Materials and Methods

SOPHIA was a prospective cohort study of women aged 31-54 years who sought care for symptomatic fibroid tumors, abnormal bleeding, and/or CPP. Details about SOPHIA’s study design, recruitment methods, and outcome measures are described in a publication that reported the participants’ baseline characteristics. In brief, women were recruited from academic, community, and public hospital practices in the San Francisco Bay area, which included those facilities affiliated with the University of California, San Francisco, San Francisco General Hospital, Kaiser Permanente Northern California, and community hospitals in San Francisco. SOPHIA was designed to explore the natural history of noncancerous gynecologic conditions, and to identify predictors of treatment use and important clinical outcomes, including HRQOL. Two recruitment cohorts (1997-1998 and 2003-2004), each of which was observed for 4 years, comprise the SOPHIA participants as a whole. We invited women who had been recruited in the first cohort to reenroll in the second cohort, which allowed extended follow-up evaluation for these participants. Potential participants were excluded if they had already undergone or were planning to have a hysterectomy, if they had been diagnosed with a gynecologic malignancy, if they could not understand verbal or written English, Spanish, or Chinese, or if they had entered menopause. The study was approved by the University of California, San Francisco, Committee on Human Research and the Institutional Review Boards at Kaiser Permanente Northern California, and San Francisco General Hospital.

We used baseline and annual interviewer-administered questionnaires to assess participants’ sociodemographic, clinical, and HRQOL characteristics and their attitudes about treatment alternatives. At the beginning of the baseline interview, participants were placed into 1 of 5 clinical groups, based on their predominant symptoms and associated diagnoses: (1) CPP only, (2) CPP and abnormal uterine bleeding, with or without leiomyomata, (3) abnormal uterine bleeding without uterine leiomyomata, (4) abnormal uterine bleeding with leiomyomata, or (5) pelvic pressure with leiomyomata. The questionnaires for the second recruitment cohort included enhanced measures of depression, sexual functioning, and urinary incontinence. We restricted our investigation to data from second-cohort participants, which included participants from the first cohort who reenrolled.

Participants were identified with CPP either at the time of study entry or based on their responses to subsequent symptom questionnaires. To be placed in the clinical groups with CPP at study entry, participants had to indicate in the recruitment interview that they were experiencing pain below the umbilicus (in any pattern) for at least 6 months. The CPP cohort also included participants who answered “yes” to the presence of pelvic pain in annual questionnaires that were conducted subsequent to the baseline assessment and indicated at least a 6-month duration.

To measure depression we used the Patient Health Questionnaire-9 (PHQ-9), which is the 9-item depression component of the Primary Care Evaluation of Mental Disorders screening instrument. The PHQ-9 has been validated for use as a case-finding instrument for depression in gynecology patients. The PHQ-9 scoring algorithm provides criteria for diagnosis of major depressive disorder (MDD) and other depressive disorder (ODD), which is a subthreshold diagnosis in which not all MDD criteria are met. The risk factors and clinical treatment of outpatients with ODD are comparable with MDD, which suggests that it is an affective disorder that is worthy of screening and treatment. We administered the PHQ-9 during the baseline assessment and with the annual assessments thereafter. For participants who entered the cohort with CPP, we used their baseline PHQ-9 scores to detect coexisting depression. For participants who were identified with CPP after study enrollment, we used their PHQ-9 scores from the same survey administration that qualified them for a CPP diagnosis. In the current investigation, we defined depression in women with CPP to include both MDD and ODD. We used questionnaire responses to identify whether depressed participants were using antidepressant medications. Participants who reported antidepressant medication use were not assigned to a depression diagnosis, unless their PHQ-9 scores met criteria for MDD or ODD.

Symptom resolution, which was the primary outcome for our analyses, was measured with the use of a single item that was assessed at baseline and in subsequent interviews: “To what extent would you say your pelvic problems have been resolved?” followed by a 4-level response (not at all, somewhat, mostly, or completely resolved). We chose this question on the basis of empiric evidence that it was a more useful predictor of hysterectomy in the first recruitment cohort than other multiitem symptom measures. Our secondary outcomes included measures of HRQOL: pelvic problem impact, satisfaction with the ability to have and enjoy sex, pelvic problem interference with sex, beliefs about current health, and the physical component summary and mental component summary scales from the Medical Outcomes Study SF-12 questionnaire. These measures are detailed in an earlier publication.

The treatments that were reported by patients were grouped into 3 categories on the basis of the most invasive treatment received: hysterectomy (removal of the uterus), uterus-preserving treatments, and other treatments. Uterus-preserving treatments were defined as alternatives to hysterectomy for CPP refractory to first-line treatments and included surgical treatments (removal or destruction of endometriosis with or without ovarian removal but excluding hysterectomy) and gonadotropin-releasing hormone (GnRH) agonist therapy. Patients who did not report these treatments or hysterectomy were categorized as “other” in our analyses and included women who received combined estrogen and progestin contraceptives, progestins (intrauterine system, depot medroxyprogesterone acetate), pain relievers that included narcotics, physical therapy, trigger point injections, or no treatments other than over the counter analgesics.

Baseline sociodemographic and clinical characteristics and HRQOL of the participants were described with the use of means, standard deviations, and percentages. We used t -tests to compare unadjusted pre-/posttreatment symptom resolution change scores between depressed and nondepressed participants and defined statistically significant differences at a probability value of < .05. For hysterectomy and uterus-preserving treatments, pre-/postsymptom resolution change scores were calculated with the most proximate data points to the index event. For other treatments, pre-/postchange scores were calculated by a comparison of the last recorded data points to baseline data.

We used stepwise multivariable linear regression to estimate the unique effects of depression on pre-/posttreatment changes in symptom resolution and our secondary outcomes. We included other explanatory variables that we hypothesized were associated with these outcomes: baseline sociodemographics, symptoms, and HRQOL; treatments used during the study, and whether the woman entered menopause during the study. Treatments were modeled as 3 binary predictors: hysterectomy, removal or destruction of endometriosis with or without ovarian removal but excluding hysterectomy, and GnRH agonist therapy. Each was compared with other treatments as defined earlier. We also adjusted for use of antidepressant medications and whether CPP was present at study enrollment or arose during the study period. Nonsignificant explanatory variables ( P > .20) were removed from the multivariable models. Depression was modeled dichotomously (MDD or ODD vs not). To test whether combining MDD and ODD influenced the regression results, we repeated these analyses using ordinal categories of depression (MDD, ODD, no symptoms) and using a strict diagnostic threshold (MDD vs not).

To estimate the effects of explanatory variables on future hysterectomy use, we fit Cox proportional hazard models, with participant age, to define the time scale. We ascertained hysterectomy events using interviewer-administered questions that were supplemented annually with medical record review to minimize missing data in participants with incomplete follow-up data. Values of the hysterectomy event outcome were either observed or censored (uterus retained at end of follow-up period). Explanatory variables were grouped into 4 categories based on how they were modeled: (1) baseline indicators of sociodemographic and clinical characteristics; (2) a time-varying indicator of menopausal status; (3) time-varying indicators of nonhysterectomy treatment use and symptom resolution; and (4) time-varying indicators of HRQOL and attitudes toward the uterus and hysterectomy. The nonhysterectomy treatment types for CPP included removal or destruction of endometriosis, GnRH agonist therapy, and prescription medications for pain (progestins/narcotics). Each type of treatment was indexed by 2 variables: a binary indicator of use before enrollment and a time-varying count of subsequent use across follow-up assessments. Preliminary Cox models tested equivalence of between- and within-person effects of time-varying indicators that described symptom resolution, pelvic problem impact, HRQOL, sexual function, and hysterectomy- and uterus-related attitudes. To accomplish this, each corresponding time-varying variable initially was replaced by 2 derived explanatory variables: a “between-person” variable of baseline values and a “within-person” variable that described individual change since baseline at each follow-up evaluation. The multivariable model was modified in 2 stages. First, a backward elimination procedure removed explanatory variables with probability values exceeding .20. Second, we tested the difference between corresponding between- and within-person parameter estimate pairs. Any nonsignificant difference ( P > .05) suggested equivalent between- and within-person effects; in that case, both corresponding explanatory variables were dropped from the model, which was replaced with the original time-varying variable, and the model was refit. Further details on our analytic procedures for predicting hysterectomy have been described in a previous publication for the cohort as a whole.

Because the data contained missing values, we fit all models to 20 multiply imputed datasets created with SAS PROC MI (version 9.1; SAS Institute Inc, Cary, NC). The imputation model included all variables included in Tables 1-3 and was fit separately for each cohort. Imputed values for binary and categoric variables were rounded and truncated to the nearest category. Parameters and standard errors were estimated by the combination of results across the 20 imputed datasets, according to the rules of Rubin and Meng and Rubin.

TABLE 1

Baseline characteristics of women with chronic pelvic pain

Variable	Measurement
Sociodemographics
Age, y a	43.2 ± 0.18
Site, n (%)
Health maintenance organization	306 (43.7)
Academic medical center	214 (30.5)
County hospital	146 (20.8)
Community practice	35 (5.0)
Race/ethnicity, n (%)
Asian	56 (8.0)
Black, African American	193 (27.5)
Latina, Latin American	129 (18.4)
White	288 (41.1)
Other	35 (5.0)
Relationship status: married/living with partner	385 (54.9)
Education, n (%)
High school or less	132 (18.9)
Some college	200 (28.5)
College degree or more	369 (52.6)
Annual household income, n (%)
<$25,000	241 (34.4)
$25,001-50,000	224 (31.9)
$50,001-100,000	185 (26.4)
>$100,000	52 (7.4)
Occupational status, n (%) b
Unemployed	24 (3.4)
Lowest prestige (16-47)	273 (38.9)
Middle prestige (47-69)	308 (43.9)
Highest prestige (68-86)	97 (13.9)
Clinical characteristics
Other symptoms at study entry, which may be concurrent, n (%)
Heavy or frequent bleeding	454 (64.8)
Symptomatic fibroid tumors (bleeding or pressure)	293 (41.8)
Symptom resolution, n (%)
Not at all	345 (49.2)
Somewhat	250 (35.7)
Mostly	102 (14.5)
Completely	4 (0.6)
Desires future pregnancy, n (%)	217 (31.0)
Depression: Patient Health Questionnaire-9, n (%)	153 (21.8)
Major depression	87 (12.4)
Other depression	66 (9.4)
Health-related quality-of-life measures a
Pelvic Problem Impact Scale c	28.7 ± 0.87
Physical Component Summary score c	44.8 ± 0.39
Mental Component Summary score c	33.5 ± 0.44
Beliefs about current health d	4.29 ± 0.05
Satisfaction with ability to have and enjoy sex e	3.29 ± 0.04
Interference of pelvic problems with sex f	2.78 ± 0.05

Only gold members can continue reading. Log In or Register to continue

Share this:

• Click to share on Twitter (Opens in new window)
• Click to share on Facebook (Opens in new window)

Related

Related posts:

1. Managing disruptive behaviors in the health care setting: focus on obstetrics services
2. Protein kinase Cβ2 inhibition reduces hyperglycemia-induced neural tube defects through suppression of a caspase 8-triggered apoptotic pathway
3. Cystatin A protease inhibitor and cysteine proteases in human cervicovaginal fluid in term pregnancy and labor
4. Phenazopyridine does not improve catheter discomfort following gynecologic surgery

Stay updated, free articles. Join our Telegram channel

Join