Diagnosis and management of neuropathic pain post-total knee replacement: A systematic review

Table 1 Summary of study designs for included studies

No.	Ref.	Year	Study type	Intervention	Control	Follow-up (months)	Outcome	Conclusion
1	Rienstra et al[18]	2021	RCT	Duloxetine peri-operative	Analgesics, NSAIDs but no neuropathic medication	12	Pain severity using KOOS pain scale and VAS, and neuropathic pain using PainDETECT at 6 weeks, 6 and 12 months after TKR	Perioperative treatment with Duloxetine did not influence chronic or neuropathic pain up to one year after TKR
2	Buvanendran et al[25]	2010	RCT	Pregabalin peri-operative	Placebo tablets	6	Pain severity using immediate post-operative NRS, and proportion with neuropathic pain using S-LANSS at 3 and 6 months	Perioperative pregabalin treatment significantly reduced the rate of neuropathic pain to 6 months
3	Albayrak et al[23]	2017	Retrospective comparative	DRG RF Ablation	Analgesics, NSAID, local treatments	8	Pain severity using VAS, and neuropathic pain using DN4 at 15 days, 1 month and 8 months after 14-day admission for treatment of chronic and neuropathic pain three years after TKR	Radiofrequency ablation of the DRG at L4 improved the proportion of patients with neuropathic pain
4	Kretzschmar[15]	2019	Observational	DRG RF stimulator	NA	36	Pain severity using VAS, in patients selected for neuropathic pain using PainDETECT	Radiofrequency stimulation of DRG at L3&4 improved neuropathic pain to 1 year and this was sustained to three years
5	Yang et al[22]	2022	Retrospective comparative	Neurectomy, interventional pain procedures	Unclear-assume Analgesics NSAIDs	2	Pain severity using VAS, did not use a neuropathic pain score, but selected based on clinical examination	Selective denervation improves neuropathic pain after TKR
6	Zhong et al[21]	2018	Retrospective comparative	Neurectomy	Analgesics, NSAID, local treatments	12	Pain severity using VAS, did not use a neuropathic pain score. Ruled out other causes of persistent post-operative knee pain	Selective denervation for those that show response to local injection improves neuropathic pain after knee surgery
7	Clendenen et al[20]	2015	Observational	Hydro-dissection of nerve	NA	9	Pain severity using VAS, did not use a neuropathic pain score. Selected patients based on clinical examination	Hydro-dissection with steroid and local anesthetic to potentially free a stuck nerve improved chronic medial knee pain after TKR

RCT: Randomized-controlled trials; DRG: Dorsal root ganglion; RF: Radiofrequency; RF: Radiofrequency; NSAID: Nonsteroidal anti-inflammatory drug; KOOS: Knee Injury and Osteoarthritis Outcome Score; S-LANSS: Self-reported Leeds Assessment of Neuropathic Symptoms and Signs; DN4: Douleur-Neuropathique-4; TKR: Total knee replacement; NRS: Numerical Rating Scale.

Table 2 Summary of quality and bias assessments for included papers

No.	Ref.	Year	Quality: Coleman A	Quality: Coleman B	Bias	Bias method
1	Rienstra et al[18]	2021	95.0	79.1	Low	ROB2
2	Buvanendran et al[25]	2010	91.7	79.1	Low	ROB2
3	Albayrak et al[23]	2017	43.3	48.8	High	ROBINS
4	Kretzschmar[15]	2019	33.3	30.2	High	Single arm used ROBINS
5	Yang et al[22]	2022	16.7	0.0	High	ROBINS
6	Zhong et al[21]	2018	45.0	53.5	High	ROBINS
7	Clendenen et al[20]	2015	40.0	65.1	High	Single arm used ROBINS

ROB: Risk of bias; ROBINS: Risk of bias in non-randomized studies of intervention.

Table 3 Summary of risk of bias assessments for included papers

ROB2	Ref.	Year	Randomization	Deviations	Outcome	Measurement	Reported result	Overall
1	Rienstra et al[18]	2021	Moderate risk	Moderate risk	Low risk	Low risk	Low risk	Low risk
2	Buvanendran et al[25]	2010	Low risk	Low risk	Low risk	Low risk	Low risk	Low risk

ROB: Risk of bias.

Table 4 Summary of risk of bias in non-randomized studies of intervention assessments for included papers

ROBINS	Ref.	Year	Confounding	Selection	Interventions	Deviation	Missing	Outcome	Reporting	Overall
3	Albayrak et al[23]	2017	Serious risk	Low risk	Moderate risk	Low risk	Serious risk	Moderate risk	Low risk	Serious risk
4	Kretzschmar[15]	2019	Serious risk				Low risk	Low risk	Low risk	Serious risk
5	Yang et al[22]	2022	Serious risk	Low risk	Moderate risk	Moderate risk	Moderate risk	Moderate risk	Low risk	Serious risk
6	Zhong et al[21]	2018	Serious risk	Low risk	Moderate risk	Low risk	Moderate risk	Moderate risk	Low risk	Serious risk
7	Clendenen et al[20]	2015	Serious risk				Low risk	Moderate risk	Low risk	Serious risk

ROBINS: Risk of bias in non-randomized studies of intervention.

Table 5 Data from publications - baseline Visual Analogue Scale pain

No.	Ref.	Year	Recruitedn I	Recruitedn C	VAS baseline I	VAS baseline I_SD	VAS baseline C	VAS baseline C_SD
1	Rienstra et al[18]	2021	31	29	5.3	2	6.6	1.9
2	Buvanendran et al[25]	2010	120	120	7.7	1.9	8	1.3
3	Albayrak et al[23]	2017	22	17	4.3	1.7	5	2.4
4	Kretzschmar[15]	2019	9	0	5.9	1.5
5	Yang et al[22]	2022	37	18	6.5	1.9	6.1	1.8
6	Zhong et al[21]	2018	22	38	6.7	1.4	6.4	2
7	Clendenen et al[20]	2015	16	0	8	1

Table 5 gives the number and the extracted pain Visual Analogue Scale at baseline noting pain severity in the intervention and control/comparator arms with the SD. The first paper used the Knee Injury and Osteoarthritis Outcome Score pain scale which required reversing the direction and division by 10 to make it comparable to the Visual Analogue Scale. Data was extracted from a figure for the Kretzschmar’s study[15]. I: Intervention; C: Control/comparator; VAS: Visual Analogue Scale.

Table 6 Data from publications - change in pain: Follow-up pain Visual Analogue Scale and difference from baseline

No.	Ref.	Year	FU number I	FU number C	VAS FU I	VAS FUI_SD	VAS FU C	VAS FU C_SD	VAS MD	VAS CI MD	VAS SMD	VAS CI SMD
1	Rienstra et al[18]	2021	20	28	2.5	1.6	2.4	1.9	0.1	-0.9 to 1.1	0.055	-0.51 to 0.62
2	Buvanendran et al[25]	2010	113	115
3	Albayrak et al[23]	2017	22	17	2	1.6	2.8	2.1	-0.8	-2.0 to 0.4	-0.428	-1.06 to 0.20
4	Kretzschmar[15]	2019	9	0	1.8	0.4			-3.2 to -0.6		-0.42 to -2.47
5	Yang et al[22]	2022	37	18	0.9	1.2	5	1.7	-4.1	-4.9 to -3.3	-2.93	-3.71 to -2.15
6	Zhong et al[21]	2018	22	38	1.2	0.3	4.1	1.4	-2.9	-3.4 to -2.3	-2.53	-3.22 to -1.84
7	Clendenen et al[20]	2015	16	0	2	2.8			-3.0 to -0.4		-0.16 to -1.26

Table 6 gives the number and the extracted pain Visual Analogue Scale (VAS) at follow-up between 6 months and 12 months after intervention noting pain severity in the intervention and control/comparator arms with the SD. The first paper used the Knee Injury and Osteoarthritis Outcome Score pain scale which required reversing the direction and division by 10 to make it comparable to the VAS scale. Data was extracted from a figure for the Kretzschmar[15] study. The MD and the 95% confidence interval are given for comparative studies to allow easy reference to the VAS minimum clinically important difference of 2. The SMD is calculated (Hedges g, as some samples were quite small) along with the 95% confidence interval. For the single arm studies, we calculated the MD/SMD using the highest and lowest values for the control arm in other studies to provide a possible range of MD and SMD. FU: Follow-up; I: Intervention; n: Number; C: Control/comparator; VAS: Visual Analogue Scale; CI: Confidence interval.

Table 7 Data from publications - log odds ratio of numbers with neuropathic pain in each trail arm

No.	Ref.	Year	FU number I	FU number C	Final_NP number _I	Final_nonNP number _I	Final_NP number _C	Final_nonNP number _C	Log odds ratio	CI Log odds ratio	Lower 95%CI	Upper 95%CI
1	Buvanendran et al[25]	2010	113	115	0	113	6	109	-2.6	-5.5 to 0.3	-5.5	0.3
2	Albayrak et al[23]	2017	22	17	0	22	10	7	-4.1	-7.1 to -1.2	-7.1	-1.2

Two papers presented data on final proportions of patients with neuropathic pain in the intervention and control arms so log odds ratio and its 95% confidence interval could be assessed. FU: Follow-up; I: Intervention; C: Control/comparator; NP: Neuropathic pain; CI: Confidence interval.