2.1 Search strategy

On August 17, 2023, the PubMed, MEDLINE, EBSCOhost, and Google Scholar databases were queried to identify all studies published between January 1, 2000 and August 17, 2023 that evaluated the utility of routine postoperative blood testing following TJA. For our query, the following search term was utilized: ("Arthroplasty, Replacement, Hip"[Mesh] OR "Arthroplasty, Replacement, Knee"[Mesh] OR "Arthroplasty, Replacement"[Mesh] OR "total joint arthroplasty" OR "total knee arthroplasty" OR "total hip arthroplasty" OR "TJA'' OR "THA" OR "TKA") AND ("Hematologic Tests"[Mesh] OR "Blood Cell Count"[Mesh] OR "complete blood count" OR "CBC" OR "routine blood test∗" OR "Diagnostic Tests, Routine"[Mesh] OR "laboratory test∗" OR "postoperative test∗" OR "Value-Based Health Care"[Mesh]).

2.2 Eligibility criteria

For article inclusion, a full-text English manuscript must have been available, and the study must have investigated the utility of postoperative laboratory testing in TJA or assessed approaches to targeted testing. Articles were excluded for not fulfilling these inclusion criteria as well as all case reports, systematic reviews, duplicate studies among databases, abstracts, articles on pre-print servers, and non-English publications.

2.3 Study selection

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines (PROSPERO registration of the study protocol: CRD42023437334, June 7, 2023) were followed for this investigation. To assess article eligibility, two independent reviewers (JRP, MSL) made initial inclusions and exclusions, with a third reviewer (CJH) helping attain consensus in the event of disagreements. 1189 unique articles were attained from the initial query after removing duplicates. 67 studies were eligible for a full-text screen following title and abstract screening. 19 studies met inclusion criteria after examining the full text. Additional review of references of included studies did not yield further articles (Fig. 1).

Close

Fig. 1 This PRISMA diagram depicts the selection process for article information.

Export to PPT

2.4 Study characteristics

A total of 19 studies evaluating 34,166 TJA procedures (TKA: n = 23,066, 67.5 %; THA: n = 11,100, 32.5 %) were included in the final analysis (Table 1). Eight studies investigated the utility of both postoperative CBCs and BMP/CMPs,^{9–11,14,16–19} nine studies investigated only postoperative CBCs,^{12,13,20–26} and two studies investigated only postoperative BMPs.^27,28 Nine studies reported results for both THA and TKA,^{16,18,20–22,24,26–28} while five studies reported results for only THA^{9,11,12,17,19} and five studies reported results for only TKA.^{10,13,14,23,25} The included studies were published between 2015 and 2023, with all being retrospective in design. All studies were conducted at single institutions analyzing patient data between January 2008 and December 2020 (see Table 2).

2.5 Risk of bias in individual studies

The Methodological Index for Nonrandomized Studies (MINORS) tool was utilized to assess for risk of bias in this investigation 33. This previously validated tool rates comparative studies on a scale from 0 to 24 with higher scores representing a greater quality study. 12 criteria related to study design, outcomes, and follow-up were rated 0 if not reported, 1 if reported but inadequate, and 2 if reported and adequate by two reviewers (JRP, MSL). Scoring differences were resolved with consensus from a third reviewer (CJH). The mean MINORS score was 18.2 ± 1.5.

2.6 Data extraction and analysis

The selected studies were analyzed for their study design, types of routine postoperative laboratory tests (CBCs, CMPs/BMPs), and outcomes associated with these tests. To assess the utility of postoperative labs, we gathered data on the frequency of abnormal preoperative and postoperative lab values, clinical interventions, and complications. Additionally, we examined predictors of abnormal test results or associated clinical interventions, as well as cost information for routine testing. Data extraction was conducted by two reviewers (JRP, MSL) who created a collaborative spreadsheet in duplicate. The results were compared for accuracy. Due to the variability in study designs, a meta-analysis was not feasible, so a narrative summary of the findings was presented.

3.1 The utility of CBCs and BMPs/CMPs

Seven studies reported frequency of preoperative anemia, with an average of 28.9 % (4.1–33.3 %).^{9–11,16,20,25,26} Nine studies reported overall frequency of postoperative anemia, with an average of 46.8 % (5.9–90.0 %) across studies.^{9,12,13,16,20,22,23,25,26} One remaining study showed that of patients who had at least one abnormal lab value (CBC or CMP), 95.8 % of them had postoperative anemia.¹⁴ All seventeen studies reported the incidence of postoperative transfusion, with an average frequency of 4.2 % (0.1–28.0 %) across studies. Four studies reported rates of 90-day complications. Two of these studies assessed a broad scope of postoperative complications and reported a mean complication rate of 4.3 % (3.1–5.2 %).^12,23 The remaining two studies specifically evaluated the frequency of postoperative acute kidney injury (AKI), with an average frequency of 2.8 % (2.0–3.8 %) between studies.^10,11

Two studies reported the frequency of preoperative electrolyte abnormalities (sodium and potassium) and showed an average frequency of 9 % (7.1–9.0 %).^19,28 All nine studies reported on the frequency of abnormal postoperative electrolyte values, with all but one reporting both sodium and potassium abnormalities. Across those eight studies, the average combined frequency of postoperative sodium and potassium abnormalities was 28.6 % (6.8–66.5 %).^{9–11,14,16,19,27,28} The one remaining study evaluated potassium values only and found that 15.5 % of patients experienced abnormal lab results.¹⁸ Eight studies reported on electrolyte-related postoperative clinical interventions in the form of electrolyte supplementations, with seven of them including both sodium and potassium-related intervention. Those seven studies showed that on average, 7.0 % (0.7–14.2 %) of patients required clinical intervention due to abnormal electrolytes.^{3,9,11,14,16,27,28} The one remaining study strictly assessed potassium-related interventions and reported that 15.5 % of all patients required postoperative clinical intervention, which meant all patients with abnormal potassium values received intervention.¹⁸

3.2 Predictors of abnormal postoperative laboratory tests and clinical intervention

Eight studies identified predictors of postoperative anemia and subsequent transfusion (Table 4).^{10,11,16–19,24,25} The most frequently noted predictors of abnormal postoperative CBCs and transfusions were preoperative Hb (13 studies, n = 26,722), increasing age (8, n = 18,444), lack of TXA use (6, n = 16,657), female sex (6, n = 16,928), low BMI (5, n = 17,031), operative time (5, n = 16,815), ASA score (4, n = 15,114), intraoperative blood loss (3, n = 2196), diabetes mellitus (2, n = 670), chronic kidney disease (2, n = 1277), and postoperative day one (POD1) Hb (1, n = 108) (see Table 3). One study found that there was no difference mean drop in Hb between anterior and posterior approaches to hip (2.12 g/dL vs. 2.24 g/dL, p = 0.59).¹⁷ Another study found no difference between the direct anterior approach and the Rottinger approach to the hip in need for transfusion (p = 0.77).⁹ Nine studies which identified predictors developed predictive models for transfusion, either using an individual factor or a multitude.^{9,12–14,19,20,23,25,26} Preoperative Hb and POD1 Hb were excellent discriminators of transfusion with maximum area under the curves (AUCs) of 0.933 and 0.943, respectively.¹⁴ However, preoperative Hb outperformed POD1 Hb in terms of specificity.²⁶ The transfusion models with multiple predictors for decision-making were able to achieve sensitivities of 64–96.5 %, specificities of 40–87 %, and AUCs of 0.802–0.92 (see Table 5).

Nine studies identified predictors of abnormal postoperative BMP/CMPs as well as subsequent fluid boluses and electrolyte infusions.^{9–11,14,16,18,19,27,28} The most frequently noted predictors of abnormal postoperative CBCs and transfusions were preoperative metabolite levels (7 studies, n = 6839), diabetes mellitus (3, n = 4887), sex (2, n = 1162), CKD (2, n = 1899), and operative time (2, n = 1108). Additionally, one study found Black race to be associated with greater risk for abnormal postoperative potassium (Odds Ratio [OR]: 1.38, 95 % Confidence Interval [CI]: 1.05–1.82) and blood urea nitrogen (OR: 1.85, 95 % CI: 1.04–3.27) following TJA in comparison to White race.²⁸ One study which identified predictors was able to test the predictive power of risk factors for albumin and potassium supplementation.¹⁴ Blood loss and preoperative potassium achieved the highest sensitivities, specificities, and AUCs for albumin and potassium supplementation, respectively.

3.3 Costs associated with postoperative laboratory testing

Studies reported between 1.74 and 3.3 postoperative labs administered per patient (Table 6).^{12,21,23,26,27} However, there was a difference in the number of tests for patients not requiring intervention (2.8–2.9 tests) compared to those who did (5.6–5.7 tests).^12,23,27 Furthermore, two studies found that a large proportion (92–97.21 %) of postoperative testing was performed on patients with normal preoperative labs.^21,26 Two studies estimated that 56–97 % of postoperative testing costs could have been saved if testing was limited to patients with abnormal preoperative values.^25,26 Additionally, patient selection algorithms utilizing other predictors of clinical intervention were estimated to create substantial savings for institutions.^9,17 Approximately $300 per patient was spent on CBC testing for patients who did not receive a transfusion^12,23 and approximately $850 per patient was spent on BMPs for patients who did not receive medical intervention.²⁷

4.1 The utility of CBCs and BMP/CMPs

This review found that routine postoperative lab testing holds little clinical utility for most patients undergoing TJA, which is supported by high rates of abnormal postoperative labs and coinciding low rates of clinical intervention across the analyzed studies. On average, the best available evidence showed that nearly half of patients experience an abnormal CBC postoperatively; however, only a fraction of these patients, typically <5 %, require transfusion. These low rates of transfusion and associated complications are partly attributable to the advent of tranexamic acid, with studies reporting transfusion rates of 0–8.1 % for TJA procedures over the past decade.^29–31 Similarly, recent studies assessing the utility of postoperative CBCs and BMPs in partial knee, partial hip, and shoulder arthroplasty concluded that the labs were frequently unnecessary and not cost-effective.^8,32–35 As advances in perioperative management continue to allow for fewer postoperative complications, these tests will continue to become increasingly inconsequential for the average patient. However, as this review showed that roughly 4 % of patients still require blood or metabolite-associated clinical intervention postoperatively, there is still value in laboratory testing for a small subset of patients. Thus, a selective approach to testing is an important step toward removing the undue burden of testing for patients and the overall healthcare system alike.

4.2 Predictors of abnormal postoperative laboratory tests and clinical intervention

Among the identified predictors for transfusion requirement such as age, sex, BMI, operative time, and ASA score, abnormal preoperative lab values were shown to offer the greatest predictive power across several studies. Of these studies, Li et al.¹⁴ performed a binary logistic regression that attained the highest AUC across analyses, with a threshold preoperative hemoglobin (Hb) level of 111.5 g/L providing a sensitivity of 91.9 %, specificity of 86.2 %, and AUC of 0.93 for predicting postoperative transfusion requirement. Utilizing this preoperative threshold in tandem with a surgeon's clinical discretion may offer even greater sensitivity, offering a straightforward means of determining which patients should have a postoperative CBC drawn. One study was able to attain a higher sensitivity at the sacrifice of a small degree of specificity, as Beckers et al.⁹ utilized a multivariable regression equation which accounted for age, gender, surgical indication, hypertension, and preoperative Hb, resulting in a sensitivity of 96.7 %, specificity of 75.5 %, and AUC of 0.92. While the moderate sacrifice in terms of specificity may well be worth the added emphasis on patient safety, this multivariate predictive model requires far more datapoints than the simple 111.5 g/L preoperative Hb threshold used by Li et al., rendering the latter the likely more enticing option. The orthopaedic literature supports the significance of these predictors, especially that of abnormal preoperative values, in the settings of hip fracture repair, lumbar spine surgery, high tibial osteotomy, and reverse total shoulder arthroplasty.^6,7,15,32 While the utilization of preoperative lab value thresholds can offer a quick and easy means of patient risk-stratification, future studies will be needed to assess the generalizability of the results of the most effective predictive models. More research is also warranted to understand factors such as race and surgical approach. As no current model offers a 100 % sensitivity, there remains a need for surgeon discretion with the use of any approach, as to tend to patients that might otherwise be overlooked.

4.3 Costs associated with postoperative laboratory testing

Routine postoperative testing was shown to lead to unnecessary expenditures for a large proportion of patients due to a lack of risk factors and need for clinical intervention. Approximately 3 tests were performed on patients who didn't receive any intervention and based on the costs of $14.39 per CMP and $6.79 per CBC in Angerame et al.,¹⁶ hospitals could save approximately $21 to $42 per patient if only the patients who required clinical intervention could be isolated for testing. These modest patient-level savings are magnified for the healthcare system, as Philips et al.² reported billing $233.10 per CBC and $569.95 per BMP. After instituting their intervention prospectively, they were able to generate $568,000 of savings within a seven-month period. Additionally, limiting testing to those patients who will require an intervention prevents delayed discharge due to abnormal test results, another national driver of TJA cost.^36–38 Kraus et al.³⁹ found that keeping patients overnight for abnormal CBCs and BMPs was not cost effective and provided little value. While there are no current clinical guidelines for routine postoperative laboratory testing, we recommend testing for patients with abnormal pre-operative lab values, increased age, high ASA score, high intraoperative blood loss, diabetes, and CKD. In tandem with strict adherence to clinical guidelines for transfusion^40,41 and IV electrolyte correction,^41,42 targeted testing can provide cost savings both directly through reduced expenditure of testing resources and indirectly by avoiding unnecessary prolongation of care episodes that can result from routine laboratory testing while maintaining patient safety.

4.4 Limitations

This study is not without limitations. First, the retrospective nature of each of the included studies confers a higher risk of bias given limited controls. Second, heterogeneity among studies prevented meta-analysis and led to exclusion of certain outcomes assessed in individual studies included in this review. The authors selected outcomes based on consistency across the included studies as well as their own clinical discretion. Additionally, the analysis was limited to routine postoperative labs (CBCs, CMPs/BMPs). As such, the utility and ideal patient selection for non-routine tests such as C-reactive protein, erythrocyte sedimentation rate, blood cultures, and coagulation studies were not detailed in this investigation. Lastly, varying perioperative care procedures across studies may limit the generalizability any one particular finding. These variations include heterogeneous transfusion thresholds, surgical approaches, proportion of inpatient and outpatient procedures, and use of perioperative prophylactics. Also, TJA patients may have comorbidities which require frequent postoperative laboratory testing, in which discussion of optimal testing protocols for these patients are not discussed in this review or any included studies. These patients require appropriate care based on clinical discretion. Nonetheless, there were a number of consistencies in findings across studies that are encouraging for the generalizability and clinical application of these results.