Predictors of Time to Healing Deep Pressure Ulcers

Transcription

1 .. '" Predictors of Time to Healing Deep Pressure Ulcers Lia van Rijswijk, RN, ET1 Marcia Polansky, ScD2 Abstract: Time to healing analysis methods (Kaplan-Meier time until healing curves) were used to compare time to healing deep pressure ulcers as a function of patient and wound characteristics at baseline and after two weeks of treatment. Time to healing was significantly reduced in patients who had a good nutritional status. Patients who were alert and coherent were also found to heal more expediently; however, mental status was not independently predictive of time to healing in the multivariable model. Larger wounds took longer to heal (median 20 days) than smaller wounds, but the difference was not statistically significant. After two weeks of treatment it was found that ulcers in patients who were 60 to 70 years old, who had a good nutritional status at baseline, and whose ulcers reduced at least 39% in size after two weeks, were found to heal much more expediently. Cox regression models showed that these factors were independently predictive of time until healing (likelihood ratio statistic on 5 DF =26.485, P <.00l). Clinical assessments, both at baseline and regular intervals thereafter, may predict treatment outcome of full-thickness pressure ulcers. INurse Consultant, Newtown, PA, MSN La Salle University School of Nursing 2Division of Biometrics, Department of Humanities and Biometrics, Hahnemann University, Philadelphia, PA Address correspondence to: Lia van Rijswijk, RN, ET 15 Willow Ct Newtown, PA Phone (215) Fax (215) Our understanding of chronic wounds, i.e., their etiology, the healing process and appropriate methods of care, has improved dramatically during the past several years. However, even though the effect of systemic and wound conditions on acute wounds have been studied extensively, their effect on healing chronic wounds is not as well understood.1,2 Patients with pressure ulcers often have a variety of systemic conditions which may influence repair. An increased understanding of the role of patient and wound variables in the healing process of these chronic

2 wounds would help clinicians and researchers alike, particularly with respect to deep pressure ulcers which may take a long time to heal. Statement of the problem and review of the literature. One of the reasons the effect of various covariates (systemic and wound conditions) on healing pressure ulcers has not been as clearly defined as acute wounds may be related to their duration and the data collection and analysis methods used to evaluate the effects of treatments. First, wound assessment definitions and communication about clinical findings are not standardized.3-s This makes it difficult to compare the results of different publications, and hampers the ability to extrapolate findings to the general population. Second, patients with chronic wounds, particularly deep wounds, are often lost to follow-up for reasons unrelated to the study. Thus, the commonly used method of comparing "healed" to "non-healed" patients i not very sensitive. Similarly, the improvement ratings used to obtain an end-point assessment for patients who withdraw from a study are not very sensitive, because they usually have not been tested for validity and reliability. Covariates which have been shown to increase the time until healing or reduce the likelihood of healing pressure ulcers include the following patient characteristics: lower protein nitrogen intake, less than adequate nutritional support or poor nutritional status, advanced age, young age, male sex, female sex, immobility and the presence of fecal incontinence.6-13 With respect to wound conditions, deep ulcers have been found to take longer to heal than partial thickness or superficial ulcers l In one study, Stage II pressure ulcers with less wound exudate at baseline were found to heal more expediently than exudating ulcers. IS Finally, a poor response to previous treatments has been found to result in a smaller proportion of healed ulcers, and deep pressure ulcers that reduce an average of 45 percent in size after two weeks of treatment have been found to be more likely to heal than those that do nopo Whether or not the above-mentioned covariates are independently associated with time to healing, or whether some are related to each other, is not completely understood at this time. To further explore the relationship between _ outcome (time to healing deep pressure ulcers). and the covariates assessed (patient and wound characteristics), a secondary analysis of previously reported study data was performed. IO Instead of comparing proportions of patients healed, analysis methods developed to study patient survival over time, i.e. Kaplan-Meier time until healing curves, were calculated for time until 50, 80 and 100 percent healing based on the ulcer tracings obtained Reduction in ulcer area was calculated as a percent reduction in area from baseline (day 0), controlled by baseline area. For patients with multiple ulcers, a single mean for the ulcers was calculated to avoid violating the assumption of independence. The median time until healing was calculated for all patients combined and as a function of each patient and ulcer characteristic at baseline and after two weeks of treatment. All calculations were performed for the time until 50, 80 and 100 percent healing to also capture patients who withdrew from the study prior to healing. To establish the independent prognostic importance of the covariates on healing time, i.e. determine which variables are most predictive, a stepwise Cox proportional hazards' model for prognostic factors of time until 100 percent healing was used.j7 Patient characteristics obtained at baseline included: age, weight, gender, number of ulcers, presence or absence of diabetes or incontinence. The following patient characteristics were obtained using 3 or 4 point rating scales: general health condition, mental status, mobility, activity level, body build and overall skin condition. The investigators rated patient nutritional status as good (including patients who received TPN), satisfactory or poor. In all tests, the latter two categories were grouped. Wound characteristics, i.e. aspect of ulcer margin, the presence of granulation (none, some or completely covering the wound) or necrotic tissue (none, some, or thick and covering the wound), and depth were assessed at baseline and every dressing change. All tests were performed using the Statistical Analysis System (SAS n,,) at 0.05 level of significance using two-tailed tests and data of all patients who in the analy- enrolled in the study were included sis.

3 , -;l; J"\ t '"c OJ. c. '0 c.g (; c. 0 a: 80% I 60% i! 40% I I 20%_[;!- I - 0% I,.f;-;t! o 14 '26' '86 98 Number of days in study 50,80, or 100 % healing after an X number of days: 50% healing curve: 75th percentile, 39 days; 50th percentile, 21 days; 25th percentile, 15 days 'C 100%1 OJ i 80% I '"cq). 60% LJ...- d-'- c: I --e1 Poor.,8 f >-++-! status 2 %-i----- j - nutritional a: 0% 1- - i ' ', '92 01, loo-ronk Number of days in study Figure % healing craves by nutritional status at baseline. Good nutritional status: median 53 days, mean 51 (SE 5.7); Poor nutrition.i--tatus status: median 90 days, mean 72,6 (SE 6.2). 80% healing curve: 75th percentile, 53 days; 50th percentile, 40 days; 25th percentile, 27 days 100% healing curve: 75th percentile, 243 days; 50th percentile, 69 days; 25th percentile, 50 days The study sample consisted of 48 patients (25 men and 22 women) with a mean age of 69 years (SO 17.2). Patients resided in an acute care (N=21), extended care (N=23) or rehabilitctio _ (N=4) facility. They presented with 56 full-thickness Stage III and IV pressure ulcers that were dressed with a hydrocolloid dressing (OuoOERM Hydroactive dressing, C0n>iatec, Skillman, NJ) for a mean of 56.0 days (SO=51.66, range 13 to 243 days). Prior to study enrollment, ulcers had existed for < 1 month (39 percent), > 3 months (21 percent) or > 9 months (14 percent). Thirteen (13) patients had diabetes mellitus and 21 were rated as having a satisfactory general health condition. The majority of patients (23) were confused, had restricted mobility (37), and a satisfactory or poor nutritional status (32). Twenty-two ulcers (46 percent) were reported as not having changed (either improved or deteriorated) prior to study enrollment. At baseline, most ulcers (24) had a maximum diameter of between 2 and 4 cm, clean ulcer margins (35), no granulation tissue (34) and some necrotic tissue (28). The mean ulcer area at baseline was 6.3 cm2 (SO=10.4). During the study, 21 (37.5 percent) ulcers healed and 16 were rated as "markedly or moderately improved". Using analysis of variance techniques, percent reduction in ulcer area after two weeks of treatment was found to differ significantly between ulcers that did and did not heal during the study (F[1,40] = 14.27,p < 0.01).10 Time to healing curves. Kaplan-Meier time until 100 percent healing curves showed that the median time to healing was 69 days. Healing was expected to be seen in 25 percent of the patients after 50 days, whereas 75 percent of patients healed after 243 days. (Figure 1) A 50 percent reduction in wound size (50 percent healing curve) occurred after 15 days, whereas half of all wounds exhibited an 80 percent reduction in ulcer area after 40 days. Effect of baseline patient covariates on time to healing. Nutritional status at baseline significantly influenced the time to healing. Patients who were rated as having a satisfactory /poor.nutritional status at baseline took a mean of 20 days longer to heal as compared to patients who had a good nutritional status (Log-Rank [I, N=40]=6.3, p=o.ol). (Figure 2) Pressure ulcers of patients who were coherent healed after a median of 53 days (Mean 54, SE-9.1) whereas patients who were confused/ disoriented healed after a median of 86 days (Mean 139, SE 28.7). A median time to healing could not be calculated for comatose patients since fewer than 50 percent healed. These differences were marginally significant (Log-Rank [2, N =40]=6.1, p=0.047). Even though a trend was observed wi th respect to patient mobility, the median time to reach 100 percent healing for completely immobile patients was 86 days (Mean 124, SE 27.5)

4 Hazard Ratio Time Variable (95% cn 95'70 CI P value Baseline poor nutritional status S After two weeks of treatment: age 70 to 79* 0.02 O.OOlS- O.3S O.OOS.. age SO age < O.OOOS - 0.2S poor nutritional status 0.08 '\ % reduction in ulcer area Referencegroup: 60 to 69 years of age..likelihood ratio statistic on 5 DF =: ,p < 0.001,for Cox proportional hazards model after two weeks of treatment compared to 53 days (Mean 49, SE 6.5) for patients with restricted mobility, the difference was not statistically significant (Log-Rank [3, N=40] = 6.1, P =0.1). Patient age was grouped into four groups to facilitate analysis: < 60 year old, 60 to 69 years old, 70 to 79 years old, and SO years old. The mean time to healing pressure ulcers in patients 60 to 69 years of age was 34.9 days (SE 3.7). Patients < 60 years of age healed after a mean of 56.4 (SE 6.6) and patients 70 to 79 and 80 years of age healed after 78.5 (SE 10.1) and 67.1 days (SE 1.9) respectively. However, these differences were not statistically significant. None of the other patient variables assessed (gender, incontinence, diabetes, general health condition, overall skin condition, number of ulcers per patient, weight, and body build) were found to influence time to healing. The effect of baseline wound covariates on time to healing. To facilitate analysis of baseline area (ulcer size) as a function of time to healing, the areas were grouped in quartiles as > cm2, > cm2, > cm2 and> cm 2. Even though a trend was observed with respect to healing, the smallest ulcers took a median of 50 days (Mean 62.3, SE 13.2) to heal whereas the largest ulcers healed after a median of 70 days (Mean 68, SE S.08), the difference was not statistically significant (Log-Rank [3, N=39J=1.2, P = 0.07). None of the ulcer parameters assessed at baseline (i.e., presence and amount of necrotic tissue, presence and amount of granulation tissue, aspect of ulcer margin, surrounding skin condition, odor emanating from the wound, presence and amount of wound exudate or pain) were found to influence the time to. healing. Multivariable analysis for covariates at baseline. When all significant and marginally significant variables were used to build a stepwise Cox proportional hazards model, only a poor nutritional status at baseline was found to be predictive of time to healing (Hazard Ratio [95 percent CIJ=0.21; 95 percent CI:0.052 to 0.85; p=0.02). (Table 1) This finding was consistent with the differences observed in the Kaplan-Meier curves. Percent reduction after two weeks as a function of time to healing. To further explore the predictive value of percent reduction in ulcer area after two weeks of treatment, reduction in ulcer area was categorized in three groups: One group

5 t.. c!! for those that increased or did not change in size during the first two weeks of treatment ( 0 per- "0 cent) and two groups for those that decreased in 80% size ( > 0 39 percent and> 39 percent). The median time to 100 percent healing could not be ' %L...._I calculated for the first group since fewer than 50!,eduet;on percent of these wounds healed. By contrast, 40% ulcers which reduced between 0 and 39 percent.g healed after a median of 70 days (Mean 64, SE 6.1) 8. 20% compared to 53 days (Mean 54, SE 8.0) for ulcers which reduced> 39 percent in size during the first two weeks (Log-Rank [2, N=30]=9.5 p=0.008). (Figure 3) Multivariable analysis for covariates after a ll: = :> 0.39% two weeks of treatment. Stepwise Cox proportional hazards model after two weeks of treat- Figure 3: 100% healing curves by % reduction in ulcer area after two weeks of treatment. ment showed that age, nutritional status, and percent reduction in ulcer area were all independent- median 53 days, mean 54.2 (SE 8.0); S. 0% reduction: > 0 - -, <-39%: median 70 days, memz 63 (SE 6.1); > 39%: ly predictive of time to healing. (Table 1) Patients median could not be calculated since <50% of patients in who were between 60 and 70 years of age, this group healed. patients who had a good nutritional status at baseline, and patients whose ulcers reduced at independently predictive of time until healing least 39 percent in size after two"weeks wete --- -aeeppressure iilcers: In the study, healing- time found to heal much more expediently. Since age was influenced by the patient's mental status, but was not a predictor of time to healing at baseline the difference was marginally significant and disbut was significant in the Cox model which was appeared when multi variable procedures were fitted using only those patients who were not yet used. However, clinically, the difference in healhealed at two weeks, additional Kaplan-Meier ing time was substantial: 33 days difference curves were constructed for this subgroup by age between patients who were alert versus patients only. Again no statistically significant differences who were confused/disoriented. Similar differamong the age groups (Log-Rank [3, N=30]=4.84, _ _ enceswere observed asa function of patient- p=0.18) were found. There appeared to be some mobility. It is possible that the size of the data set association between age and percent reduction, was too small to detect a statistically significant but this association also did not reach a statistical difference for these variables. At baseline, the significance (chi-square [6, N=37L p=0.68). patient's nutritional status had a significant effect Similarly, neither age and nutritional status On time to healing. When added to the multivari- (chi-square [3, N=40j, p=0.27) nor nutrition and able model, the effect of nutritional status on time percent reduction (chi-square [4, N=37], p=0.6),to healing remained. A relationship between the were associated. Finally, we reviewed data of all presence or severity of pressure ulcers and nutripatients who were discontinued from the study tional status has been found in several studduring the first two weeks. Of the 10 patients for ies.!o,18-20 In addition, a correlation between prowhom no data was available at the two week tein nitrogen intake/inadequate nutritional supassessment, six were deceased, two were trans- port and pressure ulcer healing has also been ferred, and two did not have their wounds traced observed.6,7,13 In this study of full-thickness at that time. Discontinued patients were evenly divided among age groups. The objective of this study was to find out if some patient and wound characteristics were I reduction = > 39% / ( 0%14 22' p =.008. Log-Rank Days in study ulcers, nutritional status, as assessed clinically by the primary investigator, was predictive of time to healing at baseline. In addition, nutritional status remained an independent predictor after two weeks, when percent reduction in ulcer area was added to the multi variable model. Studies to investigate the relationship between clinical assessments and utilization of pressure ulcer prevention scales have shown that the correlation for

6 some item specific scores is better than for who did not heal, during this study was also others.2122 In this study, the clinical assessment of found to be an independent predictor of time to patient nutritional status, despite the fact that it healing. Using the Cox proportional hazards was not measured, was found to predict time to model, it was found that percent reduction had a healing. - very high hazard ratio (7) and a low p value The effect of age on time to healing could not (p=o.ol). Thus it can be concluded that with be established at baseline. However, after two every increase in percent reduction in ulcer area, weeks, patients who were older than 70 and healing of the wound will be faster. This "early younger than 60 years of age were found to heal effect" of treatment has also been found for slower than patients who were between 60 and 70 chronic wounds on the lower legs with significant years of age. Even though advanced age has been differences reported after two and four weeks shown to influence tissue repair232 the relation- respectively.2&-30 ship between age and pressure ulcer healing In the original analysis of this data set, the proseems less clear and is certainly not linear. Our portion of healed ulcers was higher in men than results indicate that the younger patients (who in women, and higher for ulcers which had were either paraplegic, quadriplegic or acutely responded well to previous treatments, but multiill) did not heal as expediently as their older variable procedures were not performed. lo The counterparts. However, patients who were> 70 effect of these covariates on the Kaplan-Meier years of age were also healing slower than the rcurves was not significant, thus time to healing 60 to 70 year old group. Gorse and Messner also does not explain the differences reported. To furfound that younger patients with pressure ulcers ther explore the relationship between took longer to heal but further analysis revealed patient/wound variables and chronic wound. that the effect of other conditions-was more-----heaiig, -additionl studies using larger data sets important than age.? In this study, the effect of are needed. age on time to healing did exist, but also appeared to be less important than other patient characteristics since age only became predictive of time to healing at the two week evaluation The effect of patient nutritional status on time point when other predictive factors were added to healing deep pressure ulcers was found to be to the model. These results suggest that age may significant, and independent of other predictors influence time to healing, and reduced tissue per at baseline as well as two weeks into the study. fusion in the spinal cord injured or arteriosclerot- Patient age also appears to effect time to healing, ic changes in the elderly can explain the results but the non-linear relationship needs to be invesobserved. 25,26 However, larger data sets with tigated, and additional studies in this area are more patients in each group are necessary to con- needed. With respect to wound covariates, only firm these findings. percent reduction in ulcer area after two weeks of It is often assumed that large, deep wounds treatment was predictive of time until healing. will take longer to heal than small, deep wounds, Even though further research utilizing large sets even though few pressure ulcer studies have con- of data are needed to define the most optimal firmed this assumption. Robson and co-workers time period for predicting treatment outcome and reported that actual pressure ulcer volume time to healing, there appears to be sufficient evidecrease was directly proportional to the ulcer dence to suggest that weeks of ineffective treatsize whereas Skene et al. found that large leg ment modalities can be avoided if appropriate ulcers took longer to heal than small leg ulcers. 827 clinical assessments are performed at least once a When baseline area is included in calculating week. healing rates, as was the case in our study, a difference in time to healing small versus larger wounds was observed. However, variability between patients was such that this difference was not statistically significant. Percent reduction in ulcer area after two weeks, previously reported to be significantly different between patients who did and patients The authors are grateful to ConvaTec for making the data set available to them.

7 sores among hospitalized patients. Ann Intern lvfed 1986;105: Pinchofsky GD, Kaminski MV. Correlation of pressure 1. Kirsner RS, Eaglstein WHo The wound healing '-.- sores and nutritional status. JAGS 1986;34: process. In: Nemeth AJ (ed). Wound healing. 20. Bergstrom N, Braden B. A prospective study of Dermatol CIin Philadelphia, PA: W. B. Saunders, pressure sore risk. among institutionalized elderly. 1993;11: JAGS 1992;40: Polk HC Jr. Factors influencing the risk of infection 21. Van Marum RJ,Germs P, Ribbe MW. De risicoscoring after trauma. Am J SUTg 1993;165 (Suppl):25-7S. voor decubitus volgens Norton in een verpleeghuis. 3. Abruzzese R. Much ado about nothing (Editorial). Tijdschr GeTont Ger 1992;23: Dewbitus 1992;5: Xakellis GC, Frantz RA, Arteaga M, et al. A compari- 4. Bates-Jensen BM, Vredevoe DL, Brecht ML. Validity son of patient risk for pressure ulcer development and reliability of the pressure sore status tool. with nursing use of preventive interventions. JAGS Decubitus 1992;5: ;40: Lazarus GS, Cooper DM, Knighton DR, et ai. 23. Holt DR, Kirk SJ, Regan MC, et al. Effect of age on Definitions and guide-lines for assessment of wound healing in healthy human beings. SUTg wounds and evaluation of healing. ATch DeTm 1993;112: ;130: Kligman AM, Grobe GL, Balin AK Aging of skin. 6. Allman RM, Walker JM, Hart MK, et al. Air-flu- In: Finch CE, Schneider EL (eds). Handbook of the idized beds or conventional therapy for pressure (Biology of Aging. 2nd ed. New York, NY: Van sores. Ann Tntem Med 1987;107: Nostrand, 1985:82D Gorse GJ, Messner RL. Improved presure sore heal- 25. Mawson AR, Siddiqui FH, Connolly 5J, et al. Sacral ing with hydro-colloid dressings. Arch Denn 1987; transcutaneous oxygen tension levels in the spinal 123: _-- cord injured: Risk-factors for pressure ulcers? ATch 8. Robson MC, Phillips LG, Lawrence WT, et al. The Phys Med Rehab 1993;74: safety and effect of topically applied recombinant 26. Tsuchida Y. The effect of aging and arteriosclerosis basic fibroblast growth factor on the healing of on human skin blood flow. J Derm Sci 1993;5: chronic pressure sores. Ann Surg 1992;216: Gentzkow GD, Pollack SV, Kloth LC, Stubbs HA. 27. Skene AI, Smith ]M, Dore CJ, et al. Venous leg Improved healing of pressure ulcers using Derma- ulcers: A prog-nostic index to predict time to healpulse, a new electrical stimulation device. Wounds ing. 8MJ 1992;305: ;3: Van Rijswijk L, The Multi-eenter Leg Ulcer Study 10. Van Rijswijk L. Full-thickness pressure ulcers: Group. Full-thickness leg ulcers: Patient demo- Patient and wound healing characteristics. Decubitus graphics and predictors of healing. J Fam Pract 1993;6: ;36: Ferrell 5A, Osterweil D, Chris-tenson P. A random- 29. Cordts PR, Hanrahan LM, Rodriquez AA, et al. A ized trial of low-air-ioss beds for treatment of p.res- prospective, randomized trial of Unna's Boot versus sure ulcers. JAMA 1993;269: DuoDerm CGF hydro-active dressing plus compres- 12. Berlowitz DR, van B Wilking S. The short-term out- sion in the management of venous ulcers. J Vasc come of pressure sores. JAGS 1990;38: SUTg 1992;1: Breslow RA, Halfrisch], Guy OC et al. The impor- '30. Margolis D], Gross EA, Wood CR, Lazarus GS. tance of dietary protein in healing pressure ulcers. Planimetric rate of healing in venous ulcers of the JAGS 1993;41: leg treated with pressure bandage and hydrocolloid 14. Itoh M, Montemayor ]S, Matsumoto Accelerated wound healing of pressure E, et al. ulcers by dressing. J Am Ac Deml 1993;28: pulsed high peak power electromagnetic energy (Dia pulse). Decubitus 1991;4: Xakellis GC, Chrischillies EA. Hydrocolloid versus saline-gauze dressings in treating pressure ulcers: A cost-effectiveness analysis. ATch Phys Med Rehab 1992;73: Kaplan E, Meier P. Nonpara-metric estimation from incomplete observation. J Am Stat Assoc 1958; 53: Polanksy M, van Rijswijk L: Utilizing survival analysis techniques in chronic wound healing studies. WOUNDS 1994;6(4) Allman RM, Laprade CA, Noel LB, et al. Pressure