MPox
Overview of dataset
Missing data
By outcome
| variable | PosMpox | NegMPox |
|---|---|---|
| contactwMpox | 45 (60.8%) | 33 (45.8%) |
| Sex | 29 (39.2%) | 38 (52.8%) |
| orientation | 23 (31.1%) | 21 (29.2%) |
| LGB | 23 (31.1%) | 21 (29.2%) |
| HIV | 17 (23%) | 39 (54.2%) |
| pain | 5 (6.8%) | 11 (15.3%) |
| itching | 4 (5.4%) | 12 (16.7%) |
| symptoms | 3 (4.1%) | 9 (12.5%) |
| SID | 0 (0%) | 0 (0%) |
| Age | 0 (0%) | 0 (0%) |
| Gender | 0 (0%) | 0 (0%) |
| Male | 0 (0%) | 0 (0%) |
| lesions | 0 (0%) | 2 (2.8%) |
| genitallesions | 0 (0%) | 5 (6.9%) |
| facelesions | 0 (0%) | 5 (6.9%) |
| extlesions | 0 (0%) | 5 (6.9%) |
| trunklesions | 0 (0%) | 5 (6.9%) |
| systemic | 0 (0%) | 5 (6.9%) |
Correlations
Univariate (w/ missing data)
Technically some of the graphs below are bivariate, but a more formal analysis is in the next section
##
## Overall
## n 146
## Mpox = TRUE (%) 74 (50.7)
## Age (mean (SD)) 35.90 (14.49)
## Gender = F/M (%) 33/113 (22.6/77.4)
## orientation (%)
## BISEXUAL 4 ( 2.7)
## HETEROSEXUAL 49 (33.6)
## HOMOSEXUAL 49 (33.6)
## NA 44 (30.1)
## HIV (%)
## FALSE 36 (24.7)
## TRUE 54 (37.0)
## NA 56 (38.4)
## Sex (%)
## FALSE 31 (21.2)
## TRUE 48 (32.9)
## NA 67 (45.9)
## symptoms (median [IQR]) 4.00 [2.00, 7.00]
## lesions (%)
## MULTIPLE 132 (90.4)
## NONE 2 ( 1.4)
## SINGLE 10 ( 6.8)
## NA 2 ( 1.4)
## genitallesions (%)
## FALSE 72 (49.3)
## TRUE 69 (47.3)
## NA 5 ( 3.4)
## facelesions (%)
## FALSE 89 (61.0)
## TRUE 52 (35.6)
## NA 5 ( 3.4)
## extlesions (%)
## FALSE 44 (30.1)
## TRUE 97 (66.4)
## NA 5 ( 3.4)
## trunklesions (%)
## FALSE 81 (55.5)
## TRUE 60 (41.1)
## NA 5 ( 3.4)
## pain (%)
## FALSE 52 (35.6)
## TRUE 78 (53.4)
## NA 16 (11.0)
## itching (%)
## FALSE 93 (63.7)
## TRUE 37 (25.3)
## NA 16 (11.0)
## contactwMpox (%)
## FALSE 57 (39.0)
## TRUE 11 ( 7.5)
## NA 78 (53.4)
## systemic (%)
## FALSE 60 (41.1)
## TRUE 81 (55.5)
## NA 5 ( 3.4)“NA” in the table above corresponds to “unknown” in the spreadsheet. Each variable is explored in more detail in the sections below
MPox
Our primary outcome is a positive test for MPox. This
outcome did not have any missing data (as would be expected)
| Mpox | n | Percent |
|---|---|---|
| FALSE | 72 | 49.3% |
| TRUE | 74 | 50.7% |
Gender
Gender did not have missing data, and
the majority of included patients were male
| Gender | n | Percent |
|---|---|---|
| F | 33 | 23% |
| M | 113 | 77% |
Orientation
Orientation did have some missing data,
and contains multiple levels:
| orientation | n | Percent |
|---|---|---|
| BISEXUAL | 4 | 2.7% |
| HETEROSEXUAL | 49 | 33.6% |
| HOMOSEXUAL | 49 | 33.6% |
| UNKNOWN | 44 | 30.1% |
However if you cross-tab this with
gender, you can simplify these two groups
(because there were no LGBT+ women)
| BISEXUAL | HETEROSEXUAL | HOMOSEXUAL | UNKNOWN | |
|---|---|---|---|---|
| F | 0 | 25 | 0 | 8 |
| M | 4 | 24 | 49 | 36 |
| SexOrient | n | Percent |
|---|---|---|
| Female | 33 | 22.6% |
| Male Hetero | 24 | 16.4% |
| Male MSM | 53 | 36.3% |
| Male UNK | 36 | 24.7% |
We will have to be careful how we analyze sexual
orientation +
gender, since it’s prone to
multicollinearity / a singularity
Furthermore, we make an assumption that some of the men whose
orientation were not recorded denied
sexual exposures with other men. This could be done in a sensitivity
analysis…
Assumption: If sexual orientation
(orientation) for a man was not recorded, assume they are
heterosexual
| SexOrient | n | Percent |
|---|---|---|
| Female | 33 | 22.6% |
| MSM | 53 | 36.3% |
| Male | 60 | 41.1% |
HIV
A fair number of patients have missing data for
HIV status. It’s unclear if this just
wasn’t asked (will need to clarify how the data was recorded)
| HIV | n | Percent |
|---|---|---|
| N | 36 | 24.7% |
| U | 56 | 38.4% |
| Y | 54 | 37.0% |
Need to clarify what the difference between “N” and “U” means. Also need to clarify if HIV status was determined before or after the swab was sent
Sex (recent sex)
A fair number of patients have missing data for recent
sex (not to be confused with the patient’s
gender). As with HIV status, it’s unclear
how to handle missing (“unknown”) data here; was it only recorded when
it lined up with the clinical impression?
| Sex | n | Percent |
|---|---|---|
| N | 31 | 21% |
| U | 67 | 46% |
| Y | 48 | 33% |
Pain
| pain | n | Percent |
|---|---|---|
| N | 52 | 36% |
| U | 16 | 11% |
| Y | 78 | 53% |
Itching
| itching | n | Percent |
|---|---|---|
| N | 93 | 64% |
| U | 16 | 11% |
| Y | 37 | 25% |
Systemic symptoms
| systemic | n | Percent |
|---|---|---|
| N | 60 | 41% |
| U | 5 | 3% |
| Y | 81 | 55% |
MPox contact
| contactwMpox | n | Percent |
|---|---|---|
| N | 57 | 39% |
| U | 78 | 53% |
| Y | 11 | 8% |
Age
Duration of symptoms
Lesions - TODO
Bivariate
## Stratified by Mpox
## FALSE TRUE p test
## n 72 74
## Mpox = TRUE (%) 0 ( 0.0) 74 (100.0) <0.001
## Age (mean (SD)) 40.00 (17.95) 31.92 (8.42) 0.001
## Gender = M (%) 41 (56.9) 72 ( 97.3) <0.001
## orientation (%) <0.001
## BISEXUAL 2 ( 3.9) 2 ( 3.9)
## HETEROSEXUAL 39 (76.5) 10 ( 19.6)
## HOMOSEXUAL 10 (19.6) 39 ( 76.5)
## HIV = TRUE (%) 14 (42.4) 40 ( 70.2) 0.018
## Sex = TRUE (%) 13 (38.2) 35 ( 77.8) 0.001
## symptoms (mean (SD)) 14.19 (46.96) 5.47 (5.08) 0.122
## lesions (%) 0.119
## MULTIPLE 61 (87.1) 71 ( 95.9)
## NONE 2 ( 2.9) 0 ( 0.0)
## SINGLE 7 (10.0) 3 ( 4.1)
## genitallesions = TRUE (%) 21 (31.3) 48 ( 64.9) <0.001
## facelesions = TRUE (%) 17 (25.4) 35 ( 47.3) 0.012
## extlesions = TRUE (%) 49 (73.1) 48 ( 64.9) 0.381
## trunklesions = TRUE (%) 29 (43.3) 31 ( 41.9) 1.000
## pain = TRUE (%) 21 (34.4) 57 ( 82.6) <0.001
## itching = TRUE (%) 25 (41.7) 12 ( 17.1) 0.004
## contactwMpox = TRUE (%) 2 ( 5.1) 9 ( 31.0) 0.011
## systemic = TRUE (%) 24 (35.8) 57 ( 77.0) <0.001Two by Two numbers
Each subsection here will print off four parts:
- Contingency table where the “Outcome” is a MPox case, and the “Test” is the risk factor (e.g. “Test +” for itching means that the patient was reported as having itching)
- Diagnostic characteristics (e.g. sensitivity, specificity, PPV, NPV, PLR, NLR) with associated 95% CI’s
- Relative risk with associated 95% CI
- Odds ratio with associated 95% CI
Male gender
## Outcome + Outcome - Total
## Test + 72 41 113
## Test - 2 31 33
## Total 74 72 146
##
## Point estimates and 95% CIs:
## --------------------------------------------------------------
## Apparent prevalence * 0.77 (0.70, 0.84)
## True prevalence * 0.51 (0.42, 0.59)
## Sensitivity * 0.97 (0.91, 1.00)
## Specificity * 0.43 (0.31, 0.55)
## Positive predictive value * 0.64 (0.54, 0.73)
## Negative predictive value * 0.94 (0.80, 0.99)
## Positive likelihood ratio 1.71 (1.39, 2.10)
## Negative likelihood ratio 0.06 (0.02, 0.25)
## False T+ proportion for true D- * 0.57 (0.45, 0.69)
## False T- proportion for true D+ * 0.03 (0.00, 0.09)
## False T+ proportion for T+ * 0.36 (0.27, 0.46)
## False T- proportion for T- * 0.06 (0.01, 0.20)
## Correctly classified proportion * 0.71 (0.62, 0.78)
## --------------------------------------------------------------
## * Exact CIs## Relative risk for 'Male': 10.513 (3.219 - 38.187)## Odds ratio for 'Male': 27.22 (6.193 - 119.626)LGB sexual orientation
## Outcome + Outcome - Total
## Test + 41 12 53
## Test - 10 39 49
## Total 51 51 102
##
## Point estimates and 95% CIs:
## --------------------------------------------------------------
## Apparent prevalence * 0.52 (0.42, 0.62)
## True prevalence * 0.50 (0.40, 0.60)
## Sensitivity * 0.80 (0.67, 0.90)
## Specificity * 0.76 (0.63, 0.87)
## Positive predictive value * 0.77 (0.64, 0.88)
## Negative predictive value * 0.80 (0.66, 0.90)
## Positive likelihood ratio 3.42 (2.05, 5.71)
## Negative likelihood ratio 0.26 (0.14, 0.46)
## False T+ proportion for true D- * 0.24 (0.13, 0.37)
## False T- proportion for true D+ * 0.20 (0.10, 0.33)
## False T+ proportion for T+ * 0.23 (0.12, 0.36)
## False T- proportion for T- * 0.20 (0.10, 0.34)
## Correctly classified proportion * 0.78 (0.69, 0.86)
## --------------------------------------------------------------
## * Exact CIs## Relative risk for 'LGB': 3.791 (2.237 - 6.842)## Odds ratio for 'LGB': 13.325 (5.17 - 34.345)HIV status
## Outcome + Outcome - Total
## Test + 40 14 54
## Test - 17 19 36
## Total 57 33 90
##
## Point estimates and 95% CIs:
## --------------------------------------------------------------
## Apparent prevalence * 0.60 (0.49, 0.70)
## True prevalence * 0.63 (0.53, 0.73)
## Sensitivity * 0.70 (0.57, 0.82)
## Specificity * 0.58 (0.39, 0.75)
## Positive predictive value * 0.74 (0.60, 0.85)
## Negative predictive value * 0.53 (0.35, 0.70)
## Positive likelihood ratio 1.65 (1.07, 2.55)
## Negative likelihood ratio 0.52 (0.32, 0.85)
## False T+ proportion for true D- * 0.42 (0.25, 0.61)
## False T- proportion for true D+ * 0.30 (0.18, 0.43)
## False T+ proportion for T+ * 0.26 (0.15, 0.40)
## False T- proportion for T- * 0.47 (0.30, 0.65)
## Correctly classified proportion * 0.66 (0.55, 0.75)
## --------------------------------------------------------------
## * Exact CIs## Relative risk for 'HIV': 1.569 (1.109 - 2.374)## Odds ratio for 'HIV': 3.193 (1.307 - 7.804)Recent sex
## Outcome + Outcome - Total
## Test + 35 13 48
## Test - 10 21 31
## Total 45 34 79
##
## Point estimates and 95% CIs:
## --------------------------------------------------------------
## Apparent prevalence * 0.61 (0.49, 0.72)
## True prevalence * 0.57 (0.45, 0.68)
## Sensitivity * 0.78 (0.63, 0.89)
## Specificity * 0.62 (0.44, 0.78)
## Positive predictive value * 0.73 (0.58, 0.85)
## Negative predictive value * 0.68 (0.49, 0.83)
## Positive likelihood ratio 2.03 (1.29, 3.21)
## Negative likelihood ratio 0.36 (0.20, 0.66)
## False T+ proportion for true D- * 0.38 (0.22, 0.56)
## False T- proportion for true D+ * 0.22 (0.11, 0.37)
## False T+ proportion for T+ * 0.27 (0.15, 0.42)
## False T- proportion for T- * 0.32 (0.17, 0.51)
## Correctly classified proportion * 0.71 (0.60, 0.81)
## --------------------------------------------------------------
## * Exact CIs## Relative risk for 'Sex': 2.26 (1.395 - 4.013)## Odds ratio for 'Sex': 5.654 (2.109 - 15.156)Pain
## Outcome + Outcome - Total
## Test + 57 21 78
## Test - 12 40 52
## Total 69 61 130
##
## Point estimates and 95% CIs:
## --------------------------------------------------------------
## Apparent prevalence * 0.60 (0.51, 0.68)
## True prevalence * 0.53 (0.44, 0.62)
## Sensitivity * 0.83 (0.72, 0.91)
## Specificity * 0.66 (0.52, 0.77)
## Positive predictive value * 0.73 (0.62, 0.82)
## Negative predictive value * 0.77 (0.63, 0.87)
## Positive likelihood ratio 2.40 (1.67, 3.45)
## Negative likelihood ratio 0.27 (0.15, 0.46)
## False T+ proportion for true D- * 0.34 (0.23, 0.48)
## False T- proportion for true D+ * 0.17 (0.09, 0.28)
## False T+ proportion for T+ * 0.27 (0.18, 0.38)
## False T- proportion for T- * 0.23 (0.13, 0.37)
## Correctly classified proportion * 0.75 (0.66, 0.82)
## --------------------------------------------------------------
## * Exact CIs## Relative risk for 'pain': 3.167 (1.972 - 5.405)## Odds ratio for 'pain': 9.048 (3.999 - 20.468)Itching
## Outcome + Outcome - Total
## Test + 12 25 37
## Test - 58 35 93
## Total 70 60 130
##
## Point estimates and 95% CIs:
## --------------------------------------------------------------
## Apparent prevalence * 0.28 (0.21, 0.37)
## True prevalence * 0.54 (0.45, 0.63)
## Sensitivity * 0.17 (0.09, 0.28)
## Specificity * 0.58 (0.45, 0.71)
## Positive predictive value * 0.32 (0.18, 0.50)
## Negative predictive value * 0.38 (0.28, 0.48)
## Positive likelihood ratio 0.41 (0.23, 0.75)
## Negative likelihood ratio 1.42 (1.12, 1.80)
## False T+ proportion for true D- * 0.42 (0.29, 0.55)
## False T- proportion for true D+ * 0.83 (0.72, 0.91)
## False T+ proportion for T+ * 0.68 (0.50, 0.82)
## False T- proportion for T- * 0.62 (0.52, 0.72)
## Correctly classified proportion * 0.36 (0.28, 0.45)
## --------------------------------------------------------------
## * Exact CIs## Relative risk for 'itching': 0.52 (0.308 - 0.809)## Odds ratio for 'itching': 0.29 (0.129 - 0.649)Systemic symptoms
## Outcome + Outcome - Total
## Test + 57 24 81
## Test - 17 43 60
## Total 74 67 141
##
## Point estimates and 95% CIs:
## --------------------------------------------------------------
## Apparent prevalence * 0.57 (0.49, 0.66)
## True prevalence * 0.52 (0.44, 0.61)
## Sensitivity * 0.77 (0.66, 0.86)
## Specificity * 0.64 (0.52, 0.76)
## Positive predictive value * 0.70 (0.59, 0.80)
## Negative predictive value * 0.72 (0.59, 0.83)
## Positive likelihood ratio 2.15 (1.52, 3.03)
## Negative likelihood ratio 0.36 (0.23, 0.56)
## False T+ proportion for true D- * 0.36 (0.24, 0.48)
## False T- proportion for true D+ * 0.23 (0.14, 0.34)
## False T+ proportion for T+ * 0.30 (0.20, 0.41)
## False T- proportion for T- * 0.28 (0.17, 0.41)
## Correctly classified proportion * 0.71 (0.63, 0.78)
## --------------------------------------------------------------
## * Exact CIs## Relative risk for 'systemic': 2.484 (1.669 - 3.878)## Odds ratio for 'systemic': 6.007 (2.876 - 12.55)Mpox contact
## Outcome + Outcome - Total
## Test + 9 2 11
## Test - 20 37 57
## Total 29 39 68
##
## Point estimates and 95% CIs:
## --------------------------------------------------------------
## Apparent prevalence * 0.16 (0.08, 0.27)
## True prevalence * 0.43 (0.31, 0.55)
## Sensitivity * 0.31 (0.15, 0.51)
## Specificity * 0.95 (0.83, 0.99)
## Positive predictive value * 0.82 (0.48, 0.98)
## Negative predictive value * 0.65 (0.51, 0.77)
## Positive likelihood ratio 6.05 (1.41, 25.92)
## Negative likelihood ratio 0.73 (0.56, 0.94)
## False T+ proportion for true D- * 0.05 (0.01, 0.17)
## False T- proportion for true D+ * 0.69 (0.49, 0.85)
## False T+ proportion for T+ * 0.18 (0.02, 0.52)
## False T- proportion for T- * 0.35 (0.23, 0.49)
## Correctly classified proportion * 0.68 (0.55, 0.78)
## --------------------------------------------------------------
## * Exact CIs## Relative risk for 'contactwMpox': 2.332 (1.361 - 3.593)## Odds ratio for 'contactwMpox': 8.325 (1.638 - 42.315)To do
- Lesions
- Numeric variables (age, duration of symptoms)
Multivariate - todo
To do
##
## Call:
## glm(formula = Mpox ~ Male, family = "binomial", data = df)
##
## Deviance Residuals:
## Min 1Q Median 3Q Max
## -1.4240 -1.4240 0.9494 0.9494 2.3678
##
## Coefficients:
## Estimate Std. Error z value Pr(>|z|)
## (Intercept) -2.7408 0.7296 -3.757 0.000172 ***
## MaleTRUE 3.3039 0.7553 4.374 1.22e-05 ***
## ---
## Signif. codes: 0 '***' 0.001 '**' 0.01 '*' 0.05 '.' 0.1 ' ' 1
##
## (Dispersion parameter for binomial family taken to be 1)
##
## Null deviance: 202.37 on 145 degrees of freedom
## Residual deviance: 163.13 on 144 degrees of freedom
## AIC: 167.13
##
## Number of Fisher Scoring iterations: 5