*3.1. Demographics*

From January 2009 to December 2018 a total of 93 PD patients underwent STN-DBS insertion. Data from 13 patients were excluded from analysis because of incomplete data. Two patients underwent surgery under general anesthesia and were also excluded from further analysis. Anesthetic and electrophysiological data from the remaining 78 patients were analyzed, thus yielding electrophysiological date from 156 cerebral hemispheres (Table 1).

Data were first grouped into clusters depending on whether data were acquired when patients were awake or sedated. The awake cluster included data from two patient groups: the first group received no sedatives and the second group of patients received DEX or a combination of DEX and REMI which was discontinued approximately 20 min before MER. The data in the sedation cluster was sub-divided according to the PSA applied during surgery: DEX, CLONI, REMI, or a combination of DEX and REMI or REMI and CLONI (Table 1).

To control for systematic di fferences between groups, we tested whether demographic characteristics (age, disease duration, Unified Parkinson Disease Rating Scale (UPDRS), side of onset, sex and weight) were equivalent across groups. One-way ANOVA showed no di fference for any of these variables (respectively F(6,71) = 1.14, *p* = 0.35; F(6,71) = 0.59, *p* = 0.74; F(6,71) = 0.40, *p* = 0.88; F(6,71) = 0.39, *p* = 0.88; F(6,71) = 0.46, *p* = 0.83, F(6,71) = 0.39, *p* = 0.88 no correction for multiple comparisons were applied since we were here more concerned with minimizing type 2 errors).

### *3.2. Discontinuation of Sedative Agents*

In 7 patients, PSA agents (DEX *n* = 4, DEX and REMI *n* = 3) were given during the first part of the operation and were discontinued approximately 20 min before the start of MER. While these patients appeared clinically to be awake during surgery, it is possible that these compounds might still have affected the electrophysiological properties of the STN even after their discontinuation [14]. Therefore, we tested whether the group who received no PSA drugs and the group of patients in whom PSA was discontinued should be considered as distinct groups for further analysis. We conducted a *t*-test for comparison of the electrophysiological measures (firing rate (sp/sec), CV and MUA) between these groups. In SU activity there was a trend towards decreased firing rate (*t*(214) = 1.63, *p* = 0.11), but no di fference in the CV (*t*(214) = −0.18, *p* = 0.86). MUA was significantly reduced in the discontinued drug group (*t*(1571) = 5.67, *p* < 0.0001). Given these findings (Figure 2) we considered the patients in whom PSA agents were discontinued as a separate group from patients who received no PSA for further analysis.

**Figure 2.** Bar plot and t-test for within awake group analysis for firing rate, defined in spikes per seconds, coe fficient of variation and multi-unit activity, defined as activity above 300HZ. Mean and standard error are shown in each bar. \*\*\* *p* < 0.001. PSA: procedural sedation and analgesia.

### *3.3. E*ff*ect of PSA Agents on Firing Rate, Coe*ffi*cient of Variation and Multi-Unit Activity*

A one-way ANOVA was conducted between the seven groups for each electrophysiological measure. Significant, or trending di fferences between groups for all measures were observed (Firing rate, F(6,581) =2.08, *p* < 0.054; CV, F(6,581) = 3.69, *p* < 0.001; MUA, F(6,4088) = 15.93, *p* < 0.0001). To further test for di fferences between groups, a post-hoc two-sample *t*-tests was conducted to compare each PSA drug against the control group (*n* = 165) with respect to firing rate, CV and MUA.

No significant di fferences in the firing rate between groups after FDR correction for multiple comparisons were found, although DEX and REMI, and REMI in combination with CLONI showed a trend towards significance (*t*(209) = 2.64, *p* = 0.0531; *t*(269) = 2.28, *p* = 0.0695).

For CV, there were significant di fferences between the no drug group and the groups DEX (*t*(249) = −3.84, *p* = 0.00092), REMI (*t*(256) = −2.65, *p* = 0.013), DEX and REMI group (*t*(209) = −3.05, *p* = 0.0078), and REMI and CLONI (*t*(269) = −2.62, *p* = 0.013).

All PSA drug groups except REMI showed a significant decrease in MUA (*n* = 1255, PSA discontinued, *t*(1594) = 4.95, *p* < 0.0001; DEX, *t*(1820) = 6.73, *p* < 0.0001; CLONI, *t*(1594) = 4.45, *p* < 0.0001; DEX and REMI, *t*(1522) = 3.94, *p* < 0.0001; and REMI and CLONI, *t*(1934) = 3.69, *p* < 0.0001), REMI (*t*(1891) = 1.31, *p* = 0.19). All p values were corrected for multiple comparison using FDR (Figure 3).

**Figure 3.** Bar plot (error bars show standard error) and t-test for the No PSA group against each PSA group for firing rate, coefficient of variation and multi-unit activity. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001. CLONI: clonidine; DEX: dexmedetomidine; REMI: remifentanil; PSA: procedural sedation and analgesia.

To account for variance in the patient groups, we conducted a linear regression analysis to test the effect of the PSA agents with clinical and demographic variables included as factors. For DEX, CLONI, REMI, sex, recording location with respect to onset side of the disease, and left or right hemisphere were defined as categorical variables. We further included age at surgery (years), weight (Kg), UPDRS III pre-operative off medication, disease duration at surgery (months) and depth of the recording site within the STN (mm from the dorsal border) as continuous variables in the model (Table 2).


**Table 2.** Standard linear regression model for firing rate, CV and MUA.

In the analysis, we included demographic and clinical variables as sex, onset side (recording site ipsi- or contralateral to the onset side of the disease), hemisphere (right or left), age, weight, UPDRS III pre-operative off medication and disease duration. Estimate indicate the slope of the line, when negative indicates decrease and positive indicates increase. \* ipsilateral to body side with onset of disease. CLONI: clonidine; DEX: dexmedetomidine; REMI: remifentanil; PSA: procedural sedation and analgesia; UPDRS III: Unified Parkinson's Disease Rating Scale part III.

There were no significant effects of the PSA agents in the firing rate, but we could observe some trends. Notably, DEX was weakly associated with a reduced firing rate. For CV, DEX showed a significant effect and REMI showed a small effect. DEX and PSA discontinued showed significant effects in the MUA, while CLONI and REMI did not. The slope (estimates) indicated the direction of the effect. In this sense, the effect of the PSA agents in CV indicates an increase, while the slope for MUA indicates a decrease. These findings supported our previous analysis, except CLONI, where effects identified in the *t*-test were not significant in this analysis.

In addition to effects of PSA agents, we found that electrophysiological measures were affected by several clinical and demographic variables. Disease duration showed a negative effect on CV, thus CV decreased as disease advanced. Age showed a significant impact on MUA, thus MUA decreased with increased age. Hemisphere (left or right) was also significantly associated with MUA, thus right hemisphere showed higher power than the left hemisphere. Finally, STN depth had a significant impact on MUA (Table 2).

Taking into consideration that we were working with multiple observations within each patient, we also conducted a linear regression model with a random effect grouped by patient for all three electrophysiological measures. This more conservative analysis brings a reduction in the statistical power, nevertheless for CV, the effect of DEX, REMI and disease duration remained significant. For MUA, the effect of DEX, PSA discontinued, hemisphere and age also remained significant. The non-significant effect on firing rate also remained (supplementary material Table S1).
