6.77, P=0.01) after discharge ( Table 5 ).
Incorporating the individual surgeon in the regression analyses did not produce significantly better model fit (P>0.2 in all cases; ρ<0.01, where ρ is the proportion of the total variance of the data explained by surgeon identity).
Algorithm Implementation
We obtained initial tests prospectively on 173 consecutive patients for THR and 144 for TKR. Twenty-six patients were excluded after their care was transferred to an independent hospital and six were cancelled by their GP to allow investigation of their anaemia. Four were cancelled for other surgical or medical reasons. We therefore analysed data on 158 patients presenting for THR and 123 for TKR.
Seventy-three patients presented with anaemia. In total, 31 patients were treated with oral iron alone, 13 received parenteral iron (Ferric Carboxymaltose, 'Ferrinject', Vifor Pharma) and 22 received EPO (Recombinant Erythropoietin Alpha, 'Eprex', Janssen-Cilag) and iron. Fourteen non-anaemic women (whose Hb values were 12–13 g dl in combination with ferritin values suggesting functional iron deficiency and undergoing THR) were given iron in accordance with our agreed pathway. Twenty-one anaemic patients went untreated: 14 of these were for logistical reasons, for example, unavailability of an anaesthetist involved in the project for preoperative consultation; five men had only mild anaemia (Hb>12 g dl) without evidence of possible iron deficiency and were not offered treatment, in accordance with the algorithm; one patient declined because treatment would interfere with her planned holiday before surgery; and one because of needle phobia (Fig. 5). Twenty men and 46 women received treatment. Seven operations were re-scheduled for anaemia treatment. Using drug cost figures from our pharmacy, we calculated spendings of £15 070 and £1625 on EPO and parenteral iron, respectively (VAT excluded).
(Enlarge Image)
Figure 5.
Patients treated during programme implementation.
Before-and-After Comparisons
Patients and Procedures The post-implementation group was older on average than the baseline group; 74 (66–80) yr compared with 72 (65–78) yr (P=0.006). The two groups were similar in other respects ( Table 6 ). Nadir Hb (as determined by our database method) was the last pre-transfusion Hb in all 12 transfused patients in the post-implementation group. Hospital records indicate that intraoperative cell salvage was used for a total of 14 primary arthroplasties from 2008 to 2010; 11 from the baseline period and three while the programme was implemented. There were no major changes in the consultant surgical body during the period in which the algorithm was implemented. All our substantive orthopaedic surgeons had patients in both the before and after groups.
Process Measures: Preoperative Anaemia and Perioperative Hb Loss The anaemia prevalence within the post-implementation group decreased from 25.9% (73/281) at listing to 10.3% (29/281) after treatment (P<0.001), on an intention-to-treat basis. There was a significant inverse relationship between the baseline Hb and the Hb response to treatment (P=0.007; Fig. 6). Hb loss after THR decreased from 3.8 (2.9–4.9) g dl at baseline to 3.1 (2.6–4.0) g dl after algorithm implementation (P<0.001). For TKR, the Hb loss was 3.1 (1.9–4.6) g dl at baseline and 2.6 (2.0–3.3) g dl after algorithm implementation (P=0.003).
(Enlarge Image)
Figure 6.
Hb response in patients offered preoperative treatment. Scatter plot showing individuals' Hb concentration at surgical decision to list for arthroplasty, against the Hb change achieved with treatment.
Outcome Measures: Red Cell Transfusions, LOS, and Re-admissions Twelve of 158 (7.6%) patients received donor blood for THR after we implemented the algorithm, compared with 83 of 361 (23%) before (P<0.001). Our transfusion rate for TKR decreased from 6.7% (24/356) to 0% (0/123) (P=0.001). The total blood use decreased from 320 units for 717 procedures to 25 units for 281 procedures. A group of 281 patients from the baseline group would have needed [(320/717)×281] 126 units. The saving on blood, using figures obtained from our hospital blood bank, was thus [(126–25)×125] £12 625. VAT is not payable on blood.
The LOS for THR and TKR decreased from 6 (5–8) days to 5 (3–7) and 4 (3–6) days, respectively, after algorithm implementation (P<0.001). The all-cause re-admission rate to hospital within 30 days from discharge fell from 6.8% (49/717) to 4.3% (12/281) after algorithm implementation (P=0.14). The corresponding figures for re-admission within 90 days were 13.5% (97/717) before and 8.2% (23/281) after algorithm implementation (P=0.02).