Quality of Life Assessed Using Skindex-16 Scores Among Patients With Acne Receiving Isotretinoin Treatment

Key Points

Question

By what degree and time frame do patients with acne believe their quality of life improves while receiving isotretinoin?

Findings

In this case series study of 57 patients receiving isotretinoin for treatment of acne, the patients perceived rapid and significant improvements in how acne affects their quality of life, especially emotional improvements. Compared with baseline, more than 50% improvement was noticed by patients within 2 months.

Meaning

Isotretinoin is widely accepted as the most effective treatment for acne; these quality-of-life data may provide insight into how patients with acne perceive their response to isotretinoin treatment.

Abstract

Importance

Acne is a common dermatologic condition and significantly affects psychosocial health and quality of life. An international task force recommended routine use of quality-of-life measures for clinic visits associated with acne management, but this has yet to translate into clinical practice.

Objective

To assess mean Skindex-16 scores over time among patients with moderate to severe acne receiving isotretinoin treatment.

Design, Setting, and Participants

A longitudinal, retrospective case series study of Skindex-16 data collected at monthly visits from 57 consecutive patients with acne receiving isotretinoin; data were collected and evaluated between November 23, 2016, and January 22, 2019. Continuous variables were compared using quantile regression. Multivariable linear mixed models evaluated mean (95% CI) score trajectory over time.

Main Outcomes and Measure

Skindex-16 scores, including normalized scores for the emotional, symptomatic, and functional aspects of having skin disease as well as an overall score.

Results

Fifty-seven patients (31 [54.4 %] males, with median [interquartile range] age of 17.2 [15.9-18.1] years) in this case series study completed the Skindex-16 at baseline and at least once during follow-up. Baseline Skindex-16 scores were similar by sex but worse with increasing age. Emotional impact was more bothersome to patients with acne requiring isotretinoin treatment than either symptoms or functioning. Improvements of greater than 50% in overall and Emotional domain scores were seen by month 2 of receiving isotretinoin treatment (eg, overall scores decreased from 39.4 to 17.5 by month 2; a decrease of 22.0; P < .001). Qualitatively, Skindex-16 scores reached their nadir between months 3 and 5; at month 4, overall Skindex-16 scores showed a 4.4-fold improvement (from 39.4 at baseline to 8.9; P < .001) and Emotional domain scores showed a 4.8-fold improvement (from 57.7 at baseline to 11.9; P < .001).

Conclusions and Relevance

The findings of this case series suggest that patients receiving isotretinoin treatment achieve greater than a 50% improvement in quality of life by month 2 and can expect approximately 4-fold to 5-fold improvements from baseline with a full course of isotretinoin. This study shows the potential of routine administration of quality of life measures to assess patient care in dermatology.

This case series uses Skindex-16 scores to assess the association between having skin disease and quality of life among patients with acne who are receiving isotretinoin treatment.

Introduction

Acne vulgaris is prevalent, affecting nearly 75% of teenagers and young adults.¹ It affects visible body areas, such as the face and neck, and significantly affects psychosocial health and quality of life (QOL). The magnitude of this impact is similar to that seen among patients with asthma, epilepsy, diabetes, or arthritis.²

Several dermatologists scattered around the world are encouraging routine efforts to assess and quantify the QOL impact on patients in an effort to optimize both patient care and expectations.³ Recently, the Acne Core Outcomes Research Network identified key aspects associated with symptoms and QOL to standardize both physician-reported and patient-reported outcomes (PROs) in clinical trials.⁴ An international task force⁵ recently recommended routine clinical use of QOL measures for patients with acne. With increasing use of technology in health care delivery, including automated appointment reminders via text, electronic check-in kiosks, and electronic tablets for photographs and electronic signatures, the ability for patients to complete real-time PRO assessments is easier. However, very few dermatology clinics use QOL assessments in practice for various reasons, including concerns about slowing clinic flow and unfamiliarity with assessments.^6,7,8,9 To our knowledge, 6 acne-specific QOL assessments^{10,11,12,13,14,15,16} and 2 key dermatology-specific QOL assessments (Skindex-16 and Dermatology Life Quality Index [DLQI]),^17,18 have been validated. Each has its own scale and score interpretation.

Our proof-of-concept, case series study aimed primarily to show that routine QOL assessment is responsive to isotretinoin treatment by assessing results of estimated mean Skindex-16 scores over time. Secondarily, we explored differences in isotretinoin response by sex and age. These data can inform physicians and patients about the monthly association of QOL with receiving isotretinoin treatment, improving patient expectations and educational counseling.

Methods

Data Collection

This study was a longitudinal, retrospective case series investigating change in Skindex-16 scoring during the isotretinoin treatment course for acne. Skindex-16 scores were collected at baseline and at monthly follow-up visits. Additional data included patient age, sex, weight, acne type (comedonal, inflammatory, or mixed), acne severity (moderate, moderate to severe, or severe) based on an acne investigator global assessment,¹⁹ presence of scarring, and daily isotretinoin doses (measured in mg/kg/d). Acne type and severity were assessed by a single board-certified dermatologist (A.M.S.) to decrease variability. The study was approved by the University of Utah Institutional Review Board, Salt Lake City, which also waived the need for patient informed consent because it deemed of minimal risk given no patient contact and only a retrospective medical record review.

All patients were seen at a suburban, community-based, academic dermatology clinic located 15 miles north of Salt Lake City, Utah, between November 23, 2016, and January 22, 2019. Since 2016, as part of routine clinical care at our institution, all patients were given an electronic tablet at visit check-in that administers the Skindex-16,⁸ with scores uploaded automatically to the electronic medical record and immediately available for review by the dermatologist during the clinic visit.

Skindex-16 was adopted by our institution because this assessment has excellent validity, measurement model characteristics, reliability, and minimal administrative burden.²⁰ Patients aged 12 years or older who had acne were included if they completed a baseline Skindex-16 before initiating treatment with isotretinoin and at least 1 additional Skindex-16 while receiving isotretinoin. Skindex-16 has not been validated in adolescents but is used more commonly in the academic literature than Skindex-Teen.²¹ Thus, we opted to use Skindex-16 to maintain score comparability in the cohort and with the extant literature.²²

Outcome Variables

Skindex-16 consists of domain scores that assess how symptoms, emotions, and functioning from the skin issue affect the QOL of patients with acne. The overall score averages the 3 domain scores, all of which are normalized to a 0 to 100 scale, where 0 indicates that their skin condition has no impact on QOL and 100 represents maximal impact on QOL for the worse. The change in mean overall Skindex-16 score for the overall group over time was the primary outcome of interest, with the change in mean domain scores as secondary outcomes. Interactions between scores and patients’ sex and age, as well as the impact of demographic or disease characteristics with Skindex-16 scores were also evaluated.

Descriptive Statistics

Sex differences have been proposed in acne’s pathophysiology and shown in response to treatment and to QOL assessment scores.^23,24,25,26 Therefore, we compared baseline characteristics by sex. Continuous variables were reported as medians with interquartile ranges and were compared using quantile regression. Categorical variables were compared using χ². A power analysis was not performed.

These data represent multiple observations through time on different individuals; thus, our analysis approach was similar to growth-curve analysis.²⁷ Since each person varied in how acne affected their QOL at baseline and had different trajectories over time, this variation was accounted for when calculating an overall score trajectory over time. To perform this, we used linear mixed models with maximum likelihood estimation. The results were reported as estimated mean scores for the group after accounting for variation in baseline scores (random intercept effects) and random variation in slope (random slope effects) as well as other covariates, similar to multivariable regression. Time-invariant covariates included sex, age, baseline acne type, final isotretinoin dose, baseline acne severity, and presence of scarring. Because isotretinoin dose was titrated based on tolerability up to a total target goal of 120 mg/kg, the final isotretinoin dose effectively separated those patients able to tolerate a higher dose from those who could not, which was a possible confounder for treatment response.

Model Specification

An age- and sex-adjusted model with random intercept only was our base model. A more complete model added all time-invariant covariates. Additional models with sex and age interaction terms were also proposed to assess whether the slopes of mean Skindex-16 scores varied by sex or age, especially since Skindex-16 has not been validated in adolescents. Further sensitivity analyses assessed for inclusion of time, sex, and age as random slopes.

Nested models, containing different iterations of random slopes or interaction terms, were compared with baseline models using the likelihood ratio test. Models containing different covariates (ie, base model vs complete model) as well as models assessing quadratic or cubic terms were compared using the Akaike information criterion. A priori, an Akaike information criterion difference of less than 10 was considered minimal, and the models could be considered largely equivalent. This process was repeated for each Skindex-16 domain score. Assumptions of linearity, homogeneity of variance, and distribution of residuals (both of level 1 and random effects) were assessed graphically after model specification. No treatment failures were observed in this group. Missing data were assumed to be missing at random since they were primarily the result of logistical issues. Linear mixed models are robust and resistant to missing data because of their use of all observations and other modeling characteristics.^28,29,30

Final Trend Analysis

After specifying the final model, the slopes of Skindex-16 scores over time were assessed for direction and significance. Random effects representing variability around individuals’ baseline scores were also assessed. For ease of interpretation, estimated marginal means at each time point were assessed. These represented the estimated Skindex-16 score at each time point, with other explanatory variables held constant and random effects held at 0. Effectively, these represent the group’s mean score over time. A 2-sided P < .05 was considered significant for all analyses. All analyses were performed using Stata, version 14.2 (StataCorp LLC).

Results

Descriptive Data

During the study period, 69 patients were registered in iPLEDGE, a US-based national isotretinoin registry aimed at pregnancy prevention, in anticipation of initiating treatment with isotretinoin. Five patients (all female) decided not to initiate isotretinoin therapy during the 30-day waiting period. Seven patients completed Skindex-16 during follow-up but lacked a Skindex-16 score from before isotretinoin treatment began. Fifty-seven patients (31 [54%] male with a median [interquartile range] age of 17.2 [15.8-20.0] years) completed Skindex-16 at baseline and at least once during follow-up (baseline overall Skindex-16 score per patient: mean [SD], 40.5 [18.8]; range reported as absolute numbers, 2-80). Missing data from month to month were largely because of logistical issues of system onboarding by our staff (electronic tablets not charged, software glitches), not by patient refusal (overall refusal rate was 1.7%). Female patients received a higher initial and final relative weight-based daily isotretinoin dose; otherwise sex differences were not seen (Table 1). For both sexes, the Emotional domain subscore was highest and was the largest component of the overall score. Individual raw scores are plotted in eFigure in the Supplement.

Table 1. Demographic and Clinical Characteristics of 57 Patients Taking Isotretinoin for Moderate to Severe Acne.

Characteristic	Female	Male	P value
No. (%)	26 (45.6)	31 (54.4)
Age, median (IQR)a	17.3 (15.1-22.9)	17.2 (15.9-18.1)	.30
Isotretinoin dose, median (IQR), mg/kg
Initial dose	0.71 (0.63-0.75)	0.55 (0.51-0.64)	<.001
Final dose	1.05 (0.82-1.22)	0.93 (0.81-1.11)	<.001
Baseline Skindex-16 scores, median (IQR)
Overall	39.0 (29.0-52.0)	40.0 (19.0-57.0)	>.99
Emotional	64.0 (36.0-83.0)	59.8 (43.0-76.0)	.70
Symptoms	23.0 (13.0-42.0)	27.0 (17.0-38.0)	.20
Functional	16.5 (0-37.0)	17.0 (3.0-40.0)	.40
Follow-up visit, median (IQR)	4.0 (2.0-6.0)	4.0 (2.0-6.0)	>.99
Contraception, No. (%)
Abstinence	18 (69.2)	NA
OCPs and condom	5 (19.2)	NA
IUD and condom	2 (7.7)	NA
Tubal ligation and condom	1 (3.8)	NA
Acne type, No. (%)
Comedonal	2 (7.7)	0	.20
Mixed	21 (80.8)	25 (80.6)
Inflammatory	3 (11.5)	6 (19.4)
Acne severity, No. (%)
Moderate	9 (34.6)	10 (32.3)	.90
Moderate to severe	11 (42.3)	15 (48.4)
Severe	6 (23.1)	6 (19.4)
Scarring
No	11 (42.3)	15 (48.4)	.70
Yes	15 (57.7)	16 (51.6)

Abbreviations: IQR, interquartile range; IUD, intrauterine device; NA, not applicable or data not available; OCPs, oral contraceptive pills.

^aComparisons between medians were performed using quantile regression.

Final Data Analysis

For details regarding model selection, please refer to the eMethods in the Supplement. For each model, 219 observations were included. Decreases in overall Skindex-16 score (improvement in QOL) and domain scores on isotretinoin (based on negative slopes) were all significant over time (overall, −14.3; 95% CI, −17.0 to −11.6; P < .001; Emotional, −22.4; 95% CI, −25.8 to −19.0; P < .001; Symptoms, −6.7; 95% CI, −9.5 to −3.8; P < .001; and Functional, −16.5; 95% CI, −23.7 to −9.3; P < .001). In addition, the respective quadratic or cubic time terms were significant (overall, 1.7; 95% CI, 1.1-2.2; P < .001; Emotional, 2.7; 95% CI, 2.0-3.5; P < .001; Symptoms, 0.7; 95% CI, 0.1-1.3; P = .03; and Functional, 5.1; 95% CI, 1.6-8.6; P = .01 for the quadratic term, and −0.53; 95% CI, −0.99 to −0.1; P = .02 for the cubic term), suggesting a significant change in slope over time. A large random intercept SD suggested substantial variation of each person’s measurement from the group mean because of unmeasured variables (eFigure in the Supplement). Large within-subject SDs suggested significant variation in score change around each person’s mean score (ie, individual scores did not follow smooth trajectories through time).

To aid in interpretation of the curves shown in the Figure, estimated means were calculated for each score at each time point, while holding the other covariates constant (Table 2). Changes in score were most pronounced in the early months and stabilized or slightly reversed in the later months. Overall scores improved 55.7% by month 2 of isotretinoin treatment (from 39.4 to 17.5, a 22.0-point decrease; P < .001), with a mean overall benefit of 77.5% improvement in QOL for a standard course of isotretinoin (Figure, A and Table 2). Overall Skindex-16 scores reached a nadir of 8.9 at month 4 from 39.4 (a 4.4-fold decrease; P < .001). The nadir of 11.9 for Emotional domain scores also occurred at month 4 from a baseline of 57.7 (a 4.8-fold decrease; P < .001) (Table 2). Other domains showed similar associations, with the Symptoms domain showing a more modest decrease (Figure, B-D and Table 2).

Figure. Estimated Mean Skindex-16 Scores After Multivariable Adjustment.

The graphs show the mean progression of Skindex-16 scores after accounting for demographic and disease-based variables. Error bars indicate 95% CIs around the estimated mean.

Table 2. Estimated Overall Skindex-16 Score and Domain Scores After Treatment With Isotretinoin and at Each Monthly Follow-up Among 57 Patients^a.

Skindex-16 domain	Skindex-16 score, mean (95% CI)
	Before treatment (N = 57)	Month of treatment
		1 (n = 38)	2 (n = 44)	3 (n = 40)	4 (n = 26)	5 (n = 11)	6 (n = 3)
Overall	39.4 (35.9 to 43.0)	26.8 (23.8 to 29.8)	17.5 (14.3 to 20.7)	11.5 (8.3 to 14.8)	8.9 (5.3 to 12.4)	9.6 (4.4 to 14.7)	13.6 (5.2 to 22.0)
Symptoms	27.0 (23.3 to 30.7)	21.0 (17.9 to 24.1)	16.4 (13.1 to 19.6)	13.0 (9.7 to 16.4)	11.2 (7.4 to 14.7)	10.4 (5.0 to 15.9)	11.1 (2.2 to 20.1)
Emotional	57.7 (52.6 to 62.7)	38.0 (33.6 to 42.4)	23.8 (19.2 to 28.4)	15.1 (10.5 to 19.7)	11.9 (6.9 to 16.9)	14.2 (7.3 to 21.0)	21.9 (11.1 to 32.8)
Functional	24.8 (20.9 to 28.7)	12.9 (9.3 to 16.5)	8.0 (4.6 to 11.4)	6.9 (3.5 to 10.3)	6.4 (2.1 to 10.7)	3.3 (−2.7 to 9.2)b	−5.7 (−19.8 to 8.4)b

^aEstimated means were calculated from the multivariable model. Most patients reached their goal dose between months 4 and 5.

^bAlthough a negative value is not possible for Skindex-16 scores, negative values resulted from using a linear model for prediction. This number signified the continued decrease in the Functional domain score and, outside of predictions, would be bounded at 0. There are few observations and substantial variation around these later estimates.

Univariable and multivariable analyses explored the fixed effects of each demographic and clinical variable on overall Skindex-16 score within each domain score (information on interpreting regression coefficients is in a footnote to Table 3). The isotretinoin ending dosage (in mg/kg/d) was not associated with the overall score. Patients with inflammatory acne reported lower Skindex-16 scores (less QOL impact) compared with patients with comedonal and mixed-type acne. Patients with moderate to severe and severe acne had higher mean Skindex-16 scores than those with moderate acne. Similar associations were seen in the multivariable analysis of the Emotional domain score. For the symptom Skindex-16 domain, higher scores were associated with increasing age (Table 3).

Table 3. Univariable and Multivariable Linear Mixed Model Regressions.

Characteristic	Univariable model		Multivariable model
	Regression coefficient (95% CI)a	P value	Adjusted regression coefficient (adjusted 95% CI)b	Adjusted P value
Overall score
Age	0.85 (0.30 to 1.39)	.002	1.13 (0.54 to 1.72)	<.001
Isotretinoin final dose	2.56 (−11.41 to 16.54)	.72	7.92 (−4.35 to 20.19)	.21
Male sex	−3.03 (−10.08 to 4.03)	.40	1.41 (−4.90 to 7.72)	.66
Acne type
Comedonal	1 [Reference]		1 [Reference]
Mixed	−7.00 (−25.73 to 11.73)	.46	−7.97 (−23.60 to 7.65)	.32
Inflammatory	−6.66 (−27.02 to 13.70)	.52	–19.40 (−37.58 to −1.22)	.04
Acne severity
Moderate	1 [Reference]		1 [Reference]
Moderate to severe	8.20 (0.76 to 15.63)	.03	7.32 (0.59 to 14.04)	.03
Severe	13.70 (4.63 to 22.77)	.003	13.30 (4.61 to 22.09)	.003
Scarring	5.89 (–1.05 to 12.83)	.10	0.73 (–5.78 to 7.24)	.83
Emotional score
Age	0.76 (–0.04 to 1.55)	.06	1.10 (0.24 to 1.96)	.01
Isotretinoin end dose	10.65 (–8.79 to 30.09)	.28	14.60 (–3.40 to 32.62)	.11
Male sex	–6.23 (–16.07 to 3.60)	.21	–0.23 (–9.47 to 9.02)	.96
Acne type
Comedonal	1 [Reference]		1 [Reference]
Mixed	–21.60 (–47.45 to 4.31)	.10	–21.10 (–44.22 to 2.08)	.07
Inflammatory	–22.30 (–50.37 to 5.86)	.12	–31.70 (–58.53 to –4.81)	.02
Acne severity
Moderate	1 [Reference]		1 [Reference]
Moderate to severe	9.13 (–1.58 to 19.83)	.10	10.10 (0.18 to 19.90)	.05
Severe	16.60 (3.57 to 29.68)	.01	20.50 (7.67 to 33.38)	.002
Scarring	2.64 (–7.31 to 12.59)	.60	–4.30 (–13.86 to 5.26)	.38
Symptoms score
Age	0.67 (0.14 to 1.19)	.01	0.91 (0.31 to 1.50)	.003
Isotretinoin final dose	–13.60 (–26.33 to –0.88)	.04	–9.88 (–22.27 to 2.50)	.12
Male sex	0.71 (–6.00 to 7.40)	.84	1.33 (–5.03 to 7.70)	.68
Acne type
Comedonal	1 [Reference]		1 [Reference]
Mixed	9.74 (–7.68 to 27.15)	.27	5.83 (–9.86 to 21.53)	.47
Inflammatory	7.82 (–11.15 to 26.78)	.81	–6.20 (–24.49 to 12.08)	.51
Acne severity
Moderate	1 [Reference]		1 [Reference]
Moderate to severe	5.31 (–2.12 to 12.74)	.16	3.42 (–3.36 to 10.20)	.32
Severe	6.18 (–2.88 to 15.23)	.18	1.82 (–6.99 to 10.63)	.69
Scarring	4.91 (–1.66 to 11.49)	.14	3.12 (–3.44 to 9.69)	.35
Functional score
Age	1.08 (0.57 to 1.59)	<.001	1.32 (0.80 to 1.85)	<.001
Isotretinoin end dose	3.87 (–9.93 to 17.66)	.58	12.60 (1.65 to 23.57)	.02
Male sex	–1.36 (–8.37 to 5.65)	.70	3.49 (–2.17 to 9.14)	.23
Acne type
Comedonal	1 [Reference]		1 [Reference]
Mixed	1.00 (–17.28 to 19.27)	.92	–0.26 (–13.71 to 13.20)	.97
Inflammatory	3.55 (–16.38 to 23.48)	.73	–12.80 (–28.54 to 3.03)	.11
Acne severity
Moderate	1 [Reference]		1 [Reference]
Moderate to severe	8.50 (1.32 to 15.69)	.02	6.18 (–0.20 to 12.16)	.04
Severe	15.20 (6.46 to 24.00)	.001	12.40 (4.71 to 20.19)	.002
Scarring	10.70 (4.23 to 17.13)	.001	5.32 (–0.46 to 11.11)	.06

^aStandard regression coefficients represent the slopes for continuous variables and the difference in means between the reference group for that variable and the variable of interest for categorical variables. For example, age’s regression coefficient of 0.78 suggests that the overall Skindex-16 score increased by 0.78 points per 1-year increase in age. Likewise, the regression coefficient of 9.07 seen for moderate to severe acne reflects a mean difference of 9.07 points higher on the Skindex-16 scale between this group and the reference group (moderate acne).

^bAdjusted regression coefficients are interpreted in the same manner as standard regression coefficients. Results were divided into findings from overall Skindex-16 scores as well as Emotional, Symptoms, and Functional domain scores.

Discussion

In this single-site case series of patients with moderate to severe acne, it was shown that the typical patient receiving isotretinoin expected that they had achieved a 55.7% improvement in QOL during month 2 of treatment. Between treatment months 3 and 5, patients on average saw a 77.5% improvement in QOL from baseline. Acne had a notable impact on QOL, especially emotional QOL. This effect could be tracked over time, and it was responsive to treatment.

Interactions between Skindex-16 scores over time by age and sex were not significant, suggesting absence of differential treatment effects by sex or age. Likewise, multivariable analysis failed to show differences in QOL impact by sex. However, multivariable analysis did suggest that older patients and those with more severe acne experienced a more significant, negative effect on their QOL from acne.

Initial acne studies^31,32 in the early 1980s measured improvement in the mean number of cysts after isotretinoin treatment and showed a 50% improvement between week 6 and week 10 with maximal (70%-80%) improvement occurring between week 12 and week 20. These study outcomes focused on physician-reported measures. However, a 2005 article³³ found no correlation between physician-reported clinical severity and patient-reported DLQI scores in acne, similar to findings in other dermatologic diseases such as melasma.³⁴ These findings highlight that clinical severity and its consequences for QOL are often distinct outcomes, highlighting the power of PROs in helping physicians gauge improvements that are meaningful to the patient.

To our knowledge, no study has assessed outcomes from the patient’s perspective at regular intervals during the course of isotretinoin treatment, thus allowing for modeling of change in QOL through treatment course. However, some studies have used PROs before and after isotretinoin treatment. Iranian researchers reported that DLQI scores before and after isotretinoin treatment dropped from a mean of 8.0 (moderate QOL impact) to 2.4 (small QOL impact).³⁵ Two other studies^23,36 found similar improvements in mean DLQI scores before and after isotretinoin treatment, with greater improvements than for oral antibiotics or topicals alone.^23,36 Compared with data before and after isotretinoin treatment, longitudinal modeling of QOL is more powerful since it can help guide patient-physician discussions of expectations by providing insight into expected rate of QOL improvement over time. Understanding how patient factors and baseline acne features affect trajectory over time can also inform these discussions.

Both Skindex-16 and DLQI have been used in several clinical trials as part of the US Food and Drug Administration’s 2006 guidance to include PROs in clinical trials.³⁷ As expected, compared with placebo, nearly every published study on acne treatments showed significant QOL improvement.^{38,39,40,41,42} Skindex-16 has also been used to help validate other acne PRO measures.^43,44,45 In epidemiologic studies, acne had a significant impact on patients’ QOL, and objective physician-reported severity measures were not always correlate with this QOL impact.^{39,46,47,48,49,50,51,52,53,54,55,56,57}

Our findings have real-world implications by providing dermatologists with practical information to frame expectations with patients. Increased awareness of expected results helps to improve adherence.⁵⁸ A simple statement such as, “you can expect a 50% improvement in how you feel about your acne by our second follow-up visit,” provides accurate information to help patients understand that isotretinoin will likely work for them but not immediately. Visual representation similar to that shown in the Figure may help further illustrate this effect. In contrast to anecdotal evidence that many patients’ acne will get worse after initiating isotretinoin treatment, we failed to see any worsening of QOL impact but noticed 20% to 50% improvement in Skindex-16 scores at the first follow-up visit. This QOL improvement could be influenced by patient satisfaction with receiving what they consider the best available medication. Future planned studies comparing Skindex-16 with clinical measures at follow-up visits may help clarify this.

PROs may also help identify patients at risk of poorer outcomes. For example, previous studies^59,60 showed that low pretreatment Skindex-16 scores in males and high pretreatment Skindex-16 scores in females were associated with poorer adherence to acne treatments. While our data did not show sex-based interaction differences, more complex interactions between baseline scores and sex or interaction effects for other acne subtypes are plausible. In future larger studies, these interactions should be evaluated to identify subgroups who may experience differential treatment effects.

More evidence-based data are needed that show the benefits of clinical PRO use in dermatology. Implementation science in medicine has shown that physicians are resistant to change, especially changes that (1) lack significant evidence-based support, (2) are perceived as time-consuming or require training to acquire new knowledge, and (3) are not viewed as beneficial to the physician’s own practice and delivery of care.^61,62 We believe that this and similar future work using dermatology PROs in clinic will help dermatologists appreciate the clinical value of implementing PROs in clinical practice.

Limitations

These findings are limited in that they came from a single community-based academic dermatology practice and provided limited power to assess for complex interactions between patient demographic features and disease characteristics on their QOL. Similarly, limited patient follow-up seen in later months because of variability in how quickly patients achieved their target isotretinoin dose limited the accuracy of these estimates. Further data are needed to understand factors associated with QOL, especially toward the end of treatment, when adverse effects (skin dryness, lip cracking, or dermatitis) are potentially becoming more tiresome to patients and can affect reported Skindex-16 scores. In addition, residual scarring or pigmentation changes or unmet expectations may be contributory.

Future studies should include longer follow-up after isotretinoin treatment to tease out these factors. Moreover, Skindex-16 has not been validated in adolescents. However, the primary intent of this proof-of-concept study was to show that QOL can be tracked longitudinally and that Skindex-16 scores are responsive to isotretinoin treatment. Future research should establish the validity of the Skindex-16 in adolescents to support its use in the extant literature. Our group is currently investigating the Skindex-16 measurement properties and treatment effects in larger cohorts, incorporating a physician-based scoring of acne improvement, providing Skindex-16 scores to patients at each visit, and establishing better contextual features such as minimally clinical important differences of scores that can help better conceptualize score changes and improve score interpretation by both clinicians and patients. Interpretation can also be aided by the use of anchoring or distributional assessment,^63,64 but these were not pursued in this pilot study.

Conclusions

The findings of this case series suggest that patients receiving isotretinoin therapy achieve more than 50% improvement in QOL by month 2 and can expect, on average, approximately 4-fold to 5-fold improvements from baseline with a full course of isotretinoin. This study shows a potential use for routine administration of QOL assessments in informing patient care in dermatology.

Supplement.

eFigure. Raw Individual Skindex-16 Scores for Each Individual in the Study Over Time

eMethods. Model Selection