1. Introduction
Patellofemoral pain (PFP) is a multifactorial condition, which includes almost 30% of all injuries seen in sports medicine clinics [1, 2]. PFP is retropatellar or peripatellar pain during activities, in which patellofemoral joint (PFJ) is subjected to high loads [2, 3]. The annual prevalence of PFP is approximately 22.7% in the general population and 13% in women aged 18-35 years [4]. However, PFP is considered as a of the most challenging musculoskeletal condition to manage, and conservative treatment mostly leads to no long-term clinical results [5] so that 56.7% of cases report unfavorable outcomes 5-8 years after treatment [6]. 
PFP has been linked to abnormal kinematic and kinetic of lower extremity; however, the role of abnormal kinetics in the development of PFP has not been fully investigated [7, 8]. Overall, It has been suggested that loading rate is among the most important kinetic parameters for evaluation of the overload of lower limb musculoskeletal tissues [9]. Loading rate is a concept to explain the “severity” with that the force enhancement at touchdown [9]. The loading rate is determined as the slope of the primary section of the vertical ground reaction force-time curve [1]. Higher loading rates have been associated with injures, plantar fasciitis, tibia bone stress injuries, knee osteoarthritis, and PFP [8, 11]. 
Higher loading rates in PFP patients compared to healthy individuals have been explained by faulty knee kinematics [9]. Overall, during weight-bearing activities, an increase in knee flexion will be accompanied by the patellofemoral joint reaction forces (PFJRF) [12]. Consequently, PFP patients possibly utilize compensatory mechanisms at the knee to minimize symptom exacerbation and reduce PFJRF and consequently, reduce pain [12, 13, 14]. For example, reduced knee flexion is possibly a protection mechanism to decrease PFJ stress and as a result, relief pain [12, 14]. Nevertheless, it has been shown that a decrease in knee flexion can be a mechanism for higher loading rates [15]. 
Although the available evidence suggests that altered loading rate may underlie the etiology and progression of PFP, conflicting findings have been reported. For example, Silva et al. reported that patients with PFP compared to healthy individuals had an increase in loading rate while climbing stairs [9]. Esculier et al. reported a tendency to decrease loading rate in patients with PFP [16]. Further research is needed to understand these inconsistent findings related to loading rates in patients with PFP. Furthermore, understanding the factors that affect the body’s ability to absorb load may help as part of rehabilitation protocol and prevent lower extremity injuries [17, 18].
Given the mechanical connection between PFP and knee osteoarthritis, the biomechanical investigation of common movement patterns in patients with PFP possibly provide valuable information about protective mechanisms in these patients [9, 17]. Thus, exploration the potential underlying mechanics leading to PFP is critical to effectively treat and prevent this orthopedic condition [17]. Thus, the aim of this review and meta-analysis study was to answer the question “Do patients with PFP have higher loading rate compared to healthy individuals? 
2. Materials and Methods
The design and writing of this study were done in accordance with the PRISMA 2009 guidelines.
Search strategy
The research question was determined according to the PICOS framework. Papers were included if they presented outcome of clinical studies (S) on the assessment of loading rate (O) in PFP patients (P) compared with healthy individuals (C). The intervention (I) was not considered. AY and MA independently performed searching of the following electronic databases from October 2020: ScienceDirect, Scopus, Medline, and PubMed. Google Scholar functioned as a complement search engine. Search strategy was involved an extensive complex of search terms and MeSH terms (Table 1).


We had no limitation for publication date but only articles in English language were included. In general, the three categories of keywords and relevant synonyms were used. These categories were connected by “AND”, and terms of the same category were connected by “OR”, and if possible, the insignia * was used to add all the derivations of the terms. All literature collated was imported to reference management Mendeley software version 6.1 and stored alphabetically, and then, duplicates were removed automatically. 
Study selection
All articles identified were screened using the eligibility criteria by two reviewers independently (AY and MA). The inclusion and exclusion checking was implemented by two reviewers (AY and MA) according to the title and abstract of the studies included. The full text of titles and abstracts was reviewed, and those found to be eligible were included in the final review. Eligibility criteria were as follows: 
Inclusion criteria: case-control studies of human participants, which compared loading rate between PFP patients and healthy individuals during weight-bearing tasks. Furthermore, the selected articles must be published in English language. 
Exclusion criteria: letters, conference proceedings, case reports, cadaveric studies, prospective studies, no comparison of patients with PFP with healthy individuals, abstracts, reviews, clinical trials and non-English and Persian language articles were excluded. After an initial screening of title and abstract, the studies were categorized as as A) meeting (“include”), B) may meeting (“potentially”), C) not meeting the criteria (“do not include”). A full-text assessment was checked if the title and abstract were related, with those found to be eligible included in the review. Disagreement in identified papers for inclusion criteria was resolved in a consensus session.
Methodological quality assessment 
Two reviewers (AY and MA) independently evaluated the studies included by modified version of the Downs and Black checklist, which has well test–retest (r=0.88) and inter-rater (r=0.75) reliability [19]. This version has 25items, by which the following sub-groups: reporting (items 1,2,3,5,6,7, and 10), external validity (items 11 and 12), internal validity (items 16,18, and 20), and internal validity confounding (items 21,22, and 25) were evaluated [20]. The items were scored as 0 [“no”), 1 (“yes”), except item 5 for the principal confounders, which was scored as 0 (“no”), 1 (“partially”), and 2 (“yes”) [21]. Studies with quality scores of 70% or greater were classified as high quality, 40%-69% as moderate quality, and 40% or less as low quality [22]. Disagreement in identified papers for inclusion criteria was resolved in a consensus session.
Outcome measures and data extraction 
Two reviewers (AY and MA) independently extracted the demographic information: author name and year of publication, purpose, task and results. The outcome was loading rate during weight bearing tasks. Data was extracted, reviewed, and analyzed by two reviewers (AY and MA). A reviewer (MA) extracted data and this was confirmed by a second reviewer (AY). Where data were not available, the author was contacted. If data were still unable to be sourced, Web Plot Digitizer software with high reliability (Pearson’s r=0.999) and validity (r=0.989) designed to extract data from digital plot images, was used [23]. 
Statistical analyses
All analyses were performed by Review Manager software version 5.3. Standardized Mean Differences (SMD) and 95% confidence intervals (CI) were extracted to compare loading rate between PFP with healthy groups. We also used a forest plot to facilitate the comparison of results between the two groups. Meta-analyses were performed by calculating the effect size (ES) by SMD in a random-effects model. ES was categorized as: low quality (≥50%), medium quality (51%-74%), and high quality (≤75%) [24]. Random model is often utilized for small sample sizes and sample heterogeneity with selection bias common in the literature. Overall, random-effects model assumes the study populations are varied and different from each other [25]. 
The I2 statistic was used to specify the value of heterogeneity, where the percentages quantified the magnitude of heterogeneity: 25%=low, 50%=medium, and 75%=high heterogeneity [26]. The I² statistic describes the percentage of variation across studies that is due to heterogeneity rather than chance [27]. The level of statistical significance (P<0.05) was calculated via Z test. Quality rating was used to determine levels of evidence, based on a modified version of the van Tulder criteria [21]:
● Strong evidence: including at least three high-quality studies.
● Moderate evidence: including at least three moderate-/high-quality studies or two high- quality studies.
● Limited evidence: pooled findings amongst multiple low-/moderate-quality studies, or one high-quality study.
● Very limited evidence: one low-/moderate-quality study.
● Conflicting evidence: one/some studies show significant effects and one/some studies show no significant effects while the CIs of the pooled effect size lead one to accept the null hypothesis.
3. Results 
Study selection
The primary search be extracted 433 study that 41 was remove due to duplication. Then, two researchers (AY and MA) screened the remaining 392 study according to the title and abstract of the studies, which resulted in the elimination of 350 studies that were inconsistent with the purpose and eligibility criteria. 42 articles remained for full text evaluation, of which 17 were omitted, because they did not of the eligibility criteria. Seven study related to the patellofemoral joint loading rate, seven study due to lack of healthy group or comparison with other pathologies, two study due to non-reporting of loading rate values and two study that had applied various interventions in the evaluation were removed from the review process. Finally, seven study [9, 11, 16, 17, 28, 29, 30] were selected for systematic review and meta-analysis (Figure 1). Methodological quality assessment
The average score of methodological quality of studies was 72.5% (range 60%-80%), which indicates the medium quality of studies. 58% of studies (n=4) had high methodological quality [11, 16, 17, 29], 42% of studies (n=3) were as medium quality [9, 28, 30] (Table 3).


The strength of the quality studies was report particular the expression of the objectives and outcome measures. All studies have poor external reliability scores. In fact, none of the studies have identified the source of the population and how patients are selected. Also, most studies on the internal validity of confounder were poor partly. Only 4 studies reported that patients from the same population and same time period were employed [9, 11, 17, 29]. Also, only 3 studies reported the adequate adjustment for confounding [11, 16, 17]. 
Characteristics of studies
Table 2 shows the demographic information of studies.


There are a total of 490 people (mean age: 29.94 years; mass: 66.07 kg; height: 167.80 cm) in 7 study. Overall, 245 people was healthy individuals (mean age: 29.20 years; body mass: 65.50 kg; height: 168.12 cm) and 245 PFPS patients (mean age: 30.68 years; body weight: 66.65 kg; height: 167.47 cm). Only 2 studies, demographic characteristics were not reported [28, 30]. 57.14 % of studies (n=4) when running [11, 16, 28, 29]; 28.57% (n=2) when stair climbing [9, 17]; and 14.29% (n=1) when walking [30]. 
Loading rate 
7 studies [9, 11, 16, 17, 28, 29, 30] peak evaluated loading rate during weight-bearing tasks (Figure 2). Overall, results of meta-analysis showed that from statistical point of view there was no significant difference between PFP patients and healthy individuals in the loading rate (P=0.52) ; But due to the results of the forest plot; strong evidence (4 studies=HQ, 3 studies=MQ, I2=98 %) with medium effect size (SMD=0.50; 95% CI=[-1.01 to 2.00]) indicates a tendency to increase loading rates in PFP patients compared to healthy individuals.
4. Discussion
The aim of this review and meta-analysis was the evaluation of loading rate in PFP patients compared to healthy individuals during weight bearing tasks. The results systematic review and meta-analysis showed a tendency to higher loading rates be in PFP patients than health’s individuals.  
Overall, higher loading rates in patient populations with PFP have been explained by faulty kinematics of the knee [15]. Evidence suggests that PFP patient to minimize the PFJRF, reduced flexion and extensor moment of the knee [13, 31]. The PFJRF is force between the quadriceps muscles and patella tendon that increases with quadriceps muscle force and knee flexion angle [13]. At tempts to minimize the PFJRF involve alterations to one or both of these quantities [32]. In this regard, a systematic review published reported that PFP patients have decrease knee flexion and reduced the demand quadriceps for PFJRF decreases and ultimately reduces pain [4]. 
The range of motion of the lower limb when the loading phase possible influence on the forces experienced [9]. Cook et al. (1997), reported which reducing knee flexion when movement leading to increase vertical ground reaction forces and lower limb loading rates in health’s individuals [33]. Hence, although reduced knee flexion is a logical behavior by which PFJRF can be limited, it may hamper the lower extremity absorption shock mechanisms [9, 15]. While reduced knee flexion possible help protect versus the development of PFP, thus “quadriceps avoidance” behavior my lead to chronic disuse of the knee extensors [31]. Eccentric quadriceps contraction is considered to be the initial dynamic shock absorbent mechanism during weight acceptance and my its decrease had be correlation to higher loading rates [9, 34]. Such compensatory behavior in PFP patients may decrease capacity of active shock absorption via quadriceps muscle contraction and higher passive shock absorption by the viscoelastic virtues of both bone and cartilage [13, 31]. 
Higher loading rates at the tissue surface can lead to cartilage split or development of surface split, such to the osteoarthritis mechanisms [4, 9, 31]. The diminish in the peak knee extensor moment possible also have a negative effect on tibiofemoral joint loading [31]. Overall, loading rate would not had strong effects on the PFJ by nature of its vertical alignment and the fact that patellofemoral stress is initially affiliate on the value of quadriceps force and the femur direction. Thus, compensatory behaviors to exacerbation reduction of PFP possible have negative effect on the tibiofemoral joint, as a result of its horizontal direction, via axial compression [9, 17, 31]. As a result, studies have reported that there is a correlating between higher in the loading rates with the development of pain and knee osteoarthritis [17]. In this regard, the results of studies have shown that people with knee osteoarthritis be exposed severe and frequent loading of the lower extremities before the stage of the complication [4].
Clinical application
The compensatory movement pattern in PFP patients must be addressed; as it possible have correlation to higher loading rates that can lead to detrimental effects on the knee joints and functional status is consequently reduced [4, 9, 17]. Therefore, first: it is necessary to identify which factors cause this loading rate alteration to promote reestablishment of knee function and consequently reduce the disease progression [17]. Second: rehabilitation programs aimed at gait retraining, using either visual or verbal feedback, seem to be effective in reducing loading rates [34]. Third: since long-time quadriceps avoidance possible lead to quadriceps weakness and atrophy; should be considered the accession of quadriceps strengthening as a needful and effective partial in prevention and rehabilitation approaches to PFP [31].
Limitations and suggestions 
There are four limitations to this study. First, because of the case-control nature of the studies, we results is not able to differentiation between cause and effect about loading rate evaluated [9]. Second, alone one dynamic activity function was reported in studies and no comparison another weight bearing activity [17]. Assessment of movement activity which are more challenging in terms of mechanical and muscular demands, similar stairs ambulation, perhaps further contribute to the perception of compensatory mechanisms producing by PFP patient, that possible not be observed when gait [17]. Third, large heterogeneity in this meta - analysis was partly attributed to patients population and methodological differences in movements analyses [11]. Fourth, lack to evaluate the loading rate on the healthy or low - effect leg; because some studies reported that PFP patients tend to action load on healthy leg [29]. Therefore, A) it is recommended that loading rates be evaluated and compared in different challenging tasks B) on both sides in the C) athlete and general population separately. 
5. Conclusion
The PFP patient reduced knee flexor angle and extensor moment of the knee to limit the PFJRF. Such a compensatory mechanism, decrease of active shock absorption and greater passive shock absorption which finally leads to higher loading rates. Higher loading rates can have efficacy on the tibiofemoral joint that through axial compression correlating to development of pain and knee osteoarthritis. 

Ethical Considerations
Compliance with ethical guidelines
There were no ethical considerations to be considered in this research. 

Funding
The present paper was extracted from the first author's MA. thesis in Department of Corrective Exercises and Sports Injury, Faculty of Physical Education and Sport Sciences, Bu Ali Sina University, Hamedan. 

Authors' contributions
All authors equally contributed to preparing this article.

Conflict of interest
The authors declared no conflict of interest.


Reference

1. Lopes Ferreira C, Barton G, Delgado Borges L, dos Anjos Rabelo ND, Politti F, et al. Step down tests are the tasks that most differentiate the kinematics of women with patellofemoral pain compared to asymptomatic controls. Gait & Posture. 2019; 72:129-34. [DOI:10.1016/j.gaitpost.2019.05.023] [PMID]
2. Mohamadreza Ahmadi, Ali Yalfani, Farzaneh Gandomi KR. [Effect of Twelve weeks of sensorimotor training on pain, improvement proprioception, muscle strength, and postural control in men with patellofemoral pain syndrome: A randomized single-blind clinical trial (Persian)]. Scientific Journal of Rehabilitation Medicine. 2020; 10(1):1-13. [DOI:10.22037/JRM.2020.112859.2271]
3. Liew BXW, Rugamer D, Abichandani D, De Nunzio AM. Classifying individuals with and without patellofemoral pain syndrome using ground force profiles: Development of a method using functional data boosting. Gait & Posture. 2020; 80:90-5. [DOI:10.1016/j.gaitpost.2020.05.034] [PMID]
4. Yalfani A, Ahmadi M, Gandomi F, Bigdeli N. [An investigation of the lower extremity kinematics during stair ambulation in people with patellofemoral pain syndrome: A systematic review (Persian)]. Journal of Paramedical Sciences & Rehabilitation. 2021; 9(4):115-25. [DOI:10.22038/JPSR.2021.41722.1987]
5. Kernozek T, Schiller M, Rutherford D, Smith A, Durall C, Almonroeder TG. Real-time visual feedback reduces patellofemoral joint forces during squatting in individuals with patellofemoral pain. Clinical Biomechanics. 2020; 77:105050. [DOI:10.1016/j.clinbiomech.2020.105050] [PMID]
6. Ahmadi MR, Yalfani A, Gandomi F, Rashid K. [The effect of twelve week neurofeedback training on perceptual pain intensity, fear of pain, pelvic drop, and dynamic knee valgus index in men with patellofemoral pain syndrome: A randomized double-blind clinical trial (Persian)]. Sadra Medical Journal. 2020; 8(2):151-64. [DOI:10.30476/SMSJ.2020.85151.1108]
7. Paoloni M, Mangone M, Fratocchi G, Murgia M, Maria Saraceni V, Santilli V. Kinematic and kinetic features of normal level walking in patellofemoral pain syndrome: More than a sagittal plane alteration. Journal of Biomechanics. 2010; 43(9):1794-8. [DOI:10.1016/j.jbiomech.2010.02.013] [PMID]
8. Cheung RTH, Davis IS. Landing pattern modification to improve patellofemoral pain in runners: A case series. The Journal of Orthopaedic and Sports Physical Therapy. 2011; 41(12):914-9. [DOI:10.2519/jospt.2011.3771] [PMID]
9. de Oliveira Silva D, Briani RV, Pazzinatto MF, Ferrari D, Aragão FA, de Azevedo FM. Reduced knee flexion is a possible cause of increased loading rates in individuals with patellofemoral pain. Clinical Biomechanics. 2015; 30(9):971-5. [DOI:10.1016/j.clinbiomech.2015.06.021] [PMID]
10. Ueda T, Hobara H, Kobayashi Y, Heldoorn TA, Mochimaru M, Mizoguchi H. Comparison of 3 methods for computing loading rate during running. International Journal of Sports Medicine. 2016; 37(13):1087-90. [DOI:10.1055/s-0042-107248] [PMID]
11. Johnson CD, Tenforde AS, Outerleys J, Reilly J, Davis IS. Impact-related ground reaction forces are more strongly associated with some running injuries than others. The American Journal of Sports Medicine. 2020; 48(12):3072-80. [DOI:10.1177/0363546520950731] [PMID]
12. Crossley KM, Cowan SM, Bennell KL, McConnell J. Knee flexion during stair ambulation is altered in individuals with patellofemoral pain. Journal of Orthopaedic Research. 2004; 22(2):267-74. [DOI:10.1016/j.orthres.2003.08.014] [PMID]
13. Salsich GB, Brechter JH, Powers CM. Lower extremity kinetics during stair ambulation in patients with and without patellofemoral pain. Clinical Biomechanics. 2001; 16(10):906-12. [DOI:10.1016/S0268-0033(01) 00085-7] [PMID]
14. de Oliveira Silva D, Barton CJ, Briani RV, Taborda B, Ferreira AS, Pazzinatto MF, et al. Kinesiophobia, but not strength is associated with altered movement in women with patellofemoral pain. Gait & Posture. 2019; 68:1-5. [DOI:10.1016/j.gaitpost.2018.10.033] [PMID]
15. Briani RV, Pazzinatto MF, Waiteman MC, de Oliveira Silva D, de Azevedo FM. Association between increase in vertical ground reaction force loading rate and pain level in women with patellofemoral pain after a patellofemoral joint loading protocol. The Knee. 2018; 25(3):398-405. [DOI:10.1016/j.knee.2018.03.009] [PMID]
16. Esculier JF, Roy JS, Bouyer LJ. Lower limb control and strength in runners with and without patellofemoral pain syndrome. Gait & Posture. 2015; 41(3):813-9. [DOI:10.1016/j.gaitpost.2015.02.020] [PMID]
17. de Oliveira Silva D, Briani R, Pazzinatto M, Ferrari D, Aragão F, de Azevedo F. Vertical ground reaction forces are associated with pain and self-reported functional status in recreational athletes with patellofemoral pain. Journal of Applied Biomechanics. 2015; 31(6):409-14. [DOI:10.1123/jab.2015-0048] [PMID]
18. Nunes GS, Barton CJ, Viadanna Serrão F. Females with patellofemoral pain have impaired impact absorption during a single-legged drop vertical jump. Gait & Posture. 2019; 68:346-51. [DOI:10.1016/j.gaitpost.2018.12.013] [PMID]
19. Barton CJ, Levinger P, Menz HB, Webster KE. Kinematic gait characteristics associated with patellofemoral pain syndrome: A systematic review. Gait & Posture. 2009; 30(4):405-16. [DOI:10.1016/j.gaitpost.2009.07.109] [PMID]
20. Wang J, Mannen EM, Siddicky SF, Lee JM, Latt LD. Gait alterations in posterior tibial tendonitis: A systematic review and meta-analysis. Gait & Posture. 2020; 76:28-38. [DOI:10.1016/j.gaitpost.2019.11.002] [PMID]
21. Sancho I, Malliaras P, Barton C, Willy RW, Morrissey D. Biomechanical alterations in individuals with achilles tendinopathy during running and hopping: A systematic review with meta-analysis. Gait & Posture. 2019; 73(1):189-201. [DOI:10.1016/j.gaitpost.2019.07.121] [PMID]
22. Malliaras P, Barton CJ, Reeves ND, Langberg H. Achilles and patellar tendinopathy loading programmes: A systematic review comparing clinical outcomes and identifying potential mechanisms for effectiveness. Sports Medicine. 2013; 43(4):267-86. [DOI:10.1007/s40279-013-0019-z] [PMID]
23. Drevon D, Fursa SR, Malcolm AL. Intercoder Reliability and validity of WebPlotDigitizer in extracting graphed data. Behavior Modification. 2017; 41(2):323-39. [DOI:10.1177/0145445516673998] [PMID]
24. Bayartai ME, Kobayashi S, Ferreira PH, Pappas E, Pinheiro MB, Sullivan J. Heritability of motion in healthy people: A systematic review and multi-level meta-analysis. Physical Therapy in Sport. 2020; 43:8-18. [DOI:10.1016/j.ptsp.2020.01.016] [PMID]
25. Borenstein M, Hedges L, Rothstein H. Meta-Analysis Fixed effect vs. random effects [Internet]. 2007. Available from: https://www.meta-analysis.com/downloads/M-a_f_e_v_r_e_sv.pdf
26. López López L, Torres JR, Rubio AO, Torres Sánchez I, Cabrera Martos I, Valenza MC. Effects of neurodynamic treatment on hamstrings flexibility: A systematic review and meta-analysis. Physical Therapy in Sport. 2019; 40:244-50. [DOI:10.1016/j.ptsp.2019.10.005] [PMID]
27. Verbecque E, Johnson C, Rameckers E, Thijs A, van der Veer I, Meyns P, et al. Balance control in individuals with developmental coordination disorder: A systematic review and meta-analysis. Gait & Posture. 2021; 83:268-79. [DOI:10.1016/j.gaitpost.2020.10.009] [PMID]
28. Messier SP, Davis SE, Curl WW, Lowery RB, Pack RJ. Etiologic factors associated with patellofemoral pain in runners. Medicine and Science in Sports and Exercise. 1991; 23(9):1008-15. [DOI:10.1249/00005768-199109000-00003] [PMID]
29. Duffey MJ, Martin DF, Cannon DW, Craven T, Messier SP. Etiologic factors associated with anterior knee pain in distance runners. Medicine and Science in Sports and Exercise. 2000; 32(11):1825-32. [DOI:10.1097/00005768-200011000-00003] [PMID]
30. Radin EL, Yang KH, Riegger C, Kish VL, O’Connor JJ. Relationship between lower limb dynamics and knee joint pain. Journal of Orthopaedic Research. 1991; 9(3):398-405. [DOI:10.1002/jor.1100090312] [PMID]
31. Briani RV, Cannon J, Waiteman MC, de Faria Negrão Filho R, Magalhães FH, de Azevedo FM. Influence of the exacerbation of patellofemoral pain on trunk kinematics and lower limb mechanics during stair negotiation. Gait & Posture. 2021; 83:83-7. [DOI:10.1016/j.gaitpost.2020.10.008] [PMID]
32. Fok LA, Schache AG, Crossley KM, Lin YC, Pandy MG. Patellofemoral joint loading during stair ambulation in people with patellofemoral osteoarthritis. Arthritis and Rheumatism. 2013; 65(8):2059-69. [DOI:10.1002/art.38025] [PMID]
33. Cook TM, Farrell KP, Carey IA, Gibbs JM, Wiger GE. Effects of restricted knee flexion and walking speed on the vertical ground reaction force during gait. The Journal of Orthopaedic and Sports Physical Therapy. 1997; 25(4):236-44. [DOI:10.2519/jospt.1997.25.4.236] [PMID]
34. Powers CM, Heino JG, Rao S, Perry J. The influence of patellofemoral pain on lower limb loading during gait. Clinical Biomechanics. 1999; 14(10):722-8. [DOI:10.1016/S0268-0033(99) 00019-4] [PMID]