Prevalence of pain reports in pediatric primary care and association with demographics, body mass index, and exam findings: a cross-sectional study

Despite over 16 years of Joint Commission standards to assess and address patient pain, data are sparse regarding the prevalence, demographics, and body metrics associated with pain in general pediatrics. Prior studies have assessed pain in general practice or family practice, but only a fraction of those patients were children. [1, 2] Epidemiological studies in adults describe significant differences in pain prevalence among racial, socioeconomic, sex, and age categories, [3–5] however, comparable studies in pediatrics are sparse. Current scientific knowledge of outpatient pediatric pain is limited to non-routine/emergency [6] or subspecialty/disease-specific [7, 8] settings (e.g., emergency room, obesity clinic, pain clinic), or when pain is the primary complaint/diagnosis. [2, 9] Additionally, epidemiologic studies have focused on chronic pain, [10] especially through surveys at the school, [11] national, [12] or international [13] levels. Among these, general pain estimates ranged between 5% (report of chronic pain at the time of the survey) and 88% (at least one pain episode within the 3 previous months). Site-specific specific pain (e.g., headache, abdominal pain, musculoskeletal pain) rates varied nearly as much, but were also subject to many differences in assessment windows. [10] Few studies assessed the prevalence of pain assessed at the instant of data collection, but these were in school settings. None measured the prevalence of general pain at routine clinical encounters—the largest portion of pediatric healthcare. Even less is known of how clinicians respond to pain reports among children and physical exam findings from visits where pain is reported.

In a 2011 report, the Institute of Medicine requested better data on pain incidence, prevalence, and characteristics, specifically among vulnerable subpopulations, including children, people with low income, and racial and ethnic minorities. [14] This study responds, in part, to this call. Our objective was to determine prevalence of reported pain during a general pediatric encounter, and its relationship with body mass index (BMI) percentile, demographics, socioeconomic status, and season, using records from a clinical decision support system used in pediatric primary care clinics. As secondary outcomes, we investigated physician documentation of pain, and normal versus abnormal physical exam findings associated with reports of pain. Given the existing findings in adults and, to a lesser degree, children, we hypothesized differences in pain exist in pediatric patients based on obesity, sex, age, socioeconomic status, and race, and that pain is associated with an abnormal physical exam and previous pain reports.

Over the study period, 31,289 patients within the appropriate age range were seen; 26,180 met eligibility criteria and their most recent encounter was selected for analysis (see Fig. 2 for patient flow in study).

The majority of patients had normal BMI percentile (56.6%). The distribution of BMI percentile was skewed left (indicating relatively more frequent obesity); 4% were underweight and 23% obese or severely obese. Approximately 15% of patients reported pain at the index encounter. More than half (15,954 or 60.9%) of all patients had completed pain screening at the preceding visit, and 15.9% of that subset reported pain. There were on average 293 days (SD 269 days) between the index and preceding visits.

There were significant interactions on pain reporting between clinic and race. At the reference site, Black patients had lower odds of reporting pain than other races. Compared to the reference clinic, Black patients at a large clinic that saw primarily (> 80%) Black patients had 29% (CI 1.11–1.51) higher odds of reporting pain. In contrast, Black patients at Clinic 4, a Spanish-English fully bilingual clinic, had lower odds of reporting pain (OR 0.75, CI 0.59–0.96) than at the reference clinic. Within that bilingual clinic, Hispanic patients were 1.72 times (CI 1.36–2.18) more likely to report pain than Black patients.

To confirm this finding, we built a secondary model (not reported in detail here) where the race-clinic interaction was replaced by a variable indicating race-clinic concordance, which was positive when the patient’s race and the predominant race of the clinic were the same. There was a significant 27% increase odds of pain reporting in race-concordant clinics. However, when recreating Adjusted Model 2 from Table 3 (which includes pain reported at the preceding visit), the effect was not significant.

We examine the prevalence of undifferentiated pain in general pediatric outpatient care. We found about 15% of patients reported pain in our setting, which falls within prior limited clinical and non-clinical estimates among similar age groups or time frames. Previous studies in general or family practice across several countries used pain as the visit reason to estimate prevalence between 5.1 and 36%. [24–26] Within school-based surveys of varying age ranges, point prevalences have been reported between 4.8 and 27.1%. [27–29]

The obesity rates in our study sample were similar to nationally reported statistics. Obesity is linked to several pain outcomes, notably musculoskeletal complaints, headache, and chronic pain. [7, 8] A previous convenience sample showed a positive relationship between BMI and general pain report in an obesity clinic. [30] We show that obesity and severe obesity were associated with higher pain report prevalence in a pediatric population visiting a general outpatient clinic. However, underweight and overweight children did not report more pain. The correlation between obesity and pain may reflect a cycle of decreased exercise due to pain, and pain due to increased BMI from lack of exercise. [31] Unfortunately, many parents do not perceive this association in their own children. [32] Perhaps providers can use this knowledge to incentivize weight reduction to try to reduce pain. However, other psychosocial determinants of health may independently lead to pain and obesity, complicating this relationship.

Our findings are consistent with prior research on characteristics related to pain in children. Females reported pain more often than males, [2, 10] perhaps for psychosociocultural reasons, as the difference is noted across experimental pain response studies, [33] survey instruments, [12] and several pain subtypes. [10, 13] Older children generally report more location specific pain (except for abdominal pain). [10] The lower same-day pain reporting here among younger children may represent a true decrease in pain reports or sensation, a difference between acute and chronic pain among childhood ages, or even a known phenomenon of surrogates (e.g., parents or caregivers) underestimating pain. [34] Insurance at the time of visit was not associated with any change in pain reporting. Past studies demonstrate mixed findings of associations between SES and chronic pain. [10, 35, 36] However, the vast majority of our study population was on public insurance, limiting our power to assess this association. Additionally, the prevalence here includes both chronic and acute pain, and that combination may mask the prior associations with an isolated chronic pain. Increased pain reporting in winter months may reflect known seasonality of common complaints like abdominal pain, headache and respiratory complaints. [37]

In a subset analysis, the strongest predictor of pain reporting was pain reported at the preceding visit. Chronic pain is a notable burden in pediatrics, and it is possible the consecutive pain reports reflect ongoing symptoms. However, it may also be a characteristic of certain patients who are more likely to voice concerns of various common pains.

The largest racial minority in our population reported more pain when in a clinic that predominantly saw patients of that same race. We suspect the social context of patient-clinic racial and cultural concordance may offer a comfort with which to disclose more medical symptoms. Hsieh et al. [38] found increased affective and nonverbal pain expressions when there was both cultural milieu (including language) and race concordance between participants and experimenters – a context similar to the clinics in this population. While it is possible that the race of the treating provider may play a role in pain reporting, we feel this factor is minimized in our data, since the pain question was assessed in the waiting room before interaction with the provider. Further clinical research is needed in facility-level interactions on pain reporting, especially within a new era of computerized clinical decision support systems.

An electronic PSF medium was notably associated with increased pain reports, even when controlling for question response rates. We expected the patient-reported outcomes to be generally equivalent between paper and electronic reporting interfaces. [39] We are uncertain why our experience differs, and it is possible that caregivers and adolescents are more inclined to report pain when the responses go directly to the server instead of on paper, which is handed to clinic personnel. We recommend this topic for future research efforts.

Only 37% of providers rated their patients’ pain after an alert reporting a patient’s positive response. Although the provider response rate to positive pain screens appears low, it is reasonable within the context of historical [40], local [41], and other voluntary clinical reminder system precedent. [42–46] About half (51.2%) of the completed provider assessments confirmed the patient’s pain report, and these were generally rated at low pain levels. It is possible that clinicians were more inclined to respond to the prompt in order to negate what they observed was an erroneous report, which would lead to an artificially depressed rate of providers confirming a patient’s pain report. This perceived error may come from a difference between what the patients are trying to report on the screening question and what providers want to know from the alert. Or, it may simply reflect the disparity between patient, caregiver, and clinician pain assessments. Previous research on concordance of pain assessments by clinicians and patients indicate that providers significantly underestimate pain intensity in both adults and children. [47–49] Pain assessment is just the first step in a pain treatment pathway; one study showed clinicians administered pain relief to fewer than half the patients they determined to be in severe pain. [47] Clinical decision support systems aid in pain screening, but further work is needed enhance provider attention and response. Beyond alerting the clinician, decision support and provider training should focus on pain recognition disparities, and actionable and appropriate treatment recommendations.

Exam findings positively correlated with increased pain may come from common painful conditions within pediatrics, such as acute otitis media, gastroenteritis, extremity injuries, and pharyngitis. The negative correlations may result from routinely documenting an abnormal finding that is not usually painful, like heart murmur or dentition with multiple fillings. It is important to note that our sample of patients with full physical exams is a minority, since clinicians may have documented their exams elsewhere, and it is not routine practice to perform a full physical exam when evaluating a targeted chief complaint. Since our screening question assessed general, undifferentiated pain, it is possible many of the reports were for localized concerns.

The Joint Commission accreditation standard in 2001 described a “patient right” to have pain assessed and addressed. Over the past 16 years, many studies and editorials have discussed the difficulties and outcomes of the universal pain assessment intervention. Authors of these studies/editorials often find little or no effect on pain control attributable to systematic pain assessment, and there is controversy over the timing of an “opioid crisis” and the Joint Commission mandate to assess and address pain. [50] Whether systematic pain assessment improved outcomes for the patients in our study is uncertain, but pain was associated with a subset of physical findings, suggesting that there is diagnostic information in the pain measure.

Our study is not without limitations. First, CHICA PSF questions have binary (yes/no) responses, and clinicians are not required to follow-up with a standardized pain assessment. Since pain reports were generally rated by parents or providers, it is possible our findings underestimate the true pain prevalence in this population. It is possible that clinicians were documenting pain or physical exams independently in their notes, which would not be captured with our clinical decision support system. Our observational, cross-sectional study design is unable to identify causal relationships; for example, physical exam findings may be from other abnormalities that are not painful. We do not have data to differentiate well visits versus sick visits, or acute versus chronic pain, which is likely to affect pain reporting, and we emphasize our results are from routine, systematic screening. At the time of this study, the medical record data did not differentiate between race and ethnicity, and so the effect of these two distinct constructs is blurred. Finally, our data come from five urban pediatric primary care clinics serving predominantly low-income and minority families living in a single metropolitan area, so readers should exercise caution in generalizing these results.

Using a large diverse sample, we establish the first data on prevalence of pain and associated characteristics in general pediatric primary care. We show that previously reported trends in pain epidemiology in fact extend to the ambulatory practice environment. More than one in seven patients reported pain at these encounters. Our data suggest that females, White children, children 5 years of age or older, and obese or severely obese children had higher odds of reporting pain at the time of a clinical encounter. In addition, Black and Hispanic patients reported more pain in clinics with concordant majority races. Pain reports were also positively associated with electronic reporting interfaces, certain seasons, and specific exam findings. Our study answers an Institute of Medicine request for better data on childhood pain, especially within a minority and low socioeconomic status population, and provides a springboard for research into such a common symptom in primary care. Future studies should apply methods [51] that improve pain assessment workflows and efficiency in general pediatric practice, and assess clinical outcomes associated with pain assessment in children.