SSP

I first looked at the data to see the utility of using diagnosis and/or SSP scores in combination and how that might affect outcomes. First, I wanted to start by just seeing what kind of numbers we are talking about. How many cases of those with sensory issues are we missing by only inlcuding ASD? How many cases with disorders have elevated scores on SSP? Getting more information can help inform how to handle the outcome variable for SAND analysis.

Who is TD and has Sensory Issues?

First, let’s see how many cases with sensory issues we leave out by only using ASD as the outcome.

TD and Overall Sensory Issues

I start by looking at cases of TD who are scoring at clinical levels on Total SSP score.

Counts of Those Scoring Less Than or Equal to 141 (Definite Difference) on Total SSP By Disorder

Figure 1: These are the counts of kids by disorder and Total SSP score
	ASD	PMS	TD
Number with Disorder and Clinical Score	29	12	1
Total	53	28	44
Percentage with Clinical Score of Total	0.55	0.43	0.02

We can see there is one TD case with a Total Score indicating Definite Senory Issues. We can see this is participant with Stoelting ID 1047. Looking at her scores it appears she is missing an SP_VisualDomain score. Her SP_Total score is a sum of the remainder of the scores. Would she have scored out of clinical range if that score was included?

Counts of Those Scoring Less Than or Equal to 154 (Probable Difference) on Total SSP By Disorder

We may also consider those with TD scoring in a probable sensory range.

Figure 2: These are the counts of kids by disorder and Total SSP score
	ASD	PMS	TD
Number with Disorder and Clinical Score	39	18	1
Total	53	28	44
Percentage with Clinical Score of Total	0.74	0.64	0.02

Again, we can see that there is still only the one case. So, it seems that according to SP, there are very few TD cases that have overall sensory disturbances. Note also, that those with disorders are not scoring above Total SSP Clinical scores- either definite or probable at high rates.

TD and any Sensory Issue

We looked at the relationahip between disorders and overall sensory differences. We should also consider disorders and any sensory difference to see who is TD and has any type of sensory issue. To do this, I calculated number of cases who had a particular disorder and were below critical scores in the following SSP domains.

Disorder and any Definite Sensory Issue

First I looked at who met criteria for definite difference in any of the SSP scales. See the code for example of specific scores used.

ssp_any <- unlist(lapply(sheet_name,
         function(sheet){
           min_col <- which(colnames(sheet) == "SP_VisualDomain")
           max_col <- which(colnames(sheet) == "SSP_VIsAud")
           nrow(sheet[(apply(sheet, 1, function(sheet) !all(is.na(sheet[min_col:max_col])))),])
         }
  )
)

mapply(function(disorder, total){
  sand_all %>%
  filter(Dx == disorder & (SSP_Tactile <= 26 | SSP_Taste <= 11 | SSP_Mvmt <= 10 | SSP_UndrSeek <= 23 | SSP_AudFilt <= 19 | SSP_LowEnWeak <= 23 | SSP_VIsAud <= 15)) %>%
  dplyr::summarize("Number with Disorder and any Elevated Clinical Score" = n(), "Total" = total, "Percentage with Clinical Score of Total" = round(n()/total,2))  
}, dx_order, ssp_any) %>%
  kable(caption = "Figure 3: These are the counts of kids by disorder and Meeting Definite Criteria for any SSP Scale", col.names = dx_order) %>%
  kable_styling(bootstrap_options = c("striped", "hover", "condensed", "responsive"))%>%
  column_spec(4, background = "yellow")

Figure 3: These are the counts of kids by disorder and Meeting Definite Criteria for any SSP Scale
	ASD	PMS	TD
Number with Disorder and any Elevated Clinical Score	45	26	4
Total	58	28	45
Percentage with Clinical Score of Total	0.78	0.93	0.09

We can see there are now 4 kids who are TD and have any SSP definite elevated score. Let’s do the same and consider who met probable criteria on any scale.

Disorder and any Probable Sensory Issue

Figure 4: These are the counts of kids by disorder and and Meeting Probable Criteria for any SSP Scale
	ASD	PMS	TD
Number with Disorder and any ‘Probable’ Elevated Clinical Score	52	27	12
Total	58	28	45
Percentage with Clinical Score of Total	0.9	0.96	0.27

There’s now 12 TD cases who had any probable SSP scale difference.

But is that who we really want to reach with SAND? Do we expect SAND to show differences between TD cases with no sensory issue and a TD case with a probable movement or low energy difference?

Disorder and any Definite Sensory Tactile, Auditory, or Visual Scale

It would also be helpful to consider who is TD and has a sensory issue in the scales SAND measures. So, I looked at who was TD and had a definite difference in Tactile or Visual/Auditory Sensitivity.

Figure 5: These are the counts of kids by disorder and Tactile, Visual, Auditory Definite Elevated Scores
	ASD	PMS	TD
Number with Disorder and Definite Difference in Tactile, Visual, Auditory	14	4	1
Total	52	28	43
Percentage with Clinical Score of Total	0.27	0.14	0.02

There was one TD case who had either a definite Tactile, Auditory, or Visual difference. Interestingly, it seems it was the same girl, Stoelting ID 1047. I looked at her record to try to glean what could be occurring. There was no record for SAND scores. She had a VIQ of 125 and NVIQ of 105, so there was quite a lot of discrepancy there. Perhaps a visual issue?

Disorder and any Probable Sensory Tactile, Auditory, or Visual Scale

I also considered if there were TD cases with any probable Tactile, Visual, or Auditory issue.

Figure 6: These are the counts of kids by disorder and Tactile, Visual, Auditory Probable Elevated Scores
	ASD	PMS	TD
Number with Disorder and any ‘Probable’ Elevated Clinical Score	29	8	6
Total	52	28	43
Percentage with Clinical Score of Total	0.56	0.29	0.14

There are a few more TD Cases with probable Visual, Tactile, or Auditory (VTA) issues. But the other thing to notice is that the percentage of ASD and PMS kids who had even a probably VTA issue is quite low. That leads me to think about a few things:

SAND had a much stronger ability to categorize ASD using just Visual, Tactile, Auditory symptoms (see below). This suggests that ASD cases show strong symptoms in those areas, which SAND is correctly identifying.
SPP is not identifying well what they define as Visual, Tactile, Auditory symptoms in disordered cases. In fact, looking at all the above results together, SPP is not identifying that a lot of ASD and PMS cases have sensory issues.
Taken together, that suggests that SPP seems to be identifying sensory issues that may not be strongly related to ASD or PMS, but are of a different nature.
As it applies to our consideration, using SSP as an outcome measure, even restricting only to those with elevated VTA scales, would lead us to identify individuals with sensory issues, but which may not be autism-related issues.

Kids with Disorders without Sensory Issues

Conversly, in using only diagnosis as an outcome, we may also wish to consider if we are including kids with disorders who don’t have sensory issues. This would weaken the power of the SAND analysis if we are only using ASD as an outcome.

Disorders and Typical Performance

We want to identify how many people with a disorder are showing typical SSP performance.

Disorders and Typical Overall Score

I first looked to see who had a disorder and scored at or above 155 on Total SSP, indicating Typical Performance.

Figure 7: These are the counts of kids by disorder with Typical Overall Performance
	ASD	PMS	TD
Number with Disorder and Typical Score	14	9	43
Total	53	28	44
Percentage with Typical Score of Total	0.26	0.32	0.98

So, there’s a fair number of kids with a disorder who are showing a Typical Overall Score.

Kids with Disorders with Not Any Sensory Issues

Again, perhaps these cases with a disorder are only having elevated scores in one particular domain. I looked to see the number of cases of kids with disorders who had all typical SSP domain scores.

Figure 8: These are the counts of kids by disorder with all Typical Scores on SSP
	ASD	PMS	TD
Number with Disorder and all Typical Scores	7	3	32
Total	58	28	45
Percentage with Clinical Score of Total	0.12	0.11	0.71

SPP does a better job of identifying a sensory issue in any domain of those with disorders.

Kids with Disorders with Not Any Visual, Tactile, Auditory Sensory Issues

As before, we want to consider who might have a disorder but not an elevated Visual, Tactile, or Auditory Score.

Figure 9: These are the counts of kids by disorder with all Typical VTA Scores on SSP
	ASD	PMS	TD
Number with Disorder and all Typical VTA Scores	10	4	2
Total	52	28	43
Percentage with Clinical Score of Total	0.19	0.14	0.05

SAND Classification

Let’s see how this compares with SAND’s cutoff scores and disorders.

Classification and Total SAND Score

Total SAND at or above 16

First, let’s consider who scored at or above the SAND overall cutoff score of 16.

Figure 10: These are the counts of kids by disorder at or above 16 on SAND
	ASD	PMS	TD
Number with Disorder and SAND Total Score	51	28	1
Total	55	30	36
Percentage with Clinical Score of Total	0.93	0.93	0.03

The percentage of people with disorders that SAND identifies by Total Score is much higher than SSP (see Figure 1).

Total SAND at or above 13

First, let’s consider who scored at or above the SAND overall cutoff score of 13.

Figure 11: These are the counts of kids by disorder at or above 13 on SAND
	ASD	PMS	TD
Number with Disorder and SAND Total Score	52	30	3
Total	55	30	36
Percentage with Clinical Score of Total	0.95	1	0.08

This compares similarly with Figure 2.

SAND Behavior Scores

Definite Differences on Behavior

Next, we consider how many people scored at or above the definite clinical cutoff of 8 on any SAND Behavior (Hyper, Hypo, Seeking).

Figure 12: These are the counts of kids by disorder at or above 8 on at least one Sensory Behavior
	ASD	PMS	TD
Number with Disorder and SAND Behavior Score	49	30	3
Total	55	30	36
Percentage with Clinical Score of Total	0.89	1	0.08

Probable Differences on Behavior

Figure 13: These are the counts of kids by disorder at or above 7 on at least one Sensory Behavior
	ASD	PMS	TD
Number with Disorder and SAND Behavior Score	53	30	8
Total	55	30	36
Percentage with Clinical Score of Total	0.96	1	0.22

SAND Modality Scores

SAND Modality Definite Differences

We see how many scored above 8 on any Sensory Modality (Vis, Aud, Tactile)

Figure 14: These are the counts of kids by disorder at or above 8 on at least one Sensory Modality
	ASD	PMS	TD
Number with Disorder and Sand Modality Score	47	27	3
Total	55	30	36
Percentage with Clinical Score of Modality Scale	0.85	0.9	0.08

SAND Modality Probable Differences

Figure 15: These are the counts of kids by disorder at or above 8 on at least one Sensory Modality
	ASD	PMS	TD
Number with Disorder and Sand Modality Score	49	29	3
Total	55	30	36
Percentage with Clinical Score of Behavior Scale	0.89	0.97	0.08

SAND Behavior x Modality Domain

These are the counts of those who scored at least 5 on any Behavior x Modality

Figure 16: These are the counts of kids by disorder at or above 5 on at least one Sensory Behavior x Modality
	ASD	PMS	TD
Number with Disorder and Sand Behavior x Modality Score	51	30	6
Total	55	30	36
Percentage with Clinical Score on Behavior x Modality Scale	0.93	1	0.17

Summary

To answer the question about what outcome measures to use, SSP or diagnosis, first, I examined how many cases had elevated SSP scores but were TD. This would show how many cases we are leaving out by only using Dx as outcome. As was shown, there was only one case with a definite score total score above cutoff (though the data were questionable for this case). There was also only one case that had a score above definite difference cutoff criteria on domains that overlap with SAND. When using probable cutoff critera, there was still only one case with a total score above criteria, while there were 6 TD cases with a score above probable criteria in VTA domains.

I also considered how many kids are we including who have a disorder but a typical score? There are substantial percentages of kids with disorders who score typical on SSP Total. Those numbers lower when looking at those who score high on any VTA domain. According to SSP only 13% of ASD and PMS kids had a typical score in all of those domains. So, we may be including some kids with typical skills in those domains by only including Dx. On the other hand, that could be a limitation of SSP to identify all of those differences in those groups.

To see how this compares with SAND, I also got the counts based on disorders and scoring for SAND scores. Using Total SAND Score, both definite and probable, identifies a higher percentage of those with disorders than SSP, and also performs well when considering the scales and subscales. There are a small number of TD cases it flags with disorders.

SAND_SSP_Analysis

Brian Syzdek

Report Date: 2020-08-13

SSP