1 Introduction

The recent emphasis on patient-reported outcomes (PROs) in orthopaedics has brought about a search for the optimal PRO measures and methods for meaningful interpretation.^1–6 Patient expectations has emerged as one measure of interest due to its association with improved postoperative outcomes, and its potential as a point of intervention during preoperative counseling.⁷ While there is considerable heterogeneity in the definitions and tools used to measure patient expectations,^8,9 investigations have broadly examined either preoperative expectations or the degree to which those expectations were met or fulfilled postoperatively (met expectations).

Several studies have identified both preoperative expectations and met expectations as independent predictors of postoperative outcome scores. Existing literature has consistently demonstrated the role of preoperative expectations in predicting favorable outcomes and satisfaction in total knee arthroplasty patients.^10–12 Similar investigations in other extremity surgeries have reinforced the association between preoperative expectations and postoperative PROs.^7–9 Concomitantly, research has highlighted the significant influence of met expectations on postoperative PROs at both early and long-term follow-up.^13–19

Despite the abundant body of orthopaedic literature investigating the separate roles of preoperative expectations and met expectations as predictors of PROs, there is a lack of evaluation of these predictors in combination or head-to-head. Patients with high preoperative expectations may have difficulty meeting those expectations, and the influence of high met expectations on PROs may depend on patients’ level of preoperative expectations. Moreover, the precise role of expectations may depend on the type of procedure performed. There is limited literature evaluating patient expectations and PROs specifically in non-arthroplasty knee surgery patients. Therefore, combining both preoperative expectations and met expectations to evaluate PROs in a non-arthroplasty knee population is clinically relevant.

This study aimed to determine the prognostic significance of categorizing non-arthroplasty knee surgery patients into clusters based on both preoperative expectations and met expectations. We hypothesized that patients grouped into clusters characterized by “high” met expectations would have better postoperative PROs compared to those characterized by “low” met expectations. Based on the prior literature, we also hypothesized that among patients characterized by “low” met expectations, patients with “high” preoperative expectations would have superior PROs.

2 Methods

All patients who underwent non-arthroplasty knee surgery from June 2015 to May 2021 at a single urban academic institution and who were enrolled in the institution's prospective outcomes registry were retrospectively identified.²⁰ All open or arthroscopic knee procedures were included except for isolated soft tissue reconstruction, mass excision, foreign body or hardware removal, irrigation and debridement, biopsy, or injections. Exclusion criteria for participation in the registry included patients who were younger than 12 years, incarcerated at the time of enrollment, non-English speaking, or lacked a working email address. All patients provided informed consent for participation in the registry prior to undergoing surgery, and the study was monitored by the local Institutional Review Board (IRB).

Patients completed a baseline electronic survey within one week of surgery and a follow-up survey at two years postoperatively. Surveys were distributed via email or text message and were completed independently by the patient either remotely or in a private room at clinical sites. Patient expectations were measured using the expectations domain of the Musculoskeletal Outcomes Data Evaluation and Management System (MODEMS).^8,21 Preoperatively, MODEMS assesses the patient's expected results of treatment in six key areas: relieving symptoms, activities of daily living, sleep, work, exercise, and preventing disability. Items are measured on a 5-point Likert scale, with responses ranging from “not at all likely” to “extremely likely”. Postoperatively, MODEMS assesses the degree to which a patient's preoperative expectations of the results of treatment in each key area were met on a 5-point Likert scale with responses ranging from “definitely not” to “definitely yes”. The total scores for each measure were converted to a scale from 0 to 100, with a higher score representing higher levels of preoperative expectations and met expectations.

Each survey also consisted of multiple PRO measures assessing function, pain, psychosocial health, activity levels, and satisfaction. These included the Patient-Reported Outcomes Measurement Information System (PROMIS) computer adaptive tests (CAT) in six domains²²—physical function, pain interference, fatigue, social satisfaction, anxiety and depression—a numeric pain scale (NPS),²³ the International Knee Documentation Committee (IKDC) subjective knee evaluation form,²⁴ and the Marx Activity Rating Scale (MARS).²⁵ The follow-up survey also included the Surgical Satisfaction Questionnaire (SSQ-8)²⁶ and a single item assessing patient perception of complete resolution of their surgical problem, or “completely better” status.²⁷ All sociodemographic information was self-reported. Surgical procedures and American Society of Anesthesiologists (ASA) score were confirmed by chart review. Study data were collected and managed using REDCap (Research Electronic Data Capture) platform hosted at the study institution.²⁸

Patients were grouped into four clusters based on their preoperative expectations and met expectations scores using K means cluster analysis with the number of clusters (K) set to 4. Four distinct cluster groups were formed: high preoperative-high met expectations (HIGH-HIGH), low preoperative-high met expectations (LOW-HIGH), high preoperative-low met expectations (HIGH-LOW), and low preoperative-low met expectations (LOW-LOW). Subsequent analysis was performed based on cluster group. A cluster group assignment could only be made if the patient had both a preoperative expectations and met expectations score, therefore, patients without a cluster group assignment were not included in the analysis.

Categorical variables were expressed as frequency and percentage and were compared using a Pearson Chi-squared test, or Fisher's Exact test for cell counts less than 5. Continuous variables were expressed as mean and standard deviation and were compared using a Wilcoxon rank sum test for pairwise comparisons, and a Kruskal-Wallis test for comparisons across more than two groups. PROMIS scores were expressed as T-scores which were automatically calculated in REDCap upon the completion of each PROMIS CAT domain. T-scores are standardized on a scale from 0 to 100, with a score of 50 representing the reference population mean and a standard deviation of 10. A higher PROMIS score indicates more of the concept being measured. Therefore, a score of 60 on PROMIS physical function represents a score that is 1 standard deviation better than the population mean, while a score of 60 on PROMIS pain interference represents a score that is 1 standard deviation worse than the population mean. Change scores were defined as the difference between the two-year postoperative score and the baseline score.

A backwards stepwise elimination logistic regression was used to identify independent predictors of achieving completely better status. All significant associations between demographic variables and cluster group found on the bivariate analysis were included as candidate variables. Continuous variables were treated in the model as continuous. At each step, variables were chosen based on their contribution to the model's R² and the model was limited by minimum BIC. All statistical tests were two-sided with statistical significance set at p < .05. Significance for pairwise comparisons was adjusted using the Bonferroni correction. The analysis was performed using JMP Pro, Version 17 software (JMP®, Version 17.0.0. SAS Institute Inc., Cary, NC).

3 Results

A total of 1227 patients were identified who had undergone non-arthroplasty knee surgery and were enrolled in the registry within the study timeframe. Of these patients, 638 (52 %) had both preoperative expectations and met expectations scores available and were assigned to a cluster group. There was no statistically significant difference in the mean preoperative expectations score between patients who responded to the follow-up survey and patients who were lost to follow-up (87.1 vs 84.1, p = .09). The median time to follow-up was 24.9 months (IQR = 24–27).

Mean preoperative expectations and met expectations scores were significantly different between cluster groups (p < .001, Table 1). There were 350 patients (55 %) in the HIGH-HIGH group, 145 (23 %) in the LOW-HIGH group, 115 (18 %) in the HIGH-LOW group, and 28 (4 %) in the LOW-LOW group.

Cluster groups differed based on several demographic variables including age, education, employment, income, legal claim related to injury, preoperative narcotic use, body mass index (BMI), ASA score, and history of injury or prior surgery to the operative knee (p < .05, Table 2). The HIGH-LOW group on average was older than the HIGH-HIGH or LOW-HIGH groups (38.3 years vs 32.8 or 34.3, p = .003, respectively) and had a higher BMI (30.7 kg/m² vs 27.6 or 28.4, p < .001). The HIGH-LOW group also had worse mean baseline PROs compared to the HIGH-HIGH and LOW-HIGH groups in the majority of measures (p < .05, Table 3). Finally, the HIGH-HIGH group had a higher mean (SD) baseline MARS score compared to all other cluster groups (HIGH-HIGH = 62.3 (33.6), LOW-HIGH = 48.7 (36.9), HIGH-LOW = 44.1 (38.2), LOW-LOW = 32.3 (36.8); p < .001).

Across all measures, cluster groups with high met expectations had better two-year PRO scores and more improvement in score compared to cluster groups with low met expectations (Table 4). When comparing within cluster groups with similar met expectations, the HIGH-HIGH group had statistically significantly better two-year PROs compared to the LOW-HIGH group for PROMIS physical function (56.5 vs 54.2, p = .004), PROMIS pain interference (45.1 vs 47.5, p = .002), PROMIS social satisfaction (59.6 vs 57.0, p = .006) and IKDC (81.6 vs 76.4, p = .008). In contrast, the HIGH-LOW group had statistically significantly worse two-year PROs compared to the LOW-LOW group for NPS (5.0 vs 3.1, p = .002), IKDC (45.7 vs. 60.5, p = .008), and SSQ-8 (56.4 vs 73.5, p < .001).

When comparing the change scores between cluster groups with similar met expectations, the HIGH-HIGH group had a statistically significantly greater change in IKDC (38.1 vs 33.1, p = .009) and MARS (−14.4 vs −4.2, p = .004) when compared to the LOW-HIGH group. There were otherwise no statistically significant differences in the change scores between cluster groups with similar met expectations (Table 5).

The results for the PROMIS scores for each cluster group are displayed graphically in Fig. 1 as a PROMIS profile for each cluster group at each timepoint. The baseline and two-year scores are centered with the population mean of 50 set at zero to facilitate comparison. The baseline PROMIS profiles for all cluster groups are similar, whereas the two-year and change PROMIS profiles are similar only within cluster groups with similar met expectations. At two years postoperatively, cluster groups with high met expectations scored better than the population mean in all PROMIS domains, whereas cluster groups with low met expectations scored worse than the population mean in all PROMIS domains. Additionally, cluster groups with high met expectations demonstrated more improvement in PROMIS domains, whereas cluster groups with low met expectations demonstrated less improvement in PROMIS domains.

Close

Fig. 1 Promis profiles by cluster group. Each column displays the mean PROMIS scores for a given cluster group. Rows display the mean PROMIS scores at each timepoint: baseline, two-year and change from baseline (top to bottom). The baseline and two-year scores are centered with the population mean of 50 set at zero. The baseline PROMIS profiles are similar across all groups, however, the two-year and change profiles are similar only within groups with similar met expectations.

Export to PPT

Backward stepwise logistic regression for predictors of completely better status reduced the model to two variables: cluster group and ASA score. The overall model was significant (p < .001) with an R² of .19. The odds ratios for achieving completely better status based on cluster group are reported in Table 6. The HIGH-HIGH group had increased odds of achieving completely better status, while the HIGH-LOW group had decreased odds compared to all other cluster groups. The greatest difference in odds was observed between the HIGH-HIGH group and the HIGH-LOW group (OR = 16.69, 95 % CI [9.43, 29.53], p < .001). The smallest difference in odds was observed between the HIGH-LOW group and the LOW-LOW group (OR = .31, 95 % CI [.12, .78], p = .01). The LOW-HIGH group had an increased odds of achieving completely better status compared to the LOW-LOW group (OR = 3.08, 95 % CI [1.32, 7.18], p = .01).

4 Discussion

PROs have gained increasing importance in the field of orthopaedic surgery, and understanding their interaction with patient expectations is of particular interest. Numerous studies have consistently demonstrated the independent predictive value of both preoperative expectations and postoperative met expectations on PROs following orthopaedic surgery. However, few studies have analyzed preoperative expectations and met expectations in combination. This study utilized cluster analysis to group patients based on both preoperative expectations and met expectations to facilitate a multidimensional analysis of the impact of expectations on PROs. Overall, our results supported the primary hypothesis, demonstrating that high met expectations cluster groups reported better postoperative PROs and more improvement in PROs from baseline, regardless of the level of preoperative expectations. In contrast, the results contradicted the secondary hypothesis, demonstrating that among patients with low met expectations, those with high preoperative expectations reported worse postoperative PROs. High preoperative expectations cluster groups had better postoperative PROs only when those expectations were actually met. Additionally, low preoperative expectations did not preclude patients from positive outcomes, as long as those expectations were met postoperatively.

Across all four cluster groups, patients with high met expectations demonstrated better two-year outcome scores and improvement in scores when compared to patients with low met expectations, regardless of their preoperative expectations scores. Patients placed in clusters characterized by high met expectations also exhibited a significantly increased likelihood of achieving completely better status. Our findings underscore the previously established importance of met expectations in the literature.^{13,18,29–33} Munn et al. found that in a cohort of total knee arthroplasty patients, met expectations were significant modulators of pain and satisfaction one year following surgery.³³ Similarly, in a cohort including both arthroplasty and non-arthroplasty patients, Lin et al. found that met expectations predicted nearly all outcome measures two years following knee surgery.¹⁸ The present study adds to the existing literature by examining an entirely non-arthroplasty knee cohort at two-year follow-up using cluster analysis, and presents evidence suggesting that met expectations may be more important than preoperative expectations in achieving favorable outcomes.

Among patients with low met expectations, the HIGH-LOW cluster exhibited lower two-year PROs, lower satisfaction levels, and lower rates of achieving completely better status when compared to the LOW-LOW cluster. This finding contradicts our second hypothesis that patients with high preoperative expectations would have superior outcomes. Numerous studies have consistently shown that greater preoperative expectations are correlated with better patient satisfaction following knee surgery,^30,31 and with improved patient outcomes following a variety of orthopaedic surgery.^34–38 One explanation for the inferior outcomes observed in the HIGH-LOW cluster group in this study could be that the HIGH-LOW group was significantly older and had a higher proportion of patients with an ASA status greater than ASA I. However, cluster group was still a predictor of overall outcome on multivariable analysis, even when controlling for these and other potential confounding variables. Additionally, there were no significant differences in the baseline PROs between the HIGH-LOW and LOW-LOW groups.

A second explanation for the inferior outcomes observed in the HIGH-LOW group could be related to the presence of unrealistic preoperative expectations. Hafkamp et al. proposed the idea that unrealistically high expectations could result in unfulfilled expectations, thereby creating dissatisfaction.³⁹ The fact that the baseline PROs differed only in the HIGH-LOW group suggests that there may indeed be something unique about this combination of expectations. Patients with worse baseline status may harbor expectations that exceed what is realistically achievable given their underlying health, predisposing to unmet expectations. Previous studies have consistently highlighted the critical role of expectation fulfillment in patient satisfaction.^29,32,40 Additionally, several studies have underscored the influence of proper preoperative education in modifying preoperative expectations.^41–43 Engaging in preoperative discussions offers surgeons an opportunity to proactively manage patient expectations, reducing the prevalence of unrealistic anticipations and increasing the likelihood of meeting expectations post-operatively.

While the HIGH-HIGH group demonstrated superior postoperative PROs and a higher rate of achieving completely better status than all other cluster groups, the LOW-HIGH group scored similarly to the HIGH-HIGH group and had better outcomes compared to the LOW-LOW group. The superior PROs in the HIGH-HIGH group highlights the importance of striving for high preoperative expectations, provided they can be realistically met. Additionally, while high preoperative expectations may set one up for achieving a more favorable outcome, low preoperative expectations do not preclude one from achieving a similar outcome, so long as those expectations also are met. Lützner et al. showed that patients undergoing knee arthroplasty have expectations regarding the results of treatment in at least 31 different expectation areas, placing varying levels of importance on each area.⁴⁴ Fulfillment of expectation areas that patients deemed most important was associated with increased satisfaction. The results advocate for enhanced preoperative counseling to improve the likelihood of surgical success as defined by the patient's most important expectations.

We acknowledge several limitations with this study. First, our patient population was from a single outpatient hospital, which may limit the external validity of our findings. Second, the LOW-LOW group had a notably smaller sample size than the other three cluster groups, which limits the ability to detect differences in this cluster group. Third, only 52 % of the eligible patients had expectations scores available for analysis, which could introduce a selection bias favoring patients more inclined to complete surveys. However, there was no significant difference in the preoperative expectations between patients who completed follow up surveys and who did not, so we believe that the potential impact is relatively limited. Finally, while MODEMS is a validated measure of patient expectations that has been used previously, there is considerable heterogeneity in the measures used in prior studies, and its use therefore limits the ability to make direct comparisons.

5 Conclusion

Met expectations may be a better predictor of postoperative outcomes than preoperative expectations in non-arthroplasty knee surgery. Additionally, high preoperative expectations may only lead to better outcomes if those expectations are met, and low preoperative expectations do not preclude patients from achieving positive outcomes. This study highlights the importance of setting realistic preoperative expectations and focusing on achieving expectations postoperatively. These findings offer valuable insights for clinicians to manage patient expectations effectively based on individual characteristics and expected treatment outcomes.

CRediT authorship contribution statement

Brandon Leon: Investigation, Methodology, Writing – original draft. Dominic J. Ventimiglia: Conceptualization, Data curation, Investigation, Formal analysis, Methodology, Project administration, Writing – original draft, Writing – review & editing. Evan L. Honig: Conceptualization, Investigation, Methodology, Writing – review & editing. Leah E. Henry: Investigation, Writing – review & editing. Andrew Tran: Investigation, Writing – review & editing. Michael A. McCurdy: Investigation, Writing – review & editing. Jonathan D. Packer: Project administration, Writing – review & editing. Sean J. Meredith: Project administration, Writing – review & editing. Natalie L. Leong: Project administration, Writing – review & editing. R. Frank Henn: Conceptualization, Project administration, Supervision, Writing – review & editing.

Consent

Informed consent was obtained from all patients for participation in the study.

Ethics approval

This study was approved by the Institutional Review Board (IRB) Committee at the University of Maryland, Baltimore (HP-00062261).

Funding

This work was supported by a grant from The James Lawrence Kernan Hospital Endowment Fund, Incorporated (BL1941007WS). This work was also supported in part by Career Development Award Number IK2 BX004879 from the United States (U.S.) Department of Veterans Affairs Biomedical Laboratory R&D (BLRD) Service.